Comorbidities in mild autonomous cortisol secretion and the effect of treatment: systematic review and meta-analysis

Objective: To assess (1) comorbidities associated with and (2) treatment strategies for patients with adrenal incidentalomas and mild autonomous cortisol secretion (MACS; > 1.8 µg/dL ( > 50 nmol/L) cortisol level cut-off following the 1 mg dexamethasone suppression test). Design: Systematic review and meta-analysis. Methods: Seven databases were searched up to July 14, 2022. Eligible studies were (randomized) trials, cohort studies, and cross-sectional studies assessing comorbidities potentially attributable to cortisol excess or mortality in patients with adrenal incidentaloma with or without MACS or the effects of conservative or surgical management of MACS. Random-effects meta-analysis was performed to estimate pooled proportions (with 95% CIs). Results: In 30 cross-sectional and 16 cohort studies ( n = 17 156 patients in total), patients with MACS had a higher prevalence of diabetes (relative risk [RR] 1.44 [1.23-1.69]), hypertension (RR = 1.24 [1.16-1.32]), and dyslipidemia (RR = 1.23 [1.13-1.34]). All-cause mortality (adjusted for confounders) in patients with MACS, assessed in 4 studies ( n = 5921), was increased (hazard ratio [HR] = 1.54 [1.27-1.81]). Nine observational studies ( n = 856) and 2 randomized trials ( n = 107) suggest an improvement in glucometabolic control (RR = 7.99 [2.95-21.90]), hypertension (RR = 8.75 [3.99-19.18]), and dyslipidemia (RR = 3.24 [1.19-8.82]) following adrenalectomy. Conclusions: The present systematic review and meta-analysis highlight the relevance of MACS, since both cardiometabolic morbidities and mortality appeared to have increased in patients with MACS compared to patients with non-functioning incidentalomas. However, due to heterogeneous definitions, various outcomes, selective reporting, and missing data, the reported pooled estimates need to be interpreted with caution. The small number of patients in randomized trials prevents any strong conclusion on the causality between MACS and these comorbidities.


Introduction
The term mild autonomous cortisol secretion (MACS) was proposed by the recent clinical guidelines from the European Society of Endocrinology (ESE) and the European Network for the Study of Adrenal Tumors (ENSAT) for patients with adrenal incidentalomas and adrenocorticotropic hormone (ACTH)-independent cortisol hypersecretion without clinical signs of overt Cushing's syndrome (eg, catabolic symptoms such as muscle weakness, skin fragility, and striae). 1,2In this condition, cortisol production is independent of physiological hypothalamic-pituitary-adrenal (HPA) axis feedback mechanisms resulting in longstanding inappropriate exposure to cortisol in various organs and tissues. 30][11][12][13][14] However, the association between MACS and these comorbidities potentially attributable to cortisol is uncertain, and the causality is disputed.
There is considerable discussion regarding optimal testing and relevant cut-offs to diagnose MACS.The overnight 1 mg dexamethasone suppression test (1mg-DST) is considered the best, although suboptimal, test. 1,2Various diagnostic algorithms have been used for "subclinical hypercortisolism," although direct comparisons of diagnostic performance of tests are lacking.Importantly, formal comparisons between algorithms and relevant clinical outcomes are lacking.In the 2016 ESE-ENSAT guideline, the use of the 1mg-DST was recommended based on pathophysiological reasoning, simplicity, and the incorporation of the 1mg-DST in the diagnostic algorithms of most studies. 1 Given the available evidence at that time, the guidelines proposed 2 categories depending on the post-1mg-DST cortisol levels: "autonomous cortisol secretion" for cortisol levels >138 nmol/L and "possible autonomous cortisol secretion" for cortisol levels 51-138 nmol/L.During the development of the updated guideline, the cut-off of >50 nmol/L was evaluated and therefore included in the present review. 1,2urthermore, the optimal therapeutic approach for MACS is not well established.Two approaches are considered: (1) management of MACS-associated comorbidities (eg, diabetes and hypertension) and ( 2) adrenalectomy.The present study aims to systematically review the literature regarding (1) the epidemiological scope of comorbidities in MACS and (2) the effect of treatment, either conservative, medical, or surgical, on the aforementioned comorbidities.

