CushieBlog

Demographics and baseline characteristics

Demographic variable	n = 67 patients
Age, years
Mean (SD)	42.3 (13.1)
Range	20-75
Sex, n (%)
Female	56 (83.6)
CS subtype, n (%)
CD	60 (89.6)
Adrenal CS	7(10.4)
Race, n (%)
White	50 (74.6)
Black/African American	8 (11.9)
Asian	2 (3.0)
Other/unknown	7 (10.4)
24-hour UFC
Mean (SD)	391.5 (1471) µg/24 hours, 1080 (4060) nmol/24 hours
Median (IQR)	135.0 (82.7-220.0) µg/24 hours, 372 (228-607) nmol/24 hours
Range (min-max)	29-12 346 µg/24 hours, 80-34 053 nmol/24 hours
LNSC, n (%)
Normal	3 (4.5)
Abnormal	59 (88.1)
NA	5 (7.5)
Plasma ACTH
Mean (SD)	70.7 (64.1) pg/mL, 15.6 (14.1) pmol/L
Median (IQR)	56.0 (42.0-83.8) pg/mL, 12.3 (9.2-18.4) pmol/L
Range (min-max)	11-416 pg/mL (2.4-91.5 pmol/L)
Prior recurrence at baseline, n (%)	16 (23.9)
Prior transsphenoidal surgery, n (%)	16 (23.9)
1	9 (13.4)
2	7(10.4)

Abbreviations: CD, Cushing disease; CS, Cushing’s syndrome; IQR, interquartile range; LNSC, late-night salivary cortisol; NA, not available; UFC, urinary free cortisol.

Table 2.

Baseline and follow-up data

	Baseline	Longest follow-up	P-value
BMI (kg/m²)
Mean (SD)	33.2 (7.6)	30.6 (8.5)	<.001
Median (IQR)	31.6 (26.8-37.3)	29.3 (25.3-34.8)
LNSC, n (%)			<.001
Normal	3 (4.5)	30 (44.7)
Abnormal	59 (88.1)	16 (23.8)
NA	5 (7.5)	21 (31.3)
DM, n (%)			<.001
DM	28 (41.8)	13 (19.4)
Pre-DM	15 (22.4)	9 (13.4)
Hypertension, n (%)	55 (82.1)	35 (53.7)	<.001
HbA1C (%)			<.001
Total mean (SD)	6.5 (1.8)	5.7 (0.9)
DM/pre-DM mean (SD)	6.9 (1.8)	6.1 (1.0)
Antidiabetic medications, n (%)	20 (29.9)	(22.4)
1	12 (17.9)	(13.4)
2	1 (1.5)	(3.0)
3	3 (4.5)	(1.5)
Insulin	4 (6.0)	3 (4.5)
Antihypertensive medications, n (%)	34 (50.7)	(37.3)
1	15 (22.4)	(19.4)
2	10 (14.9)	(11.9)
≥3	9 (13.4)	4 (6.0)
Other medications, n (%)
Antidepressants	10 (14.9)	13 (19.4)
Anxiolytics	12 (17.9)	12(17.9)
Pain medications	16 (23.9)	23 (34.3)
Sleep medications	16 (23.9)	21 (31.3)
Treatment at most recent follow-up,^a n (%)
Transsphenoidal surgery		44 (65.7)
Medical therapy		18 (26.9)
Bilateral adrenalectomy		3 (4.5)
Radiation therapy		1 (1.5)
Adrenalectomy (adrenal CS)		7 (10.4)

Abbreviations: BMI, body mass index; CS, Cushing’s syndrome; DM, diabetes mellitus; HbA1c, hemoglobin A1c; IQR, interquartile range; LNSC, late-night salivary cortisol.

^a“n” refers to number of separate baseline-to-follow-up cases.

In total, there were 46 visits in G1, 31 in G2, and 24 in G3. At the most recent follow-up of each case, there were 24 visits in G1, 25 in G2, and 24 in G3.

The mean (range) duration from baseline to most recent follow-up was 28.3 (5-138) months in the overall cohort. The mean (range) follow-up duration since the most recent treatment was 6.3 (4-9) months for G1, 12.7 (10-18) months for G2, and 43.7 (23-120) months for G3. At their final follow-up visit, 44 patients (65.7%) achieved remission after transsphenoidal surgery (TSS), 18 (26.9%) were under medical control, 3 (4.5%) underwent bilateral adrenalectomy (BLA), 1 (1.5%) received radiation therapy (RT), and the 7 (10.4%) patients with adrenal CS underwent unilateral adrenalectomy (Table 2).

The following additional treatments were administered between this study’s baseline visit and longest follow-up: among the 44 patients treated with TSS at their latest follow-up, 1 underwent an additional TSS and 1 received medical therapy prior to TSS. Of the 18 medically managed patients at last follow-up, 8 (44.4%) had previously undergone TSS (3 of whom had 2 TSSs), and 2 of these 8 additionally received at least 1 different medication before switching to the 1 recorded at their last follow-up. Two (11.1%) other patients received 2 sequential medications before the final 1 at follow-up, and 1 (5.6%) patient was on a block-and-replace regimen with hydrocortisone (HC) after 2 TSSs and BLA. The complete treatment journey of patients on medical therapy, before and after entering the study, is shown in Fig. 1. Among the patients who underwent BLA at last follow-up, 1 had 2 prior TSSs, 1 had a sin1 gle prior TSS and received medical therapy and had 2 TSSs and received medical therapy. The patient treated with RT had 2 prior TSSs and received medical therapy.

Figure 1.

Treatment journey of the 18 patients on medical therapy at their longest follow-up. Each row represents the longitudinal treatment course of each patient before and/or after entering the study. Multiple boxes indicating medical therapy within the same patient represent different medications administered over time. Segments outlined in bold represent the follow-up period analyzed in the current cohort, from this’ study baseline to the longest available follow-up.

Abbreviations: CT, clinical trial; Keto, ketoconazole; Levo, levoketoconazole; Mety, metyrapone; Mife, mifepristone; Osilo, osilodrostat; Pasi, pasireotide.

Sixteen patients presented with recurrent disease; an additional 9 patients (13.4%) developed recurrent or persistent disease after surgery. HC replacement was administered at 21 of the longest available follow-up visits [6 due to ongoing hypopituitarism or adrenal insufficiency (AI) and 15 for temporary postoperative AI], with another 9 cases receiving replacement at intermediate follow-up visits.

All 18 patients on medical therapy at their longest follow-up received adrenal steroidogenesis inhibitors: osilodrostat (8 patients, 44.4%), metyrapone (6 patients, 33.3%), and ketoconazole (4 patients, 22.2%).

Comorbid Conditions

As shown in Table 2, mean body mass index (BMI) at baseline was 33.2 ± 7.6 kg/m². Twenty-eight (41.8%) patients presented with DM, 15 (22.4%) with prediabetes, and 24 (35.8%) without DM. Fifty-five of 67 patients (82.1%) had hypertension at baseline. At the longest follow-up, mean BMI decreased to 30.6 ± 8.5 kg/m² (P < .001), and mean HbA1c decreased to 5.7 ± 0.9% (P < .001). Thirteen patients (19.4%) continued to have DM, and 9 patients (13.4%) had prediabetes. Hypertension was present in 35 patients (53.7%), of whom 25 (71.4%) were receiving at least 1 antihypertensive medication.

LNSC levels remained abnormal in 16 patients (23.8%), although LNSC data were not available for 21 patients (31.3%). Of those, LNSC testing was not considered clinically indicated in some cases, such as patients on HC replacement for postoperative AI (n = 10) or patients with adrenal CS status postadrenalectomy (n = 3). The remaining 8 patients with missing LNSC data were on medical therapy (n = 4) or status post-TSS (n = 4).

Cushing QoL

Sixty-five patients (71 baseline to follow-up case pairs) completed the CushingQoL assessment. In the overall cohort, treatment resulted in significant improvements in mean QoL scores at all follow-up time points: mean change in G1 was 16.6 ± 18.6 (P < .001); G2, 19.1 ± 19.4 (P < .001); and G3, 16.6 ± 27.1 (P = .009) (Table 3, Fig. 2A). For longest available follow-up for each case, overall mean improvement was 18.2 ± 20.9 points (P < .001).

Figure 2.

Score trajectory for (A) Cushing QoL, (B) BDI-II, (C) STAI-State, and (D) STAI-Trait in the overall cohort based on duration of follow-up, including patients with 2 follow-up visits. Significant improvements in mean scores were observed in all assessments and all follow-up time points except in group 3 STAI-State, noted with a gray line. Group 1: 6 months posttreatment, group 2: 12 to 18 months posttreatment, group 3: ≥ 24 months posttreatment.

Abbreviations: BDI-II, Beck Depression Inventory-II; QoL, quality of life; STAI, State-Trait Anxiety Inventory.

Table 3.

Cushing QoL scores at baseline, follow-up visit, and mean score change in each time-based group for total cohort, patients who had TSS and patients on medical therapy

Category	Subgroup	n	Baseline mean	Follow-up visit mean	Mean change	SD (change)	P-value
Total cohort	Longest follow-up	71	42.4	60.6	18.2	20.9	<.001
	Group 1	45	40.6	57.2	16.6	18.6	<.001
	Group 2	30	43.5	62.6	19.1	19.4	<.001
	Group 3	23	41.2	57.9	16.6	27.1	.009
TSS	Longest follow-up	42	40.0	59.9	20.0	18.5	<.001
	Group 1	29	40.2	57.0	16.8	19.1	<.001
	Group 2	21	41.4	61.9	20.4	15.8	<.001
	Group 3	9	29.0	48.7	19.7	24.9	.045
Medical therapy	Longest follow-up	19	46.3	58.4	12.1	26.2	.059
	Group 1	9	44.6	56.7	12.1	18.5	.086
	Group 2	7	40.9	57.1	16.3	31.4	.219
	Group 3	10	56.0	62.0	6.0	27.9	.513

Abbreviations: QoL, quality of life; TSS, transsphenoidal surgery.

In the subanalysis by treatment strategy, 42 patients who completed the Cushing QoL achieved surgical remission and 19 patients were controlled on medical therapy. In the surgical cohort, improvement in scores were noted across all time groups with a mean score increase of 20.0 ± 18.5 points from baseline to the longest available follow-up (P < .001) (Figs. 3A and 4A). Among these patients, 15 had 2 follow-up visits; between them the mean score further increased by 9.6 ± 14.8 points, indicating significant QoL improvement >6 months postsurgery (P = .025). In contrast, patients under medical control at follow-up showed a mean improvement of 12.1 ± 26.2 points from baseline to the longest follow-up, which did not reach statistical significance (n = 19; P = .059) (Table 3, Figs. 3A and 4A).

Figure 3.

Mean score change in total cohort, patients after transsphenoidal surgery, and patients on medical therapy based on duration of follow-up: (A) Cushing QoL, (B) BDI-II, (C) STAI-State, (D) STAI-Trait. Direct comparison between the 2 treatment modalities was performed only in the longest available follow-up visit for each patient. Caps represent SEM. Only P-values ≤ .05 are displayed.

Abbreviations: BDI-II, Beck Depression Inventory-II; QoL, quality of life; STAI, State-Trait Anxiety Inventory.

Figure 4.

Mean scores at baseline and longest follow-up in total cohort, patients after transsphenoidal surgery, and patients on medical therapy for (A) Cushing QoL, (B) BDI-II, (C) STAI-State, (D) STAI-Trait. Caps represent SEM. Asterisks (*) indicate significant change from baseline to follow-up (P ≤ .05) and brackets significant differences between the 2 treatment modalities at the longest follow-up visit (P ≤ .05).

Abbreviations: BDI-II, Beck Depression Inventory-II; QoL, quality of life; STAI, State-Trait Anxiety Inventory.

MID achievement and predictors of improvement

In the overall cohort, CushingQoL MID was achieved in 42 of the 65 patients (64.6%) (Fig. 5). When stratified by follow-up duration, MID achievement rates were 60.8% in G1 (n = 45), 70.0% in G2 (n = 30), and 60.9% (n = 23) in G3.

Figure 5.

MID achievement rates for all patient-reported outcomes at most recent follow-up.

Abbreviations: MID, minimal important difference.

Males (n = 11) improved more than female patients (n = 54) (27.8 ± 13.0 vs 15.5 ± 21.9; P = .020) and achieved the MID more frequently (90.9% vs 59.3%; P = .045). Even though they presented with lower baseline scores compared to females (33.2 ± 16.3 vs 44.3 ± 20.7), that difference was not significant (P = .117).

Score change differed by BMI category, using as cut-off the baseline mean of our cohort (≤33.2 vs >33.2 kg/m²): patients with lower BMI (n = 34) improved considerably more than those with higher BMI (n = 31) (median score change: 26 vs 11; P = .023). Likewise, MID achievement was more common in the low-BMI group (76.5% vs 51.6%; P = .036).

Patients presenting with recurrent disease at baseline (n = 16) reported better baseline QoL than those with primary disease (n = 49) (51.6 ± 19.5 vs 39.5 ± 20.9; P = .046), and their mean improvement following treatment was smaller (7.2 ± 21.0 vs 21.0 ± 19.8; P = .022). Only 43.8% of recurrent cases achieved the MID compared to 71.4% of primary cases (P = .044).

Patients reporting symptom duration ≥3 years prior to diagnosis (n = 29) were less likely to achieve the MID compared to those with shorter symptom duration (n = 35) (48.3% vs 66.7%; P = .008).

Patients with at least 1 abnormal LNSC (n = 15) value at follow-up were less likely to meet MID compared to those with normal LNSC values (n = 28) (33.3% vs 75.0%; P = .008). Similarly, patients requiring HC replacement (after their first TSS or unilateral adrenalectomy for adrenal CS) for >6 months (n = 22) were more likely to achieve MID than those requiring ≤6 months (n = 30) (81.8% vs 50.0%; P = .019).

MID achievement rates between the TSS and medical-therapy groups differed (71.4% vs 47.4%) but did not reach significance (P = .070).

Baseline 24 hours UFC was inversely correlated with baseline CushingQoL score (ρ = −0.3; P = .035), indicating a relationship between biochemical and symptomatic disease severity.

BDI-II

Fifty-six patients (60 case pairs) were included in this subgroup. In the overall cohort, improvements in BDI-II score were seen at all follow-up time points: mean change in G1 was 4.7 ± 9.2 (P = .004); in G2, 7.7 ± 7.3 (P < .001); and in G3, 7.6 ± 10.6 (P = .008). In the overall cohort, mean improvement from baseline to the longest follow-up was 6.8 ± 8.6 points (P < .001) (Table 4, Fig. 2B). Of note, a significant 7.3-point improvement was noted between follow-up G1 (6 months) and follow-up G2 (12 months) (n = 11, P = .025), indicating continued improvement in depressive symptoms over time after treatment.

Table 4.

BDI-II scores at baseline, follow-up visit, and mean score change in each time-based group for total cohort, patients who had TSS and patients on medical therapy

Category	Subgroup	n	Baseline mean	Follow-up visit mean	Mean change	SD (change)	P-value
Total cohort	Longest follow-up	60	15.7	8.9	6.8	8.6	<.001
	Group 1	37	17.0	12.2	4.7	9.2	.004
	Group 2	26	15.2	7.5	7.7	7.3	<.001
	Group 3	18	15.9	8.3	7.6	10.6	.008
TSS	Longest follow-up	32	17.1	8.2	8.8	8.1	<.001
	Group 1	22	18.6	13.6	5.0	10.9	.043
	Group 2	17	14.7	6.7	8.0	8.1	<.001
	Group 3	6	20.5	8.3	12.2	4.7	.001
Medical therapy	Longest follow-up	18	14.4	11.0	3.4	9.9	.159
	Group 1	8	14.6	11.0	3.6	6.7	.171
	Group 2	6	18.3	10.8	7.5	7.1	.049
	Group 3	9	11.8	8.8	3.0	13.3	.517

Abbreviations: BDI-II, Beck Depression Inventory-II; TSS, transsphenoidal surgery.

Among the 32 patients who underwent TSS, improvements were noted across all follow-up time groups, with mean scores decreasing from 17.1 ± 10.9 to 8.2 ± 7.0 at the longest follow-up (P < .001). In contrast, the 18 patients treated medically did not experience a significant change (P = .159). Improvement following TSS was significantly greater than with medical therapy at longest follow-up for each case (8.8 ± 8.1 vs 3.4 ± 9.9; P = .043) (Figs. 3B and 4B).

MID achievement and improvement predictors

Thirty-eight patients (67.9%) achieved MID by their longest follow-up (Fig. 5). Twenty-nine (51.8%) patients had baseline scores ≥14 points, indicating mild or moderate depression, and 23 (79.3%) of these patients met the MID. By follow-up duration, overall MID achievement rates were 56.8% in G1 (n = 37), 76.9% in G2 (n = 26), and 72.2% in G3 (n = 18).

By treatment approach, MID was met by 75.0% of patients who had TSS (n = 32) and 38.9% of patients on medication (n = 18) (P = .012). All patients who underwent BLA (n = 4) or RT (n = 1) and 5 out of 6 patients treated for adrenal CS achieved MID.

Patients with recurrent and primary disease did not differ in terms of baseline score (P = .267). However, those with recurrent disease were less likely to achieve MID (42.9% vs 76.2%; n = 14 vs 75.6%; n = 42, P = .021).

Symptom duration prior to diagnosis was inversely correlated with BDI-II score change (ρ = −0.33, P = .016). Patients experiencing symptoms for ≥3 years (n = 24) exhibited lower MID achievement rates compared to those with shorter symptom duration (n = 31) (50.0% vs 83.9%; P = .007).

Patients with normal LNSC at follow-up had higher MID achievement rates (81.5%; n = 27 vs 45.5%; n = 11, P = .026).

STAI

STAI-S

Fifty-six patients (60 case pairs) completed the STAI-State questionnaire. All follow-up time groups exhibited improvements, although in G3 the score decrease did not reach significance. In the overall cohort, mean scores declined from 44.8 ± 14.0 to 35.3 ± 11.2 at the longest follow-up (P < .001) (Table 5).

Table 5.

