Cellular DNA Profiles of Benign and Malignant Adrenocortical Tumors
Edmund S. Cibas, M.D., L. Jeffrey Medeiros, M.D., David S. Weinberg, M.D., Arnold B. Gelb, M.D., and Lawrence M. Weiss, M.D.
We evaluated the DNA content of 43 adrenocortical neo- plasms by flow cytometry and related it to histopatholog- ic criteria of malignancy and survival. Tumor tissue was selected from paraffin blocks and processed by a modifi- cation of the Hedley technique. The tumors were classi- fied as adenomas and carcinomas by the criteria of Weiss. Aneuploid stem-lines were identified in nine of 13 (69%) of the carcinomas and in six of 30 (20%) of the adenomas. Five of the six patients with aneuploid adenomas are alive and well (mean follow-up, 59 months); the sixth was lost to follow-up. Although there was a significant correlation between ploidy and histologic diagnosis (p = 0.041), the sensitivity and specificity of aneuploidy for predicting clinical outcome were only 56% and 65%, respectively. In addition, there was no significant difference in survival between patients with diploid versus aneuploid tumors, despite a highly statistically significant difference in sur- vival between patients with a histologic diagnosis of ad- enoma versus carcinoma (p = 0.00080). We found corre- lations between ploidy and tumor size, mitotic rate, and nuclear grade (p = 0.0033, p = 0.0017, and p = 0.018, respectively). There was also a significant correlation be- tween the proliferation fraction of a tumor and its nuclear grade (p = . 0093), but not its mitotic count or clinical outcome. Because both adrenal adenomas and carcino- mas may contain abnormal DNA stem-lines, ploidy alone is not a reliable discriminator in individual cases.
Key Words: Adrenocortical tumors-Adenomas-DNA content-Flow cytometry
Am J Surg Pathol 14(10); 948-955, 1990.
From the Departments of Pathology, Brigham and Women’s Hospital, Boston, Massachusetts (E.S.C., D.S.W.); Massachu- setts General Hospital, Boston, Massachusetts (L.J.M.); Har- vard Medical School, Boston, Massachusetts (E.S.C., D.S.W., L.J.M.); and Stanford University, Stanford, California (A.B.G., L.M.W.).
Address corresondence and reprint requests to Edmund S. Cibas, M.D., Department of Pathology, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115, U.S.A.
The distinction between benign and malignant ad- renocortical tumors is often difficult. Although sev- eral recent studies have proposed histologic criteria by which such a distinction may be possible (18,32,34), borderline cases are still encountered. These indeterminate cases are often small tumors with several worrisome histologic features (14)-a category of tumor that is becoming more frequently encountered with the increasing use of noninvasive imaging techniques such as computerized tomogra- phy and magnetic resonance imaging.
In the past several years, cellular DNA content as measured by flow cytometry (FCM) has been pro- posed as a rapid, objective, quantitative, and accu- rate means to distinguish between benign and ma- lignant neoplasms of many sites, based on the as- sumption that benign tumors have normal DNA content and that an abnormal DNA content (aneu- ploidy) is present in the majority of malignant neo- plasms (2,6). In addition, several studies have dem- onstrated a relationship between aneuploidy and prognosis (24,26,29). This has been shown for breast carcinoma (7), bladder carcinoma (16), non- small-cell lung carcinoma (33), colorectal carci- noma (23), ovarian carcinoma (12), as well as other tumors. Furthermore, measurements of the tumor cell cycle have been found to correlate with prog- nosis in some tumors (3,7,26,33).
In this study, a large series of adrenocortical tu- mors, all assessed by previously proposed histo- logic criteria, were analyzed to determine the diag- nostic value of cellular DNA content as well as its predictive value as a prognostic indicator.
METHODS
Pathologic Examination
Forty-five cases of adrenocortical tumors, acces- sioned between 1972 and 1988, were retrieved from
the files of the Massachusetts General Hospital. All cases were reviewed by one of us (L.J.M.) and clas- sified as either an adrenal adenoma or carcinoma by the criteria of Weiss (34).
