Frequent antecedent nodules and diagnostic gaps in adrenocortical carcinoma

Abstract

Context

Adrenocortical carcinoma (ACC) is a rare and often fatal malignancy currently only cured by surgery. Large-scale studies have failed to capture benign to malignant transition events, and patients often present with metastatic disease. These observations suggest ACC develops rapidly without a measurable precursor lesion, forming the basis of widely implemented adrenal incidentaloma guidelines.

Objective

We aimed to characterize the prevalence of antecedent adrenal nodules, quality of preoperative diagnostic workup, and the impact on survival for adults with ACC.

Methods

We analyzed clinical data from 108 consecutive patients referred to a single tertiary academic medical center from 2000 to 2025.

Results

Median follow-up was 86.7 months. Strikingly, 31.7% harbored antecedent nodules, indolent for a median 4.68 years before ACC presentation, and ultimately associated with a higher risk of death (adjusted hazard ratio [HR], 2.9; 95% CI, 1.39-6.0). Regarding preoperative evaluation, 46.4% were referred to endocrinology, associated with more complete hormonal assessment; 31.4% had adrenal biopsy, accompanied by increased risk of progression (adjusted HR, 2.53; 95% CI, 1.24-5.2) and death (adjusted HR, 2.22; 95% CI, 1.07-4.6). Patients with high-grade disease received more interventions; those who received mitotane and/or locoregional therapies had improved survival (adjusted HR, 0.31; 95% CI, 0.14-0.71, and adjusted HR, 0.47; 95% CI, 0.23-0.98, respectively).

Conclusion

The unexpectedly high prevalence of indolent antecedent adrenal nodules suggests a prolonged, clinically premalignant phase in a subset of ACC. Initial diagnostic evaluation was often incomplete or inappropriately invasive, compromising the preoperative window and patient survival. These findings challenge current surveillance paradigms, revealing an extended but vulnerable window for surgical cure.

Adrenocortical carcinoma (ACC) is a rare and deadly cancer of the adrenal cortex [1]. Complete (R0) surgical resection is a potent predictor of survival [2, 3], serving as the only therapy with potential to cure [4]. Though more than half of all patients are diagnosed with locoregional disease, up to 75% will eventually develop incurable metastases [5]. Standard-of-care medical therapies for ACC include the adrenolytic agent mitotane with or without cytotoxic chemotherapy; in the adjuvant setting, treatment is guided by histologic features such as proliferation-based grade [6-8]. In recent years, the therapeutic landscape has expanded to include immunotherapy and tyrosine kinase inhibitors [9-12]. Patients may also benefit from adjuvant radiation, cytoreduction, and procedures targeting oligometastatic disease [5, 13-19]. These advances are gradually improving outcomes for patients, though prognosis remains dismal [5, 20].

The rarity of ACC poses significant challenges to diagnosis and management. Accurate diagnosis requires complete hormonal and imaging evaluation, necessitating a high initial index of suspicion and clinical expertise. Many adrenal tumors are discovered incidentally [21, 22], and clinicians may default to more common etiologies (eg, nonfunctional adenoma) when data are incomplete. Even when imaging suggests malignancy, the relatively higher likelihood of metastasis and pressure to obtain a tissue diagnosis frequently lead to percutaneous adrenal biopsy, a risky procedure with limited diagnostic yield, theoretical risk of biopsy tract seeding, and potential to delay definitive surgery [23-28]. Phase III randomized controlled trials for ACC are limited [6, 8, 29], and individual treatment decisions are driven by expert opinion [4, 30, 31].

Optimal care requires early referral to specialized centers with expertise in endocrine oncology and adrenal surgery, where multidisciplinary teams can provide nuanced diagnostic assessment and access to specialized therapies [4, 32]. The curative potential of primary resection is tied to surgeon experience [33]. Even in high-quality care settings, endocrine diseases remain a major training and practice gap for most nonendocrinologist physicians [34-36]. This is especially critical for ACC, where sequelae of glucocorticoid excess contribute to poor clinical outcomes and frequently occur in the most aggressive ACC subtype [1, 37-39]. As patients often initially present to generalists and community providers, these features underscore the importance of frontline education to enable accurate diagnosis and early referral.

