Adrenal Cortical Neoplasms A Study of 56 Cases
HARRY L. EVANS, MD,1 AND RENA VASSILOPOULOU-SELLIN, MD2
Fifty-six cases of adrenal cortical neoplasm with a minimum of 5 years follow-up are presented: 48 carcinomas and 8 adenomas. Adenomas typically had a maximal mitotic rate of fewer than 2 mitotic figures per 10 high-power fields (all cases), a prominent small nest growth pattern (7 cases), predominantly clear or foamy cytoplasm (6 cases), and no tumor necrosis (all cases), whereas carcinomas were characterized by at least 4 mitotic figures (often many more) per 10 high-power fields in the most active area (all cases), lack of a significant small nest growth pattern component (45 cases-solid or trabecular growth most com- mon), at least a considerable proportion of cells with eosinophilic cyto- plasm (all cases), and tumor necrosis (45 cases). Carcinomas were al- most always larger than adenomas, but two adenomas (5.9 cm and 7 cm) overlapped in size with the four smallest carcinomas (5.5 cm, 6 cm, 7 cm, and 7 cm, respectively). The patients with adenomas were older
on the average than those with carcinomas (median 58 years, range 31- 71 years versus median 41 years, range 5 months-66 years). Two ade- nomas and 19 carcinomas were functional. No patient with adenoma had recurrence of tumor after excision, whereas all but nine carcinoma patients died of tumor, after 1 to 183 months. Among carcinoma pa- tients, survival was significantly shorter when distant metastases were manifest at diagnosis (P = . 0003). There was a trend toward shorter survival with higher mitotic rates and functional tumors, but neither these nor any other parameter had a statistically significant relation- ship to survival or tumor behavior when presence/absence of metasta- ses at diagnosis was taken into account. (Key words: Adrenal neo- plasms; Adrenal cortex; Functional neoplasms; Endocrine neoplasms) Am J Clin Pathol 1996; 105:76-86.
The differential diagnosis of adrenal cortical adenoma and carcinoma has been considered problematic, and rather complex schemes have been elaborated to assist in the separation.1-3 Pathologic prognostic factors in adre- nal cortical carcinoma have been studied less exactingly, but various authors have found mitotic rate2,4 and nuclear pleomorphism5-7 to be significantly related to patient survival. To evaluate these issues, we undertook a review of cases of adrenal cortical adenoma and carci- noma at the University of Texas MD Anderson Cancer Center.
MATERIALS AND METHODS
Cases filed as adrenal cortical adenoma or carcinoma at the University of Texas MD Anderson Cancer Center were reviewed. Criteria for including a case in the study were that the adrenal neoplasm have been excised, that
From the Departments of ’ Pathology and 2 Endocrinology. The Uni- versity of Texas MD Anderson Cancer Center, Houston, Texas.
Manuscript received February 28, 1995; revision accepted July 28, 1995.
Address reprint requests to Dr. Evans: Department of Pathology, University of Texas MD Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030.
histologic material be available, that the diagnosis of adrenal cortical tumor be confirmed, and that there be at least 5 years of follow-up from the time of excision. Histologic material was examined without clinical infor- mation, and the following parameters were specifically evaluated: growth pattern, nuclear size, nuclear pleo- morphism, character of cytoplasm, mitotic rate, tumor necrosis, capsular features, and vascular invasion. Nuclear size was classed as small, medium-sized, or large, whereas nuclear pleomorphism was rated as none, slight, moderate, or marked. Mitotic rate was deter- mined by counting the number of mitotic figures in four sets of 10 consecutive high-power (×400) fields and ac- cepting the highest count observed (an American Optical One-Ten microscope was used; the area of a high-power field was 0.2 mm2). Survival curves were prepared by the Kaplan-Meier method, and comparisons were con- ducted with the Wilcoxon test. Other comparisons were made with the chi-square test. The final study group to- taled 56 cases.