Briefly, nine histopathologic features were exam- ined. Nuclear grade was assessed according to the criteria of Fuhrman et al. (13). Mitoses were eval- uated by counting 10 random high-power fields (hpf) in the area of the greatest number of mitotic figures on the five slides with the greatest number of mitoses. The counts were performed with an Amer- ican Optical microscope using a ×45 objective with a field of 0.47 mm in diameter. The presence or absence of atypical mitoses (an abnormal distribu- tion of chromosomes or an excessive number of mitotic spindles) was recorded.
The cytoplasm was evaluated for the percentage of clear or vacuolated cells resembling the normal zona fasciculata. Diffuse architecture was present if more than one-third of the tumor formed patternless sheets of cells; otherwise the architecture was re- garded as nondiffuse. Necrosis was considered to be present when it involved confluent nests of cells. Invasion of sinusoids (endothelium-lined vessels with little supporting tissue) and veins (vessels with smooth muscle in the wall) was accepted only when unequivocal. Capsular invasion consisted of nests or cords of tumor within or penetrating the capsule, with a corresponding stromal reaction.
Single points were scored for each of the follow- ing nine criteria of malignancy: (a) nuclear grade III or IV, (b) mitotic rate >5 per 50 hpf, (c) the pres- ence of atypical mitoses, (d) clear cytoplasm com- prising <25% of the tumor, (e) diffuse architecture, (f) necrosis, (g) venous invasion, (h) sinusoidal in- vasion, and (i) capsular invasion. A tumor with two or fewer criteria was categorized as an adenoma; tumors with three or more criteria were considered carcinoma. Tumor size and weight were also re- corded.
Preparation of Nuclear Suspensions
A representative block of paraffin-embedded tu- mor tissue was selected and examined for DNA content by flow cytometry. Sections with extensive necrosis or small amounts of tumor were avoided. To confirm the presence of adequate tumor in the samples analyzed for DNA content, two additional 5-um sections were cut and examined, one above and one below the section for FCM.
The suspensions of tumor cell nuclei were pre- pared from thick sections of the paraffin blocks by the method of Hedley et al. (17) as modified by Stephenson et al. (30). Briefly, two 50-um sections
were cut from the selected paraffin block and placed in a fine nylon mesh bag (Shandon Southern Instruments, Sewickley, PA, U.S.A.), then depar- affinized and rehydrated by successive immersions for 10 min each in three changes of xylene and two changes each of 100%, 95%, 70%, 50% ethanol, and distilled water. After rehydration the tissue was transferred to a test tube containing 1 ml of 0.5% pepsin (Sigma, St. Louis, MO, U.S.A.) in 0.9% NaCl adjusted to pH 1.5 with HCI and incubated for 1 h at 37° C with occasional vortexing. This incu- bation digests cytoplasmic proteins and releases whole nuclei. The reaction was quenched by the addition of 4-5 ml Hank’s balanced salt solution (HBSS) without calcium and magnesium, and the samples were sieved through a 53-um nylon mesh filter disk (Tetko, Elmsford, NY, U.S.A.) to re- move tissue fragments and cellular or nuclear clus- ters. They were washed once by centrifuging for 5 min at 1,000 rpm at room temperature and resus- pended in 0.5 ml of HBSS without calcium and mag- nesium. Specimens were stored at 4º C until mea- sured.
Measurement of DNA Content
Nuclei for each sample were counted with a he- mocytometer, and an aliquot containing 1 × 106 nu- clei was placed in a plastic test tube. To each ali- quot was added 0.5 ml of propidium iodide (PI) so- lution (50 µg/ml in NIM, Sigma) and 15 ul of RNase (final concentration 500 u/ml, ribonuclease A, bo- vine pancreas type III-A; Sigma). The nuclei were incubated at room temperature for 20-30 min. Im- mediately before analysis, the suspension was passed twice through a 26 gauge needle to disperse nuclear aggregates. The stained samples were then measured on a FACS Analyzer (Becton Dickinson, Mountain View, CA, U.S.A.) equipped with appro- priate filters to excite fluorescence at 488 nm and to detect emission above 570 nm. Electronic volume gates were left open to include all sizes of nuclei, and cellular debris was excluded by using the PI fluorescence signal rather than the volume signal to trigger data collection. Histograms were generated from 5,000 nuclei and displayed as linear fluores- cence.
