The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No.

11, 3320–3330
https://doi.org/10.1210/clinem/dgab468
Clinical Research Article

Clinical Research Article

Cardiometabolic Outcomes and Mortality

in Patients with Adrenal Adenomas in a
Population-Based Setting

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

Catherine D. Zhang,1 Dingfeng Li,1 Ravinder Jeet Kaur,1 Andreas Ebbehoj,2
Sumitabh  Singh,1 Elizabeth  J.  Atkinson,3 Sara  J.  Achenbach,3
William F. Young Jr.,1 Wiebke Arlt,4,5 Walter A. Rocca,6,7,8, and Irina Bancos1
1
Division of Endocrinology, Diabetes, Metabolism and Nutrition, Mayo Clinic, Rochester, MN, USA;
2
Department of Endocrinology and Internal Medicine, Aarhus University Hospital, Aarhus, Denmark;
3
Division of Biomedical Statistics and Informatics, Department of Health Sciences Research, Mayo
Clinic, Rochester, MN, USA; 4Institute of Metabolism and Systems Research, University of Birmingham,
Birmingham, UK; 5NIHR Birmingham Biomedical Research Centre, University of Birmingham and University
Hospitals Birmingham NHS Foundation Trust, Birmingham, UK; 6Division of Epidemiology, Department of
Health Sciences Research, Mayo Clinic, Rochester, MN, USA; 7Department of Neurology, Mayo Clinic,
Rochester, MN, USA; and 8Women’s Health Research Center, Mayo Clinic, Rochester, MN, USA
ORCiD numbers: 0000-0001-7104-7824 (A. Ebbehoj); 0000-0001-5106-9719 (W. Arlt); 0000-0001-9332-2524 (I. Bancos).

Received: 14 April 2021; Editorial Decision: 20 June 2021; First Published Online: 29 June 2021; Corrected and Typeset:
21 July 2021.

Abstract 
Context: While adrenal adenomas have been linked with cardiovascular morbidity in
convenience samples of patients from specialized referral centers, large-scale population-
based data are lacking.
Objective: To determine the prevalence and incidence of cardiometabolic disease and
assess mortality in a population-based cohort of patients with adrenal adenomas.
Design:  Population-based cohort study.
Setting:  Olmsted County, Minnesota, USA.
Patients:  Patients diagnosed with adrenal adenomas without overt hormone excess and
age- and sex-matched referent subjects without adrenal adenomas.
Main outcome measure:  Prevalence, incidence of cardiometabolic outcomes, mortality.
Results: (Adrenal adenomas were diagnosed in 1004 patients (58% women, median
age 63 years) from 1/01/1995 to 12/31/2017. At baseline, patients with adrenal adenomas
were more likely to have hypertension [adjusted odds ratio (aOR) 1.96, 95% CI 1.58-2.44],
dysglycemia (aOR 1.63, 95% CI 1.33-2.00), peripheral vascular disease (aOR 1.59, 95% CI
1.32-2.06), heart failure (aOR 1.64, 95% CI 1.15-2.33), and myocardial infarction (aOR 1.50,
95% CI 1.02-2.22) compared to referent subjects. During median follow-up of 6.8 years,

ISSN Print 0021-972X  ISSN Online 1945-7197

3320   https://academic.oup.com/jcem Printed in USA
© The Author(s) 2021. Published by Oxford University Press on behalf of the Endocrine Society. All rights reserved. For
permissions, please e-mail: [email protected]
The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11 3321

patients with adrenal adenomas were more likely than referent subjects to develop de
novo chronic kidney disease [adjusted hazard ratio (aHR) 1.46, 95% CI 1.14-1.86], cardiac
arrhythmia (aHR 1.31, 95% CI 1.08-1.58), peripheral vascular disease (aHR 1.28, 95% CI
1.05-1.55), cardiovascular events (aHR 1.33, 95% CI 1.01-1.73), and venous thromboembolic
events (aHR 2.15, 95% CI 1.48-3.13). Adjusted mortality was similar between the 2 groups.
Conclusion: Adrenal adenomas are associated with an increased prevalence and
incidence of adverse cardiometabolic outcomes in a population-based cohort.
Key Words: adrenal incidentaloma, adrenal mass, epidemiology, incidence, prevalence, cardiovascular outcomes,
cardiovascular events