RESULTS
Histopathologic Findings and Classification
Eight tumors were classified as adrenal cortical ade- noma and 48 as carcinoma. All adenomas had a maxi-
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
mal mitotic rate of less than two mitotic figures per 10 high-power fields, and none had tumor necrosis. Seven adenomas had a small-nest growth pattern either entirely or in large part (Fig. 1). The cytoplasm of all or most cells in the adenomas was clear or foamy rather than eosino- philic in six cases, and there was a substantial number of cells with clear or foamy cytoplasm in a seventh case. These seven cases had prominent small-nest growth pat- terns, and typically the cells with clear or foamy cyto- plasm were in the small nests (Fig. 1). Four of these seven adenomas had areas in which the cells had eosinophilic cytoplasm. The growth pattern in such areas was pre- dominantly solid in three cases (Fig. 2), but was nested and ribbon-like (with transversely oriented nuclei) in one case (Fig. 3-this, the seventh case mentioned above, was further distinguished by the presence of pigment in the tumor cells). The eighth adenoma had an entirely solid growth pattern and only eosinophilic cytoplasm (Fig. 4). The nuclei in the adenomas were for the most part small, but not infrequently ranged up to medium-sized, espe- cially where the cytoplasm was eosinophilic (Fig. 2). Nuclear pleomorphism was generally slight or absent, but was focally moderate, with scattered large nuclei, in
two cases (Fig. 5). Three adenomas had focal mature fat cells within them. The capsules around the adenomas varied from imperceptible (ie, the adenoma blended with the adjacent nonneoplastic adrenal without a clear divid- ing line) to thick, in some individual cases as well as in the group as a whole. Capsular invasion was not seen in the adenomas, nor was vascular invasion.
The carcinomas differed from the adenomas in several respects. All carcinomas had a mitotic rate in the most active area of at least 4 mitotic figures per 10 high-power fields, and most had considerably more. The maximal mitotic rate was less than 10 mitotic figures per 10 high- power fields in 10 cases, between 10 and 19 mitotic fig- ures (inclusive) per 10 high-power fields in 18 cases, and 20 or more mitotic figures per 10 high-power fields in 20 cases. All but three carcinomas demonstrated tumor necrosis, which was often extensive. The growth pattern in the carcinomas was generally solid or trabecular (Figs. 6 and 7). Only three carcinomas had a significant small- nest growth pattern component, although small nests were present focally in some others. In all of these cases,
small nests were combined with large nests (Fig. 8) and solid or trabecular growth and were often less well-de- fined than was usual in the adenomas (Figs. 9 and 10). A few other carcinomas had large nests without small nests in association with solid or trabecular areas. In three cases, there was a focal ribbon-like pattern (Fig. 10) sim- ilar to that in the previously described adenoma. Other unusual patterns seen in the carcinomas included a mi- crocystic arrangement (Fig. 11) and narrow cell cords in a myxoid background (Fig. 12). One carcinoma demon- strated areas with cells in single file, which was reminis- cent of lobular breast carcinoma (Fig. 13). The cells of the carcinomas for the most part had eosinophilic cyto- plasm. In a few cases some cells had clear or foamy cyto- plasm (Figs. 9 and 10), but these cells never predomi- nated. The nuclei in the carcinomas ranged from small to large, among different cases and sometimes within in- dividual cases (Fig. 14). When the nuclei were small, there was often (but not always) considerably less cyto- plasm than in the adenomas, resulting in a higher nuclear/cytoplasmic ratio (Figs. 14 and 15). Cells with
larger nuclei sometimes had very abundant cytoplasm (Fig. 16). Nuclear pleomorphism in the carcinomas ranged from absent or slight to very marked. Calcifica- tion was a frequent finding; when it was present, it was usually in areas of necrosis. The capsules around the car- cinomas varied from thin to thick. In some examples, there was clear invasion into or through the capsule, but in others there was not. Intratumoral fibrous bands were present in a minority of cases and were usually focal. Fif- teen carcinomas showed vascular (venous) invasion, which often involved large veins such as the adrenal vein, renal vein, and inferior vena cava as well as smaller veins. One carcinoma appeared to have arisen in an adenoma.
