Flow cytometric evaluation of epithelial ovarian cancer Vanessa M. Barnabei, MD, PhD,. David Scott Miller, MD,. Kenneth D. Bauer, PhD: Tariq M. Murad, MD: Alfred W. Rademaker, PhD,c and John R. Lurain, MD" Chicago, Illinois We investigated the prognostic significance of deoxyribonucleic acid content and proliferative activity of tumor cell populations as measured by flow cytometry of the tumor specimens from 115 women with epithelial ovarian cancer. Deoxyribonucleic acid aneuploidy was found in 87 of 115 (76%) of these cancers with a mean deoxyribonucleic acid index of 1.6 and S-phase fraction of 14.7%. The S-phase fraction of the 28 (24%) diploid tumors was 7.0%. Deoxyribonucleic acid ploidy was significantly correlated with survival. S-phase fraction was significantly correlated with ploidy, residual tumor, histology, grade, ascites, time to recurrence, and survival. Diploidy versus aneuploidy were the best discriminating values for deoxyribonucleic acid index and an S-phase fraction of greater or less than 18% for that parameter. Multivariate analysis revealed stage, S-phase fraction, residual tumor, and grade to be independently associated with time to recurrence, and stage, age, S-phase fraction, and largest metastases were factors associated with survival. Deoxyribonucleic acid ploidy did not significantly improve either model. These results suggest that abnormalities of deoxyribonucleic acid content and the proliferative activity of tumor cell populations are reflective of their biologic activity. (AM J OSSTET GVNECOL 1990;162:1584-92.)
Key words: DNA flow cytometry, ovarian cancer, S-phase fraction
Cancer of the ovary is the leading cause of death among gynecologic malignancies in the United States. Improvements in surgical and chemotherapeutic regimens have had minimal effect on long-term survival. Traditional parameters, such as stage, histologic grade, and tumor burden, have been used to stratify patients into broad prognostic groupS.1 Newer methods, such as the quantitation of deoxyribonucleic acid (DNA) content by flow cytometry, may prove to be more reliable in predicting outcome in ovarian cancer patients. Flow cytometry has been shown to be an objective method of quantitating cellular DNA content and proliferative activity in normal and neoplastic cell populations. 2 , 3 Blumenfeld et al! in this country and several investigators 5. lo abroad have reported a strong correlation between ploidy and outcome in women with epithelial ovarian cancer. Only one of these studies discussed ploidy abnormalities in relation to disease progression and time to recurrence. 9 We report the corFrom the Section of Gynecologic Oncology, Department of Obstetrics and Gynecology: the Department of Pathology,' and the Cancer Center, Biometry Section,' Northwestern University Medical School. Supported in part by American Cancer Society Clinical Oncology Career Development Award No. 87-101 and Biomedical Research Support Grant No. RR-05370, United States Public Health Service, National Institutes of Health, to D. S. M. Presented at the Fifty-seventh Annual Meeting of the Central Association of Obstetricians and Gynecologists, Scottsdale, Arizona, October 12-14, 1989. Reprint requests: David Scott Miller, MD, Prentice Women's Hospital and Maternity Center, 333 East Superior St., Suite #420, Chicago, IL 60611-3095.
6/6/20310
1584
relation of tumor ploidy, DNA index, and S-phase fraction with other known prognostic factors and on survival and recurrence for all stages of epithelial ovarian cancer.