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

Adrenal tumors are commonly encountered in clinical Methods
practice and are predominantly incidental findings, ob- Study Design and Setting
served in 5% to 7% of adults undergoing cross-sectional
We conducted a population-based cohort study in Olmsted
abdominal imaging with increased prevalence and inci-
County, Minnesota, USA, using the Rochester Epidemiology
dence with older age (1-4). The majority of incidentally
Project (REP). REP is a medical records-linkage system
discovered adrenal tumors—adrenal incidentalomas—are
that captures health information across different care pro-
benign adrenocortical adenomas. Overt hormone excess
viders and care settings (primary to tertiary) for residents
is rare, but mild autonomous cortisol secretion (MACS)
of Olmsted County. Researchers are able to identify per-
may be present in 30% to 50% of adrenal adenomas (2,5).
sons with specific medical diagnoses and surgical proced-
Patients with MACS lack the classic signs and symptoms of
ural codes of interest indexed in the REP and to retrieve
glucocorticoid excess that are seen in Cushing’s syndrome
medical records for persons who have provided Minnesota
(CS) but demonstrate abnormal glucocorticoid autonomy
research authorization (overall participation 98% among
that may still lead to clinical consequences over time.
Olmsted County residents) (7). Additional information
In a recent systematic review and meta-analysis on
on the Olmsted County population and on the REP are
the natural history of adrenal incidentalomas, patients
available in other publications (7-9). Study approval was
with adrenal adenomas and no overt hormone excess
obtained from the Mayo Clinic and Olmsted Medical
[nonfunctioning adrenal tumors (NFAT) or MACS] were
Center Institutional Review Boards.
found to have a high pooled prevalence and incidence of
cardiometabolic disease at baseline and at follow-up, re-
spectively, with increased morbidity in patients with MACS Study Population and Assessment of
compared to NFAT (6). The conclusions from the system- Exposure Status
atic review and meta-analysis, however, were limited by sig- We previously assembled a population-based cohort of all
nificant heterogeneity in the definitions of MACS and of Olmsted County residents diagnosed with an adrenal tumor
the clinical outcomes across the individual studies. Many from January 1, 1995 to December 31, 2017 (3). In brief,
previous studies were also smaller in size, lacked referent we identified every patient who received an International
subjects without adrenal tumors for comparison, and had Classification of Diseases, 9th or 10th edition, diagnosis
insufficient follow-up time for the assessment of de novo code for adrenal neoplasm [see Supplementary Table 1
cardiovascular events and mortality. Furthermore, the ex- in (10)] during the study timeframe using the REP infra-
isting data on adrenal adenomas and cardiometabolic risk structure. Individual medical records were reviewed to
have only been derived from specialized referral centers, confirm the presence of an adrenal tumor, assess eligi-
which limit the generalizability of the findings to the gen- bility criteria, and extract clinical information. For this
eral population. Referred patients may differ substantially study, we excluded patients younger than 20 years old, pa-
from those who are not referred in terms of comorbidities tients lacking diagnosis or procedural codes within 2 years
and subtle hormone abnormalities. prior to the adrenal tumor diagnosis, and patients with
Here, we conducted a population-based cohort study malignant adrenal disease, primary aldosteronism, CS,
to (1) determine the prevalence and incidence of meta- androgen-producing adrenal tumor, pheochromocytoma,
bolic and cardiovascular disease and (2) assess mortality in or nonadenomatous lesions (eg, adrenal cysts or hemor-
a large, population-based sample of patients with adrenal rhage). Adrenal tumors were classified as benign adenomas
adenomas. based on imaging characteristics (unenhanced computed
3322  The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11

tomography tumor attenuation <10 Hounsfield units, was diagnosed if the postdexamethasone serum cortisol
rapid contrast washout, chemical shift on magnetic reson- concentration was >1.8 mcg/dL, whereas NFAT was diag-
ance imaging, lack of growth on imaging after 6 months), nosed if the postdexamethasone serum cortisol concentra-
clinical follow-up of at least 24 months or histopathology tion was ≤1.8 mcg/dL. Given the historical timeline of the
if surgically removed. The final cohort consisted of 1004 study and the population-based study design, we suspected
patients with adrenal adenomas and no suspected overt that a significant proportion of patients diagnosed with
hormone excess (Fig. 1). an adrenal adenoma would not have been evaluated by
Patients with adrenal adenomas were then 1:1 matched an endocrinologist and would not have undergone a DST.
by age (±1  year) and sex to referent subjects randomly Thus, patients who did not undergo the 1-mg DST and
selected from Olmsted County residents without diagnosis had no features of overt hormonal excess were classified as
codes for adrenal neoplasm. The index date was defined as having adenomas with unknown cortisol secretion.
the date of the earliest adrenal adenoma diagnosis based on

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

imaging for patients and the date of matching for referent
subjects. Assessment of Outcomes
To assess the negative predictive value of the screening Definitions of cardiometabolic disease
process for the referent cohort, a random sample of 60 ref- The outcomes of interest were cardiovascular risk factors
erent subjects was selected for validation. Manual medical and disease: hypertension, dyslipidemia, dysglycemia
record and imaging review was performed and confirmed (prediabetes or type 2 diabetes mellitus), chronic kidney
the absence of an adrenal adenoma in 60 out of 60 (100%) disease, cardiovascular events, cardiac arrhythmia, per-
of the referent subjects examined. ipheral vascular disease, and heart failure. We also assessed
the prevalence and incidence of venous thromboembolic
Biochemical classification of adrenal adenomas events. Cardiovascular events were a composite of myo-
Adrenal adenomas were classified based on the results of cardial infarction, coronary intervention, or ischemic
the overnight 1 mg dexamethasone suppression test (DST), stroke, whereas venous thromboembolic events were
consistent with current clinical guidelines (5,11). MACS a composite of deep venous thrombosis or pulmonary

Persons residing in Olmsted County, Minnesota

1995-2017

Adrenal tumor diagnosis by ICD-9 and No adrenal tumor diagnosis by ICD-9 and
ICD-10 codes ICD-10 codes
n=1,881

Medical records
missing, n = 22

Records reviewed for eligibility criteria

n=1,859 Excluded, n=855*
1:1 matching to patients with adrenal
• No diagnosis or procedural codes within adenomas by age and sex
2 years of the index date, n=25 n=1,004
• Age <20 years, n=13
• Wrong diagnosis code, n=358
• No apparent adrenal mass, n=124
• Diagnosis outside of study dates, n=11
• Living outside of study area at diagnosis,
n=79
• Malignant adrenal tumor, n=111
• Overt hormone secretion, n=53
• Pheochromocytoma, n=14
• Non-adenomatous lesions (e.g. cysts or
hemorrhage), n=85
Patients with benign adrenocortical tumor and
Referent subjects without adrenal tumor
no overt hormone excess
n=1,004
n=1,004

Figure 1.  Flowchart showing the identification of study participants. *Multiple participants (n = 18) met more than 1 exclusion criteria.
The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11 3323

embolism. Hypertension was defined as an office systolic (BMI), and smoking status (current or former smoker vs
blood pressure of ≥130  mmHg and/or diastolic blood nonsmoker).
pressure of ≥80 mmHg on 2 or more occasions obtained The cumulative incidence of cardiometabolic disease was
3  months apart, consistent with the American Heart estimated using the Aalen-Johansen approach, with death
Association guidelines (12); anti-hypertensive medi- considered as a competing risk event. We used Cox regres-
cation prescription; and/or documented International sion models to test for group differences univariately and
Classification of Diseases, 9th or 10th edition, diagnosis after adjustment for potential confounders (age, sex, BMI,
code. Dyslipidemia was defined by low-density lipo- smoking status, and prevalent cardiometabolic risk factors)
protein cholesterol of ≥130  mg/dL (3.4  mmol/L), lipid- using all events starting at 1 year following the index date.
lowering medication prescription, and/or documented Multiple imputation using the mice package in R was em-
diagnosis code. Prediabetes was defined by hemoglobin ployed to address any missing data for BMI and smoking
A1c of 5.7% to 6.4% on 2 or more separate occasions status (13). A  2-tailed probability value of P  <  0.05 was

