For this retrospective analysis, we included 17 patients with PBMAH from the German Cushing’s registry who underwent AVS between 2006 to 2021 (15 treated at the LMU Klinikum, Ludwig Maximilians University Munich, Germany and 2 at the University Hospital of Würzburg, Julius Maximilians University Würzburg, Germany). These patients had documented ACTH-independent Cushing’s syndrome and bilateral adrenal masses typical for PBMAH. AVS was performed to identify a hormonally dominant side of cortisol production to guide unilateral adrenalectomy. Most of these patients (patients 1 to 12, patient 17 and patient 18) were already part of our study on the clinical role of AVS in PBMAH (14). As described also in that publication, patients underwent biochemical screening for Cushing’s syndrome by the three recommended screening tests: 1-mg dexamethasone suppression test (LDDST), late-night salivary cortisol (LNSLC) and 24h urinary free cortisol (UFC), all performed using immunoassay. Germline ARMC5 (armadillo repeat containing 5) sequencing was performed in 15 patients. Inactivating mutations of ARMC5, a putative tumor-suppressor gene, are associated with a more severe hypercortisolism, bigger adrenals with a higher number of nodules (15).The assays for the screening tests, baseline ACTH and ARMC5 status were described in the online Supplementary Material (14).

The study was approved by the LMU ethics committee (Project number: 152-10) and performed in accordance with the principles of the declaration of Helsinki. All participants gave written informed consent.

The decision to perform AVS for patients with bilateral adrenal masses was made independently by the treating endocrinologist before 2012 or by a multidisciplinary endocrine board since 2012. Our group described in detail the exact procedure previously (14). AVS was conducted without ACTH stimulation and peripheral blood samples were collected simultaneously with each of the selective blood samples. Samples were stored at -80°C until analysis. As no guideline is available for AVS performed in PBMAH patients, we interpreted the results in analogy to the Endocrine Society Practice Guideline on primary aldosteronism (PA) (16). Successful catheterization was confirmed in an exploratory analysis by a gradient of steroid concentrations between adrenal vein to peripheral vein (AV/PV, also called selectivity index, SI) greater than 2 (16–18). To analyze lateralization of hormone production we calculated a lateralization index, which is defined by the ratio of the high side to the low side of the corrected steroid of interestlevels (17). Following this definition, LI in our study is defined as:

This formula can be used for any steroid of interest, as long as the reference steroid in the denominator position is not affected by the rate of secretion of the steroid in the numerator position. In our PBMAH patient cortisol was used as steroid of interest. Recommendation for unilateral or bilateral adrenalectomy was not based on the results of LC-MS/MS presented in this study. The decision of the interdisciplinary tumour board was guided instead by the severity of CS (clinically and biochemically), the results of radiologic studies, plus AVS results measured with immunoassay as described in our previous study (14).

A panel of 15 steroids was quantified in archival AVS and peripheral EDTA-plasma samples using the commercially available MassChrom^® Steroid LC-MS/MS kit (Chromsystems, Gräfelfing, Germany) and a 1290 Infinity II ultra-high performance liquid chromatography instrument (Agilent Technologies, Santa Clara, USA) connected to a QTRAP6500+ triple quadrupole mass spectrometer (ABSciex, Framingham, USA). Sample preparation was performed via offline solid phase extraction of 500µL sample according to the instructions of the manufacturer. Before extraction, the respective stable isotope labeled steroids were added as internal standards. Twenty microliters were injected to the LC-MS/MS system, ionized with electrospray ionization (ESI) and analyzed in multiple reaction monitoring mode. Aldosterone and DHEA-S were measured in negative ESI mode. All other steroids were measured in positive ESI mode. A six-point calibration with 1/x² weighting was used for quantification of the steroids by the SciexOS software (Version 1.6.1, ABSciex, Framingham, USA). Sample containing steroid concentrations above the highest calibration were re-assayed after dilution in 0.9% saline, and results multiplied by the dilution factor. Quality control samples provided by the manufacturer were measured within each analytical run to continuously monitor performance of the LC-MS/MS measurements. We regularly participated in the national external quality assessment scheme for steroid hormones (Reference Institute for Bioanalytics, RfB, Bonn, Germany) and passed for all included steroids. The kit includes the following steroids: progesterone (Prog), 17-hydroxyprogesterone (17OHP), cortisol (F), 11-deoxycortisol (11dF), 21-deoxycortisol(21dF), cortisone (E), corticosterone (Cort), 11-deoxycorticosterone (DOC), aldosterone (Aldo), dehydroepiandrosterone (DHEA), dehydroepiandrosterone sulfate (DHEAS), dihydrotestosterone (DHT), testosterone (T), androstenedione (A) and estradiol (E2). The low limit of quantification (LLOQ) and up limit of quantification (ULOQ) for each steroid are summarized in Supplemental Data ( Table S1 ). To study the alteration of adrenal steroidogenesis pathways, conversion ratios based on adrenal size and radiological asymmetry were calculated with adrenal vein metabolite/its precursor, and conversion ratios based on ARMC5 status calculated with peripheral metabolite/its precursor, accordingly (19, 20).

