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

CD

Cushing’s Disease

CI

Confidence Interval

CRH

Corticotropin-Releasing Hormone

DFS

Disease-Free Survival

DL

Deep Learning

eTSS

Endoscopic Transsphenoidal Surgery

HR

Hazard Ratio

IPSS

Inferior Petrosal Sinus Sampling

ML

Machine Learning

MRI

Magnetic Resonance Imaging

OS

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).

  1. Download: Download high-res image (108KB)
  2. Download: Download full-size image

Fig. 1. Rates of surgical complications. (a) Intraoperative complications; (b) hormonal dysregulation rates following surgery.

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
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 sitea
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/).

  1. Download: Download high-res image (97KB)
  2. Download: Download full-size image

Fig. 2. Nomogram for predicting postoperative persistence or recurrence of Cushing’s disease (PoRP-CD). This nomogram visually represents the predictive model for assessing the risk of recurrence or persistence of Cushing’s disease following surgery. Each predictor variable—Symptom duration (months), Knosp grading, Hardy’s grading, previous pituitary surgery, and tumor site— contributes a point value that aligns with the “Linear Predictor” scale, which maps to the “Probability of Persistence” scale, allowing estimation of recurrence likelihood.

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.

  1. Download: Download high-res image (250KB)
  2. Download: Download full-size image

Fig. 3. Disease-free survival (DFS) analysis. (A) Kaplan-Meier curve of DFS for the entire cohort, showing a gradual decline over time; (B) DFS stratified by Hardy’s Grade, demonstrating significant impact of grade 3 on survival outcomes (P = 0.03); (C) DFS by tumor site, highlighting no significant association between tumor site and survival care (P > 0.05); (D) DFS based on previous surgery status, indicating a higher risk of recurrence or death in patients with prior surgical interventions (P < 0.01); (E) DFS by symptom duration, highlighting no significant association (P = 0.57).

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

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

Filed under: Cushing's, pituitary, symptoms, Treatments | Tagged: , , , | Leave a comment »

Changing face of Cushing’s Disease Over Three Decades in Pituitary Center

Posted on October 1, 2025 by MaryO

Abstract

Objective

Cushing Disease (CD) presents with typical clinical findings, even though, there is a wide spectrum of manifestations. Over the years, the sings and symptoms of Cushing’s syndrome (CS) have become more subtle and atypical forms of CS have emerged. In this study, we aimed to investigate the changes in the clinical presentation of CD in recent years.

Materials and methods

In this study, CD patients followed by our center were examined. A total of 258 patients with CD were included in the study. The clinical findings at the time of presentation, laboratory and imaging findings, treatment modalities and remission status in the first year after treatment were evaluated.

Results

The mean age of the patients included in the study was 41.3 ±13.28 years. CD patients diagnosed between 2013 and 2023 were older than those diagnosed between 1990 and 2012 (p < 0.001). There was no difference between the groups in terms of gender. Moon face, purple striae, hirsutism, and menstrual irregularities were statistically significantly less frequent in the last 10 years than in previous years (p < 0.001; p = 0.004; p < 0.001; p < 0.001, respectively). In addition, patients who applied after 2013 had lower baseline cortisol and adrenocorticotropic hormone (ACTH) levels, and a smaller median size of the pituitary adenoma. Limitations of the study include its retrospective design and the subjectivity of clinical data.

Conclusion

As the clinical presentation of Cushing’s disease changes over time, waiting for the typical Cushing’s clinic can delay diagnosis. It is important that clinicians take this into account when they suspect CD.

 This is a preview of subscription content, log in via an institution  to check access.

Filed under: Cushing's, pituitary, symptoms | Tagged: , , , , , , | Leave a comment »

Research Study for Patients Diagnosed with Cushing’s Disease and Their Caregivers

Posted on September 8, 2025 by MaryO

We’re looking for caregivers to loved ones diagnosed with Cushing’s Disease or patients diagnosed with Cushing’s Disease to participate in a research study.

✅ Who: Patients and caregivers of loved ones

⏳ What: 30-minute Online Survey

💰 Compensation: $60.00

Sign up here: https://rarepatientvoice.com/CushingsHelp/

 

Filed under: Cushing's, Interview, pituitary | Tagged: , , , , | Leave a comment »

The STUB1-TPIT Axis Regulates the Secretion of Adrenocorticotrophic Hormone in Cushing Disease

Posted on September 4, 2025 by MaryO

Abstract

Background

Cushing’s disease (CD) is a clinical syndrome caused by excessive secretion of adrenocorticotropic hormone (ACTH) from a pituitary corticotroph adenoma, resulting in adrenal cortical hyperplasia and overproduction of cortisol. The T-box transcription factor (TPIT) is crucial for regulating ACTH secretion in pituitary corticotroph adenomas. This study aims to explore the ubiquitin-mediated degradation of TPIT and identify potential pharmaceutical agents for treating CD.

Methods

The TPIT-interacting protein STUB1 was identified via mass spectrometry. The interaction between STUB1 and TPIT was confirmed using NanoBiT and GST pulldown assays. The expression of TPIT, pro-opiomelanocortin (POMC), and STUB1 was assessed by immunoblotting, dual-luciferase reporter assays, quantitative real-time PCR, RNA-sequencing, and immunohistochemistry. ACTH levels were measured by ELISA.

Results

STIP1 Homology and U-Box Containing Protein 1 (STUB1) interacts with TPIT through its TPR domain and ubiquitinates multiple sites on TPIT via the U-box domain, leading to TPIT degradation. This degradation reduces POMC expression and ACTH secretion in AtT-20 cells. Additionally, STUB1 inhibits cell proliferation both in vitro and in vivo. Clinical investigations revealed that STUB1 expression is significantly lower in ACTH-secreting corticotroph adenomas than in silent corticotroph adenomas (SCAs). A negative correlation was observed between STUB1 and TPIT protein levels, as well as POMC expression. Furthermore, NanoBiT drug screening identified that Irbesartan and Lumiracoxib increased TPIT degradation, thereby reducing POMC expression and ACTH secretion.

Conclusion

STUB1 is a promising therapeutic target for CD and drugs targeting the STUB1-TPIT complex may provide a potential treatment approach.

Graphical Abstract

E3 ubiquitin ligase STUB1 interacts with TPIT, mediating its ubiquitination degradation, subsequently inhibiting POMC transcription and ATCH secretion. Irbesartan and Lumiracoxib can promote TPIT degradation and suppress ACTH secretion by enhancing the interaction between STUB1 and TPIT.

Introduction

Cushing’s disease (CD) is a rare but clinically significant endocrine disorder characterized by excessive secretion of adrenocorticotropic hormone (ACTH) from a pituitary corticotroph adenoma. Corticotroph adenoma is characterized by expression of the transcription factor TPIT, which promotes the expression of the pro-opiomelanocortin (POMC) and ACTH secretion [1]. The excessive ACTH triggers a cascade of severe endocrine metabolic disorders and complications such as hypertension, diabetes mellitus, hyperlipidemia, osteoporosis, central obesity, and cardiovascular disease [2, 3]. These symptoms collectively impact the patient’s quality of life and increase the risk of life-threatening complications like cardiovascular disease and infections. Cushing’s disease is a highly challenging condition to manage, despite the availability of various treatment options, including surgery, medical therapy, and radiation therapy, which often fall short of achieving satisfactory outcomes. Elucidating the mechanisms underlying ACTH overproduction in Cushing’s disease is essential to develop more effective treatments and address current clinical challenges.

Ubiquitination, the covalent attachment of ubiquitin to proteins, marks them for proteasomal degradation, maintaining cellular homeostasis [4]. Dysregulation of this process can cause aberrant protein accumulation, disrupt cellular functions, and drive tumor development and progression. In corticotroph adenomas, ubiquitination also plays a significant role [5]. For example, the mutations in the USP8 gene, present in 35–62% of ACTH adenomas, increase USP8 deubiquitinase activity, leading to enhanced epidermal growth factor receptor (EGFR) deubiquitination and subsequent elevation of ACTH secretion [6, 7]. Additionally, mutations in BRAF and USP48 also contribute to POMC transcription and ACTH production [8]. Our previous research showed that ubiquitination of TPIT by TRIM65 inhibits POMC transcription and ACTH production [9]. USP11 deubiquitinates and stabilizes TPIT, thereby promoting POMC transcription and ACTH secretion [10]. While these findings highlight the critical role of ubiquitination pathways in regulating ACTH secretion in corticotroph adenomas, a more detailed understanding of the underlying mechanisms driving ACTH overproduction is still needed. Moreover, the development of targeted therapies to modulate these pathways remains a significant unmet clinical need.

STIP1 Homology and U-Box Containing Protein 1 (STUB1) is a human gene encoding the C-terminus of HSC70-interacting protein (CHIP) [11], playing a crucial role in maintaining protein homeostasis by targeting misfolded or damaged proteins for proteasomal degradation [12]. Recent studies have highlighted the involvement of STUB1 in various human malignancies, as it regulates the stability and activity of several tumor-associated transcription factors, such as CRIP1, YAP1, HIF-1α, and c-Myc, thereby influencing tumor growth, invasion, metastasis, and angiogenesis [13,14,15,16]. Moreover, STUB1 has been shown to modulate the expression of tumor-secreted factors, such as cytokines, growth factors, and matrix metalloproteinases, by ubiquitinating and regulating transcription factors such as CSF2RB, NF-κB, and STAT3 [15, 17, 18]. Furthermore, STUB1 has been implicated in the modeling of the tumor immune microenvironment through its ability to regulate transcription factors that affect the secretory functions of both tumor and immune cells [19, 20]. Despite the extensive research on STUB1’s functions in various biological processes and its emerging role in cancer biology, its specific involvement and potential mechanisms in corticotroph adenomas remain to be elucidated.