In a retrospective, real-world cohort of patients with ACC referred to a tertiary academic medical center in the midwestern United States, we examined factors shaping practice patterns and clinical outcomes. We investigated the prevalence and growth kinetics of ipsilateral adrenal nodules that predated ACC presentation. We assessed patterns in diagnostic evaluation, including rates of endocrinology referral, completeness of initial hormonal assessment, and implementation of adrenal biopsy, and the impact on clinical outcomes. We characterized hormone secretion patterns and clinical correlates, measured the prognostic value of tumor molecular profiling, and evaluated the impact of therapeutic interventions on survival. Our findings highlight the need for broader dissemination of evidence-based guidance for fast-track adrenal tumor evaluation and carry implications for the management of adrenal incidentalomas.

Methods

Patients

Our study includes an unselected series of 108 patients with ACC diagnosed in adulthood (>18 years old) who were referred to Washington University School of Medicine/Barnes-Jewish Hospital from 2000 to 2025. Patients were identified through clinical practice and institutional electronic medical record-based cohort discovery tools. We performed a retrospective chart review to collect clinical and pathologic data. Diagnosis of ACC was established through assessment by a clinical expert in pathology, oncology, and/or endocrinology. The study was conducted in accordance with the Declaration of Helsinki, with the study protocol approved by the Washington University School of Medicine Institutional Review Board.

Definition of the preoperative period

For simplicity, here we refer to the time period prior to the first intervention for ACC (surgery or medical intervention for those who did not obtain surgery) as the preoperative period.

Hormonal and grade classification

Tumor hormonal status was classified using clinical and biochemical data collected prior to the first intervention for ACC as detailed in the supplemental material [40]. Tumor grade (consensus) was determined by combining information from Ki67 staining and mitotic counts. Ki67 < 10% was classified as low grade, 10% to 19% as intermediate grade, and ≥20% as high grade. Mitotic counts <20/50 high-power field (HPF) were classified as low grade, and counts ≥20/50 HPF were classified as high grade. A sample with both measurements was classified into the highest grade strata. Weiss scores were used to support the diagnosis of malignancy but were not used to determine tumor grade.

Statistical analyses

We used nonparametric statistical tests (Wilcoxon test or Kruskal-Wallis with or without Dunn's post-hoc test) to compare continuous data from 2 or more groups, and the χ² test to compare distributions of categorical variables. We used Kaplan-Meier analysis with log-rank test or Cox proportional hazards regression models to measure the impact of clinical variables on survival. P < .05 (or, where applicable, adjusted P < .05) was considered significant for all analyses. All statistical analyses were performed, and data were plotted using R [41] and a variety of packages, as detailed in the supplemental material [40].

Results

Cohort characteristics

Clinical characteristics of our institutional cohort are summarized in Table 1 and Fig. 1. Median follow-up time was 86.7 months; overall survival, 48.7 months; progression-free survival, 11.9 months; and recurrence-free survival after R0 resection, 27.5 months (Table 1). This dataset is skewed toward more males, older patients, functional tumors, and high-grade disease compared with representative multi-institutional academic cohorts (Fig. 1A-1D) [37, 38, 42]; there were no differences in the distribution of stage at diagnosis (P = .056, χ²). Higher stage and grade were independently associated with decreased progression-free and overall survival, as expected (Fig. S1A-S1E [40]). Oncocytic histology was not associated with differences in survival (log-rank P > .05).

Table 1.

