Flow Cytometric Analysis of Nuclear DNA From Adrenocortical Neoplasms
A Retrospective Study Using Paraffin-Embedded Tissue
JAMES B. AMBERSON, MD,* E. DARRACOTT VAUGHAN JR, MD,t GEORGE F. GRAY, MD,# AND GREGORY J. NAUS, MD*
Nuclear DNA content of paraffin-embedded tissue from 48 adrenocortical neoplasms, 18 histologically normal control adrenal glands, and five hyperplastic adrenal glands was analyzed retrospectively using flow cytometry. Aneuploidy was compared with morphologic criteria as a predictor of recurrence. All 18 controls, five hyperplastic glands, and 39 neoplasms were diploid. Nine neoplasms were aneuploid. Com- pared with their diploid counterparts, aneuploid neoplasms were more likely to weigh more than 50 g (P <0.0001) and to have three or more histologic features of carcinoma (P < 0.0001). Thirty-six neoplasms were followed clinically for at least 2 years (range 24 to 120 months, mean = 64.6 months) or until local recurrence, metastasis, or death. Five were clinically malignant. Neoplasms which recurred or metastasized were more apt to be aneuploid (P < 0.005) than those showing no evidence of further disease during the follow-up period. They were also more likely to weigh more than 50 g (P < 0.005) and to have three or more histologic features of carcinoma (P < 0.0025). However, neither aneuploidy, large size, nor unfavorable histology result was a consistent feature in every malignant neoplasm. Flow cytometric DNA content analysis appears to be as effective a predictor of clinical outcome as size and histology and may be of particular value when the morphologic features are ambiguous. Cancer 59:2091-2095, 1987.
T HE DISTINCTION BETWEEN benign and malignant ad- renocortical neoplasms can be difficult. A combi- nation of clinical, gross, and microscopic features are usually considered before making a diagnosis. Although most neoplasms can be classified as either adenoma or carcinoma, some must be regarded as being of borderline or indeterminant malignancy. Flow cytometric DNA content analysis has been advocated as a useful adjunct to the histologic classification of adrenal neoplasms,’ al- though the requirement of fresh tissue has precluded analysis of a large series of these rare neoplasms. The re- cent development of a method for analyzing nuclear DNA content by flow cytometry using paraffin-embedded tissue now permits the retrospective analysis of these and other uncommon neoplasms.2 The current study investigated
the prognostic value of nuclear DNA content analysis by flow cytometry using paraffin-embedded archival material from a large series of adrenocortical neoplasms. DNA ploidy was compared with size and histology as a predictor of recurrence.
Materials and Methods
Subjects
Fifty-three consecutively encountered cases indexed as adrenal cortical adenoma and carcinoma were retrieved from the surgical pathology files of The New York Hos- pital from the period 1974 to 1985. The age and sex of the patient, associated hormonal syndrome(s), subsequent clinical course, and weight of the neoplasm were obtained from clinical records and pathology reports. Malignant neoplasms were defined as those producing local recur- rence or metastases. Criteria for inclusion in the study were: (1) absence of coexistent visceral carcinoma, and (2) verification of adrenocortical neoplasm upon review of slides and clinical information. Two neoplasms were excluded because of coexistent carcinoma and one because of uncertain primary site. All 50 acceptable adrenocortical neoplasms were submitted for flow cytometric DNA con- tent analysis. Eighteen histologically normal adrenal glands, obtained between 1974 and 1985, from nephrec-
From the *Department of Pathology and fDivision of Urology, De- partment of Surgery, Cornell University Medical College, The New York Hospital Medical Center, New York, New York, and Department of Pathology, Vanderbilt University School of Medicine, Nashville, Ten- nessee.
The authors thank Ms. Freda Karp for technical support, and Mr. Douglas Weston for secretarial support.
Address for reprints: James B. Amberson, MD, Cornell University Medical College, Department of Pathology, 1300 York Avenue, New York, NY 10021.
Accepted for publication January 15, 1987.
tomy specimens removed for Stage I renal carcinoma served as controls. Five hyperplastic glands were also studied as examples of benign proliferative lesions. Four were from patients with primary aldosteronism (Conn’s syndrome), and one was from a patient with excessive glucocorticoid production (Cushing’s syndrome).