Here, we found that STUB1-mediated ubiquitination and subsequent degradation of TPIT led to an inhibition of ACTH secretion, becoming a potential therapeutic target for the management of Cushing’s disease. The drugs Irbesartan or Lumiracoxib, which promoted the binding of STUB1 and TPIT, hold promise as viable avenues for the treatment of CD.

Materials and methods

Cell culture and reagents

A Mycoplasma Stain Assay kit (C0296, Beyotime, China) was used to test all cell lines for mycoplasma contamination prior to use. HEK-293T and AtT-20 cells were kindly provided by the Cell Bank, Chinese Academy of Sciences. In cell culture, the HEK293T cells (Passage number < 20) were cultivated in Dulbecco’s Modified Eagle Medium (DMEM; Basal Media, L110KJ) supplemented with 10% (v/v) fetal bovine serum (FBS; Basal Media, S660JY). The AtT-20 cells (Passage number < 20) were cultured in Ham’s F-12K (Kaighn’s) medium (Basal Media, L450KJ) supplemented with 2.5% FBS and 15% horse serum (Absin, abs989). All cell lines were maintained in a humidified atmosphere with 5% CO 2 at 37 ℃.

Isolation and cultivation of pituitary tumor cells

Freshly obtained corticotroph tissues were rinsed with Hank’s Balanced Salt Solution (HBSS, Gibco, 14,025,092) and cut into small pieces with a sterile scalpel. The fragments were digested in HBSS supplemented with 100 U/ml collagenase, type IV (Gibco, 17104019) at 37 ℃ for 6–8 h. Following digestion, the dispersed cells were filtered using a 100 μm cell strainer (Beyotime Biotechnology, FSTR100) and resuspended in ACK Lysing Buffer (Gibco, A1049201) for 5 min. The cells were subsequently resuspended in DMEM supplemented with 10% FBS and plated in a cell culture dish (BIOFIL, TCD010100). After 48 h of incubation, the floating tumor cells were utilized for subsequent experiments.

Plasmids

Supplementary Table 1 provides additional details. The pLVshRNA-EGFP(2A)Puro cloning vector was used to generate the shRNAs targeting genes at various regions. The sequences of the shRNAs can be found in Supplementary Table 2.

Transfection

To prepare the polyethyleneimine (PEI) transfection reagent, 1 g of Polyethylenimine Linear (PEI) with a molecular weight of 40,000 (Yeasen, 40816ES03) was precisely weighed and dissolved in 900 mL of ultrapure water within a 1 L glass beaker. The pH of the solution was carefully adjusted to a range of 6.80 to 6.90 using a sodium hydroxide solution. Subsequently, the solution was diluted to a final volume of 1 L. Finally, the solution was filtered through a 0.22 µm filter to obtain a clear and sterile 1 mg/mL PEI solution.

For cell transfection, HEK293T cells were seeded in culture dishes one day prior to transfection, with a target cell confluency of 70% to 90% at the time of transfection. For the PEI transfection procedure, a 1.5 mL centrifuge tube was prepared. Into this tube, 200 µL of Opti-MEM serum-free medium (Gibico, 31,985,070), 6 µg of plasmid DNA, and 18 µL of the PEI reagent were added, thereby maintaining a 1:3 mass/volume ratio of plasmid to PEI. The components were thoroughly mixed by vortexing for 30 s, followed by a brief centrifugation to collect the liquid at the bottom of the tube. The mixture was then incubated at room temperature for 15 min before being added dropwise to a 6 cm culture dish (BIOFIL, TCD010060). Subsequent experimental operations can be performed 24 to 48 h later.

Mass spectrometry (MS)

Approximately 10 million primary corticotroph tumor cells were infected with Flag-TPIT adenovirus at a multiplicity of infection (MOI) of 10. The cells were lysed using a Triton X-100 lysis buffer (150 mM NaCl, 50 mM Tris, and 1% Triton X-100, pH 7.5), and the immunoprecipitation complex was isolated utilizing anti-DYKDDDDK affinity beads (Smart-lifesciences Biotechnology). The beads were washed four times with Triton X-100 lysis buffer and the proteins were eluted with elution buffer (8 M urea, 100 mM Tris, pH 8.0). The eluted proteins were subsequently identified using MS.

The 10 million HEK-293T cells were transfected with Flag-TPIT for 48 h, followed by lysis. Immunoprecipitation of Flag-TPIT and its bound proteins was performed using 100 μL DYKDDDDK affinity beads. The beads were then washed using lysis buffer and boiled in SDS-PAGE sample buffer for 5 min at 95 ℃. The interacting proteins in the gel were identified by immunoblotting and MS.

Proteomic data analysis was done by Shanghai Luming Biological Technology Co., Ltd. using an Easy-NLC1200 nano-HPLC system. Peptide samples were re-dissolved in Buffer A (0.1% formic acid aqueous solution) and separated on a 75 μm × 150 mm RP-C18 column at 300 nL/min. They were cleaned by a gradient with blank solvent for 30 min. The eluates were analyzed by Q-Exactive MS in DDA mode, scanning 300–1600 m/z. The 20 strongest fragment profiles were collected after each full scan using HCD with an NCE of 28 and dynamic exclusion of 25 s. MS1 resolution was 45,000 at m/z 200, AGC target 1E6, and max injection time 50 ms. MS2 resolution was 15,000, AGC target 1E5, and max injection time 50 ms.

Co-immunoprecipitation (Co-IP)

The cells in 6 cm dish were lysed with 800 μL Triton X-100 lysis buffer containing a protease inhibitor cocktail and sonicated. 100 μL of the lysate was used as input, and the remaining 700 μL lysates were then incubated overnight with antibody (1–2 μg), followed by incubation with 30 μL rProtein A/G Beads (Smart-lifesciences Biotechnology, SA032025) for 1 h at room temperature. The immunoprecipitates were then washed five times with lysis buffer and analyzed by immunoblotting.

Immunoblotting

The total proteins were extracted with Triton X-100 lysis buffer supplemented with protease inhibitors (NCM Biotech, P003) and subsequently sonicated with a Qsonica Q700 Sonicator. The protein concentrations were measured using a BCA protein assay kit (Yoche Biotech, YSD-500T), with absorbances at 562 nm measured using a Biotek 800 TS microplate reader. Aliquots of whole-cell lysates were mixed with SDS-PAGE loading buffer and separated on SDS-PAGE, followed by transfer onto polyvinylidene difluoride membranes (Millipore, ISEQ10100). The antibodies used for probing the blots are listed in Supplementary Table 3.

Ubiquitination assay

HEK-293T cells were cultured in 6 cm plates and co-transfected with the total 6 μg plasmids for 24 h and then lysed with RIPA lysis buffer (50 mM Tris–HCl, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.5% sodium pyrophosphate, 0.1% SDS, pH 7.4) containing a protease inhibitor cocktail. The whole-cell lysates were then incubated overnight with 30 μL anti-DYKDDDDK affinity beads. The beads were washed five times with lysis buffer and analyzed by immunoblotting using the specified antibodies.

NanoBiT assay

HEK-293T cells were cultured in 6-well plates and transfected with plasmids containing LgBiT and SmBiT. The cells were lysed using cell lysis buffer (Vazyme, DL101–01), and centrifuged, and the proteins in the supernatant were quantified using a BCA protein assay kit. Equal volumes and amounts of protein were added to a white 96-well plate (Absin, abs7016), followed by the addition of the same volume of 20 μM luminescent substrate Furimazine (Topscience, T15359). The luminescence was measured using the BioTek Synergy Neo2 Hybrid Multimode Reader.

Immunofluorescence microscopy

The AtT-20 cells were infected with lentivirus containing Flag-TPIT. Subsequently, 100,000 of the infected cells were seeded in a 12-well plate that had been pre-coated with Matrigel (Absin, abs9410). The cells were fixed with 4% formaldehyde (Beyotime, P0099) for 20 min and then washed with PBS (Servicebio, G4202). Following this, the cells were permeabilized using a cold 0.2% Triton X-100 solution for 15 min and incubated overnight at 4 ℃ with anti-STUB1 and Flag antibodies after blocking with 3% BSA. The cells were then incubated with secondary antibodies conjugated with Alexa 488 (Cell Signaling Technology, 4412S) or Alexa 555 (Cell Signaling Technology, 4409S) for 1 h at 37 ℃. Nuclei were stained using DAPI (Beyotime, C1002), and the stained sections were evaluated and imaged using a Zeiss LSM880 lightning confocal microscope (Carl Zeiss AG, Oberkochen, Germany).

GST pulldown assay

The GST or His-tagged proteins were purified and GST pull-down assays were performed as previously described [21].

Nuclear-cytoplasmic separation

Nuclear and cytoplasmic proteins were separated using the Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime, P0027).

In vitro ubiquitination assay

The GST-TPIT and STUB1-His proteins were purified, and subsequently subjected to an in vitro ubiquitination assay using the Ubiquitinylation kit (Enzo Life Sciences, BML-UW992).

Stable cell lines

The HEK-293T cells were cultured in 10 cm plates. For transfection, 6 μg of psPAX2, 3 μg of pMD2.G, 9 μg of the targeting plasmid, and 54 μL of PEI transfection reagent were added to 1 mL of Opti-MEM medium. The mixture was vortexed and incubated at room temperature for 15 min before being added to the cell culture dishes. After 6-8 h of infection, the cell supernatant was removed and replaced with fresh cell culture medium. After 48 of transfection, media containing the virus were collected and filtered through 0.45μm nitrocellulose filters (Millipore, SLHV033RS). The virus was then concentrated using PEG8000 (5 × PEG8000: 150 mm NaCl; 25% PEG8000) and used to infect AtT-20 cells. After 6-8 h of infection, the cell supernatant containing the lentivirus was removed and replaced with fresh cell culture medium. After 48 h of infection, the medium was changed to one containing puromycin to select the infected cells. The stably transfected cells were selected using puromycin at 1 μg/ml for 1 week (Beyotime, ST551).

