- Email: [email protected]
Long-term survival with advanced ovarian cancer: An analysis of 5-year survivors in the Australian trial comparing combination versus sequential chlorambucil and cisplatin therapy
GYNECOLOGIC
ONCOLOGY
47, 292-297
(1992)
Long-Term Survival with Advanced Ovarian Cancer: An Analysis of 5Year Survivors in the Australian Trial Comparing Combination versus Sequential Chlorambucil and Cisplatin Therapy M. H. N. TATTERSALL, Department
of Gynaecological
Oncology,
King
C. E. SWANSON, AND H. J. SOLOMON
George
V and Royal
Received
April
Of 369patientswith apparentadvancedovarian cancerentered on a randomizedtrial betweenSeptember1979and June 1983, 56 survived more than 5 years. Central pathology review confirmed the diagnosisin 318 cases(15 casesnot ovarian primary, 36 slidesdid not reachcentral review). Two hundred ninety-eight were invasive cancerand 34 of these(11.4%) patients were alive at 5 years. The remaining20 caseswere tumorsof low malignant potential and 17 (85%) were alive at 5 years. The clinical, pathological, and treatment characteristicsof theselong-term survivors has beenascertained.No signiiicant difference in survival hasemergedbetweenthe two treatment arms being comparedcombinationchlorambuciland cisplatin versussequentialchlorambucil followed by cisplatin on treatment failure. Of 298 cases of invasive cancerthere were 133with residuumlessthan 2 cm after initial surgery and 20 of these(15%) survived 5 years compared with 14/165 (8.5%) of thosewith more tumor residuum. Clinical responsewas a poor indicator of survival. Only 7 of 46 (15%) patients with completeclinical responsesurvived 5 years; however, of 44 patients with complete surgical response,29 (65.9%)werealive at 5 years. Theseresultshighlight the improved survival prospectsof womenwith low-potentialmalignancy, even in advancedstage,comparedto thosewith invasive tumors and make a strong casefor central pathology review of all casesof apparent ovarian cancer entered on randomized trials. o 1~ Academic
Press, Inc.
INTRODUCTION Ovarian cancer is the most common fatal gynaecological cancer in the developed world. The majority of patients present with advanced-stage disease (FIG0 Stages III-IV) and the 5-year survival rate for these patients remains poor. Cisplatin-based combination chemotherapy has been reported to cause improved tumor response compared to single alkylating agent treatment but metaanalysis of all randomized trials comparing single 292 0090-8258/92 $4.00 Copyright 0 1992 by Academic Press, All rights of reproduction in any form
Inc. reserved.
Prince
Alfred
Hospital,
Camperdown,
NSW,
2050 Australia
8, 1991
with multidrug treatment has shown only a modest effect on survival [l] by combination chemotherapy. Pretreatment prognostic factors associated with improved survival include tumor characteristics and disease extent [2]. Tumor histological type and grade [2-71 and tumor cell DNA content [8] are found to be prognostically important in some studies. In addition tumor stage and the residual tumor volume after initial surgery influence not only the complete tumor response rate to chemotherapy but also survival. Additional prognostic factors have been reported by some studies--e.g., patient age [3-91 and performance status [4,6,9], the largest residual mass after surgery [5,6], the presence of multiple residual lesions after surgery [S], and the presence of ascites [3,10]. With so many factors influencing outcome, and with the majority of studies reporting small numbers of patients, it is not surprising that the impact of current treatments on survival is not well established [ll]. Most groups treat patients with all types of advancedstage epithelial invasive ovarian cancer in the same way without reference to tumor histology or grade. The Princess Margaret Hospital group [2] have analyzed histological factors predicting a prolonged survival and reported that both grade and histological type could be used to stratify patients into different risk categories that might appropriately be treated differently. Few study groups have adopted the Toronto formula and as far as we are aware, no other group has analyzed survival in their patient population according to the Toronto criteria. We entered 369 patients with Stages III and IV ovarian cancer on a prospective randomized trial comparing combination versus sequential therapy with chlorambucil and cisplatin. The initial results of this trial were reported when the median follow-up was about 36 months [12]. Cox model multivariate analysis of factors including treat-
SURVIVAL
IN OVARIAN
ment regimen, disease stage, bulk of residual disease at trial entry, type of primary surgery (debulking treatment versus other), tumor grade, patient age, and evidence of myelosuppression on chemotherapy revealed that residual tumor bulk and evidence of myelosuppression were the two major independent prognostic variables predicting survival. Allowing for these variables, tumor grade and patient age were marginally of prognostic significance. We now report the results of further analysis of patients who are long-term survivors in this study. Patient entry on this trial ceased in June 1983 and thus all patients have been followed for a minimum of 5 years from study entry. PATIENTS
AND METHODS
Three hundred sixty-nine patients with Stages III and IV ovarian adenocarcinoma were entered on this trial from all states of Australia and from New Zealand. Eligible patients had Stage III or IV ovarian adenocarcinoma based on local interpretation of tumor histology, were age 70 years or less, and were judged suitable to receive chemotherapy. Patients were excluded on grounds of poor general medical condition, a past history of other cancers, or if geography made treatment or follow-up difficult. Randomization was accomplished by telephone to the trial center. Patients were stratified prior to randomization according to the clinical stage of the disease (Stage III and IV), according to the extent of tumor remaining after surgical resection, largest mass up to 2 cm or greater, and according to whether there was measurable residual tumor. The chemotherapy comprised oral chlorambucil, 7.5 mg/m2 daily for 14 days and cisplatin, 50 mg/m’ iv with appropriate fluid administration and antiemetic treatment. Patients receiving combination chemotherapy received the cisplatin on the 14th day of chlorambucil administration. Treatment courses were repeated at intervals of 4 weeks. Evaluation by clinical examination and appropriate radiographs, isotope scans, and ultrasound was carried out prior to randomization and monthly (clinical examination) or at three monthly intervals during therapy. After completion of therapy (12 months) patients in complete clinical and pathological remission were followed up at 3-month intervals. Tumor response status was assessed by standard WHO criteria. In the absence of disease progression, patients randomized to combination chemotherapy received eight courses of treatment after which cisplatin was ceased and courses of oral chlorambucil were continued for a total of 12 months. Patients randomized to sequential treatment received courses of chlorambucil alone for 12 months or until disease progression. Patients with progressive disease documented either clinically or at secondlook laparotomy received cisplatin if initially randomized
293
CANCER
to sequential treatment or were treated on nonprotocol lines if initially randomized to combination chemotherapy. Patients who had no clinical evidence of residual tumor 12 months after trial entry were assessed regarding their suitability for second-look laparotomy. These patients either had had no clinical evidence of tumor from trial entry or had achieved a complete clinical response to chemotherapy. The outcome of surgical reexploration in these patients has been reported in another manuscript [131.