Age and Sex Distribution
The age distribution of patients with adenoma and carcinoma is shown in Figure 17. Patients with adenoma varied from 31 to 71 years of age, with a median of 58 years, whereas carcinoma patients ranged from 5 months to 66 years of age, with a median of 41 years. Among the adenoma patients, four were male and four were female, for a sex ratio of 1:1. In the carcinoma group, there were
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
19 males and 29 females, resulting in a sex ratio of ap- proximately 2:3. Thus, patients with carcinoma tended to be younger than those with adenoma, and were more likely to be female.
Tumor Laterality
Five adenomas involved the right adrenal and three the left. Among the carcinomas, there were 20 on the right side, 26 on the left, and 2 of unknown laterality.
Tumor Function and Presentation
One patient with adenoma had Cushing’s syndrome, one had aldosteronism, and the remainder had no evi- dence of tumor function. Eleven carcinoma patients had Cushing’s syndrome, seven had virilization (which was mixed with Cushing’s syndrome in five), one had aldo- steronism, and the others had no functional symptom- atology. Nonfunctional carcinomas generally presented with abdominal mass or pain, whereas nonfunctional ad- enomas were incidental findings on radiographic studies or, in one case, at surgery.
Treatment
Seven adenomas were treated by adrenalectomy and one by nephroadrenalectomy. In the latter case, there was also a transitional cell carcinoma of the renal pelvis. Carcinomas were treated by adrenalectomy in 29 cases and nephroadrenalectomy in 19. Occasionally, the spleen or the tail of the pancreas was also removed. Two patients with carcinoma had liver metastases excised at the initial operation, and one had a wedge excision of a lung metastasis. In a few cases, surgery was followed by radiation or chemotherapy with opDDD (mitotane). Re- current and metastatic carcinomas were treated with sur- gery, radiotherapy, and chemotherapy (usually opDDD) in varying combinations.
Metastases at Diagnosis
Nine patients with carcinoma had distant metastases clinically manifest at diagnosis: three in liver, two in lung, two in liver and lung, and two in bone. One of these patients and two others had involvement of regional
lymph nodes by metastatic tumor at the initial opera- tion.
Tumor Size
The adenomas ranged in size from 1.5 cm to 7 cm (maximum diameter), with a median of 4.5 cm. Carci- nomas varied from 5.5 cm to 25 cm, with a median of 15 cm. Five carcinomas were of unknown size, including four simply described as “large.” Among carcinomas of known size, all but 10 were 10 cm or larger. The two largest adenomas (5.9 cm and 7 cm) overlapped in size with the four smallest carcinomas (5.5 cm, 6 cm, 7 cm, and 7 cm, respectively).
Follow-up and Clinicopathologic Correlation
All patients with adenoma remained tumor-free dur- ing the follow-up period, which ranged from 69 to 177 months (median 87 months). The overall survival of the carcinoma patients is shown in Figure 18. Thirty-nine patients with carcinoma died of tumor, after intervals varying from 1 to 183 months, and one additional pa-
tient was living with uncontrolled locally recurrent and metastatic tumor at the latest follow-up of 86 months. Median survival of the carcinoma patients was 33 months. The most important prognostic factor in the carcinoma group was the presence or absence of distant metastases at diagnosis (Fig. 19); survival of patients with such metastases was shorter to a highly significant degree (P = . 0003). When presence or absence of distant metas- tases at diagnosis was taken into account, other clinical and pathologic variables did not have a statistically sig- nificant relationship with survival; however, there was a trend toward shorter survival with higher mitotic rates (Fig. 20) and functional tumors (Fig. 21) among patients without distant metastases at diagnosis. Of the two pa- tients who had involved regional lymph nodes but not distant metastases at diagnosis, one died of tumor and the other did not. Local recurrence developed in 22 pa- tients (including 3 patients who had grossly incomplete excision) and was controlled by subsequent therapy in only 1 of those patients. Distant metastasis occurred in 36 patients, including the 9 patients who had distant me- tastases at diagnosis. Two patients who had solitary lung metastases survived without evidence of tumor after ex-
cision of the metastases. The interval to local recurrence varied from 0 months (grossly incomplete excision) to 123 months with a median of 12 months for patients who had recurrence, whereas the interval to distant me- tastasis ranged from 0 months (distant metastases at di- agnosis) to 85 months with a median of 12 months for patients who had distant metastases. There was no statis- tically significant relationship between local recurrence and any clinical or pathologic parameter, nor was there any in regard to distant metastasis when patients with distant metastases at diagnosis were excluded from con- sideration.