Clinical characteristics

WashU ACC cohort, n = 108
Sex, n (%)
Female	55 (50.9)
Male	53 (49.1)
Age, years
Median (range)	58.3 (25.1-89.8)
Clinical setting at diagnosis, n (%)
Academic medical center	24 (22.2)
Academic affiliate	12 (11.1)
Government affiliate	3 (2.78)
Community	65 (60.2)
Unknown	4 (3.7)
Hormone excess, n (%)
Silent	14 (13)
Functional	69 (63.9)
Unknown	25 (23.1)
Presenting symptoms, n (%)
Constitutional	7 (6.48)
Hormone excess	34 (31.5)
Mass effect	63 (58.3)
Incidental	30 (27.8)
Unknown	2 (1.85)
ENSAT/AJCC, n (%)
Stage I	5 (4.63)
Stage II	33 (30.6)
Stage III	40 (37)
Stage IV	29 (26.9)
Unknown	1 (0.926)
Grade (consensus), n (%)
Low	17 (15.7)
Intermediate	11 (10.2)
High	54 (50)
Unknown	26 (24.1)
Histology, n (%)
Conventional	90 (84.1)
Oncocytic	15 (14)
Sarcomatoid	2 (1.87)
Primary resection status, n (%)
R0	55 (50.9)
R1	20 (18.5)
R2	1 (0.926)
RX	12 (11.1)
Unsuccessful	2 (1.85)
Not attempted	18 (16.7)
Follow-up time, months
Median (95% CI)	86.7 (59.8-137.6)
Overall survival, months
Median (95% CI)	48.7 (27.3-87.2)
Progression-free survival, months
Median (95% CI)	11.9 (7.23-24)
Recurrence-free survival after R0, months
Median (95% CI)	27.5 (22.5-Inf)
Pathogenic germline alteration, n (%)
No	31 (28.7)
Yes	6 (5.55)
Unknown	71 (65.7)
Smoker status at diagnosis, n (%)
Never	49 (45.4)
Former	32 (29.6)
Active	26 (24.1)
Unknown	1 (0.926)

Recurrence-free survival after R0 only includes patients without stage IV disease at diagnosis.

Abbreviations: ACC, adrenocortical carcinoma; ENSAT/AJCC, American Joint Committee on Cancer/European Network for the Study of Adrenal Tumors; Inf, infinity; WashU, Washington University School of Medicine/Barnes-Jewish Hospital.

Figure 1.

Cohort characteristics and benchmarking to published datasets. We compared demographic data from this cohort (WashU) to a combination of datasets published in the literature [37, 38, 42]. Reference datasets were chosen based on the following criteria: from expert tertiary care centers, included data points of interest, and provided sufficient documentation to support mutual exclusivity (including with the WashU cohort). The WashU cohort is biased in favor of older male patients (A, B), hormone excess (C), and high-grade disease (D). Patients typically presented in community hospital settings and were referred to academic medical centers for surgery (E). The WashU cohort had comparable R0 resection rates to reference datasets (F); patients with metastatic disease were excluded here, as 1 reference dataset reported resection status only for locoregional disease. Including all patients who received surgical resection of the primary tumor in the WashU cohort, R0 resection rates were higher at academic centers (G). In A, C, D, and F, the number of patients in each group is specified as “n = …” above each bar and includes all patients for whom relevant data were available. In B, violin plots display continuous data distributions with embedded box plots (point, patient; line, median; box, interquartile range; whiskers, 1.5× interquartile range). Statistical tests with P values are reported below each panel; for pairwise tests of continuous variables, P values are also signified with brackets where ****P < .0001, ***P < .001, **P < .01, *P < .05, and P ≥ .05 (not significant).

Abbreviation: WashU, Washington University School of Medicine/Barnes-Jewish Hospital.

Less than 20% of patients with germline genetic testing possessed a pathogenic variant in a tumor susceptibility gene, consistent with other studies (P = .48, χ²) [37]. More than half of patients had current or former cigarette use, a known risk factor for adrenal tumors [43-45]; 24.1% were active smokers at diagnosis, comparable to national averages (Table 1) [46]. Those more recently diagnosed with ACC were less likely to be active smokers (diagnosed 1996-2018 vs 2019-2025, P = .0298, χ²) despite older age at diagnosis (P = .0164, Wilcoxon). Males had more smoking exposure measured in pack-years (P = .0074, Wilcoxon).

A total of 60.2% of patients presented in a community healthcare setting (Table 1); in contrast, 77% had primary surgical resection at an academic medical center (Fig. 1E). Over 60% of patients who had surgery received R0 resection, comparable to published cohorts (Fig. 1F); R0 resection was more likely in academic settings (Fig. 1G). In 16.7% of patients, surgery was not attempted (Table 1). Hospital setting at diagnosis or surgery was not associated with differences in progression-free or overall survival (P > .05 for all models, Cox regression).

Antecedent nodules

Having established baseline characteristics of our cohort, we next investigated the evolutionary timeline of patients’ tumors. Patients reached clinical attention due to any combination of mass effect, hormone excess, and constitutional symptoms; in 27.8% of cases, ACC was otherwise asymptomatic and discovered incidentally on imaging (Table 1), which is in line with the literature [22, 47]. Most incidentalomas are benign [22, 48], and malignant transformation to ACC after a prolonged indolent period is thought to be exceedingly rare [21, 49], though it has been observed [50].