Tumor histology was reviewed centrally by a panel of gynaecological tumor pathologists. In some instances, either the pathology was not received and centrally reviewed (36) or the diagnosis was changed to nonovarian primary origin (15), or to proliferating tumor of low potential malignancy (20) by this panel. After these exclusions, there were 298 patients with histologically confirmed invasive ovarian adenocarcinoma. Statistical analysis used BMDP statistical software to prepare and compare Kaplan-Meier survival curves. Univariate analyses were supplemented by stepwise Cox model multivariate survival comparison, selecting default conditions for the inclusion and exclusion of terms and the MPLR stepping option. RESULTS
Three hundred sixty-nine patients were entered on this trial when accrual closed at the end of June 1983. Followup information until the end of June 1989 has been obtained. Table 1 documents the characteristics of patients grouped according to their randomization. Figure 1 shows the survival curves for all entered patients according to treatment groups. The median survival for patients in the sequential group is 16 months compared to 17.5 months for the combination treatment. No significant survival difference has emerged between the arms. Fifty-one patients with histologically confirmed ovarian cancer survived 5 years from trial entry. Of the 51 patients 28 were entered in the sequential therapy arm and 23 in the combination chemotherapy arm (Table 1). In the total population 20 of the 30 5-year survivors in the sequential therapy arm have still not received cisplatin treatment because their disease has not relapsed following chlorambucil treatment, and 10 of these patients had tumors of low-potential malignancy on pathological review. Twenty patients entered on trial had tumors characterized by central pathology review as of low-potential malignancy, and 17 of these patients survived 5 years (85%). In comparison although 34 other patients survived 5 years, only 29 of these were from the 298 patients (10%) with invasive ovarian tumors confirmed by central pathology review (Fig. 2). The other 5 long-term survivors
294
TATTERSALL, SWANSON, AND SOLOMON TABLE 1 Pretreatment Characteristics of Patient Groups Total population Total
Sequential
No. of patients
369
186
FIG0 stage III IV
280 89
142
Residual disease <2 cm >2 cm
Patients surviving >5 yr Combination
Total
Sequential
Combination
183
56
30
26
138 4.5
51
44
5
27 3
24 2
172 197
82 104
90
39 17
18 12
21
93
Measurable disease
166
86
80
13
9
4
Histology, review Low malignant potential Invasive ovary
20 298
13 148
7 150
17 34”
10 18
7 16
188 181
101 85
87 96
34 22
19
15 11
5
Age
<55 >55
11
’ Survival >5 yr, four unknown, one nonovarian.
included 4 of the 36 who had no central pathology review and 1 of the 15 with nonovarian histology. Twenty of these long-term survivors were from 133 patients (15%) with tumors who had less than 2 cm residuum and, 14 were from 165 patients (8.5%) with invasive tumors and greater than 2 cm residuum. The initial surgical treatment influenced long-term survival, with only 2 of the 55 patients (3.6%) whose initial surgery was biopsy only surviving 5 years compared to 51/314 (16%) who had debulking surgery. The extent of residual disease prior to chemotherapy also influenced survival. Overall 37 of 172 patients (21.5%) with <2 cm residuum survived 5 years compared to 16/197 (8%) of those with more residuum. Altogether 38 of the 5-year survivors (72%) had no clinically measurable disease at
study entry, this group coming from a total of 136 patients entered on trial without measurable disease (28%). Of those patients with clinically measurable disease 7/46 (15%) who achieved a clinical CR still survive. There were 44 patients who were classified as surgical complete responders, and 29 (65%) survived 5 years or more (Table 2).
Treatment-related acute toxicity remains as reported in our earlier paper. However, since that report, one patient in the sequential therapy arm has died of leukemia which was plausibly related to therapy. Cox’s proportional hazards model was used to ascertain those variables independently associated with survival in this study. The proportional hazards model including
- Sequenttal - Comblnatlon
----01 0
1
100
200
300
400
500
Weeks Sequential FIG. 1. Patient survival after randomization. chlorambucil --, cisplatin, - - - combination chlorambucil and cisplatin.
01 0
---I I
100
300 200 Weeks
400
500
FIG. 1. Patient survival according to tumor histology on central Tumor of low-potential malignancy, - - - inpathology review. vasive ovarian tumors.