The location of metastases is shown in Table 1 for all patients. Lung and liver were by far the most com- mon sites (26 and 25 patients, respectively) and were followed by bone (8 patients). Seven patients had re- gional lymph node metastases (including the three pa- tients who had them at initial surgery), and four pa- tients had metastatic involvement of distant lymph nodes. Metastasis to the opposite adrenal also oc- curred in four patients.
DISCUSSION
The differentiation of adrenal cortical carcinoma from adenoma has been studied in the greatest detail by Hough and colleagues,’ Weiss,3 and Slooten and associ- ates.2 Hough and colleagues developed a histologic index of malignancy based on the presence or absence of seven findings that were more frequent in metastasizing tu- mors: (1) a diffuse (solid) growth pattern; (2) vascular in- vasion; (3) tumor necrosis; (4) broad intratumoral fi- brous bands; (5) capsular invasion; (6) an average of more than one mitotic figure per 10 high-power fields; and (7) moderate to marked nuclear pleomorphism. In determining the histologic index, these findings were weighted in proportion to the relative strength of their association with metastasis. Hough and associates’ also proposed a nonhistologic index of malignancy deter- mined in a similar manner and based on the factors of tumor weight greater than 100 g, urinary ketosteroids >10 mg/g creatinine per 24 hours, absence of response
to ACTH, virilization (with or without Cushing’s syn- drome) or lack of tumor function, and patient weight loss greater than 10 pounds. The authors found that combin- ing the two indices resulted in a sharper separation of metastasizing and monmetastasizing tumors than the use of either index alone. The scheme of Slooten and co- workers2 was similar to this, although it was exclusively histologic. These authors weighted seven parameters ac- cording to their value in discriminating metastasizing from nonmetastasizing neoplasms and determined a his- tologic index for a given case by adding the discriminat- ing values of the parameters present. The parameters used by Slooten and coworkers were the following: (1) a mitotic rate greater than two mitotic figures per 10 high- power fields (discriminating value 9); (2) regressive changes including tumor necrosis (discriminating value 5.7); (3) abnormal nucleoli (discriminating value 4.1); (4) capsular or vascular invasion (discriminating value 3.3); (5) moderate to marked nuclear hyperchromatism (discriminating value 2.6); (6) moderate to marked nuclear atypia (discriminating value 2.1); and (7) abnor- mal growth pattern (discriminating value 1.6). The au- thors concluded that a tumor should be considered ma-
lignant when its histologic index was >8 and benign when it was <8. Weiss’s3 system was somewhat simpler because weighting was not involved; rather, he proposed that nine criteria be sought, and that a case be scored according to the number of criteria present. The criteria employed in this system were as follows: (1) a mitotic rate greater than five mitoses per 50 high-power fields in the most active areas of the tumor; (2) atypical mitoses; (3) venous invasion (smooth muscle in wall of vessel); (4) clear cells comprising 25% or less of the tumor; (5) tumor necrosis; (6) nuclear grade of III or IV by Fuhrman’s method for renal carcinoma;8 (7) diffuse (solid) architec- ture in more than 1/3 of the tumor; (8) invasion of sinu- soidal structures (no mural smooth muscle); and (9) cap- sular invasion. Weiss3 found that adenomas had two or fewer of these criteria and that carcinomas had four or more (later changed to three or more4).