Unexpectedly, on review of cross-sectional imaging, we identified an antecedent ipsilateral adrenal nodule in 31.7% of patients (Fig. 2A; antecedent nodules defined as in Fig. S2A [40]). This comprised 32/54 (59.2%) of patients for whom adrenal imaging prior to diagnosis of ACC was available. Compared with those without antecedent nodules, these patients were nearly a decade older at ACC diagnosis (Fig. 2B), were less likely to present with symptoms of hormone excess or mass effect (P = .035, χ²), and had smaller primary tumors (Fig. 2C). There were no significant differences in major characteristics including sex, stage, hormonal activity, smoking exposure, grade, and oncocytic histology (P > .05 for all comparisons, χ² or Wilcoxon test). Antecedent nodules were evenly distributed across patients with earlier and more recent diagnoses of ACC (diagnosed 1996-2018 vs 2019-2025; P = .661, χ²).

Figure 2.

Antecedent nodules were prevalent, indolent for years, and associated with a higher risk of death. (A) Among patients for whom adequate data at the time of diagnosis were available, 31.7% had documentation of an antecedent, ipsilateral adrenal nodule. (B and C) Patients with adrenocortical carcinoma (ACC) derived from antecedent adrenal nodules were diagnosed with ACC at older ages with smaller primary tumors. (D) Top, line plot depicts the growth of adrenal tumors over time, colored by individual patients; the final point reflects the maximum dimension documented on the surgical specimen or the last radiologic timepoint prior to nonsurgical intervention for ACC. Normalized time is calculated as years elapsed since the initial time point; median, 4.68 years. Bottom, swimmer plot depicts radiology read corresponding to each time point. (E) Final growth is calculated as the change in maximum dimension between the final radiologic time point and baseline, divided by total elapsed time in years. The horizontal line is drawn at 0.5 cm/year. Growth rate of ACC derived from antecedent nodules was greater than that of longitudinally observed contralateral benign adrenocortical tumors (ACT), with significant overlap below the horizontal line. (F) Final growth is calculated as the percent change in maximum dimension at the final radiologic time point relative to baseline. The horizontal line is drawn at 20% tumor growth. Here, growth rates of ACC are highly variable compared to ACT, though with no significant statistical difference and significant overlap above the horizontal line. (G) Forest plot depicting Cox regression. Though ACC derived from antecedent nodules was not predictive of survival in univariate analysis, it was associated with significantly higher risk of death in multivariable models, including stage and grade. Shading specifies variables included in the model; point, hazard ratio (HR); whiskers, 95% CI. The reference variable in each strata has a hazard ratio = 1 and no P value annotated.

The median latency period (defined as the time between first documentation of antecedent adrenal nodule and date of ACC diagnosis) was 4.68 years (range, 0.20-16.5 years). Tumor growth patterns were heterogeneous, and 32.2% were suspected benign during tumorigenesis (Fig. 2D). Though in a minority of cases, benign nature was suspected using prior radiology reports, in 78.3% of reads, radiologists applied validated radiologic criteria to support this assessment (Fig. S2B [40]).

A subset of patients also had contralateral benign adrenocortical tumors (ACT), subject to prolonged surveillance and evenly distributed across patients with and without antecedent ipsilateral nodules (Fig. S2C [40]). There were significant differences in the growth rates of ACT and ACC arising from antecedent nodules (Fig. 2E and 2F; Fig. S2D and S2E [40]). However, applying typical imaging criteria to triage risk of malignancy (eg, monitor for an increase in tumor size by 0.5 cm or 20% [21]) would neither capture all antecedent ACC nor distinguish ACC from ACT, particularly with extended observation windows (Fig. 2E and 2F; Fig. S2D and S2E [40]).

ACC arising from antecedent nodules was not associated with a higher hazard for progression or death in univariate analysis; however, in multivariable models including stage and grade, antecedent ACC and longer latency periods were associated with a higher risk of death, even when corrected for age (Fig. 2G; Fig. S2F [40]).

