Neoplasms of endometrial stroma: Histopathologic and flow cytometric analysis with clinical correlation

Neoplasms of Endometrial Stroma: Histopathologic and Flow Cytometric Analysis With Clinical Correlation CAREY Z. AUGUST, MD, KENNETH D. BAUER, PHD, JOHN LURAIN, MD, AND TARIQ MURAD, MD, PHD Ten neoplasms of endometrial stroma (one stromal nodule, four endolymphatic stromal myoses, and five stromal sarcomas) were compared using clinical data as well as histopathologic and flow cytometric parameters. None of the patients with stromal nodules or endolymphatic stromal myosis had extrauterine disease at presentation or tumors displaying a mitotic rate > lO/lO hpf (highpower fields), nuclear pleomorphism, atypical mitotic figures, DNA aneuploidy, or a high proliferative index (> 10% S phase cells). The stromal nodules were circumscribed and behaved in a benign fashion. The patients with endolymphatic stromal myosis had infiltrative tumors that behaved as low-grade cancers with good responses to therapy. Four of the five patients with stromal sarcoma had extrauterine disease at presentation as well as tumors characterized by a mitotic rate much > 10 mitoses/lO hpf, nuclear pleomorphism, atypical mitotic figures, DNA aneuploidy, and a high proliferative index. These four patients had aggressive disease with poet response to therapy. The fifth of the stromal sarcomas had a high mitotic rate, but lacked the other features linked with aggressive behavior; this patient has responded well to treatment. The mitotic count may not be the most useful criterion for predicting biologic behavior in endometrial stromal tumors since it does not always reflect an increased rate of cell turnover as demonstrated by the percentage of cells in the S phase. DNA analysis by flow cytometry yields a more accurate picture of tumor behavior. HUM PATHOL 20:232-237. 0 1989 by W.B. Saunders Company.

Neoplasms of endometrial stroma fall into three groups: stromal nodule (SN), endolymphatic stromal myosis (ESM), and stromal sarcoma (SS). In general, these tumors are easily separated into the three groups on the basis of their clinical course as well as their morphology. There has been some confusion in the literature, however, about the histologic definitions, histogenesis, and, most importantly, the biologic behavior of these tumors. For example, synonyms that have been applied to ESM include both “benign stromal endometriosis” and “malignant stroma1 endometriosis.“’ The purpose of the present study is to examine a group of patients with endometrial stromal tumors using clinical, morphologic, and flow cytometric parameters in order to identify From the Departments of Pathology and Obstetrics and Gynecology, Northwestern University Medical School, Chicago. Accepted for publication July 13, 1988. Presented in part in poster format at the 1988 International Academv of Patholortv annual meeting, Washineton, DC. Key ‘zuords: stromal nodule, Strom>1 sarcoma, endolymphatic stromal myosis, endometrium. Address correspondence and reprint requests to Carey Z. August, MD, Passavani Pavilion, Room-300, Northwestern Memorial Hospital, Superior St and Fairbanks Ct, Chicago, IL 60611. 0 1989 by W.B. Saunders Company. 0046-8 177189/2003-0007$5.0010