Research questions and study aims
The present systematic reviews and meta-analyses are incorporated in the 2023 ESE-ENSAT adrenal incidentaloma guideline. 2For the present review, 2 research questions were formulated: (1) Is MACS associated with an increased cardiovascular, metabolic, and fracture risk in patients with adrenal mass(es)?, and (2) What is the effectiveness of surgery versus conservative/medical treatment in patients with adrenal mass(es) and MACS with an increased cardiovascular, metabolic, and fracture risk?Two separate systematic reviews of the literature were performed to answer these questions.These reviews were performed while adhering to the Declaration of Helsinki.Results were reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement. 15The present reviews were not registered on PROSPERO, as they were part of the clinical practice guideline development.The present reviews were not assessed by the medical ethical review committee as no individual patients were studied.

Eligibility criteria
Original studies on patients with adrenal mass(es), in which endocrine work-up for glucocorticoid excess was performed -irrespective of definitions of "autonomous cortisol secretion" used-were eligible for inclusion.Comparison between patients based on biochemical profiles (including post-1mg-DST serum cortisol levels) needed to be incorporated.
For the first systematic review (Q1), comparing patients with non-functioning adrenal incidentaloma and patients with adrenal incidentaloma and MACS, cross-sectional studies and longitudinal clinical studies (ie, randomized or nonrandomized trials and cohort studies) assessing at least one of the following clinically relevant outcomes were eligible: mortality, cardiovascular events, cardiometabolic comorbidities (eg, glucose intolerance, diabetes mellitus, hypertension, and dyslipidemia), vertebral fractures, and health-related quality of life (HR-QoL).
For the second systematic review (Q2), comparing conservative (ie, active surveillance or pharmacological interventions) versus surgical approaches in patients with adrenal incidentaloma and MACS, longitudinal clinical studies (ie, randomized or non-randomized trials and cohort studies) assessing at least one of the following major outcomes were eligible: mortality, cardiovascular events, cardiometabolic comorbidities, vertebral fractures, and HR-QoL.
To minimize risk of selection bias, ≥ 10 patients per study arm had to be included.Only articles written in English were considered.

Data extraction and study outcomes
Identified references were entered in EndNote X20 (Thomson Reuters, Philadelphia, PA, USA).First, studies were screened by title and abstract, followed by the full-text screening of all eligible articles.Two independent reviewers (I.C.M.P. and O.M.D.) reviewed potentially relevant articles in detail.Following full-text screening, data were extracted from included studies.
The following data were extracted: study design, number of patients, study population, study period, eligibility criteria, subgroups based on biochemical profiles or treatment strategies, follow-up duration, and outcome parameters (number of patients/events per group and effect sizes).In some studies, data were solely presented as percentages (resulting in backcalculation of absolute numbers for that outcome), or solely presented in figures without mentioning absolute patient numbers (resulting in the estimation of the numbers to the best of the author's abilities).If data were presented only according to patient categories (eg, pre-, and post-menopausal women), the data were combined prior to analyses.For the second systematic review, improvement or complete resolution of the assessed comorbidities was recorded.For the assessment of improvement, all reported data (ie, absolute numbers, as well as general statements without further specification) were extracted and considered.
Reported effect estimates adjusted for confounders were extracted as well.Unfortunately, uncorrected effect sizes (bar the outcomes of mortality) were used for further analyses, since not all studies performed correction for (multiple) confounding factors.For the HR-QoL outcomes, absolute values as outcomes of the validated questionnaires were extracted.

Risk of bias assessment using Grading of Recommendations, Assessment, Development, and Evaluations
7][18] In Tables S1   and S2, all components of the GRADE system (ie, imprecision, inconsistency, indirectness, and publication bias, respectively) and the final grading of all outcomes are summarized.

Statistical analysis
An overview of reported outcomes of all included studies is summarized in Table S3 (for question 1) and Table S4 (for question 2), respectively.If multiple studies described (partially) overlapping populations, only the largest study reporting on the outcome of interest was included, resulting in the meta-analyses containing a subset of the included studies for both question 1 and question 2. A random-effects logistic regression model was used for pooled percentages and relative risks (RR), accompanied by 95% CIs.In order to prevent exclusion of studies with 0 events, results were modified to include 0.5 events.Subgroup analyses based on type of study (cohort study vs randomized controlled trial [RCT]), and cutoff levels were performed.All analyses were performed using Stata 16.1 (Stata Corp., College Station, TX, United States).