SURVIVAL
IN OVARIAN
295
CANCER
TABLE 2 Treatment Characteristicsof Patient Groups Total population Total
Sequential
Clinical response CR PR SD PD
46 29 61 73
19 17 32 46
Pathological response CRS PRS
44 21 126 86 151
Treatment-related < 2.500 WBC
Patients surviving >5 yr Combination
Total
Sequential
Combination
21 12 29 27
I 2 2 -
3 2 2 -
4 -
21 8
23 13
29 6
15 5
14 1
34 24 47
92 62 104
21 17 28
7 6 10
14 9 18
myelosuppression
treatment arm, age, FIG0 stage, residual disease, histology type, tumor grade, and treatment-related myelosuppression found significant independent associations with improved survival for the following in descending order of significance: tumors with low-potential malignancy (P = 0.001, relative risk (RR), 0.07, 95%, confidence interval (CI) 0.02-0.22), occurrence of thrombocytopaenia during therapy (P = 0.012, RR = 0.64 (95%), CI = 0.47-0.86), age less than 55 years (P = 0.04, RR = 0.76 (95%), CI = 0.58-0.99), and for welldifferentiated tumors (P = 0.05, RR = 0.49 (95%), CI = 0.25-0.97). Restricting this analysis to only baseline characteristics of the patients in the model, i.e., age, FIG0 stage, residual disease, histology and tumor grade, showed that only tumors of low potential malignancy were independently associated with longer survival in this series (P = 0.001, RR = 0.10 (95%), CI = 0.04-0.26). In view of this association of survival with tumor type and grade, another analysis was performed classifying patients according to the scheme proposed by Dembo et al. [2], i.e., (a) well-differentiated serous and clear cell tumors versus (b) mutinous and endometrioid tumors of any grade versus (c) moderately poorly differentiated serous and clear cell tumors and undifferentiated tumors. Two hundred fourteen patients could be classified in this manner and a univariate survival analysis shows an advantage for a-type patients (median 171 weeks) compared with b- and c-type patients (medians 65 and 67 weeks, respectively), although this was not statistically significant (P = 0.10, generalized Wilcoxon statistic). The proportional hazards model including treatment arm, age, FIG0 stage, residual disease, myelosuppression during therapy, and Dembo classification showed that occurrence of thrombocytopenia during therapy (P =
0.01, PR = 0.62 (95%), CI = 0.44-0.88) and residual disease less than 2 cm (P = 0.02, RR = 0.70 (95%), CI 0.52-0.96) were associated with improved survival. The Dembo classification was found to be just significantly associated with poorer survival (P = 0.05, RR = 1.30, 95% CI 0.99-1.72), indicating that Dembo b- and c-types are not at a significantly poorer risk. DISCUSSION
There have been significant changes in the management of ovarian epithelial cancers in the past 15 years. Formal surgical staging is now regarded as an essential prerequisite to proper management. In addition, combination chemotherapy regimens incorporating cisplatin have become standard treatment for those women with advancedstage disease although clinical trial results have not demonstrated clear survival advantages for these more intensive chemotherapy programs. Few randomized trials comparing different chemotherapy regimens in advancedstage ovarian cancer have reported survival differences between the treatment arms, although higher tumor response rates have commonly been reported for the more intensive treatment regimens. There has been a recent movement toward less complicated and less toxic chemotherapy regimens which lately have included cisplatin or its less toxic analogue and cyclophosphamide and current trials are comparing different treatment durations and/or dose regimens. Several recent reports have examined prognostic factors for prolonged survival in advanced-stage epithelial ovarian cancer [2-101. None of these studies has identified chemotherapy regimens or schedule as a significant prognostic factor, although one study has reported a strong correlation between the use of cisnlatin and the nercent-
296
TATTERSALL,
SWANSON,
age of patients with a tumor residuum of less than 2 cm, cisplatin has been applied particularly in patients s&e tumors have been “initially debulked” [14]. .Most trials comparing chemotherapy regimens in advanced-stage epithelial cancer have been handicapped by small patient numbers, by lack of central pathology review, and by premature reporting of treatment outcome. Largely for these reasons and because of differences in management approach in different countries and institutions, the British Medical Research Council is currently undertaking a formal metaanalysis of all randomized trials in ovarian cancer in which single-agent treatment has been compared with multiagent treatment. The results of this analysis may help to establish the status of combination chemotherapy in this tumor type and clarify in which patient group this approach is advantageous. Only when these questions have been resolved is it likely that differences of opinion about cisplatin dose, number of treatment cycles, and timing of second-look surgery can be properly resolved. We report here the results of extended follow-up of patients with advanced ovarian cancer entered on a trial comparing sequential therapy with chlorambucil and cisplatin versus combination chemotherapy with the same drugs. The trial patients have been followed for a minimum of 6 years since trial entry. No difference in survival between the treatments being compared has emerged nor has a patient group been identified in whom one treatment approach is superior. However, several important factors correlating with survival prospects have emerged. Of greatest prognostic significance is the prolonged survival of patients with tumors of low-potential malignancy, whichever treatment they received. Our study also reveals that only 2 of 55 patients whose initial surgery was biopsy only survived 5 years compared to 51/314 who had some degree of debulking surgery. The group who had the greatest chance of 5-year survival were women with no macroscopic tumor remaining after surgery. These observations suggest that tumor removal prior to commencing chemotherapy is an important element influencing long-term survival prospects. Our results indicate that treatment-related myelosuppression is significantly associated with survival prospects, suggesting that chemotherapy at a myelotoxic dose is advantageous compared to less toxic treatment. We analyzed our results according to the schema proposed by Dembo et al. [2], which separates patients into those with (a) well-differentiated serous and clear cell tumors, (b) mutinous and endometrioid tumors of any grade, and (c) moderately poorly differentiated serous and clear cell tumors. Two hundred fourteen patients could be classified in this manner and a univariate survival analysis showed an advantage for a-type patients (median
AND SOLOMON
survival 171 weeks) compared with b- and c-type patients (median survival 65 and 67 weeks, respectively), but this was not statistically significant (P = 0.10, generalized Wilcoxon statistic). We conclude that tumor histology, and in particular identifying patients with tumors of low-potential malignancy is the most important prognostic factor for prolonged survival in Stages III and IV ovarian epithelial cancer. This finding is important since most trials have either not identified these patients or have excluded them from trial. The prognostic importance of tumor ploidy which has been reported by several groups in part may be influenced by the inclusion of low-potential malignant tumor in the diploid group that have these improved survival prospects [ 14-171. Clearly central pathology review by an experienced gynecological pathologist is a useful adjunct to the analysis of long-term survival of Stages III and IV ovarian cancer. Similar conclusions have been reached in regard to survival analyses in patients with earlier stage ovarian cancer [18]. REFERENCES 1.