STAI scores at baseline, follow-up visit, and mean score change in total cohort, patients who had TSS and patients on medical therapy

Outcome	Category	Subgroup	n	Baseline mean	Follow-up visit mean	Mean change	SD (change)	P-value
STAI-State	Total cohort	Longest follow-up	60	44.8	35.3	9.6	12.5	<.001
		Group 1	40	45.9	36.6	9.3	12.3	<.001
		Group 2	25	46.2	35.3	10.8	10.8	<.001
		Group 3	17	42.4	36.1	6.3	13.8	.078
	TSS	Longest follow-up	33	44.4	34.3	10.1	12.3	<.001
		Group 1	24	44.4	35.8	8.6	11.9	.002
		Group 2	16	43.7	33.9	9.8	11.9	.005
		Group 3	7	46.0	37.9	8.1	12.1	.126
	Medical therapy	Longest follow-up	17	47.2	37.4	9.8	14.7	.014
		Group 1	9	50.9	37.2	13.7	13.7	.017
		Group 2	5	56.4	39.8	16.6	8.4	.012
		Group 3	8	36.3	34.6	2.0	14.9	.715
STAI-Trait	Total cohort	Longest follow-up	58	46.0	37.3	8.6	12.6	<.001
		Group 1	36	47.9	40.3	7.6	12.0	<.001
		Group 2	26	45.7	36.0	9.6	10.9	<.001
		Group 3	16	46.7	36.9	9.8	13.2	.010
	TSS	Longest follow-up	31	47.5	36.7	10.7	12.2	<.001
		Group 1	22	47.9	40.6	7.3	11.5	.008
		Group 2	16	46.3	35.9	10.4	11.4	.002
		Group 3	6	54.0	37.8	16.2	7.5	.003
	Medical therapy	Longest follow-up	18	45.1	38.8	6.2	13.4	.065
		Group 1	8	49.5	39.8	9.8	14.0	.089
		Group 2	6	47.5	36.2	11.3	10.9	.052
		Group 3	8	39.3	37.5	1.8	12.7	.709

Abbreviations: STAI, State-Trait Anxiety Inventory; TSS, transsphenoidal surgery.

By treatment modality, state anxiety improved in both the TSS group (10.1 ± 12.3; n = 33; P < .001) and patients on medical therapy (9.8 ± 14.7; n = 17; P = .014) (Figs. 3C and 4C).

MID achievement and improvement predictors

Overall, 30 of 56 (53.5%) patients achieved MID in STAI-State at their longest follow-up visit (Fig. 5). By follow-up duration, MID achievement rates were 52.5% in G1 (n = 40), 56.1% in G2 (n = 25), and 64.7% in G3 (n = 17).

A negative correlation was observed between STAI-S score change and baseline age (ρ = −0.3, P = .029). Patients >40 years old at baseline (n = 29), improved less than younger patients (n = 27) [median score change: 5 vs 13 (P = .017)] and were less likely to meet the MID, with results approaching statistical significance (41.4% vs 66.7%, P = .058).

STAI-T

Fifty-three patients (58 case pairs) were evaluated. In the overall cohort, mean score change from baseline to longest follow-up was 8.6 ± 12.6 points (P < .001). In time-based subgroups the following score reductions were noted: G1: 7.6 ± 12.0 (P < .001), G2: 9.6 ± 10.9 (P < .001), G3: 9.8 ± 13.2 (P = .010) (Fig. 2D). Among patients treated with TSS (n = 31), significant improvement was seen in every subgroup. Patients receiving medical therapy (n = 18) showed numerical but not statistically significant improvement (P = .065) (Table 5, Figs. 3D and 4D).

MID achievement and improvement predictors

STAI-Trait MID was achieved by 28 (52.8%) patients at the longest follow-up (Fig. 5). By follow-up duration, MID achievement rates were 44.4% in G1, 53.8% in G2, and 68.8% in G3.

Patients ≤40 years at baseline (n = 26) improved more than those aged >40 years (n = 27), with results approaching significance [median score change: 14 vs 4 (P = .060)].

Patients with ≥2 Follow-up Visits

Twenty-eight patients had multiple follow-up visits; we stratified by follow-up duration (<2 years vs ≥2 years) [Table S1 (30)].

Cushing QoL

Significant improvements were noted in all groups with pairwise comparisons revealing higher scores in both first and second follow-up, with the mean score changing by 14.9 (P = .002) and 21.5 (P < .001) points, respectively, in total cohort.

BDI-II

Although the overall trajectory demonstrated significant improvement, pairwise comparisons showed no significant changes between baseline and first follow-up. Improvement was noted between baseline and the second follow-up visit (P < .001) and between the 2 treated visits (P = .021) (Table 6).

Table 6.

BDI-II mean scores and pairwise comparisons in patients with 2 follow-up visits

Comparison	Mean score A	Mean score B	Mean difference	P-value
Baseline vs follow-up 1	16.9	13.0	4.846	.200
Baseline vs follow-up 2	16.9	7.1	9.731	<.001
Follow-up 1 vs follow-up 2	13.0	7.1	4.885	.021

Abbreviations: BDI-II, Beck Depression Inventory-II.

STAI-S

Overall, the mean score decreased from 45.9 ± 13.0 at baseline to 38.3 ± 12.4 at the first follow-up and to 36.1 ± 10.9 at the second follow-up (P = .005). In cases with follow-up ≥2 years (n = 13), the score trajectory did not change significantly from baseline (P = .187). In contrast, patients with total follow-up <2 years (n = 11) exhibited significant improvement (P = .008).

STAI-T

Overall, the mean score decreased from 49.2 ± 9.0 at baseline to 39.8 ± 11.6 at first follow-up and further to 36.4 ± 10.5 at second follow-up (P < .001). Significant improvement noted from baseline to both follow-up visits in both subgroups (P < .001).

Regression Analyses for Predictors of Change

In all measurements, after controlling for age and sex, baseline score was an independent predictor of greater change (P < .001) (Table 7). Patients with more impaired QoL, or severe depression and anxiety at baseline, had more room for improvement.

Table 7.

Predictors of mean score change from baseline to most recent follow-up of each patient in univariable and multivariable linear regression analysis

Outcome	Parameter	Univariable analysis			Multivariable analysis
		Estimate	SE	P-value	Estimate	SE	P-value
Cushing QoL score change	Baseline score	−0.50	0.11	<.001	−0.47	0.11	<.001
	Baseline age	−0.05	0.20	.797	−0.04	0.19	.825
	Male sex	12.11	6.83	.081	7.49	6.68	.267
	Baseline age ≤40 (vs >40)	−3.43	5.23	.515	−4.90	4.89	.321
	Normal LNSC (vs abnormal)	−19.98	6.4	.004	−19.39	5.26	.001
	HC replacement >6 months (vs ≤6 months)	10.06	5.90	.095	12.35	4.96	.016
	Primary disease at baseline (vs recurrent)	−13.19	5.86	.028	−6.63	5.60	.241
	Baseline BMI ≤33.2 kg/m² (vs >33.2 kg/m²)	−8.72	5.1	.095	−6.53	4.71	.171
	Symptom duration ❤ years (vs ≥3 years)	−4.60	5.25	.384	−4.55	4.70	.337
	Treatment (TSS vs medical therapy)	−7.87	5.8	.185	−4.23	5.41	.473
BDI-II score change	Baseline score	0.57	0.09	<.001	0.58	0.09	<.001
	Baseline age	−0.08	0.09	.402	0.02	0.08	.797
	Male sex	−0.59	3.07	.848	0.80	2.53	.752
	Baseline age ≤40 (vs >40)	−3.96	4.82	.429	−0.52	2.02	.800
	Normal LNSC (vs abnormal)	−3.01	3.06	.332	−3.27	1.87	.090
	HC replacement >6 months (vs ≤6 months)	0.06	2.577	.980	2.33	1.90	.226
	Primary disease at baseline (vs recurrent)	−4.76	2.63	.076	−2.66	2.17	.224
	Baseline BMI ≤33.2 kg/m² vs >33.2 kg/m²	−3.79	2.29	.104	−1.41	1.90	.462
	Symptom duration ❤ years (vs ≥3 years)	−5.61	2.23	.015	−3.49	1.78	.055
	Treatment (TSS vs medical therapy)	−5.46	2.60	.041	−3.94	2.02	.057
STAI-State score change	Baseline score	0.57	0.09	<.001	0.56	0.09	<.001
	Baseline age	−0.22	0.13	.104	−0.11	0.12	.338
	Male sex	−5.70	4.37	.197	−4.39	3.69	.239
	Baseline age ≤40 (vs >40)	−5.94	3.30	.078	−3.75	2.73	.175
	Normal LNSC (vs abnormal)	−2.15	3.95	.589	−4.47	2.89	.131
	HC replacement >6 months (vs ≤6 months)	0.72	3.45	.836	4.42	2.81	.123
	Primary disease at baseline (vs recurrent)	2.41	3.91	.743	2.14	2.91	.465
	Baseline BMI ≤33.2 kg/m² (vs >33.2 kg/m²)	−2.36	3.38	.488	−0.93	2.56	.716
	Symptom duration ❤ years (vs ≥3 years)	−5.67	3.33	.095	−3.26	2.46	.192
	Treatment (TSS vs medical therapy)	−1.50	3.91	.970	−2.77	2.97	.355
STAI-Trait score change	Baseline score	0.58	0.11	<.001	0.56	0.12	<.001
	Baseline age	−0.20	0.13	.128	−0.07	0.11	.562
	Male sex	−3.09	4.57	.502	−0.83	4.13	.841
	Baseline age ≤40 (vs >40)	−5.45	3.36	.111	−2.55	3.03	.405
	Normal LNSC (vs abnormal)	−6.52	4.23	.133	−6.74	3.44	.059
	HC replacement >6 months (vs ≤6 months)	4.63	3.52	.195	7.11	2.87	.018
	Primary disease at baseline (vs recurrent)	−2.07	3.90	.597	−0.34	3.42	.921
	Baseline BMI ≤33.2 kg/m² (vs >33.2 kg/m²)	−4.95	3.38	.150	−2.59	3.00	.393
	Symptom duration ❤ years (vs ≥3 years)	−5.78	3.37	.093	−4.35	2.80	.127
	Treatment (TSS vs medical therapy)	−4.49	3.74	.236	−3.39	3.11	.281

Each predictor in multivariable analysis was separately evaluated, adjusting for baseline age, sex, and baseline score. In models exploring baseline age <40 years as a categorical variable, continuous baseline age was not included in the multivariable model. Statistically significant results (P ≤ .05) are indicated in bold.

Abbreviations: BDI-II, Beck Depression Inventory-II; BMI, body mass index; HC, hydrocortisone; LNSC, late-night salivary cortisol; QoL, quality of life; STAI, State-Trait Anxiety Inventory; TSS, transsphenoidal surgery.

Cushing QoL

Normal LNSC at follow-up and >6 months of postoperative HC replacement were predictors of QoL score improvement and MID achievement even after adjustment for baseline score, age, and sex. Lower baseline BMI and male sex, although significant in univariable analysis, were no longer significant in the multivariable linear model. However, a BMI < 33.2 kg/m² (P = .034) and symptom duration ❤ years prior to diagnosis (P = .005) remained statistically significant predictors of reaching the MID in the multivariable logistic model (Table 8, Fig. 6). To determine if treatment modality modified the effect of LNSC, we built a model including baseline QoL score, age, sex, follow-up LNSC, and treatment type (TSS vs medical therapy). In this multivariable model, normal LNSC remained a significant predictor of improvement (P = .023).

Figure 6.

MID achievement predictors after multivariable analysis for (A) Cushing QoL, (B) BDI-II, (C) STAI-State. Each predictor was analyzed in a separate logistic regression model after adjustment for baseline score, age, and sex. Predictors for trait anxiety are not shown, as a longer duration of postoperative HC replacement was a significant predictor only in the linear multivariable regression model.

Abbreviations: BDI-II, Beck Depression Inventory-II; HC, hydrocortisone; LNSC, late-night salivary cortisol; QoL, quality of life; STAI, State-Trait Anxiety Inventory; TSS, transsphenoidal surgery.

Table 8.

https://academic.oup.com/jcem/advance-article/doi/10.1210/clinem/dgaf598/8307075?login=false

Predictors of MID achievement from baseline to most recent follow-up of each patient in univariable and multivariable logistic regression models

Outcome	Parameter	Univariable analysis			Multivariable analysis
		Estimate	SE	P-value	Estimate	SE	P-value
Cushing QoL MID achievement	Baseline score	0.94	0.02	<.001	0.94	0.02	<.001
	Baseline age	1.01	0.02	.548	1.02	0.03	.410
	Male sex	6.89	1.09	.076	3.82	1.16	.249
	Baseline age ≤40 (vs >40)	1.01	0.52	.987	1.27	0.62	.704
	Normal LNSC (vs abnormal)	6.00	0.70	.011	22.82	1.17	.007
	HC replacement >6 months (vs ≤6 months)	4.50	0.66	.023	14.49	0.99	.007
	Primary disease at baseline (vs recurrent)	3.21	0.60	.050	1.78	0.68	.400
	Baseline BMI ≤33.2 kg/m² (vs >33.2 kg/m²)	3.05	0.54	.039	4.33	0.69	.034
	Symptom duration ❤ years (vs ≥3 years)	4.29	0.56	.010	9.07	0.78	.005
	Treatment (TSS vs medical therapy)	2.79	0.57	.074	2.36	0.68	.209
BDI-II MID achievement	Baseline score	1.08	0.04	.064	1.08	0.04	.042
	Baseline age	1.02	0.02	.510	1.01	0.03	.613
	Male sex	5.28	1.10	.130	5.76	1.14	.126
	Baseline age ≤40 (vs >40)	1.11	0.57	.854	1.05	0.63	.937
	Normal LNSC (vs abnormal)	5.28	0.78	.033	14.86	1.25	.030
	HC replacement >6 months (vs ≤6 months)	2.00	0.65	.288	2.32	0.71	.236
	Primary disease at baseline (vs recurrent)	4.27	0.65	.026	2.67	0.71	.165
	Baseline BMI ≤33.2 kg/m² (vs >33.2 kg/m²)	1.94	0.58	.255	1.55	0.66	.504
	Symptom duration < 3 years (vs ≥3 years)	5.20	0.64	.010	5.74	0.70	.012
	Treatment (TSS vs medical therapy)	4.71	0.63	.014	4.19	0.69	.039
STAI-State MID achievement	Baseline score	1.17	0.04	<.001	1.19	0.05	<.001
	Baseline age	0.97	0.02	.241	0.96	0.03	.261
	Male sex	1.95	0.71	.347	3.17	1.00	.249
	Baseline age ≤40 (vs >40)	2.83	0.56	.061	5.87	0.89	.048
	Normal LNSC (vs abnormal)	2.02	0.73	.337	2.41	1.04	.396
	HC replacement >6 months (vs ≤6 months)	0.94	0.59	.943	2.66	0.97	.313
	Primary disease at baseline (vs recurrent)	1.21	0.62	.757	2.15	0.92	.408
	Baseline BMI ≤33.2 kg/m² (vs >33.2 kg/m²)	2.05	0.54	.189	1.57	0.82	.584
	Symptom duration < 3 years (vs ≥3 years)	1.39	0.55	.52	0.98	0.77	.980
	Treatment (TSS vs medical therapy)	1.95	0.62	.279	1.44	0.78	.634
STAI-Trait MID achievement	Baseline score	1.17	0.05	<.001	1.17	0.05	<.001
	Baseline age	0.98	0.02	.295	0.97	0.03	.342
	Male sex	2.33	0.75	.257	4.16	1.02	.161
	Baseline age ≤40 (vs >40)	2.12	0.56	.175	2.32	0.76	.265
	Normal LNSC (vs abnormal)	1.78	0.71	.416	1.48	0.96	.686
	HC replacement >6 months (vs ≤6 months)	1.58	0.60	.450	4.21	0.95	.130
	Primary disease at baseline (vs recurrent)	2.45	0.61	.138	2.06	0.90	.421
	Baseline BMI ≤33.2 kg/m² (vs >33.2 kg/m²)	1.98	0.54	.202	1.11	0.79	.891
	Symptom duration < 3 years (vs ≥3 years)	1.09	0.53	.866	0.99	0.71	.984
	Treatment (TSS vs medical therapy)	1.39	0.60	.585	1.18	0.82	.839

Each predictor in multivariable analysis was separately evaluated, adjusting for baseline age, sex and baseline score. In models exploring baseline age <40 years as a categorical variable, continuous baseline age was not included in the multivariable model. Statistically significant results (P ≤ .05) are indicated in bold.

Abbreviations: BDI-II, Beck Depression Inventory-II; BMI, body mass index; HC, hydrocortisone; LNSC, late-night salivary cortisol; MID, minimal important difference; QoL, quality of life; STAI, State-Trait Anxiety Inventory; TSS, transsphenoidal surgery.

BDI-II

Symptom duration ❤ years (P = .012), normal LNSC at follow-up (P = .030), and TSS (P = .039) instead of medical therapy (for CD) were statistically significant predictors of MID achievement in the multivariable logistic models even after adjusting for age, sex, and baseline score (Table 8, Fig. 6).

STAI-S

In the multivariable logistic model adjusted for sex and baseline score, age <40 predicted higher odds of MID achievement (P = .041) (Table 8, Fig. 6).

STAI-T

After adjustments for sex and baseline score, age group <40 was no longer a predictor of improvement. Although nonsignificant in univariable screening, duration of postoperative HC replacement >6 months emerged as a significant predictor of score change, though not MID achievement, after adjusting for age, sex, and baseline score (Tables 7 and 8).

Discussion

In a clinical practice cohort of patients with CS followed prospectively before and over time up to 11.5 years after surgical remission and/or biochemical control from medical treatment, we identified significant improvements in mean QoL, depression, and anxiety scores in the overall cohort, but only half of patients achieved clinically meaningful improvements in anxiety, as assessed by MID, and about two-thirds of the cohort achieved clinically meaningful improvements in QoL and depression at their most recent follow-up. When assessed by treatment strategy, surgery resulted in statistically significant improvements in all 3 measures, whereas medical therapy resulted in statistically significant improvements in state anxiety but not QoL or depression. These findings may be impacted by the smaller cohort size of the medically treated patients and more complex treatment journeys in the medically vs surgically treated patients. Overall, in this cohort of treated, biochemically controlled patients, several predictors of improvements were identified, including age, baseline BMI, duration of symptoms prior to treatment, duration of HC requirement after surgery, and LNSC normalization with treatment.

PRO studies in CS have shown that patients with CS are at risk for mood disorders and impaired QoL at diagnosis and that improvement posttreatment is often partial, delayed, or inconsistent, even after biochemical remission (3-12). The most recent prospective study confirmed persistent deficits in QoL and depressive symptoms up to 1 year postsurgery, with mean BDI-II scores remaining in the clinically significant range (9). As for anxiety, a prospective study reported high baseline anxiety in patients with CD, and, although it improved after surgery, a proportion continued to experience anxiety up to 1 year posttreatment (14). Neuroimaging supports a biological basis for these symptoms, with brain abnormalities (hippocampal atrophy, cortical thinning, white matter damage) seen after biochemical cure possibly explaining the long-term emotional and cognitive deficits in some patients (12, 15). As for previously reported predictors of improvement, male sex, lower BMI at follow-up (4, 31, 32), LNSC normalization (17), and shorter duration of cortisol exposure (32, 33) emerged as independent predictors of better QoL. Persistent hormone deficits or arginine vasopressin deficiency were related to worse depression (9) while increased age and male sex predicted less anxiety (31). While some studies suggest that hypopituitarism and HC replacement are associated with poorer outcomes (11, 34), others found no significant difference (35). Limitations of these studies include the cross-sectional design (4, 31-36), small cohort sizes (9), and lack of long-term follow-up >12 months (37), especially in the setting of clinical trials (17).