Material and methods Patient data. The medical records of 115 women with epithelial ovarian cancer diagnosed and treated at Northwestern Memorial Hospital between 1978 and 1987 were reviewed with respect to multiple clinical parameters, as shown in Table I. Pathologic studies. Original histopathologic slides from the above patients were reviewed by one of us (T. M. M.) to ensure that there was adequate tumor tissue in the blocks used for flow cytometry. All cases were reviewed without knowledge of the clinical outcome. After confirming the diagnosis, the cases were classified according to their histologic types as presented by the World Health Organization in 1973 11 and reiterated by Scully in 1977 .12 Tumors of low malignant potential will be reported elsewhere. The tumors were graded as well-differentiated (grade 1), moderately differentiated (grade 2), and poorly differentiated (grade 3) tumors. The well-differentiated carcinomas were characterized by well-formed papillae with fibrovascular cords and no solid components. Grade 2 tumors were partially solid and glandular. The poorly differentiated tumors were recognized when the carcinoma was primarily solid with little or no glandular formation. Mitoses and nuclear pleomorphism were more often seen in high-grade tumors than low-grade
Flow cytometry in ovarian cancer
Volume 162 Number 6
tumors, but these features were not used primarily for grading. For flow cytometry study the blocks were selected when the entire tissue was represented by the malignancy. In those cases in which benign tissue was noted adjacent to the tumor, the cancerous area was identified by a permanent marker on the paraffin block. Flow cytometry. The corresponding paraffinembedded blocks were dewaxed with xylene and rehydrated with a modification of the method originally described by Hedley et al. u. 14 The tissue was then pepsin treated and resuspended in Hepes-Hanks media. After treatment with ribonuclease I (I80 U/ml), the cells were stained with propidium iodide (50 mg/ml) as detailed previously.I4 Cellular DNA content was measured on a Coulter Epics 752 flow cytometer (Coulter Electronics, Hialeah, Fla.). The data were presented as DNA fluorescence histograms from which cell cycle analysis (percent G I, percent S, percent G,M), DNA ploidy, and DNA index were calculated. The coefficient of variation about the major GIGO peak on the DNA histogram ranged from 1.0% to 6.9% in these studies (mean, 4.42%). Cases with a coefficient of variation greater than 7.5% were excluded from DNA ploidy and cell cycle analysis. DNA aneuploidy was defined only on the basis of the presence of two distinct G II Go peaks in the DNA histogram. DNA index was defined as the mean fluorescence of the DNA aneuploid peak divided by that of the DNA diploid peak. This measurement thus provides an indication of the degree of alteration or change in DNA content in relation to diploid cases (DNA index = 1.0). The percentage of S = phase cells was calculated by the MCYCLE software (Phoenix Systems, Inc.) according to a modification of the method of Dean and JetL IS Proliferative activity in this investigation was reported on the basis of the percentage of cells in S phase because this calculation has been reported to be a more precise estimate than calculations of the percentage of G 2 M. IG Statistical methods. Time to recurrence and survival were calculated from the date of primary operation. The tests used to compare DNA index and S-phase fraction with clinical parameters, such as stage, grade, histology, and residual tumor, were X2 and analysis of variance. 17 Kaplan-Meier product limit estimates,I8 Breslow's generalized Wilcoxon test,I9 and Cox regression analysis were used to evaluate the association of Sphase fraction and the other factors to recurrence and survival. I" Hazard ratios indicate the change in the risk of recurrence or death per unit change in percent S phase. Stepwise Cox multivariate regression analysis was used to assess the relative order of importance of clinical parameters to recurrence and survival time. Data analysis was performed with BMDP statistical software.'o Statistical significance was defined as values of
p < 0.05.
1585
Table I. Patient data Personal history Age Race Gravidity I parity Menarche LM P I menopause Cancer history Contraceptiv~ use Cancer family history Surgical data Primary operation Primary tumor diameter Largest metastasis diameter Size largest residual tumor Structures left Complications Ascites EBL Transfusion Second-look procedure Largest tumor diameter Size largest residual tumor Structures left Complications Ascites EBL Transfusion Histopathologic data Stage Grade Histology Ca 125 Estrogen receptor Progesterone receptor DNA index Cell cycle analysis Chemotherapy Dates Disease status Weight BSA Ca 125 Dosages Toxicity LMP, Last menstrual period; EBL, estimated blood loss; BSA, bodv surface area.
Results
Of the 115 patients evaluated, 17 (15%) had stage I disease according to International Federation of Gynecology and Obstetrics criteria, 13 (11 %) had stage II disease, 61 (53%) had stage III, and 24 (21 %) had stage IV disease. After primary tumor debulking, 29 (25%) patients were left with no residual tumor, 43 (37%) had implants 2 cm. The correlation of stage and residual tumor is shown in Table II. As seen in Table III, most cancers were serous histologically and poorly differentiated. DNA aneuploidy as defined by a DNA index other than 1.0 was found in 76% of these tumors (871115). The mean ± SE DNA index of the aneuploid cancers
1586
8arnabei et al.
June 1990
Am
J Obstet Gym'col
P 188F. 98·:L
E
R C
88
i:
68
\,.