Evaluation of DNA Content
A sample was defined as diploid if there was a single G0/G1 peak. Samples were called aneuploid if there was more than one distinct GO/G1 peak, and the peak with the lowest DNA content was assumed to represent diploid cells. The DNA index (DI) of aneuploid populations was calculated by dividing
the mean fluorescence intensity of each aneuploid G0/G1 peak, expressed in arbitrary units (channel numbers from 0 to 256), by the mean fluorescence intensity of the diploid peak. Peaks with DI at or close to 2.0 were considered aneuploid if they con- tained more than 20% of the total population of nu- clei analyzed; less than 20% was considered within the possible range of G2M nuclei and coincident doublets. Samples with more than one aneuploid peak were further subclassified as multiploid. The decision to call a tumor diploid or aneuploid was made without knowledge of clinical data, histologic diagnosis, or outcome.
The proliferation fraction (PF) was defined as the percentage of cells in S + G2 + M phases of the cell cycle. This was readily calculated for diploid tumors. The PF of aneuploid tumors was calculated only in the absence of multiple aneuploid peaks, which contain overlapping cell populations. The aneuploid PF was arbitrarily calculated by ignoring the diploid population. This was done with the rec- ognition that the diploid population often over- lapped the aneuploid population, leading to an over- estimation of cells in the aneuploid PF.
Statistical Methods
Null hypothesis of no difference between cyto- metric, histologic, and clinical parameters was tested. To analyze categorical data such as ploidy versus histologic diagnosis, Fisher’s exact test was performed because of the relatively small number of cases and expected frequencies (10). When data was from an ordered category such as nuclear grade, either the Mann-Whitney test (corrected for ties) or Kruskal-Wallis test (25) was applied, de- pending on whether two or more groups were in- volved (27). Two-sided p values were used for all tests. Individual « levels were set at 0.01 to de- crease the risk of false positives arising from mul- tiple comparisons (15). This level represents a com- promise with the risk of falsely conceding ß error since the number of cases is small. Sensitivity, specificity, and other indices of efficiency in explor- ing the use of flow cytometry as a diagnostic test were calculated according to standard definitions (10).
Estimates of survival distributions were com- puted from the follow-up data using the product- limit method (20). Breslow’s version of the gener- alized Wilcoxon statistic was used to test the equal- ity of survival curves (5). Statistical analyses and graphical output were generated with the following software packages: BMDP (9), Statview SE + graphics (11), and MacDraw (8).
RESULTS
Acceptable DNA histograms were obtained after the examination of a single paraffin block in 41 of 45 cases. In two additional cases, the interpretation of an initially questionable DNA profile was clarified as unequivocally aneuploid after the examination of additional (one or two) blocks of tumor. The re- maining two of the original 45 cases (4%) yielded uninterpretable histograms due to excessive debris despite the examination of additional blocks; these two cases were eliminated from further consider- ation.
The clinical and gross pathologic features of the remaining 43 cases are summarized in Table 1. There were 30 adenomas and 13 carcinomas. Fol- low-up was available in 32 cases (74%), 20 adeno- mas and 12 carcinomas. Median follow-up time was 54 months (61 months for patients not dead of dis- ease), with a range of 6 to 152 months. None of the patients with an adenoma developed recurrence or metastasis, as compared to nine of the 13 patients with carcinoma.
