Kim, Hee Seung MD*; Ju, Woong MD, PhD†; Jee, Byung Chul MD, PhD‡; Kim, Yong Beom MD, PhD‡; Park, Noh Hyun MD, PhD*; Song, Yong Sang MD, PhD*; Kim, Seung Cheol MD, PhD†; Kang, Soon-Beom MD, PhD*; Kim, Jae Weon MD, PhD*
Ovarian cancer is the second most common malignancy of female genital tract with 204,200 new cases and 124,700 deaths in the world.1 Approximately 90% of ovarian cancers consists of epithelial ovarian cancer (EOC) derived from tissues that come from the coelomic epithelium or mesothelium. Epithelial ovarian cancer progresses into its advanced-stage disease in more than two-thirds of all patients because of a lack of related symptoms and no effective screening tools in early-stage disease.2
The primary standard treatment is made up of staging laparotomy including maximal cytoreduction and adjuvant taxane- and platinum-based chemotherapies.3 In particular, pelvic or para-aortic lymph node metastasis identified by surgery is known to be more frequent in EOC than in other gynecologic cancers, and approximately 20% of patients with disease within the pelvis have been shown to be classified as stage IIIC disease according to the International Federation of Gynecology and Obstetrics (FIGO) due to pelvic or para-aortic lymph node metastasis.4
Although the association between pelvic or para-aortic lymph node metastasis and poor prognosis has been established,5 there is no consensus among gynecologic oncologists about the efficacy of systematic lymphadenectomy (SL) for improved overall survival (OS).6,7 In particular, 2 randomized controlled trials (RCTs) have not shown the significance of SL for OS,8,9 whereas some observational studies have suggested that SL may be a favorable factor for OS in EOC.6,7,10
Therefore, the current study was designed to evaluate the role of SL in EOC using a meta-analysis of relevant studies that compared the efficacy for OS between SL and unsystematic lymphadenectomy (USL).
MATERIALS AND METHODS
For this meta-analysis, a literature search of the National Library of Medicine and National Institutes of Health (PubMed), EMBASE, and Cochrane Controlled Trials Register (CENTRAL) electronic database was performed independently by 2 reviewers. The literature search was limited to the period of January 1, 1995, through December 31, 2008. We also scanned bibliographies of relevant articles to identify additional studies.
We performed the computerized literature search using the free text search terms "ovarian cancer," "ovarian neoplasm," "ovarian tumor," "ovarian carcinoma," "lymph node," "lymph node sampling," "lymph node dissection," and "lymphadenectomy" for the outcome factors. All terms were expanded to include all subcategories in an attempt to obtain all published research that fit the selection criteria. No financial conflict of interest existed with any commercial entity whose products are described, reviewed, evaluated, or compared in the current study.
To be included in the meta-analysis, retrieved studies had to fulfill the following inclusion criteria: (1) EOC and (2) comparison of OS between SL and USL. Exclusion criteria included (1) non-EOC such as germ cell tumor, fallopian tubal cancer, and primary peritoneal carcinomatosis; (2) studies where the comparison of OS was not performed between SL and USL; and (3) non-English literature because of lack of accessibility and reading. All resulting citation abstracts were reviewed for potential eligibility, and the full article texts were obtained for further evaluation in cases in which abstracts did not provide enough details for the determination of eligibility.
A total of 1338 potentially relevant studies were identified based on the above search terms. All of the studies retrieved from the databases were independently evaluated. After screening the titles and abstracts, 814 studies were excluded owing to nonovarian diseases (n = 402), preclinical studies (n = 172), review articles (n = 130), and case reports (n = 110). Further assessment for more detailed information identified 308 ineligible studies associated with no comparison between SL and USL (n = 223) and non-EOC (n = 85). After we reviewed full articles of the remaining studies, 7 studies were excluded because of insufficient data for survival (n = 5) and the absence of full length articles (n = 2). Finally, 2 RCTs and 7 observational studies were scrutinized in full text as appropriate (Fig. 1).6-14
Extraction of Data
The following data were independently abstracted for the current study: first author, year of publication, design of study, clinical stage, histology, debulking surgery, number of patients, definition of SL or USL, the observed number of events on the experimental intervention (O), the log-rank expected number of events on the experimental intervention (E), and the variance of the log-rank statistic (V).
