- © 2007 by American Society of Clinical Oncology
Adjuvant Chemotherapy for Early-Stage Ovarian Cancer: Review of the Literature
- From the Department of Gynecologic Oncology, The Norwegian Radium Hospital; Faculty Division, the Norwegian Radium Hospital, Faculty of Medicine, University of Oslo, Oslo, Norway
- Address reprint requests to Claes Tropé, MD, PhD, Department of Gynecologic Oncology, the Norwegian Radium Hospital, Montebello, 0310 Oslo, Norway; e-mail: c.g.trope{at}medisin.uio.no
Abstract
Purpose This overview summarizes studies with acceptable quality and validity and presents a synthesis of the effectiveness on adjuvant therapy after surgery for early ovarian cancer (EOC) patients.
Methods The literature published between 1970 and 2006 was identified systematically by computer-based searches in MEDLINE and Cochrane library.
Results Twenty-two prospective randomized studies were analyzed, which included 4,626 patients. No difference between adjuvant chemotherapy (AC) and radiotherapy was found. There is agreement on that patients with stage IA, grade 1 tumors have excellent survival and do not need postsurgical therapy. The International Collaborative Ovarian Neoplasm 1/Adjuvant Chemotherapy in Ovarian Neoplasm trials were the first to show an effect on survival of AC, but in patients with adequate surgical staging, there was no additional effect of AC. For patients who are staged incompletely at the time of initial surgery, completion of the staging procedure with either laparoscopy or laparotomy is a reasonable approach before a final decision is made regarding the need for AC. If full staging cannot be performed due to medical contraindication or patient refusal, consideration of AC is reasonable in selected patients. Using prognostic variables such as grade, International Federation of Gynecology and Obstetrics substage, pretreatment of CA-125 ≤ 30 U/mL, and DNA ploidy, it is possible to divide patients into risk groups to avoid overtreatment. Gynecologic Oncology Group study 157 suggests that it may be possible to minimize chemotherapy-induced toxicity by using three instead of six cycles of AC, although it is not known fully whether this will compromise effectiveness.
Conclusion Future randomized studies in EOC will include the investigation of new targeted therapies and new prognostic factors in adequately staged patients.
INTRODUCTION
Ovarian cancer is the second most common gynecologic cancer, with an incidence of about 15 cases per 100,000 women in Western countries, and approximately 205,000 new cases and 125,000 deaths worldwide annually.1 Although the incidence varies from one study to another, approximately one third of the patients present with localized early-stage (EOC) disease classified as International Federation of Obstetrics and Gynecology (FIGO) stages I and II.2 The prognosis for these women is much better than for women with a higher stage of disease at the time of diagnosis. The 5-year survival rates reported in the literature for patients with EOC vary, in part because of differences in surgical staging and grade of differentiation, and inclusion of borderline tumors; 5-year survival rates range from 50% to 95% and 30% to 80% for patients with stage I and II disease, respectively.1,3-5 In Norway, the population-based overall relative 5-year survival rate has improved between 1975 and 1999 from 80% to 90% for stage I and from 42% to 62% for stage II disease.6 The improvement in relative survival has occurred during the period in which adjuvant chemotherapy (AC) has been used in treatment of EOC. However, surgery has also been improved during the same period. The improvement in survival might be due to earlier diagnosis, better anesthesia, and the intensive use of medicines and antibiotics. Furthermore, the discrimination between borderline tumors and invasive ovarian cancer has been better in this time period. All of these factors might have contributed to the observed moderate increase in survival. However, it is not possible to calculate the relative significance of these factors from historical data. Although the long-term survival is good compared with patients with advanced disease, approximately 10% to 50% of women with EOC will experience relapse or die as a result of their disease4,7-9 Because of these suboptimal long-term survival figures, major efforts have been made to develop adjuvant therapies, to improve surgical staging, and to identify prognostic factors that can predict patient outcome.
AC most often involves the systemic administration of chemotherapeutic agents to patients without evidence of residual tumor after removal of the primary tumor. This approach is based on data from 1950 to 1960, showing an inverse relationship between response to chemotherapy and number of tumor cells. The possibility of improved survival in patients with minimal disease after surgery, together with a poorer response in advanced disease, provided a good rationale for the use of AC. There are two major problems regarding adjuvant therapy in EOC.10 The first problem is determining prognostic factors that can predict the presence of micrometastasis (ie, disease that is no longer localized); the second problem is to determine AC treatments that are effective in controlling micrometastatic disease with tolerable short- and long-term adverse effects. Patients with a statistically significant risk for having persistent disease should be treated with AC. Actually, only a fraction of the patient population treated has micrometastatic disease and can potentially benefit from the treatment. Therefore, the role of AC in patients with EOC remains controversial, with the exception of patients with stage II disease, who in our opinion should be classified as having advanced-stage disease, for which all patients should have postoperative AC with carboplatin combined with paclitaxel (CP).