The microscopic findings are summarized in Ta- ble 2. As reported previously (34), no one criterion
| Adenomas (n = 30) | Carcinomas (n = 13) | |
|---|---|---|
| Age | ||
| Median | 51 | 52 |
| Range | 5-78 | 21-76 |
| Sex | ||
| Male | 14 | 5 |
| Female | 16 | 8 |
| Side | ||
| Left | 11 | 7 |
| Right | 17 | 6 |
| Unknown | 2 | 0 |
| Clinical syndrome | ||
| Cushing's | 9 | 7 |
| Aldosteronism | 6 | 0 |
| Sexual precocity | 1 | 0 |
| Hypertension | 1 | 0 |
| No function | 9 | 6 |
| Unknown | 4 | 0 |
| Size (greatest diameter) | ||
| Median | 3.5 cm | 10 cm |
| Range | 1.5-9.5 cm | 4.5-18 cm |
| Weight (n = 22) | ||
| Median | 36 g | 592 g |
| Range | 12-230 g | 142-1480 g |
| Outcome | ||
| A-NED | 18 | 2 |
| D-NED | 2 | 1 |
| AWD | 0 | 2 |
| DOD | 0 | 7 |
| Unknown | 10 | 1 |
A-NED = alive, no evidence of disease; D-NED = dead, no evidence of disease; AWD = alive with disease; DOD = dead of disease.
| Adenomas (n = 30) | Carcinomas (n = 13) | |
|---|---|---|
| Nuclear grade | ||
| 1 | 7 | 0 |
| 11 | 4 | 1 |
| III | 2 | 4 |
| IV | 17 | 8 |
| Mitotic rate (per 50 hpf) | ||
| 0-5 | 30 | 4 |
| Greater than 5 | 0 | 9 |
| Atypical Mitoses | ||
| Absent | 30 | 5 |
| Present | 0 | 8 |
| Cytoplasm | ||
| Greater than 25% clear | 25 | 1 |
| 0-25% clear | 5 | 12 |
| Architecture | ||
| Nondiffuse | 29 | 4 |
| Diffuse | 1 | 9 |
| Necrosis | ||
| Absent | 22 | 2 |
| Present | 8 | 11 |
| Venous invasion | ||
| Absent | 30 | 6 |
| Present | 0 | 7 |
| Sinusoidal invasion | ||
| Absent | 30 | 4 |
| Present | 0 | 9 |
| Capsule invasion | ||
| Absent | 30 | 5 |
| Present | 0 | 8 |
alone could be used to distinguish a malignant from a benign tumor.
Twenty-eight tumors were diploid, and 15 (35%) were aneuploid (Fig. 1). The mean coefficient of variation (CV) for the GO/G1 peaks of diploid tu- mors was 4.7 (range, 2.9-8.6). The diploid GO/G1 peaks of aneuploid tumors had a mean CV of 5.1 (range, 2.7-9.1), and the aneuploid GO/G1 peaks had a mean CV of 5.6 (range, 2.4-9.3).
Four of the 15 aneuploid tumors (27%) were mul- tiploid, each containing two abnormal peaks. The median DNA index for aneuploid peaks was 1.50 (range, 1.13-2.73).
The PF was calculated for all diploid tumors and for the 11 aneuploid tumors with only a single an- euploid peak. The mean PF was 9.7% (range, 2.8-19.4%) for diploid tumors and 17.7% (range, 7.3-36.9%) for aneuploid tumors.
Table 3 summarizes the histologic, ploidy, prolif- eration, and survival data for all cases. Nine of the 13 carcinomas (69%) were aneuploid, as were six of the 30 adenomas (20%). Figure 2 displays the six aneuploid histograms derived from benign tumors. Follow-up information was available for five of these six patients, all of whom are alive with no evidence of disease (mean follow-up, 59 months).
There was significant correlation between ploidy
1000
0
0
100
200
320
0
0
100
200
and histologic diagnosis (Fisher’s exact test: p = 0.041). However, the sensitivity and specificity of aneuploidy for predicting clinical outcome were 56% and 65%, respectively, in contrast to a sensi- tivity of 100% and a specificity of 87% of histologic diagnosis for predicting clinical outcome. Inasmuch as the prevalence of metastases was 28% in this patient population with adrenal neoplasms, the pre- dictive value of aneuploid versus diploid results for predicting clinical outcome was 38% and 75%, re- spectively. Table 4 shows the 2 x 2 observed fre- quency table for histologic diagnosis and ploidy.