In our study, QoL, depression, and anxiety improved following treatment, but the patterns varied by domain and follow-up duration.

As for QoL, interestingly, patients with recurrent disease showed better baseline QoL scores than those with primary disease, possibly due to posttreatment surveillance, resulting in earlier diagnosis at recurrence vs initial presentation. Although patients on medical therapy showed a trend toward improvement with treatment, results did not reach significance, potentially due to sample size or the increased (better) baseline scores in patients with recurrent disease and thus those receiving medical treatment. Most patients on medical therapy had persistent or recurrent disease and have experienced longer, more complex treatment journeys (as depicted in Fig. 1) compared to those in surgical remission, which also may impact QoL and mood outcomes. Notably, in patients with 2 follow-up visits, QoL continued to significantly improve 6 months posttreatment in those treated surgically but not in the total cohort.

Multivariable analysis revealed several predictors of QoL improvement after treatment. LNSC normalization was independently associated with approximately 20 times higher odds of achieving the MID, indicating the clinical importance of recovery of cortisol circadian rhythm for treated CS patients and the need for further work to identify medical therapies and regimens that can facilitate this. Postoperative HC replacement for more than 6 months after surgery (indicating a longer hypothalamic-pituitary-adrenal axis recovery) was also associated with greater QoL improvement. This finding complements prior work showing an association between duration of postoperative HC replacement and long-term remission (38, 39). Lower baseline BMI and shorter symptom duration were predictive of MID achievement, though not of mean score change.

As for depression, patients with 2 follow-ups had a distinct pattern: no significant change between baseline and first follow-up but significant improvement between the 2 follow-up visits. This suggests that depression may take longer to improve, with more evident change >6 months after biochemical control, which contrasts prior work suggesting that anxiety takes longer than depression to improve (14). The delayed trajectory could reflect the structural brain changes seen in CS even in remission, which are partially reversible (12, 40). Our data showed that symptom duration > 3 years prior to diagnosis reduced MID achievement, consistent with the literature linking diagnostic delay to persistent depression (33). A normal follow-up LNSC was associated with approximately 15 times higher odds of achieving the MID after adjustment, again emphasizing the need to attempt LNSC normalization while on medical therapy (9, 17).

As for anxiety, to date, no prospective study has assessed anxiety longitudinally using STAI, the gold standard for measuring and differentiating between trait and state anxiety (29). Our results confirm that anxiety improves after treatment; however, state and trait show different patterns. State anxiety was the only domain overall to improve significantly in the medical therapy group, while trait anxiety showed only a trend. Although age <40 predicted greater anxiety improvements in both, this remained significant only for state anxiety after adjustment in the logistic model. Trait anxiety improvements were predicted by longer postoperative HC replacement in the linear multivariable model, again suggesting that a shorter recovery time of the HPA axis may be an early indicator for identifying patients who require a closer follow-up. A normal LNSC at follow-up approached significance in the multivariable linear model, suggesting the importance of circadian rhythm recovery in trait anxiety improvement as well.

Across all measures, we found no baseline or outcome differences between pituitary and adrenal CS or between those on or off HC replacement at their last follow-up. Of note, our cohort was predominantly CD patients, and the small number of adrenal CS patients may limit the ability to detect a difference in the 2 cohorts.

Overall, discrepancies between mean change and MID achievement, as reflected in the linear and logistic models, respectively, highlight the importance of reporting both metrics when available, as they may capture different but clinically useful predictors.

We also observed differences between score change and MID achievement across different time groups within the same questionnaire. In STAI-State, G2 (12-18 months since most recent treatment) had greater score reductions than G3 (24 months or more posttreatment)—though change in G3 was nearly significant. However, a higher proportion of patients in G3 achieved MID. Looking at our data, G3 had the highest SD of mean change, indicating greater heterogeneity in treatment response, likely due to broader range of follow-up duration or higher medical therapy rates among patients: 45.5% (n = 10) in G3 vs 22.6% (n = 6) in G2% and 20% (n = 8) in G1. This variability in state anxiety is reflected in the subgroup of patients with 2 follow-up visits: those followed for >2 years showed no significant improvement, while those with <2 years did. Differential responses to long-term medical therapy, higher rates of loss to follow-up among postsurgical patients, or the negative impact of time on state anxiety symptoms may explain this. For BDI-II we used a percentage-based MID, which likely contributed to greater alignment with mean changes, and accounted for individual variability and baseline severity, factors especially relevant when applying generic tools in disease-specific contexts.

Of note, in the cohort overall, the mean follow-up score was within the normal range for depression (<14 for BDI-II) and anxiety (<40 for STAI) (41). This is an encouraging finding that, on average, patients with treated CS may have rates of depression and anxiety that are not clinically significant. Nevertheless, as shown in Table 2, rates of antidepressant, anxiolytic, pain, and sleep medication use did not decrease with treatment but instead were stable or increased numerically, although they were not statistically significant. Similarly, case-control studies have reported higher depression and anxiety levels in patients with CS in remission when compared to healthy controls, even if the mean scores were within the normal range for both groups (15, 42). Whether this difference is clinically significant still remains inconclusive. Taken together, these results emphasize the importance of multidisciplinary pituitary centers that integrate formal psychological services, including psychiatric care and social work support, to monitor and promote long-term mental health in this population.

Inclusion of both surgically and medically treated patients may be considered a limitation to the study, since it introduces heterogeneity in the cohort. However, including patients undergoing a range of treatments allows for analysis of CS cohorts as seen in a real-world practice rather than a controlled clinical trial setting, thus providing clinically valuable information. Another limitation of the study is the use of clinically available, rather than centralized, hormone assays, again introducing variability in our data. As this cohort included patients treated at our center, their endocrine testing followed standard of care, which did not include sending samples to a centralized laboratory. The use of antidepressants in a minority of patients could potentially affect depression scores. However, this is an unavoidable reality in patients with CS, and their use was stable over time (14.9% at baseline vs 19.4% at follow-up, P = .49). Given our prospective study design, which captured each patient’s change relative to their own baseline, and adjustment for baseline scores in multivariable models, any confounding is likely limited.

Despite these limitations, our data contribute to the literature as the largest clinical practice cohort to date that prospectively characterizes QoL and mood disturbances in CS patients, before and over time after achieving biochemical control. By incorporating 3 longitudinal time points, we identified that the greatest improvements occur within the first 6 months for QoL and anxiety, while depression improves more gradually beyond that point. Another strength of our approach is the use of score change and MID as outcomes when exploring potential predictors of improvement and not remission score per se, enabling more precise tracking of each patient’s progress and supporting an individualized approach by accounting for baseline severity.

In summary, this prospective analysis of mood and Qol in a clinical practice cohort of patients with CS showed that effective treatment of hypercortisolism improves depression, anxiety, and QoL, but one-third to one-half of patients do not experience clinically meaningful improvements in these measures. We identified predictors of improvement that highlight the need for early detection of CS and treatment strategies that allow for recovery of cortisol circadian rhythm. Psychological recovery in CS is heterogeneous, domain-specific, and not always aligned with biochemical normalization. Our findings support a model of care that extends beyond endocrine remission, integrating psychosocial follow-up and individualized treatment.

Acknowledgments

We would like to thank the people with Cushing’s syndrome who contributed their valuable time to this research.

Funding

This research was funded by the National Institutes of Health/National Cancer Institute Support Grant P30 CA008748.

Filed under: Clinical trials, Cushing's, pituitary, symptoms | Tagged: depression, diabetes, hypertension, immune suppression, medication, mood, obesity, osteoporosis, post-op, surgery | Leave a comment »

The CuPeR Model: A Dynamic Online Tool for Predicting Cushing’s Disease Persistence and Recurrence After Pituitary Surgery

Posted on October 30, 2025 by MaryO

Abstract

Objective

Predicting postoperative persistence and recurrence of Cushing’s disease (CD) remains a clinical challenge, with no universally reliable models available. This study introduces the CuPeR model, an online dynamic nomogram developed to address these gaps by predicting postoperative outcomes in patients with CD undergoing pituitary surgery.

Methods

A retrospective cohort of 211 patients treated for CD between 2010 and 2024 was analyzed. Key patient and tumor characteristics, imaging findings, and treatment details were evaluated. Multivariate logistic regression identified independent predictors of postoperative persistence or recurrence of CD (PoRP-CD), which were then incorporated into the CuPeR model using stepwise selection based on Akaike Information Criterion. Internal validation was performed using a testing dataset, and a user-friendly online nomogram was developed to facilitate immediate, patient-specific risk estimation in clinical practice.

Results

The final predictive model identified four key factors: symptom duration, MRI Hardy’s grade, tumor site, and prior pituitary surgery. Longer symptom duration and a history of prior surgery significantly increased the risk of recurrence, while bilateral tumor location reduced this risk. The model demonstrated an area under the receiver operating characteristic curve (AUC-ROC) of 0.70, with 83% accuracy, specificity of 96%, and sensitivity of 33%.

Conclusions

The CuPeR model may offer a practical tool for predicting PoRP-CD, enhancing preoperative decision-making by providing personalized risk assessments.

Keywords

Cushing’s disease

Transsphenoidal surgery

Nomogram

Recurrence

Disease Persistence

Abbreviations

ACTH

Adrenocorticotropic Hormone

AIC

Akaike Information Criterion

AUC

Area Under the Curve

BMI

Body Mass Index

Cushing’s Disease

Confidence Interval

CRH

Corticotropin-Releasing Hormone

DFS

Disease-Free Survival

Deep Learning

eTSS

Endoscopic Transsphenoidal Surgery

Hazard Ratio

IPSS

Inferior Petrosal Sinus Sampling

Machine Learning

MRI

Magnetic Resonance Imaging

Overall Survival

PoRP-CD

Persistent or Recurrent Cushing’s Disease

SIADH

Syndrome of Inappropriate Antidiuretic Hormone Secretion

TSS

Transsphenoidal Surgery

UFC

Urinary Free Cortisol

Introduction

Cushing’s disease (CD) is a rare endocrine disorder, with an annual incidence rate of approximately 0.24 cases per 100,000 individuals [1]. Transsphenoidal surgery (TSS), performed using either endoscopic or microscopic approaches, remains the cornerstone of treatment for CD. Notably, meta-analytical studies have reported that TSS achieves remission and provides long-term disease control in 71–80 % of patients [[2], [3], [4]]. The remaining cases may experience persistent disease despite surgery, while others may face disease recurrence despite initial remission. In such cases, additional treatment options include second pituitary surgery, pituitary irradiation, targeted medical therapies, and bilateral adrenalectomy, each with varying success rates ranging from 25 % for medical therapy to 100 % for bilateral adrenalectomy [5].

To date, no single predictive factor has proven effective in reliably forecasting treatment outcomes in patients with CD [6]. This underscores the critical need for developing predictive models to assess the likelihood of postoperative recurrence or persistence of Cushing’s disease (PoRP-CD). However, only a limited number of studies have addressed this gap. Notably, two studies from Peking Union Medical College Hospital attempted to tackle this issue using machine learning (ML) and deep learning (DL) approaches [6,7]. These studies utilized demographic, clinical, and paraclinical variables to construct predictive models, with DL approaches showing potential to enhance predictive accuracy [7]. While the results of these models were promising, their applicability in routine clinical practice remains limited. Both studies focused exclusively on patients undergoing their initial transsphenoidal surgery, making them less applicable for cases involving patients with a prior history of pituitary surgery or radiotherapy. Furthermore, these models incorporated both preoperative and postoperative parameters, such as changes in cortisol and adrenocorticotropic hormone (ACTH) levels. However, serum cortisol, ACTH, and comprehensive endocrine testing should be available before any treatment decisions are made, and each patient should ideally be reviewed by a multidisciplinary tumor board, including neurosurgery, radiology, endocrinology, and oncology, prior to pituitary surgery. As such, more comprehensive and practical predictive tool that can support timely clinical decision-making and accommodate a broader range of patient scenarios in the management of CD.

The current study was designed to address these critical limitations and provide a more practical solution for predicting postoperative outcomes in CD. Applying one of the largest available CD cohorts, we incorporated a wide array of patient and tumor characteristics, imaging findings, and treatment details to develop a robust and comprehensive predictive model. This model offers treating surgeons reliable insights into the likelihood of tumor recurrence or persistence. By providing individualized risk predictions, the model is intended to assist clinicians in considering different therapeutic options before pituitary surgery, complementing—but not replacing—standard multidisciplinary decision-making. To further enhance its utility in clinical practice, we also developed an interactive online dynamic nomogram, allowing individualized predictions of postoperative persistence or recurrence.

Methods

Study design, patients, and endpoints

The experimental protocol was approved by the Institutional Review Board of Shahid Beheshti University of Medical Sciences (Tehran, Iran). This retrospective study investigates the clinical outcomes of pituitary surgery in patients with CD underwent pituitary surgery between 2010 and 2024 in the neurosurgery department at Loghman Hakim Hospital. Surgeries were conducted by a group of experienced neurosurgeons under the supervision of the first author (G.S). The primary objective of this study was to develop and validate a predictive model for assessing the risk of PoRP-CD. The secondary objectives were (a) to summarize patient and tumor characteristics; (b) to report surgical outcomes and remission rates following surgery; and (c) to analyze patient survival. This study was performed in accordance with the Declaration of Helsinki, and adheres to the reporting guidelines outlined in the STROBE Statement. Due to retrospective nature of the study informed consent was waived by Shahid Beheshti University of Medical Sciences Ethics Committee. All methods were performed in accordance with the relevant guidelines and regulations.

Preoperative assessments

The “index surgery” was set to the most recent pituitary surgery. Before the index surgery, patients underwent comprehensive clinical evaluations, including biochemical and neurological assessments as well as visual field examinations. This research utilized the Endocrine Society Clinical Practice Guideline to establish the diagnosis of CD [8]. Three main steps were involved in the diagnostic process: in the first step, the focus was on detecting hypercortisolemia, which was determined by examining 24-hour urinary free cortisol levels (normal: <60 mcg/24 h), as well as plasma and salivary cortisol profiles. Low-dose dexamethasone suppression testing was performed using the 2 mg/48 h protocol, which was the standard practice in our institution during the study period (2010 onward) [8]. The second step aimed to confirm ACTH-dependent cause of hypercortisolemia, through measuring plasma ACTH levels. The final step aimed to distinguish Cushing’s disease from ectopic sources of ACTH. This was performed using a high-dose dexamethasone suppression test (8 mg overnight), with a plasma cortisol suppression exceeding 50 % typically considered indicative of a pituitary origin [9].

Next, the patients were subjected to thin-slice (3 mm) 1.5-tesla dynamic pituitary gland magnetic resonance imaging (MRI) with gadolinium contrast. The MR evaluation adhered to a strict protocol, requiring an independent agreement of treating neurosurgeon and radiologist to confirm the diagnosis. MR scans were categorized according to the Hardy and Knosp classifications [10]. Normal scans required to demonstrate the absence of direct signs, including inhomogeneity in the pituitary, as well as indirect signs such as a deviation of the pituitary stalk, bulging or erosion of the Sella contour. In cases where the CD was confirmed but pituitary MRI was inconclusive, bilateral inferior petrosal sinus sampling (IPSS) was performed per standard protocol under corticotropin-releasing hormone (CRH) stimulation [11]. Patients with macroadenoma or signs of elevating the optic chiasm were candidates for Humphrey visual field examination.

Surgical approach

Patients underwent endoscopic transsphenoidal approach using conventional “Two Nostrils–Four Hands” technique [12]. Given the diminutive size and deep-seated location of most adenomas, locating the adenoma emerged as a formidable challenge, particularly when the tumor remained not visualized in pre-operative imaging studies. The surgical procedure entailed extensive drilling of the Sellar floor laterally up to the carotid artery on both sides, providing a comprehensive view of the medial wall of the cavernous sinus and exposure of the anterior and posterior intercavernous sinuses. The exploration of the entire Sella commenced in the region where the original tumor had been localized. Upon identification of a tumor, a selective adenomectomy was performed, accompanied by a thorough inspection of the pituitary gland to detect and eliminate any potential tumor remnants. The removal of any pseudo capsule was executed meticulously.

The primary surgical objective was selective adenomectomy, with further exploration guided by the side recommended by IPSS in cases where no adenoma was initially observed. The exploration involved making a plus-like incision on the corresponding half of the gland, enabling deep exploration to leave no part unexplored. In instances where creamy material suggestive of a tumor was drained after a pituitary incision, a tissue biopsy was obtained, although it was not conclusively considered a tumor. Exploration continued on the opposite side in such cases.

When no distinct adenoma was found, a peri-glandular inspection was conducted to visualize the medial wall of the cavernous sinus, diaphragm, and Sellar floor, aiming to detect an ectopic microadenoma. If an apparent tumor remained undetected, the procedure was repeated on the contralateral side, and a vertical medial incision on the pituitary gland adjacent to the pituitary stalk and neurohypophysis was made as a final effort for tumor detection. In the absence of identified pathology during the surgical procedure, hemi-hypophysectomy was considered on the side where IPSS had detected the gradient or on the side with an apparent or suspicious MRI finding. Considering the typical central location of corticotroph cells in the pituitary gland, microadenoma exploration extended posteriorly and medially to confirm extirpation.

Postoperative assessments

In this study, the patients were closely monitored for signs of diabetes insipidus and syndrome of inappropriate antidiuretic hormone secretion (SIADH). Serum sodium levels, urine-specific gravity, and volume were checked regularly. Following surgery, morning cortisol levels were measured on the first day, and other anterior pituitary hormones were evaluated on day 3. Hydrocortisone therapy was initiated based on the patient’s symptoms, signs of adrenal insufficiency, and low cortisol levels. The first postoperative check-up occurred two weeks after surgery, followed by another at three months, which included a comprehensive assessment of pituitary hormones. This evaluation was repeated every three months for two years and then annually. Additionally, patients underwent a dynamic 1.5-Tesla pituitary MRI at six months post-surgery and annually thereafter, with a minimum follow-up period of 12 months.

Remission was defined as having low cortisol levels, indicated by early morning serum cortisol level ≤ 5 μg/dL within two days post-surgery [13]. Persistent CD was characterized by ongoing hypercortisolism, and postoperative recurrence refers to the reappearance of CD symptoms despite initial remission marked by hypercortisolemia. In case of persistence or recurrence, patients were candidates for second-line treatment options selected by their physicians, including revision surgery, targeted medical therapy, pituitary radiotherapy, or bilateral adrenalectomy. Disease-free survival (DFS) was defined as the time from the index surgery to the first occurrence of disease recurrence or death from any cause, while overall survival (OS) was defined as the time from the index surgery to death from any cause.