Sample Preparation
All flow cytometric DNA analyses was performed on paraffin-embedded tissue according to a slightly modified version of the method of Hedley,2 as described below. Tissue from one paraffin block was available for analysis in 38 neoplasms and from two or more blocks in 12. Three 50 um sections were cut from each paraffin-embed- ded block, placed in 10 ml glass centrifuge tubes, and dewaxed at room temperature with two changes of xylene (5 ml. 10 minutes each). The sections were then rehydrated in a sequence of decreasing concentrations of ethanol: two changes of 5 ml of 100% ethanol (10 minutes each), followed by one change each of 5 ml of 95% ethanol (10 minutes), 70% ethanol (10 minutes), and 50% ethanol (10 minutes). The samples were washed twice in distilled water and suspended in 1 ml of 0.5% pepsin (P-7012, Sigma Chemical Company, St. Louis, MO) in 0.9% NaCl solu- tion adjusted to pH 1.5 with 2N HC1. The tubes were incubated in a 37℃ water bath for 30 minutes. After washing twice in 2 ml of Dulbecco’s phosphate buffered saline, the samples were filtered through 74um nylon mesh and resuspended in 1 ml of 50 µg/ml propidium iodide in 10 mmol tris buffer (pH 7.0) containing 5 mmol MgCl2 and 1 mg/ml sodium azide. Ribonuclease 1A (R-4875 Sigma Chemical Company, St. Louis, MO) was added to each sample to give a final concentration of 120 ug/ml. The samples were placed in a water bath at 37℃ for 60 minutes and then refiltered through 74 um nylon mesh before use.
Flow Cytometric Analysis
Nuclear DNA content was measured in 50,000 cells per sample (flow rate = 100 cells/second at room tem- perature) using a Coulter EPICS V flow cytometer (Coul- ter Electronics Inc., Hialeah, FL) equipped with a 5 W argon laser operating at 300 mW at 488 nm in light reg- ulation mode. The red fluorescent emission was measured using a 515 nm long pass interference and a 515 nm long pass absorbance filter. No other parameters were mea- sured. Fresh unfixed human peripheral blood lymphocytes from normal volunteers were stained by the same method described above. Their modal GO/G1 peak signals were set at channel 60 on a scale of 0 to 255 by adjusting the high voltage to the photomultiplier tube. DNA histograms were displayed as a plot of fluorescence (relative DNA content) versus the number of cells measured.
Since the duration of formalin fixation and age of the
paraffin blocks may potentially alter nuclear fluorescence and result in relatively large intersample variation com- pared with fresh tissue, internal standards such as human lymphocytes or trout erythrocytes cannot be used to iden- tify a diploid GO/G1 peak and DNA index.3,4 However, other studies have reported that the presence of stromal and inflammatory cells within a neoplasm can serve as a built-in diploid control. The presence of such cells could be identified in all our samples by the use of parallel mi- croscopic sections. Furthermore, every effort was made to measure those blocks with nonneoplastic as well as neoplastic adrenal tissue in order to assure the presence of diploid cells. This effort was successful in all neoplasms weighing less than 50 g, but was not possible in larger neoplasms which had obliterated the nonneoplastic ad- renal glands.
Neoplasms with only a single GO/G1 peak were con- sidered diploid (Fig. 1A). Neoplasms were classified as aneuploid if there was an additional distinct GO/G1 peak (Fig. 1B). No attempt was made to separate tetraploid from diploid neoplasms. The DNA index for aneuploid populations was calculated as the quotient of the modal channel of the aneuploid peak divided by the modal channel of the diploid peak. The two main limitations of this method of analysis result from the assumption that the first GO/G1 peak belongs to a diploid cell population: (1) hypodiploid aneuploid cell populations, which are un- usual in most solid tumors, cannot be distinguished from hyperdiploid aneuploid populations, and (2) a sample consisting of a pure population of aneuploid cells, an un- likely occurence for reasons described above, would be indistinguishable from a diploid population.
Morphologic Analysis
Neoplasms were classified according to criteria based on size and histology. They were classified by size as fol- lows: large tumors: weight greater than 100 g; borderline tumors: weight 51 g to 100 g; and small tumors: weight 50 g or less.
Neoplasms were classified histologically according to the system of Weiss3 which included the following features: nuclear Grade III/III (our Grade III being equivalent to Weiss’ Grades III and IV), greater than 5 mitoses per 50 high power fields, atypical mitoses, clear cells comprising 25% or less of the neoplasm, diffuse architecture, micro- scopic necrosis, and invasion of venous, sinusoidal, or capsular structures. Histologic carcinoma: four or more histologic features; histologic borderline tumor: three his- tologic features; and histologic adenoma: two or less his- tologic features.