A statistically significant difference in survival exists between patients with adenoma versus carci- noma (Breslow test, p = 0.00080) (Fig. 3). No sta- tistically significant difference in survival was seen between patients with diploid and aneuploid tumors (Fig. 4).
Aneuploidy showed a statistically significant cor- relation with increased tumor size (Mann-Whitney test, p = 0.0033), increased mitotic rates (p = 0.0017), and high nuclear grade (p = 0.018). A sig- nificant correlation was found between the prolifer- ation fraction of a tumor and its nuclear grade (Kruskal-Mann test, p = 0.0093),but not its mitotic count or clinical outcome.
| Case | No. of criteria | Diagnosis | Ploidy | Prolif. fraction | F/U status | F/U time (mo.) |
|---|---|---|---|---|---|---|
| 1 | 1 | Adenoma | Diploid | 9.7 | NED | 120 |
| 2 | 2 | Adenoma | Diploid | 11.7 | NED | 122 |
| 3 | 1 | Adenoma | Diploid | 12.9 | - | |
| 4 | 1 | Adenoma | Diploid | 3.9 | NED | 101 |
| 5 | 0 | Adenoma | Diploid | 5.9 | NED | 152 |
| 6 | 1 | Adenoma | Diploid | 14.5 | D-NED | |
| 7 | 0 | Adenoma | Diploid | 5.9 | NED | 16 |
| 8 | 2 | Adenoma | Diploid | 9.3 | NED | 54 |
| 9 | 0 | Adenoma | Diploid | 5.7 | NED | 88 |
| 10 | 1 | Adenoma | Diploid | 9.7 | ||
| 11 | 1 | Adenoma | Diploid | 8.1 | NED | 12 |
| 12 | 1 | Adenoma | Diploid | 16.2 | NED | 69 |
| 13 | 1 | Adenoma | Diploid | 9.2 | D-NED | 67 |
| 14 | 1 | Adenoma | Diploid | 2.8 | - | |
| 15 | 0 | Adenoma | Diploid | 5.1 | NED | 51 |
| 16 | 1 | Adenoma | Diploid | 2.8 | - | |
| 17 | 0 | Adenoma | Diploid | 7.9 | - | |
| 18 | 0 | Adenoma | Diploid | 8.3 | - | - |
| 19 | 1 | Adenoma | Diploid | 13.8 | - | |
| 20 | 2 | Adenoma | Diploid | 10.0 | NED | 15 |
| 21 | 0 | Adenoma | Diploid | 6.4 | - | |
| 22 | 1 | Adenoma | Diploid | 11.7 | NED | 6 |
| 23 | 2 | Adenoma | Diploid | 11.9 | - | |
| 24 | 2 | Adenoma | Diploid | 13.2 | NED | 13 |
| 25 | 2 | Adenoma | Aneuploid | 33.8 | NED | 81 |
| 26 | 2 | Adenoma | Aneuploid | 36.9 | NED | 132 |
| 27 | 2 | Adenoma | Aneuploid | 10.5 | - | |
| 28 | 1 | Adenoma | Aneuploid | 26.5 | NED | 42 |
| 29 | 2 | Adenoma | Aneuploid | mult. | NED | 22 |
| 30 | 2 | Adenoma | Aneuploid | 21.1 | NED | 16 |
| 31 | 8 | Carcinoma | Diploid | 19.7 | DOD | 29 |
| 32 | 4 | Carcinoma | Diploid | 3.2 | AWD | 101 |
| 33 | 9 | Carcinoma | Diploid | 19.4 | DOD | 12 |
| 34 | 8 | Carcinoma | Diploid | 14.0 | DOD | 39 |
| 35 | 9 | Carcinoma | Aneuploid | 17.5 | DOD | 66 |
| 36 | 4 | Carcinoma | Aneuploid | mult. | D-NED | 89 |
| 37 | 4 | Carcinoma | Aneuploid | mult. | DOD | 25 |
| 38 | 7 | Carcinoma | Aneuploid | 7.3 | DOD | 67 |
| 39 | 7 | Carcinoma | Aneuploid | 8.6 | NED | - |
| 40 | 4 | Carcinoma | Aneuploid | 8.3 | NED | 67 |
| 41 | 8 | Carcinoma | Aneuploid | mult. | DOD | 14 |
| 42 | 7 | Carcinoma | Aneuploid | 11.6 | AWD | 16 |
| 43 | 6 | Carcinoma | Aneuploid | 13.1 | - | - |
F/U = followup; NED = alive, no evidence of disease; D-NED = dead, no evidence of disease; AWD = alive with disease; DOD = dead of disease; mult. = multiple aneuploid peaks.