Cycloheximide (CHX) Chase Assay

The 500,000 stable cells were seeded into a 6-well plate with 3 mL of culture medium. The cells were treated with 100 μg/ml CHX (ApexBio, A8244), collected at specified time intervals, and analyzed by immunoblotting.

Dual-luciferase gene reporter assays

The pGL4.15 vector was utilized to create the luciferase reporter vector Pomc-Luc by inserting the mouse Pomc promoter sequence (–646 to + 65). HEK-293T cells were transfected with pGL4.15-Pomc-Luc, pRL-TK, Flag-TPIT, and Myc-STUB1 (wild-type or mutants). The dual-luciferase reporter assay was conducted using the Dual Luciferase Reporter Assay Kit (Vazyme, DL101-01).

Quantitative real-time PCR (qRT-PCR)

Prepare 100,000–200,000 cells, and total RNA was extracted from cells using a total RNA extraction kit (NCM Biotech, M5105). The concentration of RNA was determined using a DS-11 microphotometer. Subsequently, cDNA synthesis was carried out using the cDNA synthesis kit (ABclonal, RK20429). qRT-PCR amplification was performed using a SYBR Green qPCR kit (ABclonal, RK21203) in accordance with the manufacturer’s instructions. The qRT-PCR was conducted using an ABI 7500 Real-Time PCR System (Applied Biosystems). The primer sequences used for qRT-PCR are listed in Supplementary Table 4.

Luminescent cell viability assay

The numbers of AtT-20 cells were measured using a Beckman Coulter Z2 cell counter. Following this, a total of three thousand cells in 100 μL media were plated per well in white 96-well plates (Absin, abs7016). Subsequently, the cell viability was assessed using the CellCounting-Lite 2.0 Luminescent Cell Viability Assay (Vazyme, DD1101-03) with the BioTek Synergy Neo2 Hybrid Multimode Reader.

Enzyme-linked immunosorbent assay (ELISA)

A total of 40,000 AtT-20 cells in 1 mL of culture medium were seeded into 12-well plates. The supernatants were collected after 24 h and then centrifuged at 4 ℃ at 2000 g for 20 min. The ACTH levels were measured by an ELISA kit (Immunoway, KE1520).

Xenograft model

The Ethical Review Board at Shanghai Jiao Tong University School of Medicine’s Ruijin Hospital approved of the study’s animal experimentation. The protocols followed the Institutional Animal Care and Use Committee’s (IACUC) guidelines. Four-week-old nu/nu female mice (The number of mice in each group was 5) were procured and maintained in a controlled specific pathogen-free environment. One million AtT-20 cells were combined with Matrigel (Yeasen, 40187) in a 100-μL volume and then injected subcutaneously. Xenograft tumor volumes were assessed by measuring two perpendicular diameters with calipers and individually calculated using the formula: volume = a × b2/2 (where a denotes length and b denotes width). The weights of the mice and the dimensions of the tumors were measured twice a week. Subsequently, the mice were euthanized, and the tumors were excised, measured, and photographed.

Patient and tissue samples

Samples of pituitary adenomas (PAs) were obtained from patients undergoing surgical procedures at Ruijin Hospital between 2016 and 2023. The study received ethical approval from the Ethical Review Board of Ruijin Hospital, affiliated with Shanghai Jiao Tong University School of Medicine. All patients whose tumor tissues were utilized in this study provided written informed consent. Information on the patients is provided in Supplementary Table 5.

RNA sequencing and GTEx data analysis

The raw mRNA read counts were normalized using Fragments Per Kilobase of exon model per Million mapped fragments (FPKM). Specifically, read counts from 107 normal pituitary samples obtained from the GTEx V8 database were normalized by FPKM. To address batch effects between our data and the GTEx normal data, the normalizeBetweenArrays function in the limma package in R, a tool commonly utilized for powering differential expression analyses in RNA-sequencing and microarray studies, was used. The information on the expression of the genes (TPIT, POMC, and STUB1) is shown in Supplementary Table 5.

Immunohistochemistry (IHC)

The immunohistochemistry was performed as previously described [21]. For IHC staining of FFPE tumor tissue sections, dewax and rehydrate sections sequentially in dewaxing solutions (Ⅰ, Ⅱ, Ⅲ, 10 min each), anhydrous ethanol (Ⅰ, Ⅱ, Ⅲ, 5 min each), and distilled water (5 min). Perform antigen retrieval by boiling in sodium citrate buffer (pH 6.0) for 30 min using a microwave histoprocessor, then cool and wash in PBS (pH 7.4, 3 × 5 min). Block endogenous peroxidase with 3% hydrogen peroxide (25 min, RT, dark), then wash in PBS (3 × 5 min). Block with 3% BSA (or rabbit serum for goat primary antibodies, 30 min, RT). Incubate with primary antibody in PBS overnight at 4 ℃. Wash in PBS (3 × 5 min), then incubate with HRP-labeled secondary antibody (50 min, RT). Wash again in PBS (3 × 5 min), add DAB solution, and monitor under a microscope until brownish-yellow color appears, then rinse with tap water. Restain nuclei with Gill’s hematoxylin (3 min), wash, differentiate, and rinse with running water. Dehydrate in 75% alcohol, 85% alcohol, anhydrous ethanol (twice), n-butanol, and xylene (each for 5 min), then seal with mounting medium. Examine under a white light microscope.The expression of STUB1 and TPIT was assessed using the H-Score method [H-SCORE = ∑ (PI × I) = (percentage of cells with weak intensity × 1) + (percentage of cells with moderate intensity × 2) + (percentage of cells with strong intensity × 3)], using QuantCenter 2.3 software. The quantification of expression in each sample was conducted in 10 randomly selected fields (at a magnification of 400×) for each case, by two impartial observers who were unaware of the participants’ medical characteristics. The information on the H-Scores (TPIT and STUB1) is shown in Supplementary Table 5.

High-throughput Nanobit drug screening

The HEK-293T cells in a 10 cm dish were transfected with 3 µg of Smibt-TPIT and 3 µg of Lgibt-STUB1 plasmids for 24 h. Then, 25 μL of media containing 5000 cells were seeded into CulturPlate-384 (PerkinElmer, 6007680) using the Multidrop Combi (Thermo Fisher Scientific). Subsequently, an equal volume of media containing the luminescent substrate furimazine (20 μM) was added to each well of the 384-well plate. The luminescence emitted by Lgbit-Smbit was quantified using the Explorer high-throughput screening platform (PerkinElmer). The drugs, including 1913 compounds from the Approved Drug Library, were then introduced at final concentrations of 1 μM. The luminescence was measured at 15, 30, 60, and 120 min. The data can be found in Supplementary Table 6. This procedure was conducted at the National Center for Translational Medicine,·Shanghai.

Surface plasmon resonance (SPR)

Surface Plasmon Resonance (SPR) binding assays were performed on a Biacore system (Cytiva) at 25 ℃. The ligand protein was immobilized onto a CM5 sensor chip via amine coupling: chip surface activation with NHS/EDC (10 μL/min, 420 s), protein coupling (30 and 50 μg/mL in pH 4.0 acetate buffer, 5 μL/min, 600 s + 300 s), and blocking with ethanolamine (10 μL/min, 420 s), achieving 8000 RU immobilization. Small molecule analytes (in PBST buffer) were injected using the LMW Kinetics method (flow rate: 30 μL/min; contact: 60 s; dissociation: 300 s) across flow cells 4 (reference) and 3 (ligand). Real-time binding responses were recorded in resonance units (RU).

Statistical analysis

The data were analyzed using GraphPad Prism version 7 (GraphPad Software, La Jolla, CA, USA) and are presented as means ± standard deviation (SD). Statistical analyses encompassed two-tailed t-tests, one-way analysis of variance (ANOVA), two-way ANOVA, and Pearson’s correlation coefficient (r). Statistical significance was denoted by p < 0.05 and is visually represented in the figures by one asterisk (p < 0.05), two asterisks (p < 0.01), or three asterisks (p < 0.001).

Results

The ubiquitin ligase STUB1 interacts with TPIT

The T-box transcription factor (TPIT) is crucial for regulating ACTH secretion in pituitary corticotroph adenomas [1, 22]. To elucidate the regulatory network of ubiquitination for TPIT, co-immunoprecipitation (Co-IP) and mass spectrometry analysis (MS) were employed to identify ubiquitin ligases and deubiquitinases associated with TPIT in HEK-293T and corticotropin cells. Ten potential ubiquitin ligases and four potential deubiquitinases were identified to interact with TPIT (Fig. 1A). The interactions between TPIT and the candidate ubiquitin ligases were further verified by Co-IP. It was found that STUB1, Tripartite motif-containing protein 28 (TRIM28), and PLRG1 interacted with TPIT, while the others did not (Fig. 1B, Supplementary Fig. 1A). Ubiquitin ligases can mediate the ubiquitination of substrates and promote their degradation through the proteasome pathway [23]. STUB1 and TRIM28 were identified as factors that promote the ubiquitination of TPIT, while PLRG1 did not exhibit this effect (Supplementary Fig. 1B). Additionally, both STUB1 and TRIM28 were found to decrease the protein levels of Flag-TPIT, with STUB1 showing greater efficacy in this regard (Supplementary Fig. 1C). Consequently, STUB1 was chosen as a potential candidate gene for further investigation. Binding between endogenous STUB1 and TPIT was observed in AtT-20 and corticotropin cells by Co-IP analysis (Fig. 1C). Subsequent NanoBiT assays using Smbit-TPIT and Lgbit-STUB1 also confirmed the interaction between STUB1 and TPIT (Fig. 1D, E). Furthermore, immunofluorescence analysis demonstrated the co-localization of TPIT and STUB1 proteins in AtT-20 cells (Fig. 1F). In vitro experiments revealed that purified His-STUB1 directly interacted with the glutathione S-transferase (GST)-TPIT fusion protein, but not with GST alone (Fig. 1G).