2.
3.
4. 5.
6.
7.
8.
9.
10.
Advanced Ovarian Career Trialists Group. Chemotherapy in advanced ovarian cancer: An overview of randomized clinical trials, Br. Med. J. 33, 884-893 (1991). Dembo, A. J., and Bush R. S. Choice of postoperative therapy based on prognostic factors, Int. J. Radiat. Oncol. Biol. Phys. 88, 893-897 (1982). Sigurdsson, K., Alm, P., and Gullberg, B. Prognostic factors in malignant epithelial ovarian tumors, Gynecol. Oncol. 15, 370-380 (1983). Klein, B., Falkson, G., and Smit, C. F. Advanced ovarian cancer: Factors influencing survival, Cancer 55, 1829-34 1985). Redman, J. R., Petrasi, G. R., Saigo, P. E., Geller, N. L., and Hakes, T. E. Prognostic factor in advanced ovarian carcinoma, J. Clin. Oncol. 4, 515-523 (1986). Gall, S., Bundy, B., Beecham, J., Whitney, C., Hanesley, H., Lifshitz, S., and Adcock, L. L. Therapy of Stage 3 (optimal) epithelial carcinoma of the ovary with melphalan or melphalan plus corynebacterium parvum, Gynecol. Oncol. 25, 26-36 (1986). Krag, K. J., Canellos, G. P., Knapp, R. C., Parker, M. L.,, Welch, W. R., Klatt, M., and Anderson, T. Predictive factors for longterm survival in patients with advanced ovarian cancer, Gynecol. Oncol. 34, 88-93 (1989). Friedlander, M. L., Hedley, D. W., Taylor, I. W., Russel, P., Coates, A. S., and Tattersall, M. H. N. Influence of cellular DNA content on survival in advanced ovarian cancer, Cancer Res. 44, 397-401 (1984). Bruckner, H. W., Cohen, C. M., Feuer, E., Wallach, R. C., Kabekow, B., and Holland, J. F. Prognostic factors: Cisplatin regimens for patients with ovarian cancer after failure of chemotherapy, Obstet. Gynecol. 69, 114-120 (1987). Sutton, G. P., Stehman, F. B., Einhorn, L. H., Rath, L. M., Blessing, J. A., and Ehrlich. Ten year follow-up of patients receiving cisplatin, doxorubicin and cyclophosphamide chemotherapy for advanced ovarian cancer, J. Clin. Oncol. 7, 223-229 (1989).
SURVIVAL
IN OVARIAN
11. Dembo, A. J. Controversy over combination chemotherapy in advanced ovarian cancer: What we learn from reports of matured data, 1. C/in. Oncol. 4, 1573-1576 (1986). 12. Gynecological Group of the Clinical Oncological Society of Australia and the Sydney Branch of the Ludwig Institute for Cancer Research. Chemotherapy of advanced ovarian adenocarcinoma: A randomized comparison of combination versus sequential therapy using chlorambucil and cisplatin, Gynecol. Oncol. 23, 1-13 (1986). 13. Gynaecological Group, Clinical Oncological Society of Australia and the Sydney Branch Ludwig Institute for Cancer Research. The role of surgical reexploration following chemotherapy in ovarian cancer, Aust. NJ J. Obstet. Gynaecol. 28, 121-126 (1988). 14. Voest, E. E., van Houwelingen, T. C., and Neijt, J. P. A meta analysis of prognostic factors in advanced ovarian cancer with median survival and overall survival (measured with the log (relative
15.
16.
17.
18.
CANCER
297
risk)) as main objectives, Eur. J. Cancer Clin. Oncol. 25, 711-720 (1989). Iversen, O., and Skaarland, E. Ploidy assessment of benign and malignant ovarian tumour by flow cytometry: A clinicopathlogic study, Cancer 60, 82-87 (1987). Erba, E., Ubezeo, P., Pepe, S., Vaghi, V., Marsoni, S., Torri, W., Mangioni, C., Landoni, F., and D’Incalci, M. Flow cytometric analysis of DNA content in human ovarian cancers, Br. J. Cancer 60, 45-50 (1989). Jakobsen, A., and Bichel, P. Ploidy level, histopathological differentiation and response to chemotherapy in serous ovarian cancer, Eur. J. Cancer Clin. Oncol. 25, 1589-1593 (1989). Young, R. C., Walton, L. A., Ellenberg, S. S., Homesley, P., Wilbanks, G. D., Deeken, D. G., Miller, A., Park, R., and Major, F. Adjuvant therapy in Stage I and Stage II epithelial ovarian cancer, N. Engl. J. Med. 322, 1021-1927 (1990).