Is this exhaustive an evaluation needed for determin- ing malignancy in adrenal cortical neoplasms? We think not in most cases. Heinbecker and colleagues9 in 1957 had noted in a more informal manner that frequent mi- toses, vascular invasion, and capsular invasion were firm indicators of malignancy and that tumor necrosis, calci-
fication (in our experience usually associated with necro- sis), and nuclear pleomorphism were suggestive. Tang and Gray10 basically confirmed these observations and also pointed out the architectural differences between ad- enomas and carcinomas (ie, the presence of nests of cells in most adenomas and of a solid or trabecular pattern in most carcinomas). They also emphasized the difference in tumor size between adenomas and carcinomas, which was so great that there was little overlap. We believe that most adrenal cortical neoplasms can be readily diag- nosed as benign or malignant if one keeps in mind that adenomas have very few if any mitotic figures, whereas carcinomas generally have at least a moderate number in some area, that adenomas typically lack tumor necrosis, whereas most carcinomas demonstrate it, that adenomas most often have a prominent small nest growth pattern component, whereas carcinomas usually are entirely or predominantly solid or trabecular, and that most adeno- mas have at least a substantial proportion of cells with clear or foamy cytoplasm (which are usually arranged in small nests), whereas cells with eosinophilic cytoplasm normally are more prevalent (or are solely present) in
carcinomas. Although we observed some overlap in tu- mor size in the 5.5-7 cm range and rare small carcino- mas have been reported by others,11 it is clear that ade- nomas are almost always smaller than carcinomas. Certainly it would seem reasonable to consider any adre- nal cortical tumor measuring 10 cm or more malignant on that basis alone. We do not deny that there are very uncommon “borderline” cases in which a systematic ap- proach might be helpful. Probably the most “difficult” case in our series was that of the adenoma with a solid growth pattern and solely eosinophilic cytoplasm, espe- cially because this was the adenoma that measured 7 cm. In the Weiss system,3 which we prefer over the other schemes because we consider it easiest to use, this case would have been scored at two points (for solid growth and lack of clear cytoplasm), the maximum allowed for an adenoma.
Cagle and colleagues12 suggested that pediatric adrenal cortical neoplasms may have histologic features nor- mally associated with malignancy, such as mitotic figures and tumor necrosis, but yet be benign. In their study, only tumor weight >500 g was strongly related to malig-
nant behavior. However, Ribeiro and associates13 had several fatal carcinomas that were substantially smaller than 500 g in their series of adrenal cortical carcinomas in children. There were only three patients in the pediat- ric age group in our study, two 10-year-old girls, both of
| Location | No. of Patients |
|---|---|
| Lung | 26 |
| Liver | 25 |
| Bone | 8 |
| Lymph nodes | |
| Regional | 7 |
| Distant | 4 |
| Opposite adrenal | 4 |
| Diaphragm | 1 |
| Omentum | 1 |
| Breast | 1 |
| Orbit | 1 |
| Small bowel | 1 |
| Brain | 1 |
| Spleen | 1 |
whom died of tumor, and a 5-month-old girl, who sur- vived. All three had large tumors with tumor necrosis and eight or more mitotic figures per 10 high-power fields. We do not believe these cases provide sufficient
18
15
Adenoma
Carcinoma
No. of Patients
12
9
6
3
0
0-9
10-19
20-29
30-39
40-49
50-59
60-69
70-79
Age
1
TOTAL
DEAD
.8
48
39
Proportion Surviving
.6
4
.2
0
0
24
48
72
96
120
144
168
192
216
240
Months
grounds for us to gi … an opinion as to whether the cri- teria for malignanc 1.Duld be different in children than in adults.
The pathologic p ,nostic factors that have received the most support in adrenal cortical carcinoma are mi- totic rate an : nuclear pleomorphism. Weiss and col- leagues4 fo .. hat patient survival was significantly shorter whe
· mitotic rate was higher, and Slooten
and associa: Iso determined that mitotic rate was prognostical. nportant. In our study, survival was reduced by a atistically significant extent when the maximal mitotic rate was 20 mitotic figures per 10 high-power fields or greater if the entire group of pa- tients was considered. However, if patients with mani- fest distant metastases at diagnosis were evaluated sep- arately, the difference was no longer statistically significant either without or with such metastases (there was no significant difference regardless of dis-
1
TOTAL
DEAD
Metastases at Diagnosis
9
9
.8
Proportion Surviving
No Metastases at Diagnosis
39
30
.6
.4
.2
0
0
24
48
72
96
120
144
168
192
216
240
Months
1
TOTAL
DEAD
≥20 Mitoses/10 HPF
14
12
.8
Proportion Surviving
<20 Mitoses/10 HPF
25
18
.6
.4
.2
0
0
24
48
72
96
120
144
168
192
216
240
Months
tant metastases at diagnosis between patients whose tumors had <10 mitotic figures per 10 high-power fields and those whose tumors had 10 to 19 mitotic figures per 10 high-power fields). Neither Weiss and as- sociates4 nor Slooten and coworkers2 took distant me- tastases at diagnosis into account when they deter- mined the prognostic significance of mitotic rate, despite the fact that this has been found repeatedly to be an important negative factor in patient survival14-18 (as it was in our study). Other authors have advocated histologic grading of adrenal cortical carcinoma based principally on the degree of nuclear pleomorphism and have argued that this was prognostically signifi- cant.5-7 We did not find nuclear pleomorphism to be a prognostic determinant when all patients were in-
1
TOTAL
DEAD
Functional Tumor
8
16
14
Proportion Surviving
Nonfunctional Tumor
23
16
.6
4
.2
0
0
24
48
72
96
120
144
168
192
216
240
Months
cluded in the evaluation or when distant metastases at diagnosis were considered.