232

the most useful factors for predicting the natural course of the tumors and response to treatment. MATERIALS AND METHODS All cases diagnosed as SN, ESM, or endometrial SS accessed in the surgical pathology department at Northwestern Memorial Hospital (Chicago) between 1976 and 1987 were reviewed. These included consultations sent to the department, as well as material from the hospital. After confirmation of diagnoses, ten patients were selected: one with an SN, four with ESM, and five with SS. Clinical parameters, including age, menstrual status, parity, presenting symptoms, treatment, and outcome, were obtained from the patients’ records (operative reports and gross descriptions were used to determine the extent of the tumor at presentation). Hematoxylin-eosin (H&E)-stained sections were examined; the number of slides available ranged from one to 11 per case. The following histologic parameters were considered: nuclear pleomorphism, degree of cellularity, degree of vascularity, necrosis, number of mitoses, presence of atypical mitoses, and presence of tumor giant cells. Features constituting nuclear pleomorphism included the following: a threefold or greater variation in the size of nuclei, irregularity of nuclear outline, hyperchromatism, chromatin clumping, and variably enlarged nucleoli. The number of mitotic figures per 10 high-power fields (hpf) was determined simultaneously by two pathologists (CA. and T.M.) in the most active areas of the tumors. At least 20 hpf were assessed in every case; in cases with sufficient material available, 40 hpf were assessed. For flow cytometric analysis, paraffin-embedded tissue was deparaffinized and dissociated using a previously reported modification of the method developed by Hedley et aL2 Approximately five 50-km sections were taken from paraffin blocks containing tissue that had been fixed in buffered 10% formalin. The blocks chosen for flow cytometry were the same ones used for assessing mitotic rate in H&E-stained slides, and additional H&E-stained sections were examined from the same blocks after the 50+m sections were taken to insure uniformity of the material. Portions of tumor were dissected from non-tumorous endometrium using a blunt needle and scalpel; the nontumorous endometrium was used as control material for each case in which it was obtainable. Next, the tissue was deparaffinized, rehydrated, and washed with distilled water. This tissue was finely minced, suspended in 0.1% pepsin solution, and incubated for 30 minutes at 37°C. Pepsin proteolysis was stopped by the addition of 0.1 mL of pepstatin A, and the digested material was filtered through a 37qm nylon monofilament mesh. Hank’s balanced salt solution plus 10 mmoUL of Hepes (HBSS-HEPES) was added to the filtrate. This mixture was centrifuged for ten minutes at 2OOxg and at 4”C, and the nuclear pellet was resuspended in 2 mL of HBSS-HEPES. Nuclear recovery was

NEOPIASMS OF ENDOMETRIAL STROMA [August et al)

assessed in several cases; the mean nuclear recovery was 3.39 X 10’ nuclei/g, with a range of 6.6 X IO6 to 6.1 X 10’ nuclei/g of deparaffinated tissue. Following tissue dissociation, the nuclei samples were stained with propidium iodide as described by Bauer et al.2 The specimens were maintained in a light-shielded tube at 4°C for at least one hour before flow cytometric (FCM) analysis. The fluorescence of propidium iodide (PI)-stained cells was monitored on an EPICS 752 flow cytometer (Coulter, Hialeah, FL) equipped with a 5-watt argon ion laser. Immediately before FCM analysis, the cell suspensions were filtered through a 37-Frn nylon monofilament mesh. PI-stained cells were excited at 488 nm, 500 mW, and fluorescence was monitored through a 515-nm laser interference filters. Fluorescent microspheres were used to assess instrument performance at the beginning of each run and as an external fluorescent standard. A total of 2.5 X lo4 cells was evaluated from each specimen except for case no. 4, for which 1.3 x 10’ cells were evaluated. The mean coefficient of variation of the G,/G, population in each sample for the entire study was 4.5 1. All samples were electronically gated on peak u integrated fluorescence signals to minimize retention of cell aggregates in the analyzed DNA distributions. DNA histogram display and analysis were performed using a Terak 8600 minicomputer (Terak, Scottsdale, AZ) and software developed by Salzman et aI9 as modified by Robinson and Leary of the University of Rochester. Cell cycle analysis was performed using a simple rectangle model previously described, 4 adapted to a Model XT personal computer (International Business Machines, Boca Raton, FL). In all histograms, a background of cellular debris was observed to contribute to the cell cycle distribution. To correct for this artifact, an exponential curve was plotted between the debris regions to the left of the G,/G, cells and to the right of the G,M cells, and the area under this curve was electronically subtracted from the DNA distribution. Cell proliferation in this investigation is expressed in terms of the percent of S phase cells. An aneuploid population was considered to be present only when two distinct GdG, populations were evident. The DNA index was calculated by dividing the mean fluorescence channel number of the aneuploid Go/G1 population by the mean channel number of the diploid Go/G, population.