ONCOLOGY
47, 292-297
(1992)
Long-Term Survival with Advanced Ovarian Cancer: An Analysis of 5Year Survivors in the Australian Trial Comparing Combination versus Sequential Chlorambucil and Cisplatin Therapy M. H. N. TATTERSALL, Department
of Gynaecological
Oncology,
King
C. E. SWANSON, AND H. J. SOLOMON
George
V and Royal
Received
April
Of 369patientswith apparentadvancedovarian cancerentered on a randomizedtrial betweenSeptember1979and June 1983, 56 survived more than 5 years. Central pathology review confirmed the diagnosisin 318 cases(15 casesnot ovarian primary, 36 slidesdid not reachcentral review). Two hundred ninety-eight were invasive cancerand 34 of these(11.4%) patients were alive at 5 years. The remaining20 caseswere tumorsof low malignant potential and 17 (85%) were alive at 5 years. The clinical, pathological, and treatment characteristicsof theselong-term survivors has beenascertained.No signiiicant difference in survival hasemergedbetweenthe two treatment arms being comparedcombinationchlorambuciland cisplatin versussequentialchlorambucil followed by cisplatin on treatment failure. Of 298 cases of invasive cancerthere were 133with residuumlessthan 2 cm after initial surgery and 20 of these(15%) survived 5 years compared with 14/165 (8.5%) of thosewith more tumor residuum. Clinical responsewas a poor indicator of survival. Only 7 of 46 (15%) patients with completeclinical responsesurvived 5 years; however, of 44 patients with complete surgical response,29 (65.9%)werealive at 5 years. Theseresultshighlight the improved survival prospectsof womenwith low-potentialmalignancy, even in advancedstage,comparedto thosewith invasive tumors and make a strong casefor central pathology review of all casesof apparent ovarian cancer entered on randomized trials. o 1~ Academic
Press, Inc.
INTRODUCTION Ovarian cancer is the most common fatal gynaecological cancer in the developed world. The majority of patients present with advanced-stage disease (FIG0 Stages III-IV) and the 5-year survival rate for these patients remains poor. Cisplatin-based combination chemotherapy has been reported to cause improved tumor response compared to single alkylating agent treatment but metaanalysis of all randomized trials comparing single 292 0090-8258/92 $4.00 Copyright 0 1992 by Academic Press, All rights of reproduction in any form
Inc. reserved.
Prince
Alfred
Hospital,
Camperdown,
NSW,
2050 Australia
8, 1991
with multidrug treatment has shown only a modest effect on survival [l] by combination chemotherapy. Pretreatment prognostic factors associated with improved survival include tumor characteristics and disease extent [2]. Tumor histological type and grade [2-71 and tumor cell DNA content [8] are found to be prognostically important in some studies. In addition tumor stage and the residual tumor volume after initial surgery influence not only the complete tumor response rate to chemotherapy but also survival. Additional prognostic factors have been reported by some studies--e.g., patient age [3-91 and performance status [4,6,9], the largest residual mass after surgery [5,6], the presence of multiple residual lesions after surgery [S], and the presence of ascites [3,10]. With so many factors influencing outcome, and with the majority of studies reporting small numbers of patients, it is not surprising that the impact of current treatments on survival is not well established [ll]. Most groups treat patients with all types of advancedstage epithelial invasive ovarian cancer in the same way without reference to tumor histology or grade. The Princess Margaret Hospital group [2] have analyzed histological factors predicting a prolonged survival and reported that both grade and histological type could be used to stratify patients into different risk categories that might appropriately be treated differently. Few study groups have adopted the Toronto formula and as far as we are aware, no other group has analyzed survival in their patient population according to the Toronto criteria. We entered 369 patients with Stages III and IV ovarian cancer on a prospective randomized trial comparing combination versus sequential therapy with chlorambucil and cisplatin. The initial results of this trial were reported when the median follow-up was about 36 months [12]. Cox model multivariate analysis of factors including treat-
SURVIVAL
IN OVARIAN
ment regimen, disease stage, bulk of residual disease at trial entry, type of primary surgery (debulking treatment versus other), tumor grade, patient age, and evidence of myelosuppression on chemotherapy revealed that residual tumor bulk and evidence of myelosuppression were the two major independent prognostic variables predicting survival. Allowing for these variables, tumor grade and patient age were marginally of prognostic significance. We now report the results of further analysis of patients who are long-term survivors in this study. Patient entry on this trial ceased in June 1983 and thus all patients have been followed for a minimum of 5 years from study entry. PATIENTS
AND METHODS
Three hundred sixty-nine patients with Stages III and IV ovarian adenocarcinoma were entered on this trial from all states of Australia and from New Zealand. Eligible patients had Stage III or IV ovarian adenocarcinoma based on local interpretation of tumor histology, were age 70 years or less, and were judged suitable to receive chemotherapy. Patients were excluded on grounds of poor general medical condition, a past history of other cancers, or if geography made treatment or follow-up difficult. Randomization was accomplished by telephone to the trial center. Patients were stratified prior to randomization according to the clinical stage of the disease (Stage III and IV), according to the extent of tumor remaining after surgical resection, largest mass up to 2 cm or greater, and according to whether there was measurable residual tumor. The chemotherapy comprised oral chlorambucil, 7.5 mg/m2 daily for 14 days and cisplatin, 50 mg/m’ iv with appropriate fluid administration and antiemetic treatment. Patients receiving combination chemotherapy received the cisplatin on the 14th day of chlorambucil administration. Treatment courses were repeated at intervals of 4 weeks. Evaluation by clinical examination and appropriate radiographs, isotope scans, and ultrasound was carried out prior to randomization and monthly (clinical examination) or at three monthly intervals during therapy. After completion of therapy (12 months) patients in complete clinical and pathological remission were followed up at 3-month intervals. Tumor response status was assessed by standard WHO criteria. In the absence of disease progression, patients randomized to combination chemotherapy received eight courses of treatment after which cisplatin was ceased and courses of oral chlorambucil were continued for a total of 12 months. Patients randomized to sequential treatment received courses of chlorambucil alone for 12 months or until disease progression. Patients with progressive disease documented either clinically or at secondlook laparotomy received cisplatin if initially randomized
293
CANCER
to sequential treatment or were treated on nonprotocol lines if initially randomized to combination chemotherapy. Patients who had no clinical evidence of residual tumor 12 months after trial entry were assessed regarding their suitability for second-look laparotomy. These patients either had had no clinical evidence of tumor from trial entry or had achieved a complete clinical response to chemotherapy. The outcome of surgical reexploration in these patients has been reported in another manuscript [131.