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

in patients not on diabetes treatment and/or documented considered statistically significant for all tests. Statistical
diagnosis code. Diabetes mellitus was defined as hemo- analysis was conducted in SAS, version 9.4 (SAS Institute,
globin A1c ≥ 6.5% on 2 or more separate occasions and/ Cary, NC, USA) and R version 3.6.2 (R Foundation for
or documented diagnosis code. Chronic kidney disease, Statistical Computing, Vienna, Austria).
myocardial infarction, coronary intervention, ischemic
stroke, heart failure, cardiac arrhythmia, peripheral vas-
cular disease, pulmonary embolism, and deep venous Results
thrombosis were defined based on the relevant diagnosis Study Population and Prevalence of
or surgical procedural codes [see Supplementary Tables Cardiometabolic Disease
2 and 3 in (10)]. From 1995 to 2017, 1004 patients were diagnosed with
an adrenal adenoma, and 22% underwent biochemical
Calculation of prevalence and incidence of screening with the 1-mg DST. A total of 141 patients (14%)
cardiometabolic disease and mortality were diagnosed as NFAT, 81 (8%) as MACS, and the re-
The prevalence of cardiometabolic disease was assessed mainder (n = 782, 78%) as adenomas with unknown cor-
over a baseline period, consisting of the 6  months pre- tisol secretion. The median age of diagnosis was 63 years
ceding the index date, to minimize the risk of surveillance (range, 21-96), and 582 (58%) were women. Patients with
bias related to increased medical attention at the time of adrenal adenomas had higher BMI (31 kg/m2 vs 29 kg/m2,
abdominal imaging. The incidence of cardiometabolic P  <  0.001) and were more likely to be current or former
disease was assessed starting at 1  year following the smokers than age- and sex-matched referent subjects (70%
index date until death, migration out of the community, vs 54%, P < 0.001) (Table 1). In the 5 years preceding the
last contact with the REP, or adrenalectomy (n  =  30). index date, rates of abdominal cross-sectional imaging
Study participants with an event or comorbidity at base- were similar between the 2 cohorts (22% in cases and 20%
line or prior to the 1-year mark were excluded from the in referents).
follow-up analysis for the particular event or condition. At baseline, patients with adrenal adenomas had a
As surveillance bias is not a concern with mortality, higher unadjusted prevalence of hypertension (79% vs
overall mortality was assessed starting at the index date 65%, P < 0.001), dyslipidemia (66% vs 59%, P = 0.002),
for all study participants. dysglycemia (43% vs 28%, P < 0.001), myocardial infarc-
tion (8% vs 5%, P = 0.006), cardiovascular events (16%
vs 12%, P  =  0.008), cardiac arrhythmia (32% vs 23%,
Statistical Analysis P  <  0.001), peripheral vascular disease (21% vs 13%,
Continuous variables were presented using the median P  <  0.001), and heart failure (11% vs 6%, P  <  0.001)
and range, and comparisons were made between the compared to referent subjects without adrenal adenomas.
exposed and the unexposed groups using the Kruskal- After adjusting for age, sex, BMI, and smoking status, the
Wallis rank sum test. Categorical variables were pre- prevalence of hypertension (aOR 1.96, 95% CI 1.58-2.44),
sented using counts and proportions, and comparisons dyslipidemia (aOR 1.23, 95% CI 1.01-1.51), dysglycemia
were made using the chi-square test. We used logistic re- (aOR 1.63, 95% CI 1.33-2.00), myocardial infarction
gression models to test for baseline group differences in (aOR 1.50, 95% CI 1.02-2.22), peripheral vascular disease
the prevalence of cardiometabolic disease univariately, (aOR 1.59, 95% CI 1.32-2.06), and heart failure (aOR
and after adjustment for age, sex, body mass index 1.64, 95% CI 1.15-2.33) remained increased in patients
3324  The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11

Table 1.  Clinical characteristics of study participants with or without adrenal adenomas at baseline

Characteristic Adrenal adenoma (n = 1004) No adrenal adenoma (n = 1004) P-value

Age, median (range), years 63 (21-96) 63 (21-96) 0.976

Sex, female, n (%) 582 (58.0) 582 (58.0) 1.000
Race, white, n (%) 939 (93.5) 945(94.1) 0.578
Body mass index, mean (SD), kg/m2 31.2 (7.8) 29.0 (6.2) <0.001
Abdominal imaging within 5 years prior to index date, n (%) 223 (22.2) 200 (19.9) 0.208
Follow-up time, median (range), years 6.8 (0-21.9) 7.2 (0-22.0) —
Adrenocortical adenoma
  Unilateral adenoma, n (%) 876 (87.3) — —
  Adenoma size, median (range), mm 15 (5-85)a — —
  Serum cortisol following 1 mg DST, n (%)

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

  ≤1.8 mcg/dL 141 (14) — —
  >1.8 mcg/dL 81 (8) — —
   No DST performed 782 (78) — —
Cardiovascular risk factors, n (%)
  Current or former smoker 652 (70.1) 434 (53.6) <0.001
 Hypertension 770 (78.5) 614 (64.9) <0.001
 Dyslipidemia 644 (65.8) 575 (58.9) 0.002
 Prediabetes 146 (15.4) 101 (10.5) <0.001
  Diabetes mellitus 261 (27.5) 167 (17.4) <0.001
 Dysglycemiab 410 (43.1) 268 (28.0) <0.001
  Chronic kidney disease 65 (6.5) 52 (5.2) 0.216
Cardiovascular disease, n (%)
  Myocardial infarction 78 (7.8) 48 (4.8) 0.006
  Coronary intervention 75 (7.5) 56 (5.6) 0.086
  Ischemic stroke 66 (6.6) 55 (5.5) 0.302
  Cardiovascular eventsc 157 (15.6) 116 (11.6) 0.008
  Atrial fibrillation 91 (9.1) 73 (7.3) 0.142
  Cardiac arrhythmia 323 (32.2) 226 (22.5) <0.001
  Heart failure 105 (10.5) 60 (6.0) <0.001
  Peripheral vascular disease 208 (20.7) 133 (13.2) <0.001
Deep venous thrombosis 29 (2.9) 17 (1.7) 0.073
Pulmonary embolism 30 (3.0) 13 (1.3) 0.009
Deep venous thrombosis or pulmonary embolism 49 (4.9) 26 (2.6) 0.007