Fig. 1

figure 1

The ubiquitin ligase STUB1 interacts with TPIT. A Flowchart for the identification of ubiquitin ligases and deubiquitinases interacting with TPIT. TPIT and its interacting proteins were immunoprecipitated and identified by mass spectrometry. B, C Interaction of STUB1 with TPIT. External (B) and endogenous (C) Co-IP experiments were performed to detect the interaction between STUB1 and TPIT. D, E The NanoBiT assay was performed to detect the binding of STUB1 and TPIT. Data are presented as mean ± SD values. n = 3. ***p < 0.001. F Co-localization of Flag-TPIT and STUB1 in AtT20 cells. Scale bar, 5 μm. G The direct interaction between STUB1 and TPIT was determined by GST pulldown assay

Thus, these data indicated that the ubiquitin ligase STUB1 interacts directly with TPIT.

STUB1 enhances the ubiquitination of TPIT

STUB1 is an E3 ubiquitin ligase that mediates the ubiquitination of various transcription factors, including CRIP1, YAP1, HIF-1α, and c-Myc [13,14,15]. Likewise, STUB1 increased the ubiquitination of TPIT, while knockdown of STUB1 decreased TPIT ubiquitination (Fig. 2A, B, Supplementary Fig. 2A, B). STUB1 is also known to mediate ubiquitin-dependent degradation of target proteins from membranes, the cytoplasm, and the nucleus [24]. The subcellular localization of STUB1 in HEK-293T cells was predominantly observed in the cytoplasm, with some presence in the nucleus (Supplementary Fig. 2C). Subsequent nuclear-cytoplasmic fractionation experiments revealed that STUB1 facilitated TPIT ubiquitination in both cellular compartments (Fig. 2C). Additionally, in vitro ubiquitination assay also confirmed that STUB1 promoted TPIT ubiquitination (Fig. 2D). Further exploration of the STUB1-TPIT interaction region was conducted through Co-IP experiments, which indicated that STUB1 lacking the TPR domain were unable to interact with TPIT (Fig. 2E, F). NanoBiT and ubiquitination assays demonstrated that STUB1, lacking the TPR or U-box domains, was unable to ubiquitinate TPIT (Fig. 2G, H). To further elucidate the specific lysine (K) sites on TPIT targeted for ubiquitination by STUB1, we generated various TPIT mutants (3KR, 4KR, 5KR, and 12KR) as outlined in a previous study [11]. Subsequent ubiquitination assays revealed that TPIT mutants with the 3KR, 5KR, and 12KR mutations exhibited decreased ubiquitination by STUB1 (Fig. 2I-J).

Fig. 2

figure 2

STUB1 enhances the ubiquitination of TPIT. A STUB1 promoted the ubiquitination of TPIT. HEK293T cells were transfected with the indicated plasmids, and cell lysates were analyzed by ubiquitination assays. B Knockdown of STUB1 reduced TPIT ubiquitination. C TPIT ubiquitination was determined by nuclear-cytoplasmic fractionation. D In vitro ubiquitination assays showing ubiquitination of purified TPIT by purified STUB1. E Schematic showing wild-type STUB1 and its truncated forms. F STUB1 lacking the TPR domain was unable to interact with TPIT. The interaction between Flag-TPIT and Myc-tagged STUB1 or it truncated mutants was detected by Co-IP. G, H NanoBiT (G) and ubiquitination assays (H) showed that STUB1 lacking the TPR or U-box domains failed to ubiquitinate TPIT. Data are presented as mean ± SD values. n = 3. ***p < 0.001. I Schematic showing wild-type TPIT and its mutants. Mutation oflysine (K) to arginine (R). J Ubiquitination assays showed reduced ubiquitination of TPIT with the 3KR, 5KR, and 12KR mutations by STUB1

In summary, our findings suggest that STUB1 interacts with TPIT via the TPR domain and employs the U-box domain to ubiquitinate multiple sites on TPIT.

STUB1 mediates proteasomal degradation of TPIT

In order to investigate the regulatory effect of STUB1 on TPIT expression, STUB1 was overexpressed in HEK-293T and AtT-20 cells. The results demonstrated that the overexpression of STUB1 led to a decrease in TPIT protein levels in both cell lines (Fig. 3A, B, Supplementary Fig. 3A). However, the inhibition of proteasomal activity with the proteasome inhibitor MG132 reversed this reduction in TPIT protein levels induced by STUB1 in HEK293T and AtT-20 cells, suggesting that the downregulation of TPIT by STUB1 occurs through the proteasomal pathway (Fig. 3C, D). Moreover, the absence of the TPR or U-box domains in STUB1 resulted in the inability to decrease TPIT protein levels (Fig. 3E). Additionally, knockdown of Stub1 led to an increase in TPIT protein levels in AtT-20 cells (Fig. 3F, Supplementary Fig. 3B). To further investigate the role of STUB1 in regulating TPIT stability, cycloheximide (CHX) chase assays were conducted, revealing that depletion of STUB1 extended the half-life of Flag-TPIT in HEK-293T cells (Fig. 3G). Similarly, knockdown of Stub1 extended the half-life of TPIT in AtT-20 cells (Fig. 3H).

Fig. 3

figure 3

STUB1 mediates proteasomal degradation of TPIT. A, B Overexpression of STUB1 reduced the protein levels of TPIT in HEK-293T (A) and AtT-20 (B) cells. C, D Proteasome inhibitor MG132 (20 μM) prevented STUB1-mediated degradation of TPIT in HEK-293T (C) and AtT-20 (D) cells. E Immunoblotting showed that STUB1 lacking the TPR or U-box domains failed to reduce TPIT protein levels in HEK-293T cells. F Knockdown of Stub1 increased TPIT protein levels in AtT-20 cells. Stable AtT-20 cells with Stub1 knockdown were immunoblotted. G, H CHX chase assays showed that knockdown of STUB1 extend the half-life of TPIT in HEK-293T (G) and AtT-20 (H) cells. The cells were treated with CHX (100 μg/mL) for 0, 3, or 6 h. The relative ratio of TPIT/actin was determined by ImageJ. Data are presented as mean ± SD values. n.s., not significant; n = 3. ***p < 0.001

These findings collectively demonstrate that STUB1 promotes the proteasomal degradation of TPIT.

STUB1 reduces expression of POMC and secretion of ACTH

TPIT and PITX1 were identified as promoters of Pomc gene transcription [25]. Subsequent investigation into the impact of STUB1 on Pomc transcription utilized dual-luciferase reporter assays, revealing that STUB1 inhibited TPIT-mediated Pomc transcription while having no effect on PITX1-mediated Pomc transcription (Fig. 4A). Additionally, qPCR analysis demonstrated that STUB1 suppressed Pomc transcription without affecting Tpit transcription in AtT-20 cells (Fig. 4B, Supplementary Fig. 3C). In order to examine the impact of STUB1 on POMC protein levels, immunoblotting was conducted, revealing that STUB1 decreased the levels of POMC protein, with this effect being reversed by the proteasomal inhibitor MG132 (Fig. 4C, D). Additionally, knockdown of Stub1 increased Pomc transcription without affecting Tpit transcription in AtT-20 cells (Fig. 4E, Supplementary Fig. 3D). Knockdown of Stub1 resulted in elevated POMC protein expression in AtT-20 cells (Fig. 4F). Additionally, the impact of STUB1 on proliferation in AtT-20 cells was investigated, revealing that overexpression of STUB1 suppressed cell proliferation, whereas knockdown of Stub1 enhanced cell proliferation (Supplementary Fig. 4A, B). Importantly, the effects of STUB1 on the ACTH secretion were then evaluated using enzyme-linked immunosorbent assay (ELISA), showing that overexpression of STUB1 inhibited the secretion of ACTH, while Stub1 knockdown increased the secretion of ACTH in AtT-20 cells (Fig. 4G, H).

Fig. 4

figure 4

STUB1 reduces the expression of POMC and secretion of ACTH. A Dual-luciferase reporter assay demonstrated that STUB1 suppressed the transcription of Pomc mediated by TPIT. Data are presented as mean ± SD values. n.s., not significant; n = 3. ***p < 0.001. B qPCR showed that STUB1 suppressed the transcription of Pomc in AtT-20 cells. Data are presented as mean ± SD values. n = 3. ***p < 0.001. C, D STUB1 reduced POMC protein levels (C) with this effect reversed by the proteasomal inhibitor MG132 (D) in AtT-20 cells. E, F Knockdown of Stub1 enhanced the mRNA (E) and protein (F) expression of POMC in AtT-20 cells. Data are presented as mean ± SD values. ***p < 0.001. G, H STUB1 regulated the secretion of ACTH in AtT-20 cells. The secretion of ACTH from stable AtT-20 cells with Stub1 overexpression (G) or knockdown (H) was measured by ELISA. Data are presented as mean ± SD values. n = 3. ***p < 0.001

Collectively, these data indicated that STUB1 downregulates the expression of POMC and inhibits the secretion of ACTH.

STUB1 inhibits TPIT and POMC expression, as well as ACTH secretion in vivo

To investigate the impact of STUB1 on TPIT, POMC, and ACTH in vivo, STUB1-overexpressing or control AtT-20 cells were implanted into nude mice. The results showed that STUB1 overexpression significantly inhibited xenograft tumor growth compared to the control group (Fig. 5A, B). Furthermore, immunoblotting analysis revealed that STUB1 overexpression led to a significant reduction in the protein levels of TPIT and POMC (Fig. 5C). Consistently, immunohistochemistry analysis demonstrated that STUB1 overexpression resulted in reduced levels of POMC (Fig. 5D, E). Most importantly, STUB1 overexpression significantly inhibited the secretion of ACTH in vivo (Fig. 5F).