Staging
Accurate surgical staging is crucial to the assessment of prognosis of EOC. Trimbos et al9 propose two types of EOC: completely staged (small or minimal likelihood of residual disease), and incompletely staged (presence of residual disease ± 24% depending on grade). A report by Young et al11 showed that staging often was performed carelessly. They performed prospective, systematic restaging within 4 weeks after referral of 100 EOC patients. Thirty-one patients were upstaged, and 23 of these had stage III disease. Information was often missing about the omentum, the retroperitoneal nodes, and the peritoneal-diaphragmatic sites.11,12
Even though surgical staging is part of the FIGO staging system, in the literature only 10% to 30% of patients with EOC have undergone proper surgical staging.13 Involved nodes are present in 5% to 25% of cases of stage I disease.14-16 Maggioni et al17 performed a randomized trial of patients with ovarian carcinoma confined to the pelvis on macroscopic examination; the study compared systematic lymphadenectomy versus lymph node sampling only. The main result from this study was that a statistically significantly higher proportion of patients was found to have metastatic involvement of pelvic and/or aortic nodes in the lymphadenectomy group compared with the lymph node sampling group (22% v 9%, respectively; P = .007). This indicates an upstaging of these apparent EOCs to stage IIIC disease. Systematic lymphadenectomy was associated with a statistically insignificant improvement in disease-free survival (DFS) and overall survival (OS). Despite these data, there is still a lack of convincing evidence that systematic lymphadenectomy has survival benefit. In 351 patients with stage I ovarian cancer, Zanetta et al18 showed that the extent of the surgical staging was a statistically independent prognostic factor for both DFS and OS.
Three prospective observational studies have been published in which patients did not receive AC after surgery.9,19,20 These studies demonstrate the natural course for patients with EOC, and support emphasis on the importance of proper staging of ovarian cancer. On the basis of these findings, the Norwegian Radium Hospital (NRH) defined an optimal staging procedure listed in Table 1, which is similar to that from the European Organisation for Research and Treatment of Cancer (EORTC) and Gynecologic Oncology Group (GOG).
NRH Surgical Staging Procedures in EOC
Low-Risk and High-Risk EOC
Patients with EOC are divided into risk groups. The excellent prognosis of patients with low-risk stage IA grade 1 tumor treated with complete surgery alone is widely recognized.3 They have a very good prognosis with approximately 95% 5-year OS, and fertility-sparing surgery is possible.21,22 The main question is how to define high-risk EOC. Earlier reports on EOC have identified FIGO substage, grade of differentiation, histologic type (eg, clear cell carcinoma), dense adhesions, large volume of ascites, rupture before or during surgery, extracapsular growth, and age as independent prognostic factors.20,23-25 The largest retrospective multivariate analysis in stage I invasive epithelial ovarian cancer, which included a total of 1,545 women from six different databases from Canada and Europe, concluded that the most important independent prognostic factors were degree of differentiation followed by rupture before surgery, rupture during surgery, FIGO substage IB versus IA, and age.5 Furthermore, FIGO stage IB disease had a prognosis similar to that of stage IC. Vergote et al5 found that clear cell cancer did not have a poorer prognosis than the seropapillary type. The high-risk group including patients with stage IA or IB with unfavorable histology and/or grade 3, and stage Ic, are candidates for adjuvant treatment. The optimal adjuvant treatment for this group has not yet been established.7
Purpose
Controlled prospective randomized clinical studies are considered a prerequisite for documentation of treatment effect. The EOC studies to date have been too small and lacked power to detect or reject expected treatment effects. The aim of this study was to determine whether AC adds survival benefit in patients with EOC.
METHODS
Search Strategy and Selection Criteria
This overview summarizes studies with acceptable quality and valid results, and presents a synthesis of the available documentation of AC and adjuvant radiotherapy in the primary treatment of EOC using qualitative methodology. The scientific literature, published in the time period 1970 to 2006, was identified systematically by computer-based searches in MEDLINE, and a supplementary search was made in the Cochrane library. Additional manual searches in relevant journal indexes and reference lists of retrieved articles were performed. The search term EOC was used in combination with AC/adjuvant radiotherapy, surgery, and prospective randomized studies as subheadings. Limitations to the following study design were made: only prognostic randomized controlled studies, other controlled studies, and meta-analyses were used. Only studies published in English or with an abstract in English, German, or French were included. The resulting studies were examined and sorted in three phases. Phase 1 sorted via lists and article summaries. The two authors (independently of each other) studied all of the abstracts from search. Relevant abstracts went on to phase 2, which involved the examination of full-text articles. Phase 3 was a critical examination of the selected studies.26
Systematic and Critical Evaluation of Available Literature
The evaluation was carried out for each study regarding its scientific quality and validity of the results by type of study according to a three-stage scale: high (++), moderate (+), and low (−). Quality/validity (QV score) assessment included the use of well-known checklists27,28 (Table 2).