Abbreviations: DST, dexamethasone suppression test.

a
An 85-mm adrenal adenoma was associated with hemorrhage and subsequently decreased in size on follow-up imaging scans.
b
Dysglycemia: composite of prediabetes or diabetes mellitus.
c
Cardiovascular events: composite of myocardial infarction, coronary intervention, or ischemic stroke.

with adrenal adenomas compared to referent subjects atrial fibrillation [125/871 (16%) vs 93/911 (11%)],
(Table 2). cardiac arrythmia [242/594 (46%) vs 224/730 (33%)],
heart failure [148/849 (20%) vs 125/927 (13%)], and
peripheral vascular disease [226/711 (35%) vs 200/836
Incidence of Cardiometabolic Disease (25%)] (Fig. 2).
Patients with adrenal adenomas were followed for a In a multivariable model adjusted for age, sex, BMI,
median of 6.8  years (range, 0-22) and referent subjects smoking status, and prevalent cardiometabolic risk factors,
for 7.2  years (range, 0-22) after the index date. The adrenal adenomas were independently associated with an
unadjusted 10-year cumulative incidence of new meta- increased risk for chronic kidney disease [adjusted hazard
bolic and cardiovascular disease was higher in patients ratio (aHR) 1.46, 95% CI 1.14-1.86], cardiovascular
with adrenal adenomas compared to referent subjects: events (aHR 1.33, 95% CI 1.01-1.73), cardiac arrhythmia
dysglycemia [119/700 (18%) vs 88/818 (14%)], chronic (aHR 1.31, 95% CI 1.08-1.58), and peripheral vascular
kidney disease [167/891 (19%) vs 112/930 (12%)], car- disease (aHR 1.28, 95% CI 1.05-1.55) (Table 3). In add-
diovascular events [131/793 (19%) vs 200/863 (14%)], ition, patients with adrenal adenomas were more than
The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11 3325

Table 2.  Associations between adrenal adenomas and cardiometabolic diagnoses at baseline (case-control analyses)

Cardiometabolic conditions Unadjusted ORa (95% CI) Age and Sex-adjusted OR (95% CI) Model 1b OR (95% CI)

Hypertension 2.11 (1.74-2.57) 2.36 (1.91-2.91) 1.96 (1.58-2.44)

Dyslipidemia 1.34 (1.12-1.60) 1.39 (1.15-1.69) 1.23 (1.01-1.51)
Dysglycemiac 1.89(1.56-2.29) 1.92 (1.58-2.34) 1.63 (1.33-2.00)
Chronic kidney disease 1.27 (0.87-1.85) 1.28 (0.87-1.89) 1.06 (0.71-1.57)
Myocardial infarction 1.68 (1.16-2.43) 1.72 (1.18-2.51) 1.50 (1.02-2.22)
Coronary intervention 1.27 (0.96-1.95) 1.39 (0.96-2.01) 1.19 (0.82-1.74)
Ischemic stroke 1.21 (0.84-1.76) 1.22 (0.84-1.79) 1.19 (0.81-1.76)
Cardiovascular eventsd 1.42 (1.10-1.84) 1.48 (1.12-1.94) 1.33 (1.00-1.76)
Atrial fibrillation 1.27 (0.92-1.75) 1.30 (0.93-1.81) 1.16 (0.82-1.64)
Cardiac arrhythmia 1.63 (1.34-1.99) 1.72 (1.40-2.13) 1.55 (1.24-1.92)

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

Heart failure 1.84 (1.32-2.56) 1.92 (1.36-2.69) 1.64 (1.15-2.33)
Peripheral vascular disease 1.71 (1.35-2.17) 1.81 (1.41-2.33) 1.59 (1.32-2.06)
Deep venous thrombosis 1.72 (0.94-3.16) 1.73 (0.94-3.18) 1.70 (0.91-3.19)
Pulmonary embolism 2.35 (1.22-4.53) 2.35 (1.22-4.53) 1.97 (1.00-3.87)
Deep venous thrombosis or pulmonary embolism 1.93 (1.19-3.13) 1.94 (1.19-3.14) 1.70 (1.03-2.80)

Abbreviation: OR, odds ratio.

a
Odds ratios were calculated for each cardiometabolic diagnosis in patients with adrenal adenoma vs referent subjects.
b
Model 1: adjusted for age, sex, body mass index, and cigarette smoking (ever/never). Imputed data were used for participants missing body mass index or smoking
data (n = 89 for patients with adrenal adenomas and n = 252 for referent subjects).
c
Dysglycemia: composite of prediabetes or diabetes mellitus.
d
Cardiovascular events: composite of myocardial infarction, coronary intervention, or ischemic stroke.