Tumor histology was reviewed centrally by a panel of gynaecological tumor pathologists. In some instances, either the pathology was not received and centrally reviewed (36) or the diagnosis was changed to nonovarian primary origin (15), or to proliferating tumor of low potential malignancy (20) by this panel. After these exclusions, there were 298 patients with histologically confirmed invasive ovarian adenocarcinoma. Statistical analysis used BMDP statistical software to prepare and compare Kaplan-Meier survival curves. Univariate analyses were supplemented by stepwise Cox model multivariate survival comparison, selecting default conditions for the inclusion and exclusion of terms and the MPLR stepping option. RESULTS
Three hundred sixty-nine patients were entered on this trial when accrual closed at the end of June 1983. Followup information until the end of June 1989 has been obtained. Table 1 documents the characteristics of patients grouped according to their randomization. Figure 1 shows the survival curves for all entered patients according to treatment groups. The median survival for patients in the sequential group is 16 months compared to 17.5 months for the combination treatment. No significant survival difference has emerged between the arms. Fifty-one patients with histologically confirmed ovarian cancer survived 5 years from trial entry. Of the 51 patients 28 were entered in the sequential therapy arm and 23 in the combination chemotherapy arm (Table 1). In the total population 20 of the 30 5-year survivors in the sequential therapy arm have still not received cisplatin treatment because their disease has not relapsed following chlorambucil treatment, and 10 of these patients had tumors of low-potential malignancy on pathological review. Twenty patients entered on trial had tumors characterized by central pathology review as of low-potential malignancy, and 17 of these patients survived 5 years (85%). In comparison although 34 other patients survived 5 years, only 29 of these were from the 298 patients (10%) with invasive ovarian tumors confirmed by central pathology review (Fig. 2). The other 5 long-term survivors
294
TATTERSALL, SWANSON, AND SOLOMON TABLE 1 Pretreatment Characteristics of Patient Groups Total population Total
Sequential
No. of patients
369
186
FIG0 stage III IV
280 89
142
Residual disease <2 cm >2 cm
Patients surviving >5 yr Combination
Total
Sequential
Combination
183
56
30
26
138 4.5
51
44
5
27 3
24 2
172 197
82 104
90
39 17
18 12
21
93
Measurable disease
166
86
80
13
9
4
Histology, review Low malignant potential Invasive ovary
20 298
13 148
7 150
17 34”
10 18
7 16
188 181
101 85
87 96
34 22
19
15 11
5
Age
<55 >55
11
’ Survival >5 yr, four unknown, one nonovarian.
included 4 of the 36 who had no central pathology review and 1 of the 15 with nonovarian histology. Twenty of these long-term survivors were from 133 patients (15%) with tumors who had less than 2 cm residuum and, 14 were from 165 patients (8.5%) with invasive tumors and greater than 2 cm residuum. The initial surgical treatment influenced long-term survival, with only 2 of the 55 patients (3.6%) whose initial surgery was biopsy only surviving 5 years compared to 51/314 (16%) who had debulking surgery. The extent of residual disease prior to chemotherapy also influenced survival. Overall 37 of 172 patients (21.5%) with <2 cm residuum survived 5 years compared to 16/197 (8%) of those with more residuum. Altogether 38 of the 5-year survivors (72%) had no clinically measurable disease at
study entry, this group coming from a total of 136 patients entered on trial without measurable disease (28%). Of those patients with clinically measurable disease 7/46 (15%) who achieved a clinical CR still survive. There were 44 patients who were classified as surgical complete responders, and 29 (65%) survived 5 years or more (Table 2).
Treatment-related acute toxicity remains as reported in our earlier paper. However, since that report, one patient in the sequential therapy arm has died of leukemia which was plausibly related to therapy. Cox’s proportional hazards model was used to ascertain those variables independently associated with survival in this study. The proportional hazards model including
- Sequenttal - Comblnatlon
----01 0
1
100
200
300
400
500
Weeks Sequential FIG. 1. Patient survival after randomization. chlorambucil --, cisplatin, - - - combination chlorambucil and cisplatin.
01 0
---I I
100
300 200 Weeks
400
500
FIG. 1. Patient survival according to tumor histology on central Tumor of low-potential malignancy, - - - inpathology review. vasive ovarian tumors.
