- Email: [email protected]
Combined morphologic and cytochemical grading of serous ovarian tumors
Combined morphologic and cytochemical grading of serous ovarian tumors K. Erhardt, M.D., G. Auer, M.D., E. Bjorkholm, M.D., G. Forsslund, B.M., B. Moberger, M.D., C. Silfverswird, M.D., G. Wicksell, M.T., and A. Zetterberg, M.D. Stockholm, Sweden The potentiality of DNA analysis to complement morphologic evaluation in classifying serous ovarian tumors as adenoma, borderline malignancy, or invasive adenocarcinoma was investigated in a series of 54 tumors. The DNA analyses were performed on histologic tumor sections. The primary diagnoses were borderline tumor in 24 cases and invasive adenocarcinoma in 30 (World Health Organization classification). When the specimens were reviewed, 17 of the 54 tumors were reclassified, after which the series consisted of 9 adenomas, 24 borderline tumors, and 21 invasive adenocarcinomas. Rising histologic malignancy grade was associated with increasing numbers of cells showing high DNA content. The DNA levels in the adenomas thus were within the diploid range of a normal cell population. They were somewhat higher in the borderline tumors and were highest in the invasive adenocarcinomas. Though no clear-cut intergroup demarcation was discernible, there was a subgroup of adenocarcinomas with greatly elevated DNA levels, indicating high biologic malignancy. The observations suggested that DNA analyses can complement histologic malignancy grad1ng and can be useful for the recognition of highly malignant tumors amoung invasive adenocarcinomas. (AM J OBSTET GYNECOL 1985;151:356-61.)
Key words: Serous ovarian tumors, morphology, DNA
Despite modern methods of treatment for ovarian cancer, the mortality from this disease remains high. The prognosis may differ widely from case to case and mainly depends on the histologic tumor pattern. Thus tumors fulfilling the World Health Organization criteria for borderline malignancy 1 carry an appreciably better prognosis than do invasive adenocarcinomas. 2 Some studies have demonstrated 91% to 93% survival in the borderline cases, but only 34% in invasive adenocarcinoma. 3 • 4 The histologic diagnosis therefore is important for judging the malignancy potential. According to recommendations from the International Federation of Gynecology and Obstetrics5 and the World Health Organization, 1 the histologic appearance of the primary ovarian tumor should provide the basis for classification and malignancy grading. Though the importance of universally similar, or preferably identical, criteria for histologic tumor diagnosis is dear, earlier studies of various tumor types"-8 have shown that different pathologists may disagree in their interpretation of the same sections. A more objective method for correct malignancy grading of tumors is therefore highly desirable. From the Department of Tumor Pathology, Karolinska Institute and Hospital, and the Department of Gynecologic Oncology, Radiumhemmet, Karolinska Hospital. Received for publication February 17, 1984; accepted August 29, 1984.
Reprint requests: K. Erhardt, Department of Tumor Pathology, Karolinska Institute and Hospital, S-104 OJ Stockholm 60, Sweden.
356
As early as 1966, Taylor9 demonstrated that the levels of DNA in adenocarcinoma of the ovary were related o the histologic grade of malignancy. Increasing aneuploidy (nondiploidy) thus was seen to accompany increasing degree of malignancy. A recent study has confirmed these findings. 10 In the present study, DNA levels in serous ovarian tumors were compared with histologic assessment of the same tumors. The aim was to evaluate the usefulness of DNA analysis as an objective and accordingly reproducible technique in assisting World Health Organization grouping as adenoma, borderline malignancy, or invasive carcinoma. The DNA analyses were made on sectioned tumor specimens, which permitted direct comparison between the histologic evaluation and the DNA measurements in the tumor cells. Material and methods Patients. The patients had been referred to the Radiumhemmet for postoperative treatment of serous malignant or borderline ovarian tumors during the period 1975 to 1980. The International Federation of Gynecology and Obstetrics clinical stage of the tumors was lA-C or IIA. 5 The primary surgical treatment was performed at various hospitals in the Stockholm region, where the tumors had been histologically classified and graded. The histologic specimens were reviewed at the Karolinska Hospital by an experienced pathologist (C. S.), who was unaware of the earlier diagnosis and of the
Grading of serous ovarian tumors 357
Volume 151 Number 3
patients' clinical fate. This review excluded from the series tumors which did not fulfill the World Health Organization criteria for serous ovarian tumor. The 54 remaining tumors were thereafter grouped according to World Health Organization1 and International Federation of Gynecology and Obstetrics5 classification as serous adenoma (lA), serous borderline tumor (IB) or invasive adenocarcinoma (IC). The adenocarcinomas were then graded by malignancy, with guidance from the degree of cellular pleomorphism, nuclear atypia, mitosis frequency, and solid growth, as highly, moderately, or poorly differentiated tumors 11 (Table 1). During the follow-up period (mean, 5 years; range, 2.5 to 8 years), eight patients died because of the ovarian tumor. All deaths occurred within 2.5 years of the primary diagnosis. Histological material. Paraffin-embedded tissue resected at the primary operation was used in the analyses. A tumor-representative block was selected in each case. Two consecutive sections of 4 tJ.m thickness were cut from each block. After deparaffining, one section was stained with hematoxylin-eosin solution. The other was stained according to the Feulgen technique (acid hydrolysis with 5N hydrochloric acid, 22° C, 60 minutes) after refixation in 10% neutral buffered formalin.12 With guidance from the hematoxylin-eosin-stained sections, representative areas in the Feulgen preparations were selected for photography. Usually 10 fields were photographed in each case. The fields were chosen from different areas where the tumor tissue was well preserved histologically and where the architecture of the tumor tissue allowed DNA analysis. The general goal was to select representative fields throughout the entire tissue section. The selection was done by an experienced pathologist (K. E.), without foreknowledge of the previous diagnosis in the case or of the clinical course. A Leitz photomicroscope was used, with 40 X /1.0 oil objective (refractory index, 1.518) in monochromatic light and wavelength of 546 nm. The sensitivity of the film (Kodak Technical Pan 2415) was 29 DIN and the developing time was 4 minutes at 22° C (Kodak D19). DNA measurements. On the developed films the DNA content of individual cells was measured. The photographic cytophotometric method used was a modification of that described by Adams 13 in 1968. The method is based on light transmission from the Feulgen-stained nuclei, the blackness of which is taken as a measure of the DNA content. Comparative studies between this method and conventional rapid-scanning microspectrophotometry 14 showed close conformity. In each histologic slide 100 tumor cells were analyzed for DNA content. Within the photographed tumor areas individual tumor cells were selected by means of
Table I. Grading of ovarian tumors in study Histologic classification
Cells >2.5c
I
(%)
Case No.