twice as likely to develop a new deep venous thrombosis or with MACS compared to NFAT throughout the follow-up
pulmonary embolism (aHR 2.15, 95% CI 1.48-3.13) com- period (3% vs 2% at 5 years, 20% vs 9% at 10 years, and
pared to referent subjects. 37% vs 19% at 15 years) with an age, sex, BMI, smoking,
and prevalent cardiometabolic risk factors adjusted hazard
ratio of 2.01 (95% CI 0.92-4.41).
Mortality
During the study, 259 patients with adrenal adenomas Discussion
and 196 referent subjects died. The unadjusted overall
In a population-based setting, we found that patients with
mortality was higher in patients with adrenal aden-
adrenal adenomas had both a higher prevalence at base-
omas compared to referent subjects throughout the
line and a higher risk of developing de novo cardiovas-
follow-up period (15% vs 10% at 5 years, 28% vs 21%
cular events and comorbid conditions at follow-up than
at 10  years, and 41% vs 35% at 15  years; P  <  0.001.)
age- and sex-matched referent subjects. The majority of
(Fig. 2), but there were no significant differences in mor-
patients in our cohort did not undergo standard of care
tality risk after adjusting for age, sex, BMI, and smoking
hormonal workup, reflecting the timeline of the study, the
status (aHR 1.17, 0.96-1.42) and number of preva-
population-based study design, and the knowledge gaps
lent cardiometabolic risk factors (aHR 1.11, 95% CI
regarding the diagnostic workup of adrenal adenomas in
0.92-1.35).
the general medical community. This prevented us from re-
lating the findings to the presence and degree of MACS in
the study participants.
Comparison of PATIENTS with NFAT vs MACS This is the first study to assess the association of adrenal
We examined cardiometabolic outcomes in patients with adenomas with cardiometabolic outcomes in a population-
NFAT (n  =  141) vs MACS (n  =  81) for participants who based cohort. We found that patients with adrenal aden-
underwent biochemical screening for cortisol excess [see omas without overt hormone excess had a high baseline
Supplementary Tables 4 and 5 in (10)]. There were no prevalence of hypertension, dyslipidemia, dysglycemia,
notable differences in the prevalence and incidence of myocardial infarction, peripheral vascular disease, and
cardiometabolic disease between the two groups; however, heart failure, beyond that observed in age- and sex-matched
the number of incident events in the analyses was small. referent subjects without adrenal tumors, even after adjust-
The unadjusted overall mortality was higher in patients ment for known confounders. These findings are consistent
3326  The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11

Strata Adrenal adenomas Referent

A Dysglycemia B Chronic Kidney Disease

Dysglycemia (%)

60 60

CKD (%)
40 40

20 20

0 0
1 5 10 15 20 1 5 10 15 20
Years Following Index Date Years Following Index Date
AA 642 416 206 53 5 AA 888 582 292 78 7

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

Ref 773 546 262 87 4 Ref 930 671 332 101 9

C Cardiac Arrhythmia D Peripheral Vascular Disease

60 60
Arrhythmia (%)

PVD (%)
40 40

20 20

0 0
1 5 10 15 20 1 5 10 15 20
Years Following Index Date Years Following Index Date
AA 568 347 153 34 3 AA 696 451 209 48 5
Ref 730 479 209 59 4 Ref 836 553 269 81 7

E Myocardial Infarction or Coronary Intervention F Cardiovascular Event

60 60
CV Event (%)
MI or CI (%)

40 40

20 20

0 0
1 5 10 15 20 1 5 10 15 20
Years Following Index Date Years Following Index Date
AA 840 552 282 86 5 AA 791 513 255 76 5
Ref 911 648 318 104 8 Ref 863 604 295 95 8

G Venous Thromboembolic Events H Overall Mortality

60 60
Mortality (%)
PE/DVT (%)

40 40

20 20

0 0
1 5 10 15 20 1 5 10 15 20
Years Following Index Date Years Following Index Date
AA 898 613 304 86 4 AA 963 677 352 106 7
Ref 963 693 351 112 10 Ref 990 717 366 117 10

Figure 2.  Cumulative incidence of cardiometabolic outcomes and mortality during follow-up, starting at 1 year after the index date. Study partici-
pants with the outcome of interest at baseline or prior to the 1-year mark were excluded. (A) Dysglycemia (composite of prediabetes or diabetes
mellitus); (B) chronic kidney disease; (C) cardiac arrhythmia; (D) peripheral vascular disease; (E) myocardial infarction or coronary intervention; (F)
cardiovascular events (composite of myocardial infarction, coronary intervention, or ischemic stroke); (G) venous thromboembolic events (com-
posite of deep venous thrombosis or pulmonary embolism); and (H) overall mortality. Abbreviations: AA, adrenal adenoma group; CI, coronary
intervention; CKD, chronic kidney disease; CV event, cardiovascular event; DVT, deep venous thrombosis; MI, myocardial infarction; PE, pulmonary
embolism; PVD, peripheral vascular disease; Ref, referent group.
Table 3.  Adrenal adenomas and the risk for incident cardiometabolic outcomes

Outcome Adrenal adenoma No adrenal adenoma Model 1a HRb (95% CI) Model 2c HR (95% CI)

At risk, n Events, n 10-year cumulative incidence At risk, n Events 10-year cumulative incidence

Hypertension 109 50 0.86 271 137 0.76 0.81 (0.58-1.13) 0.81 (0.58-1.13)
Dyslipidemia 291 123 0.57 276 148 0.49 1.06 (0.82-1.37) 1.04 (0.80-1.35)
Dysglycemiad 700 119 0.18 818 88 0.14 1.41 (1.06-1.87) 1.32 (0.99-1.76)
Chronic kidney disease 891 167 0.19 930 112 0.12 1.57 (1.23-2.01) 1.46 (1.14-1.86)
Myocardial infarction 884 72 0.09 940 50 0.06 1.53 (1.06-2.22) 1.45 (1.02-2.13)
Coronary intervention 873 69 0.09 928 51 0.07 1.47 (1.02-2.13) 1.37 (0.94-1.98)
Ischemic stroke 890 83 0.10 930 63 0.08 1.28 (0.91-1.78) 1.18 (0.84-1.64)
Cardiovascular eventse 793 131 0.19 863 200 0.14 1.45 (1.11-1.89) 1.33 (1.01-1.73)
Atrial fibrillation 871 125 0.16 911 93 0.11 1.37 (1.04-1.79) 1.29 (0.98-1.69)
Cardiac arrhythmia 594 242 0.46 730 224 0.33 1.35 (1.12-1.63) 1.31 (1.08-1.58)
Heart failure 849 148 0.20 927 125 0.13 1.34 (1.05-1.70) 1.23 (0.96-1.58)
Peripheral vascular disease 711 226 0.35 836 200 0.25 1.35 (1.11-1.65) 1.28 (1.05-1.55)
DVT 921 74 0.09 970 35 0.04 2.10 (1.39-3.16) 2.07 (1.37-3.14)
The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11