Fig. 5

figure 5

STUB1 inhibits TPIT and POMC expression, as well as ACTH secretion in vivo. A, B STUB1 overexpression inhibited tumor growth in vivo. The TRIM21-overexpression or control AtT-20 cells were implanted subcutaneously in nude mice. Xenograft tumors were imaged (A), measured (A) and weighed (B). Data are presented as mean ± SD values. n = 5. ***p < 0.001. (C) STUB1 overexpression inhibited TPIT and POMC expression in vivo. The STUB1-overexpressing and control xenograft tumors were harvested and immunoblotted. D, E Immunohistochemistry of xenograft tumors. Representative IHC images of proteins in STUB1-overexpressing and control xenograft tumors. Scale bar, 50 μm. Semi-quantitative immunohistochemical analysis of proteins. Data are presented as mean ± SD values. F STUB1 inhibited the secretion of ACTH in vivo. The secretion of ACTH from nude mice with xenograft tumors was measured by ELISA. Data are presented as mean ± SD values. n = 5. ***p < 0.001

Taken together, these findings suggest that STUB1 acts to inhibit cell proliferation, decrease the expression of TPIT and POMC, and reduce ACTH secretion in vivo.

STUB1 is decreased in ACTH-secreting corticotroph adenoma and negatively correlated with TPIT protein

To explore the role of STUB1 in the corticotroph pituitary, its mRNA expression in human normal pituitary and corticotroph adenomas was analyzed. Compared to normal pituitary glands, the expression of STUB1 was found to be downregulated in corticotroph adenomas (Supplementary Fig. 5A). Additionally, corticotroph adenomas are divided into ACTH-secreting corticotroph adenomas and silent corticotroph adenomas (SCAs) [2]. SCAs demonstrate immunopositivity for ACTH in the absence of biochemical and clinical signs of hypercortisolism [26]. Compared to SCAs, STUB1 was downregulated in ACTH-secreting corticotroph adenomas (Fig. 6A). Additionally, the relationships between the expression of STUB1, TPIT, and POMC were then analyzed. The mRNA expression of STUB1 was observed to be negatively correlated with the mRNA expression of POMC, while no significant association was seen with the mRNA levels of TPIT in 56 corticotroph adenomas (Fig. 6B, Supplementary Fig. 5B). Furthermore, immunohistochemistry (IHC) showed that nuclear-localized STUB1 was downregulated in ACTH-secreting corticotroph adenomas, accompanied by increased levels of TPIT, POMC, and ACTH compared to SCAs (Fig. 6C). Then a semi-quantitative immunohistochemical analysis of nuclear-localized STUB1 was performed, demonstrating that STUB1 protein levels were also downregulated in ACTH-secreting corticotroph adenoma compared to SCAs (Fig. 6D). The nuclear-localized STUB1 was negatively correlated with the mRNA expression of POMC, but not significantly associated with TPIT mRNA levels in corticotroph adenomas (Fig. 6E, Supplementary Fig. 5C). The relationship between the protein expression of STUB1 and TPIT was then analyzed. Immunohistochemical analysis showed that nuclear-localized STUB1 was negatively correlated with TPIT protein, which was verified by immunoblotting (Fig. 6F, G).

Fig. 6

figure 6

STUB1 is decreased in ACTH-secreting corticotroph adenoma and negatively correlated with TPIT protein. A STUB1 expression was decreased in ACTH-secreting corticotroph adenoma. The expression of STUB1 in functioning (n = 9) and silent (n = 47) corticotroph adenomas was analyzed by RNA-seq. Data are presented as mean ± SD values. **p < 0.01. B Correlation between mRNA expression of STUB1 and POMC in corticotroph adenomas. C Representative immunohistochemical images showing expression of STUB1, TPIT, POMC, and ACTH in functioning and silent corticotroph adenomas. D Semi-quantitative immunohistochemical analysis of nuclear-localized STUB1 in functioning and silent corticotroph adenomas. E Correlation between nuclear-localized STUB1 protein and mRNA levels of POMC in corticotroph adenomas. F, G Correlation between STUB1 and TPIT protein levels in corticotroph adenomas, shown by immunohistochemistry (F) and immunoblotting (G)

Taken together, the findings indicate that STUB1 is decreased in ACTH-secreting corticotroph adenomas and is negatively correlated with POMC expression and TPIT protein.

Irbesartan and Lumiracoxib enhance the interaction between STUB1 and TPIT, reducing POMC expression and ACTH secretion

Enhancing the interaction between STUB1 and TPIT has the potential to increase TPIT ubiquitination and degradation, offering a potential therapeutic approach for CD. To investigate potential drug candidates that can enhance this interaction, a NanoBiT system was developed to evaluate the binding between Lgbit-STUB1 and Smbit-TPIT, followed by a drug screening process involving 1913 FDA-approved compounds (Fig. 7A, B). A number of drugs were found to enhance or reduce the interaction between STUB1 and TPIT, and the top 10 enhancing drugs and 2 reducing drugs were selected for further verification (Fig. 7C). Co-IP assays showed that Irbesartan, Bosentan, and Lumiracoxib enhanced the interaction between STUB1 and TPIT (Fig. 7D). Moreover, Irbesartan and Lumiracoxib promoted the ubiquitination of TPIT, but Bosentan failed (Fig. 6E). Further investigation into the impact of Irbesartan and Lumiracoxib on TPIT, POMC, and ACTH levels revealed that both compounds decreased TPIT protein levels, which was reversed by the proteasomal inhibitor MG132 (Fig. 7F). Additionally, Irbesartan and Lumiracoxib suppressed Pomc transcription but did not affect Tpit and Stub1 transcription in AtT-20 cells (Fig. 7G, Supplementary Fig. 6A, B). Furthermore, both Irbesartan and Lumiracoxib demonstrated a reduction in TPIT and POMC protein levels, while showing no impact on STUB1 (Fig. 7H). Subsequent evaluation of ACTH secretion using ELISA indicated that both Irbesartan and Lumiracoxib significantly suppressed ACTH secretion in AtT-20 cells. However, the knockdown of STUB1 partially mitigated this inhibitory effect (Fig. 7I). Cell toxicity analysis confirmed the non-toxic nature of Irbesartan and Lumiracoxib on AtT-20 cells (Supplementary Fig. 6A, B), suggesting that the inhibition of ACTH secretion by these compounds was not due to cell death. Furthermore, evidence suggests that Irbesartan and Lumiracoxib have the capability to directly bind to the STUB1 protein (Supplementary Fig. 8A, B). Moreover, these compounds have the potential to augment the efficacy of established therapies, such as Pasireotide (Supplementary Fig. 9A).

Fig. 7

figure 7

Irbesartan and Lumiracoxib enhance the interaction between STUB1 and TPIT, reducing POMC expression and ACTH secretion. A Schematic showing drug screening using the NanoBiT system. The NanoBiT assay was used to detect the interaction between Lgbit-STUB1 and Smbit-TPIT. A total of 1913 FDA-approved drugs were screened. B Interaction between STUB1 and TPIT. C Heatmap of candidate drugs influencing the interaction between STUB1 and TPIT. D The effects of Irbesartan, Bosentan, and Lumiracoxib on the STUB1-TPIT interaction were assessed by coimmunoprecipitation. E The effects of Irbesartan, Bosentan, and Lumiracoxib on TPIT ubiquitination were assessed by ubiquitination assays in HEK-293T cells. F Irbesartan and Lumiracoxib both reduced the protein levels of TPIT in AtT-20 cells while the proteasomal inhibitor MG132 (20 μM) reversed this effect. G Irbesartan and Lumiracoxib both suppressed Pomc in AtT-20 cells. Cells were treated with Irbesartan (1 μM) and Lumiracoxib (1 μM) and Pomc expression was measured by qPCR. Data are presented as mean ± SD values. n = 3. ***p < 0.001. H AtT-20 cells were treated with Irbesartan (1 μM) and Lumiracoxib (1 μM), and immunoblotted. I Irbesartan and Lumiracoxib inhibited the secretion of ACTH in AtT-20 cells, while knockdown of STUB1 partly reversed this effect. AtT-20 cells were treated with Irbesartan and Lumiracoxib. ACTH levels were measured by ELISA. Data are presented as mean ± SD values. n = 3. *p < 0.05, ***p < 0.001

Collectively, these findings indicate that Irbesartan and Lumiracoxib facilitate the interaction between STUB1 and TPIT, leading to the degradation of TPIT, thereby decreasing POMC expression and ACTH secretion. Consequently, Irbesartan and Lumiracoxib hold promise as potential therapeutic agents for CD.

Discussion

Ubiquitination of proteins plays a key role in regulating the stability and activity of various tumor-related proteins [5]. TPIT, as the main transcriptional activator of the POMC gene, is crucial for the occurrence and development of CD [27]. Our previous study identified the ubiquitin ligase TRIM65 for TPIT through a yeast two-hybrid technique [9]. This study utilized Co-IP and mass spectrometry techniques to investigate potential ubiquitin ligases and deubiquitinases for TPIT and identified two novel E3 ligases, STUB1 and TRIM28, which interact with TPIT to regulate its degradation. Further experiments found that STUB1 suppresses ACTH secretion by degrading TPIT and inhibits cell proliferation. Additionally, STUB1 was decreased in ACTH-secreting corticotroph adenoma compared to SCAs. STUB1 may become a potential therapeutic target for managing CD.

It is important to acknowledge a potential limitation regarding the transcriptomic data. The observed imbalance in cohort sizes between functional corticotroph adenomas and SCAs in our RNA-seq analysis may introduce bias into the identification of differential signatures. Larger, balanced multi-centre collections will be essential in future studies to confirm the generalizability and robustness of the identified transcriptomic profiles distinguishing these adenoma subtypes.