Age ,(yr)
Original
1 2 3 4 5 6 7 8 9 10
74 74 72 69 69 63 56 58 51 75 73 60 69 69 62 60 59 52 51 53 53 50 51 49 46 46 43 42 40 38 34 37 32 41 40 40 40 74 54 62 58 75 63 55 55 60 70 67 55 64 60 55 54 50
IB IB IB IB IB IB IB IB IC IB IB IB IC IB IB IB IB IC IB IB IC IB IC IB IB IC IB IB IB IC IC IB IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC
11
12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
Revised*
lA lA lA lA lA lA lA lA lA IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC
(h) (h) (h) (h) (m) (m) (m) (m) (m) (m) (l) (l)
(I) (I) (I) (l)
(I) (l)
(I) (I) (I)
25 44 22 32 42 20 25 36 43 65 42 50 75 67 66 47 56 56 55 17 37 47 72 39 48 21 60 29 59 44 62 29 37 56 26 53 57 28 81 63 94 99 90 93 77 98 69 96 96 99 92 82 97 95
*Differentiation represented by h, high; m, moderate; I, low.
morphologic criteria. All tumor cells that could be analyzed (e.g., no overlapping of the Feulgen-stained nuclei), were measured at random. To define the normal diploid (2c) value, approximately 50 control cells-fibroblasts and/or endothelial cells-were similarly analyzed in normal tissue adjoining the tumor. To permit
358 Erhardt et al.
February 1, 1985 Am J Obstet Gynecol
comparisons between the DNA levels in different tumors, DNA content was expressed in c units, 2c being defined from the median value (P50) found in the control cells. An upper limit of 2.5c was set for diploid cell populations, so as to distinguish them from nondiploid types. This limit was empirical, based on the observation that it is not exceeded by most (>90%) of the control cells. The percentage of tumor cells with DNA >2.5c was taken as a measure of nondiploidy. Fig. I illustrates an example of how the DNA values from a tumor were recalculated according to the internal standard. The upper histogram shows the distribution pattern of the control cells, the intermediate that of the uncorrected tumor cell population, and the lower that of the profile after correction.
The DNA patterns are exemplified in Fig. 2. The tumors which at review of the histologic specimens were classified as adenoma (lA) showed DNA values around diploid level. In the borderline tumors (IB) the values were slightly higher but still within the DNA limits in normal proliferating cells. Among the invasive carcinomas (I C) there were some with DNA patterns resembling those of borderline tumors, while others were composed of cells with very high and scattered DNA contents which significantly exceeded the values in normal proliferating cells. At the histologic review, most of the invasive adenocarcinomas with DNA pattern resembling borderline malignancy were graded as highly differentiated, whereas the invasive carcinomas with DNA content indicating severe aneuploidy were graded as poorly or moderately differentiated. For each of the 54 tumors in the series, the proportion of cells with DNA content exeeding 2.5c was calculated as a measure of nondiploidy (Table 1). The association between the primary diagnosis and nondiploidy is illustrated in Fig. 3 and that between the review diagnosis and nondiploidy in Fig. 4. The distribution of malignancy grades-high, moderate, or poor differentiation-within group IC is also shown in Fig. 4. Tumor populations with comparable degrees of nondiploidy were found in the groups primarily classified as IB or IC (Fig. 3). The latter group, however, also contained tumors with very high proportions (>80%) of nondiploid cells. A similar relationship between tumor DNA content and histologic evaluation was found when the classification was revised (Fig. 4), i.e., lowest DNA levels in adenomas (lA) and highest in adenocarcinomas (I C). Among the invasive adenocarcinomas, those of moderate or poor differentiation showed the highest DNA values, whereas the DNA content in the highly differentiated tumors did not essentially differ from the values in the borderline group (IB). In all of the eight patients who died of ovarian malignancy, both the primary and the reviewed morphologic diagnosis were IC. At review of the specimens the differentiation grade was found to be moderate or poor. In all eight tumors there was severe nondiploidy, with more than 80% nondiploid cells (Figs. 3 and 4).
Results The 54 cases remaining in the series after histologic review of the original specimens are surveyed in Table I. Initially 24 of the tumors were classified as of borderline malignancy (IB) and 30 as invasive adenocarcinoma (IC). The revised classification was adenoma (lA) in 9 cases, borderline tumor (IB) in 24 and invasive adenocarcinma (IC) in 21 cases. Of the nine adenomas, eight had previously been classified as borderline tumors and one as adenocarcinoma. Eight tumors initially judged to be adenocarcinomas were reclassified as borderline malignancies.
Comment In recent years many new treatment modalities have been introduced for malignant ovarian tumors, particularly as a result of progress in radiotherapy and chemotherapy. Several prospective randomized studies have been undertaken to evaluate such regimens. As in earlier studies, the prognosis has been shown to be determined by a variety of factors, including type of tumor, clinical stage and the histologic grade of malignancy. Recently a multivariate analysis" demonstrated the histologic malignancy grade to be well correlated to survival within the clinical stages lA-C and IIA.