Study inclusion
The flowcharts of study inclusion for the 2 systematic reviews are shown in Figure 1.The initial search yielded 1059 abstracts (Q1) and 291 abstracts (Q2) respectively, of which 52 (Q1) and 8 (Q2) publications were assessed in detail.A total of 46 (34 new) studies for Q1 (comorbid conditions based on biochemical cortisol profile) and 11 studies (7 new) for Q2 (therapeutic approaches for patients with autonomous cortisol secretion) were included.

Description of included studies
Included studies were published between 2004 and 2022.The predominant method used to define MACS was cortisol levels >1.8 µg/dL post-1mg-DST, especially in studies published after publication of the 2016 guideline. 1

Treatment of MACS (question 2)
8][69][70][71][72][73] The summary of included studies is listed in Table S4.The conservative treatment modalities ranged from pharmacotherapeutic interventions for comorbidities to watchful waiting, with different approaches not being standardized.The quality of evidence from these studies is low to very low, mainly due to confounding factors (as outlined in Table S2).Two studies were randomized,

S90
European Journal of Endocrinology, 2023, Vol.189, No. 4 although treatment allocation was not concealed, and none of the studies reported blinded outcome assessment. 65,66urthermore, most studies were downgraded for imprecision, due to the small number of events.Differences in diagnostic protocols used to define autonomous cortisol secretion, definitions of outcomes, and duration of follow-up were heterogeneous between and within studies.Moreover, in none of these 11 included studies, evaluation and treatment of concomitant comorbidities following intervention were standardized.

Mortality
Six studies reported on all-cause mortality. 11,49,55,56,59,62In one study, 49 analyses performed were not adjusted for confounding factors, and therefore, not included in further analyses.In another study, solely reporting on absolute cortisol levels, a significantly increased hazard ratio (HR) for mortality was observed with every 10 nmol/L increase in cortisol levels following 1mg-DST (HR = 1.10 [95% CI 1.01-1.19]). 11The remaining 4 studies (including 5921 patients) are summarized in Figure 2, with all point estimates being >1.0 regardless of the cut-off levels used for the definition of MACS and pooled HR being increased in patients with MACS (1mg-DST cortisol level cut-off of >1.8 µg/dL): HR = 1.54 (95% CI 1.27-1.81). 55,56,59,62The leading causes of death were cardiovascular 11,49,56 and malignancies. 55,59rdiovascular events Twelve studies (6 cross-sectional studies and 6 cohort studies) reported on the prevalence or incidence of acute cardiovascular events, 11,36,41,52,53,55,56,60,61 with different definitions of cardiovascular events being used (eg, myocardial infarction and stroke).Using cortisol levels >1.8 µg/dL following 1mg-DST, point estimates for the prevalence of cardiovascular events were >1.0 in 3/4 studies comparing patients with MACS and non-functioning adenomas, whereas point estimates for CVA and MI varied greatly (Figure S1).In 5 cohort studies, 11,52,53,55,56 the incidence of cardiovascular events during follow-up was higher in MACS.

HR-QoL
Two studies reported on several QoL outcomes, of which most subscales were not different between patients with MACS and patients with non-functioning adenomas, 40,48 bar the components disability and stress Sheehan Disability Scale (SDS) being higher in patients with MACS. 40Two studies reported on mental health and cognitive function status, which was not affected by the presence of MACS, but patient frailty was significantly higher in patients with MACS compared to nonfunctioning adenomas (age-, BMI-, and sex-adjusted). 40,74gure 2. Meta-analysis of all-cause mortality in patients with adrenal incidentaloma.HR with CIs for all-cause mortality as reported by the individual studies, following adjustment for >2 confounders were included.Because of the asymmetry in the reported CIs, the CI tolerance was adjusted for performance of the random-effects maximum likelihood (REML) meta-analysis of the HR.The squares and horizontal lines correspond to the HR with the adjusted 95% CIs.The diamond represents the pooled HR and 95% CI based on the 1mg-DST cut-off of >1.8 µg/dL (50 nmol/L) for MACS.The reference line represents the estimate of 1.0 meaning no difference between the patient groups.MACS, mild autonomous cortisol secretion.

Mortality and major cardiovascular events
None of the included studies reported on the risk of mortality, or major cardiovascular events.