Statistical Analysis
All statistical comparisons were made using Fisher’s exact test.
Relative number of cells
A
Relative number of cells
B
1.0
1.0
0.5
0.5
50
100
150
200
50
100
150
200
Channel number (relative DNA content)
Channel number (relative DNA content)
Results
The general features of the 50 adrenocortical neoplasms are shown in Table 1. Only three occurred in patients aged less than 20 years (aged 2.5, 7, and 19 years). Clinical follow-up of at least 2 years or until documentation of either local or distant recurrence was available for 36 cases (range for nonrecurrent neoplasms = 24 to 120 months, mean = 64.6 months). Five neoplasms were clinically malignant. All occurred in adults, the youngest patient being aged 43 years. Two malignant neoplasms had no associated endocrine syndrome: one was found to have invaded the kidney, aorta, and renal vessels at the time of diagnosis, and the second was diagnosed in a patient in whom lung metastases were discovered 1 month fol- lowing removal of the primary. The third was associated with Cushing’s disease and was followed 5 years after re- section by pulmonary metastases. The fourth presented with virilization and recurred 7 years later. The fifth was associated with Conn’s syndrome and recurred after 1 year.
All 18 control, all five hyperplastic, and 39 neoplastic adrenal glands were diploid. Their coefficients of variation for G0/G1 peaks were 7.8% (range 4.6% to 12.7%), 5.8% (range 4.0% to 8.7%), and 5.7% (range 3.4% to 10.2%), respectively. Among the 39 diploid neoplasms were six with more than one block available for analysis. All blocks from these six neoplasms, three of which weighed more than 50 g, were diploid. However, as reported by others, 2,3 there was considerable interspecimen variability in fluo- rescence intensity of diploid nuclei. Specifically, modal GO/G1 channels varied from a high of 57 to a low of 31 when the modal GO/G1 channel for fresh unfixed human lymphocytes was set at channel 60. In general, the older the paraffin block, the weaker the intensity of nuclear flu- orescence. However, there was no evidence to suggest loss of stoichiometric binding of propidium iodide in older samples: the G2-M to GO/G1 modal channel ratios did
not vary with sample age. They remained 2.00 (range 1.83 to 2.07), 1.98 (range 1.87 to 2.02), and 2.00 (range 1.95 to 2.05) for control, hyperplastic, and the 39 diploid neo- plastic glands, respectively.
Nine neoplasms were aneuploid. All had three or more peaks. Six had single or mean DNA indices clustering around 1.3 (1.24, 1.27, 1.28, 1.29, 1.32, and 1.37) and were classified as hypotriploid aneuploid. Three had single or mean DNA indices around 1.7 (1.63, 1.68, and 1.69) and were labelled hypertriploid aneuploid (Fig. 2). In six cases more than one block was available for study. From block to block, the DNA indices never varied more than 0.12 index units within a given aneuploid neoplasm.
The histograms from two small aldosterone producing tumors could not be interpreted because of excessive fluorescent debris and coefficients of variation greater than 15%.
Table 2 compares DNA ploidy and morphologic cri- teria. Eight of nine aneuploid neoplasms versus three of 39 diploid neoplasms weighed more than 50 g (P
| Age | |
|---|---|
| Mean | 50 yrs |
| Range | 2.5-88 yrs |
| Sex | |
| Male | 24 |
| Female | 26 |
| Clinical syndrome | |
| Conn's | 27 |
| Cushings | 11 |
| Virilizing | 3 |
| Feminizing | 2 |
| Mixed | 0 |
| None | 7 |
| Neoplasm size (weight) | |
| Median | 22 g |
| Range | 4-1000 g |
4
Number of cases
HYPOTRIPLOID
HYPERTRIPLOID
N=6
N=3
3
2
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
DNA index
< 0.0001). Seven of nine aneuploid neoplasms versus two of 39 diploid neoplasms had three or more histopathologic features of carcinoma (P < 0.0001). Weight and histo- pathologic features were also strongly correlated: 8/11 neoplasms weighing more than 50 g versus 1/37 weighing 50 g or less had three or more histopathologic features of carcinoma (P < 0.0001).