Statistical analysis

Categorical variables were expressed as numbers and percentages, and continuous variables as mean, range, and standard deviation. The distribution of variables was checked using the Shapiro-Wilk test, which showed a deviation from normal distribution. Contingency tables were used for categorical variables with Pearson’s Chi-squared or Fisher’s Exact test used to examine their association with outcomes for univariate analyses. For continuous variables, the unpaired t-test was applied to compare means between two independent groups when the data met the assumption of normality. Analyses were conducted with R Statistical Software v4.4.0 (“Puppy Cup”). All statistical inferences were two-sided, and P < 0.05 were considered statistically significant.

Model development and internal validation

The dataset was split by “caret package” into a training set (70 %) and a testing set (30 %) using stratified sampling to ensure representative proportions of outcomes. Binary logistic regression was used to identify predictors of PoRP-CD. Patients with adequate follow-up data were included in the analysis. The variables with a marginal level of association (P < 0.15) in the univariate analysis were further included in the multivariate logistic regression analysis to identify the independent predictors of PoRP-CD. Imported factors included demographic, medical history, imaging and pathology results, and treatment details. To identify predictors of PoRP-CD, a multivariable logistic regression model was developed using stepwise selection based on Akaike Information Criterion (AIC). Model performance, including sensitivity, specificity, and area under the receiver operating characteristic curve (AUC-ROC), was evaluated using internal validation on the test dataset.

Nomogram creation and deployment

A nomogram was constructed using the validated logistic regression model. The nomogram was then integrated into a web-based application using the “Shiny package” in R program. The dynamic nomogram allows clinicians to input patient data and obtain individualized risk predictions for PoRP-CD.

Survival analysis

Survival analysis was conducted to evaluate DFS across various patient subgroups. The log-rank test was applied to assess statistical differences in survival distributions between subgroups. Cox proportional hazard regression was used to estimate hazard ratios (HR) and 95 % confidence intervals (CI). The “survival” and “survminer” R packages were applied in this section.

Results

Patients and tumors characteristics

A total of 211 patients with CD had been treated by a group of experienced neurosurgeons under the supervision of the first author (G.S) between March 2010 and January 2024 in the neurosurgery department at Loghman Hakim Hospital. Table 1 summarizes the baseline characteristics of patients at the timepoint of index surgery. The patients had a mean age of 35.9 ± 12.1 years (range: 11–67), among which 21 patients (9.9 %) were in the pediatric age range, and 165 (78.1 %) were female. Obesity was the most common patients’ symptoms (45.9 %), and physical examination reported centripetal obesity (84.3 %), moon face (75.8 %), and striae (64.4 %) as the most common clinical manifestations. Compared to the adult patients, pediatrics had less common hypertension on physical examination (35.2 vs. 5.9 %) and medical history of diabetes mellitus (36.8 vs. 4.7 %) (P < 0.05). The majority of patients (63.9 %, 135/211) had not received any prior treatment. Among those who had, surgery alone was the most common approach (n = 57, 27.0 %), performed once in 50 patients (23.6 %), twice in 6 patients (2.8 %), and three times in a single patient.

Table 1. Baseline characteristics of adult and pediatric patients with Cushing’s disease.

Demographics	Total n = 211	Adults n = 190	Pediatrics n = 21	P	Medical Hx	Total n = 211	Adults n = 190	Ped. n = 21	P	Drug-Family Hx	Total n = 211	Adults n = 190	Ped. n = 21	P
Age; mean-SD (y)	35.9–12.1	38.3–10.2	14.8–1.7	0<.001	Hypertension	97 (45.9)	92 (48.4)	5 (23.8)	0.31	Cabergoline	3 (1.4)	3 (1.5)	0	1.0
Sex; female	165 (78.1) ^a	149 (78.4)	16 (76.1)	0.78	Diabetes mellitus	71 (33.6)	70 (36.8)	1 (4.7)	0<.001	Ketoconazole	12 (5.6)	12 (6.3)	0	0.61
Marital status; married	105 (70.0) ^b	103 (76.8) ^b	2 (12.5) ^b	0<.001	Dyslipidemia	45 (21.3)	42 (22.1)	3 (14.2)	0.56	Metyrapone	0	−	−	−
Smoking status; active–passive-non	17 (10)-27(17)-113(72) ^b	17 (11)–23(15)-101(70) ^b	0–4(25)-12(75) ^b	0.70	Prior pituitary surgery	57 (27.0))	51 (26.8)	6 (28.5)	1.0	Pasireotide	0	−	−	−
Height; mean-SD (cm)	163.9–8.7	163.8–8.9	165.1–6.6	0.59	Fatty liver	37 (17.5)	32 (16.8)	5 (23.8)	0.36	Somatostatin	0	−	−	−
Weight; mean-SD (Kg)	74.1–22.5	74.6–22.5	69.3–23.1	0.58	Thromboembolism	6 (2.8)	6 (3.1)	0	1.0
BMI; mean-SD (Kg/m²)	28.8–6.1	29.0–6.2	27.6–5.5	0.72	DVT	3 (1.4)	3 (1.5)	0	1.0	FH of Cushing	5 (2.3)	4 (2.1)	1 (4.7)	0.43
Symptom duration; mean-SD (m)	30.7–41.2	32.0–43.2	20.0–14.2	0.78	MEN	1 (0.4)	1 (0.5)	0	1.0	FH of MEN	1 (0.4)	1 (0.5)	0	1.0
Presenting Symptoms
Obesity	75 (45.9) ^b	66 (45.2) ^b	9 (52.9) ^b	0.61	Striae	10 (6.1) ^b	8 (5.4) ^b	2 (11.7) ^b	0.27	Headache	4 (2.4) ^b	3 (2.0) ^b	1 (5.8) ^b	0.35
Menstrual disorders	16 (9.8) ^b	13 (8.9) ^b	3 (17.6) ^b	0.22	Edema	7 (4.2) ^b	7 (4.7) ^b	0	1.0	Diabetes mellitus	3 (1.8) ^b	3 (2.0) ^b	0	1.0
Hypertension	12 (7.3) ^b	12 (8.2) ^b	0	0.61	Muscular weakness	7 (4.2) ^b	6 (4.1) ^b	1 (5.8) ^b	0.54	Bone fracture	3 (1.8) ^b	3 (2.0) ^b	0	1.0
Blurred vision	10 (6.1) ^b	9 (6.1) ^b	1 (5.8) ^b	1.0	Moon face	6 (3.6) ^b	6 (4.1) ^b	0	1.0	Other	10 (6.1) ^b	10 (6.8) ^b	0	0.60
Clinical Manifestations
Acanthosis nigricans	35 (16.5)	34 (17.8)	1 (4.7)	0.12	Easy bruising	103 (48.8)	92 (48.4)	11 (52.3)	0.91	Male pat. hair loss	111 (52.6)	100 (52.6)	11 (52.3)	1.0
Acne	68 (32.2)	58 (30.5)	10 (47.6)	0.16	Ecchymosis	58 (27.5)	50 (26.3)	8 (38.0)	0.37	dysmenorrhea	96 (45.4)	84 (44.2)	12 (57.1)	0.49
Ankle edema	105 (49.7)	96 (50.5)	9 (42.8)	0.57	Exophthalmia	50 (23.7)	47 (24.7)	3 (14.2)	0.27	Moon face	160 (75.8)	141 (74.2)	19 (90.4)	0.69
Backache	66 (31.2)	60 (31.5)	6 (28.5)	0.88	Facial plethora	97 (45.9)	85 (44.7)	12 (57.1)	0.33	Osteoporosis	25 (11.8)	25 (13.1)	0	0.14
Blurred vision	70 (33.2)	67 (35.2)	3 (14.2)	0.27	Fatigue	146 (69.2)	130 (68)	16 (76.1)	0.76	Prox. myopathy	94 (44.5)	86 (45.2)	8 (38.0)	0.63
Buffalo hump	123 (58.3)	107 (56.3)	16 (76.1)	0.43	Fracture	12 (5.6)	12 (6.3)	0	0.61	Skin atrophy	81 (38.4)	73 (38.4)	8 (38.0)	1.0
Centripetal obesity	178 (84.3)	159 (83.6)	19 (90.4)	0.50	Headache	109 (51.6)	97 (51.0)	12 (57.1)	1.0	Striae	136 (64.4)	119 (62.6)	17 (80.9)	0.55
Cerebrospinal fluid leakage	5 (2.3)	4 (2.1)	1 (4.7)	0.41	Hirsutism	104 (49.3)	92 (48.4)	12 (57.1)	0.72	Supraclav. fat pad	38 (18.0)	33 (17.3)	5 (23.8)	0.67
Cranial nerve palsy	3 (1.4)	3 (1.5)	0	1.0	Hyperpigmentation	38 (18.0)	37 (19.4)	1 (4.7)	0.12	Visual field defect	24 (11.3)	22 (11.5)	2 (9.5)	1.0
Diplopia	18 (8.5)	15 (7.8)	3 (14.2)	0.41	Hypertension	69 (32.7)	67 (35.2)	2 (9.5)	0.009	Weight gain	108 (51.1)	95 (50.0)	13 (61.9)	0.39
Prior Treatments
Treatment naïve	135 (63.9)	122 (64.2)	13 (61.9)	1.0	Pituitary surgery alone	39 (18.4)	33 (17.3)	6 (28.5)	0.23	Radiotherapy alone	6 (2.8)	5 (2.6)	1 (4.7)	0.47
Medication alone	5 (2.3)	5 (2.6)	0	1.0	Combination therapy	17 (8.1)	17 (8.9)	0	0.22	Adrenalectomy alone	11 (5.2)	10 (5.2)	1 (4.7)	1.0
Hormonal Assessments
Hypothyroidism	24 (31.1) ^b	24 (31.1) ^b	0	0.09	GH deficiency	6 (8.8) ^b	6 (8.8) ^b	0	1.0	Hypogonadism	7 (9.8) ^b	7 (9.8) ^b	0	1.0
Panhypopituitarism	2 (2.5) ^b	2 (2.5) ^b	0	1.0
Imaging Features
Hardy’s grading (sphenoid bone invasion) 0 1 2 3 4	37 (21.1) ^b 102 (58.2) ^b 27 (15.4) ^b 4 (2.2) ^b 5 (2.8) ^b	35 (22.7) ^b 88 (57.1) ^b 23 (14.9) ^b 3 (1.9) ^b 5 (3.2) ^b	2 (9.5) 14 (66.7) 4 (19.0) 1 (4.7) 0	0.45	Hardy’s staging (suprasellar extension) A B C D E	36 (20.4) ^b 86 (48.8) ^b 14 (7.9) ^b 4 (2.2) ^b 36 (20.4) ^b	34 (21.9) ^b 73 (47.1) ^b 14 (9.0) ^b 2 (1.2) ^b 32 (20.6) ^b	2 (9.5) 13 (62) 0 2 (9.5) 4 (19.0)	0.07	Knosp grading 0 1 2 3 4	152 (82.6) ^b 13 (7.0) ^b 7 (3.8) ^b 4 (2.1) ^b 8 (4.3) ^b	135 (82.8) ^b 10 (6.1) ^b 6 (3.6) ^b 4 (2.4) ^b 8 (4.9) ^b	17 (80.9) 3 (14.2) 1 (4.7) 0 0	0.46
Tumor size Microadenoma Macroadenoma MR-negative	122 (58.6) ^b 50 (24.0) ^b 36 17.3) ^b	111 (59.3) ^b 42 (22.4) ^b 34 (18.1) ^b	11 (52.3) 8 (38.0) 2 (9.5)	0.28	Sphenoid shape Sellar Presellar Conchal	205 (97.6) ^b 3 (1.4) ^b 2 (0.9) ^b	184 (97.3) ^b 3 (1.5) ^b 2 (1.0) ^b	21 (100) 0 0	1.0	Multifocality Unifocal Multifocal	113 (80.1) 28 (19.8)	97 (79.5) 25 (20.4)	16(84.2) 3 (15.7)	0.79
Invasion ^c No invasion Cavernous sinus Carotid Dura Clivus	185 (88.5) ^b 12 (5.7) ^b 3 (1.4) ^b 6 (2.8) ^b 3 (1.4) ^b	165 (87.7) ^b 11 (5.8) ^b 3 (1.5) ^b 6 (3.1) ^b 3 (1.5) ^b	20 (95.2) ^b 1 (4.8) ^b 0 0 0	1.0	Tumor site Right lobe Left lobe Bilateral Central Stalk	22 (15.6)) ^b 16 (11.3)) ^b 51 (36.1) ^b 49 (34.7) ^b 3 (2.1) ^b	20 (16.2) ^b 13 (10.5) ^b 43 (34.9) ^b 45 (36.5) ^b 2 (1.6) ^b	2 (11.1) ^b 3 (16.6) ^b 8 (44.4) ^b 4 (22.2) ^b 1 (5.5) ^b	0.38	Empty sella No Yes	207 (98.1) 4 (1.8)	187 (98.4) 3 (1.5)	20(95.2) 1 (4.7)	0.34
Pituitary apoplexy No Yes	185 (97.3) ^b 5 (2.6) ^b	167 (98.2) ^b 3 (1.7) ^b	18 (90.0) ^b 2 (10.0) ^b	0.08	Kissing carotids No Yes	209 (99.0) 2 (0.9)	188 (98.9) 2 (1.0)	21 (100) 0	1.0

a: the numbers in parentheses represent the percentage for each patient group.
b: percentage after ruling out missing data.
c: one patient had invasion to cavernous sinus and carotid and another one had clivus and dural invasion.

A comprehensive preoperative hormonal assessment was conducted on 77 patients (36.4 %), revealing hormonal dysregulation in 28 patients (36.3 %). Hypothyroidism was the most common abnormality, affecting 35 % of those assessed (24 out of 77). On MRI scans, most tumors were microadenomas (58.6 %), with fewer macroadenomas (24.0 %) and some cases with no detectable tumor (17.3 %). Tumors were commonly localized bilaterally (36.1 %) or centrally (34.7 %), and most were unifocal (80.1 %). Knosp grading indicated no cavernous sinus invasion in the majority (82.6 %), with only 6.4 % showing grades 3–4. According to Hardy’s grading, most patients had mild sphenoid bone invasion, predominantly grade 1 (58.2 %). For Hardy’s staging of suprasellar extension, nearly half were at stage B (48.8 %), with smaller groups in stages A and E (20.4 % each), and fewer in stages C and D. Other MRI findings are summarized in Table 1. There was no significant difference between adult and pediatric patients in terms of hormonal and imaging findings (P > 0.05). Pathology reports were available for 36 patients. The most common finding was sparse cellularity, observed in 11 patients (30.6 %) followed by dense cellularity identified in 9 patients (25 %). Crooke cell changes were the least common, present in 7 patients (19.4 %). Nine specimens (25 %) had no tumor identified in the sample submitted to pathology.

Treatment details and outcomes

A total of 36 patients (17.1 %) underwent preoperative IPSS, among which 13 had right lateralization, 13 left, 4 bilateral, 3 central, 2 central-right, and 1 central-left. Pituitary surgery was predominantly performed using the endoscopic transsphenoidal (eTSS) approach (98.5 %, 208/211), while the transplanum approach was used in 3 patients (1.5 %). Adenomectomy was the most common surgical procedure (n = 187, 88.6 %), followed by total hypophysectomy in 17 patients (8.1 %) and hemi-hypophysectomy in 7 patients (3.3 %). In addition, four patients in the total hypophysectomy group and one patient in the adenomectomy group also underwent hypophyseal stalk resection. Information on disease persistence or recurrence was available for 204 patients. Median follow-up of patients was 58.4 months (range: 4.5–170.4 months) after index surgery. In total, 23 patients (11.2 %) experienced persistent disease following the index surgery, while 10 patients (4.9 %) had disease recurrence, with a median time to recurrence of 7 months (range: 1–78 months). The median recurrence-free interval for the entire cohort was 37 months.

The surgical complication rates were as follows (Fig. 1A): cerebrospinal fluid leaks were observed in 22 patients (10.4 %), followed by cranial nerve injury in 7 patients (3.3 %) and meningitis in 5 patients (2.3 %). Carotid injury and intracerebral bleeding each occurred in 3 patients (1.4 %). Nasal bleeding, the need for a ventriculoperitoneal shunt, and embolic events were each reported in 1 patient (0.4 %). Perioperative mortality was observed in one female patient (0.4 %) due to an iatrogenic carotid injury. This patient had previously undergone three pituitary surgeries and received radiotherapy at the pituitary site. Hormonal dysregulation following surgery included hypothyroidism in 99 patients (46.9 %), diabetes insipidus in 76 patients (36 %), hypogonadism in 28 patients (13.2 %), growth hormone deficiency in 10 patients (4.7 %), and panhypopituitarism in 7 patients (3.3 %) (Fig. 1B).

Multivariate analysis on the predictors of Persistent/Recurrent Cushing’s disease

To identify potential predictive factors for PoRP-CD, we conducted a comprehensive binary logistic regression analysis, examining key clinical and imaging variables (Table 2). In the univariate analysis, factors including symptom duration (OR [odds ratio] 1.01, 95 % CI [confidence interval] 1.00–1.02, P = 0.04), MRI Hardy’s grade (OR 1.62, 95 % CI 0.98–2.69, P = 0.05), and previous pituitary surgery (OR 3.56, 95 % CI 1.39–9.07, P = 0.007) demonstrated significant association with PoRP-CD. MR-reported tumor size showed increased odds of recurrence with an increased tumor size (OR for microadenoma vs. no tumor: 2.41, 95 % CI: 0.50–11.53; OR for macroadenoma vs. no tumor: 4.15, 95 % CI 0.80–21.42), though the effect was not statistically significant (P > 0.05). To impede missing the marginal significant factors, three factors with P values between 0.05 and 0.15 were also included in the multivariate analysis, including “MRI Knosp grading”, “MR-reported tumor site”, and “previous pituitary radiotherapy”. In the multivariate analysis, “symptom duration” was positively correlated with recurrence, with an odds ratio (OR) of 1.03 (95 % CI: 1.01–1.06, P = 0.01), indicating a higher risk of recurrence with prolonged symptoms. Additionally, a history of “previous pituitary surgery” was significantly associated with recurrence, with an OR of 4.67 (95 % CI: 1.04–20.89, P = 0.04). Other factors, including tumor grading, tumor site, and previous radiotherapy, did not reach statistical significance.

Table 2. Regression analysis of patient and tumor’s factors related to postoperative persistence or recurrence in Cushing disease.