DISCUSSION
Previous studies applying flow cytometric analy- sis of DNA content to the assessment of adrenocor- tical neoplasms have had conflicting results. Origi- nally, there was great enthusiasm that DNA content measurements could make a significant contribu- tion to the pathologic evaluation of these tumors. For example, Klein and colleagues reported in 1985 that all seven benign adrenal neoplasms and four normal adrenal glands were diploid, but that four adrenocortical carcinomas contained cells with ane- uploid stem-lines (21). Furthermore, they used flow cytometry to confirm a diagnosis of carcinoma in a borderline adrenal neoplasm (22).
One year later, Bowlby et al. reported that all of 16 adrenocortical adenomas had a diploid DNA pat- tern, while five of six adrenocortical carcinomas were aneuploid, including all three with metastases (4). The next year, Taylor et al. reported flow cy- tometric observations on adrenocortical tumors oc- curring in children (31): Seven tumors occurring in patients without metastases were diploid, and five tumors were aneuploid; four of the five subse- quently metastasized. The results of these three studies suggested that DNA content abnormalities detected by flow cytometry correlate with histo- logic assessment and can provide an objective and accurate measure of the biologic potential of these tumors.
400
400
0
0
100
200
0
®
100
200
300
300
0
®
100
200
0
0
100
200
300
300
0
®
100
200
0
0
100
200
Unfortunately, recent studies have shown that there may not be a perfect relationship between aneuploidy, histologic assessment, and even clini- cal outcome. In agreement with previous studies, Amberson and collegues found in 1987 that neo- plasms that recurred or metastasized were more apt to be aneuploid than those showing no evidence of further disease during the follow-up period (p < 0.005) (1). Interestingly, they found two tumors with histopathologic features of adenoma that had an aneuploid DNA content. One of these tumors recurred several years after surgical excision, sug- gesting that the results of flow cytometry were su- perior to histopathology in predicting the outcome of this case. However, when Joensuu and Klemi examined 164 adenomas from endocrine organs,
| Benign | Malignant | Totals | |
|---|---|---|---|
| Diploid | 24 | 4 | 28 |
| Aneuploid | 6 | 9 | 15 |
| Totals | 30 | 13 | 43 |
they found aneuploidy in 25% or more of cases, depending on the organ site examined (19). Of 17 adrenal adenomas studied, nine (53%) were un- equivocally aneuploid. Eleven of these adenomas were smaller than 5 cm in largest diameter. Follow- up, which was available in 16 of the 17 patients, showed no tumor recurrence or metastases in 66 total years of follow-up. The authors concluded that DNA aneuploidy is compatible with benign histol- ogy. From a clinical point of view, they questioned its use as a definitive criterion for malignancy.
Our study examined a large number of adrenal neoplasms using strict histologic criteria in which a long period of clinical follow-up was available. Like the previous investigators, we found a significant correlation between histologic diagnosis and ploidy. However, the sensitivity and specificity of ploidy for predicting clinical outcome were low, and no significant differences in survival were seen be- tween patients with diploid and those with aneu- ploid tumors, despite the presence of a highly sta- tistically significant difference in survival between patients with a histologic diagnosis of adenoma and those with carcinoma. Our findings were similar to
Benign versus Malignant
1.0
0.8
Probability
0.6
0.4
Malignant Benign
0.2
0
100
200
Time (months)
those of Joensuu and Klemi; six of 30 (20%) adeno- mas were aneuploid. Follow-up information was available for five of six of these patients, and none showed evidence of disease over a total of 295 months of follow-up.