~ 7.
R 58
E C U
p=8.19
."1-(> ......:
48
·oo. :......•....................... : D IPLO ID .'.~ ....__.-•.••-1. :··L ............... n.~~~. ____•
R 38 R
E N C E
28
ANEUPLOID
18 8
n=87 28
8
48
188
88
68
128
MONTHS
Fig. 1. Time to recurrence product limit curves for diploid (n index, p = 0.04.
P
=
28) versus aneuploid (n
p=I'I.1'I4
C 81'1 N
'.
'.
71'1
T GI'I
S
U R
87) DNA
188
E 91'1 R
E
=
51'1
.\. '·L....,
41'1
r~r;..
DIPLOID
~- .•--:
~
31'1 IJ I 21'1
...
roo.
: ... _----- •... _.•.......•....••. -
__
...-.--_._--
L··-L.
IJ
ANEUPLOID
A 11'1 L
1'1
n=2B .-_ ......... _-------_. .
n=87 1'1
21'1
61'1
41'1
11'11'1
81'1
141'1
121'1
1GI'I
MONTHS
Fig. 2. Product limit survival curves for diploid (n 0.19.
p=
=
28) versus aneuploid (n
= 87)
DNA index,
Table II. Stage versus residual tumor Residual tumor
Stage
No.
II III
I
15 10 4
Total
29
IV
I
>2 em
1870 (n = ~5). P < 0.001.
lB8
p E R C B8 E 7B N T GB
8
,,"\.. I
..-
I:·L
' ..
,
3B
...-.--.-.....-.
l •.• "1....
'.,
,
V I 2B V A UI L
B
I~.~
'.
51'1
U 41'1
R
p(B.BBl
'.
%8(lB
•.. -.
%8>1B ~------.--- .. 11=25
1 • • _ • • • ____ • • • _______ ...
1'1
2B
41'1
GB
BB
lB8
11=89 12B
14B
1GB
MONTHS Fig. 4_ Product limit survival curves for S-phase fraction < 18o/c (n 0.001.
P
1870 (II
= 25),
and 3 of other causes). Of the remaining 32 patients, 24 are alive without evidence of disease and 8 are alive with disease. Median time to recurrence was 13 months and median length of survival was 26 months. Tumor ploidy and S-phase fraction were correlated with time to recurrence and survival. DNA ploidy correlated with survival (p = 0.04), whereas S-phase fraction was correlated with time to recurrence (hazard ratio = 1.04, P = 0.0008) and survival (hazard ratio = 1.04, P = 0.0013). Time to recurrence and survival time versus DNA index were evaluated by product-limit survival analysis. Tumor diploidy versus aneuploidy was the best prog-
1588 Barnabei et al.
June 1990 Am J Obstet Gynecol
Table III. Histologic type versus grade Grade
J
Serous Endometrioid Mucinous Clear cell Undifferentiated
%
4 7 6
No.
I
%
24 6 4 1 1
17
Total X2 ,
I
No.
Histologic type
3
2
8
I
No.
%
No.
I
%
38 10 2 10 33
36
60
Total
Unspecified
59
2
2
No.
I
%
67 23 10 4 11
58 20 9 3 10
115
100
P < 0.0001.
Table IV. Stage, DNA ploidy, and S-phase fraction % S-phase
Aneuploid Stage
I II III IV
I
No.
%
10/17 9/13 49/61 19/24
P = 0.29
Table VII. Grade, DNA ploidy, and S-phase fraction
59 69 80 80
Faction (mean ± SE)
9.5 10.2 13.1 15.9
± 2.1
± 1.9 ± 0.9 ± 2.1
I
Residual tumor
No.
Ocm :s2cm >2cm
20/29 34/43 30/38
P=
%
69 79 79 0.55
9.7 ± 1.5 13.8 ± 1.2 15.0 ± 1.4 p = 0.027
Table VI. Histologic type, DNA ploidy, and S-phase fraction
No.
Serous Mucinous Endometrioid Clear cell Adenocarcinoma
56/67 7/10 13/23 2/4 9/ II
P=
1% 84 70 56 50 82 0.069
1
9117 27/36 49/60
2 3
% S-phase
fraction (mean ± SE)
14.9 9.1 10.7 5.1 10.9
P=
P=
%
fraction (mean ± SE)
53 75 82 0.053
7.2 ± 1.7 13.5 ± 1.4 14.3 ± 1.0 p = 0.006
% S-phase
Aneuploid
fraction (mean ± SE)
Histologic type
No.