Parameters	Univariate Analysis		Multivariate Analysis
Parameters	OR (95 % CI)	P	OR (95 % CI)	P
Age	0.97 (0.94–1.01)	0.23
Sex (male vs. female)	1.17 (0.39–3.50)	0.77
Smoking (active smoker vs. non)	0.78 (0.65–10.28)	0.77
Family history of CD (positive vs. negative)	0.01 (0–Inf)	0.99
Family history of MEN (positive vs. negative)	0.01 (0–Inf)	0.99
Preoperative BMI	1.03 (0.94–1.13)	0.43
Symptom duration	1.01 (1.00–1.02)	0.04 **	1.03 (1.01–1.06)	0.01 **
Preop serum ACTH (high vs. normal)	0.88 (0.13–6.00)	0.90
Preop free serum cortisol (high vs. normal)	1.18 (0.40–3.45)	0.74
Preop urine free cortisol (high vs. normal)	0.15 (0.01–2.98)	0.21
Knosp grading (ref: grade 0)	1.41 (0.93–2.15)	0.10 *	1.56 (0.61–3.97)	0.34
Hardy’s grading (ref: grade 0)	1.62 (0.98–2.69)	0.05 **	1.98 (0.54–7.21)	0.29
Hardy’s staging (ref: stage A)	2.97 (0.61–14.38)	0.17
Tumor size Macro vs. no tumor Micro vs. no tumor	4.15 (0.80–21.42) 2.41 (0.50–11.53)	0.17
Multifocality (multifocal vs. unifocal)	1.68 (0.44–6.42)	0.44
MR-based tumor site^a Bilateral vs. central Left vs. central Right vs. central Stalk vs. central	0.16 (0.01–1.53) 0.82 (0.18–4.40) 0.49 (0.09–2.82) 5.33 (0.37–144.16)	0.14 *	0.34 (0.02–3.95) 0.23 (0.01–3.12) 5.36 (0.19–146.38) (0.0002–0.67)	0.03 ** 0.39 0.27 0.31
Invasion (pos. vs. neg.)	1.18 (0.31–4.51)	0.80
Surgical approach (transplanum vs. eTSS)	6.21 (0.37–103.55)	0.20
Surgical type (adenomectomy vs. hypophysectomy)	1.55 (0.46–5.22)	0.47
Histopathology Dense type vs. Crooke’s cell adenoma Normal appearing vs. Crooke’s cell adenoma Sparse type vs. Crooke’s cell adenoma	2.00 (0.09–69.06) 0.80 (0.04–23.23) 0.28 (0.01–9.45)	0.56
Ki-67 (>3% vs. ≤ 3 %)	1.34 (0.14–12.64)	0.79
Previous pituitary surgery (yes vs. no)	3.56 (1.39–9.07)	0.007 **	4.67 (1.04–20.89)	0.04 **
Previous pituitary radiotherapy (yes vs. no)	3.36 (0.89–12.62)	0.07 *	3.63 (0.28–46.07)	0.31
Postop decrease in BMI	0.90 (0.73–1.03)	0.22

Abbreviations: ACTH − Adrenocorticotropic Hormone; BMI − Body Mass Index; CD − Cushing’s Disease; CI − Confidence Interval; eTSS − Endoscopic Transsphenoidal Surgery; Inf − Infinity; MEN − Multiple Endocrine Neoplasia; MR − Magnetic Resonance; OR − Odds Ratio; PoRP-CD − Persistent or Recurrent Cushing’s Disease; Preop − Preoperative; Postop − Postoperative.

^aMR-reported.

* Significant at the level of 0.15.

** Significant at the level of 0.05.

The stepwise selection–in both forward and backward directions–retained four predictors— symptom duration, Hardy’s grading, tumor site, and prior surgery —for the final model. The final multivariate model with four predictors of “symptom duration”, “MRI Hardy’s grading”, “tumor site”, and “previous pituitary surgery” demonstrated significant associations for “symptom duration” (OR 1.03, 95 % CI 1.005–1.05, P = 0.02), previous pituitary surgery (OR 4.61, 95 % CI 1.12–22.0, P = 0.03), and a certain tumor site; tumors located bilaterally had significantly lower odds of recurrence compared to central tumors (OR 0.01, 95 % CI 0.0002–0.45, P = 0.02). On the testing dataset, the four-factor model achieved an AUC of 0.70, specificity of 96 %, and sensitivity of 33 %. The model’s accuracy in predicting PoRP-CD is 83 %.

Predicting persistent or recurrent Cushing’s disease–The CuPeR nomogram

A nomogram was developed based on the multivariate model comprising four key predictors: “Symptom duration”, “MRI Hardy’s grading”, “Previous pituitary surgery”, and “MRI-reported tumor site” (Fig. 2). This nomogram visually represents the impact of each predictor on the likelihood of PoRP-CD. The total score derived from the nomogram aligns with the probability scales, allowing for estimation of the risk of PoRP-CD. Higher cumulative points correspond to an increased likelihood of persistent or recurrent disease. To facilitate individualized predictions of postoperative persistence or recurrence, we developed an online dynamic nomogram (link: https://cushing.shinyapps.io/cuper/).

Survival analysis

Survival analysis demonstrated a steady, gradual decline in DFS across the entire cohort, with the median DFS not reached despite substantial follow-up (Fig. 3A). Among the predefined variables, Hardy’s Grade 3 was associated with a significantly worse DFS compared with Grade 0 (HR = 6.02, 95 % CI: 1.09–33.02, P = 0.03) (Fig. 3B), whereas other Hardy’s Grades did not reach statistical significance (P > 0.05). Regarding tumor site, no site was a statistically significant risk factor for DFS; stalk tumors showed a trend toward poorer DFS but did not reach significance (HR = 5.09, 95 % CI: 0.84–30.63, P = 0.07) (Fig. 3C). Patients with a history of previous pituitary surgery had significantly worse DFS (HR = 4.72, 95 % CI: 2.29–9.75, P < 0.01) (Fig. 3D). In contrast, symptom duration was not associated with poor DFS (HR = 1.26, 95 % CI: 0.56–2.81, P = 0.57) (Fig. 3E). A similar analysis on OS was not performed, as only five events were recorded among the 211 patients (2.36 %), rendering meaningful statistical analysis infeasible.

Discussion

In this large cohort study, we developed the CuPeR model, a comprehensive predictive tool for PoRP-CD, by analyzing diverse patient and tumor characteristics, imaging findings, and treatment details. This model identified four key predictors—symptom duration, MRI Hardy’s grade, tumor site, and previous pituitary surgery. Multivariate analysis revealed that longer symptom duration and a history of prior surgery significantly increased recurrence risk, while bilateral tumor location was associated with a reduced risk. Validated with an AUC of 0.70 and 83 % accuracy on the testing dataset, the model offers significant clinical utility by providing treating surgeons with valuable insights into postoperative outcomes.

This study is among the few to develop a predictive model for estimating PoRP-CD (Table 3). Previous efforts, such as those by Liu et al. [6] and Fan et al. [7], employed machine learning and deep learning methodologies, respectively, demonstrating promising results (AUCs of 0.78 and 0.86). However, both studies were limited in their applicability to many clinical settings, as they focused solely on patients undergoing initial surgeries and incorporated postoperative parameters, which are unavailable for preoperative decision-making. By addressing these gaps, our study contributes a more practical tool for use in diverse clinical scenarios. Moreover, the findings of this study align with predictors identified in prior research. For instance, factors such duration of symptoms and history of previous pituitary surgery have been highlighted as critical for recurrence [6,14]. Importantly, our inclusion of MRI-based predictors and preoperative variables ensures the model’s relevance during preoperative planning, distinguishing it from previous approaches.

Table 3. Studies on predictive models or patients and tumors predictive factors of post-operative remission of Cushing’s disease.

Empty Cell	Year	Country	Study Size	Methods	Main Findings	Ref.
Predictive Models
Comprising 8 factors: age, disease coarse, morning serum ACTH (preop), morning serum cortisol (preop), urine free cortisol (preop), morning serum ACTH nadir (postop), morning serum cortisol nadir (postop), urine free cortisol nadir (postop)	2019	China	354	Machine-learning using Random Forest algorithm	Sensitivity 87 %, specificity 58 % AUC 0.78	[6]
Comprising 5 factors: age, disease coarse, morning serum ACTH (postop), morning serum cortisol nadir (postop), urine free cortisol nadir (postop)	2021	China	354	Deep-learning using factorization‑machine based neural approach	AUC 0.86	[7]
Predictive Factors
Serum cortisol < 35 nmol/L (6–12 w after surgery)	1993	UK	11	Prospective	Favorable long-term remission rate	[15]
Serum 11-deoxycortisol > 150 nmol/L after metyrapone test at 14 days post-surgery	1997	Netherlands	29	Retrospective	Higher risk of recurrence Sensitivity 100 %, specificity 75 %	[16]
Serum cortisol < 2 μ/dL (3–8 d after surgery)	2001	Japan	49	Retrospective	Recurrent disease in 4 % of patients	[17]
MRI-based tumor size and cavernous sinus invasion	2003	Italy	26	Retrospective	Unfavorable factors of persistent disease	[18]
No histological evidence of adenoma	2007	US	490	Retrospective	Lower remission rate	[19]
Long-term hypocortisolism after surgery (≥13 m)	2017	India	230	Retrospective	Favorable for remission Sensitivity 46 %, specificity 100 %	[20]
Greater decrease in BMI after surgery Lower DHEAS before surgery	2017	Taiwan	41	Retrospective	Favorable factors for higher remission	[21]
High serum ACTH/cortisol ratio before surgery	2018	Turkey	119	Retrospective	Risk factor for disease recurrence	[22]
USP8 mutation	2018	Germany	48	Retrospective	Higher recurrence rate	[23]
Serum cortisol < 107 nmol/L after betamethasone suppression test following surgery	2018	Sweden	28	Interventional	Sensitivity 85 %, specificity 94 % AUC 0.92	[24]
Tumor visualization on MRI before surgery	2022	Spain	40	Retrospective	Favorable factor for remission	[25]

Abbreviations: ACTH − Adrenocorticotropic Hormone; AUC − Area Under the Curve; BMI − Body Mass Index; DHEAS − Dehydroepiandrosterone Sulfate; MRI − Magnetic Resonance Imaging; PoRP-CD − Persistent or Recurrent Cushing’s Disease; Preop − Preoperative; Postop − Postoperative; USP8 − Ubiquitin Specific Peptidase 8.

Several other studies aimed to explore the predictive value of single predictors. Braun et al. (2020) summarized the predictors for CD remission following TSS in a systematic review. Key predictors include pre-surgical identification of the tumor via MRI and the absence of adenoma invasion into the cavernous sinus. Postoperative hormonal levels, particularly low cortisol (< 2 µg/dL) and ACTH levels (< 3.3 pmol/L) as well as low cortisol levels (< 35 nmol/L) at 6–12 weeks post-surgery and sustained hypocortisolism requiring long-term replacement therapy, were significant indicators of remission. Additionally, post-surgical decreases in BMI contributed to favorable outcomes. Other reported predictors included a high level of surgical expertise, younger patient age, non-mutant USP8 corticotroph tumors, and swift recovery from postoperative adrenal insufficiency [5].

This study has certain limitations that should be acknowledged. The reliance on retrospective data may result in potential biases in variable selection and data completeness. While the model demonstrated good predictive accuracy, its limited sensitivity may restrict its ability to identify all high-risk patients. Moreover, the model has not been externally validated in independent cohorts, which limits its generalizability to other clinical settings. Despite these limitations, the study possesses significant strengths that underscore its contribution to the field. Applying one of the largest CD cohorts, it provides a robust statistical foundation and enhances the reliability of the findings. The comprehensive inclusion of diverse patient and tumor characteristics, imaging findings, and treatment details resulted in a clinically relevant and well-rounded predictive model. Notably, this model stands out for its applicability to a broader spectrum of patients, including those with prior surgeries or radiotherapy, addressing a gap left by earlier studies. Furthermore, the development of an online dynamic nomogram bridges the gap between research and clinical practice, allowing personalized predictions and aiding surgeons in making informed decisions before pituitary surgery.

Although this study incorporated long-term follow-up (median 58 months) to define persistence and recurrence and to internally validate the model, external validation in prospective, multi-institutional cohorts remains essential to confirm its broader applicability. Although the CuPeR model incorporates a wide array of clinical, radiological, biochemical, and demographic variables, other potential prognostic factors were not included and may warrant consideration in future studies. For instance, the presence of osteoporosis, degree of tumor invasion, and early recovery of the adrenal axis during the postoperative period have all been reported as relevant predictors of outcomes in Cushing’s disease [26]. Moreover, the role of surgical expertise is critical, as higher surgeon and institutional experience are strongly associated with improved remission and lower recurrence rates [27]. Incorporating novel parameters, such as genetic markers or advanced imaging techniques, could further enhance the predictive accuracy and clinical utility of the model. Prospective implementation of the nomogram in routine clinical workflows will provide valuable insights into its performance and its potential to improve patient outcomes.

Conclusions

This study introduced a practical, predictive model for estimating the risk of postoperative persistence and recurrence in Cushing’s disease, possibly offering a reliable tool for preoperative planning. By integrating key clinical predictors into an interactive online dynamic nomogram, the CuPeR model may provide surgeons with personalized risk assessments to aid in preoperative planning. Its focus on preoperative data ensures broader applicability, paving the way for tailored therapeutic strategies and improved patient outcomes in diverse clinical scenarios.

Funding details

None.

CRediT authorship contribution statement

Guive Sharifi: Supervision, Conceptualization. Elham Paraandavaji: Investigation, Data curation. Nader Akbari Dilmaghani: Investigation, Data curation. Tohid Emami Meybodi: Investigation, Data curation. Ibrahim Mohammadzadeh: Investigation, Data curation. Neginalsadat Sadeghi: Investigation, Data curation. Amirali Vaghari: Visualization. Behnaz Niroomand: Visualization. Seyed Mohammad Tavangar: Resources. Mohammad reza Mohajeri Tehrani: Validation. Zahra Davoudi: Resources. Marjan Mirsalehi: Writing – review & editing. Seyed Ali Mousavinejad: Validation, Resources. Farzad Taghizadeh-Hesary: Writing – review & editing, Writing – original draft.

Informed consent

Not applicable.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

None.

The data that support the findings of this study are available on request from the corresponding author.

References

[1]

G. Giuffrida, S. Crisafulli, F. Ferraù, et al.

Global Cushing’s disease epidemiology: a systematic review and meta-analysis of observational studies

J Endocrinol Invest, 45 (6) (2022), pp. 1235-1246, 10.1007/s40618-022-01754-1

View in Scopus Google Scholar
[2]

F. Roelfsema, N.R. Biermasz, A.M. Pereira

Clinical factors involved in the recurrence of pituitary adenomas after surgical remission: a structured review and meta-analysis

Pituitary, 15 (1) (2012), pp. 71-83, 10.1007/s11102-011-0347-7

View in Scopus Google Scholar
[3]

L.H.A. Broersen, N.R. Biermasz, W.R. Van Furth, et al.

Endoscopic vs. microscopic transsphenoidal surgery for Cushing’s disease: a systematic review and meta-analysis

Pituitary, 21 (5) (2018), pp. 524-534, 10.1007/s11102-018-0893-3

View in Scopus Google Scholar
[4]

S. Petersenn, A. Beckers, D. Ferone, et al.

Therapy of endocrine disease: outcomes in patients with Cushing’s disease undergoing transsphenoidal surgery: systematic review assessing criteria used to define remission and recurrence

Eur J Endocrinol, 172 (6) (2015), pp. R227-R239, 10.1530/EJE-14-0883

View in Scopus Google Scholar
[5]

L.T. Braun, G. Rubinstein, S. Zopp, et al.

Recurrence after pituitary surgery in adult Cushing’s disease: a systematic review on diagnosis and treatment

Endocrine, 70 (2) (2020), pp. 218-231, 10.1007/s12020-020-02432-z

View in Scopus Google Scholar
[6]

Y. Liu, X. Liu, X. Hong, et al.

Prediction of recurrence after transsphenoidal surgery for Cushing’s disease: the use of machine learning algorithms

Neuroendocrinology, 108 (3) (2019), pp. 201-210, 10.1159/000496753

Google Scholar
[7]

Y. Fan, D. Li, Y. Liu, M. Feng, Q. Chen, R. Wang

Toward better prediction of recurrence for Cushing’s disease: a factorization-machine based neural approach

Int J Mach Learn Cybern, 12 (3) (2021), pp. 625-633, 10.1007/s13042-020-01192-6

View in Scopus Google Scholar
[8]

L.K. Nieman, B.M.K. Biller, J.W. Findling, et al.

The diagnosis of Cushing’s syndrome: an endocrine society clinical practice guideline

J Clin Endocrinol Metab, 93 (5) (2008), pp. 1526-1540, 10.1210/jc.2008-0125

View in Scopus Google Scholar
[9]

H.L. Dichek, L.K. Nieman, E.H. Oldfield, H.I. Pass, J.D. Malley, G.B. Cutler

A comparison of the standard high dose dexamethasone suppression test and the overnight 8-mg dexamethasone suppression test for the differential diagnosis of adrenocorticotropin-dependent Cushing’s syndrome

J Clin Endocrinol Metab, 78 (2) (1994), pp. 418-422, 10.1210/jcem.78.2.8106630

View in Scopus Google Scholar
[10]

M. Araujo-Castro, A. Acitores Cancela, C. Vior, E. Pascual-Corrales, B.V. Rodríguez

Radiological Knosp, revised-Knosp, and Hardy–Wilson classifications for the prediction of surgical outcomes in the endoscopic endonasal surgery of pituitary adenomas: study of 228 cases

Front Oncol, 11 (2022), Article 807040, 10.3389/fonc.2021.807040

View in Scopus Google Scholar
[11]

A.M. Landolt, A. Valavanis, J. Girard, A.N. Eberle

Corticotrophin‐releasing factor‐test used with bilateral, simultaneous inferior petrosal sinus blood‐sampling for the diagnosis of pituitary‐dependent Cushing’s disease

Clin Endocrinol, 25 (6) (1986), pp. 687-696, 10.1111/j.1365-2265.1986.tb03624.x

View in Scopus Google Scholar
[12]

P. Castelnuovo, A. Pistochini, D. Locatelli

Different surgical approaches to the sellar region: focusing on the “two nostrils four hands technique

Rhinology, 44 (1) (2006), pp. 2-7

View in Scopus Google Scholar
[13]

F. Esposito, J.R. Dusick, P. Cohan, et al.

Early morning cortisol levels as a predictor of remission after transsphenoidal surgery for Cushing’s disease

J Clin Endocrinol Metab, 91 (1) (2006), pp. 7-13, 10.1210/jc.2005-1204

View in Scopus Google Scholar
[14]

A. Stroud, P. Dhaliwal, R. Alvarado, et al.

Outcomes of pituitary surgery for Cushing’s disease: a systematic review and meta-analysis

Pituitary, 23 (5) (2020), pp. 595-609, 10.1007/s11102-020-01066-8

View in Scopus Google Scholar
[15]

G.C. Toms, M.I. McCarthy, M.J. Niven, C.H. Orteu, T.T. King, J.P. Monson

Predicting relapse after transsphenoidal surgery for Cushing’s disease

J Clin Endocrinol Metab, 76 (2) (1993), pp. 291-294, 10.1210/jcem.76.2.8432771

View in Scopus Google Scholar
[16]

M.O. Van Aken, W.W. De Herder, A. Van Der Lely, F.H. De Jong, S.W.J. Lamberts

Postoperative metyrapone test in the early assessment of outcome of pituitary surgery for Cushing’s disease

Clin Endocrinol, 47 (2) (1997), pp. 145-149, 10.1046/j.1365-2265.1997.2541051.x

View in Scopus Google Scholar
[17]

T. Imaki, T. Tsushima, N. Hizuka, et al.

Postoperative plasma cortisol levels predict long-term outcome in patients with Cushing’s disease and determine which patients should be treated with pituitary irradiation after surgery

Endocr J, 48 (1) (2001), pp. 53-62, 10.1507/endocrj.48.53

View in Scopus Google Scholar
[18]

S. Cannavo, B. Almoto, C. Dall’Asta, et al.

Long-term results of treatment in patients with ACTH-secreting pituitary macroadenomas

Eur J Endocrinol Published online September 1 (2003), p. 195, 10.1530/eje.0.1490195

View in Scopus Google Scholar
[19]

N. Pouratian, D.M. Prevedello, J. Jagannathan, M.B. Lopes, M.L. Vance, E.R. Laws

Outcomes and management of patients with Cushing’s disease without pathological confirmation of tumor resection after transsphenoidal surgery

J Clin Endocrinol Metab, 92 (9) (2007), pp. 3383-3388, 10.1210/jc.2007-0208

View in Scopus Google Scholar
[20]

P. Bansal, A. Lila, M. Goroshi, et al.

Duration of post-operative hypocortisolism predicts sustained remission after pituitary surgery for Cushing’s disease

Endocr Connect, 6 (8) (2017), pp. 625-636, 10.1530/EC-17-0175

View in Scopus Google Scholar
[21]

C.H. Kuo, S.R. Shih, H.Y. Li, et al.