~
60 40 20
Control cells
a:
w
m ::::e
::::)
z
...J ...J
60 40 20
Tumor cells
w
u
60 40 20
Tumor cells
2c DNA
4c
6c
8c
(rei. units)
Fig. 1. Example of DNA distribution patterns of normal fibroblasts and endothelial cells (control cells) and of tumor cells before recalculation according to the median value of the control cells (intermediate histogram) and after recalculation (lower histogram).
Grading of serous ovarian tumors 359
Volume 151 Number 3
1b
1a
~
40 20
a: w
m :::t ::J
z
40 20
_,_,
w
u
40 20
1c
~~
~~ ~~~ 2c
4c
2c
6c
DNA
4c
6c
m I
at 9
2c
4c
6c
Be
( rei. units >
Fig. 2. Examples of DNA distribution patterns in serous ovarian tumors with morphologic diagnosis of adenoma (!A), borderline tumor (!B) or invasive adenocarcinoma (!C). The adenocarcinomas are subdivided according to tumor differentiation: h, high; m, moderate: l, low.
For comparability of treatment results, both within single randomized series and between series at different treatment centers, homogeneity of histologic classification and of malignancy grading are decisive. For the individual patient, too, such malignancy grading is of great importance. The high risk of side effects from currently employed chemotherapy underlines the need for standardized malignancy grading as a basis for choice of postoperative management. Great efforts have therefore been made to define reproducible histologic criteria and to disseminate them among pathologists, e.g., by the World Health Organization. 1 Despite all attempts to achieve objectivity, however, histologic diagnosis based on light microscopy remains a subjective assessment that is influenced by the examiner's personal experience in the field. Although this influence is of minor significance in regard to tumors of typical appearance, diagnostic routines frequently must deal also with tumors that cannot readily be assigned to the groups described in manuals. Measurement of DNA content, with standardized procedure, is an objective and hence reproducible method. In a recent study of serous ovarian tumors there was a close relationship between the nuclear DNA content and the malignancy potential of the tumors. The biologically highly malignant tumors were identifiable with surprisingly good accuracy. The question posed for the present study was whether DNA analysis could also provide a basis for grouping of serous ovarian tumors as adenoma, borderline malignancy, or adenocarcinoma. We therefore compared the DNA levels with the histologic evaluation. To make the comparison as direct as possible, the
DNA analyses were performed on sections from the tissue specimens that had been used for the histologic examinations. The study comprised 54 tumors primarily diagnosed as serous borderline malignancy or serous invasive adenocarcinoma. Review of the specimens led to revision of the diagnosis in 17 (31%) of the series, a figure which agrees well with previous reports. 6· 8 The distribution pattern of DNA (Fig. 2) showed that in the tumors reclassified as adenoma (lA) the values mostly were within the limit (<2.5c) for a normal diploid cell population. Between adenoma and borderline tumor, however, there was no clear demarcation in regard to DNA pattern. The borderline tumors thus included some that were comparable with adenoma in this respect and some with fairly abundant cells showing DNA levels of >2.5c. These latter nevertheless did not exceed the tetraploid (4c) values that can be seen in normal prolife-rating cell populations. Because of the imprecise transition in DNA values from adenoma to borderline malignancy, the method does not permit definite classification of a tumor as benign (lA) or of borderline type (IB), implying risk of subsequent metastasis and death from the disease. The same demarcation is also recognized as being difficult in histologic evaluation. Among the invasive adenocarcinomas there were some with DNA distribution curves indistinguishable from those of borderline tumors, but there also was a group with strongly aneuploid cell populations, containing cells with very high and scattered amounts of DNA. At the histologic review all of these aneuploid tumors were graded as adenocarcinoma of moderate
360 Erhardt et al.
February 1, 1985 Am J Obstet Gynecol
1a
1c
1b
1a
1b
1c h
100 100 -
u
•o~
It)
u
N'
"
1-
z 1z
80
-------------- -------------- ____!l_!_____ _
oo
0
60
0
0
z
------------- -------------- ----
0 0
0 0
00 0 0
z
0 0 0 00 0 0 00
0
(,)
60
20
00
0 0
0
w
z
ocP
0
40 -
::t:
o oa 0
~
0
en ...J
00 0 0
(,)
...J
~
(,)
20 -
w
0
~--------~----------.__________
Fig. 3. Percentage of cells with DNA content >2.5c of normal control cells in tumors orginally classified as borderline malignancy (IB) or invasive adenocarcinoma (/C). Open circles, survivors; solid circles, deaths because of tumor.
or poor differentiation, and all of the deaths in the follow-up period occurred in this group. The tumors judged at review to be highly differentiated invasive adenocarcinoma showed DNA distribution curves of the type found in borderline tumors. The clinical behavior of these adenocarcinomas likewise resembled that of the borderline tumors in that no patient died because of the tumor. Although the number of cases was too small to permit firm conclusions, the results indicate the possibility that serous invasive cancer showing DNA content within the limits for normal tissues can be managed in the same way as borderline tumors. Compared with histologic assessment, analysis of DNA content by means of standardized procedure is objective and accordingly a more reproducible method. DNA analysis, however, does not always permit differentiation between normal tissue and tumor tissue. Many tumors are composed of entirely diploid cell populations, and the DNA curves therefore are identical with those of normal tissue. The finding of a diploid cell population in a tumor indicates only a slow growth rate and by no means precludes metastasis. 15 Conventional histologic examination thus remains decisive for the distinction between adenoma and borderline tumor. The present study suggests that, despite the subjective factor in histologic malignancy grading, its reliability in this distinction is relatively high. The histologic demarcation between tumors of borderline malignancy and true invasive adenocarcinoma is likewise subjectively influenced. Many tumors his-
0
-'--0 0
0
1-
0
_..Q
0
<
oo
40
...J
80
0
0
en ...J
"
11-
ooo
< z
N
w
0
o oo
0
(,)
•
~
w
m
0
cP
0
0
0 0 00
0
0
0
0
0 0
~
0 ~--------~--------~~--~--
Fig. 4. Percentage of cells with DNA content >2.5c of normal control cells in tumors reclassified as adenoma (lA), borderline malignancy (IB), or invasive adenocarcinoma (IC). The adenocarcinomas are subdivided according to tumor differentiation: h, high; m, moderate; l, low.