Table VIII. Ascites, DNA ploidy, and S-phase fraction
% S-phase
Aneuploid
1
Grade
P = 0.059
Table V. Residual tumor, DNA ploidy, and S-phase fraction Aneuploid
% S-phase
Aneuploid
± 1.0 ± 2.4 ± 1.8
± 0.7 ± 2.6
0.017
nostic discriminator for DNA index. The median time to recurrence was 14 months for the diploid tumors and 11 months for the aneuploid tumors (p = 0.19) (Fig. 1). The median patient survival was 36 months
%
fraction (mean ± SE)
81 68 P = 0.19
14.5 ± 1.1 11.0 ± 1.2 p = 0.042
I
Ascites
No.
Present Absent
52/64 25/37
for the diploid tumors and 23 months for aneuploid tumors (p = 0.04) (Fig. 2). An S-phase fraction of 18% was the best prognostic discriminating level of that variable. The median time to recurrence was 17 months for those fractions < 18% and 5 months for those> 18% (p < 0.001) (Fig. 3). The median survival was 32 months for those < 18% and 12 months for those> 18% (p < 0.001) (Fig. 4). The other parameters listed in Table I were correlated with outcome by univariate analysis. Stage and residual tumor, the diameter of the largest metastases, ascites, and size of residual tumor were significantly associated with time to recurrence and survival (p < 0.001). Primary tumor diameter was associated with time to recurrence (p = 0.04). Survival was also assocociated with age (p = 0.0097). On the basis of results of univariate analysis and other studies,! 10 factors were chosen for analysis by Cox multivariate stepwise regression. The variables that
Flow cytometry in ovarian cancer
Volume lfi2 l\umbt'r 6
were independently associated with time to recurrence were stage, S-phase fraction, residual tumor, and grade Crable IX). There was no significant improvement in the model by the addition of DNA ploidy. However, stage, age, S-phase fraction, and largest metastasis diameter were independently associated with survival Crable X). DNA ploidy did not significantly improve the survival model. Comment
The application in recent years of DNA flow cytometry has provided supplemental information regarding prognosis for women with epithelial ovarian cancer. Our results demonstrate that tumor ploidy and S-phase fraction are prognostic factors in epithelial ovarian cancer. This is a consistent finding in other recent reports,5lO although the importance of flow cytometry results in multivariate analysis varies from study to study. In our analysis, S-phase fraction was predictive of recurrence and survival, with 18% being the most sensitive predictor for that variable. This suggests that proliferative activity significantly influences the clinical behavior of these tumors, such that tumors with a high percentage of actively dividing cells behave in a more aggressive manner. Review of the literature reveals a considerable range in the frequency of aneuploidy detected in epithelial ovarian tumors, from a low of 52% to a high of 81 %. Factors that may influence the frequency of aneuploidy include the stage of disease (i.e., inclusion of early stage disease), the type of tissue analyzed (i.e., fresh versus paraffin blocks), the area of tumor analyzed, and the sensitivity of the techniques and instrumentation used. For this study, the original histopathologic slides were reviewed by a single individual (T.M.M.) without knowledge of clinical outcome; areas of high tumor density were marked for flow cytometry. This process may be more time-consuming than the use of fresh tissue or random blocks, but we believe that it lends consistency to our results and provides an accurate assessment of tumor aneuploidy in our patient population. Of the recent studies in the literature, only Rodenburg et al." have addressed the issue of flow cytometry on progression-free interval. In their analysis, as in ours, ploidy was dropped out as an independent prognostic factor in multivariate analysis. However, S-phase fraction, stage, residual tumor, and grade were predictive of time to recurrence and survival by multivariate analysis in this study. The dismal survival of patients with aneuploid tumors and those with high S-phase fraction suggests inherent differences exist between these tumors and diploid tumors. Although flow cytometrv quantitates cellular DNA content and proliferative activity, it can
1589
Table IX. Cox multivariate analysis of time to recurrence Step
Variable
p to Enter
2 3 4 5 6 7 8 9 10 11
Stage S-phase fraction Residual tumor Grade Chemotherapy type Ascites Age Ploidv Histo'iogic type Metastasis diameter Primary tumor diameter
7.5 were excluded. In the studies I have seen on ovarian cancer, the aneuploidy percentage in fresh tissue is less than that found in paraffin-embedded tissue, although the peaks are actually clearer. The coefficient of variability is better, but there is a decreased percentage of aneuploidy. That is why I think our process of choosing tumors on the basis of their high tumor density is more indicative of the amount of aneuploidy in these tumor cells. Finally, I think that until we are more successful in identification of patients with ovarian cancer before their disease is advanced, we must use whatever other information is available to improve survival in these women. I think the application of DNA flow cytometry will have clinical usefulness in the future for the evaluation and treatment of patients with epithelial ovarian cancer.