The adrenal volume was measured in three dimensions and calculated by height x width x depth. One patient provided only axial planes. Therefore, in this patient adrenal size was assessed by the maximum diameter in cm. The definition of adrenal asymmetry was a difference > 30% between adrenal volumes or, if not available, the maximum diameter.

We analyzed correlation using a two-tailed Spearman correlation coefficient. Mann Whitney test was used to assess the differences of peripheral steroids and clinical parameters based on ARMC5 status, and conversion ratios based on adrenal size and ARMC5 status. The differences of conversion ratios between groups based on radiological asymmetry were evaluated by Wilcoxon test. Steroid concentrations below LLOQ and without peak were calculated as 0.5*LLOQ. A value of p<0.05 was considered statistically significant. We used Microsoft Excel 365 for data calculation and GraphPad Prism 8 for the statistical analyses.

16/17 patients were females, 15 patients were older than 55 years (see Supplementary Table S1 ). The biochemical evaluation demonstrated the typical features of adrenal hypercortisolism. ARMC5 status was evaluated in 15 patients. 6 patients showed mutations in the ARMC5 gene, while 9 patients had the wildtype.

Correlations between adrenal volume, baseline ACTH, the three diagnostic tests for CS and the peripheral steroids were analyzed in 15 postmenopausal patients ( Figure 1 ); the male (patient 4) and one premenopausal female (patient 6) were excluded from this analysis. The concentrations of aldosterone (7 patients), 21-deoxycortisol (4 patients), estradiol (8 patients), DHT (8 patients), 11-deoxycorticosterone (2 patients) and progesterone (8 patients) were below LLOQ or detection in >10 % of the samples ( Supplementary Table S2 ). Therefore, we excluded these steroids from further analysis. Adrenal volume showed negative correlations with plasma cortisone concentrations (r=-0.62, p<0.05). UFC did not correlate with plasma cortisol but did correlate positively with 17-hydroxyprogesterone (r=0.64, p<0.05), DHEA (r=0.57, p<0.05), DHEAS (r=0.82, p<0.001), androstenedione (r=0.73, p<0.01) and testosterone (r=0.54, p<0.05) in peripheral plasma samples. The association of steroid concentrations with germline ARMC5 mutation status is presented in Table 1 . Patients with germline ARMC5 mutations had lower plasma androstenedione concentrations than wild-type patients.

An ideal reference hormone to be used during AVS for correction of dilution effects in adrenal veins should have a rather short half-life and should be secreted independently of the underlying adrenal pathology. Therefore, the selection of the best reference hormone was based on three steps in our study (1): The candidate hormone should show a concentration gradient between adrenal and peripheral vein (selectivity index) ≥ 2 (2); The candidate hormone should not correlate with cortisol, which would indicate a co-secretion by the cortisol-producing tumor (3); The candidate hormone has a short half-time.