tologically classified as invasive cancer are of low malignancy, which is reflected in patient survival times comparable with those for borderline tumor. The main importance of DNA analysis in ovarian tumors is its high specificity in distinguishing between high and low degrees of malignancy. DNA analysis, as an objective technique, thus constitutes a useful diagnostic complement, partly as an aid in histologically doubtful cases, but above all for identification of the highly malignant tumor forms. A combination of histologic evaluation and DNA analysis therefore appears to offer prospects of heightened diagnostic accuracy in serous ovarian tumors. REFERENCES 1. Serov SF, Scully RE, Sobin LH. Histological typing of ovarian tumors: international histological typing of tumors, vol 9. Geneva: World Health Organization, 1973. 2. Santesson L, Kottmeier HL. General classification of ovarian tumors. New York: Springer-Verlag, 1968. (Gentil F, Junqueira AC, eds. Ovarian cancer, monograph series, vol II).
3. Bjorkholm E, Pettersson F, Einhorn N, Krebs I, Nilsson B, Tjernberg B. Long-term follow-up and prognostic factors in ovarian carcinoma. Acta Radio! (Oncol) 1982; 21(6):413-9. 4. Nikrui N. Survey of clinical behavior of patients with borderline epithelial tumors of the ovary. Gynecologic Oncology 1981;12:107-19. 5. Kottmeier HL, ed. Annual report on the results of treatment in gynecological cancer, vol 18. F.I.G.O. Radiumhemmet, Stockholm: Editional Office, 1982. 6. Baak JPA, Lindeman J, Overdiep SH, Langley FA. Dis-
Grading of serous ovarian tumors
Volume 151 Number 3
7. 8. 9. 10. 11.
agreement of histopathological diagnosis of different pathologists in ovarian tumors-with some theoretical considerations. Eur 1 Obstet Gynecol Reprod Bio 1982; 13:51-5. Woodrugg RD. Reviewing histologic diagnosis of lymphoma. Arch Pathol Lab Med 1981;105:573-6. Wallgren A, Silfverswiird C, Hultborn A. Carcinoma of the breast in women under 30 years of age. Cancer 1977; 40:916-23. Taylor HC. Studies in the clinical and biological evolution of adenocarcinoma of the ovary. Br 1 Obstet Gynaecol 1966;827-42. Erhardt K, Auer G, Bjorkholm E, eta!. Significance of nuclear DNA content in serous ovarian tumors. Cancer Res 1984;44:2198-202. Sorbe B, Frankendal B, Veress B. Importance of histologic grading in the prognosis of epithelial ovarian carcinoma. Obstet Gynecoll982;59(5):576-82.
12. Gaub 1. Auer G, Zetterberg A. Quantitative cytochemical aspects of a combined Feulgen-naphthol yellowS staining procedure for the simultaneous determination of nuclear and cytoplasmic proteins and DNA in mammalian cells. Exp Cell Res 1975;92:323-32. 13. Adams LR. Photographic cytophotometric method which avoids distributional error. Acta Cytol 1968;12(1):3-8. 14. Caspersson T, Kudynowski J. Cytochemical instrumentation for cytopathological work. Int Rev Exp Patholl980; 21:1-54. 15. Sigurdson IC, Aim P, Gullberg B. Prognostic factors in malignant epithelial ovarian tumors. Gynecol On col 1983; 15:370-380. 16. Auer G, Arrhenius E, Granberg P-0, Fox C. Comparison of DNA distributions in primary human breast cancers and their metastases. Eur 1 Cancer 1980;16:273-8.
The effect of a prostaglandin synthetase inhibitor on the hormonal profile and the endometrium in women B. M. Landgren, M.D., V. Lundstrom, M.D., P. Eneroth, M.D., and E. Johannisson, M.D. Stockholm, Sweden, and Geneva, Switzerland A prostaglandin synthetase inhibitor, naproxen, was given continuously throughout the menstrual cycle at a dose of 250 mg twice daily to 10 healthy fertile women (group 1) and at a dose of 1000 mg per day to eight women in the secretory phase (group 2). Blood samples were withdrawn three times a week during a control cycle and during the treatment cycle. Luteinizing hormone, follicle-stimulating hormone, prolactin, estradiol, and progesterone were analyzed. Endometrial biopsies were taken in the secretory phase of the control cycle and in the treatment cycle of group 1 and on the first day of menstruation of group 2. Naproxen treatment did not suppress ovulation in any cycle and did not affect the corpus luteum function either in group 1 or in group 2. In only one of the endometrial samples taken in the secretory phase (group 1) was the density of the lysosomes increased in the treatment cycle compared to the control cycle. The specimen taken during the early menstrual period (group 2) showed an increase in glandular epithelium and height following naproxen treatment. Furthermore, a significant increase in the number of plasmolemmal vesicles per square micrometer was observed in the capillary endothelial cells after the administration of 1000 mg of naproxen per day. This suggests that the transcellular exchange of water-soluble molecules in the endothelial cells was more active after the administration of a prostaglandin synthetase inhibitor than in the control group. In spite of the significant morphologic changes observed in the naproxen-treated material the onset of menstrual bleeding could not be prevented. The mechanism of the onset of menstruation needs to be further investigated. (AM J OBSTET GYNECOL 1985;151 :361-8.)