SURVIVAL
IN OVARIAN
295
CANCER
TABLE 2 Treatment Characteristicsof Patient Groups Total population Total
Sequential
Clinical response CR PR SD PD
46 29 61 73
19 17 32 46
Pathological response CRS PRS
44 21 126 86 151
Treatment-related < 2.500 WBC
Patients surviving >5 yr Combination
Total
Sequential
Combination
21 12 29 27
I 2 2 -
3 2 2 -
4 -
21 8
23 13
29 6
15 5
14 1
34 24 47
92 62 104
21 17 28
7 6 10
14 9 18
myelosuppression
treatment arm, age, FIG0 stage, residual disease, histology type, tumor grade, and treatment-related myelosuppression found significant independent associations with improved survival for the following in descending order of significance: tumors with low-potential malignancy (P = 0.001, relative risk (RR), 0.07, 95%, confidence interval (CI) 0.02-0.22), occurrence of thrombocytopaenia during therapy (P = 0.012, RR = 0.64 (95%), CI = 0.47-0.86), age less than 55 years (P = 0.04, RR = 0.76 (95%), CI = 0.58-0.99), and for welldifferentiated tumors (P = 0.05, RR = 0.49 (95%), CI = 0.25-0.97). Restricting this analysis to only baseline characteristics of the patients in the model, i.e., age, FIG0 stage, residual disease, histology and tumor grade, showed that only tumors of low potential malignancy were independently associated with longer survival in this series (P = 0.001, RR = 0.10 (95%), CI = 0.04-0.26). In view of this association of survival with tumor type and grade, another analysis was performed classifying patients according to the scheme proposed by Dembo et al. [2], i.e., (a) well-differentiated serous and clear cell tumors versus (b) mutinous and endometrioid tumors of any grade versus (c) moderately poorly differentiated serous and clear cell tumors and undifferentiated tumors. Two hundred fourteen patients could be classified in this manner and a univariate survival analysis shows an advantage for a-type patients (median 171 weeks) compared with b- and c-type patients (medians 65 and 67 weeks, respectively), although this was not statistically significant (P = 0.10, generalized Wilcoxon statistic). The proportional hazards model including treatment arm, age, FIG0 stage, residual disease, myelosuppression during therapy, and Dembo classification showed that occurrence of thrombocytopenia during therapy (P =
0.01, PR = 0.62 (95%), CI = 0.44-0.88) and residual disease less than 2 cm (P = 0.02, RR = 0.70 (95%), CI 0.52-0.96) were associated with improved survival. The Dembo classification was found to be just significantly associated with poorer survival (P = 0.05, RR = 1.30, 95% CI 0.99-1.72), indicating that Dembo b- and c-types are not at a significantly poorer risk. DISCUSSION
There have been significant changes in the management of ovarian epithelial cancers in the past 15 years. Formal surgical staging is now regarded as an essential prerequisite to proper management. In addition, combination chemotherapy regimens incorporating cisplatin have become standard treatment for those women with advancedstage disease although clinical trial results have not demonstrated clear survival advantages for these more intensive chemotherapy programs. Few randomized trials comparing different chemotherapy regimens in advancedstage ovarian cancer have reported survival differences between the treatment arms, although higher tumor response rates have commonly been reported for the more intensive treatment regimens. There has been a recent movement toward less complicated and less toxic chemotherapy regimens which lately have included cisplatin or its less toxic analogue and cyclophosphamide and current trials are comparing different treatment durations and/or dose regimens. Several recent reports have examined prognostic factors for prolonged survival in advanced-stage epithelial ovarian cancer [2-101. None of these studies has identified chemotherapy regimens or schedule as a significant prognostic factor, although one study has reported a strong correlation between the use of cisnlatin and the nercent-
296
TATTERSALL,
SWANSON,
age of patients with a tumor residuum of less than 2 cm, cisplatin has been applied particularly in patients s&e tumors have been “initially debulked” [14]. .Most trials comparing chemotherapy regimens in advanced-stage epithelial cancer have been handicapped by small patient numbers, by lack of central pathology review, and by premature reporting of treatment outcome. Largely for these reasons and because of differences in management approach in different countries and institutions, the British Medical Research Council is currently undertaking a formal metaanalysis of all randomized trials in ovarian cancer in which single-agent treatment has been compared with multiagent treatment. The results of this analysis may help to establish the status of combination chemotherapy in this tumor type and clarify in which patient group this approach is advantageous. Only when these questions have been resolved is it likely that differences of opinion about cisplatin dose, number of treatment cycles, and timing of second-look surgery can be properly resolved. We report here the results of extended follow-up of patients with advanced ovarian cancer entered on a trial comparing sequential therapy with chlorambucil and cisplatin versus combination chemotherapy with the same drugs. The trial patients have been followed for a minimum of 6 years since trial entry. No difference in survival between the treatments being compared has emerged nor has a patient group been identified in whom one treatment approach is superior. However, several important factors correlating with survival prospects have emerged. Of greatest prognostic significance is the prolonged survival of patients with tumors of low-potential malignancy, whichever treatment they received. Our study also reveals that only 2 of 55 patients whose initial surgery was biopsy only survived 5 years compared to 51/314 who had some degree of debulking surgery. The group who had the greatest chance of 5-year survival were women with no macroscopic tumor remaining after surgery. These observations suggest that tumor removal prior to commencing chemotherapy is an important element influencing long-term survival prospects. Our results indicate that treatment-related myelosuppression is significantly associated with survival prospects, suggesting that chemotherapy at a myelotoxic dose is advantageous compared to less toxic treatment. We analyzed our results according to the schema proposed by Dembo et al. [2], which separates patients into those with (a) well-differentiated serous and clear cell tumors, (b) mutinous and endometrioid tumors of any grade, and (c) moderately poorly differentiated serous and clear cell tumors. Two hundred fourteen patients could be classified in this manner and a univariate survival analysis showed an advantage for a-type patients (median
AND SOLOMON
survival 171 weeks) compared with b- and c-type patients (median survival 65 and 67 weeks, respectively), but this was not statistically significant (P = 0.10, generalized Wilcoxon statistic). We conclude that tumor histology, and in particular identifying patients with tumors of low-potential malignancy is the most important prognostic factor for prolonged survival in Stages III and IV ovarian epithelial cancer. This finding is important since most trials have either not identified these patients or have excluded them from trial. The prognostic importance of tumor ploidy which has been reported by several groups in part may be influenced by the inclusion of low-potential malignant tumor in the diploid group that have these improved survival prospects [ 14-171. Clearly central pathology review by an experienced gynecological pathologist is a useful adjunct to the analysis of long-term survival of Stages III and IV ovarian cancer. Similar conclusions have been reached in regard to survival analyses in patients with earlier stage ovarian cancer [18]. REFERENCES 1.