Metabolic and cardiovascular comorbidities
5][66][67][68][69][70][71][72][73] In the first RCT, including 45 patients in total, in 5/8 patients with type 2 diabetes mellitus, glucometabolic control improved after surgery (with normalization of glucose tolerance in 2 patients), compared to 0/6 patients with diabetes in the conservative group. 65In the second and largest RCT to date including 62 patients in total, 56 patients could be evaluated at least 6 months after randomization.In these patients, rates of improvement in glycemic control were 28% (surgery) and 3.3% (conservative), respectively. 66However, one has to acknowledge that only 11 patients (19.6%) had diabetes and additional 16 (28.6%)impaired fasting glucose or impaired glucose tolerance.Comparable results were reported in the cohort studies, with all included studies reporting point estimates >1.0 indicating better outcomes after surgical treatment: pooled RR = 8.0 (2.9-21.9)2,73 Notably, in the largest RCT, improvement in BP control was shown in 68% of surgically treated patients and in only 13.4% in the conservative arm; 66 again, one-third of patients did not have hypertension.As summarized in Figure 6B, all individual studies reported point estimates of >1.0 for improvement or normalization of BP following surgery, with pooled RR being 8.8 (4.0-19.2).
72,73 Chiodini et al. 67 reported similar numbers of patients improving LDL-cholesterol levels in both treatment groups, whereas in the other studies, none of the conservatively treated patients, in contrast to the surgically treated patients, showed improvement of lipids. 65,68,69,72,73Pooled RR for the improvement of dyslipidemia in patients with MACS was 3.2 (1.2-8.8)comparing surgery to conservative management, as is shown in Figure 6C.

Risk of vertebral fractures
Solely 1 study reported on vertebral (micro)fractures. 71atients who underwent adrenalectomy presented with less new vertebral (micro)fractures compared to patients managed conservatively during follow-up of 39.9 ± 20.9 months (9.4% vs 52.2%).

Discussion
In this systematic review, the prevalence of comorbidities potentially attributable to cortisol in patients with MACS compared to patients with non-functioning adrenal incidentalomas was assessed.Generally, the prevalence of cardiometabolic diseases was increased in patients with MACS, with the prevalence of diabetes, hypertension, and dyslipidemia being 15%-40% higher compared to patients with nonfunctioning adenomas.Moreover, mortality in patients with MACS was increased.Regarding effects of treatment for MACS, the evidence is rather limited, although data do imply that adrenalectomy lowered the risk of comorbid conditions.
Cortisol autonomy might be a biological continuum without clear separation between non-functioning adenomas and functioning adenomas associated with cortisol excess. 1,2,58In the 2016 ESE-ENSAT guidelines, the term "autonomous cortisol secretion" was introduced for patients with adrenal incidentaloma and cortisol levels after 1mg-DST of >138 nmol/L, with "possible autonomous cortisol secretion" for cortisol levels after 1mg-DST between 50 and 138 nmol/L. 1 One main message of this systematic review is that patients with possible autonomous cortisol have a similar prevalence of comorbidities potentially attributable to cortisol, which is clearly higher than in patients with non-functioning adenomas.Therefore, the new guideline defines MACS by cortisol >1.8 µg/dL post-1mg-DST. 2Moreover, the present review showed that any cortisol cut-off value post-1mg-DST has an overall poor accuracy to predict prevalent or incident comorbidities.Several studies have reported higher rates of comorbidities with increasing post-1mg-DST cortisol levels, with the relationship being non-linear. 41,42,55,56However, with the present evidence, a cortisol level cut-off of ≤50 nmol/L (≤1.8 µg/dL) following 1mg-DST appears to discern patients at risk for suffering from comorbidity-related clinically relevant cortisol overproduction.In the future, the additional relevance of ACTH suppression on MACS-associated comorbidities should be investigated.Notably, the 1mg-DST-as all assays using a cut-off level-is subject to misclassifications based on interand intra-assay variations and biological variations.To further assess the robustness of the cut-off level including their validity when used for treatment decisions, individual patient data (IPD) analysis is warranted.