Evaluation of Quality and Validity
The review of the literature has estimated whether the statistical analyses have been carried out adequately under the circumstances of the study aim, design, and number of patients. The possibility that the demands of the analyses have changed during the time period of investigation has been taken into account. In most of the articles included for critical evaluation of clinical efficacy, the relevance was usable for the main analyses. The method of randomization, balance of prognostic factors, and completeness of follow-up; intervention (treatment procedure) and outcome measures used in the study; intention-to-treat analysis; and whether the study helps to answer our key questions were also investigated.10 The following statistical analyses have been used in the articles being reviewed: multivariate analyses (mainly Cox regression analyses), Fisher's exact test, χ2 test, Kaplan-Meier survival curves, and log-rank test. Statistical power was calculated for all randomized controlled studies; generally studies with approximately 80% power to unveil a two-sided 15% difference in observed survival were assessed as having acceptable (ie, high or moderate) quality. The level of significance was set at P < .05 and all tests were two tailed. Primary end points reporting clinical effectiveness relevant for this systematic review included OS and DFS. The OS and DFS were expressed as hazard ratios (HRs). Types of bases for practical recommendations were adapted from a European state-of-the-art online instrument for clinical oncologists30 (Table 3). Earlier systematic reviews in EOC have been done by Högberg et al,31 Winter-Roach et al,10 and the Foundation for Scientific and Industrial Research (SINTEF) report.26
Types of Bases for Practical Recommendations
RESULTS
In the past, many randomized trials have enrolled patients with EOC to test the value of adjuvant therapies (eg, external radiotherapy, intraperitoneal installation of radionuclides such as gold-198 or phosphorus-32 (32P), single alkylating agents, or platinum-based single or combination chemotherapy) in preventing recurrence. This review is an update of earlier reviews10,26,31 of the literature of trials of such therapies for EOC. We found 22 randomized studies.3,32-52 However, nine of these studies33-35,37-40,42,45 were of low quality because they had methodologic flaws such as the omission of a control arm, inclusion of borderline tumors, incomplete surgical staging, or the inclusion of patients with stage II and III disease with minimal residual disease. Most of these trials included too few patients for conclusive results and will not be addressed.26-28 No disagreements between the two authors were found.
Is the Watch-and-Wait Approach Adequate After Comprehensive Surgical Staging of Low-Risk EOC?
The comparison of survival results in nonrandomized studies using different types of AC and a control arm is difficult (level of type 3 evidence, type R basis).30 Most prior randomized trials compared two or three different treatment modalities, and nearly all have had very low power because of the small number of patients included and too few events. However, without an untreated observation group, the efficacy of any AC cannot be firmly established. Only two randomized prospective studies including a control arm have been published for low-risk EOC; both were based on a type 2 level of evidence.3,48 QV scores are listed in Table 4. In the first study by the GOG,3 81 patients with low-risk disease (grade 1 or 2 tumors; FIGO 1973 stage IA or IB) were randomly assigned to receive 12 cycles of orally administered adjuvant melphalan therapy and 81 patients received no treatment. Unfortunately, at subsequent central pathology review, 30% of the patients were found to have borderline tumors. After a median follow-up period exceeding 6 years, no significant difference in OS (94% v 98%) or DFS (91% v 98%) was found. One patient in the melphalan arm died as a result of aplastic anemia. A study by the Gruppo Italiano Colloborative Oncologics Ginecologica/Gruppo Interregionale Cooperativo di Oncologia Ginecologica48 including low-risk disease (Table 4) showed a significant DFS advantage in the cisplatin group (83%; 41 patients) compared with the untreated group (65%; 44 patients; P = .03). However, when the controls were treated with cisplatin at relapse, they had the same overall 5-year survival as the group receiving AC cisplatin treatment (82% and 88%, respectively). These results have been confirmed with a median follow-up of more than 120 months.53 This suggests, however, that eight of 10 women in the cisplatin-treated group had been overtreated. If survival after relapse is compared, the patients in the up-front cisplatin group did much worse than patients from the nontreated group. The authors suggested that salvage treatment was more effective in the observation arm than in the AC arm, but also in this trial, patient numbers were too small for reliable conclusions.54
Adjuvant Chemotherapy Versus No Additional Treatment
Is High-Risk EOC Overtreated Surgically and Systemically?
The outcome of high-risk patients has not been favorable. In the literature, 30% to 50% of these patients had a recurrence within 5 years after their primary surgery.7,8,55 Three large prospective European trials have been conducted to determine whether AC in high-risk patients would prolong DFS and OS.32,49,50 QV scores are listed in Table 4. Between 1992 and 1997, the Nordic Cooperative Ovarian Cancer Group32 performed a randomized study in patients with high-risk epithelial ovarian cancer (stage I) comparing carboplatin and observation (control group) on a type 2 level of evidence (Tables 4 and 5). The study was closed prematurely due to poor accrual. Of 230 patients who underwent surgery, 162 were eligible (68 patients refused to participate in the observation arm [nontreatment]). High risk was defined as grade 2 or 3 tumor, all clear cell tumors, and all DNA aneuploid tumors, independent of grade. Patients were randomly assigned to observation (81 patients) or to AC carboplatin (area under the time-concentration curve = 7) for six cycles. With a median follow-up of 60 months, recurrence was registered in 46 patients: 25 in AC group and 21 in the control group. The estimated 5-year OS and DFS rates were 86% v 85% and 70% v 71% for the AC and control groups, respectively (not significant). Grade of differentiation, ploidy, and FIGO substages were the three independent prognostic factors.