Key words: Endometrium, hormones, prostaglandin synthetase inhibitor Animal studies have suggested that prostaglandins are mediators of ovulation as well as luteolysis. 1• 2 A
From the Department of Obstetrics and Gynecology, Karolinska H ospital, Stockholm, and the Centre de Cytologic et de Depistage du Cancer, Geneva. This study was supported by a grant from Astra Syntex, Sodertalje, Sweden. ReceivedforpublicationApril4, 1984; accepted September 26, 1984. Reprint requests: Dr. B. M. Landgren, Department of Obstetrics and Gynecology, Karolinska Hospital, S-104 OJ Stockholm, Sweden.
similar function of prostaglandins in women has not yet been convincingly demonstrated. However, in vitro studies by Hamberger et aP·• indicate that prostaglandin F2a in the presence of norepinephrine may induce luteolysis in human corpora lutea in vitro as well as in VIVO.
Administration of prostaglandin synthetase inhibitors to a subhuman primate has been demonstrated to interfere with rupture of the follicular membrane without affecting luteinization, leading to the "blocked 361
Key words: Serous ovarian tumors, morphology, DNA
Despite modern methods of treatment for ovarian cancer, the mortality from this disease remains high. The prognosis may differ widely from case to case and mainly depends on the histologic tumor pattern. Thus tumors fulfilling the World Health Organization criteria for borderline malignancy 1 carry an appreciably better prognosis than do invasive adenocarcinomas. 2 Some studies have demonstrated 91% to 93% survival in the borderline cases, but only 34% in invasive adenocarcinoma. 3 • 4 The histologic diagnosis therefore is important for judging the malignancy potential. According to recommendations from the International Federation of Gynecology and Obstetrics5 and the World Health Organization, 1 the histologic appearance of the primary ovarian tumor should provide the basis for classification and malignancy grading. Though the importance of universally similar, or preferably identical, criteria for histologic tumor diagnosis is dear, earlier studies of various tumor types"-8 have shown that different pathologists may disagree in their interpretation of the same sections. A more objective method for correct malignancy grading of tumors is therefore highly desirable. From the Department of Tumor Pathology, Karolinska Institute and Hospital, and the Department of Gynecologic Oncology, Radiumhemmet, Karolinska Hospital. Received for publication February 17, 1984; accepted August 29, 1984.
Reprint requests: K. Erhardt, Department of Tumor Pathology, Karolinska Institute and Hospital, S-104 OJ Stockholm 60, Sweden.
356
As early as 1966, Taylor9 demonstrated that the levels of DNA in adenocarcinoma of the ovary were related o the histologic grade of malignancy. Increasing aneuploidy (nondiploidy) thus was seen to accompany increasing degree of malignancy. A recent study has confirmed these findings. 10 In the present study, DNA levels in serous ovarian tumors were compared with histologic assessment of the same tumors. The aim was to evaluate the usefulness of DNA analysis as an objective and accordingly reproducible technique in assisting World Health Organization grouping as adenoma, borderline malignancy, or invasive carcinoma. The DNA analyses were made on sectioned tumor specimens, which permitted direct comparison between the histologic evaluation and the DNA measurements in the tumor cells. Material and methods Patients. The patients had been referred to the Radiumhemmet for postoperative treatment of serous malignant or borderline ovarian tumors during the period 1975 to 1980. The International Federation of Gynecology and Obstetrics clinical stage of the tumors was lA-C or IIA. 5 The primary surgical treatment was performed at various hospitals in the Stockholm region, where the tumors had been histologically classified and graded. The histologic specimens were reviewed at the Karolinska Hospital by an experienced pathologist (C. S.), who was unaware of the earlier diagnosis and of the
Grading of serous ovarian tumors 357
Volume 151 Number 3
patients' clinical fate. This review excluded from the series tumors which did not fulfill the World Health Organization criteria for serous ovarian tumor. The 54 remaining tumors were thereafter grouped according to World Health Organization1 and International Federation of Gynecology and Obstetrics5 classification as serous adenoma (lA), serous borderline tumor (IB) or invasive adenocarcinoma (IC). The adenocarcinomas were then graded by malignancy, with guidance from the degree of cellular pleomorphism, nuclear atypia, mitosis frequency, and solid growth, as highly, moderately, or poorly differentiated tumors 11 (Table 1). During the follow-up period (mean, 5 years; range, 2.5 to 8 years), eight patients died because of the ovarian tumor. All deaths occurred within 2.5 years of the primary diagnosis. Histological material. Paraffin-embedded tissue resected at the primary operation was used in the analyses. A tumor-representative block was selected in each case. Two consecutive sections of 4 tJ.m thickness were cut from each block. After deparaffining, one section was stained with hematoxylin-eosin solution. The other was stained according to the Feulgen technique (acid hydrolysis with 5N hydrochloric acid, 22° C, 60 minutes) after refixation in 10% neutral buffered formalin.12 With guidance from the hematoxylin-eosin-stained sections, representative areas in the Feulgen preparations were selected for photography. Usually 10 fields were photographed in each case. The fields were chosen from different areas where the tumor tissue was well preserved histologically and where the architecture of the tumor tissue allowed DNA analysis. The general goal was to select representative fields throughout the entire tissue section. The selection was done by an experienced pathologist (K. E.), without foreknowledge of the previous diagnosis in the case or of the clinical course. A Leitz photomicroscope was used, with 40 X /1.0 oil objective (refractory index, 1.518) in monochromatic light and wavelength of 546 nm. The sensitivity of the film (Kodak Technical Pan 2415) was 29 DIN and the developing time was 4 minutes at 22° C (Kodak D19). DNA measurements. On the developed films the DNA content of individual cells was measured. The photographic cytophotometric method used was a modification of that described by Adams 13 in 1968. The method is based on light transmission from the Feulgen-stained nuclei, the blackness of which is taken as a measure of the DNA content. Comparative studies between this method and conventional rapid-scanning microspectrophotometry 14 showed close conformity. In each histologic slide 100 tumor cells were analyzed for DNA content. Within the photographed tumor areas individual tumor cells were selected by means of
Table I. Grading of ovarian tumors in study Histologic classification
Cells >2.5c
I
(%)
Case No.