In order to evaluate specific steroids as reference hormones, a ratio of adrenal vein/peripheral vein called selectivity index (SI) was calculated for each steroid. SI > 2 was selected as the traditional cutoff to assess the success of AVS cannulation (16–18), in analogy to AVS for primary aldosteronism ( Table 2 ). In the right AVS of patient 17, only aldosterone showed an SI > 2. We therefore concluded a failure of cannulation of the right adrenal vein of patient 17, and this sample was excluded from further analysis. 17-hydroxyprogesterone, 11-deoxycortisol, corticosterone, DHEA, androstenedione and 11-deoxycorticosterone indicated successful catheterization in >90% of the cases. Thus, correlation analyses between these six steroids and cortisol were performed ( Figure 2 ). Corticosterone (r=0.71, p<0.0001), 11-deoxycortisol (r=0.57, p<0.0005), 17-hydroxyprogesterone (r=0.59, p<0.001) and 11-deoxycorticosterone (r=0.57, p<0.001) showed significant correlations with cortisol, indicating a co-secretion with the cortisol-producing tumor. Androstenedione (r=0.27, p=0.13) and DHEA (r=0.20, p=0.27) had no correlation with cortisol. Compared with androstenedione (T1/2 _A is about 30 min) (21), DHEA (T1/2 _DHEA is 15 to 30 min) (22, 23) has a shorter half-life. As a result, we concluded that DHEA could be a viable reference hormone, meeting all the criteria we defined previously.

We calculated the lateralization index to investigate differences in steroid fingerprints between the adrenals of each patient. Figure 3 shows the lateralization indices of the nine steroids, which could be measured adequately, using DHEA as reference hormone. The results indicate that the secretion of all measured steroids could be affected in PBMAH: every measured steroid showed a lateralization (LI above 2) in at least one patient. On the other hand, not every patient with PBMAH showed a lateralization of steroid production: 3 patients (patient 6, 14 and 15) had no lateralization at all when lateralization was defined as LI above 2. If defining LI above 4 as a relevant lateralization in steroid production, only 5 patients fulfilled this criterion by at least one steroid. However, out of these, in two patients (patient 7 and patient 10) the dominant side was inconsistent: In patient 7, the secretion of cortisol was mainly left (LI=82.7), while 11-deoxycortisol (LI=2.0), corticosterone (LI=2.5) and 11-deoxycorticosterone (LI=2.7) were mainly produced by the right adrenal. In patient 10, DHEAS (LI=2.4) production lateralized to the left side, whereas 11-deoxycortisol (LI=2.4), corticosterone (LI=4.8), and 11-deoxycorticosterone (LI=11.5) lateralized to the right side.

Additionally, conversion ratios (steroid/precursor) based on adrenal size and on ARMC5 status values were calculated to identify possible differences in affected steroid pathways. All conversion ratios among the pathway of adrenal steroidogenesis were analyzed ( Supplementary data, Figure S1 ). For conversion ratios involving sex hormones, only the values of the postmenopausal women were included. 14 adrenals had a volume ≥ 30 ml, 18 adrenals had a volume< 30 ml, 12 patients showed radiological asymmetry. Aldosterone/corticosterone and cortisone/cortisol were higher in the group of adrenals with volumes< 30 ml ( Figures 4A, B ) and radiologically small adrenals ( Figures 4D, E ). Androstenedione/dehydroepiandrosterone was higher in the group of adrenals with volumes ≥ 30 ml ( Figure 4C ) and radiologically large adrenals ( Figure 4F ).

ARMC5 mutation carriers showed lower conversion ratios of androstenedione/17-hydroxyprogesterone and higher testosterone/androstenedione than ARMC5-wildtype patients (p<0.05, Figures 5A, B ).