STUB1 is recognized as a tumor suppressor in liver, prostate, lung, and gastric cancers through the degradation of key oncogenic proteins, such as JMJD1A, YAP1, AR1, and YTHDF2 [13, 14, 28,29,30]. STUB1 is mainly localized cytoplasm and highly expressed in skeletal muscle, heart, pancreatic, and brain tissue [11, 31]. Notably, STUB1 possesses the ability to target cell surface, cytoplasmic, nuclear, or secreted proteins for ubiquitin-dependent degradation [24]. Studies have indicated that the loss of nuclear, but not cytoplasmic, STUB1 is associated with increased tumor aggressiveness and reduced survival in patients with breast cancer [32]. TPIT is mainly localized in the nucleus, and the staining in the current study revealed that STUB1 is expressed and localized in both the cytoplasm and nucleus. This study found that the nuclear-localized STUB1 was upregulated in SCAs and negatively correlated with the expression of POMC. JG98, a novel allosteric HSP70 inhibitor, and inflammation-associated stimulation can promote the nuclear translocation of STUB1 [33, 34]. In the future, drugs and immunotherapy could be explored to promote the nuclear translocation of STUB1, thereby enhancing its ubiquitination and degradation of TPIT.

The tumor size is usually that of a microadenoma, and there is no space-occupying effect in CD. Its harm is mainly the result of changes in the peripheral system caused by excessive secretion of ACTH. Therefore, we propose the therapeutic strategy of targeting the specific transcription factor TPIT, leading to inhibition of POMC transcription and reduced ACTH secretion, and thus controlling the harmful effects of CD. Pharmacological treatment of CD is usually aimed at suppressing cortisol production or blocking its action. Commonly used medications include dopamine agonists, somatostatin analogs, mifepristone, ketoconazole, and metopirone [35]. It is important to note that pharmacological treatment is usually long-term and may be associated with side effects, such as hypertension, edema, and electrolyte imbalances [36]. In contrast, the present study screened ubiquitin ligases to regulate the levels of TPIT, thus influencing the expression of POMC and the secretion of ACTH. Furthermore, two drugs approved by the Food and Drug Administration [37], Irbesartan and Lumiracoxib, were identified to promote the interaction between STUB1 and TPIT, leading to the degradation of TPIT, thereby downregulating POMC expression and ACTH secretion. While Irbesartan and Lumiracoxib show promise in promoting STUB1-mediated TPIT degradation, it is crucial to consider their known primary targets (e.g., angiotensin II receptor type 1 and COX-2, respectively) [38,39,40]. Our observation that STUB1 knockdown only partially reversed the inhibitory effect of these drugs on ACTH secretion suggests they may also act through additional, STUB1-independent pathways to suppress ACTH. The precise nature of these alternative mechanisms requires further exploration in future studies.

Irbesartan is an antihypertensive drug commonly used in clinical practice, while Lumiracoxib is a selective COX-2 inhibitor. Irbesartan has been studied and found to have no adverse effects on cancer risk, while also demonstrating promising potential in overcoming drug resistance in pancreatic cancer and activating the immune system in patients with colorectal cancer [38,39,40]. Lumiracoxib has demonstrated antiproliferative effects on human non-small cell lung cancer cell lines and may have the potential to reduce the risk of skin cancer [41, 42]. However, it is important to note that Lumiracoxib, along with several other drugs, has been withdrawn from the market due to its potential to cause liver damage [43]. Therefore, further investigation from a structural biology perspective is warranted, together with in vivo studies on animal models to evaluate the efficacy of these drugs and lay the groundwork for pre-clinical trials. Beyond its role in suppressing ACTH secretion via TPIT degradation, our data also demonstrates that STUB1 inhibits corticotroph tumor cell proliferation. STUB1 is likely to suppress AtT-20 growth through targets other than TPIT. Identifying additional STUB1 substrates relevant to corticotroph tumor growth is an important avenue for future research. Furthermore, the use of the proteolysis-targeting chimera (PROTAC) technique to degrade proteins crucial for tumor development has emerged as a potential cancer treatment strategy [44, 45]. TPIT as a pituitary-specific transcription factor and the identification of its ubiquitin ligases provides the foundation for the development of PROTAC drugs.

In summary, as TPIT is a key regulatory factor in the occurrence and development of CD, regulation of its stability and activity through the ubiquitination pathway may provide new targets for tumor treatment. The study discovered a novel mechanism of STUB1-mediated TPIT ubiquitination and degradation and preliminarily verified the therapeutic potential of targeting the STUB1/TPIT interaction. In addition, exploring the use of immunotherapy and other means to promote the redistribution of STUB1 in cells and enhance its ubiquitination of TPIT is also a novel direction worthy of attention. In the future, in-depth research on the ubiquitination-mediated regulation of TPIT and the development of more efficient and specific small-molecule regulators will provide new avenues for the precise treatment of CD.

Conclusion

In this study, we discovered that a novel E3 ubiquitin ligase STUB1 interacts with transcription factor TPIT, leading to proteasomal degradation of TPIT. Moreover, STUB1 could inhibit POMC expression, ATCH secretion and cell proliferation in AtT-20 cells. Additionally, STUB1 was decreased in ACTH-secreting corticotroph adenoma compared to SCAs and negatively correlated with TPIT protein, as well as expression of POMC. Furthermore, drug screening identified the Irbesartan and Lumiracoxib promoted the interaction between STUB1 and TPIT, leading to proteasomal degradation of TPIT and the suppression of ACTH secretion.

Data availability

The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive [46] in National Genomics Data Center [47], China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA005096) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa-human.

References

  1. Lamolet B, Pulichino AM, Lamonerie T, Gauthier Y, Brue T, Enjalbert A, Drouin J. A pituitary cell-restricted T box factor, Tpit, activates POMC transcription in cooperation with Pitx homeoproteins. Cell. 2001;104:849–59.

    CAS PubMed Google Scholar

  2. Melmed S, Kaiser UB, Lopes MB, Bertherat J, Syro LV, Raverot G, Reincke M, Johannsson G, Beckers A, Fleseriu M, et al. Clinical biology of the pituitary adenoma. Endocr Rev. 2022;43:1003–37.

    PubMed PubMed Central Google Scholar

  3. Melmed S. Pituitary-tumor endocrinopathies. N Engl J Med. 2020;382:937–50.

    CAS PubMed Google Scholar

  4. Liu F, Chen J, Li K, Li H, Zhu Y, Zhai Y, Lu B, Fan Y, Liu Z, Chen X, et al. Ubiquitination and deubiquitination in cancer: from mechanisms to novel therapeutic approaches. Mol Cancer. 2024;23:148.

    PubMed PubMed Central Google Scholar

  5. Deng L, Meng T, Chen L, Wei W, Wang P. The role of ubiquitination in tumorigenesis and targeted drug discovery. Signal Transduct Target Ther. 2020;5:11.

    CAS PubMed PubMed Central Google Scholar

  6. Reincke M, Sbiera S, Hayakawa A, Theodoropoulou M, Osswald A, Beuschlein F, Meitinger T, Mizuno-Yamasaki E, Kawaguchi K, Saeki Y, et al. Mutations in the deubiquitinase gene USP8 cause Cushing’s disease. Nat Genet. 2015;47:31–8.

    CAS PubMed Google Scholar

  7. Ma ZY, Song ZJ, Chen JH, Wang YF, Li SQ, Zhou LF, Mao Y, Li YM, Hu RG, Zhang ZY, et al. Recurrent gain-of-function USP8 mutations in Cushing’s disease. Cell Res. 2015;25:306–17.

    CAS PubMed PubMed Central Google Scholar

  8. Chen J, Jian X, Deng S, Ma Z, Shou X, Shen Y, Zhang Q, Song Z, Li Z, Peng H, et al. Identification of recurrent USP48 and BRAF mutations in Cushing’s disease. Nat Commun. 2018;9: 3171.

    PubMed PubMed Central Google Scholar

  9. Yao H, Xie W, Dai Y, Liu Y, Gu W, Li J, Wu L, Xie J, Rui W, Ren B, et al. TRIM65 determines the fate of a novel subtype of pituitary neuroendocrine tumors via ubiquitination and degradation of TPIT. Neuro Oncol. 2022;24:1286–97.

    CAS PubMed PubMed Central Google Scholar

  10. Zhang T, Liu Y, Liu F, Guo K, Tang R, Ye J, Xue L, Su Z, Wu ZB. X-linked ubiquitin-specific peptidase 11 (USP11) increases susceptibility to Cushing’s disease in women. Acta Neuropathol Commun. 2025;13:22.

    CAS PubMed PubMed Central Google Scholar

  11. Ballinger CA, Connell P, Wu Y, Hu Z, Thompson LJ, Yin LY, Patterson C. Identification of CHIP, a novel tetratricopeptide repeat-containing protein that interacts with heat shock proteins and negatively regulates chaperone functions. Mol Cell Biol. 1999;19:4535–45.

    CAS PubMed PubMed Central Google Scholar

  12. Paul I, Ghosh MK. The E3 ligase CHIP: insights into its structure and regulation. Biomed Res Int. 2014;2014: 918183.

    PubMed PubMed Central Google Scholar

  13. Wang J, Zhou Y, Zhang D, Zhao W, Lu Y, Liu C, Lin W, Zhang Y, Chen K, Wang H, Zhao L. CRIP1 suppresses BBOX1-mediated carnitine metabolism to promote stemness in hepatocellular carcinoma. EMBO J. 2022;41: e110218.

    CAS PubMed PubMed Central Google Scholar

  14. Tang DE, Dai Y, Lin LW, Xu Y, Liu DZ, Hong XP, Jiang HW, Xu SH. STUB1 suppresseses tumorigenesis and chemoresistance through antagonizing YAP1 signaling. Cancer Sci. 2019;110:3145–56.