Age ,(yr)
Original
1 2 3 4 5 6 7 8 9 10
74 74 72 69 69 63 56 58 51 75 73 60 69 69 62 60 59 52 51 53 53 50 51 49 46 46 43 42 40 38 34 37 32 41 40 40 40 74 54 62 58 75 63 55 55 60 70 67 55 64 60 55 54 50
IB IB IB IB IB IB IB IB IC IB IB IB IC IB IB IB IB IC IB IB IC IB IC IB IB IC IB IB IB IC IC IB IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC
11
12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
Revised*
lA lA lA lA lA lA lA lA lA IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IB IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC IC
(h) (h) (h) (h) (m) (m) (m) (m) (m) (m) (l) (l)
(I) (I) (I) (l)
(I) (l)
(I) (I) (I)
25 44 22 32 42 20 25 36 43 65 42 50 75 67 66 47 56 56 55 17 37 47 72 39 48 21 60 29 59 44 62 29 37 56 26 53 57 28 81 63 94 99 90 93 77 98 69 96 96 99 92 82 97 95
*Differentiation represented by h, high; m, moderate; I, low.
morphologic criteria. All tumor cells that could be analyzed (e.g., no overlapping of the Feulgen-stained nuclei), were measured at random. To define the normal diploid (2c) value, approximately 50 control cells-fibroblasts and/or endothelial cells-were similarly analyzed in normal tissue adjoining the tumor. To permit
358 Erhardt et al.
February 1, 1985 Am J Obstet Gynecol
comparisons between the DNA levels in different tumors, DNA content was expressed in c units, 2c being defined from the median value (P50) found in the control cells. An upper limit of 2.5c was set for diploid cell populations, so as to distinguish them from nondiploid types. This limit was empirical, based on the observation that it is not exceeded by most (>90%) of the control cells. The percentage of tumor cells with DNA >2.5c was taken as a measure of nondiploidy. Fig. I illustrates an example of how the DNA values from a tumor were recalculated according to the internal standard. The upper histogram shows the distribution pattern of the control cells, the intermediate that of the uncorrected tumor cell population, and the lower that of the profile after correction.
The DNA patterns are exemplified in Fig. 2. The tumors which at review of the histologic specimens were classified as adenoma (lA) showed DNA values around diploid level. In the borderline tumors (IB) the values were slightly higher but still within the DNA limits in normal proliferating cells. Among the invasive carcinomas (I C) there were some with DNA patterns resembling those of borderline tumors, while others were composed of cells with very high and scattered DNA contents which significantly exceeded the values in normal proliferating cells. At the histologic review, most of the invasive adenocarcinomas with DNA pattern resembling borderline malignancy were graded as highly differentiated, whereas the invasive carcinomas with DNA content indicating severe aneuploidy were graded as poorly or moderately differentiated. For each of the 54 tumors in the series, the proportion of cells with DNA content exeeding 2.5c was calculated as a measure of nondiploidy (Table 1). The association between the primary diagnosis and nondiploidy is illustrated in Fig. 3 and that between the review diagnosis and nondiploidy in Fig. 4. The distribution of malignancy grades-high, moderate, or poor differentiation-within group IC is also shown in Fig. 4. Tumor populations with comparable degrees of nondiploidy were found in the groups primarily classified as IB or IC (Fig. 3). The latter group, however, also contained tumors with very high proportions (>80%) of nondiploid cells. A similar relationship between tumor DNA content and histologic evaluation was found when the classification was revised (Fig. 4), i.e., lowest DNA levels in adenomas (lA) and highest in adenocarcinomas (I C). Among the invasive adenocarcinomas, those of moderate or poor differentiation showed the highest DNA values, whereas the DNA content in the highly differentiated tumors did not essentially differ from the values in the borderline group (IB). In all of the eight patients who died of ovarian malignancy, both the primary and the reviewed morphologic diagnosis were IC. At review of the specimens the differentiation grade was found to be moderate or poor. In all eight tumors there was severe nondiploidy, with more than 80% nondiploid cells (Figs. 3 and 4).
Results The 54 cases remaining in the series after histologic review of the original specimens are surveyed in Table I. Initially 24 of the tumors were classified as of borderline malignancy (IB) and 30 as invasive adenocarcinoma (IC). The revised classification was adenoma (lA) in 9 cases, borderline tumor (IB) in 24 and invasive adenocarcinma (IC) in 21 cases. Of the nine adenomas, eight had previously been classified as borderline tumors and one as adenocarcinoma. Eight tumors initially judged to be adenocarcinomas were reclassified as borderline malignancies.
Comment In recent years many new treatment modalities have been introduced for malignant ovarian tumors, particularly as a result of progress in radiotherapy and chemotherapy. Several prospective randomized studies have been undertaken to evaluate such regimens. As in earlier studies, the prognosis has been shown to be determined by a variety of factors, including type of tumor, clinical stage and the histologic grade of malignancy. Recently a multivariate analysis" demonstrated the histologic malignancy grade to be well correlated to survival within the clinical stages lA-C and IIA.
~
60 40 20
Control cells
a:
w
m ::::e
::::)
z
...J ...J
60 40 20
Tumor cells
w
u
60 40 20
Tumor cells
2c DNA
4c
6c
8c
(rei. units)
Fig. 1. Example of DNA distribution patterns of normal fibroblasts and endothelial cells (control cells) and of tumor cells before recalculation according to the median value of the control cells (intermediate histogram) and after recalculation (lower histogram).
Grading of serous ovarian tumors 359
Volume 151 Number 3
1b
1a
~
40 20
a: w
m :::t ::J
z
40 20
_,_,
w
u
40 20
1c
~~
~~ ~~~ 2c
4c
2c
6c
DNA
4c
6c
m I
at 9
2c
4c
6c
Be
( rei. units >
Fig. 2. Examples of DNA distribution patterns in serous ovarian tumors with morphologic diagnosis of adenoma (!A), borderline tumor (!B) or invasive adenocarcinoma (!C). The adenocarcinomas are subdivided according to tumor differentiation: h, high; m, moderate: l, low.