REFERENCES
1. Hough AJ, Hollifield JW, Page DL, Hartmann WH. Prognostic factors in adrenal cortical tumors. Am J Clin Pathol 1979; 72: 390-399.
2. Slooten H, Schaberg A, Smeenk D, Moolenaar AJ. Morphologic characteristics of benign and malignant adrenocortical tumors. Cancer 1985; 55:766-773.
3. Weiss LM. Comparative histologic study of 43 metastasizing and nonmetastasizing adrenocortical tumors. Am J Surg Pathol 1984;8:163-169.
4. Weiss LM, Medeiros LJ, Vickery AL. Pathologic features of prog- nostic significance in adrenocortical carcinoma. Am J Surg Pa- thol 1989; 13:202-206.
5. Hogan TF, Gilchrist KW, Westring DW, Citrin DL. A clinical and pathologic study of adrenocortical carcinoma. Cancer 1980;45: 2880-2883.
6. Karakousis CP, Rao U, Moore R. Adrenal adenocarcinomas: His- tologic grading and survival. J Surg Oncol 1985; 29:105-111.
7. Nakano M. Adrenal cortical carcinoma. Acta Pathol Jpn 1988; 38: 163-180.
8. Fuhrman SA, Lasky LL, Limas C. Prognostic significance of mor- phologic parameters in renal cell carcinoma. Am J Surg Pathol 1982;6:655-663.
9. Heinbecker P, O’Neal LW, Ackerman LV. Functioning and non- functioning adrenal cortical tumors. Surg Gynecol Obstet 1957; 105:21-33.
10. Tang CK, Gray GF. Adrenocortical neoplasms: Prognosis and morphology. Urology 1975; 5:691-695.
11. Gandour MJ, Grizzle WE. A small adrenocortical carcinoma with aggressive behavior. Arch Pathol Lab Med 1986; 110:1076- 1079.
12. Cagle PT, Hough AJ, Pysher TJ, et al. Comparison of adrenal cor- tical tumors in children and adults. Cancer 1986; 57:2235-2237.
13. Ribeiro RC, Neto RS, Schell MJ, et al. Adrenocortical carcinoma in children: A study of 40 cases. J Clin Oncol 1990; 8:67-74.
14. Bodie B, Novick AC, Pontes JE, et al. The Cleveland Clinic expe- rience with adrenal cortical carcinoma. J Uro/ 1989; 141:257- 260.
15. Didolkar MS, Bescher RA, Elias EG, Moore RH. Natural history of adrenal cortical carcinoma. Cancer 1981;47:2153-2161.
16. Henley DJ, van Heerden JA, Grant CS, Carney JA, Carpenter PC. Adrenal cortical carcinoma: A continuing challenge. Surgery 1983;94:926-931.
17. Luton J-P, Cerdas S, Billaud L, et al. Clinical features of adreno- cortical carcinoma, prognostic factors, and the effect of mito- tane therapy. N Engl J Med 1990; 322:1195-1201.
18. Sullivan M, Boileau M, Hodges CV. Adrenal cortical carcinoma. J Uro/ 1978; 120:660-665.
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016