Meta-Analysis of Four Randomized Trials: Adjuvant Cisplatin-Based Chemotherapy Versus NAC in Early Ovarian Cancer
When the patients in the control group experienced relapse, they were treated with carboplatin and showed a tendency to better OS; however, this finding was not statistically significant. The 5-year OS rate for the observation arm compared with the AC arm was 40% and 18%, respectively. This suggests that women in the AC group had been overtreated. However, the authors believed the results were inconclusive because of the small study size.
Between November 1990 and January 2000, EORTC/Adjuvant Chemotherapy in Ovarian Neoplasm (ACTION) collaborators, including 40 EORTC institutions from 10 countries, ran a prospective, unblinded, randomized phase III trial comparing platinum-containing AC versus observation after surgery in 448 patients with high-risk EOC on a type 1 level of evidence (QV scores are listed in Table 4).50 The study had strict guidelines for surgery. If all of the staging requirements were met, the staging performance was considered to be optimal.
Patients were centrally randomly assigned to either the AC arm or observation arm. Randomization was stratified according to institution, FIGO stage, and grade of differentiation (Tables 4 and 5). The primary end point was OS, whereas clinical recurrence was a secondary end point.
With a median follow-up of 5.5 years, the difference in OS between the two arms was not statistically significant (HR = 0.69; 95% CI, 0.44 to 1.08; P = .10; Table 5). DFS, however, was statistically significantly improved in the AC arm (HR = 0.63; 95% CI, 0.43 to 0.92; P = .02; Table 5).
Only one third of the patients (n = 151) were optimally staged. In patients randomly assigned to the observation arm, optimal staging was associated with a statistically significant improvement in OS and DFS (HR = 2.31; 95% CI, 1.08 to 4.96; P = .03 and HR = 1.82; 95% CI, 1.02 to 3.24; P = .04, respectively). This was not observed in the AC arm. In patients without optical staging (n = 297), a statistically significant improvement in OS and DFS was observed in the AC arm (HR = 1.75; 95% CI, 1.04 to 2.95; P = .03 and HR = 1.78; 95% CI, 1.15 to 2.77; P = .009, respectively). No benefit of AC was seen in the optimally staged patients, suggesting the significance of complete staging. The benefit of AC seemed to be limited to patients with nonoptimal staging (ie, patients with more risk of unappreciated residual disease). Within the ACTION trial, completeness of surgical staging, histologic type, and tumor grade were independent prognostic factors for both DFS and OS. In the suboptimally staged patients, however, AC did improve the outcome.50
The International Collaborative Ovarian Neoplasm 1 (ICON1) trial was performed between August 1991 and January 2000. A total of 477 EOC patients were entered onto the ICON1 trial from 84 centers in five countries (United Kingdom, Ireland, Brazil, Switzerland, and Italy). A total of 241 patients were randomly assigned to AC platinum-based chemotherapy, and 236 patients were assigned to observation on a type 1 level of evidence.49 There were no staging criteria. The only inclusion criterion was uncertainty of the attending physician whether AC was indicated (Tables 4 and 5). The QV scores are listed in Table 4.
The primary end points were similar to those in the ACTION trial. Women who received AC had better OS than women who did not (HR = 0.66; 95% CI, 0.45 to 0.97; P = .03). The 5-year OS was 9% better (79% for the AC arm compared with 70% for the nontreatment group; 95% CI, 1% to 15%). AC also improved DFS (HR = 0.65; 95% CI, 0.46 to 0.91; P = .01). There was an 11% difference (95% CI, 3% to 18%) in favor of the treatment arm of DFS (73% compared with 62%, respectively; Tables 4 and 5). At the Annual Meeting of the International Society of Gynecologic Cancer in 2006 (Santa Monica, CA), the OS follow-up was presented: the long-term analysis of ICON1, with a median follow-up of 9 years, showed the survival benefit in favor of AC did not reach statistical significance (71% v 64%; P = .1; M. Swart, personal communication, March 2007).
Together, ACTION and ICON1 randomly assigned a total of 925 EOC patients ICON1, n = 477; ACTION, n = 448). A total of 465 patients were randomly assigned to platinum-based AC and 460 patients were randomly assigned to observation on a type 1 level of evidence.49,50,56 OS at 5 years was 82% in the AC arm and 74% in the observation arm (8% difference; 95% CI, 2% to 12%; HR = 0.67; 95% CI, 0.50 to 0.90; P = .008). DFS at 5 years was also better in the AC arm compared with the observation arm (76% v 65%; 11% difference; 95% CI, 5% to 16%; HR = 0.64; 95% CI, 0.50 to 0.82; P = .001). To our knowledge, this is the first evidence that indicates AC treatment is significantly better than treatment at relapse. Subgroup analyses (pretreatment age, tumor stage, histologic cell type, and differentiation of grade) provided no evidence of a difference in the size of effect of AC on survival in any pretreatment subcategory. However, it is interesting that in well-differentiated and stage IA tumors, there seemed to be a benefit of AC (but they were not completely staged).