For comparability of treatment results, both within single randomized series and between series at different treatment centers, homogeneity of histologic classification and of malignancy grading are decisive. For the individual patient, too, such malignancy grading is of great importance. The high risk of side effects from currently employed chemotherapy underlines the need for standardized malignancy grading as a basis for choice of postoperative management. Great efforts have therefore been made to define reproducible histologic criteria and to disseminate them among pathologists, e.g., by the World Health Organization. 1 Despite all attempts to achieve objectivity, however, histologic diagnosis based on light microscopy remains a subjective assessment that is influenced by the examiner's personal experience in the field. Although this influence is of minor significance in regard to tumors of typical appearance, diagnostic routines frequently must deal also with tumors that cannot readily be assigned to the groups described in manuals. Measurement of DNA content, with standardized procedure, is an objective and hence reproducible method. In a recent study of serous ovarian tumors there was a close relationship between the nuclear DNA content and the malignancy potential of the tumors. The biologically highly malignant tumors were identifiable with surprisingly good accuracy. The question posed for the present study was whether DNA analysis could also provide a basis for grouping of serous ovarian tumors as adenoma, borderline malignancy, or adenocarcinoma. We therefore compared the DNA levels with the histologic evaluation. To make the comparison as direct as possible, the
DNA analyses were performed on sections from the tissue specimens that had been used for the histologic examinations. The study comprised 54 tumors primarily diagnosed as serous borderline malignancy or serous invasive adenocarcinoma. Review of the specimens led to revision of the diagnosis in 17 (31%) of the series, a figure which agrees well with previous reports. 6· 8 The distribution pattern of DNA (Fig. 2) showed that in the tumors reclassified as adenoma (lA) the values mostly were within the limit (<2.5c) for a normal diploid cell population. Between adenoma and borderline tumor, however, there was no clear demarcation in regard to DNA pattern. The borderline tumors thus included some that were comparable with adenoma in this respect and some with fairly abundant cells showing DNA levels of >2.5c. These latter nevertheless did not exceed the tetraploid (4c) values that can be seen in normal prolife-rating cell populations. Because of the imprecise transition in DNA values from adenoma to borderline malignancy, the method does not permit definite classification of a tumor as benign (lA) or of borderline type (IB), implying risk of subsequent metastasis and death from the disease. The same demarcation is also recognized as being difficult in histologic evaluation. Among the invasive adenocarcinomas there were some with DNA distribution curves indistinguishable from those of borderline tumors, but there also was a group with strongly aneuploid cell populations, containing cells with very high and scattered amounts of DNA. At the histologic review all of these aneuploid tumors were graded as adenocarcinoma of moderate
360 Erhardt et al.
February 1, 1985 Am J Obstet Gynecol
1a
1c
1b
1a
1b
1c h
100 100 -
u
•o~
It)
u
N'
"
1-
z 1z
80
-------------- -------------- ____!l_!_____ _
oo
0
60
0
0
z
------------- -------------- ----
0 0
0 0
00 0 0
z
0 0 0 00 0 0 00
0
(,)
60
20
00
0 0
0
w
z
ocP
0
40 -
::t:
o oa 0
~
0
en ...J
00 0 0
(,)
...J
~
(,)
20 -
w
0
~--------~----------.__________
Fig. 3. Percentage of cells with DNA content >2.5c of normal control cells in tumors orginally classified as borderline malignancy (IB) or invasive adenocarcinoma (/C). Open circles, survivors; solid circles, deaths because of tumor.
or poor differentiation, and all of the deaths in the follow-up period occurred in this group. The tumors judged at review to be highly differentiated invasive adenocarcinoma showed DNA distribution curves of the type found in borderline tumors. The clinical behavior of these adenocarcinomas likewise resembled that of the borderline tumors in that no patient died because of the tumor. Although the number of cases was too small to permit firm conclusions, the results indicate the possibility that serous invasive cancer showing DNA content within the limits for normal tissues can be managed in the same way as borderline tumors. Compared with histologic assessment, analysis of DNA content by means of standardized procedure is objective and accordingly a more reproducible method. DNA analysis, however, does not always permit differentiation between normal tissue and tumor tissue. Many tumors are composed of entirely diploid cell populations, and the DNA curves therefore are identical with those of normal tissue. The finding of a diploid cell population in a tumor indicates only a slow growth rate and by no means precludes metastasis. 15 Conventional histologic examination thus remains decisive for the distinction between adenoma and borderline tumor. The present study suggests that, despite the subjective factor in histologic malignancy grading, its reliability in this distinction is relatively high. The histologic demarcation between tumors of borderline malignancy and true invasive adenocarcinoma is likewise subjectively influenced. Many tumors his-
0
-'--0 0
0
1-
0
_..Q
0
<
oo
40
...J
80
0
0
en ...J
"
11-
ooo
< z
N
w
0
o oo
0
(,)
•
~
w
m
0
cP
0
0
0 0 00
0
0
0
0
0 0
~
0 ~--------~--------~~--~--
Fig. 4. Percentage of cells with DNA content >2.5c of normal control cells in tumors reclassified as adenoma (lA), borderline malignancy (IB), or invasive adenocarcinoma (IC). The adenocarcinomas are subdivided according to tumor differentiation: h, high; m, moderate; l, low.
tologically classified as invasive cancer are of low malignancy, which is reflected in patient survival times comparable with those for borderline tumor. The main importance of DNA analysis in ovarian tumors is its high specificity in distinguishing between high and low degrees of malignancy. DNA analysis, as an objective technique, thus constitutes a useful diagnostic complement, partly as an aid in histologically doubtful cases, but above all for identification of the highly malignant tumor forms. A combination of histologic evaluation and DNA analysis therefore appears to offer prospects of heightened diagnostic accuracy in serous ovarian tumors. REFERENCES 1. Serov SF, Scully RE, Sobin LH. Histological typing of ovarian tumors: international histological typing of tumors, vol 9. Geneva: World Health Organization, 1973. 2. Santesson L, Kottmeier HL. General classification of ovarian tumors. New York: Springer-Verlag, 1968. (Gentil F, Junqueira AC, eds. Ovarian cancer, monograph series, vol II).