    CAS PubMed PubMed Central Google Scholar

  15. Wang Y, Ren F, Wang Y, Feng Y, Wang D, Jia B, Qiu Y, Wang S, Yu J, Sung JJ, et al. CHIP/Stub1 functions as a tumor suppressor and represses NF-κB-mediated signaling in colorectal cancer. Carcinogenesis. 2014;35:983–91.

    PubMed Google Scholar

  16. Liu F, Ying J, Yang K, Xiong X, Yang N, Wang S, Zhao W, Zhu H, Yu M, Wu J, et al. Deciphering the regulatory mechanisms and biological implications of ARID1A missense mutations in cancer. Cell Rep. 2024;43: 114916.

    CAS PubMed Google Scholar

  17. Koochaki SHJ, Słabicki M, Lumpkin R, Zou C, Belizaire R, Fischer ES, Ebert BL. A STUB1 ubiquitin ligase/CHIC2 protein complex negatively regulates the IL-3, IL-5, and GM-CSF cytokine receptor common β chain (CSF2RB) protein stability. J Biol Chem. 2022;298: 102484.

    CAS PubMed PubMed Central Google Scholar

  18. Jang KW, Lee KH, Kim SH, Jin T, Choi EY, Jeon HJ, Kim E, Han YS, Chung JH. Ubiquitin ligase CHIP induces TRAF2 proteasomal degradation and NF-κB inactivation to regulate breast cancer cell invasion. J Cell Biochem. 2011;112:3612–20.

    CAS PubMed Google Scholar

  19. Apriamashvili G, Vredevoogd DW, Krijgsman O, Bleijerveld OB, Ligtenberg MA, de Bruijn B, Boshuizen J, Traets JJH, D’Empaire Altimari D, van Vliet A, et al. Ubiquitin ligase STUB1 destabilizes IFNγ-receptor complex to suppress tumor IFNγ signaling. Nat Commun. 2022;13: 1923.

    CAS PubMed PubMed Central Google Scholar

  20. Ng S, Lim S, Sim ACN, Mangadu R, Lau A, Zhang C, Martinez SB, Chandramohan A, Lim UM, Ho SSW, et al. STUB1 is an intracellular checkpoint for interferon gamma sensing. Sci Rep. 2022;12: 14087.

    CAS PubMed PubMed Central Google Scholar

  21. Liu YT, Liu F, Cao L, Xue L, Gu WT, Zheng YZ, Tang H, Wang Y, Yao H, Zhang Y, et al. The KBTBD6/7-DRD2 axis regulates pituitary adenoma sensitivity to dopamine agonist treatment. Acta Neuropathol. 2020;140:377–96.

    CAS PubMed Google Scholar

  22. Asa SL, Faiman GH, Mohamed A, Mete O. Multilineage pituitary neuroendocrine tumors expressing TPIT and SF1: a clinicopathological series of six tumors. Endocr Pathol. 2024;35:349–53.

    CAS PubMed Google Scholar

  23. Swatek KN, Komander D. Ubiquitin modifications. Cell Res. 2016;26:399–422.

    CAS PubMed PubMed Central Google Scholar

  24. Paul I, Ghosh MK. A CHIPotle in physiology and disease. Int J Biochem Cell Biol. 2015;58:37–52.

    CAS PubMed Google Scholar

  25. Drouin J. 60 YEARS OF POMC: Transcriptional and epigenetic regulation of POMC gene expression. J Mol Endocrinol. 2016;56:T99-t112.

    CAS PubMed Google Scholar

  26. Drummond J, Roncaroli F, Grossman AB, Korbonits M. Clinical and pathological aspects of silent pituitary adenomas. J Clin Endocrinol Metab. 2019;104:2473–89.

    PubMed Google Scholar

  27. Drouin J. The corticotroph cells from early development to tumorigenesis. J Neuroendocrinol. 2022;34: e13147.

    CAS PubMed Google Scholar

  28. Xu S, Fan L, Jeon HY, Zhang F, Cui X, Mickle MB, Peng G, Hussain A, Fazli L, Gleave ME, et al. P300-mediated acetylation of histone demethylase JMJD1A prevents its degradation by ubiquitin ligase STUB1 and enhances its activity in prostate cancer. Cancer Res. 2020;80:3074–87.

    CAS PubMed PubMed Central Google Scholar

  29. Liao Y, Liu Y, Yu C, Lei Q, Cheng J, Kong W, Yu Y, Zhuang X, Sun W, Yin S, et al. HSP90β impedes STUB1-induced ubiquitination of YTHDF2 to drive sorafenib resistance in hepatocellular carcinoma. Adv Sci. 2023;10: e2302025.

    Google Scholar

  30. Zhang P, Li C, Li H, Yuan L, Dai H, Peng Z, Deng Z, Chang Z, Cui CP, Zhang L. Ubiquitin ligase CHIP regulates OTUD3 stability and suppresses tumour metastasis in lung cancer. Cell Death Differ. 2020;27:3177–95.

    CAS PubMed PubMed Central Google Scholar

  31. Jiang B, Shen H, Chen Z, Yin L, Zan L, Rui L. Carboxyl terminus of HSC70-interacting protein (CHIP) down-regulates NF-κB-inducing kinase (NIK) and suppresses NIK-induced liver injury. J Biol Chem. 2015;290:11704–14.

    CAS PubMed PubMed Central Google Scholar

  32. Luan H, Mohapatra B, Bielecki TA, Mushtaq I, Mirza S, Jennings TA, Clubb RJ, An W, Ahmed D, El-Ansari R, et al. Loss of the nuclear pool of ubiquitin ligase CHIP/STUB1 in breast cancer unleashes the MZF1-cathepsin pro-oncogenic program. Cancer Res. 2018;78:2524–35.

    CAS PubMed PubMed Central Google Scholar

  33. Xu P, Yang JC, Ning S, Chen B, Nip C, Wei Q, Liu L, Johnson OT, Gao AC, Gestwicki JE, et al. Allosteric inhibition of HSP70 in collaboration with STUB1 augments enzalutamide efficacy in antiandrogen resistant prostate tumor and patient-derived models. Pharmacol Res. 2023;189: 106692.

    CAS PubMed PubMed Central Google Scholar

  34. Chen Z, Barbi J, Bu S, Yang HY, Li Z, Gao Y, Jinasena D, Fu J, Lin F, Chen C, et al. The ubiquitin ligase Stub1 negatively modulates regulatory T cell suppressive activity by promoting degradation of the transcription factor Foxp3. Immunity. 2013;39:272–85.

    CAS PubMed Google Scholar

  35. Fleseriu M, Auchus R, Bancos I, Ben-Shlomo A, Bertherat J, Biermasz NR, Boguszewski CL, Bronstein MD, Buchfelder M, Carmichael JD, et al. Consensus on diagnosis and management of Cushing’s disease: a guideline update. Lancet Diabetes Endocrinol. 2021;9:847–75.

    PubMed PubMed Central Google Scholar

  36. Feelders RA, Newell-Price J, Pivonello R, Nieman LK, Hofland LJ, Lacroix A. Advances in the medical treatment of Cushing’s syndrome. Lancet Diabetes Endocrinol. 2019;7:300–12.

    PubMed Google Scholar

  37. Safdari HA, Pandey S, Shukla AK, Dutta S. Illuminating GPCR signaling by cryo-EM. Trends Cell Biol. 2018;28:591–4.

    CAS PubMed Google Scholar

  38. Zhou T, Xie Y, Hou X, Bai W, Li X, Liu Z, Man Q, Sun J, Fu D, Yan J, et al. Irbesartan overcomes gemcitabine resistance in pancreatic cancer by suppressing stemness and iron metabolism via inhibition of the Hippo/YAP1/c-Jun axis. J Exp Clin Cancer Res. 2023;42:111.

    CAS PubMed PubMed Central Google Scholar

  39. Titmuss E, Milne K, Jones MR, Ng T, Topham JT, Brown SD, Schaeffer DF, Kalloger S, Wilson D, Corbett RD, et al. Immune activation following Irbesartan treatment in a colorectal cancer patient: a case study. Int J Mol Sci. 2023. https://doi.org/10.3390/ijms24065869.

    PubMed PubMed Central Google Scholar

  40. Effects of telmisartan, irbesartan, valsartan, candesartan, and losartan on cancers in 15 trials enrolling 138,769 individuals. J Hypertens 2011, 29:623-635.

  41. Hao JQ, Li Q, Xu SP, Shen YX, Sun GY. Effect of lumiracoxib on proliferation and apoptosis of human nonsmall cell lung cancer cells in vitro. Chin Med J (Engl). 2008;121:602–7.

    CAS PubMed Google Scholar

  42. Ma Y, Yu P, Lin S, Li Q, Fang Z, Huang Z. The association between nonsteroidal anti-inflammatory drugs and skin cancer: different responses in American and European populations. Pharmacol Res. 2020;152: 104499.

    CAS PubMed Google Scholar

  43. Singh D, Khan MA, Siddique HR. Unveiling the therapeutic promise of natural products in alleviating drug-induced liver injury: present advancements and future prospects. Phytother Res. 2024;38:22–41.

    PubMed Google Scholar

  44. Békés M, Langley DR, Crews CM. Protac targeted protein degraders: the past is prologue. Nat Rev Drug Discov. 2022;21:181–200.

    PubMed PubMed Central Google Scholar

  45. Kelm JM, Pandey DS, Malin E, Kansou H, Arora S, Kumar R, Gavande NS. Protac’ing oncoproteins: targeted protein degradation for cancer therapy. Mol Cancer. 2023;22: 62.