2.
3.
4. 5.
6.
7.
8.
9.
10.
Advanced Ovarian Career Trialists Group. Chemotherapy in advanced ovarian cancer: An overview of randomized clinical trials, Br. Med. J. 33, 884-893 (1991). Dembo, A. J., and Bush R. S. Choice of postoperative therapy based on prognostic factors, Int. J. Radiat. Oncol. Biol. Phys. 88, 893-897 (1982). Sigurdsson, K., Alm, P., and Gullberg, B. Prognostic factors in malignant epithelial ovarian tumors, Gynecol. Oncol. 15, 370-380 (1983). Klein, B., Falkson, G., and Smit, C. F. Advanced ovarian cancer: Factors influencing survival, Cancer 55, 1829-34 1985). Redman, J. R., Petrasi, G. R., Saigo, P. E., Geller, N. L., and Hakes, T. E. Prognostic factor in advanced ovarian carcinoma, J. Clin. Oncol. 4, 515-523 (1986). Gall, S., Bundy, B., Beecham, J., Whitney, C., Hanesley, H., Lifshitz, S., and Adcock, L. L. Therapy of Stage 3 (optimal) epithelial carcinoma of the ovary with melphalan or melphalan plus corynebacterium parvum, Gynecol. Oncol. 25, 26-36 (1986). Krag, K. J., Canellos, G. P., Knapp, R. C., Parker, M. L.,, Welch, W. R., Klatt, M., and Anderson, T. Predictive factors for longterm survival in patients with advanced ovarian cancer, Gynecol. Oncol. 34, 88-93 (1989). Friedlander, M. L., Hedley, D. W., Taylor, I. W., Russel, P., Coates, A. S., and Tattersall, M. H. N. Influence of cellular DNA content on survival in advanced ovarian cancer, Cancer Res. 44, 397-401 (1984). Bruckner, H. W., Cohen, C. M., Feuer, E., Wallach, R. C., Kabekow, B., and Holland, J. F. Prognostic factors: Cisplatin regimens for patients with ovarian cancer after failure of chemotherapy, Obstet. Gynecol. 69, 114-120 (1987). Sutton, G. P., Stehman, F. B., Einhorn, L. H., Rath, L. M., Blessing, J. A., and Ehrlich. Ten year follow-up of patients receiving cisplatin, doxorubicin and cyclophosphamide chemotherapy for advanced ovarian cancer, J. Clin. Oncol. 7, 223-229 (1989).
SURVIVAL
IN OVARIAN
11. Dembo, A. J. Controversy over combination chemotherapy in advanced ovarian cancer: What we learn from reports of matured data, 1. C/in. Oncol. 4, 1573-1576 (1986). 12. Gynecological Group of the Clinical Oncological Society of Australia and the Sydney Branch of the Ludwig Institute for Cancer Research. Chemotherapy of advanced ovarian adenocarcinoma: A randomized comparison of combination versus sequential therapy using chlorambucil and cisplatin, Gynecol. Oncol. 23, 1-13 (1986). 13. Gynaecological Group, Clinical Oncological Society of Australia and the Sydney Branch Ludwig Institute for Cancer Research. The role of surgical reexploration following chemotherapy in ovarian cancer, Aust. NJ J. Obstet. Gynaecol. 28, 121-126 (1988). 14. Voest, E. E., van Houwelingen, T. C., and Neijt, J. P. A meta analysis of prognostic factors in advanced ovarian cancer with median survival and overall survival (measured with the log (relative
15.
16.
17.
18.
CANCER
297
risk)) as main objectives, Eur. J. Cancer Clin. Oncol. 25, 711-720 (1989). Iversen, O., and Skaarland, E. Ploidy assessment of benign and malignant ovarian tumour by flow cytometry: A clinicopathlogic study, Cancer 60, 82-87 (1987). Erba, E., Ubezeo, P., Pepe, S., Vaghi, V., Marsoni, S., Torri, W., Mangioni, C., Landoni, F., and D’Incalci, M. Flow cytometric analysis of DNA content in human ovarian cancers, Br. J. Cancer 60, 45-50 (1989). Jakobsen, A., and Bichel, P. Ploidy level, histopathological differentiation and response to chemotherapy in serous ovarian cancer, Eur. J. Cancer Clin. Oncol. 25, 1589-1593 (1989). Young, R. C., Walton, L. A., Ellenberg, S. S., Homesley, P., Wilbanks, G. D., Deeken, D. G., Miller, A., Park, R., and Major, F. Adjuvant therapy in Stage I and Stage II epithelial ovarian cancer, N. Engl. J. Med. 322, 1021-1927 (1990).