Preoperative diagnostic workup

As a substantial proportion of patients experienced a prolonged indolent period prior to ACC presentation, the preoperative window represents a crucial opportunity for potentially curative intervention. For >90% of patients, imaging at the time of ACC presentation had features that were unequivocally concerning for malignancy (eg, >4 cm, heterogeneous, >20 HU, ¹⁸F-fluorodeoxyglucose-avid, and/or irregular boundaries). Patients with imaging features of malignancy were more likely to have tumor-associated symptoms (P = .012, χ²). Preoperative endocrinology referral was more likely among those with presenting symptoms attributable to hormone excess (P = .0036, χ²) or physical exam findings of Cushing syndrome (P = .033, χ²).

Ultimately, only 46.4% of patients were seen preoperatively by endocrinology (Fig. 3A), regardless of hospital setting at diagnosis or surgery (P > .05, χ²). While 84.4% of patients had adequate initial hormonal workup for pheochromocytoma (collection of metanephrines), only 53.4% of patients had adequate workup for ACC (measurement of cortisol and dehydroepiandrosterone sulfate for all patients, and aldosterone for patients with hypertension). Endocrinology evaluation was associated with a more complete hormonal workup for both entities (Fig. 3B and 3C). There was no association between the nature of presenting symptoms and adequacy of hormonal workup for ACC (P = .363, χ²).

Figure 3.

Nature of diagnostic workup prior to first intervention for adrenocortical carcinoma (ACC). (A) Most patients were not referred to endocrinology prior to the first intervention. (B and C) Endocrinology referral was associated with a more complete hormonal workup for pheochromocytoma and ACC. The proportion of patients with adequate workup is indicated above the bars. Adequate pheochromocytoma evaluation included the collection of metanephrines; adequate ACC evaluation included measurement of cortisol and dehydroepiandrosterone sulfate for all patients and aldosterone for patients with hypertension. (D) Types of biopsies prior to first intervention, with inset showing pathology reads of adrenal biopsies. (E) Reasons for adrenal biopsy were often inappropriate. Primary indicates primary team decision. Intra-op indicates being obtained during unsuccessful primary resection. (F) Adrenal biopsies were obtained irrespective of adequate pheochromocytoma workup. (G) Adrenal biopsy was associated with a higher risk of progression and death as a univariate and remained significant in multivariable models with stage and grade.

Abbreviations: ACA, adrenocortical adenoma; ACT, adrenocortical tumor.

Preoperative biopsies were obtained in 39% of patients; 80.5% of these were adrenal biopsies, of which 30.3% failed to diagnose ACC (Fig. 3D). Adrenal biopsies were obtained for a variety of reasons, although 63.6% of the time they were obtained inappropriately (Fig. 3E). Of some concern, adrenal biopsies were obtained regardless of adequacy of preoperative pheochromocytoma workup (Fig. 3F). There was no association between adrenal biopsy and preoperative endocrinology referral, hospital setting at diagnosis or surgery, or imaging features (P > .05 for all comparisons, χ²). Adrenal biopsy was associated with a higher risk of progression and death in univariate and multivariable models (Fig. 3G). Notably, adrenal biopsies were often performed in patients with advanced disease (84.8% stage III-IV; P = .00616, χ²), including those who were not surgical candidates (Fig. 3E).

Hormone secretion patterns

We assigned hormonal status to patients using collected clinical and laboratory data as detailed in the supplemental materials [40]. The distribution of major hormonal secretion patterns (predominant androgen secretion, predominant cortisol secretion, cortisol and androgen secretion, hormonally silent, other) was comparable to representative academic cohorts (P = .0643, χ²) [37, 38, 42]. There were no differences in major hormone secretion patterns when we compared antecedent ACC with that arising de novo (P = .429, χ²); we thus characterized hormone secretion patterns across the unified cohort (Fig. 4A). Any androgen secretion was associated with larger tumor size (Fig. 4B). Patients with combined cortisol and androgen secretion frequently had metastatic disease at diagnosis; patients with predominant cortisol secretion tended to present with locally advanced disease (Fig. 4C). Patients with predominant cortisol secretion generally had less smoking exposure than those with other hormonal subtypes (Fig. 4D). There was no association between hormonal secretion pattern and grade (P = .582, χ²) or oncocytic histology (P = .327, χ²).

Figure 4.