Key words: DNA flow cytometry, ovarian cancer, S-phase fraction
Cancer of the ovary is the leading cause of death among gynecologic malignancies in the United States. Improvements in surgical and chemotherapeutic regimens have had minimal effect on long-term survival. Traditional parameters, such as stage, histologic grade, and tumor burden, have been used to stratify patients into broad prognostic groupS.1 Newer methods, such as the quantitation of deoxyribonucleic acid (DNA) content by flow cytometry, may prove to be more reliable in predicting outcome in ovarian cancer patients. Flow cytometry has been shown to be an objective method of quantitating cellular DNA content and proliferative activity in normal and neoplastic cell populations. 2 , 3 Blumenfeld et al! in this country and several investigators 5. lo abroad have reported a strong correlation between ploidy and outcome in women with epithelial ovarian cancer. Only one of these studies discussed ploidy abnormalities in relation to disease progression and time to recurrence. 9 We report the corFrom the Section of Gynecologic Oncology, Department of Obstetrics and Gynecology: the Department of Pathology,' and the Cancer Center, Biometry Section,' Northwestern University Medical School. Supported in part by American Cancer Society Clinical Oncology Career Development Award No. 87-101 and Biomedical Research Support Grant No. RR-05370, United States Public Health Service, National Institutes of Health, to D. S. M. Presented at the Fifty-seventh Annual Meeting of the Central Association of Obstetricians and Gynecologists, Scottsdale, Arizona, October 12-14, 1989. Reprint requests: David Scott Miller, MD, Prentice Women's Hospital and Maternity Center, 333 East Superior St., Suite #420, Chicago, IL 60611-3095.
6/6/20310
1584
relation of tumor ploidy, DNA index, and S-phase fraction with other known prognostic factors and on survival and recurrence for all stages of epithelial ovarian cancer.
Material and methods Patient data. The medical records of 115 women with epithelial ovarian cancer diagnosed and treated at Northwestern Memorial Hospital between 1978 and 1987 were reviewed with respect to multiple clinical parameters, as shown in Table I. Pathologic studies. Original histopathologic slides from the above patients were reviewed by one of us (T. M. M.) to ensure that there was adequate tumor tissue in the blocks used for flow cytometry. All cases were reviewed without knowledge of the clinical outcome. After confirming the diagnosis, the cases were classified according to their histologic types as presented by the World Health Organization in 1973 11 and reiterated by Scully in 1977 .12 Tumors of low malignant potential will be reported elsewhere. The tumors were graded as well-differentiated (grade 1), moderately differentiated (grade 2), and poorly differentiated (grade 3) tumors. The well-differentiated carcinomas were characterized by well-formed papillae with fibrovascular cords and no solid components. Grade 2 tumors were partially solid and glandular. The poorly differentiated tumors were recognized when the carcinoma was primarily solid with little or no glandular formation. Mitoses and nuclear pleomorphism were more often seen in high-grade tumors than low-grade
Flow cytometry in ovarian cancer
Volume 162 Number 6
tumors, but these features were not used primarily for grading. For flow cytometry study the blocks were selected when the entire tissue was represented by the malignancy. In those cases in which benign tissue was noted adjacent to the tumor, the cancerous area was identified by a permanent marker on the paraffin block. Flow cytometry. The corresponding paraffinembedded blocks were dewaxed with xylene and rehydrated with a modification of the method originally described by Hedley et al. u. 14 The tissue was then pepsin treated and resuspended in Hepes-Hanks media. After treatment with ribonuclease I (I80 U/ml), the cells were stained with propidium iodide (50 mg/ml) as detailed previously.I4 Cellular DNA