clinical status). In all patients with ESM, tumor invaded the myometrium but was confined to the uterus. Except for patient no. 6, whose tumor invaded the myometrium but was confined to the uterus, all patients with SS had extrauterine disease at presentation. The patient with SN received no further treatment after the diagnosis of presumed SN was made from the curettings. The ESM patients were all treated by hysterectomy. All patients with SS were also treated by hysterectomy, except patient no. 10, who refused any therapy after biopsy of a vaginal nodule. Postsurgical therapies are summarized in Table 1. Follow-Up No evidence of further progression of disease has occurred in the SN patient. Patient no. 5 is the only ESM case with a recurrence. This patient presented with pelvic and intraabdominal masses 14 years after hysterectomy. She underwent partial tumor resection and bilateral salpingooophorectomy, followed by treatment with megestrol acetate and CYVADIC chemotherapy (cyclophosphamide, vincristine, adriamycin, dacarbazine); she had a complete clinical response and has no evidence of disease 3 years later. Of the SS cases, patient no. 10 was lost to follow-up after refusing therapy for stage IV disease. Except for patient no. 6, who remains free of disease 3 years after pelvic irradiation and CYVADIC chemotherapy, the other three SS patients had rapidly progressive disease typical of high-grade malignant stromal lesions, culminating in death at 8, 5, and 15 months from diagnosis, respectively. Histology The histologic findings are listed in Table 2. The histologic pattern of SN is well-described in the literature.5 In our patients, it was circumscribed, free of vascular infiltration, and composed of small, spindle-shaped cells that bore a striking resemblance to those of normal proliferative endometrial stroma. In addition, small blood vessels resembling spinal arterioles were present. For the most part, the histology of ESM was similar, except that these were infiltrative growths sometimes displaying vascular permeation (Figs 1 and 2). None of these tumors displayed nuclear pleomorphism, atypical mitotic figures, or tumor giant cells. The vascular pattern was especially prominent in two lesions. Necrosis was seen in only one case (patient no. 5). By definition, all cases of SN or ESM had < 10 mitoses/lO hpf.5 Tumors were designated SS by the criterion of Norris and Taylor5 (ie, mitotic rate > 10 mitoses/lO hpf in the most active areas). All stromal sarcomas, except for patient no. 6, displayed similar histologic features. They showed marked nuclear pleomorphism, atypical mitotic figures, and a prominent vascular pattern (Fig 3). Some, but not all, SS cases showed necrosis and tumor giant cells.

RESULTS Clinical Information All patients with either an SN or an ESM were young (age range, 23 to 36 years), except for patient no. 5, who was 72 years old. All patients with SS were older (age range, 50 to 80 years), except for patient no. 6, who was 33 years old. Three of the five patients with SN or ESM were parous, one was nulliparous, and the parity was unknown in one. Three of the five SS patients were parous, one was nulliparous, and the parity was unknown in one. Although hysterectomy was not performed to confirm the diagnosis of SN in patient no. 1, the diagnosis was made presumptively because the proliferation of cells was circumscribed rather than infiltrative, and because the same circumscribed proliferation was noted retrospectively in uterine curettings taken 5 years earlier (without change in the patient’s 233

HUMANPATHOLOGY

Volume 20, No. 3 (March1989)

TABLE 1. Clinical Information Patient No.

Age (yrl

Parity

Diagnosis

Postsurgical Therapy

Extent No myometrial invasion Inner 113 myometrial invasion Mid l/3 myometrial invasion Mid l/3 myometrial invasion

None

None

Alive and well at 5 yr

None

None

NED at 3.5 yr

Progestin

None

NED at 4.5 yr

None

14 yr

Radiation

None

NED at 3 yr 9 mo

*

DOD at 8 mo

Parous Parous

ss ss

Extrauterine Extrauterine

disease disease

* *

DOD at 5 mo DODat 15mo

Nulliparous

ss

Extrauterine

disease

CYVADIC, radiation-therapy CYVADIC, radiation therapy CYVADIC Radiation therapy, CYVADIC, tamoxifen Refused treatment

None

ss

Outer l/3 myometrial invasion Outer l/3 myometrial invasion Extrauterine-disease

CYVADIC and progestin therapy for recurrence; no further recurrence in 3 yr NED at 1.5 yr

*

Lost to follow-up

adriamycin,

and dacarbazine;

1

27

Parous

SN

2

36

Parous

ESM

3

36

Nulliparous

ESM

4

37

Nulliparous

ESM

5

72

Parous

ESM

6

33

?

ss

7

50

Parous

8 9

68 67

10

80

Abbreviations: CYVADIC, cyclophosphamide, disease. * Metastatic disease present at diagnosis.

vincristine,

The histology of the tumor from patient no. 6 differed from the other SS cases (Fig 4). Although the tumor was deeply invasive with an infiltrative growth pattern and had 37 mitoses/lO hpf in its most active area, there was no nuclear pleomorphism. The cells had the appearance of normal proliferative endometrial stromal cells. Necrosis, tumor giant cells, and atypical mitotic figures were absent. DNA content analysis of these specimens was undertaken further to examine the relationship between histology and DNA ploidy as well as histology and proliferative activity. The results of this analysis are summarized in Table 3. In the SN and the three analyzable cases of ESM (Fig 5B), DNA aneuploidy was not detected. In contrast, DNA aneuploidy was observed in four of five SS cases (Fig 5D). The only SS case not demonstrating DNA aneuploidy was that of patient no. 6 (Fig 5C). Analysis of proliferative activity revealed 3.9% * 3.9% and 19.7% + 11.0% S phase cells (mean * SD) TABLE2.

