Adrenocortical Tumors in Brazilian Children: Immunohistochemical Markers and Prognostic Factors

Tumors affecting the adrenal cortex in children are rare, corresponding to an incidence of 0.3 case per 1000 000 persons per year and representing 0.2% of all childhood tumors.1 There are 2 peaks of incidence, in those younger than 5 years and in those aged between 40 and 50 years.1-3

Pathologically, adrenocortical tumors (ACTs) are generally divided into adenomas (benign histologie results) and carcinomas (malignant histologie results). However, their behavior in childhood is usually unpredictable, and distinction based only on histologie classification is difficult.4-6 Because of the lack of an established relationship between histologic results and prognosis, some authors recommend that ACTs always be classified as carcinomas.7

Immunomarkers have been extensively studied in these tumors. Although their role in predicting outcome has not yet been elucidated, the p53 protein has been, so far, the most investigated.8-12 The Ki-67 nuclear antigen (as a proliferative index), the c-Erb-B2 antigen of the cell membrane (as a marker of growth factor receptor), and the Bcl-2 antigen of cell cytoplasm (as an apoptosis marker) have been associated with several types of tumors, but the role of cell cycle regulatory proteins, such as p53 and Ki-67, in ACTs remains controversial, because the rarity of the disease limits adequate study populations to correlate the expression of these proteins with prognosis.13-24

The objective of this study was to analyze if there were any associations of clinical features, histologie classification, and presence of immunomarkers with prognosis of these rare tumors in a consecutive series of children treated by the same surgical team in 2 Brazilian medical services.

PATIENTS AND METHODS

A retrospective review of the medical charts of 33 consecutive children with ACTs was done. The patients had been treated and followed up on in the Pediatrie Endocrinology Service and the Pediatrie Surgery Service of the State University of Campinas and Boldrini Children's Cancer Center since 1980. The principles outlined in the Declaration of Helsinki were followed, and the Ethics Committee of the State University of Campinas School of Medical Sciences approved the publication of these data. Clinical data were reviewed, such as age at diagnosis, sex, time between first symptoms and diagnosis, clinical signs and symptoms, tumor position, type of surgery, staging, and follow-up.

Tumor staging was done according to the classification by Sullivan et al.25 The volume and weight of each tumor were reviewed, and the histologie sections were classified using histologie and pathologic criteria according to the methods of Hough et al,26 Weiss,27 Van Slooten et al,28 Cagle et al,29 and Ribeiro et al,30 to assess their prognostic value.

Paraffin-embedded blocks were sent for an additional immunohistochemical study using the streptavidin-biotin-peroxidase technique31 to analyze the following markers: p53, Ki-67, c-ErbB2, and Bcl-2. The primary antibodies used were clone DO7, code M701,1:100 (DAKO, Carpinteria, Calif) for p53; clone MIB-I, code 0505, 1:200 (Immunotech, Westbrook, Me) for Ki-67; code AO485, 1:300 (DAKO) for c-Erb-B2; and clone 124, code M0887, 1:40 (DAKO) for Bcl-2. All antibodies had a positive and negative control when stained on a slide. When the results were negative, the immunohistochemical reaction was repeated.

Cochran Q test was used to compare the results of the histologie classifications and the immunomarker analysis to assess the differences in the proportions of the responses. The Wilcoxon signed rank test for related samples was used to compare the number of atypical cells found according to the classifications by Weiss,27 Van Slooten et al,28 and Hough et al.26

Patients with recurrence, those whose tumors were incompletely resected, and those who had died were studied using eventfree survival (EPS) analysis, and overall survival was evaluated by the Kaplan-Meier method. The curves were analyzed, when possible, by the log-rank test, with P < .05 being considered statistically significant.

RESULTS

The age at diagnosis ranged from 2 to 96 months (median age, 27 months). Twenty-one children (64%) were female, and 12 (36%) were male. The time between first symptoms and diagnosis ranged from 1 to 24 months (median, 6 months).