3. Bjorkholm E, Pettersson F, Einhorn N, Krebs I, Nilsson B, Tjernberg B. Long-term follow-up and prognostic factors in ovarian carcinoma. Acta Radio! (Oncol) 1982; 21(6):413-9. 4. Nikrui N. Survey of clinical behavior of patients with borderline epithelial tumors of the ovary. Gynecologic Oncology 1981;12:107-19. 5. Kottmeier HL, ed. Annual report on the results of treatment in gynecological cancer, vol 18. F.I.G.O. Radiumhemmet, Stockholm: Editional Office, 1982. 6. Baak JPA, Lindeman J, Overdiep SH, Langley FA. Dis-
Grading of serous ovarian tumors
Volume 151 Number 3
7. 8. 9. 10. 11.
agreement of histopathological diagnosis of different pathologists in ovarian tumors-with some theoretical considerations. Eur 1 Obstet Gynecol Reprod Bio 1982; 13:51-5. Woodrugg RD. Reviewing histologic diagnosis of lymphoma. Arch Pathol Lab Med 1981;105:573-6. Wallgren A, Silfverswiird C, Hultborn A. Carcinoma of the breast in women under 30 years of age. Cancer 1977; 40:916-23. Taylor HC. Studies in the clinical and biological evolution of adenocarcinoma of the ovary. Br 1 Obstet Gynaecol 1966;827-42. Erhardt K, Auer G, Bjorkholm E, eta!. Significance of nuclear DNA content in serous ovarian tumors. Cancer Res 1984;44:2198-202. Sorbe B, Frankendal B, Veress B. Importance of histologic grading in the prognosis of epithelial ovarian carcinoma. Obstet Gynecoll982;59(5):576-82.
12. Gaub 1. Auer G, Zetterberg A. Quantitative cytochemical aspects of a combined Feulgen-naphthol yellowS staining procedure for the simultaneous determination of nuclear and cytoplasmic proteins and DNA in mammalian cells. Exp Cell Res 1975;92:323-32. 13. Adams LR. Photographic cytophotometric method which avoids distributional error. Acta Cytol 1968;12(1):3-8. 14. Caspersson T, Kudynowski J. Cytochemical instrumentation for cytopathological work. Int Rev Exp Patholl980; 21:1-54. 15. Sigurdson IC, Aim P, Gullberg B. Prognostic factors in malignant epithelial ovarian tumors. Gynecol On col 1983; 15:370-380. 16. Auer G, Arrhenius E, Granberg P-0, Fox C. Comparison of DNA distributions in primary human breast cancers and their metastases. Eur 1 Cancer 1980;16:273-8.
The effect of a prostaglandin synthetase inhibitor on the hormonal profile and the endometrium in women B. M. Landgren, M.D., V. Lundstrom, M.D., P. Eneroth, M.D., and E. Johannisson, M.D. Stockholm, Sweden, and Geneva, Switzerland A prostaglandin synthetase inhibitor, naproxen, was given continuously throughout the menstrual cycle at a dose of 250 mg twice daily to 10 healthy fertile women (group 1) and at a dose of 1000 mg per day to eight women in the secretory phase (group 2). Blood samples were withdrawn three times a week during a control cycle and during the treatment cycle. Luteinizing hormone, follicle-stimulating hormone, prolactin, estradiol, and progesterone were analyzed. Endometrial biopsies were taken in the secretory phase of the control cycle and in the treatment cycle of group 1 and on the first day of menstruation of group 2. Naproxen treatment did not suppress ovulation in any cycle and did not affect the corpus luteum function either in group 1 or in group 2. In only one of the endometrial samples taken in the secretory phase (group 1) was the density of the lysosomes increased in the treatment cycle compared to the control cycle. The specimen taken during the early menstrual period (group 2) showed an increase in glandular epithelium and height following naproxen treatment. Furthermore, a significant increase in the number of plasmolemmal vesicles per square micrometer was observed in the capillary endothelial cells after the administration of 1000 mg of naproxen per day. This suggests that the transcellular exchange of water-soluble molecules in the endothelial cells was more active after the administration of a prostaglandin synthetase inhibitor than in the control group. In spite of the significant morphologic changes observed in the naproxen-treated material the onset of menstrual bleeding could not be prevented. The mechanism of the onset of menstruation needs to be further investigated. (AM J OBSTET GYNECOL 1985;151 :361-8.)
Key words: Endometrium, hormones, prostaglandin synthetase inhibitor Animal studies have suggested that prostaglandins are mediators of ovulation as well as luteolysis. 1• 2 A
From the Department of Obstetrics and Gynecology, Karolinska H ospital, Stockholm, and the Centre de Cytologic et de Depistage du Cancer, Geneva. This study was supported by a grant from Astra Syntex, Sodertalje, Sweden. ReceivedforpublicationApril4, 1984; accepted September 26, 1984. Reprint requests: Dr. B. M. Landgren, Department of Obstetrics and Gynecology, Karolinska Hospital, S-104 OJ Stockholm, Sweden.
similar function of prostaglandins in women has not yet been convincingly demonstrated. However, in vitro studies by Hamberger et aP·• indicate that prostaglandin F2a in the presence of norepinephrine may induce luteolysis in human corpora lutea in vitro as well as in VIVO.
Administration of prostaglandin synthetase inhibitors to a subhuman primate has been demonstrated to interfere with rupture of the follicular membrane without affecting luteinization, leading to the "blocked 361