Patient

No.

: 3 4 5 ; 8 9 10 Symbols:

Follow-Up

Recurrence

Diagnosis

Mitosis per 10 hpf

SN ESM ESM ESM ESM ss ss ss ss ss

1 1 1 4.5 8 37 49 73 60 19

+ , present;

therapy

NED, no evidence

of disease;

DOD, dead

of

in the three analyzable cases of ESM and the five cases of SS, respectively. These results suggest a lower proliferative rate in ESM v SS. Using 10% S phase as a cutoff for dividing the cases into low proliferative activity (< 10% S phase) and high proliferative activity (a 10% S phase), all ESM cases demonstrated low proliferative activity, while four of five SS cases demonstrated high activity. The only SS case not demonstrating high proliferative activity was patient no. 6. DISCUSSION For the most part, neoplasms of endometrial stroma are readily separable morphologically as well as clinically. The SN has a well-defined, pushing border and shows no vascular invasion. In their study of 53 endometrial stromal neoplasms, Norris and Taylor5 found that the average size of an SN was 4.4

Histologic Findings Atypical Mitotic Figures

Nuclear Pleomorphism

Prominent Vascularity

_ -

_ _ _ -

+ -

+ + + +

+ + + +

+ + + +

- , absent.

234

_ + _

Necrosis + _ _ + -

Tumor Giant Cells _ + + -

NEOPLASMS OF ENDOMETRiAl STROMA (August et al)

FIGURE 1. Patient no. 3. ESM, composed of uniform, spindleshaped cells with a spiral arteriole vascular pattern, (H&E; magnification x 230.1

FIGURE 3. Patient no. 7. SS, displaying marked nuclear pleomorphism, tumor giant cells, and atypical mitotic figures. (H&E; magnification x 230.)

cm (as opposed to 5.8 cm for ESM and 7.4 cm for SS). For all intents and purposes, the SN is a benign lesion, but any individual microscopic field may look the same as a field of ESM. The morphologic difference between these two lesions is the growth pattern. ESM, as opposed to SN, has an infiltrative growth pattern and may show vascular infiltration. However, both are composed of cells bearing a striking resemblance (by electron microscopy as well as light microscopy6-8) to midproliferative endometrial stroma1 cells, and both show a vascular pattern similar to the endometrial spiral arterioles. The time-honored tradition of using the mitotic count is of no value in separating SN from ESM. Generally, both have 0 to 3 mitoses/lO hpf, although Norris and Taylor showed that the range of mitotes in their 18 SN cases was 1 to 15 mitoses/lO hpf.5 Thus, the features of circumscription and lack of vascular invasion seem to be the most useful in separating SN from ESM. The next problem is separating the ESM from the SS. In terms of clinical outcome, the two groups separate fairly well. In general, lesions designated as ESM have a low but definite risk of recurrence or metastasis (as great as 50% in some series’), and these events may take place as many as 28 years after

hysterectomy.g However, because these recurrences or metastases may take place late, long-term followup is necessary for these patients. The 5-year survival rate for ESM patients was 100% in the series of Norris and Taylor. In contrast, tumors they classified as SS had only a 55% 5-year survival rate.5 These tumors are much more likely to recur and/or metastasize with a rapidly progressive course to death. Several features have been used to separate these

FIGURE 2. Patient no. 5. Infiltrative growth pattern of ESM. Neoplastic cells are seen permeating between myometrial fibers. (H&E; magnification x 90.1

235

FIGURE 4. Patient no. 6. Designated SS because of the high mltotic rate, as deplcted here. Note the uniformity of the cells and the lack of necrosis, atypical mitoses, and nuclear pleomorphism. [H&E. Magnifications: top, x 90; bottom, x 230.1

HUMAN PATHOLOGY

TABLE 3. Patient No.