Two reviewers compared the results of the abstraction from all 9 studies for accuracy and came to an agreement on any discrepancies. In 2 studies with disagreement, a third reviewer served as the tiebreaker.
The aim of the meta-analysis was to compare OS between patients treated with SL and those treated with USL. Conducting a meta-analysis using summary information from eligible studies is often problematic because the most appropriate summary statistics are typically not presented. Among some approaches for resolving this problem, the method by Tierney et al15 was used in this meta-analysis. We estimated O-E and 5 from Kaplan-Meier OS curves to calculate hazard ratio (HR) and 95% confidence interval (CI).
Heterogeneity was assessed using Higgins I2, which measures the percentage of the total variation across studies that is due to heterogeneity rather than chance.16 I2 is evaluated as follows: I2 = (Q − df)/Q × 100%, where Q is Cochran heterogeneity statistic and df is its degrees of freedom. The value of I2 ranges from 0% (no observed heterogeneity) to 100% (maximal heterogeneity). I2 > 50% may be considered to represent substantial heterogeneity.
For identifying publication bias, a funnel plot was represented, which is a scatter plot of HRs of individuals studies on the x axis against the SE of log HR of each study on the y axis. If there is no publication bias, HRs of small-scale studies scatter widely at the bottom of the graph, with the spread narrowing among large-scale studies. The funnel plot resembles a symmetrical inverter funnel in the absence of publication bias, whereas publication bias makes the funnel plot asymmetrical.
This meta-analysis was performed using Review Manager Version 5.0 (the Nordic Cochrane Centre, Copenhagen, Denmark), and we used only the fixed-effect meta-analysis method because it is the only method available in Review Manager Version 5.0 for 'O-E and V' outcomes. P < 0.05 was considered statistically significant.
The clinical characteristics of a total of 21,919 patients from these 9 studies are depicted in Table 1. Among them, 3 studies have demonstrated the role of SL in FIGO stage I-II EOC,7,8,14 whereas 5 studies and 1 subanalysis have shown it in FIGO stage III-IV EOC.6,9-13 Systematic lymphadenectomy was defined as follows: (1) pelvic and para-aortic SL10-14; (2) more than 11 resected pelvic and para-aortic lymph nodes6,7; (3) pelvic (≥20) and para-aortic (≥15) resected lymph nodes8; and (4) pelvic (≥25) and para-aortic (≥15) resected lymph nodes.9 On the other hand, USL was defined as follows: (1) not performed10-13; (2) 10 or less resected pelvic and para-aortic lymph nodes6,7; (3) random removal of pelvic and para-aortic lymph nodes8; (4) removal of all macroscopic (≥1 cm) lymph nodes9; and (5) lymph node exploration or sampling.14
This meta-analysis using a fixed-effect model demonstrated that SL improved OS compared with USL in all-stage disease (HR, 0.72; 95% CI, 0.68-0.76). After we excluded 2 Surveillance, Epidemiology and End Results (SEER) studies owing to the deviation of weight,6,7 SL also showed increased OS when compared with USL in all-stage disease (HR, 0.84; 95% CI, 0.72-0.98; Fig. 2). When we performed this meta-analysis according to the study design, 2 RCTs demonstrated that there was no difference in OS between SL and USL (HR, 0.86; 95% CI, 0.70-1.07). On the other hand, 7 observational studies showed that SL improved OS compared with USL (HR, 0.71; 95% CI, 0.67-0.75). However, there was no difference in OS between SL and USL after 2 SEER studies were excluded (HR, 0.81; 95% CI, 0.64-1.03; Fig. 3).6,7
In 3 studies in which only 7158 patients with FIGO stage I-II disease were included, SL improved OS (HR, 0.80; 95% CI, 0.70-0.92). Two observational studies demonstrated that SL increased OS in FIGO stage I-II disease (HR, 0.81; 95% CI, 0.70-0.93), whereas 2 studies except for 1 SEER study showed no difference in OS between SL and USL (HR, 0.85; 95% CI, 0.62-1.16; Fig. 4).7
On the other hand, 6 studies in which only 14,700 patients with FIGO stage III-IV disease were enrolled showed that SL increased OS (HR, 0.70; 95% CI, 0.67-0.75), and 5 studies except for 1 study from SEER also demonstrated that SL improved OS (HR, 0.83; 95% CI, 0.69-0.99).6 Furthermore, SL was a significant factor for improved OS in 5 observational (HR, 0.70; 95% CI, 0.66-0.74) and 4 observational studies except for 1 SEER study (HR, 0.76; 95% CI, 0.59-0.97; Fig. 5).6
Among all studies, 4 included only 997 patients who had undergone optimal debulking surgery. Optimal debulking surgery was defined as follows: (1) the reduction of all tumor lesions to 1 cm or less in diameter8-10 and (2) no residual tumor after surgery.14 Systematic lymphadenectomy was also a significant factor for improved OS in patients with all-stage disease who underwent optimal debulking surgery (HR, 0.83; 95% CI, 0.69-0.99). On the other hand, there was no difference in OS between SL and USL in FIGO stage I-II (HR, 0.85; 95% CI, 0.62-1.16) and III-IV diseases (HR, 0.82; 95% CI, 0.62-1.02; Fig. 6).
