Annals of Internal Medicine | 2019
Natural History of Adrenal Incidentalomas With and Without Mild Autonomous Cortisol Excess
Abstract
The incidental detection of adrenal masses by imaging is increasingly common in clinical practice owing to the ever-increasing use of cross-sectional imaging. Adrenal incidentalomas are found in about 5% of the population, and incidence increases with age (1). When an adrenal incidentaloma is detected, a systematic biochemical work-up is required to exclude excess secretion of adrenal hormones. In most cases, further radiologic work-up also ensues to exclude cancer. The mainstay management strategy for both overt hormone excess and cancer is adrenalectomy (2). Whereas most adrenal incidentalomas are benign nonfunctioning adrenal tumors (NFATs), about 15% are associated with hormone excess (3). A frequent finding in patients with adrenal incidentalomas is failure to suppress cortisol sufficiently after dexamethasone administration in the absence of typical clinical signs of cortisol excess (3), a condition often called subclinical Cushing syndrome or, as recently coined, mild autonomous cortisol excess (MACE). Generally, NFATs do not require surgery, whereas whether MACE adenomas should be removed is controversial (2, 4, 5), despite several studies suggesting that MACE may result in increased risk for cardiometabolic disease and death (68). Whether NFATs should undergo long-term follow-up is also debatable because of the substantial health economic burden and increased radiation exposure from potentially unnecessary repeated imaging. However, NFATs may be associated with increased cardiometabolic risk during follow-up (9, 10), and marked increases in adenoma size and development of clinically overt hormone excess over time have been reported (1114). Adrenal incidentaloma guidelines from the National Institutes of Health and the American Association of Clinical Endocrinologists recommend repeated imaging and hormonal assessment for up to 5 years for tumors characterized at diagnosis as benign without overt hormonal hypersecretion (15, 16). Recent European guidelines discourage repeated imaging and hormonal assessment but recommend annual screening for overt Cushing syndrome in patients with MACE (2). These considerable differences in recommendations reflect the inadequacy of evidence that informs the follow-up of adrenal incidentalomas without an immediate indication for surgery. Most previous studies that addressed the natural history of NFAT and MACE were done in small cohorts of heterogeneous populations and used varied biochemical and radiologic assessments. Therefore, we sought to summarize the available literature and undertake a systematic review and meta-analysis of the natural history of NFAT and MACE adenomas to better guide their management. Our objectives were to determine the following: 1) the proportion of NFAT and MACE adenomas that increase in size, the degree of growth during follow-up, and the proportion of adenomas that show malignant transformation; 2) the proportion of NFAT and MACE adenomas that change their hormone production during follow-up; 3) the prevalence of hypertension, obesity, dyslipidemia, type 2 diabetes, and cardiovascular events in patients with NFAT or MACE at baseline and their incidence during follow-up; and 4) incident all-cause and cardiovascular mortality in patients with NFAT or MACE during follow-up. Methods The meta-analysis was done according to a predesigned protocol (developed on 16 January 2017), which included methods and inclusion criteria and was developed in conformance with the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) statement (17). Data Sources and Searches We did a comprehensive electronic search of multiple databases, including Ovid MEDLINE In-Process & Other Non-Indexed Citations, Ovid MEDLINE, Ovid Embase, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, and Scopus, for titles published between 1 January 1990 and 20 February 2019. This search strategy (Supplement Table 1) was designed and carried out by an experienced reference librarian (L.P.) with input from a study investigator with experience in conducting systematic reviews (I.B.). Controlled vocabulary supplemented with keywords was used to search for studies of the natural history of NFAT and MACE during follow-up. No language restriction was applied. We further searched bibliographies of included articles to identify any eligible studies that the electronic search may have missed. Supplement. Supplementary Material Study Selection We included original studies that evaluated populations of patients with NFAT, MACE, or both. Studies could be prospective or retrospective and had to include adults (aged 18 years) with NFAT or MACE (as defined by authors) (Supplement Table 2), have a sample size of at least 20 conservatively managed patients with at least 12 months of longitudinal follow-up, and report outcomes of interest at baseline and the end of follow-up. We excluded all