Radiotherapy and AC Versus Radiotherapy Alone
The lack of effective control with pelvic radiation therapy led to studies of whole abdominal radiation (WAR). Six randomized trials with a type 2 level of evidence comparing WAR versus noncisplatinum chemotherapy as AC for EOC have been reported in the literature.34,35,41,42,44,46 Three of the reports34,35,42 were of low quality (QV−) and three of the reports41,44,46 were of moderate quality (QV+). All of the described studies have been criticized because of incomplete surgical staging, unusual patient classification, too few patients included, and imbalance of risk factors between randomization arms. No significant differences between WAR and AC were found; therefore, WAR should not be recommended as AC in EOC. The long-term complications were also more severe after WAR than after AC. Although the article was of low quality, the study by Smith et al34 had a great impact, given that most institutions in the United States abandoned postoperative external radiotherapy in favor of AC.34
AC Versus Radiotherapy
Six studies with moderate or high quality (which included 1,067 women) with a type 2 level of evidence have compared adjuvant radiotherapy versus AC (QV scores are listed in Table 6). Four of the studies used intraperitoneal 32P as adjuvant treatment,3,43,48,52 whereas in the other two studies,46,47 WAR therapy was used. Cisplatin-based AC was used in five of the trials and melphalan was used in one trial (Table 6). It is not possible from the collected documentation to conclude whether AC has any effect or whether AC is more effective than adjuvant radiotherapy. However, the conclusion of these studies from the United States, Italy, and Norway was that cisplatin-containing regimens are preferable to radiotherapy or intraperitoneal 32P as AC for EOC patients because of lower toxicity and relative ease of administration.
Adjuvant Chemotherapy Versus Radiotherapy
Duration of Chemotherapy
Between March 1995 and May 1998, 457 EOC patients were included in a randomized phase III trial (GOG157) of three versus six cycles of AC with CP from GOG15751 and a type 1 level of evidence (the QV score was high). A total of 427 (93%) patients were considered eligible after centralized pathologic and medical review. Surgical staging was to be performed according to the GOG standards, which are similar to EORTC and NRH standards (Table 1). Of the 213 eligible patients in the three-cycle arm, 155 patients (73%) had complete surgery, and of the 214 patients in the six-cycle arm, 146 patients (68%) had complete surgery; a total of 299 (70%) of 427 patients had been assigned the correct surgical stage and 69% had stage I disease. The proportion of patients completing the three-cycle regimen was 96% compared with 83% for the six-cycle regimen. The toxicities from treatment were statistically significantly higher in the six-cycle arm (neurotoxicity, granulocytopenia, and anemia). The recurrence rate after six cycles was 24% lower (HR = 0.761; 95% CI, 0.51 to 1.13; P = .18) than that for three cycles, and the estimated probability of recurrence within 5 years was 20.1% (six cycles) versus 25.4% (three cycles; not significant). The overall death rate was similar for these regimens (HR = 1.02; 95% CI, 0.662 to 1.57). The conclusion from this trial was that three cycles compared with six cycles of CP do not significantly alter the recurrence rate in high-risk EOC, but six cycles are associated with more toxicity. An unexpectedly large number of patients (n = 126; 30%) had incomplete or inadequately documented surgical staging. In the subset of patients for whom complete surgical staging was documented, the cumulative incidence of recurrence within 5 years was 23% and 20%, respectively, for three-cycle and six-cycle regimens. The cumulative incidence of recurrence within 5 years was 22% for those who did have a documented complete staging procedure and 26% for those who did not (not significant).
Meta-Analyses
Meta-analyses have been performed by Winter-Roach et al10 for those trials with complete surgical staging procedures comparing AC versus radiotherapy (Appendix Table A1, online only). The meta-analyses showed no significant difference between the effects of AC and radiotherapy on OS or DFS. The main analysis of OS showed an HR of 0.85 (95% CI, 0.62 to 1.17) and the DFS showed an HR of 0.94 (95% CI, 0.56 to 1.59). In the subgroups comparing AC versus WAR, AC versus 32P, or platinum-based AC versus 32P, radiotherapy showed no statistical advantage for any modality. Winter-Roach et al10 and ICON1 and ACTION collaborative groups56 have performed separate meta-analyses in the patients randomly assigned between AC and no treatment. Three fourths of the patients came from the two trials ICON149 and ACTION.50 Winter-Roach10 tested five studies (four with cisplatin-based AC and one with melphalan). The ICON1/ACTION group excluded the melphalan study and used data from four cisplatin-based studies (Table 5). Both meta-analyses showed a significant beneficial effect of AC on OS (Winter-Roach,10 HR = 0.71; 95% CI, 0.63 to 0.80 and ICON1/ACTION, HR = 0.72; 95% CI, 0.55 to 0.94)56 and on DFS (Winter-Roach,10 HR = 0.68; 95% CI, 0.54 to 0.79 and ICON1/ACTION, HR = 0.66; 95% CI, 0.53 to 0.83).56 There was no evidence of heterogeneity between the trials. However, it should be noted that these meta-analyses include a preponderance of inadequately staged patients.