Flow Cytometric Data Proliferative Activity

Diagnosis

Aneuploidy

SN ESM ESM ESM ESM ss ss ss ss ss

Not present

: 3 4 5 6 7 8 9 10

Volume 20, No. 3 (March 1989)

(% S Phase)

2.8% s C4% 1.3% 2.0% 7.1% 34.6% 21.4% 24.4% 11%

Not present Not present Not present Not present Present Present Present Present

-=C10% s

> 10% s

tumors in various studies. Grossly, ESM is more likely to be a plaque-like proliferation or a thickening of the uterine wall, and SS is more likely to be a polypoid growth. However, these gross features can be variable. Hart and Yoonessi’O suggested that intravascular growth is more commonly a feature of ESM than SS, whereas Norris and Taylor5 found it in 57% of their SS cases as compared with 30% of their ESM cases. Mitotic counts have probably been the most universally accepted means of differentiating between the two neoplastic conditions. The cutoff, as established by Norris and Taylor, is 10 mitoses/lO hpf. Using the mitotic rate as their primary criterion, Norris and Taylor were able to readily identify two groups of patients with very different clinical outcomes5 Evans, l* however, felt that these tumors could be better separated without relying primarily on the mitotic counts. He thus designated two tumors: endometrial stromal sarcoma (ESS), and poorly differentiated endometrial sarcoma (PDES). His main criterion for separating the tumors was the histologic pattern: ESS is a tumor consisting of small- to medium-sized, uniform cells with a prominent “spiral

_

Lol 100

200

RED

0

:!I; 100

700

+

FLUORESCENCE

(DNA CONTENT)

FIGURE 5. DNA histograms showing [A and B) diploid DNA content _~ and a low S phase fraction of normal endometrium and ESM, respectively (patient no. 51, (C) diploid DNA content and low S phase fraction of the most mitotically active area of patient no. 6, and [D) DNA aneuploidy and high S phase fraction of SS [patient no. 8).

236

arteriole” vascular pattern, whereas PDES is a tumor consisting of large, hyperchromatic cells without a distinctive vascular pattern. Although Evans noted that most cases of ESS had a low mitotic rate and most cases of PDES had a high mitotic rate, he set no mitotic count cutoff as a criterion and noted two cases of ESS with more than 20 mitoses/lO hpf. The ESS and PDES cases showed similar clinical behavior to ESM and SS cases, respectively. Although this study is limited by the small number of patients and, in some instances, by a lack of long-term follow-up, it did demonstrate that the mitotic count is not always indicative of increased cell proliferation, and it has identified a peculiar variant of endometrial stromal tumor. For patient no. 6, the count of 37 mitoses/lO hpf was not a reliable measure of cell growth, since analysis of the cell cycle kinetics from the DNA histogram revealed that only 7.1% of the cells were in S phase. Not only does this support Evans’ view that the magnitude of the mitotic count may not be a sufficient criterion for the classification of proliferative and neoplastic stromal lesions of the endometrium, but it also raises some interesting questions about the interpretation of mitotic counts. Patient no. 6 was diagnosed and treated as an SS case based on an apparent high mitotic rate and response to therapy was good. She has been free of tumor for 3 years, whereas all the other SS patients studied had rapidly progressive adverse courses and three of them have died. Other features of patient no. 6 are also atypical for SS: she was young, did not present with extrauterine tumor, and the component cells in the lesion did not have the cytologic features of malignancy. This is in sharp contrast to the patients who had fulminant courses, and whose tumors, in addition to a high mitotic rate, showed nuclear polymorphism and many atypical mitotic figures. Thus, our findings support the classification approach of Evans. I1 Using his criterion of a tumor composed of uniform cells resembling those of normal proliferative endometrial stroma (regardless of mitotic count), patient no. 6 would be classified as an ESS case. As noted above, the ESS of Evans’ classification has similar biologic behavior to the ESM of the classification of Norris and Taylor. The biologic behavior of the tumor of patient no. 6 is more compatible with ESM than with SS. Cytologic features, as noted by Evans, may be of greater value in predicting response to therapy than the mitotic count alone. Since both of these parameters are based on a subjective evaluation, quantitative measurements by flow cytometry of the rate of cell proliferation would be a useful objective adjunct. Careful studies of the mitotic event in cells in culture by time-lapse photomicrography have established that an increased mitotic index may be due to an increased rate of cell proliferation, a prolonged duration of the mitotic phase, or a combination of both. In tumor cellsi and, more rarely, in normal cellsi a lengthened mitotic phase has been documented. This would, of course, lead to the presence of increased mitoses in tissues which would be erro-