Clinical outcome compared with DNA ploidy and morphologic criteria is summarized in Table 3. Aneuplo- idy was noted in four of five malignant versus three of 31 clinically benign neoplasms (P < 0.005). Four of five ma- lignant neoplasms versus three of 31 benign neoplasms weighed more than 50 g (P < 0.005). Four of five malig- nant neoplasms versus two of 31 benign neoplasms had three or more histopathologic features of carcinoma (P < 0.0025). Thus, using aneuploidy as a predictor of clinical
malignancy, DNA ploidy determination had a false pos- itive rate of 42.9% and a false negative rate of 3.4% com- pared with false positive and false negative rates of 42.9% and 3.4% for weight, and 33.3% and 3.3% for histology. Three of the malignant neoplasms were hypertriploid aneuploid. All three weighed more than 100 g and had four or more histopathologic features of carcinoma. One was associated with Cushing’s syndrome and two with no hormonal syndrome. The two other malignant tumors were not obvious carcinomas by morphology. One was hypotriploid aneuploid, weighed 100 g, and had only two histopathologic features of carcinoma. The second was diploid, weighed less than 50 g, and had three histouath- ologic features of carcinoma. This neoplasm was asure aldosterone producer.
Discussion
Both neoplasm size and histology have been advocated as useful predictors of recurrence in adrenocortical neo- plasms. Tang and Gray5 found neoplasm weight to be the single most useful prognostic feature; all carcinomas in their series weighed more than 95 g, while no adenoma weighed more than 40 g. Similarly, Schteingart et al.6 found that adenomas rarely weighed more than 50 g. Neoplasm weight in excess of 100 g was also an important adverse prognostic factor in the series of Hough et al.7 In a recent report Cagle et al.8 found size to be the only morphologic predictor of biologic behavior in pediatric adrenocortical neoplasms.
The use of histologic features to predict behavior in adrenocortical neoplasms has been controversial. Series by Heinbecker et al.9 Schteingart et al.,6 and Weiss4 all
| Tumor weight | Histopathologic features | |||||
|---|---|---|---|---|---|---|
| Small ≤50 g | Borderline 51 g-100 g | Large >100 g | ||||
| <2 Adenoma | 3 Borderline | >4 Carcinoma | ||||
| DNA ploidy | ||||||
| Diploid | 36 | 2 | 1 | 37 | 2 | - |
| Hypotriploid Aneuploid | 1 | 2 | 3 | 2 | 1 | 3 |
| Hypertriploid Aneuploid | - | - | 3 | - | - | 3 |
| Histopathologic features | ||||||
| ≤2 adenoma | 36 | 2 | 1 | |||
| 3 borderline | 1 | 1 | 1 | |||
| ≥4 carcinoma | - | 1 | 5 | |||
| DNA ploidy | Tumor weight | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Diploid | Hypotriploid Aneuploid | Hypertriploid Aneuploid | Small ≤50 g | Borderline 51 g-100 g | Large >100 g | <2 Adenoma | Histopathologic 3 Borderline | features >4 Carcinoma | |
| Clinical outcome | |||||||||
| Benign | 28 | 3 | 0 | 28 | 1 | 2 | 29 | 1 | 1 |
| Malignant | 1 | 1 | 3 | 1 | 1 | 3 | 1 | 1 | 3 |
found microscopic features of value in distinguishing be- nign from malignant neoplasms. However, they were un- able to find one histologic feature that consistently sepa- rated carcinomas from adenomas. Some investigators have questioned the value of microscopic findings in di- agnosing adrenocortical carcinoma. Lewinsky et al.,10 re- porting on a series of nonfunctioning adrenocortical neo- plasms, found histologic appearance to be an unreliable indicator of metastatic potential. Other reports have stressed that histologic appearance is less effective than neoplasm size and clinical features in predicting out- come.5,7,8
Rapid analysis of nuclear DNA content by flow cytom- etry has been used in the study of an increasing number of solid tumors11,12 and has been advocated as an objective prognostic factor in adrenocortical neoplasms. Klein et al.,1 in a study of two adenomas and four carcinomas, found that the adenomas were diploid and the carcinomas aneuploid. Three of the carcinomas metastasized. Whether or not flow cytometric DNA analysis provided additional discriminating power compared with conven- tional criteria of malignancy is unclear, since the authors do not discuss how the diagnoses of adenoma and carci- noma were made.
In our study, nuclear DNA content was compared with neoplasm classification as determined by conventional morphologic criteria. Allowing for limitations in sample size, aneuploidy was as strongly associated with clinical malignancy as were large size and histologic features of carcinoma. Measurement of DNA content may be par- ticularly useful in neoplasms of uncertain malignant po- tential. For example, one neoplasm in our series was bor- derline by weight and an adenoma by histology. This neo- plasm was aneuploid and recurred several years after removal. In such cases with ambiguous morphologic findings, the detection of aneuploidy should be regarded as an ominous prognostic sign.
