This study was approved by the Human Research Ethics Committee at Chiba University (approval number: 828), and all patients provided written informed consent. We retrospectively analyzed the steroid profiles of 7 patients with cortisol-producing ACC and 25 CPA patients who were admitted to our hospital between 2013 and 2018. The final diagnosis had been ascertained by histology and evidence of metastasis in ACC, and by imaging, biochemical, and clinical follow-up showing no evidence of adrenal tumor growth and metastasis in CPA. Endocrinologically, overt Cushing’s syndrome was determined in patients with signs or symptoms of excess hormones, increased 24-hour urinary free cortisol, and high plasma cortisol that could not be suppressed below 5 μg/dl with overnight dexamethasone at doses of 1 mg. Subclinical Cushing’s syndrome was determined in patients with no signs or symptoms of excess hormones, subnormal suppression following overnight dexamethasone (1 mg; >1.8 μg/dl), and normal 24-hour urinary free cortisol [23, 24].
We analyzed five important clinicopathological parameters in cortisol-producing ACC: ENSAT classification, Weiss score, Ki67-index derived by pathological immunohistological staining, tumor size, and overall survival. We further examined the correlation of these five clinical parameters to steroid metabolites.
Then, we identified the steroid metabolites that correlated with pathological findings or clinical parameters related to staging and disease prognosis in ACC. The ENSAT staging system consists of stages I (T1N0M0), II (T2N0M0), III (T1–2N1M0 or T3–4N0–1M0), and IV (TanyNanyM1, metastatic ACC) . The Weiss score is comprised of nine histological criteria: (i) high nuclear grade, (ii) mitotic rate greater than five per 50 high power fields (HPF), (iii) atypical mitotic figures, (iv) eosinophilic tumor cell cytoplasm (greater than 75% of tumor cells), (v) diffuse architecture (greater than 33% of the tumor), (vi) necrosis, (vii) venous invasion, (viii) sinusoidal invasion, and (ix) capsular invasion . A tumor is labeled malignant when it meets three or more of these histological criteria . The Ki67-index is evaluated using an immunohistochemical assessment of cell proliferation by the detection of Ki67 antigen in neoplastic cell populations; a Ki67-index of 10% or more is diagnosed as high risk .
Clinical samples and hormonal assays
We compared 12 serum steroid metabolites using an immunoassay, which is a more rapid and less expensive method compared to LC-MS/MS, in cortisol-producing ACC and CPA. In addition, 24-hour urine excretion of individual 17-ketosteroids as androgen secretions were examined. These serum and urine metabolites, with the exception of the largest produced end products (cortisol, aldosterone, and DHEAS), were not regularly measured but examined in all subjects in our study before starting treatment with mitotane, ketoconazole, metopirone, or other drugs that influence the hormonal evaluation. Under basal conditions, fasting blood was withdrawn after a 15 minute rest between 8:00 and 9:00 AM. The day before, 24-hour urine specimens were collected for periods of two to three days, and urinary free cortisol and 17-ketosteroid fractions were measured. Urinary and serum cortisol were measured by radioimmunoassay (RIA). 11-deoxycortisol, 11-deoxycorticosterone, corticosterone, aldosterone, pregnenolone, 17-hydroxypregnenolone, and androstenedione were measured by RIA. Progesterone, 17-hydroxyprogesterone, and testosterone were measured by electrochemical luminescence immunoassay (ECLIA). Urinary 17-ketosteroid fractions were measured by GC-MS. Sample analysis was completed by a Japanese clinical analytical laboratory (SRL, Inc., Tokyo, Japan).
Shapiro-Wilk test showed that the endocrinological data were not normally distributed. Hence, pairwise comparisons were performed using the Mann-Whitney U-test. The results were expressed as a median (interquartile ranges), and a value of P < 0.05 was considered statistically significant. The Chi-Square statistic is used for testing relationships between categorical variables. Receiver operating characteristics (ROC) curves were generated for steroid metabolites that displayed relatively significant differences between ACC and CPA. The area under the ROC curve (AUC) was also calculated. A perfect classifier has AUC = 1, and a completely random classifier has AUC = 0.5. Sensitivity and specificity were calculated at cut-off values providing the highest sensitivity. The correlation between the clinicopathological parameters and individual steroid metabolites in ACC was calculated by Pearson's correlation coefficient (R). R values between 0.7 and 1.0 together with P < 0.05 can be considered highly correlated. Statistical analysis was performed using SPSS Statistics for Windows (SPSS Inc., Chicago, IL, USA).