The significant correlation between DNA ploidy and nuclear grade in this study was not surprising, because nuclear grade is, in part, a crude visual assessment of DNA content. Just as nuclear grade correlates only imperfectly with clinical behavior in
Diploid versus Aneuploid
1.0
0.8
Probability
0.6
0.4
Aneuploid
· Diploid
0.2
0
100
200
Time (months)
adrenocortical neoplasms, however, so too did DNA ploidy correlate poorly with survival. There was no correlation between the proliferation frac- tion of a tumor and its mitotic rate, but this may be a consequence of the way we assessed the mitotic rate. Rather than counting mitotic figures in random fields-a procedure that would more closely simu- late flow cytometric analysis-we evaluated only selected fields with the highest number of mitoses.
In summary, because ploidy analysis has a low sensitivity and specificity for predicting outcome, and because we found no significant difference in survival between patients with diploid and patients with aneuploid tumors, we cannot recommend a strong role for flow cytometry in the assessment of the biologic potential of these tumors. Flow cyto- metric evaluation might be of some value when the histologic evaluation cannot be adequately per- formed-e.g., when only one section of a large tu- mor is available and the status of the capsule, the presence or absence of venous and sinusoidal inva- sion, and the mitotic rate cannot be assessed. Flow cytometry might also have a limited role when the clinical, gross, and histologic findings are equivocal or contradictory. However, we cannot recommend basing a diagnosis in an individual patient solely on the results of DNA ploidy analysis.
REFERENCES
1. Amberson JB, Vaughan ED, Gray GF, Naus GJ. Flow cy- tometric analysis of nuclear DNA from adrenocortical neo- plasms: a retrospective study using paraffin-embedded tis- sue. Cancer 1987;59:2091-5.
2. Barlogie B, Raber MN, Schuman J, et al. Flow cytometry in clinical cancer research. Cancer Res 1983;43:3982-97.
3. Bauer KD, Merkel DE, Winter JN, et al. Prognostic impli- cations of ploidy and proliferative activity in diffuse large cell lymphomas. Cancer Res 1986;46:3173-8.
4. Bowlby LS, DeBault LER, Abraham SR. Flow cytometric analysis of adrenal cortical tumor DNA: relationship be- tween cellular DNA and histopathologic classification. Can- cer 1986;58:1499-505.
5. Breslow N. A generalized Kruskal-Wallis test for comparing k samples subject to unequal patterns of censorship. Bio- metrika 1970;57:579-94.
6. Buchner TH, Hiddemann W, Wormann B, et al. Differential pattern of DNA-aneuploidy in human malignancies. Pathol Res Pract 1985;179:310-7.
7. Clark GM, Dressler LG, Owens MA, Pounds G, Oldaker T, McGuire WL. Prediction of relapse or survival in patients with node-negative breast cancer by DNA flow cytometry. N Engl J Med 1989;320:627-33.
8. Cutter M. MacDraw, version 1.7. Cupertino: Apple Com- puter, 1985.
9. Dixon WJ, ed. BMDP statistical software. Berkeley: Univ. of California Press, 1985:555-94.
10. Feinstein AR. Clinical biostatistics. St. Louis: CV Mosby, 1977.
11. Feldman D, Gagnon J, Hoffman R, Simpson J. Statview SE + graphics. Berkeley: Abacus Concepts, 1988.
12. Friedlander ML, Hedley DW, Swanson C, Russell P. Pre- diction of long-term survival by flow cytometric analysis of
cellular DNA content in patients with advanced ovarian can- cer. J Clin Oncol 1988;6:282-90.
13. Fuhrman SA, Lasky LC, Limas C. Prognostic significance of morphologic parameters in renal cell carcinoma. Am J Surg Pathol 1982;6:655-63.