NEOPIASMS OF ENDOMEKW

STROMA (August et al]

REFERENCES

neously interpreted as indicative of augmented cell proliferation. In patient no. 6, we may be dealing with a diploid SS complicated by an abnormally extended cell cycle. The fact that SS is a relatively slow-growing tumor may serve to explain the apparent response to therapy and its non-aggressive biological behavior, which sets it in sharp contrast to the other SS lesions in this study. On the other hand, it may not be an SS lesion at all, but rather an atypical variant of ESM. The mitotic count, the classical assessment of proliferative activity, is limited by the tedious nature of the assay, as well as the fact that the determination is based exclusively on the analysis of a comparatively small number of cells comprising a rare cell subpopulation. To our knowledge, the only other study using a more sophisticated approach than counting was of that by Goldfarb et aLi in which the technique Feulgen DNA microspectrophotometry was used to measure DNA content in two cases of ESM. Although this study was able to establish that the cases lacked aneuploid cell populations, only 100 cells were examined in each case. In contrast, DNA flow cytometry allows for objective measurements based on each of thousands of cells in a matter of minutes. In our study, we were able to measure DNA content in as many as 25,000 nuclei per tumor. Such capabilities suggest a more statistically precise assessment of proliferative activity using DNA content FCM, which would thereby serve as a useful adjunct to the histoThe full capacity of this pathologic classification. technology in this context, however, will only be realized when DNA measurements are coupled with the quantitation of other features, such as protein products of oncogenes, l5 proliferation-associated proteins,2 and nucleolar proteins. l6

1. Piver MS, Rutledge FN, Copeland L, et al: Uterine endolymphatic stromal myosis: A collaborative study. Obstet Gynecol 64:173-178, 1984 2. Bauer KD, Clevenger CV, Endow RK, et al: Simultaneous nuclear antigen and DNA content quantitation using paraffinembedded colonic tissue and multiparameter flow cytometry. Cancer Res 146:2428-2434, 1986 3. Salzman CC, Hiebert RD, Crowell JM: Data acquisition and display for a high-speed cell sorter. Comput Biomed Res 11:77-88, 1978 4. Baisch H, Gohde W, Linden W: Analysis of PCP-data to determine the fraction of cells in various phases of the cell cycle. Radiat Environ Biophys 12:31-39, 1975 5. Norris HJ, Taylor HB: Mesenchymal tumors of the uterus: I. A clinical and pathological study of 53 endometrial stromal tumors. Cancer 19:755-766, 1966 6. Akhtar M, Kim PY, Young I: Ultrastructure of endometrial stromal sarcoma. Cancer 35:406-412, 1975 7. Becker W, Stegner HE: A light and electron microscopic study of endometrial sarcomas of the uterus. Virchows Arch [A] 368:141-156, 1975 8. Mazur MT, Askin FB: Endolymphatic stromal myosis: Unique presentation and ultrastructural study. Cancer 42:26612667, 1978 9. Hunter WC: Uterine stromal endometriosis (stromatosis): New examples, variants, follow-up reportsComparison with personally observed instance of hemangiopericytoma. Am J Obstet Gynecol 83:1564-1573, 1962 10. Hart WR, Yoonessi M: Endometrial stromatosis of the uterus. Obstet Gynecol 49:393-403, 1977 11. Evans HL: Endometrial stromal sarcoma and poorly differentiated endometrial sarcoma. Cancer 60:2170-2182, 1982 12. Moorhead PS, Hsu TC: Cytologic studies of Hela, a strain of human cervical carcinoma. III. Durations and characteristics of the mitotic phases. JNCI 16:1047-1066, 1956 13. Fell HB: Mitosis in the mouse: A study of living and fixed cells in tissue cultures. Q J Micro Sci 90:355-380, 1949 14. Goldfarb S, Richart RM, Okagaki T: Nuclear DNA content in endolymphatic stromal myosis. Am J Obstet Gynecol 106:524-529, 1970 15. Watson JV: Oncogenes, cancer, and analytical cytology. Cytometry 7:400-410, 1986 16. Anastasi I, Bauer KD, Variakoiis D: DNA aneuploidy in Hodgkin’s disease! A multiparameter flow-cytometric analysis with cytologic correlation. Am J Path01 128:573-582, 1987

Acknowledgment. The authors are grateful to Leei Yang for technical assistance and to Dr Dante Scarpelli for critical evaluation of the manuscript.

237