DNA Ploidy As a New Prognostic Factor
New prognostic variables, especially DNA ploidy, have been shown to be of independent prognostic significance in EOC.32,57-63 In a study from the NRH, the prognostic value of DNA ploidy was analyzed in 219 stage I patients57: 51% had diploid tumors and 49% had aneuploid tumors. The 5-year DFS rate for the patients with diploid tumors was 90%, compared with 64% for patients with aneuploid tumors (P = .0001). The only way to confirm the prognostic significance of DNA ploidy is to test this variable in prospective, randomized trials of stage I patients. So far, the only such study including DNA ploidy was done by the Nordic Cooperative Ovarian Cancer Group.32 The 162 patients were analyzed by a multivariate analysis identifying grade of differentiation, DNA ploidy, and FIGO substage as the only independent prognostic factors. DNA ploidy adds objective independent prognostic information regarding DFS and OS of patients with EOC.57,61-63 A patient with a stage Ia, grade 1 to 2 differentiated DNA diploid tumor has 10-year DFS of 98.2%, compared with a patient with a DNA aneuploid grade 3 tumor or clear cell stage Ib/Ic tumor who has a 10-year DFS of only 16.4%.
DISCUSSION
There is agreement among gynecologic oncologists on a type C basis that all patients should be optimally staged according to the NRH, EORTC, and GOG criteria (Table 1). Zanetta et al18 reported a significant OS and DFS advantage for EOC patients who were operated on by a gynecologic oncologist compared with a nongynecologic oncologist. However, it is disappointing that only 10% to 30% of patients worldwide who are presumed to have EOC have complete surgical staging.18,51,64,65 This limited staging success indicates that even in the large gynecologic oncology centers optimal staging is not always performed.65 Our policy at NRH is that all patients with presumptive EOC should be fully staged. If full staging is not possible, due to medical contraindication or patient refusal, consideration of AC is reasonable in selected patients. However, for selected patients who were staged incompletely at the time of initial surgery, completion of the staging procedure with either laparoscopy or laparotomy is another reasonable approach, before a final decision can be made regarding the need for AC.
Restaging should be done only in patients with a high risk of recurrence to identify those patients who do not require postoperative AC. The laparoscopic procedure can be used safely for the restaging of apparent EOC by teams that are experienced in ovarian and advanced laparoscopic surgery.66 Green67 (among others) opposes this policy, and questions the restaging of nonoptimally staged patients because the small gain in OS is not worth the effort and risk. He believes that the differences between small macroscopic and microscopic diseases are overcome by AC.
No differences in OS and DFS between the observation arm and AC arm were found in the optimally staged patients enrolled onto the ACTION trial.50 This lack of survival benefit suggests that women who had nonoptimal staging benefit from AC. Optimally staged patients randomly assigned to observation were observed to have a significantly better OS and DFS than nonoptimally staged patients. Similar differences were not seen in the patients in the AC arm. However, final conclusions cannot be drawn from this subgroup analysis because of the small number of events.64
Therefore, Trimbos (personal communication, February 2007) performed a meta-analysis from studies listed in Table 5 for OS and compared AC versus controls in adequately and inadequately staged EOC (Figs 1 and 2). In the adequately staged EOC trials, the HR was 0.91 (95% CI, 0.51 to 1.61) and AC produced a better outcome (not statistically significant). In the inadequately staged EOC trials, the HR was 0.68 (95% CI, 0.52 to 0.89). These results show that AC produces significantly better results when the surgical staging was inadequate. However, Trimbos (personal communication, February 2007) has shown that the survival curves of two thirds of the inadequately staged patients in the observation arm of the ACTION trial and all patients in the observation arm of the ICON trial were superimposed. This leads to the conclusion that the ICON population can be characterized as a poorly staged group of patients. In the ACTION trial and in the GOG157 trial, 151 and 301 patients, respectively, were optimally staged. When the optimally staged group (one third of patients) from the ACTION trial is compared with the optimally staged group (70%) of patients in the GOG157 trial, the recurrence rate was similar (22%). These results are supported by the matches in frequency of grade of differentiation and cell type among the study populations (grade 3, 35% and 32%, and serous/clear cell, 50% and 53%, respectively, in the ACTION and GOG157 trials). When the OS rates for the one third of the ACTION trial group with complete surgical staging are compared with patients in the GOG157 trial after three and six postoperative cycles of CP, the survival curves were superimposed (Fig 3) . This result makes the effect of AC in adequately staged patients even more doubtful, given that no statistical significant effect of AC was shown in this group in the ACTION trial. Because of the results of the meta-analysis in the optimally staged patients and lack of a difference in OS between ACTION and GOG157, we believe many gynecologic oncologists may consider the use of AC as an overtreatment. Tropé et al32 and Bolis et al48 have shown that salvage treatment was more effective in the optimally staged observation arm than in the optimally staged AC arm. They suggested that salvage treatment should be postponed until the time of recurrence. The number of patients involved in their analysis was small, but it is of interest that the same differences in the effectiveness of salvage chemotherapy were found in the ACTION trial.56 Kolomainen et al25 have also shown that patients with stage I disease can be treated successfully with a cisplatin-based salvage chemotherapy regimen after a period of observation. Their results showed an 80% 10-year DFS rate for patients in the observation group.