Hormone secretion patterns and clinical correlates. (A) Hormone secretion patterns are broadly categorized into major groups in the outer ring. (B) Patients with any androgen production had larger primary tumors. (C) Patients with combined cortisol and androgen secretion tended to present with metastatic disease; those with predominant cortisol secretion tended to present with locally advanced disease. (D) Patients with predominant cortisol secretion had significantly less smoking exposure than patients with androgen secretion or hormonally silent tumors. Kruskal-Wallis with significant comparisons by Dunn's post-hoc test (according to Benjamini-Hochberg adjusted P) indicated by brackets; all other pairwise comparisons had adjusted P ≥ .05. (E) Using Cox regression analysis, any cortisol secretion as a univariate was associated with a high risk of progression and death. In models including all major hormonal secretion patterns, predominant androgen secretion was protective against progression.

Any cortisol secretion was an adverse prognostic factor (Fig. 4E), and physical exam findings of Cushing syndrome were associated with a higher hazard for progression (hazard ratio, 1.81; 95% CI, 1.09-3.02; P = .0272, Cox regression). In models accounting for all major hormonal secretion patterns, predominant androgen secretion was protective against progression (Fig. 4E). These findings are comparable with other cohorts (Table S1 [40]) [37, 38, 47, 51-55].

Clinical genomics

A diagnosis of ACC is a clinical indication for germline genetic testing as ACC can be a feature of several tumor susceptibility syndromes [56, 57]. Thirty-four percent of patients had some form of germline genetic testing, and 45.4% had tumor genetic testing; both were more prevalent among those more recently diagnosed with ACC (Fig. S3A and S3B [40]). In almost all cases, genetic testing was performed using targeted panels (Table S2 [40]). Germline genetic testing was not differentially directed to younger individuals (P = .584, Wilcoxon). Patients harboring pathogenic germline variants presented at similar ages (Fig. S3C [40]), though with less disease burden (Fig. S3D and S3E [40]) compared to those without. Tumor genetic testing was directed to individuals with higher-grade (thus prognostically poor; Fig. S1C and S1D [40]) disease (Fig. S3F [40]).

The landscape of genomic and molecular alterations in this cohort is depicted in Fig. 5. A total of 54.1% of individuals had 1 or more tumor genetic alterations in a recurrent ACC driver gene [37, 58], and 10.6% of individuals with tumor genetic testing had no pathogenic variants. Confirming parity at the molecular level, there were no differences in tumor mutational burden, programmed death ligand 1 (PD-L1) expression, presence of germline pathogenic variants, or presence of known driver alterations when comparing antecedent ACC with that arising de novo (P > .05 for all comparisons, χ² or Wilcoxon test). Higher tumor mutational burden was associated with higher tumor PD-L1 expression (P = .0278, Wilcoxon test comparing PD-L1 < 1% to ≥1%). Tumor mutational burden was not associated with smoking exposure (P = .454, Kruskal-Wallis).

Figure 5.

Landscape of genomic alterations. Heatmaps depicting clinical genomics data. Adrenocortical carcinoma (ACC) samples are signified by columns, and genes targeted for alteration are signified by rows in the tumor (top, restricted to genes appearing in >1 patient and/or known to be recurrently altered in ACC [37, 58]) or germline (bottom). Tumor mutational burden (TMB, mutations/Mb) is depicted above the heatmaps. Samples are color-coded by antecedent nodule status, major hormone secretion pattern, sample site, programmed death-ligand 1 (PD-L1) percentage, and microsatellite instability status in the top rows according to the legend on the right. Tumor or germline alterations are also colored according to the legend on the right. In the top heatmap, the light gray box signifies that the gene was not reported to possess a pathogenic variant in the tumor. Some patients had more than 1 site submitted for sequencing; samples that come from the same patient are signified below the heatmap as follows: A1-A2 come from patient A, B1-B3 come from patient B, etc. Missing data are signified by white boxes.

Abbreviation: VUS, variant of uncertain significance.

Most common driver alterations activated the cell cycle (39.3% of individuals) and/or Wnt pathway (26.2% of individuals) as reported in pan-genomic studies [37, 58]. These alterations were associated with reduced progression-free survival (Fig. 6A), consistent with the literature [37, 38, 58]. The Cancer Genome Atlas-ACC cohort identified an inverse relationship between adrenal differentiation and immune infiltration [37], suggesting reduced immune checkpoint activation in functional tumors, a finding borne out in subsequent reports [59-61]. Similarly, we observed that patients with functional tumors had reduced tumor PD-L1 expression (Fig. 6B).