Heterogeneity and Publication Bias
Tests for heterogeneity demonstrated that there was no significant between-study variation (I2 = 0%-43%). Furthermore, the funnel plot for 9 eligible studies in this meta-analysis revealed that all studies were distributed evenly across the graph, suggesting no publication bias in the meta-analysis (Fig. 7).
The aim of the current study was to investigate the efficacy of SL for OS in EOC using a meta-analysis, and all 9 studies showed that SL was efficient for improving OS in all-stage disease. Although 7 observational studies showed increased OS by SL in all-stage disease (HR, 0.71; 95% CI, 0.67-0.75), 2 RCTs and 5 observational studies except for 2 SEER studies demonstrated that there was no difference in OS between SL and USL.
These controversial findings were caused by major limitations such as lack of RCTs, the deviation of weight by 2 large-scale SEER studies.6,7 Only 2 RCTs have been published up to now,8,9 and 2 observational SEER studies had weighted with the results of this meta-analysis.6,7 As a result, 2 RCTs, which have shown that SL may not improve OS in EOC, have not affected the results of this meta-analysis. However, the meta-analytic results using 2 RCTs and 5 observational studies except for 2 SEER studies showed the possibility that SL improved OS with marginal significance (HR, 0.86 and 0.81; 95% CI, 0.70-1.07 and 0.64-1.03), and the results from SEER should also be considered as important because of the large numbers of patients with EOC.
When we performed subanalyses according to disease status, the efficacy of SL on increased OS could not be determined in FIGO stage I-II disease. Although 1 RCT showed no difference in OS between SL and USL (HR, 0.78; 95% CI, 0.55-1.11),8 3 observational studies demonstrated increased OS by SL in FIGO stage I-II disease (HR, 0.80; 95% CI, 0.70-0.92). On the other hand, SL was a significant factor for increased OS in FIGO stage III-IV disease (HR, 0.70; 95% CI, 0.67-0.75). Although 1 RCT demonstrated no difference of OS between SL and USL (HR, 0.91; 95% CI, 0.70-1.19),9 this meta-analysis showed that SL might be important for improving OS in FIGO stage III-IV disease. These findings suggest the possibility that the efficacy of SL may be limited for OS in early-stage EOC, whereas SL may be efficient to increase OS in advanced-stage EOC.
Previous studies have shown that SL may be important for detecting occult lymph node metastasis. In 2 RCTs, patients treated with SL had a higher rate of positive lymph nodes at histological examination than those who underwent the removal of macroscopic lymph nodes (22% vs 9% in clinically early-stage disease; 70% vs 42% in clinically advanced-stage disease),8,9 suggesting the possibility that there may be more opportunity to detect occult lymph node metastasis by SL in advanced-stage than in early-stage EOC. This hypothesis is supported by a previous study where the percentage of lymph node metastasis has been reported to be 5% to 24% in clinically early-stage disease and 48% to 75% in peritoneally advanced-stage disease.17
Furthermore, occult lymph node metastasis could lead to chemoresistance if it would not be removed. The "pharmacologic sanctuary hypothesis" for poor prognosis in EOC suggests that lymph node metastasis may be resistant to chemotherapy because of diminished blood supply,18 suggesting that SL may be a favorable factor for prolonged OS in clinically advanced-stage EOC where patients have more chance to have occult lymph node metastasis.