On the other hand, in the absence of demonstrable aneuploid DNA content, one cannot automatically as- sume that a given neoplasm will behave in a benign fash- ion. Neoplasm size and histology must also be considered. For example, one diploid neoplasm, a small aldosterone producer with borderline histology, proved to be clinically malignant. One might argue that this neoplasm had an occult population of tetraploid cells and, therefore, was not truly diploid. Although this possibility cannot be en- tirely ruled out, it seems unlikely. When three separate samples of neoplasm were analyzed, the percentage of cells in the G2-M peak was never more than 2% nor did this peak have its own replicating octaploid peak. It is also unlikely that a near-diploid aneuploid peak was hidden in the GO/G1 peak, since in the latter the coeffi- cients of variation were less than 4% for each sample.
The DNA indices of our aneuploid neoplasms were distributed into those around 1.30 (hypotriploid) and those around 1.70 (hypertriploid). In the series of Klein et al.1 the distribution is similar with relative DNA indices of 1.35, 1.65, 1.70, and 1.80 (DNA indices obtained by dividing their ploidy values by 2). Although our results suggest that hypertriploid neoplasms are more likely to recur than hypotriploid neoplasms (Table 3), more cases need to be analyzed before the clinical significance of this distinction can be fully evaluated. Studies of other solid neoplasms have suggested that DNA indices in the tri- ploid-tetraploid range are associated with a worse prog- nosis than those in the near diploid-hypotriploid range.11,12
This study demonstrates that flow cytometric DNA analysis can be easily and accurately performed on par- affin-embedded tissue from adrenocortical neoplasms. Neoplasm ploidy appears to be as effective a predictor of clinical outcome as size and histology; however, neither diploid DNA content, small size, nor the lack of histologic features associated with carcinoma can guarantee a benign clinical course. Flow cytometry may be most useful as an adjunctive study when a diagnosis cannot be made on morphologic grounds alone.
REFERENCES
1. Klein FA, Kay S, Ratliff JE, White FKH, Newsome HH. Flow cytometric determinations of ploidy and proliferation patterns of adrenal neoplasms: An adjunct to histological classification. J Urol 1985; 134: 933-935.
2. Hediey DW, Friedlander MI, Taylor IW, Rugg CA, Musgrove EA. Method for analysis of cellular DNA content of paraffin-embedded pathological material using flow cytometry. J Histochem Cytochem 1983; 31:1333-1335.
3. Schutte B, Reynders MMJ, Bosman FT, Blijham GH. Flow cy- tometric determination of DNA ploidy level in nuclei isolated from par- affin-embedded tissue. Cytometry 1985; 6:26-30.
4. Weiss L. Comparative histologic study of 43 metastasizing and nonmetastasizing adrenocortical tumors. Am J Surg Pathol 1984; 8:163- 169.
5. Tang CK, Gray GF. Adrenocortical neoplasms: Prognosis and morphology. Urology 1975; 5:691-695.
6. Schteingart DE, Oberman HA, Friedman BA, Conn JW. Adre- nocortical neoplasms producing Cushing’s syndrome. Cancer 1968; 22: 1005-1013.
7. Hough AJ, Holifield JW, Page DL, Hartmann WH. Prognostic factors in adrenal cortical tumors: A mathematical analysis of clinical and morphologie data. Am J Clin Pathol 1979; 72:390-399.
8. Cagle PT, Hough AJ, Pysher J et al. Comparison of adrenal cortical tumors in children and adults. Cancer 1986; 57:2235-2237.
9. Heinbecker P, O’Neal LA, Ackerman LV. Functioning and non- functioning adrenal cortical tumors. Surg Gynecol Obstet 1957; 105:21- 33.
10. Lewinsky BS, Gregor KM, Symington T, Neville AM. The clinical and pathological features of “non-hormonal” adrenocortical tumors: Report of 20 new cases and review of literature. Cancer 1974; 33:778- 790.
11. Barlogie B, Raber MN, Schumann J et al. Flow cytometry in clinical cancer research. Cancer Res 1983; 43:3982-3997.
12. Barlogie B, Johnston DA, Smallwood L et al. Prognostic impli- cations of ploidy and proliferative activity in human solid tumors. Cancer Genet Cytogenet 1982; 6:17-28.