14. Gandour MJ, Grizzle WE. A small adrenocortical carcinoma with aggressive behavior: an evaluation of criteria for ma- lignancy. Arch Pathol Lab Med 1986;110:1076-9.
15. Godfrey K. Comparing the means of several groups. N Engl J Med 1985;313:1450-6.
16. Gustafson H, Tribukait B, Esposti PL. DNA profile and tumour progression in patients with superficial bladder tu- mours. Urol Res 1982;10:13-8.
17. Hedley DW, Friedlander ML, Taylor IW, Rugg CA, Mus- grove EA. Method for analysis of cellular DNA content of paraffin-embedded pathological material using flow cytom- etry. J Histochem Cytochem 1983;31:1333-5.
18. Hough AJ, Hollifield JW, Page DL, Harmann WH. Prognos- tic factors in adrenal cortical tumors: a mathematical analy- sis of clinical and morphologic data. Am J Clin Pathol 1979;72:390-9.
19. Joensuu H, Klemi PJ. DNA aneuploidy in adenomas of en- docrine organs. Am J Pathol 1988;132:145-51.
20. Kaplan EL, Meier P. Nonparametric estimation from incom- plete observations. J Am Stat Assoc 1958;53:457-81.
21. Klein FA, Kay S, Ratliff JE, White FKH, Newsome HH. Flow cytometric determinations of ploidy and proliferation patterns of adrenal neoplasms: an adjunct to histologic clas- sification. J Urol 1985a;134:862-6.
22. Klein FA, Miller NL, Hackler RH. Flow cytometry in fem- inizing adrenocortical carcinoma. J Urol 1985b;134:933-5.
23. Kokal W, Sheibani K, Terz J, Harada JR. Tumor DNA con- tent in the prognosis of colorectal carcinoma. JAMA 1986; 255:3123-7.
24. Koss LG, Czerniak B, Herz F, Wersto RP. Flow cytometric measurements of DNA and other cell components: a critical appraisal. Hum Pathol 1989;20:528-48.
25. Kruskal WH, Wallis WA. Use of ranks in one-criterion vari- ance analysis. J Am Stat Assoc 1952;47:583-621; Errata 1953;48:907-11.
26. Merkel DE, Dressler LG, McGuire WL. Flow cytometry, cellular DNA content, and prognosis in human malignancy. J Clin Oncol 1987;5:1690-703.
27. Moses LE, Emerson JD, Hosseni H. Analyzing data from ordered categories. In: Bailar JC III, Mosteller F, eds. Med- ical Uses of Statistics. Waltham: MA, NEJM Books, 1986.
28. Seckinger D, Sugarbaker E, Frankfurt O. DNA content in human cancer. Application in pathology and clinical medi- cine. Arch Pathol Lab Med 1989;113:619-26.
29. Stephenson RA, Gay H, Fair WR, Melamed MR. Effect of section thickness on quality of flow cytometric DNA content determinations in paraffin-embedded tissues. Cytometry 1986;7:41-4.
30. Taylor SR, Roederer M, Murphy RF. Flow cytometric DNA analysis of adrenocortical tumors in children. Cancer 1987; 59:2059-63.
31. Van Slooten H, Schaberg A, Smeenk D, Moolenaar AJ. Morphologic characteristics of benign and malignant adre- nocortical tumors. Cancer 1985;55:766-73.
32. Volm M, Drings P, Mattern J, Sonka J, Vogt-Moykopf I, Wayss K. Prognostic significance of DNA patterns and re- sistance-predictive tests in non-small cell lung carcinoma. Cancer 1985;56:1396-403.
33. Weiss LM. Comparative histologic study of 43 metastasizing and non-metastasizing adrenocortical tumors. Am J Surg Pathol 1984;8:163-9.
34. Weiss LM, Medeiros LJ, Vickery AL. Pathologic features of prognostic significance in adrenocortical carcinoma. Am J Surg Pathol 1989;13:202-6.