Adrenocorticotropic hormone levels before treatment predict recurrence of Cushing’s disease

J Formos Med Assoc, 116 (6) (2017), pp. 441-447, 10.1016/j.jfma.2016.08.008

View PDF View article View in Scopus Google Scholar
[22]

A. Selek, B. Cetinarslan, Z. Canturk, et al.

The utility of preoperative ACTH/cortisol ratio for the diagnosis and prognosis of Cushing’s disease

J Neurosci Rural Pract, 09 (01) (2018), pp. 106-111, 10.4103/jnrp.jnrp_308_17

View in Scopus Google Scholar
[23]

A. Albani, L.G. Pérez-Rivas, C. Dimopoulou, et al.

The USP 8 mutational status may predict long‐term remission in patients with Cushing’s disease

Clin Endocrinol, 89 (4) (2018), pp. 454-458, 10.1111/cen.13802

View in Scopus Google Scholar
[24]

E. Uvelius, P. Höglund, S. Valdemarsson, P. Siesjö

An early post-operative ACTH suppression test can safely predict short- and long-term remission after surgery of Cushing’s disease

Pituitary, 21 (5) (2018), pp. 490-498, 10.1007/s11102-018-0902-6

View in Scopus Google Scholar
[25]

M. Araujo-Castro, M. Marchán Pinedo, M. Fernández-Argüeso, et al.

Presurgical predictive factors of surgical remission in Cushing’s disease. Study of 32 cases

Endocrinol Diabetes Nutr, 69 (8) (2022), pp. 584-590, 10.1016/j.endinu.2021.07.004

View PDF View article View in Scopus Google Scholar
[26]

M. Ünal, A. Selek, M. Sözen, et al.

Recurrent Cushing’s disease in adults: predictors and long-term follow-up

Horm Metab Res, 55 (08) (2023), pp. 520-527, 10.1055/a-2047-6017

Finding PDF…View in Scopus Google Scholar
[27]
Belkacem S. Prognostic factors for remission in Cushings disease after pituitary surgery bout 100 cases. Endocr Abstr. Published online May 15, 2021. doi:10.1530/endoabs.73.AEP525.

Google Scholar

https://www.sciencedirect.com/science/article/pii/S2214623725000353

Filed under: Cushing's, pituitary, symptoms, Treatments | Tagged: Cushing's Disease, pituitary, post-op, The CuPeR Model | Leave a comment »

The Outcome of Abnormal Glucose Metabolism and Its Clinical Features in Patients With Cushing’s Disease After Curative Surgery

Posted on July 27, 2025 by MaryO

Abstract

Objective

To investigate the outcomes of abnormal glucose metabolism and its clinical characteristics in patients with Cushing’s disease (CD) who achieved biochemical remission after surgery.

Methods

Patients diagnosed with CD who achieved biochemical remission and underwent regular follow-up after surgery were enrolled. Pre- and postoperative clinical data were collected and analyzed.

Result

151CD patients were included, of whom 80 (53 %) had preoperative abnormal glucose metabolism, including 56 with diabetes mellitus (DM) and 24 with impaired glucose regulation (IGR). At one year after surgery, 57 patients exhibited improved glucose metabolism, accompanied by a significant reduction in the homeostasis model assessment of insulin resistance (HOMA-IR). Improvements were mainly observed at 3 and 6 months after surgery. At one-year after surgery, there were 20 patients with diabetes and 16 with IGR. Compared to those with NGT, these individuals exhibited a higher prevalence of hypertension, hyperlipidemia, fatty liver, and abnormal bone metabolism.

Conclusion

CD patients demonstrated a high incidence of abnormal glucose metabolism. Notably, approximately two-thirds demonstrated improved glucose metabolism one year after curative surgery, with the greatest improvements observed at 3- to 6-month postoperative follow-up.

Introduction

Cushing’s disease (CD) is characterized by excessive endogenous cortisol production caused by pituitary adrenocorticotropic hormone adenoma and is the main cause of Cushing’s syndrome (CS). Surgical resection of the tumor is the preferred treatment. Prolonged exposure to hypercortisolism increases the risk of metabolic abnormalities, including obesity, hypertension, glucose and lipid abnormalities, osteoporosis, etc. Additionally, it significantly elevates the risk of infection, thrombosis, and hypokalemia. Abnormal glucose metabolism is a common complication of CS, with an incidence ranging from 13.1 % to 47 %[1], and diabetes is an independent risk factor for mortality in CD patients[2].

Previous clinical studies have found that metabolic abnormalities such as diabetes, hypertension, and hyperlipidemia improve in CS patients who achieve biochemical remission after surgical treatment. However, the concept of improvement in glucose metabolism, the incidence of improvement, and its related factors are inconsistent in various reports. Previous studies primarily assessed the outcome of glucose metabolism based on plasma glucose results at a single fixed follow-up time after surgery. The lack of regular follow-up data makes it difficult to clearly understand the trend of postoperative plasma glucose changes, and there are no clinical data on when glucose metabolism begins to improve or change. Therefore, this study retrospectively analyzed the follow-up data of patients with Cushing’s disease in our hospital before and after surgery, and monitored the changes in glucose metabolism, to explore the characteristics and clinical features of such changes in patients with Cushing’s disease who achieved remission from CD following surgery..

Access through your organization

Check access to the full text by signing in through your organization.

Access through your organization

Section snippets

Subjects

This study enrolled hospitalized patients with Cushing’s disease at Huashan Hospital, Fudan University from January 2014 to February 2020. Inclusion criteria were as follows: (1) Age ≥ 18 years; (2) diagnosis of Cushing’s disease according to the 2021 Consensus on the Diagnosis and Management of Cushing’s Disease, confirmed by pathology[3]; (3) biochemical remission after transsphenoidal surgery; (4) complete preoperative data and regular follow-up visits (including visits at 1, 3, 6, and

Patients’ baseline characteristics

A total of 168 patients with CD were admitted to Huashan Hospital from 2014 to 2020 with pathological diagnosis and regular postoperative follow-up; however, 17 patients were excluded due to no biochemical remission after surgery or relapse during follow-up (Fig. 1). Ultimately, 151 patients (32 males and 119 females) were included in this study. The baseline characteristics of the included patients were shown in Table 1. There were 80 cases (53 %) complicated with abnormal glucose metabolism

Discussion

CD was a rare disease often associated with abnormal glucose metabolism. Based on medical history and OGTT screening, we found that over half (53 %) of CD patients exhibited abnormal glucose metabolism before surgery, with 37.1 % being diagnosed with diabetes. Previous studies have shown that the prevalence of diabetes in CS patients ranged from 13.1 % to 47 %, and most reports falling between 35 % and 45 %, which is consistent with our findings [1,12,13]. However, it should be noted that CD

Author contributions

Q.C. analyzed the data and wrote the manuscript. Q.C., Y.L., X.L., Q.S., W.S., and H.Z. collected the data. Y.L., Z.Z., M.H., S.Z., and H.Y. recruited patients. J.Z., Y.S., and S.Z. conducted the study design and revised the manuscript. All authors read and approved the final manuscript.

CRediT authorship contribution statement

Qiaoli Cui: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Yujia Li: Writing – original draft, Investigation, Formal analysis, Data curation. Xiaoyu Liu: Investigation, Formal analysis, Data curation. Quanya Sun: Investigation, Data curation. Wanwan Sun: Investigation, Formal analysis, Data curation. Min He: Project administration, Investigation. Jie Zhang: Writing – review & editing, Supervision, Funding

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We are indebted to the patients who participated in this study and all the doctors who contributed to the diagnosis and treatment of these patients. This work was supported by grants from the Multidisciplinary Diagnosis and Treatment (MDT) demonstration project in research hospitals (Shanghai Medical College, Fudan University, NO: DGF501069/017), National Science and Technology Major Project (NO: 2023ZD0506800，2023ZD0506802), 2023 Ningbo International Cooperation Program (NO: 2023H024).

References (16)

R.N. Clayton et al.

Mortality in patients with Cushing’s disease more than 10 years after remission: a multicentre, multinational, retrospective cohort study

Lancet Diabetes Endocrinol

(2016)
M. Fleseriu et al.

Consensus on diagnosis and management of Cushing’s disease: a guideline update

Lancet Diabetes Endocrinol

(2021)
C. Scaroni et al.

Glucose Metabolism Abnormalities in Cushing Syndrome: from Molecular Basis to Clinical Management

Endocr Rev

(2017)
C. Dai et al.

Surgical outcome of transsphenoidal surgery in Cushing’s disease: a case series of 1106 patients from a single center over 30 years

Endocrine

(2022)
E. Valassi et al.

Delayed remission after transsphenoidal surgery in patients with Cushing’s disease

J Clin Endocrinol Metab

(2010)
X. He et al.

Glucocorticoid Withdrawal Syndrome following treatment of endogenous Cushing Syndrome

(2022)
E.V. Varlamov et al.

Perioperative Management of a Patient with Cushing Disease

J Endocr Soc

(2022)
Q. Cui et al.

The recovery time of hypothalamic-pituitary-adrenal axis after curative surgery in Cushing’s disease and its related factor

Endocrine

(2023)

There are more references available in the full text version of this article.

https://www.sciencedirect.com/science/article/abs/pii/S0168822725003961

Filed under: Cushing's, pituitary, Treatments | Tagged: diabetes mellitus, glucose, pituitary, post-op, remission, surgery | Leave a comment »

Postoperative Initiation of Thromboprophylaxis in Patients with Cushing’s Disease (PIT-CD):

Posted on June 28, 2025 by MaryO

Abstract

Background

Pituitary surgical intervention remains the preferred treatment for Cushing’s disease (CD) while postoperative venous thromboembolism (VTE) is a significant risk. Whether to prescribe pharmacological thromboprophylaxis presents a clinical dilemma, balancing the benefit of reducing VTE risk with the potential for increasing hemorrhagic events in these patients. Currently, strong evidence and established protocols for routine pharmacological thromboprophylaxis in this population are lacking. Therefore, a randomized, controlled trial is warranted to determine the efficacy and safety of combined pharmacological and mechanical thromboprophylaxis in reducing postoperative VTE risk in patients with CD.

Methods

This investigator-initiated, multi-center, prospective, randomized, open-label trial with blinded outcome assessment aims to evaluate the efficacy and safety of combined pharmacological and mechanical thromboprophylaxis compared to mechanical thromboprophylaxis alone in postoperative patients with CD. A total of 206 patients diagnosed with CD who will be undergoing transsphenoidal surgery will be randomized in a 1:1 ratio to receive either combined pharmacological and mechanical thromboprophylaxis (intervention) or mechanical thromboprophylaxis only (control). The primary outcome is the risk of VTE within 12 weeks following surgery.

Discussion

This trial represents a significant milestone in evaluating the efficacy of combined pharmacological and mechanical prophylaxis in reducing VTE events in postoperative CD patients.

Trial registration

ClinicalTrials.gov Identifier: NCT04486859, first registered on 22 July 2020.

Peer Review reports

Administrative information

Note: the numbers in curly brackets in this protocol refer to SPIRIT checklist item numbers. The order of the items has been modified to group similar items (see http://www.equator-network.org/reporting-guidelines/spirit-2013-statement-defining-standard-protocol-items-for-clinical-trials/).

Title {1}	Postoperative Initiation of Thromboprophylaxis in patients with Cushing’s Disease (PIT-CD): a randomized control trial
Trial registration {2a and 2b}	ClinicalTrials.gov Identifier: NCT04486859, first registered on 22 July 2020 WHO Trial Registration Data Set (Supplement)
Protocol version {3}	Date: 1 July 2021, Version 5.0
Funding {4}	The trial is supported by Clinical Research Plan of SHDC (SHDC2020CR2004A).
Author details {5a}	Nidan Qiao, Min He, Zhao Ye, Wei Gong, Zengyi Ma, Yifei Yu, Zhenyu Wu, Lin Lu, Huijuan Zhu, Yong Yao, Zhihong Liao, Haijun Wang, Huiwen Tan, Bowen Cai, Yerong Yu, Ting Lei, Yan Yang, Changzhen Jiang, Xiaofang Yan, Yanying Guo, Yuan Chen, Hongying Ye, Yongfei Wang, Nicholas A. Tritos, Zhaoyun Zhang, Yao Zhao.
Name and contact information for the trial sponsor {5b}	Investigator initiated trial, principal investigators, post-production correspondence: Yao Zhao (YZ), Department of Neurosurgery, Huashan Hospital, Fudan University, 12 mid Wulumuqi Rd, Shanghai 200040, China. Email: zhaoyao@huashan.org.cn Zhaoyun Zhang (ZZ), Department of Endocrinology, Huashan Hospital, Fudan University, 12 mid Wulumuqi Rd, Shanghai 200040, China. Email: zhangzhaoyun@fudan.edu.cn
Role of sponsor {5c}	The trial sponsor holds responsibility for all key elements of the trial’s execution, including its design, data collection, management, analysis, interpretation of results, and reporting. An independent Data Safety Monitoring Board monitors data safety and participant protection to ensure the trial’s integrity and the safety of participants.

Introduction

Background and rationale {6a}

Cushing’s disease (CD) is characterized by hypercortisolism resulting from an adrenocorticotropic hormone-secreting pituitary adenoma [1]. Tumor-directed surgical intervention remains the preferred treatment for this condition. Patients with Cushing’s disease commonly experience a hypercoagulable state due to activation of the coagulation system [2], suppression of anticoagulation and fibrinolytic pathways, and enhanced platelet activation, significantly increasing their risk of venous thromboembolism (VTE). Postoperative VTE risk is further exacerbated by factors such as intravenous medications, blood loss, and prolonged bed rest. Multiple studies report postoperative VTE risks in patients with CD ranging from 3 to 20% [2,3,4,5].

The Endocrine Society and Pituitary Society recommends considering perioperative thromboprophylaxis as a strategy to reduce VTE risk in patients with CD [1, 6]. However, this recommendation was based on a single study that investigated perioperative prophylactic anticoagulation in patients with Cushing’s syndrome [7]. The study was limited by its small sample size, single-center nature, and retrospective design. Crucial details such as the optimal timing for initiation, choice of anticoagulant, and duration of therapy were not established. Recent surveys of European and US centers indicate that thromboprophylaxis protocols are not routinely employed, and there is considerable heterogeneity in prophylactic practices across centers [8, 9].

The primary risk associated with thromboprophylaxis is postoperative hemorrhage. In patients with CD, although the risk of bleeding is significantly lower than after a typical craniotomy, complications such as intrasellar hemorrhage and nasal bleeding may still occur. Due to its retrospective nature, the aforementioned study cannot conclusively determine whether the benefits of thromboprophylaxis outweigh its risks. Consequently, guidelines from hematology and neurosurgical societies have concluded that the current evidence is insufficient to support a standardized VTE prophylaxis regimen for neurosurgical patients [10,11,12]. Nevertheless, both the American Society of Hematology and European guidelines suggest that a combination of pharmacological and mechanical prophylaxis may be justified for higher-risk subgroups [10, 13].

Objectives {7}

Due to conflicting recommendations and lack of a definitive study to determine whether the benefits outweigh the risks regarding the use of pharmacological antithrombotic prophylaxis in patients with CD following pituitary surgery, we initiated this study, called Postoperative Initiation of Thromboprophylaxis in Patients with Cushing’s Disease (PIT-CD). The aim of this study is to evaluate whether the combined use of pharmacological and mechanical prophylaxis reduces VTE events compared to mechanical prophylaxis alone in postoperative CD patients.

Trial design {8}

Our hypothesis was that pharmacological prophylaxis in combination with intermittent pneumatic compression would be superior to intermittent pneumatic compression alone.

The PIT-CD study is an open-label, multicenter, prospective, randomized clinical trial with open-label treatment designed to assess the efficacy of combined pharmacological and mechanical prophylaxis compared to mechanical prophylaxis alone. Patients are randomized in a 1:1 ratio. The patient flow is illustrated in Fig. 1.

Methods: participants, interventions and outcomes

Study setting {9}

This study was initiated in tertiary centers across China with expertise in managing patients with CD. Currently, seven centers (see Supplements) are actively recruiting patients for the study.

Eligibility criteria {10}

Inclusion criteria

Patients are eligible for inclusion if they meet the following criteria:

1.Age between 18 and 65 years (inclusive)
2.Diagnosed with CD and scheduled to undergo transsphenoidal surgery
3.Either newly diagnosed or recurrent disease

A diagnosis of CD is confirmed based on the following criteria:

A.Twenty-four-hour urine free cortisol > upper normal boundary and low-dose dexamethasone suppression test (overnight or over two days): serum cortisol > 1.8 µg/dL
B.8 AM serum adrenocorticotropic hormone > 20 pg/mL
C.High-dose dexamethasone suppression test: serum cortisol or 24-h urine cortisol suppression > 50%
D.Inferior petrosal sinus sampling (IPSS) indicates elevated adrenocorticotropic hormone central gradient consistent with secretion from a central source

Patients are diagnosed with CD if both criteria A and B are met, in addition to either C or D. In patients with tumors smaller than 6 mm on MRI, IPSS indicating a central source is essential.