Overall survival: meta-analysis of randomized controlled trials of adjuvant chemotherapy (AC) versus controls in inadequately staged early ovarian cancer. European Organisation for Research and Treatment of Cancer/Adjuvant Chemotherapy in Ovarian Neoplasm trials, two thirds of the patients (events in 23 of 148 patients with AC versus 37 of 147 with no adjuvant chemotherapy [NAC])50; International Collaborative Ovarian Neoplasm 1, Medical Research Council (MRC), all patients (events in 42 of 241 patients with AC versus 61 of 236 patients with NAC49; Tropé et al32, the Nordic Cooperative Ovarian Cancer Group (NOCOVA), all patients (events in nine of 81 patients with AC versus nine of 81 patients with NAC. HR, hazard ratio. (From JB Trimbos, personal communication, February 2007, with permission.)
Overall survival: meta-analysis of randomized controlled trials of adjuvant chemotherapy (AC) versus controls in adequately staged early ovarian cancer. European Organisation for Research and Treatment of Cancer/Adjuvant Chemotherapy in Ovarian Neoplasm trials, one third of patients (events in 10 of 76 patients with AC versus eight of 75 patients with no adjuvant chemotherapy [NAC])50; Bolis et al,48 all patients (events in eight of 42 patients with AC versus nine of 41 patients with NAC); Young et al,3 all patients (events in two of 43 patients with AC versus four of 38 patients with NAC. HR, hazard ratio. (From JB Trimbos, personal communication, February 2007, with permission.)
Overall survival in early ovarian cancer after complete surgical staging: only one third of patients in the Adjuvant Chemotherapy in Ovarian Neoplasm trial had complete surgery versus 70% in the Gynecologic Oncology Group (GOG) trial 157 with three and six postoperative cycles of carboplatin + paclitaxel (CP). (From JB Trimbos, personal communication, February 2007, with permission.)
The 5-year survival rate for properly staged patients with stage IA grade 1 tumors is good (93% to 95%). Thus it is generally accepted by gynecologic oncologists that AC is not necessary on a type C basis.64,65,68,69 Furthermore, young fertile women with true stage IA grade 1 disease can be treated with conservation of the contralateral ovary, thus preserving their fertility.24,25
We have evaluated the treatment outcome for each of the main adjuvant treatment modalities that have been evaluated in randomized clinical trials. Until the results from the ICON1 and ACTION prospective randomized trials were published, no studies in EOC had shown the effect of adjuvant treatment (using either radiotherapy or chemotherapy). However, all trials have been insufficient for reliable statistical analyses because too few patients were randomly assigned and insufficient surgical staging was performed. After a median follow-up of more than 4 years from the combined ICON1/ACTION studies, there was a significant advantage for the AC arm in terms of 5-year OS (82% v 74%; P = .008) and 5-year DFS interval (76% v 65%; P = .001) on a type 1 level of evidence.56 Two meta-analyses including five trials10 and four trials50 comparing no treatment versus cisplatin-based AC demonstrated an even better effect of AC on 5-year OS and 5-year DFS. The two meta-analysis including 1,234 patients would seem to be definitive proof of the benefit of platinum-based AC for all patients with high-risk EOC.
The main strength of ICON1/ACTION analysis is the large number of patients included; despite the inclusion criteria differences between the two trials, the outcomes of the end points were remarkably similar.65 Many medical oncologists may interpret the results of the meta-analysis as a basis for considering platinum-based AC as the standard treatment to the majority of high-risk EOC patients. Hence, all nonoptimally staged patients or patients with incomplete data have to be treated with platinum-based chemotherapy.10,50,65 However, there are considerable differences between the two trials; (eg, patient inclusion criteria, prognostic factors, chemotherapy drugs and doses, and type of surgery). There are several weak points in ICON1/ACTION studies. Neither of these two trials used centralized histopathology review of diagnostic type or tumor grade. The ACTION trial included only median- and high-risk EOC. ICON1, however, included 32% of low-risk stage IA grade 1 tumors, which should not be included in high-risk EOC studies. This increases the risk for inclusion of borderline tumors. The recruitment time in both studies was nearly 10 years. Another weak point is that only one sixth of patients in ICON1/ACTION were optimally staged. Therefore, we agree with Young,65 Vergote and Trimbos,64 and Colombo-Pecorelli68 that the proof is not definitive; we are still not convinced that AC improves OS in adequately staged patients in FIGO stage I.
The type of AC and number of cycles in high-risk EOC have been debated for a long time. In this review, only the GOG15751 study has used the standard combination of CP for advanced ovarian cancer. They concluded that after complete surgical staging, three cycles of CP is a reasonable treatment for women with high-risk, early-stage EOC. Three additional cycles (for a total of six cycles) of CP chemotherapy are likely to produce only a small reduction in the risk of recurrence but cause increased toxicity.
We do not believe that AC is indicated in the majority of stage I tumors that are adequately staged. In small and selective groups of very high risk patients, we consider the use of adjuvant CP. We believe it is appropriate to extrapolate from the experience in advanced ovarian cancer70,71 that the addition of CP is warranted in high-risk EOC. The NRH guidelines are listed in Table 7.