The tumor was found on the right side in 17 children (52%) and on the left side in 16 children (48%). The main clinical findings in this series are shown in Table 1. Virilization alone was present in 30 children (91%), virilization plus Gushing syndrome in 7 children (21%), and Gushing syndrome alone in 1 child (3%). There were 2 asymptomatic patients (6%). Fifteen children were hypertensive, with the disease in 3 children uncontrolled by diet and the usual antihypertensive drugs. As previously reported, those children were subsequently treated effectively with ketoconazole.32 Three patients had metastasis at diagnosis (1 in the liver and 2 in the lungs). According to the criteria of Sullivan et al,25 18 patients (55%) were classified as having stage 1 disease, 9 patients (27%) as having stage 2 disease, 3 patients (9%) as having stage 3 disease, and 3 patients (9%) as having stage 4 disease.

Two children had biopsies only performed because of unresectable tumors. Complete surgical resection was attempted in 31 patients, but rupture with tumor spillage occurred in 4 of them. Intraoperative complications consisted of bleeding in 2 patients, with the bleeding secondary to a vena cava injury in one of them.

Tumor weight varied from 7 to 3700 g (median, 70 g), and tumor volume varied from 6 to 2300 mL (median, 72 mL). Seven patients had tumor volumes exceeding 200 mL. There was an association between tumor volume exceeding 200 mL and poor survival (P = .04) (Figure). According to the criteria of Hough et al,26 21 tumors were classified as adrenocortical carcinomas and 12 were of undetermined type (Table 2). Based on the histologicclassification proposed by Weiss,27 there were 31 adrenocortical carcinomas and 2 adenomas; however, following the criteria of Van Slooten et al,28 all tumors were classified as adrenocortical carcinomas. When analysis was based on the criteria proposed by Cagle et al29 and by Ribeiro et al,30 there were 4 and 15 tumors, respectively, that were considered to have malignant behavior. There were no statistically significant differences among the proportions of malignant tumors disclosed by these classifications.

Follow-up ranged from 5 to 158 months (median, 27 months). Four patients died (2 of progression of the disease, 1 on the third postoperative day, and 1 without an established cause). Five patients experienced recurrence (2 locally, 2 in the lungs, and 1 in the brain).

Factors associated with good outcome were female sex (P = .02), stage 1 and 2 disease (P = .01), complete resection of the tumor (P = .04), tumor weight of 100 g or less (P = .002), and localized disease as defined by Ribeiro et al30 (P = .004). Tumor volume of 200 mL or less was associated with better overall survival (P = .04) but not with EFS.

Factors showing no correlation with prognosis included age (P = .06), laterality of the tumor (P = .82), presence of hypertension (P = .73), and disease classification according to the methods of Hough et al26 (P = .24), Weiss27 (P = .49), Van Slooten et al,28 or Cagle et al29 (P = .49).

Immunohistochemical analysis revealed that all tumors were negative for c-Erb-B2 and p53, 1 tumor was positive for Ki-67, and 9 tumors (27%) were positive for Bcl-2 (Table 3). Results of the immunohistochemical tests were inconclusive regarding their prognostic significance because of the small number of tumors expressing the markers.

COMMENT

Although rare in most parts of the world, ACTs are frequent in Brazil, especially in the southern part of the country.30,33 The present study confirms previous epidemiologic findings regarding age at diagnosis (all patients were 8 years or younger), sex (ACTs are more common among girls), laterality of the tumor (both sides are equally affected), and virilization as the main clinical syndrome among symptomatic patients.

As in most rare tumors, issues regarding prognostic factors, treatment, and outcome are still open to investigation. In addition, the tumor biology and definitive criteria for malignancy are lacking for ACTs, especially those seen in childhood.