Exclusion criteria

Patients will be excluded from the study if they meet any of the following criteria:

1.History of VTE before surgery or within 24 h post-surgery
2.Acute bacterial endocarditis
3.Major bleeding events within the previous 6 months
4.Thrombocytopenia
5.Active gastrointestinal ulcers
6.History of stroke
7.High risk of bleeding due to clotting abnormalities
8.Participation in other clinical trials within the last three months
9.Contraindications to rivaroxaban (e.g., renal dysfunction with eGFR < 50 mL/min)
10.Presence of malignant diseases
11.Severe mental or neurological disorders
12.Presence of intracranial vascular abnormalities
13.Contraindications to mechanical prophylactic anticoagulation
14.Pregnancy
15.Any other condition that researchers deem inappropriate for study participation (e.g., oral contraceptive use, history of thrombophilia)

Who will obtain informed consent? {26a}

Patients with CD are provided with detailed information about the clinical trial, including known and foreseeable risks and potential adverse events. Investigators are required to thoroughly explain these details to the patients or their guardians if the patients lack capacity to provide consent. Following a comprehensive explanation and discussion, both the patients or their guardians and the investigators sign and date the informed consent form.

Additional consent provisions for collection and use of participant data and biological specimens {26b}

N/A. Biological specimens are unnecessary in this trial. Participant data was not intended to be included in any other ancillary studies.

Interventions

Explanation for the choice of comparators {6b}

Participants in the control arm of the study will be required to use a limb compression system twice daily, for 30 min each session, from the 2nd to the 7th day post-surgery. The intermittent pneumatic compression devices are the standard of care in the prevention deep vein thrombosis in many literatures [14, 15].

Intervention description {11a}

Participants in the intervention arm of the study will be required to use the same limb compression system, also for 30 min twice daily from the 2nd to the 7th day post-surgery. Additionally, participants will receive subcutaneous injections of low molecular weight heparin (4000 IU) once daily from the 2nd to the 4th day post-surgery. Starting on the 5th day and continuing through the 28th day post-surgery, participants will take oral rivaroxaban tablets (10 mg) once daily.

Criteria for discontinuing or modifying allocated interventions {11b}

Participants have the right to withdraw their consent at any time without providing a reason, thereby terminating their participation in the study. Any withdrawal and the reasons, if known, will be documented. Criteria for premature termination include the following: occurrence of the primary outcome (patients will still be monitored for safety for 12 weeks), failure to meet inclusion criteria, fulfillment of exclusion criteria, or loss of contact.

Strategies to improve adherence to interventions {11c}

Several strategies will be employed to maintain adherence to interventions in this trial. Participants will receive thorough preoperative education on the importance of pharmacological and mechanical prophylaxis in preventing VTE if they are assigned to the intervention arm or the importance of mechanical prophylaxis if they are assigned to the control arm. Detailed instructions on the use of the limb compression system and administration of rivaroxaban will be provided. Pill counts will be performed to document adherence in the intervention group.

Relevant concomitant care permitted or prohibited during the trial {11d}

N/A. Participants in both groups will receive treatment according to the current standard-of-care.

Provisions for post-trial care {30}

Participants experiencing adverse events will be followed until the events are resolved. Other participants will be regularly followed in accordance with clinical routine clinical practice. Participants in the trial are compensated in the event of trial-associated harms.

Outcomes {12}

Primary outcome

The primary outcome of the study is the risk of venous thromboembolism (VTE) within 12 weeks after surgery. VTE is defined as either deep vein thrombosis (DVT) or pulmonary embolism (PE), regardless of whether the cases are symptomatic or asymptomatic.

Secondary outcomes

The secondary outcomes are as follows: (1) risk of DVT within 12 weeks after surgery; (2) risk of PE within 12 weeks after surgery; (3) risk of symptomatic DVT, symptomatic PE, or symptomatic VTE within 12 weeks after surgery; (4) risk of VTE-associated mortality within 12 weeks after surgery; (5) risk of all-cause mortality within 12 weeks after surgery.

“Symptomatic” is defined as the presence of one or more of the following symptoms attributed to VTE: pain or swelling in the affected leg; chest pain, dyspnea, or decreased oxygen saturation.

Safety outcomes

Safety outcomes include the following: (1) major bleeding; (2) minor bleeding; (3) hemorrhage-associated surgery; (4) hemorrhage-associated readmission; (5) coagulation disorders (APTT or INR > 2.5 normal upper boundary); (6) thrombocytopenia; (7) increase in liver function tests.

Major bleeding is defined according to the Subcommittee on Control of Anticoagulation of the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis [16]. This includes fatal bleeding; bleeding that is symptomatic and occurs in a critical area or organ; extrasurgical site bleeding causing a fall in hemoglobin level of 20 g/L or more, or leading to transfusion of two or more units of whole blood or red cells; surgical site bleeding that requires a second intervention.

Participant timeline {13}

A schema of all trial procedures and clinical visits is summarized in Table 1.

Table 1 Schedule of enrolment, interventions and assessments

From https://trialsjournal.biomedcentral.com/articles/10.1186/s13063-025-08923-6

Sample size {14}

Our estimates are based on a retrospective study examining the effects of preventive anticoagulation during the perioperative period in Cushing syndrome [7]. This study reported that the risk of postoperative VTE was lower in patients receiving preventive anticoagulants (6%) compared to those who did not (20%). Therefore, we assume that the risk of the primary outcome in the control group is 20%, while in the intervention group it is 5% within 12 weeks. Based on these assumptions, we calculated the required sample size for each group to be is 93 using PASS software, with an alpha level of 0.05 and a power of 0.9. Accounting for an estimated 10% dropout rate, the total number of patients required is 206.

Recruitment {15}

Clinical investigators will receive training on communicating with potential patients and their relatives, documenting screening logs, and other standard operating procedures during the kick-off meeting at each participating center. All centers will recruit patients competitively, and recruitment progress will be monitored to track the process. The estimated recruitment rate is 8 to 10 patients per month, with an expected recruitment period of 2 years.

Assignment of interventions: allocation

Sequence generation {16a}

The randomization procedure is computer- and web-based, and is stratified by age (≤ 35 years old vs. > 35 years old), sex (female vs. male) and disease duration (≤ 2 years vs. > 2 years).

Concealment mechanism {16b}

Participants are randomized using a web-based randomization system (edc.fudan.edu.cn). This system maintains allocation concealment by withholding the randomization code until screening is complete.

Implementation {16c}

Investigators will enroll participants, with the stratified block algorithms generating a random allocation sequence. Participant assignment through the randomization system is not subject to influence by the clinical investigators.

Assignment of interventions: blinding

Who will be blinded {17a}

This is an open-label trial, meaning that both the treating physicians and the participants are aware of the treatment allocation. However, a separate group of clinical outcome assessors (Clinical Event Committee, CEC), who are blinded to the treatment allocation, will determine the clinical outcomes. Similarly, lower limbs ultrasound and pulmonary computed tomography angiography (CTA) assessments will be adjudicated by an Independent Review Committee (IRC) that is blinded to the treatment allocation. Statisticians remain blinded to treatment allocation prior to the final analysis, and the interim analyses will be conducted by a separate team from the one undertaking the final analysis.

Procedure for unblinding if needed {17b}

N/A. The design is open label.

Data collection and management

Plans for assessment and collection of outcomes {18a}

Deep vein thrombosis (DVT) will be assessed using bilateral lower limb ultrasound. Asymptomatic participants will undergo evaluation at prespecified intervals (day 4, day 7, week 4, and week 12 post-intervention), while symptomatic individuals will receive immediate imaging upon presentation of clinical manifestations such as unilateral or bilateral lower extremity edema or pain. Pulmonary embolism (PE) screening will be performed via pulmonary computed tomography angiography (CTA) at day 7 in asymptomatic cases, with expedited assessment triggered by acute symptoms (e.g., chest pain, dyspnea) or radiographic evidence of DVT detected during lower limb ultrasonography. These events will be adjudicated by an Independent Review Committee (IRC). A CEC will be convened to assess other outcomes.

Plans to promote participant retention and complete follow-up {18b}

The initial intervention for participants takes place during the patient’s inpatient stay, during which researchers will provide detailed information about the required procedures. Participants will undergo routine follow-up at 4 weeks and 12 weeks post-surgery, with VTE-related follow-up arranged during these routine visits. Transportation and examination expenses for follow-up visits are reimbursable.

Data of those who discontinue will also be documented.

Data management {19}

Data will be kept, both on paper and in electronic databases, for at least 5 years. Data will be entered by clinical investigators using electronic case report forms (eCRFs) on a web-based platform (http://crip-ec.shdc.org.cn). The investigators will be introduced to the platform and trained in data entry during the initial kick-off meeting before the recruitment of the first study participant. Access to the study database will be restricted to authorized clinical investigators, who will use a personal ID and password to gain entry.

Confidentiality {27}

When adding a new participant to the database, identifying data (e.g., Chinese name) are entered on a form that is printed but not saved on the server. On this form, participants will be represented by a unique ID. The printed form is kept in a locked space accessible only to the principal investigator and may be used to unblind personal data if necessary.

Plans for collection, laboratory evaluation and storage of biological specimens for genetic or molecular analysis in this trial/future use {33}

N/A. There will be no biological specimens collected.

Statistical methods

Statistical methods for primary and secondary outcomes {20a}

The primary analysis will be conducted on the full analysis data set, adhering to the intention-to-treat principle, which includes all patients randomized in the study. Generalized linear models (GLMs) with binomial distribution will be employed to analyze primary, secondary, and safety outcomes. Treatment effects for these outcomes will be quantified as risk differences (RDs) with corresponding 95% confidence intervals (CIs). Additionally, odds ratios with 95% confidence intervals will be calculated using a logistic regression model, and hazard ratios with 95% confidence intervals will be calculated using a Cox Proportional model.

Safety analyses will be based on all randomized patients who have received the study treatment. The risk and percentages of adverse events (AEs) and serious adverse events (SAEs) will be summarized by treatment group. Instances of subject death will be summarized and listed. All analyses will be performed using the SAS system, version 9.4.

Interim analyses {21b}

The Data Safety Monitoring Board (DSMB) plans to convene the interim analysis meeting after randomization and 12-week follow-up visits are completed for 103 participants. The significance level for interim analysis (primary outcome) is set at 0.001 according to the Haybittle–Peto boundary principle.

Based on these analyses, the DSMB will advise the steering committee on whether the randomized comparisons in this study have demonstrated a clear benefit of the intervention. If the p-values from the interim analysis for both groups are less than 0.001, recruitment will be halted, and the study will meet the criteria for early termination. If the p-values are greater than or equal to 0.001, recruitment will continue until the planned sample size is achieved, with the final analysis significance level set at 0.049.

Methods for additional analyses (e.g., subgroup analyses) {20b}

For both primary and secondary outcomes, pre-specified subgroup analyses will be conducted based on sex, age, disease duration, and magnitude of urine free cortisol elevation.

Methods in analysis to handle protocol non-adherence and any statistical methods to handle missing data {20c}

The primary analysis will be conducted on the intention-to-treat data set, which includes all randomized patients and is based on the treatment arm to which they were assigned, regardless of the therapy they actually received. A per-protocol analysis will also be performed to account for non-adherence. If appropriate, multiple imputation will be used to address any missing data in the dataset. The prespecified statistical analysis plan (SAP), developed by independent biostatisticians blinded to treatment allocation, will be prospectively registered on ClinicalTrials.gov prior to database lock.

Plans to give access to the full protocol, participant-level data and statistical code {31c}

The trial was prospectively registered in ClinicalTrials.gov with the Identifier NCT04486859. Updates to reflect significant protocol amendments will be submitted. The statistical analysis protocol will also be updated prior to database locking. The datasets and statistical code are available from the corresponding author upon reasonable request.

Oversight and monitoring

Composition of the coordinating centre and trial steering committee {5d}

The trial steering committee is composed of four Chinese experts and two international experts from outside of China. Investigators in participating centers are required to attend a training course during a kick-off event organized by the principal investigator. Each investigator must confirm that they have been properly introduced to trial-specific procedures. An IRC will adjudicate primary outcomes. An independent CEC will be responsible for ensuring high-quality outcomes and minimizing inconsistencies or bias in the clinical trial data.

Composition of the data monitoring committee, its role and reporting structure {21a}

The Data Safety Monitoring Board (DSMB) consists of three members, including one statistician. The DSMB will regularly receive blinded statistical reports and monitor serious adverse events throughout the trial to assess patient safety and determine if the trial should be terminated prematurely due to safety concerns.

An initial DSMB meeting will be conducted to ensure that DSMB members fully understand the research protocol, review and approve the DSMB charter, assess the monitoring plans for safety and efficacy data, and discuss the statistical methods, including stopping rules. A second DSMB meeting will be conducted to review the interim analysis. The interim analyses and the treatment allocation data will be provided by an independent trial statistician and provided confidentially to the DSMB chairman. An ad hoc DSMB meeting may be convened by either the principal investigators or the DSMB if imminent safety issues arise during the trial.

Adverse event reporting and harms {22}

Adverse events (AEs) and serious adverse events (SAEs) are defined according to the ICH GCP guidelines. All AEs and SAEs reported by study participants or observed by investigators within the study period must be documented in the eCRF and reported to the DSMB. Additionally, SAEs must be reported to the IRB.

Anticipated adverse events, including both major and minor bleeding events (e.g., epistaxis necessitating readmission), as well as coagulation disorders, thrombocytopenia, and elevated liver function tests, will be prospectively monitored in all trial participants. Unanticipated adverse events (not pre-specified in Section {12}) will be captured through spontaneous reporting. All adverse event data will be classified and graded according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0 to ensure consistency. For reporting, we will disclose all protocol-specified adverse events from Section {12}, alongside any unanticipated events higher than Grade 3.

Frequency and plans for auditing the trial conduct {23}

The trial conduct will be regularly audited to ensure compliance with the study protocol and Good Clinical Practice guidelines. Audits will be conducted by independent monitors from Shanghai Shenkang Hospital Developing Centers. These audits will involve reviewing study documentation, informed consent forms, source data verification, and adherence to the protocol. Audits will also assess data entry accuracy and the overall management of the trial. The frequency of these audits will be determined based on the recruitment rate, safety concerns, and previous audit findings.

Plans for communicating important protocol amendments to relevant parties (e.g., trial participants, ethical committees) {25}

Any modifications to the study protocol will require protocol amendments, which will be promptly submitted for approval to the Institutional Review Board. These changes will only be implemented after receiving approval from the Institutional Review Board. Once approved, ClinicalTrials.gov will be updated to reflect any significant changes. If necessary, protocol training to implement the amendments will be provided by the study team to participating centers.

Dissemination plans {31a}

After database closure and data analysis, the trial statistician will prepare a report detailing the main study results. Following this, a meeting of the investigators will be convened to discuss the findings before drafting a scientific manuscript for peer review and publication in a major scientific journal. Additionally, efforts will be made to present the results at key international conferences of neuroendocrine societies.

Discussion

This trial represents a significant milestone in evaluating the efficacy of combined pharmacological and mechanical prophylaxis in reducing VTE events in postoperative CD patients. To date, no similar randomized controlled trials have addressed this specific clinical question.

Transnasal transsphenoidal pituitary tumor resection is the preferred surgical approach for patients with CD. Compared to craniotomy, transsphenoidal surgery has a significantly lower risk of bleeding. The published literature indicates a bleeding risk of 0.02% following transsphenoidal surgery [17], whereas the incidence of intracranial hemorrhage after craniotomy ranges from 1% to 1.5% [18]. Therefore, for clinical practicality and safety, this study will exclusively include patients undergoing transsphenoidal resection.

Early meta-analyses indicated that low molecular weight heparin is generally safer, with a relatively lower bleeding risk compared to rivaroxaban, particularly when used for thrombosis prevention after hip and knee replacement surgeries [19]. However, recent studies have shown that rivaroxaban may have no significant difference in major bleeding and non-major bleeding risks compared to enoxaparin in thromboprophylaxis following non-major orthopedic surgeries of the lower limbs [20]. Given the risk of postoperative bleeding and the potential bleeding side effects of oral medications, LMWH was chosen for initial postoperative treatment because of its relatively lower bleeding risk. As patients prepare for discharge, the more convenient oral medication was selected for ongoing prophylaxis.

Patients who develop early VTE on the first day after surgery or despite anticoagulant use will be included in a further post hoc analysis. This will help identify risk factors for VTE. This analysis will aim to determine why VTE occurred despite anticoagulant use and explore whether specific factors, such as hypertension, diabetes, body mass index, or disease duration, are associated with increased risk. Based on our findings, recommendations may include earlier initiation of prophylaxis, dosage adjustments, or extended duration of treatment for high-risk patients.

Trial status

This protocol is based on trial protocol version 5.0, dated July 1, 2021. The first patient was enrolled in December 2020, and the final patient is expected to be enrolled by the end of 2024. While the original plan anticipated completing recruitment by December 2022, the COVID-19 pandemic significantly impacted many districts and cities in China, leading to lockdowns that have severely delayed the implementation and recruitment for this trial.

Data availability {29}

Data will be made available from the corresponding author upon reasonable request.

Abbreviations

CD:: Cushing’s disease
VTE:: Venous thromboembolism
DVT:: Deep vein thrombosis
PE:: Pulmonary embolism
CEC:: Clinical events committee
IRC:: Independent Review Committee
CTA:: Computed tomography angiography
eCRFs:: Electronic case report forms
AE:: Adverse events
SAE:: Severe adverse events
DSMB:: Data Safety Monitoring Board

References

Fleseriu M, Auchus R, Bancos I, Ben-Shlomo A, Bertherat J, Biermasz NR, et al. Consensus on diagnosis and management of Cushing’s disease: a guideline update. Lancet Diabetes Endocrinol. 2021;9(12):847–75. https://doi.org/10.1016/S2213-8587(21)00235-7.

Article PubMed PubMed Central Google Scholar
Feelders RA, Nieman LK. Hypercoagulability in Cushing’s syndrome: incidence, pathogenesis and need for thromboprophylaxis protocols. Pituitary. 2022;25(5):746–9. https://doi.org/10.1007/s11102-022-01261-9.

Article CAS PubMed PubMed Central Google Scholar
White AJ, Almeida JP, Filho LM, Oyem P, Obiri-Yeboah D, Yogi-Morren D, et al. Venous Thromboembolism and Prevention Strategies in Patients with Cushing’s Disease: A Systematic Review. World Neurosurg. 2024;S1878–8750(24):01460–8. https://doi.org/10.1016/j.wneu.2024.08.090.

Article Google Scholar
Waqar M, Chadwick A, Kersey J, Horner D, Kearney T, Karabatsou K, et al. Venous thromboembolism chemical prophylaxis after endoscopic trans-sphenoidal pituitary surgery. Pituitary. 2022;25(2):267–74. https://doi.org/10.1007/s11102-021-01195-8.