content was measured on a Coulter Epics 752 flow cytometer (Coulter Electronics, Hialeah, Fla.). The data were presented as DNA fluorescence histograms from which cell cycle analysis (percent G I, percent S, percent G,M), DNA ploidy, and DNA index were calculated. The coefficient of variation about the major GIGO peak on the DNA histogram ranged from 1.0% to 6.9% in these studies (mean, 4.42%). Cases with a coefficient of variation greater than 7.5% were excluded from DNA ploidy and cell cycle analysis. DNA aneuploidy was defined only on the basis of the presence of two distinct G II Go peaks in the DNA histogram. DNA index was defined as the mean fluorescence of the DNA aneuploid peak divided by that of the DNA diploid peak. This measurement thus provides an indication of the degree of alteration or change in DNA content in relation to diploid cases (DNA index = 1.0). The percentage of S = phase cells was calculated by the MCYCLE software (Phoenix Systems, Inc.) according to a modification of the method of Dean and JetL IS Proliferative activity in this investigation was reported on the basis of the percentage of cells in S phase because this calculation has been reported to be a more precise estimate than calculations of the percentage of G 2 M. IG Statistical methods. Time to recurrence and survival were calculated from the date of primary operation. The tests used to compare DNA index and S-phase fraction with clinical parameters, such as stage, grade, histology, and residual tumor, were X2 and analysis of variance. 17 Kaplan-Meier product limit estimates,I8 Breslow's generalized Wilcoxon test,I9 and Cox regression analysis were used to evaluate the association of Sphase fraction and the other factors to recurrence and survival. I" Hazard ratios indicate the change in the risk of recurrence or death per unit change in percent S phase. Stepwise Cox multivariate regression analysis was used to assess the relative order of importance of clinical parameters to recurrence and survival time. Data analysis was performed with BMDP statistical software.'o Statistical significance was defined as values of
p < 0.05.
1585
Table I. Patient data Personal history Age Race Gravidity I parity Menarche LM P I menopause Cancer history Contraceptiv~ use Cancer family history Surgical data Primary operation Primary tumor diameter Largest metastasis diameter Size largest residual tumor Structures left Complications Ascites EBL Transfusion Second-look procedure Largest tumor diameter Size largest residual tumor Structures left Complications Ascites EBL Transfusion Histopathologic data Stage Grade Histology Ca 125 Estrogen receptor Progesterone receptor DNA index Cell cycle analysis Chemotherapy Dates Disease status Weight BSA Ca 125 Dosages Toxicity LMP, Last menstrual period; EBL, estimated blood loss; BSA, bodv surface area.
Results
Of the 115 patients evaluated, 17 (15%) had stage I disease according to International Federation of Gynecology and Obstetrics criteria, 13 (11 %) had stage II disease, 61 (53%) had stage III, and 24 (21 %) had stage IV disease. After primary tumor debulking, 29 (25%) patients were left with no residual tumor, 43 (37%) had implants 2 cm. The correlation of stage and residual tumor is shown in Table II. As seen in Table III, most cancers were serous histologically and poorly differentiated. DNA aneuploidy as defined by a DNA index other than 1.0 was found in 76% of these tumors (871115). The mean ± SE DNA index of the aneuploid cancers
1586
8arnabei et al.
June 1990
Am
J Obstet Gym'col
P 188F. 98·:L
E
R C
88
i:
68
\,.
~ 7.
R 58
E C U
p=8.19
."1-(> ......:
48
·oo. :......•....................... : D IPLO ID .'.~ ....__.-•.••-1. :··L ............... n.~~~. ____•
R 38 R
E N C E
28
ANEUPLOID
18 8
n=87 28
8
48
188
88
68
128
MONTHS
Fig. 1. Time to recurrence product limit curves for diploid (n index, p = 0.04.
P
=
28) versus aneuploid (n
p=I'I.1'I4
C 81'1 N
'.
'.
71'1
T GI'I
S
U R
87) DNA
188
E 91'1 R
E
=
51'1
.\. '·L....,
41'1
r~r;..
DIPLOID
~- .•--:
~
31'1 IJ I 21'1
...
roo.