In the present series, age at diagnosis was not an indicator of outcome; the survival rate was the same in children younger or older than 24 months. However, in other series, age has been reported to be a prognostic factor. Humphrey et al34 found a 54% survival in children with ACTs and age younger than 7 years (75% of them were younger than 2 years) and only a 17% survival in children aged between 9 and 16 years. Sabbaga et al,33 in a series of 55 children, reported an age younger than 24 months as an indicator of good prognosis (83% survival for children younger than 24 months versus 45% survival for children older than 24 months).

The median time from initial symptoms to diagnosis in this series was 6 months, and this did not appear to affect survival. However, Sabbaga et al33 reported a 70% survival in children with ACTs and onset of symptoms less than 6 months before the diagnosis; when symptoms were present for more than 6 months, survival dropped to 9%.

Staging of the tumor is an important prognostic factor and, as in other studies,35,36 we found statistically significant improved survival among patients with stage 1 and 2 disease according to the classification by Sullivan et al.25

Complete tumor resection is considered curative for most children with ACTs, while rupture of the tumor seems to be associated with early recurrence and poor outcome.30,33

Weight and volume of the tumor are believed to be factors that affect prognosis. Tumor weights exceeding 100 g,26,30 150 g,28 and 500 g29 were reported to be associated with poor outcome and have been included as criteria for malignancy. Tumor volume exceeding 200 mL also showed correlation with poor outcome.30 In the present study, children with tumor weight of 100 g or less had a significantly better EFS prognosis. Tumor volume of 200 mL or less was associated with increased overall survival, although no change in the EFS was found.

Ribeiro et al30 found that, at ages younger than 3Vi years, the interval between the first symptoms and diagnosis was more than 6 months and that levels of urinary corticosteroids were significantly associated with prognosis. However, when tumor volume was included in the analysis, it was the only factor that retained prognostic significance. Results of histologie analysis of the tumors in that series of patients showed no relationship with prognosis. Similarly, in the present study, immunohistochemical analysis of p53, Ki-67, c-Erb-B2, and Bcl-2 was not of great value in predicting outcome.

Protein p53 expression has been associated with ACTs, ranging in incidence from 23% to 31% of ACTs in several series.8-10,24,37 Although p53 is considered more frequent in carcinomas than adenomas, no consistent correlation with prognosis has been found in children.9,10,12,24,37 On the other hand, a correlation with poor outcome has been suggested in adults with ACTs.14 The observation of 100% negativity of p53 expression from the immunohistochemical analysis in our series may be due to the brief half-life of this protein, with consequent difficulties in detecting it.

Ribeiro et al,38 in 2001, described a germline point mutation in exon 10 of the p53 gene, leading to an Arg337His amino acid substitution in 97% of children with ACTs from southern Brazil. Also in 2001, Latronico et al39 found the identical mutation in 77.7% of children and 13.5% of adults with ACTs from southeastern Brazil. According to Latronico et al, the presence of this mutation was related to an unfavorable prognosis in most of the adults but not in the children.

A strong correlation between expression of Ki-67 and malignancy of ACTs in adults has been proposed by Vargas et al14 and by Goldblum et al.15 In the present series, there was only one child with a positive Ki-67 result, thus preventing any conclusion about the role of this nuclear antigen in predicting outcome.

To our knowledge, there are no reports in the literature about the expression of c-Erb-B2 in ACTs in adults or in children. Our study failed to disclose any association of this protein with ACTs. Although the Bcl-2 protein was found to be positive in 27% (n = 9) of the tumors in the present series, it did not appear to have any effect on overall survival or EPS.

From the data herein presented, we conclude that a good prognosis in children with ACTs includes the following factors: female sex, early stage of disease, complete surgical resection, and tumor weight of 100 g or less or tumor volume of 200 mL or less. Use of existing histologie classification systems or results of specific immunohistochemical analyses could not predict prognosis of this dis ease in childhood.

We thank Silvia Brandalise, MD, PhD, and Marcelo Alvarenga, MD, for their clinical and histopathologic assistance, and Sherian Kae Bowyer for her English-language assistance.

© 2005 College of American Pathologists Provided by ProQuest LLC. All Rights Reserved.