Figure 6.

Clinical correlates of molecular profiling and survival benefits conferred by therapies. (A) Kaplan-Meier curves demonstrating that tumor cell cycle and/or Wnt pathway alterations are associated with shorter progression-free survival; log-rank P shown in the lower left corner. (B) Adrenocortical carcinoma (ACC) with high programmed death-ligand 1 (PD-L1; ≥1%) were more likely to be hormonally silent. (C) Additional therapeutic interventions were directed to patients with higher-grade disease. Localized therapies include procedural interventions targeting specific disease sites (eg, targeted radiation of the adrenal bed or surgical resection of recurrences). (D and E) Multivariable Cox regression analysis models the impact of systemic therapies (D) or localized therapies (E) on death. Across systemic therapies, mitotane exposure (which typically occurred in the adjuvant setting) was protective against death. Cytotoxic chemotherapy was associated with a higher risk of death, likely due to inadvertent therapy selection bias among patients with metastatic disease, as detailed in the text. Exposure to localized therapies was protective against death.

Therapeutics

As clinical and genomic features of ACC arising from antecedent nodules were ultimately indistinguishable from ACC arising de novo, we analyzed survival outcomes by therapeutic modality across the unified cohort. A total of 65.7% received systemic therapies, which included mitotane (54.6% of all patients) and cytotoxic chemotherapy (45.4% of all patients). Localized therapies (including interventions like targeted radiation and surgical resection of recurrences) were utilized in 56.5%. A total of 65.3% of patients with locoregional disease at diagnosis received adjuvant therapy. Neoadjuvant cytotoxic chemotherapy was administered in 5/108 individuals; 2/5 individuals also received neoadjuvant mitotane. Patients with high grade disease were more likely to receive additional therapeutic interventions beyond surgery (Fig. 6C). In multivariable models accounting for stage and grade at diagnosis, mitotane and localized therapies were protective against death (Fig. 6D and 6E). Exposure to cytotoxic chemotherapy was associated with a higher hazard for death; however, stage IV as a parameter lost significance in this model, and patients with metastatic disease were more likely to receive cytotoxic chemotherapy (P = .022, χ²), suggesting this therapeutic intervention captured those with intrinsically more aggressive disease.

Discussion

ACC is a rare malignancy for which surgery remains the only intervention with the potential to cure. Prompt recognition with rapid referral to expert tertiary care centers is essential to securing even the possibility of disease eradication. Adding complexity, the gold standard initial workup for adrenal tumors is tissue-specific, prioritizing noninvasive modalities and specialized biochemical tests over pursuit of a pathologic diagnosis [21]. These features complicate clinical decision-making for frontline and community providers, who will make the majority of diagnoses in acute settings yet must assemble adequate minimum data to support escalation of care.

We were surprised to observe that nearly a third of patients possessed antecedent ipsilateral adrenal nodules, with latency periods extending to greater than 16 years (Fig. 2). Patients with antecedent nodules presented at older ages, without symptoms, and with smaller primary tumors that eventually evolved to more aggressive ACC (Fig. 2B, 2C, and 2G; Fig. S2F [40]). These lesions could not consistently be distinguished from ACT using standard surveillance approaches (Fig. 2D-2F; Fig. S2B, S2D, and S2E [40]). The prevalence of indolent antecedent nodules in our study is dramatically higher than previously reported rates approaching 4% to 5% [62, 63], possibly due to our longer observation period and higher proportion of older individuals (Fig. 1B). Taken together, these findings suggest that benign to malignant transformation may be more common than previously appreciated in adrenocortical carcinogenesis [21, 49, 50], accounting for shared recurrent somatic alterations in ACC and ACT [37, 42, 58] and reinforcing the clinical relevance of murine models that recapitulate this transition [64-67]. We did not observe an association between antecedent nodule-derived ACC and other characteristics, including hormone secretion patterns and molecular profiles, suggesting that evolution to malignancy is stochastic. The key to leveraging clinical benefit from this observation is to identify features that can distinguish truly benign lesions from those harboring occult malignant potential.