No correlation between adrenal volume and peripheral cortisol level was found. Even though UFC better reflects the 24-hour time-integrated cortisol secretion compared to random serum cortisol, still no correlation between adrenal volume and UFC was observed indicating variable steroidogenic efficiency between PBMAH cases. This is in line with the findings of Wurth et al. in a cohort of 44 patients with PBMAH, in which adrenal volume was calculated based on computed tomography scans (24). 17-hydroxyprogesterone, DHEA, DHEAS, androstenedione and testosterone showed a positive correlation with UFC. In contrary, no correlation or even negative correlation (DHEAS) with serum cortisol was seen. Also, serum cortisol was not correlated with UFC. A possible explanation could be, that cortisol secretion levels vary throughout the day and UFC reflects the daily cortisol output. As a genetic disease, PBMAH is reported to be associated with germline mutation in ARMC5 gene. Mutation carriers are found to have an more severe Cushing phenotype than patients with wild-type ARMC5 (25). In the cohort described by Espiard et al., germline ARMC5-mutation is associated with lower ACTH, and higher UFC and higher cortisol after dexamethasone suppression test compared to wild-type PBMAH patients (26). This was not observed in our cohort ( Supplementary Table S3 ), probably due to the relatively small cohort size.

At the moment cortisol is commonly used as a reference hormone to correct for dilution effects during AVS for PA. According to the experience of AVS in PA, a successful adrenal vein cannulation is traditionally defined by a SI >2 when AVS is performed without ACTH stimulation (16–18). In PA, the selection of cortisol as reference hormone is based on the assumption that cortisol is entirely secreted by the normal zona fasciculata and not overproduced by the aldosterone-producing lesion, which however has limitations in case of pronounced aldosterone and cortisol co-secretion (17, 27). Metabolites with long half-life are slowly cleared from circulation, and have decreasing adrenal to peripheral gradients, thus, this will impair the interpretation of AVS results. As a result, we selected the reference hormone based on a three-step approach accounting for these factors. According to our findings, we identified DHEA as the most promising reference hormone in patients with PBMAH.

AVS was firstly introduced to guide surgical decision making in ACTH-independent Cushing’s syndrome by colleagues from the Mayo Clinic (5). In the following years, AVS was performed occasionally in adrenal CS by several centers. Due to the limited sample size and outcome data, our study was not performed to evaluate the optimal LI cut off, rather, the study intended to evaluate the LI of all steroids to evaluate adrenal laterality in patients with PBMAH. Our results nicely illustrate that cortisol lateralization based on AVS is valuable in most patients with PBMAH. Interestingly, corticosterone and 11-deoxycortisol showed also pronounced lateralization effects. So taken together one can say that cortisol is one of the most dysregulated hormones in PBMAH pathology, but all other steroids could be affected, too.

Lower conversion ratios of cortisone/cortisol (catalyzed by HSD11B2) and aldosterone/corticosterone (catalyzed by aldosterone synthase, CYP11B2) were observed in larger adrenals both based on adrenal volumes and radiological asymmetry. HSD11B2 is not expressed in normal adrenals (28), but expressed in adrenal adenomas (28, 29). Therefore, this result is unexpected. However, up to our knowledge the expression status of HSD11B2 in adrenals of PBMAH patients have not been investigated yet.CYP11B2 was undetectable by immunohistochemistry in the tumor of adrenal CS in a study by Nishimoto et al. (30). Similar immunohistochemical analyses have not yet been done in PBMAH, but it could be speculated that lower synthesis of aldosterone in adrenals affected by PBMAH is possibly caused by CYP11B2 repression. Higher conversion ratio of androstenedione/dehydroepiandrosterone (catalyzed by CYP17A1 or CYB5A) in larger adrenals indicates possible dysregulation in androgenic steroids in PBMAH. Taken together, even though PBMAH is primarily associated with cortisol excess, there are co-secretion of steroids of the mineralocorticoid and androgen pathways. In addition, we observed patients with germline ARMC5 variants have lower androstenedione/17-hydroxyprogesterone and higher testosterone/androstenedione conversion ratios. In line with these findings, peripheral androstenedione concentrations in germline ARMC5 mutation carriers were decreased, indicating that the androgen steroid pathway is dysregulated in PBMAH patients carrying a germline ARMC5 mutation, possibly through decreased ACTH. In line with this notion, lower peripheral DHEAS levels measured by LC-MS/MS in germline ARMC5 mutation carriers were described in another cohort (31). We saw the same tendency in our cohort, but this finding failed to be significant (119 vs 314 ng/ml, p=0.06).