Guidelines for Treatment Outside Protocols of Stage I Epithelial Ovarian Cancer Patients Referred to the Norwegian Radium Hospital
Retrospective and prospective studies on DNA ploidy in stage I disease from our32,57,61 and other institutions60,62,63 have shown that DNA ploidy is the second most important independent prognostic factor after degree of differentiation. The conclusion from our studies was that DNA ploidy combined with histology grade and type allows separation of patients into risk groups. We propose that DNA ploidy should be included with degree of differentiation in the FIGO classification and used for classifying patients into risk groups (Table 7). Unfortunately, DNA ploidy has not been accepted worldwide, although flow cytometry or image cytometry is an easily accessible tool and should be used by the pathologists. From the largest retrospective study of stage I (n = 1,545)5 disease, the most important independent prognostic factors were degree of differentiation followed by rupture before surgery, rupture during surgery, FIGO substage (IB v IA), and patient age. The importance of grade and substage was confirmed by the ICON1/EORTC-ACTION trials showing that FIGO IB had similar survival as substage IC. In most studies clear cell tumors are not graded, but the Silverberg grading system should now be used routinely. If conventional/classic factors are used under conditions of adequate surgical staging, EOC patients can be classified into three new risk groups, as proposed by Vergote and Amant72: low risk (I grade 1), medium risk (IB or IC grade 1, IA grade 2), and high risk (IB or IC grade 2 or 3, and clear cell). Recently, from data of 600 surgically staged patients with FIGO stage I EOC treated in 11 centers in Australia, United States, and Europe in a multivariate analysis, Obermair et al73 showed that preoperative CA-125 ≤ 30 U/mL was an independent prognostic factor for overall survival.73 Obermair et al showed that EOC patients with a preoperative value of CA-125 ≤ 30 U/mL had extremely good survival and might not require AC.
The results of this review of EOC may lead to many controversies. It is often said that the ICON1 trial is a trial for the real world.65,67 Guthrie,74 on the behalf of the ICON1 collaborators, believes that all patients with EOC are likely to benefit from AC. We believe that this real-world trial suggests overtreatment of the majority of the EOC patients; for example (adapted from the SINTEF report26), if a study shows a 5-year survival of 87% in the adjuvant-treated group compared with 80% in the nontreated control group, this means that of 100 treated patients only seven experienced benefit from the treatments. Despite adjuvant treatment, 13 patients died as a result of treatment-related toxicity and their disease. Furthermore, for 80 patients the treatment was unnecessary. These patients were exposed to acute and possibly long-term toxicity. The time period was extended unnecessarily, including the treatment and the notification of illness; at the same time, patients had to focus on their disease.
In a Norwegian study75 including 70 (54%) stage I patients, 22% had chronic fatigue. Five-year survivors of EOC had more somatic and mental morbidity, more fatigue, poorer quality of life, and used more medication and health services than controls (compared with an age-representative sample of the general female population).
The relative risk for developing secondary malignancies is increased from 3.3 to 6.5 after platinum therapy, depending on dose.76 This supports our belief that the use of AC should be reduced for patients who are likely to be long-term survivors.
We agree with Winter-Roach,10 Vergote and Trimbos,64 Young,65 and Vergote and Amant,72 who indicate that the focus of next trials should be to identify the patients who do not need AC and search for improved treatment for the patients who do need AC. Many institutions, including our own, use different molecular markers such as gene expression, microarray profiles,77 and serum protein78 in an attempt to improve methods to separate patients into risk groups.
Only a future randomized trial with optimal surgical staging versus AC can provide definitive conclusions. However, such a study does not seem feasible because of the small number of recurrences among patients with optimal surgical staging. The same is probably true regarding the introduction of new cytostatic drugs or molecular-targeted drugs into trials of EOC. Such studies will not be large enough to show any statistically significant improved survival.
The GOG is currently running a prospective randomized trial comparing three cycles of AC with CP followed by either weekly paclitaxel for 6 months or observation.51 It seems more interesting to include high-risk EOC in the phase III randomized studies in advanced ovarian cancer; therefore, our institution decided to participate in ICON7, which is a randomized controlled phase III trial in patients with high-risk, early-stage (FIGO stage I or IIA clear cell or grade 3 carcinoma) or advanced-stage (FIGO stage IIB or higher, all grades and all histologic subtypes) epithelial ovarian carcinoma, primary peritoneal carcinoma, or fallopian tube carcinoma, to evaluate the addition of bevacizumab to standard chemotherapy with CP. Primary aims are to determine whether the addition of bevacizumab to standard chemotherapy improves DFS and OS compared with standard chemotherapy alone.
AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
The author(s) indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS
Conception and design: Claes Tropé
Provision of study materials or patients: Claes Tropé, Janne Kaern
Collection and assembly of data: Claes Tropé, Janne Kaern
Data analysis and interpretation: Claes Tropé, Janne Kaern
Manuscript writing: Claes Tropé, Janne Kaern
Final approval of manuscript: Claes Tropé, Janne Kaern
Appendix
Meta-Analysis Results From Six Prospective Randomized Studies Comparing Chemotherapy Versus Radiotherapy
Acknowledgments
We thank the scientific secretary, Gry Seppola, for technical and organizational help.
Footnotes
-
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
- Received February 1, 2007.
- Accepted April 23, 2007.