PE 925 41 0.05 978 22 0.03 1.85 (1.09-3.13) 1.89 (1.11-3.21)

DVT or PE 898 90 0.16 963 43 0.05 2.16 (1.49-3.31) 2.15 (1.48-3.13)
Overall mortality 961 233 0.25 967 196 0.20 1.17 (0.96-1.42) 1.11 (0.92-1.35)

Abbreviations: DVT, deep venous thrombosis; HR, hazard ratio; PE, pulmonary embolism.
a
Model 1: adjusted for age, sex, body mass index, and cigarette smoking (ever/never).
b
Hazard ratios were calculated to assess the risk of incident outcomes in patients with adrenal adenomas vs referent subjects, starting at 1 year after the index date. Participants with the outcome of interest at baseline or prior
to the one-year mark were excluded from the analysis for the particular condition.
c
Model 2: adjusted for Model 1 variables plus number of prevalent cardiovascular risk factors (hypertension, dyslipidemia, dysglycemia, chronic kidney disease, peripheral vascular disease, heart failure, atrial fibrillation, and
cardiovascular events).
d
Dysgylcemia: composite of prediabetes or diabetes mellitus.
e
Cardiovascular events: composite of myocardial infarction, coronary intervention, or ischemic stroke.
3327

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

3328  The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11

with the high cardiometabolic disease burden reported in study) in 198 patients with adrenal adenomas and MACS
convenience samples of patients with NFAT and MACS (14) and of incident atrial fibrillation in 102 patients with
derived from university hospitals or specialized endocrin- adrenal adenomas and MACS (15) compared to patients
ology clinics (2,6,14-18). with NFAT.
The prevalence of cardiometabolic conditions in our In our cohort of 1004 patients, we found that patients
cohort was higher than the pooled prevalence reported in with adrenal adenomas had a higher incidence of de novo
the recent systematic review and meta-analysis on the nat- chronic kidney disease, cardiovascular events, cardiac ar-
ural history of adrenal incidentalomas (6). This difference rhythmia, and peripheral vascular disease than referent
may be due to several reasons. We used comprehensive subjects without adrenal tumors, even after adjustment for
definitions for the classification of hypertension, diabetes known confounders. The risk for incident dysglycemia was
mellitus, and dyslipidemia that incorporated laboratory lower than the 2-fold excess risk reported by Lopez et  al
results, medication prescription, and blood pressure meas- in patients with NFAT compared to patients without ad-

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

urements in addition to documented diagnosis codes. renal tumors (18), possibly due to the already high base-
Furthermore, the REP infrastructure allowed us to follow line prevalence of diabetes in our cohort. These findings
study participants across multiple care providers and care extend on the existing data from convenience samples of
settings, which likely resulted in more complete ascertain- patients to characterize the cardiometabolic risk associ-
ment of clinical outcomes. This capability may be particu- ated with adrenal adenomas on long-term follow-up in a
larly relevant for chronic diseases that are more likely to be population-based cohort.
diagnosed and followed in the primary care setting. We found that patients with adrenal adenomas also had
Patients with adrenal adenomas were more likely to a 2-fold increased risk of incident venous thromboembolic
be current or former smokers than age and sex-matched events compared to referent subjects without adrenal tu-
referent subjects (70% vs 54%). This finding is consistent mors. Because hypercoagulability is a well-known feature
with the results of 2 recent studies that examined the of CS (22), we hypothesize that venous thromboembolism
cross-sectional association of smoking with adrenal ad- risk occurs on a continuum, with higher rates of pulmonary
enomas (19,20). In particular, Olsen et  al found that pa- embolism and/or deep venous thrombosis with greater de-
tients with adrenal adenomas had a higher prevalence of grees of glucocorticoid autonomy. However, we were un-
current smoking compared to the general population and able to confirm this hypothesis in our study due to the low
that smoking was associated with larger adrenal aden- rates of hormonal screening.
omas, more bilateral adrenal disease, and increased rates In the 222 (22%) patients with adrenal adenomas and
of autonomous cortisol secretion (19). Collectively, our available DST results, we did not observe a meaningful
studies suggest that smoking is a potential risk factor for difference in incident cardiometabolic disease or venous
the development of adrenal adenomas. Case selection bias thromboembolism when patients with NFAT were com-
could be an alternative explanation, because smokers may pared to patients with MACS. However, the number of in-
be more likely to undergo imaging evaluation for various cident events in the analyses was smaller than that reported
reasons than non-smokers. However, the frequency of in a convenience sample of comparable size (14), which po-
cross-sectional imaging was similar in our adrenal ad- tentially limited our ability to detect an effect. In the 782
enoma and referent cohorts in the 5  years preceding the (78%) patients who did not undergo biochemical screening
index date. Smoking is also an established risk factor for with the 1-mg DST, a significant proportion may have had
the development of cardiovascular disease and was ad- MACS based on estimates of hormonal excess in previously
justed for in our analysis. published adrenal adenoma studies (2,5). In particular, a
The existing data on the incidence of cardiometabolic recent study found that 50% of patients with incidentally
outcomes in patients with adrenal adenomas are more detected adrenal adenomas had postdexamethasone serum
limited than the data about prevalence. There are only a cortisol concentrations >1.8 mcg/dL in a prospective co-
few longitudinal studies with extended follow-up time and hort (2). Thus, unrecognized cortisol excess may have con-
most are smaller in size and lack referent subjects without tributed to the increased cardiometabolic burden observed
adrenal tumors, making comparisons to our findings dif- in the adrenal adenoma group. However, we could not re-
ficult (14-16,18,21). Among the larger studies, Lopez late our findings to the presence and degree of mild cor-
et  al found an increased risk of incident composite dia- tisol secretion because of the low number of patients who
betes (prediabetes or diabetes) among 166 patients with underwent appropriate hormonal evaluation.
NFAT compared to patients without adrenal tumors (18). We found that the unadjusted overall mortality was
Di Dalmazi et  al reported a higher risk of incident car- higher in patients with adrenal adenomas compared to ref-
diovascular events (equivalent composite outcome as our erent subjects without adrenal tumors. However, mortality
The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11 3329