Through our lens as a tertiary referral center, we gained insight into practice patterns with meaningful implications for diagnosis, treatment selection, and surveillance. Though most patients ultimately had biochemical evidence of hormone excess (Fig. 4A), only a minority presented with concordant symptoms, resulting in fewer preoperative endocrinology referrals and incomplete hormonal workup (Fig. 3A-3C). Despite abundant literature cautioning against this practice [23-28], adrenal biopsies were routinely pursued without clinical indication, were inconsistently diagnostic, and were associated with adverse survival, even in multivariable models (Fig. 3D-3G). Though short of implying causation, these observations strongly suggest that this procedure poses greater risk than benefit for patients with ACC.

ACC is heterogeneous, with survival outcomes strongly linked to tumor molecular profiles [37, 38, 58]. Unlike other cancers, aggressive ACC subtypes are more differentiated, bearing somatic alterations that drive steroidogenesis and glucocorticoid production [37, 39]. Consistent with this, we observed that hormone secretion patterns—coarse clinical proxies for the steroidogenesis signature—were similarly associated with disease extent and survival (Fig. 4B, 4C, and 4E). The minimal smoking exposure among patients with predominant cortisol secretion (Fig. 4D) may allude to specific genomic instability mechanisms driving these tumors not captured by measurement of tumor mutational burden alone [37].

Most patients did not receive genetic testing, likely reflecting our prolonged observation window and early limited access to the technology (Fig. S3A and S3B [40]). Clinical targeted genomics reaffirmed known driver alterations (Fig. 5) with prognostic import (Fig. 6A), similar to prior reports [68]. Consistent with emerging evidence that steroidogenesis shapes the immune microenvironment [37, 59-61], we observed that hormonally silent tumors had higher PD-L1 expression (Fig. 6B). These and microsatellite unstable tumors constitute the minority (Figs. 1C, 4A, and 5), potentially justifying the generally poor response of ACC to immunotherapy (Fig. 6D). The increasing application of genetic testing is encouraging (Fig. S3A and S3B [40]), creating opportunities to delineate differences between germline-associated and apparently sporadic ACC, as well as to identify predictive biomarkers.

Treatment patterns reflected contemporary management, with broad application of systemic therapies and locoregional interventions, though cytotoxic chemotherapy was more frequently administered in stage IV contexts. Although patients with high-grade disease were more likely to receive these interventions (Fig. 6C), exposure to mitotane or localized therapies was protective against death (Fig. 6D-6E). Our results align with trial data suggesting that the patients who benefit from adjuvant mitotane may not include those with low-grade disease [6] and support studies suggesting a survival benefit from localized therapies [5, 14-19]. Exposure to cytotoxic chemotherapy was associated with higher mortality (Fig. 6D); this is likely attributable to confounding by indication in our center, though it highlights the need for objective data to enable therapy selection.

This study bears limitations inherent to its retrospective, single-center design. These include referral bias favoring aggressive disease, evolving clinical practice, and a small sample size, characteristic of rare disease. These limitations are offset by extended follow-up and depth of real-world data spanning 25 years of care. The identification of antecedent nodules in a subset of ACC and association with survival are subject to ascertainment bias, and the true prevalence of antecedent nodules is likely higher than what we observed. As suspected benign lesions are not routinely biopsied, it is not possible to map the clonal evolution of antecedent ACC and definitively rule out the possibility of a benign/malignant collision tumor. However, the absence of collision tumors on final pathology and growth kinetics of antecedent ACC imitating benign ACT (Fig. 2E and 2F; Fig. S2D and S2E [40]) argue against this phenomenon.

Our findings underscore the importance of early endocrinology involvement, caution against adrenal biopsy absent exceptional clinical indications, and suggest clinical benefit from mitotane and locoregional interventions in high-grade disease. The strikingly high prevalence of antecedent nodules identifies a subset of ACC with clinically silent disease evolution that evades current surveillance paradigms. Multicenter studies will be essential to validate this observation and nominate biomarkers that can recognize malignant potential at a time when the window for curative resection is still open.

Funding

D.R.M. is supported by the Oliver Langenberg Physician-Scientist Training Program at Washington University School of Medicine and NIH T32 DK007120.

Disclosures

D.R.M. is an inventor on granted and pending patent applications owned by the Regents of the University of Michigan describing compositions and methods for characterizing or treating cancer. All other authors have no conflicts to declare.

Data availability

Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will, on request, detail the restrictions and any conditions under which access to some data may be provided.