Article CAS PubMed Google Scholar
Wagner J, Langlois F, Lim DST, McCartney S, Fleseriu M. Hypercoagulability and Risk of Venous Thromboembolic Events in Endogenous Cushing’s Syndrome: A Systematic Meta-Analysis. Front Endocrinol (Lausanne). 2019;9:805. https://doi.org/10.3389/fendo.2018.00805.

Article PubMed Google Scholar
Nieman LK, Biller BM, Findling JW, Murad MH, Newell-Price J, Savage MO, et al. Treatment of Cushing’s Syndrome: An Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab. 2015;100(8):2807–31. https://doi.org/10.1210/jc.2015-1818.

Article CAS PubMed PubMed Central Google Scholar
Boscaro M, Sonino N, Scarda A, Barzon L, Fallo F, Sartori MT, et al. Anticoagulant prophylaxis markedly reduces thromboembolic complications in Cushing’s syndrome. J Clin Endocrinol Metab. 2002;87(8):3662–6. https://doi.org/10.1210/jcem.87.8.8703.

Article CAS PubMed Google Scholar
White AJ, Almeida JP, Petitt JC, Yogi-Morren D, Recinos PF, Kshettry VR. Significant Variability in Postoperative Thromboprophylaxis in Cushing’s Disease Patients: A Survey of the North American Skull Base Society and the AANS/CNS Joint Tumor Section. J Neurol Surg B Skull Base. 2023;85(5):540–5. https://doi.org/10.1055/s-0043-1772698.

Article PubMed PubMed Central Google Scholar
Isand K, Feelders R, Brue T, Toth M, Deutschbein T, Reincke M, et al. High prevalence of venous thrombotic events in Cushing’s syndrome: data from ERCUSYN and details in relation to surgery. Eur J Endocrinol. 2024;190(1):75–85. https://doi.org/10.1093/ejendo/lvad176.

Article PubMed Google Scholar
Anderson DR, Morgano GP, Bennett C, Dentali F, Francis CW, Garcia DA, et al. American Society of Hematology 2019 guidelines for management of venous thromboembolism: prevention of venous thromboembolism in surgical hospitalized patients. Blood Adv. 2019;3(23):3898–944. https://doi.org/10.1182/bloodadvances.2019000975.

Article PubMed PubMed Central Google Scholar
Nyquist P, Bautista C, Jichici D, Burns J, Chhangani S, DeFilippis M, et al. Prophylaxis of Venous Thrombosis in Neurocritical Care Patients: An Evidence-Based Guideline: A Statement for Healthcare Professionals from the Neurocritical Care Society. Neurocrit Care. 2016;24(1):47–60. https://doi.org/10.1007/s12028-015-0221-y.

Article CAS PubMed Google Scholar
Raksin PB, Harrop JS, Anderson PA, Arnold PM, Chi JH, Dailey AT, et al. Congress of Neurological Surgeons Systematic Review and Evidence-Based Guidelines on the Evaluation and Treatment of Patients With Thoracolumbar Spine Trauma: Prophylaxis and Treatment of Thromboembolic Events. Neurosurgery. 2019;84(1):E39–42. https://doi.org/10.1093/neuros/nyy367.

Filed under: Cushing's, pituitary, Treatments | Tagged: Cushing's Disease, pituitary, post-op, thromboprophylaxis | Leave a comment »

Cardiometabolic Complications After Cushing’s Disease Remission

Posted on April 30, 2025 by MaryO

Abstract

Background and aim

Cushing’s disease (CD) is associated with phenotypic traits and comorbidities that may persist after the normalization of cortisol levels. Medical therapy is usually given in recurrent or persistent CD after transsphenoidal surgery. We aimed to investigate the impact of long-term normalization of daily cortisol secretion on clinical picture and cardiometabolic comorbidities, comparing surgical remission to medical treatment.

Methods

Monocentric retrospective study, two- and five-years observation. Sixty CD patients, with sustained normal 24-h urinary free cortisol (UFC) levels, divided group 1 (surgical remission, n = 36) and group 2 (medical remission, n = 24).

Results

Patients were different after achieving eucortisolism with surgery or medical treatment. Phenotypic traits: round face, dorsocervical fat pad, and bruisability persisted more prominently in the group 2, however abdominal obesity and muscle weakness persisted in both groups, especially in those patients with increased late-night salivary cortisol (LNSC). Hypertension: greater improvement was observed in group 1 (-31% vs. -5%, p = 0.04). Diabetes: less prevalent in group 1 after 2 years (2/36 vs. 9/24, p = 0.002), with a corresponding reduction in glucose-lowering treatments and persistence of impaired LNSC in diabetic patients (p < 0.001). Dyslipidemia: remained widespread in both groups, with minimal improvement over time (-22% in surgical and − 6% in medical cohort).

Conclusions

Surgical remission leads to faster and sustained improvements in clinical phenotype. However, obesity, arterial hypertension, and dyslipidemia do not completely revert in five years, especially during medical treatment. Most comorbidities persist despite UFC normalization, due to impaired LNSC: the recovery of cortisol rhythms confirms the remission of hypercortisolism.

Introduction

Cushing’s disease (CD) is caused by an adrenocorticotropic hormone (ACTH)-secreting pituitary tumor, resulting in persistent endogenous hypercortisolism. The cortisol excess leads to a typical clinical picture: round face, facial plethora, buffalo hump, cutaneous striae rubrae, easy bruising, proximal myopathy, weight gain with visceral obesity, hirsutism and acne [1,2,3]. Moreover, several comorbidities are cortisol-related: metabolic syndrome (visceral obesity, arterial hypertension, glucose intolerance or diabetes, and dyslipidemia), acquired thrombophilia, osteoporosis or vertebral fractures, immunological impairments with increased infection susceptibility, and psychiatric disorders [4]. The sum of physical changes and comorbidities leads to a reduced life expectancy and a worsening of the quality of life [5]. Pituitary trans-sphenoidal surgery (TSS) is the first-choice CD treatment [1]. Despite high remission rates (up to 90% in referral centers) [6], the risk of recurrence varies from 10 to 47% [7], especially in series with long-term follow-up. If surgery fails or is not feasible, cortisol excess can be managed with medical therapy. Not rarely, patients on cortisol-lowering therapy experience fluctuations of their cortisol levels, making outcome evaluations difficult and hardly standardized. The goals of CD treatment are to normalize cortisol levels, and to reduce the burden of comorbidities. The most used biochemical marker in clinical practice is urinary free cortisol (UFC), which estimates the cumulative daily secretion of cortisol, but does not offer information about cortisol rhythm [8].

In this study we compared two groups of CD patients with sustained normalization of 24-h UFC due either to post-surgical or medical cortisol-lowering therapy remission. The aim of the study was to analyze the impact of long-term normalization of hypercortisolism in terms of UFC, achieved with surgical or medical treatment, on endocrine parameters, cortisol-related clinical picture and comorbidities, in a five-years observation period of patients with CD.

Materials and methods

Subjects

Sixty CD patients were enrolled (75% female); the median age at diagnosis was 41 years (interquartile range [IQR] 32–52), followed at the Endocrinology Unit of Padua University Hospital from 2000 to 2021. This observational study was conducted in accordance with the STROBE (STrengthening the Reporting of OBservational studies in Epidemiology) guidelines [9]. The study, following the guidelines in the Declaration of Helsinki, was approved by the ethics committee of Padova University Hospital (PITACORA, protocol No. AOP3318, ethics committee registration 5938-AO-24), and all patients gave informed consent. All data are included in the Repository of the University of Padova [10].

The first normalized UFC is considered as the starting point of observation at follow-up (two or five years). The cohort was divided into two cohorts: group 1 achieved CD remission after surgery, and group 2 achieved long-term eucortisolism during medical therapy. The inclusion criterion was 24-h UFC levels (mean of two collections) below the upper limit of normality during the observational period. Postoperative long-term adrenal insufficiency requiring substitutive glucocorticoid treatment (with hydrocortisone or cortisone acetate tablets) 12 months after surgery or new-onset hypopituitarism were considered exclusion criteria. The group 1 was made of 36 patients (69% female) in remission after successful TSS. The second group consisted of 24 patients (83% female) on long-term medical treatment for CD persistence (n = 17) or relapse (n = 4) after surgery and three patients in primary medical therapy for poor surgical eligibility, as shown in Fig. 1. Within group 2, nine patients underwent previous radiotherapy without efficacy, at least 5 years before reaching adequate biochemical control with medical treatment; none developed hypopituitarism. 14/24 patients (58%) were treated with a monotherapy and 11/24 (46%) with combined therapies during the observation period. Details on medical therapies are shown in Table 1. In particular, 3 patients were treated with metyrapone + pasireotide s.c., 1 with metyrapone + ketoconazole, 2 with ketoconazole and cabergoline, 1 with metyrapone + cabergoline, 1 with metyrapone + ketoconazole + cabergoline, 1 with metyrapone + ketoconazole + pasireotide s.c., 1 with metyrapone + ketoconazole + pasireotide s.c. + cabergoline. Metyrapone and ketoconazole were administered two/three times a day, pasireotide s.c. twice daily and cabergoline once daily in the evening.

Table 1 Cortisol-lowering drugs, dose, and time in treatment of subjects treated with a single and combined lines of therapy

All 60 patients completed at least 2 years of follow-up; a long-term 5-years evaluation was available in 43 patients of the original cohort (32 after surgery and 11 with medical therapy). Baseline characteristics of the two cohorts are reported in Table 2.

Table 2 Baseline characteristics of the two groups and previous treatment modalities

Data collection and study design

Two researchers retrieved clinical and biochemical data independently from the local digital medical records. We considered as baseline visit the clinical and endocrine evaluation performed with active hypercortisolism. Therefore, the baseline visit consists in the pre-surgical evaluation in group 1, and in the post-surgical confirmation of active hypercortisolism in those in medical treatment (or diagnosis in case of primary treatment, group 2).

We considered clinical and biochemical outcomes during routine follow-up at two- and five-years in each group, starting from surgical remission or the beginning of a stable normalization of UFC under medical therapy. CD diagnosis was based on at least two parameters among 24-h UFC above the upper normal limit (ULN, at least two collections), unsuppressed cortisol levels (> 50 nmol/L) after 1 mg overnight dexamethasone test (1 mg-DST) or late-night salivary cortisol (LNSC) > ULN (at least two samples). In all subjects, CD diagnosis was considered in case of normal-high ACTH levels, positive response to dynamic tests (corticotropin-releasing hormone or desmopressin test, high-dose dexamethasone test), and, two cases, with petrosal sinus sampling (BIPSS) [11]. Long-term remission after TSS was defined through normal UFC, combined with serum cortisol levels < 50 nmol/L in the first month after surgery and need of glucocorticoid replacement therapy. A relapse of CD was defined as the reappearance of the typical signs and symptoms of CD associated with the alteration of at least two first-line screening tests. Presence/absence of clinical signs of CD (round face, facial rubor, buffalo hump, bruising, cutaneous red striae, acne, hirsutism and oligo/amenorrhea in females) were evaluated during outpatient visits by expert endocrinologists. The presence of hirsutism in females was measured according to the Ferriman–Gallwey score > 8 (extent of hair growth in 9 locations was rated 0–4). Proximal muscle strength was diagnosed if patients were not able to stand up from a low seated position with anteriorly extended arms. Bodyweight, body mass index (BMI), waist and hip circumference, systolic (SBP), and diastolic blood pressure (DBP) were assessed with calibrated tools. Overweight was diagnosed in patients with BMI 25–30 kg/m², obesity with BMI > 30 kg/m². Visceral obesity was diagnosed as waist circumference ≥ 94 cm in men and ≥ 80 cm in women, or with a waist/hip ratio (WHR) ≥ 1 according to International Diabetes Federation criteria. Arterial hypertension was diagnosed for SBP above 140 mm Hg and/or DBP above 90 mm Hg and/or in patients on antihypertensive drugs. Diabetes mellitus (DM) was diagnosed according to American Diabetes Association criteria or when patients were taking antidiabetic medication. Dyslipidemia was diagnosed when low-density lipoprotein (LDL) calculated cholesterol was ≥ 100 mg/dL and hypertriglyceridemia when triglycerides were ≥ 150 mg/dL or when patients were on lipid-lowering medication. The presence of carotid vascular disease (CVD) has been assessed by supra-aortic vessels duplex ultrasound. Cushing’s cardiomyopathy (CCM) was diagnosed by doppler echocardiography with evidence of impaired relaxation and left ventricular filling pattern. The medical history was checked for cardiovascular disease (acute coronary syndrome, ACS) in all cases. A shortened activated partial thromboplastin time (aPTT < 29 s) defined pro-thrombotic status.

Assays

All biochemical analyses were carried out in an ISO15189:2012-accredited clinical laboratory [12], cortisol levels have been measured in urine or saliva with a mass-spectrometry home-made validated method. UFC was determined by a home-brew liquid chromatography-mass spectrometry (LC-MS/MS) method (intra-assay/interassay coefficient of variation [CV] < 6%/< 8%) since 2011 [13], previously by a radio-immunometric assay (Radim, intra-assay/interassay CV < 3%/< 9%). The patients were instructed to discard the first morning urine void and to collect all urine for the next 24 h, so that the morning urine void on the second day was the final collection. The sample was kept refrigerated from collection time until it was analyzed: normal range for UFC is 16–168 nmol/24 h.

Salivary cortisol was measured by a radio-immunometric assay (Radim, intra-assay/interassay CV < 3%/< 9%) until 2014 [14], after then by LC-MS/MS method (intra-assay/interassay CV < 6%/< 8% [15]). In order to prevent food or blood contamination, samples were collected at least 30 min after subjects had eaten, brushed their teeth, smoked or assumed liquorice; undertaken using Salivette^® devices containing a cotton swab with or without citric acid (Sarstedt, Nümbrecht, Germany). The sample was stored at − 80 °C, before analyses [15].

The 1-mg DST test was performed orally assuming 1 mg of dexamethasone between 11 P.M. and midnight, sampling serum cortisol the next morning at 8 A.M. Serum dexamethasone levels, routinely evaluated since 2017, were adequate in all cases [16]. Serum cortisol (RRID: AB_2810257) and ACTH (RRID: AB_2783635) were determined by immune-chemiluminescence assay (Immulite 2000, Siemens Healthcare). Dynamic second-line tests and BIPSS were performed according to international standards.

Statistical analysis

Data were analyzed using SPSS Software for Windows, version 24.0 (SPSS Inc). Data are reported as medians and interquartile range or as percentages. The comparison between continuous variables was performed by non-parametric Wilcoxon test or Mann–Whitney test, as appropriate. The comparison between categorical variables was performed by the χ2 test. The correlation between continuous variables was performed by linear regression analysis. The level of significance for the overall difference between the groups was tested with one-way ANOVA. A p value < 0.05 was considered statistically significant.

Results

Endocrine evaluation

At baseline the two groups were similar for morning serum/salivary cortisol, LNSC, cortisol after 1 mg DST and morning ACTH levels (Table 3); UFC levels were higher in the surgical cohort (p < 0.001). Endocrine parameters were not influenced by sex and BMI. At baseline, all patients had impaired salivary cortisol rhythm with increased LNSC and inadequate cortisol suppression after 1-mg DST. At two years the recovery of salivary cortisol rhythm was observed in 97% of patients after surgery and 50% of patients during medical therapy. The only patient who did not show recovery of cortisol rhythm in the surgical cohort had LNSC of 5.4 nmol/L (range 0.5–2.6 nmol/L), with adequate cortisol suppression after 1-mg DST and sustained normal UFC: it was considered a false-positive due to residual minor depression state.

Table 3 Biochemical pattern at baseline and during the follow-up

Adequate cortisol suppression after 1-mg DST (both with normal UFC and LNSC) was observed in 34 out of 36 patients (94%) in the surgical cohort; the two patients who did not show complete cortisol suppression after 1-mg DST had cortisol levels of 60 and 119 nmol/l, respectively. On the contrary, as per selection criteria, none of the patients in group 2 presented suppressed cortisol after 1-mg DST.

At 5 years follow-up, all cases in the surgical cohort had suppressed cortisol after 1-mg DST and normal salivary cortisol rhythm, whereas in group 2 9% had suppressed cortisol after 1-mg DST and 36% recovered salivary cortisol rhythm. At 5 years, UFC and salivary cortisol levels (either morning or late night) were similar in the two groups, while the median value of serum cortisol after 1-mg DST remained not adequately suppressed (median 75 nmol/L, from 18 to 257 nmol/L) during medical therapy (See Table 3). In group 2, patients on combined therapy had higher UFC (102 vs. 76 nmol/24h p = 0.03) and LNSC (2.4 vs. 1.9 p = 0.05) at 5 years, compared to patients on monotherapy.

Hirsutism, abdominal obesity, round face and facial rubor were prevalent in group 1 at baseline. On the contrary, the abdominal obesity, facial rubor and easy bruising were most commonly found in the medical cohort. The prevalence of facial rubor, buffalo hump and bruisability was higher after medical than surgical remission after 2 years of eucortisolism; at 5 years the prevalence of buffalo hump and bruisability was higher in patients under drug therapy as well (Table 4; Fig. 2). Higher levels of UFC at baseline were observed in all patients with proximal myopathy (p < 0.001).

Table 4 Two- and five-years changes in clinical phenotype from baseline in group 1 and group 2

Fig. 2

Signs and symptoms of hypercortisolism at baseline (grey bars), two-years (orange bars) and five-years (blue bars) follow up after surgical (TSS) or medical remission (MED)

Full size image

Arterial hypertension

Arterial hypertension (AH) was the most frequent comorbidity in both groups at baseline, with similar distribution in the two groups (Table 5). The prevalence of AH decreased after two years in both groups, especially in the surgical cohort (64% vs. 44% in group 2, p < 0.001; 75% vs. 71% p = 0.003), with no further improvement after five years. Overall, hypertensive patients were older at diagnosis (45yrs vs. 31y; p < 0.001) and with larger BMI (29 vs. 25 kg/m²; p = 0.03). Median UFC, morning salivary cortisol and LNSC, and 1-mg DST were not different in patients with/without AH at baseline and at 2 years. SBP and DBP values were similar in the two cohorts and were not correlated to UFC, LNSC or 1-mg DST throughout the follow-up. At 2 years, hypertensive patients had higher levels of morning salivary cortisol and LNSC with impaired rhythm (respectively 10.4 vs. 6 nmol/L, p = 0.01 and 3.2 vs. 1 nmol/l, p = 0.007). SBP and DBP values did not change during the five-years observation time in both groups; however, the number of anti-hypertensive drugs was higher in group 2 than in group 1 (p = 0.007). Overall patients treated with metyrapone showed higher values of DBP at 2 years (mean 89.4 vs. 81.7 mmHg, p = 0.01), the prevalence of AH did not differ from patients with other medical treatments.

Table 5 Two- and five-years changes in cardio-metabolic cortisol-related comorbidities of CD from baseline in group 1 and group 2