: ... _----- •... _.•.......•....••. -
__
...-.--_._--
L··-L.
IJ
ANEUPLOID
A 11'1 L
1'1
n=2B .-_ ......... _-------_. .
n=87 1'1
21'1
61'1
41'1
11'11'1
81'1
141'1
121'1
1GI'I
MONTHS
Fig. 2. Product limit survival curves for diploid (n 0.19.
p=
=
28) versus aneuploid (n
= 87)
DNA index,
Table II. Stage versus residual tumor Residual tumor
Stage
No.
II III
I
15 10 4
Total
29
IV
I
>2 em
1870 (n = ~5). P < 0.001.
lB8
p E R C B8 E 7B N T GB
8
,,"\.. I
..-
I:·L
' ..
,
3B
...-.--.-.....-.
l •.• "1....
'.,
,
V I 2B V A UI L
B
I~.~
'.
51'1
U 41'1
R
p(B.BBl
'.
%8(lB
•.. -.
%8>1B ~------.--- .. 11=25
1 • • _ • • • ____ • • • _______ ...
1'1
2B
41'1
GB
BB
lB8
11=89 12B
14B
1GB
MONTHS Fig. 4_ Product limit survival curves for S-phase fraction < 18o/c (n 0.001.
P
1870 (II
= 25),
and 3 of other causes). Of the remaining 32 patients, 24 are alive without evidence of disease and 8 are alive with disease. Median time to recurrence was 13 months and median length of survival was 26 months. Tumor ploidy and S-phase fraction were correlated with time to recurrence and survival. DNA ploidy correlated with survival (p = 0.04), whereas S-phase fraction was correlated with time to recurrence (hazard ratio = 1.04, P = 0.0008) and survival (hazard ratio = 1.04, P = 0.0013). Time to recurrence and survival time versus DNA index were evaluated by product-limit survival analysis. Tumor diploidy versus aneuploidy was the best prog-
1588 Barnabei et al.
June 1990 Am J Obstet Gynecol
Table III. Histologic type versus grade Grade
J
Serous Endometrioid Mucinous Clear cell Undifferentiated
%
4 7 6
No.
I
%
24 6 4 1 1
17
Total X2 ,
I
No.
Histologic type
3
2
8
I
No.
%
No.
I
%
38 10 2 10 33
36
60
Total
Unspecified
59
2
2
No.
I
%
67 23 10 4 11
58 20 9 3 10
115
100
P < 0.0001.
Table IV. Stage, DNA ploidy, and S-phase fraction % S-phase
Aneuploid Stage
I II III IV
I
No.
%
10/17 9/13 49/61 19/24
P = 0.29
Table VII. Grade, DNA ploidy, and S-phase fraction
59 69 80 80
Faction (mean ± SE)
9.5 10.2 13.1 15.9
± 2.1
± 1.9 ± 0.9 ± 2.1
I
Residual tumor
No.
Ocm :s2cm >2cm
20/29 34/43 30/38
P=
%
69 79 79 0.55
9.7 ± 1.5 13.8 ± 1.2 15.0 ± 1.4 p = 0.027
Table VI. Histologic type, DNA ploidy, and S-phase fraction
No.
Serous Mucinous Endometrioid Clear cell Adenocarcinoma
56/67 7/10 13/23 2/4 9/ II
P=
1% 84 70 56 50 82 0.069
1
9117 27/36 49/60
2 3
% S-phase
fraction (mean ± SE)
14.9 9.1 10.7 5.1 10.9
P=
P=
%
fraction (mean ± SE)
53 75 82 0.053
7.2 ± 1.7 13.5 ± 1.4 14.3 ± 1.0 p = 0.006
% S-phase
Aneuploid
fraction (mean ± SE)
Histologic type
No.
Table VIII. Ascites, DNA ploidy, and S-phase fraction
% S-phase
Aneuploid
1
Grade
P = 0.059
Table V. Residual tumor, DNA ploidy, and S-phase fraction Aneuploid
% S-phase
Aneuploid
± 1.0 ± 2.4 ± 1.8
± 0.7 ± 2.6
0.017
nostic discriminator for DNA index. The median time to recurrence was 14 months for the diploid tumors and 11 months for the aneuploid tumors (p = 0.19) (Fig. 1). The median patient survival was 36 months
%
fraction (mean ± SE)
81 68 P = 0.19
14.5 ± 1.1 11.0 ± 1.2 p = 0.042
I
Ascites
No.