S96
European Journal of Endocrinology, 2023, Vol.189, No. 4 adrenal incidentaloma.This comparison is necessary to judge the true impact of the adrenal tumor on the different comorbidities, since certain diseases (eg, diabetes) might increase the risk for adrenal tumors (and not solely vice versa). 79Nevertheless, this review highlighted that MACS is associated with negative cardiovascular risk profiles, including evidence for an association between MACS and an increased risk of cardiovascular events and excess mortality.Since observational studies-by design-cannot prove causality, interventional studies are of key importance.This review suggests that surgery may improve the cardiovascular risk profile, although evidence is weak.5][66][67][68][69][70][71][72][73] In addition to the limitations mentioned above, none of the published studies evaluated the endpoints in a standardized manner, and medical treatment was not standardized, which is of importance to compare and interpret the effect of therapeutic interventions.Importantly, no data were available on more clinically relevant endpoints (eg, mortality or major cardiovascular events) following surgery or other treatment strategies.Moreover, post-abdominal surgery weight loss might influence the cardiovascular risk profile, leading to non-hormonal effects influencing the study outcome.In the included RCTs comparing adrenalectomy to conservative management for MACS, the influence of post-abdominal (adrenal) surgery weight loss on the reported outcomes is not a general rule, as-surprisingly -50% and 76% of patients, respectively, did not lose weight post-surgery. 65,66Future studies should focus on the indications of adrenalectomy in patients with MACS.
The present systematic review and meta-analysis were part of the recent update to the clinical guideline on adrenal incidentalomas 1,2 and highlight the clinical importance of the condition MACS.In conclusion, MACS is associated with an increased risk for multiple metabolic or cardiovascular comorbidities and mortality.Moreover, adrenalectomy of the adrenal adenoma causing MACS appeared to be beneficial.Due to the limited sample size, the heterogeneous definitions of outcome measurements, selective reporting, and significant proportions of missing data, the reported pooled estimates need to be interpreted with caution.In general, the quality of the evidence was low, hampering making strong conclusive remarks and highlighting the need for future prospective studies.

Figure 1 .
Figure 1.Flowchart of article inclusion.All articles derived from the searches for question 1 (A) and question 2 (B) were screened based on title and abstract, followed by full-text screening when applicable.n, number of articles.

Figure 3 .
Figure 3. Summary of RR for prevalent diabetes in patients with adrenal incidentaloma comparing non-functioning adenomas to MACS.RR with CIs for the prevalence of diabetes as reported were included in the random-effects maximum likelihood (REML) meta-analysis based on the different cut-offs for MACS following the 1mg-DST.The squares and horizontal lines correspond to the RR with 95% CIs.The diamond represents the pooled HR and 95% CI.The reference line represents the estimate of 1.0 meaning no difference between the patient groups.DST, dexamethasone suppression test; MACS, mild autonomous cortisol secretion.

Figure 4 .
Figure 4. Summary of RR for prevalent hypertension in patients with adrenal incidentaloma comparing non-functioning adenomas to MACS.RR with CIs for the prevalence of hypertension as reported were included in the random-effects maximum likelihood (REML) meta-analysis based on the different cut-offs for MACS following the 1mg-DST.The squares and horizontal lines correspond to the RR with 95% CIs.The diamond represents the pooled HR and 95% CI.The reference line represents the estimate of 1.0 meaning no difference between the patient groups.DST, dexamethasone suppression test; MACS, mild autonomous cortisol secretion.

Figure 5 .
Figure 5. Summary of RR for prevalent dyslipidemia in patients with adrenal incidentaloma comparing non-functioning adenomas to MACS.RR with CIs for the prevalence of dyslipidemia as reported were included in the random-effects maximum likelihood (REML) meta-analysis based on the different cut-offs for MACS following the 1mg-DST.The squares and horizontal lines correspond to the RR with 95% CIs.The diamond represents the pooled HR and 95% CI.The reference line represents the estimate of 1.0 meaning no difference between the patient groups.DST, dexamethasone suppression test; MACS, mild autonomous cortisol secretion.

Figure 6 .
Figure 6.Summary of RR for improvement in glucometabolic and cardiovascular parameters in patients with adrenal incidentaloma and MACS following treatment.RR with CIs for the improvement in (A) glucometabolic control, (B) hypertension, and (C) dyslipidemia as reported were included in the random-effects maximum likelihood (REML) meta-analysis based on the different cut-offs for MACS following the 1mg-DST.The squares and horizontal lines correspond to the RR with 95% CIs.The upper two diamonds represent the pooled HR and 95% CI for the RCT and cohort studies, respectively.The lowest diamond represents the pooled HR and 95% CI combining the RCT and cohort studies.The reference line represents the estimate of 1.0 meaning no difference between the patient groups.MACS, mild autonomous cortisol secretion.