did not differ after adjusting for smoking and cardiovas- study participants who had the outcome of interest at base-
cular morbidity, suggesting that these factors are potential line or prior to the 1-year mark. In addition, we examined
contributors to the excess mortality risk in patients with the frequency of cross-sectional imaging in cases and refer-
adrenal adenomas. One previous study from an academic ents in the 5 years preceding the index date and found no
medical center found an increased all-cause mortality risk major differences. Second, because Olmsted County resi-
(aHR 1.14, 95% CI 1.003-1.29) in 969 patients with ad- dents are predominately Caucasian (94%), our findings
renal incidentalomas compared to controls without adrenal may not apply to populations with different ethnic and
tumors (23). However, several important covariates, such socioeconomic characteristics. Third, while the 1-mg DST
as smoking status, were not considered in their analyses. is the guideline recommended test to screen for cortisol ex-
A  few studies have also reported increased mortality in cess, additional data may be needed to verify the diagnosis
patients with adrenal adenomas based on post-DST cor- of autonomous cortisol secretion. In particular, the DST
tisol concentrations (14,17,24), but a systematic review may have false-positive results due to noncompliance,

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

and meta-analysis of 9e studies found that mortality was variability in dexamethasone absorption and metabolism,
similar between patients with MACS and NFAT on a mean interference of certain medications, variability in cortisol
follow-up of 5.5 years (6). assays and cutoffs used, and other factors. Last, there may
The strengths of this study include the large sample size, be residual confounding in our analyses, including unrec-
extended longitudinal follow-up, and ability to capture ognized primary aldosteronism, despite adjustment for
clinical outcomes across health systems and care settings important confounders in our models.
through the REP infrastructure. The population-based study In summary, our findings suggest that adrenal ad-
design avoids the selection and referral biases of published enomas are associated with a high prevalence of
studies from university hospital or specialized endocrinology cardiometabolic disease at the time of diagnosis and
clinics and increases the generalizability of our results. with an increased incidence during follow-up of de novo
The first limitation of this study was that a minority of cardiovascular events, peripheral vascular disease, car-
patients (22%) underwent diagnostic workup with the 1-mg diac arrhythmia, chronic kidney disease, and venous
overnight DST. The low rate of hormone screening suggests thromboembolic events compared to referent subjects
a potential gap in the care of patients with adrenal aden- without adrenal tumors. Overall mortality did not differ
omas. Although the onset of the study preceded the first between the 2 groups after adjusting for the higher fre-
published guidelines on the management of incidentally quency of baseline smoking and cardiometabolic disease
discovered adrenal tumors in 2002 (25), rates of hormonal in patients with adrenal adenomas. Finally, screening
workup continued to remain low (3) even after guidelines rates for hormone excess were low in our population-
were readily available (5,11,25,26). This could be due to based cohort, suggesting knowledge gaps in the workup
several reasons. First, most patients in our cohort were man- of adrenal adenomas among general health care pro-
aged by nonspecialists, who may not be familiar with the viders. Patients with adrenal incidentalomas should
standard adrenal workup. Second, adrenal adenomas may undergo timely screening for hormone excess at the time
not be given appropriate attention, because providers and of diagnosis and assessment for cardiometabolic risk
patients are more focused on the nonadrenal problem that during follow-up.
prompted the imaging scan. However, timely evaluation
for subtle cortisol excess is important because of the poten-
tial cardiometabolic consequences associated with adrenal Acknowledgments
adenomas. Funding Support: This research was supported by the National
Several other limitations apply to this study. First, Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)
of the National Institutes of Health (NIH) USA under award
there may be an inherent bias related to the need
K23DK121888 (to I.B). The study was conducted using the Roch-
for cross-sectional imaging in patients with adrenal ester Epidemiology Project medical record-linkage system, which
incidentalomas. Individuals who undergo imaging evalu- is supported by the National Institute on Aging of the NIH under
ation may be sicker or more likely to receive medical care award numbers R01 AG034676 and AG052425. The views ex-
than those who do not. Thus, the association observed be- pressed are those of the author(s) and not necessarily those of the
National Institutes of Health USA.
tween adrenal adenomas and cardiometabolic risk may in
part be attributed to the underlying medical disorder that
prompted the initial cross-sectional imaging, or to sur- Additional Information
veillance bias related to closer medical follow-up during Correspondence: Irina Bancos, MD, Division of Endocrinology,
that time. To minimize these biases, we assessed incidence Diabetes, Metabolism and Nutrition, Mayo Clinic, 200 First Street
starting at 1 year following the index date and excluded SW, Rochester, MN 55905, USA. Email: [email protected].
3330  The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 11

Disclosures: I.B. reports advisory board participation and/or con- Management of High Blood Pressure in Adults: A Report of the
sulting with Corcept Therapeutics, Strongbridge, Sparrow Pharma- American College of Cardiology/American Heart Association
ceutics, and HRA Pharma outside the submitted work. All other au- Task Force on Clinical Practice Guidelines. Hypertension.
thors have nothing to disclose. 2018;71(6):e13-e115.
Data Availability: Some or all data sets generated during and/ 13. van Buuren S, Groothuis-Oudshoorn K. mice: multivariate im-
or analyzed during the current study are not publicly available but putation by chained equations in R. J Stat Softw. 2011;45(3):67.
are available from the corresponding author on reasonable request. 14. Di Dalmazi G, Vicennati V, Garelli S, et al. Cardiovascular events
and mortality in patients with adrenal incidentalomas that are
either non-secreting or associated with intermediate phenotype
References or subclinical Cushing’s syndrome: a 15-year retrospective study.
1. Bovio  S, Cataldi  A, Reimondo  G, et  al. Prevalence of adrenal Lancet Diabetes Endocrinol. 2014;2(5):396-405.
incidentaloma in a contemporary computerized tomography 15. Di Dalmazi G, Vicennati V, Pizzi C, et al. Prevalence and incidence
series. J Endocrinol Invest. 2006;29(4):298-302. of atrial fibrillation in a large cohort of adrenal incidentalomas:

Downloaded from https://academic.oup.com/jcem/article/106/11/3320/6311110 by guest on 07 March 2023

2. Reimondo  G, Castellano  E, Grosso  M, et  al. Adrenal
a long-term study. J Clin Endocrinol Metab. 2020;105(8):e2770
incidentalomas are tied to increased risk of diabetes: find- -e2777.
ings from a prospective study. J Clin Endocrinol Metab. 16. Petramala  L, Olmati  F, Concistrè  A, et  al. Cardiovascular and
2020;105(4):dgz284. metabolic risk factors in patients with subclinical Cushing.
3. Ebbehoj  A, Li  D, Kaur  RJ, et  al. Epidemiology of adrenal tu- Endocrine. 2020;70(1):150-163.
mours in Olmsted County, Minnesota, USA: a population-based 17. Debono M, Bradburn M, Bull M, Harrison B, Ross RJ, Newell-
cohort study. Lancet Diabetes Endocrinol. 2020;8(11):894-902. Price J. Cortisol as a marker for increased mortality in patients
4. Song  JH, Chaudhry  FS, Mayo-Smith  WW. The incidental ad- with incidental adrenocortical adenomas. J Clin Endocrinol
renal mass on CT: prevalence of adrenal disease in 1049 con- Metab. 2014;99(12):4462-4470.
secutive adrenal masses in patients with no known malignancy. 18. Lopez D, Luque-Fernandez MA, Steele A, Adler GK, Turchin A,
AJR Am J Roentgenol. 2008;190(5):1163-1168. Vaidya A. “Nonfunctional” adrenal tumors and the risk for inci-
5. Fassnacht  M, Arlt  W, Bancos  I, et  al. Management of ad-
dent diabetes and cardiovascular outcomes: a cohort study. Ann
renal incidentalomas: European Society of Endocrinology Intern Med. 2016;165(8):533-542.
Clinical Practice Guideline in collaboration with the European 19. Olsen H, Kjellbom A, Löndahl M, Lindgren O. High prevalence
Network for the Study of Adrenal Tumors. Eur J Endocrinol. of smoking in patients with adrenal incidentalomas: causality or
2016;175(2):G1-G34. case selection? Eur J Endocrinol. 2020;183(3):335-341.
6. Elhassan YS, Alahdab F, Prete A, et al. Natural history of adrenal 20. Yousaf  A, Patterson  J, Hobbs  G, et  al. Smoking is associated
incidentalomas with and without mild autonomous cortisol ex- with adrenal adenomas and adrenocortical carcinomas: a na-
cess: a systematic review and meta-analysis. Ann Intern Med. tionwide multicenter analysis. Cancer Treat Res Commun.
2019;171(2):107-116. 2020;25:100206.
7. St Sauver JL, Grossardt BR, Yawn BP, et al. Data resource pro- 21. Morelli V, Reimondo G, Giordano R, et al. Long-term follow-up
file: the Rochester Epidemiology Project (REP) medical records- in adrenal incidentalomas: an Italian multicenter study. J Clin
linkage system. Int J Epidemiol. 2012;41(6):1614-1624. Endocrinol Metab. 2014;99(3):827-834.
8. Sauver  JLS, Grossardt  BR, Leibson  CL, Yawn  BP, Melton  LJ 22. Wagner  J, Langlois  F, Lim  DST, McCartney  S, Fleseriu  M.

III, Rocca  WA. Generalizability of epidemiological findings Hypercoagulability and risk of venous thromboembolic events
and public health decisions: an illustration from the Rochester in endogenous Cushing’s syndrome: a systematic meta-analysis.
Epidemiology Project. Mayo Clin Proc. 2012;87(2):151-160. Front Endocrinol (Lausanne). 2018;9:805.
9. Rocca WA, Yawn BP, St Sauver JL, Grossardt BR, Melton LJ 3rd. 23. Taya M, Paroder V, Bellin E, Haramati LB. The relationship be-
History of the Rochester Epidemiology Project: half a century tween adrenal incidentalomas and mortality risk. Eur Radiol.
of medical records linkage in a US population. Mayo Clin Proc. 2019;29(11):6245-6255.
2012;87(12):1202-1213. 24. Patrova J, Kjellman M, Wahrenberg H, Falhammar H. Increased
10. Zhang  CD, Li  D, Kaur  RJ, et  al. Supplemental data:
mortality in patients with adrenal incidentalomas and autono-
cardiometabolic outcomes and mortality in patients with ad- mous cortisol secretion: a 13-year retrospective study from one
renal adenomas: a population-based cohort study. Uploaded center. Endocrine. 2017;58(2):267-275.
April 13, 2021. Figshare. doi:10.6084/m9.figshare.14408573.v3 25. NIH state-of-the-science statement on management of the clinic-
11. Vaidya A, Hamrahian A, Bancos I, Fleseriu M, Ghayee HK. The ally inapparent adrenal mass (“incidentaloma”). NIH Consensus
evaluation of incidentally discovered adrenal masses. Endocr State Sci Stat. 2002;19(2):1-25.
Pract. 2019;25(2):178-192. 26. Zeiger M, Thompson G, Duh Q-Y, et al. American Association
12. Whelton  PK, Carey  RM, Aronow  WS, et  al. 2017 ACC/AHA/ of Clinical Endocrinologists and American Association of
AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Endocrine Surgeons medical guidelines for the management of
Guideline for the Prevention, Detection, Evaluation, and adrenal incidentalomas. Endocr Pract. 2009;15(Suppl 1):1-20.