Present Absent
52/64 25/37
for the diploid tumors and 23 months for aneuploid tumors (p = 0.04) (Fig. 2). An S-phase fraction of 18% was the best prognostic discriminating level of that variable. The median time to recurrence was 17 months for those fractions < 18% and 5 months for those> 18% (p < 0.001) (Fig. 3). The median survival was 32 months for those < 18% and 12 months for those> 18% (p < 0.001) (Fig. 4). The other parameters listed in Table I were correlated with outcome by univariate analysis. Stage and residual tumor, the diameter of the largest metastases, ascites, and size of residual tumor were significantly associated with time to recurrence and survival (p < 0.001). Primary tumor diameter was associated with time to recurrence (p = 0.04). Survival was also assocociated with age (p = 0.0097). On the basis of results of univariate analysis and other studies,! 10 factors were chosen for analysis by Cox multivariate stepwise regression. The variables that
Flow cytometry in ovarian cancer
Volume lfi2 l\umbt'r 6
were independently associated with time to recurrence were stage, S-phase fraction, residual tumor, and grade Crable IX). There was no significant improvement in the model by the addition of DNA ploidy. However, stage, age, S-phase fraction, and largest metastasis diameter were independently associated with survival Crable X). DNA ploidy did not significantly improve the survival model. Comment
The application in recent years of DNA flow cytometry has provided supplemental information regarding prognosis for women with epithelial ovarian cancer. Our results demonstrate that tumor ploidy and S-phase fraction are prognostic factors in epithelial ovarian cancer. This is a consistent finding in other recent reports,5lO although the importance of flow cytometry results in multivariate analysis varies from study to study. In our analysis, S-phase fraction was predictive of recurrence and survival, with 18% being the most sensitive predictor for that variable. This suggests that proliferative activity significantly influences the clinical behavior of these tumors, such that tumors with a high percentage of actively dividing cells behave in a more aggressive manner. Review of the literature reveals a considerable range in the frequency of aneuploidy detected in epithelial ovarian tumors, from a low of 52% to a high of 81 %. Factors that may influence the frequency of aneuploidy include the stage of disease (i.e., inclusion of early stage disease), the type of tissue analyzed (i.e., fresh versus paraffin blocks), the area of tumor analyzed, and the sensitivity of the techniques and instrumentation used. For this study, the original histopathologic slides were reviewed by a single individual (T.M.M.) without knowledge of clinical outcome; areas of high tumor density were marked for flow cytometry. This process may be more time-consuming than the use of fresh tissue or random blocks, but we believe that it lends consistency to our results and provides an accurate assessment of tumor aneuploidy in our patient population. Of the recent studies in the literature, only Rodenburg et al." have addressed the issue of flow cytometry on progression-free interval. In their analysis, as in ours, ploidy was dropped out as an independent prognostic factor in multivariate analysis. However, S-phase fraction, stage, residual tumor, and grade were predictive of time to recurrence and survival by multivariate analysis in this study. The dismal survival of patients with aneuploid tumors and those with high S-phase fraction suggests inherent differences exist between these tumors and diploid tumors. Although flow cytometrv quantitates cellular DNA content and proliferative activity, it can
1589
Table IX. Cox multivariate analysis of time to recurrence Step
Variable
p to Enter
2 3 4 5 6 7 8 9 10 11
Stage S-phase fraction Residual tumor Grade Chemotherapy type Ascites Age Ploidv Histo'iogic type Metastasis diameter Primary tumor diameter
7.5 were excluded. In the studies I have seen on ovarian cancer, the aneuploidy percentage in fresh tissue is less than that found in paraffin-embedded tissue, although the peaks are actually clearer. The coefficient of variability is better, but there is a decreased percentage of aneuploidy. That is why I think our process of choosing tumors on the basis of their high tumor density is more indicative of the amount of aneuploidy in these tumor cells. Finally, I think that until we are more successful in identification of patients with ovarian cancer before their disease is advanced, we must use whatever other information is available to improve survival in these women. I think the application of DNA flow cytometry will have clinical usefulness in the future for the evaluation and treatment of patients with epithelial ovarian cancer.