    PubMed PubMed Central Google Scholar

  46. Chen T, Chen X, Zhang S, Zhu J, Tang B, Wang A, Dong L, Zhang Z, Yu C, Sun Y, et al. The genome sequence archive family: toward explosive data growth and diverse data types. Genom Proteom Bioinform. 2021;19:578–83.

    Google Scholar

  47. Database Resources of the National Genomics Data Center. China National Center for Bioinformation in 2022. Nucleic Acids Res. 2022;50:D27-d38.

    Google Scholar

Download references

Acknowledgements

We extend our heartfelt thanks to the dedicated team at the Central Laboratory of Ruijin Hospital and the high-throughput screening platform of the National Research Center for Translational Medicine (Shanghai) for their invaluable assistance in the library screening process.

Funding

This research was generously supported by the Postdoctoral Fellowship Program of CPSF under Grant Number GZC20251410 (F.L.). Additional funding was provided by the National Natural Science Foundation of China through grants 82002627 (Y.L.), 82472640 (Z.B.W.), 82373131 (S.L.), and 82172605 (L.X.). Additional support was provided by the Academic Star program of Ruijin Hospital School of Medicine, Shanghai Jiao Tong University, under grant number 2024XS034 (Y.L.), the Fundamental Research Funds for Central Universities (No.YG2023ZD06), the Health Care Leader program of the Shanghai Municipal Health Commission (No.2022LJ006) (Z.B.W.), the National Research Center for Translational Medicine with grants NRCTM (SH) 2023-15 (Z.B.W.), and the Qiusuo Funds for Distinguished Young Scholars (L.X.). This project is supported by open project research of the first affiliated hospital of henan university (grant number KFZD25003).

Author information

Author notes

  1. Fang Liu, Yanting Liu and Tao Zhang contributed equally to this work.

Authors and Affiliations

  1. Department of Neurosurgery, Center of Pituitary Tumor, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, 200025, People’s Republic of ChinaFang Liu, Yanting Liu, Tao Zhang, Ning Huang, Desheng Chen, Xiaobin Wang, Li Xue, Shaojian Lin & Zhe Bao Wu
  2. Department of Neurosurgery, First Affiliated Hospital of Wenzhou Medical University, Wenzhou, 325000, People’s Republic of ChinaYanting Liu, Tao Zhang, Ning Huang, Shaojian Lin & Zhe Bao Wu
  3. Department of Neurosurgery, Center for Immune-Related Diseases, Shanghai Institute of Immunology, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, 200025, People’s Republic of ChinaLi Xue, Shaojian Lin & Zhe Bao Wu
  4. The First Affiliated Hospital, Henan University, Kaifeng, 475004, People’s Republic of ChinaXiaobin Wang, Jiangong Ma & Zhe Bao Wu
  5. Department of Biochemistry and Molecular Cell Biology, State Key Laboratory of Oncogenes and Related Genes, Shanghai Key Laboratory for Tumor Microenvironment and Inflammation, School of Medicine, Shanghai Jiao Tong University, Shanghai, 200025, People’s Republic of ChinaFang Liu
  6. Shanghai Center for Brain Science and Brain-Inspired Technology, Shanghai, 200022, People’s Republic of ChinaLi Xue

Contributions

F.L., Y.L. and T.Z. performed the experiments, analysed the data, and co-wrote the manuscript. N.H. and D.C. performed CO-IP and immunoblotting experiments. L.X. performed NanoBit assay. X.W., S.L. and J.M. performed and analyzed the IHC experiments. Z.B.W. conceived the idea, designed and supervised the study, analysed the data and co-wrote the manuscript.

Corresponding author

Correspondence to Zhe Bao Wu.

Ethics declarations

Ethics approval and consent to participate

Informed consent was obtained from all participants, and approval was granted by the Ethics Committee of the Ruijin Hospital, affiliated with Shanghai Jiao Tong University School of Medicine (2019-39). All animal experiments were conducted in strict accordance with the guidelines for the care and use of laboratory animals, and ethical approval was obtained from the relevant institutional review board (RJ2023009).

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1 of The STUB1-TPIT axis regulates the secretion of adrenocorticotrophic hormone in cushing disease

Supplementary Materials
Supplementary Figure 1 STUB1, TRIM28, and PLRG1 interact with TPIT. (A)
Interactions between TPIT and candidate ubiquitin ligases in HEK-293T cells were
assessed by coimmuniprecipitation. Except for the HUWE1 protein carrying a FLAG
tag, all other co-IP experiments employed TPIT fused to FLAG. FLAG-tagged
proteins were consistently used as the bait to assess protein interactions. (B) STUB1
and TRIM28 promoted the ubiquitination of TPIT. HEK-293T cells were transfected
with indicated plasmids followed by ubiquitination assays. (C) STUB1 and TRIM28
both reduced the protein levels of Flag-TPIT in HEK-293T cells. Stably transfected
HEK-293T cells expressing Flag-TPIT were transfected with the indicated plasmids
before immunoblotting.
Supplementary Figure 2 STUB1 is mainly localized in the cytoplasm in HEK-
293T cells. (A-B) qPCR validation of STUB1 overexpression or knockdown
efficiency in HEK293T. Data are presented as mean ± SD values. n = 3. ***p <
0.001. (C) Localization of Myc-STUB1 in HEK-293T cells. Scale bar, 5 μm.
Supplementary Figure 3 Stub1 expression has no effect on the mRNA expression
of Tpit. (A-B) qPCR validation of STUB1 overexpression or knockdown efficiency in
AtT-20. Data are presented as mean ± SD values. n = 3. ***p < 0.001. (C-D) qPCR
showed that overexpression (A) or knockdown (B) of Stub1 had no effect on the
mRNA expression of Tpit in AtT-20 cells. Data are presented as mean ± SD values. n
= 3. n.s., not significant.
Supplementary Figure 4 STUB1 regulates cell proliferation in AtT-20 cells. (A-B)
Stub1 inhibited proliferation of AtT20 cells. Stably transfected AtT-20 cells with
Stub1 overexpression (A) or knockdown (B) were assessed by luminescent cell

 

viability assays. Data are presented as mean ± SD values. n = 3. ***p < 0.001.
Supplementary Figure 5 STUB1 is downregulated in corticotroph adenomas. (A)
STUB1 is downregulated in corticotroph adenomas. Bulk RNA sequencing was
conducted on corticotroph adenomas (n = 56) compared to normal pituitary tissues (n
= 107). Data are presented as mean ± SD values. ***p < 0.001. (B) Correlation
between the mRNA expression of STUB1 and TPIT in corticotroph adenomas. (C)
Correlation between nuclear-localized STUB1 and mRNA expression of TPIT in
corticotroph adenomas.
Supplementary Figure 6 Irbesartan and Lumiracoxib have no effect on mRNA
expression of Tpit and Stub1. (A-B) qPCR showed that Irbesartan and Lumiracoxib
had no effect on mRNA expression of Tpit (A) and Stub1 (B) in AtT-20 cells. Data are
presented as mean ± SD values. n = 3. n.s., not significant.
Supplementary Figure 7 Irbesartan and Lumiracoxib are non-toxic to AtT-20
cells. (A-B) AtT-20 cells were treated with varying concentrations of Irbesartan and
Lumiracoxib for 24 h or 48 h and cell viability was measured by luminescent cell
viability assays. Data are presented as mean ± SD values. n = 3.
Supplementary Figure 8 Irbesartan and Lumiracoxib could directly bind to
STUB1.
(A-B) STUB1 proteins were immobilized on a CM5 chip as the solid phase.
Irbesartan (A) and Lumiracoxib (B) were subsequently serially diluted to generate
concentration gradients, facilitating their interaction with the immobilized protein.
Supplementary Figure 9 Irbesartan and Lumiracoxib synergize with pasireotide
to suppress ACTH secretion.
(A-B) AtT-20 cells were treated with Irbesartan (1 μM), Lumiracoxib (1 μM) and
pasireotide (10 nM). ACTH levels were measured by ELISA. Data are presented as

Filed under: Cushing's | Tagged: , , , , , , | Leave a comment »

Double Synchronous Functional Pituitary Adenomas Causing Acromegaly and Subclinical Cushing Disease

Posted on July 1, 2025 by MaryO

Abstract

Double pituitary adenomas with growth hormone (GH) and adrenocorticotropic hormone (ACTH) secretion are very rare. They are responsible for acromegaly with hypercortisolism. Subclinical corticotropic adenomas are exceptional.
Herein, we report the case of a patient with double functional pituitary adenomas causing acromegaly and subclinical Cushing’s disease. A 45-year-old woman was referred to our Department for suspected acromegaly. Her past medical history included diabetes mellitus treated with oral antidiabetic drugs and hypertension.
On physical examination, she had a large prominent forehead, thickened lips, increased interdental spacing, prognathism, and enlarged hands and feet. No signs of hypercortisolism were found. Biological investigations showed an elevated insulin growth factor-1 (IGF-1) level at 555 ng/mL, a GH nadir after 75 g oral glucose tolerance test at 2 ng/mL, a morning cortisol level at 158 ng/mL, an ACTH level at 64 pg/mL, a thyroid stimulating hormone (TSH) level at 2.26 mIU/L, and a free thyroxine (FT4) level at 12.8 pmol/L. Cortisol level after low-dose dexamethasone suppression test was 86 ng/mL.
The diagnosis of acromegaly associated with Cushing’s disease was established. Pituitary magnetic resonance imaging showed a pituitary macroadenoma with no clear limits. The patient underwent transsphenoidal tumor resection. The pathological examination revealed two separate pituitary adenomas. The positivity to ACTH and GH was 100% and 80%, respectively.
This case emphasizes the necessity of an evaluation of all the pituitary axes in case of adenoma in order not to miss a double hormonal secretion or more even in the absence of suggestive clinical signs.

Filed under: Cushing's, pituitary, Rare Diseases | Tagged: , , , , , , , , | Leave a comment »

Next Page »