nonoriginal studies and case reports. We included the following outcomes of interest for which we identified follow-up data for at least 100 patients: change in size of adrenal tumor, malignant transformation, change in hormone secretion, development of hypertension, obesity or weight gain, dyslipidemia, type 2 diabetes, cardiovascular events, and death during follow-up (Supplement Table 3 defines these outcomes). We excluded outcomes with fewer than 100 identified patients, which included bone loss, vertebral fractures, and visceral fat accumulation. Six reviewers independently screened identified studies for eligibility. Conforming to the predefined inclusion criteria, they reviewed the titles and abstracts of identified studies in duplicate and removed all studies that did not fulfill the inclusion criteria at this stage. When reviewers disagreed, studies progressed to the next stage. In this phase, 2 reviewers independently screened full-text articles to assess eligibility for final inclusion, and discrepancies were resolved through discussion and consensus. Data Extraction and Quality Assessment Data extracted from each study included first author, year of publication, country where the study was done, participant demographics, inclusion and exclusion criteria, study design, and prevalence of outcomes of interest at baseline and the end of follow-up. One investigator (Y.S.E.) combined all data extraction forms and resolved discrepancies at this stage by discussing them with the reviewers and referring back to the full-text articles. Pairs of reviewers independently assessed the quality of each included study. The modified NewcastleOttawa Scale for risk-of-bias assessment of observational studies was applied to assess the following domains (Supplement Table 4): selection (whether the cohort of patients with adrenal adenomas was representative), ascertainment (whether glucocorticoid autonomy and ascertainment of outcomes were described clearly), reporting (whether the proportion of patients affected by the studied outcome at baseline was reported clearly), causality (adequacy of follow-up, adequacy of the proportion of baseline patients assessed at follow-up, and similarity of outcome assessment between baseline and follow-up), and funding and conflicts of interest (whether information on funding and conflicts of interest was reported clearly) (18). One investigator (Y.S.E.) combined all of the quality assessment forms and resolved discrepancies by discussion with assessors. Data Synthesis and Analysis We analyzed only outcomes of interest for which the systematic review identified at least 100 patients. For continuous data, we used means and SDs from each study; when studies did not report these directly, we converted available data measures to means and SDs. We then pooled the mean differences from follow-up to baseline across the studies using the random-effects model, with heterogeneity estimated from the MantelHaenszel model. We estimated an exact binomial CI following the ClopperPearson procedures (19). We followed the FreemanTukey double arcsine transformation (20) to stabilize the variances. For dichotomous data, we calculated the event rate from each study group. We then pooled the log-transformed event rates using the DerSimonian and Laird random-effects models, with heterogeneity estimated from the MantelHaenszel model. All analyses were done using the random-effects model to produce proportion estimates and 95% CIs. Heterogeneity was assessed using the I 2 statistic, which represents the proportion of variability that is not attributable to chance. I 2 values more than 50% indicate substantial heterogeneity. Statistical analyses were done using OpenMeta [Analyst] (Brown University) (21). All values were 2-tailed, and a P value less than 0.05 was set as the threshold for statistical significance. Role of the Funding Source The funders had no role in study design, data collection or analysis, decision to publish, or preparation of the manuscript. Results Characteristics of Included Studies The database search yielded 1111 studies, in addition to 28 identified from bibliographies of screened articles. After screening, 32 studies were included in the analysis (Supplement Table 2). The PRISMA flow diagram outlines the numbers of studies screened, included, and excluded (Figure 1) (22). Of the 32 included studies (7, 10, 1214, 2349), 17 were retrospective and 15 were prospective. Eligible studies were mainly done in Europethat is, Italy (n= 13), Turkey (n= 4), Greece (n= 4), Finland (n= 2), the United Kingdom (n= 1), France (n= 1), Sweden (n= 1), and Spain (n= 1)but also in the United States (n= 2), Japan (n= 1), South Korea (n= 1), and Hong Kong (n= 1). Included studies reported follow-up outcomes of interest on 4121 patients (2873 with NFAT, 784 with MACE, and 464 with either NFAT or MACE; for the latter, it was not possible to disaggregate the baseline or follow-up outcomes on the basis of hormone assessment). The mean age of patie