Nevertheless, occult lymph node metastasis should also be considered as a risk factor for poor OS in patients with clinically early-stage EOC because up to 30% of the patients have been reported to be upstaged during the restaging procedure with lymphadenectomy.19 Thus, occult lymph node metastasis led to controversial results about the efficacy of SL on OS despite fewer opportunity to detect occult lymph node metastasis in clinically early-stage EOC, and more relevant RCTs will clarify the role of SL in future. Besides, the definitions of SL and USL were different among all studies in this meta-analysis. Because the number of resected lymph nodes has been reported to be associated with prognosis in EOC,6 the various extent of SL can be a bias for evaluating the efficacy of SL in EOC using a meta-analysis.
On the other hand, SL was a significant factor for OS in patients with all-stage disease who underwent optimal debulking surgery (HR, 0.83; 95% CI, 0.69-0.99). When we performed subanalyses according to disease status, SL could not improve OS statistically in FIGO stage I-II and III-IV diseases. However, SL increased OS with marginal significance in FIGO stage III-IV disease (HR, 0.82; 95% CI, 0.66-1.02), suggesting the possibility that SL may be more important to improve OS in patients with advanced-stage disease who underwent optimal debulking surgery than in the patients with early-stage diseases. Because patients with FIGO stage IIIC EOC due to positive lymph nodes have only been shown to have a more favorable prognosis compared with other stage IIIC patients who underwent optimal cytoreduction,20 SL for removing occult lymph node metastasis can increase OS in 20% of stage IIIC patients who had disease within the pelvis with pelvic or para-aortic lymph node metastasis when compared with other FIGO stage IIIC patients.4
Although this meta-analysis suggests the possibility that SL may be efficient for improving OS in patients with advanced-stage EOC because SL may contribute to the detection of occult lymph node metastasis with chemoresistance, the efficacy of SL on OS is still unknown because of the lack of related RCTs. For this purpose, the Arbeitsgemeinschaft Gynaekologische Onkologie (AGO) ovarian cancer study group is preparing a prospective randomized multicenter trial, titled the Lymphadenectomy In Ovarian Neoplasms trial, for evaluating the efficacy of pelvic or para-aortic SL in patients with FIGO stage IIB-IV disease who had no clinically obvious or enlarged lymph nodes (AGO-OVAR OP. 3). Until the results from more RCTs are in fact available, these findings should be interpreted carefully considering the limitation in this meta-analysis, and patients with EOC should be informed in detail about the pros and cons of SL.
The authors thank the Medical Research Collaborating Center in Seoul National University Hospital for statistical analysis. Moreover, the authors thank the committee of the 41st Annual Meeting of the Society of Gynecologic Oncologists for giving them the opportunity to present their abstract as a poster.
1. Sankaranarayanan R, Ferlay J. Worldwide burden of gynaecological cancer: the size of the problem. Best Pract Res Clin Obstet Gynaecol. 2006;20:207-225.
2. Heintz AP, Odicino F, Maisonneuve P, et al. Carcinoma of the ovary. FIGO 6th Annual Report on the Results of Treatment in Gynecological Cancer. Int J Gynaecol Obstet. 2006;95(Suppl 1):S161-S192.
3. Bookman MA, Greer BE, Ozols RF. Optimal therapy of advanced ovarian cancer: carboplatin and paclitaxel vs. cisplatin and paclitaxel (GOG 158) and an update on GOG0 182-ICON5. Int J Gynecol Cancer. 2003;13:735-740.
4. Onda T, Yoshikawa H, Yokota H, et al. Assessment of metastases to aortic and pelvic lymph nodes in epithelial ovarian carcinoma. A proposal for essential sites for lymph node biopsy. Cancer. 1996;78:803-808.
5. Takeshima N, Hirai Y, Umayahara K, et al. Lymph node metastasis in ovarian cancer: difference between serous and non-serous primary tumors. Gynecol Oncol. 2005;99:427-431.
6. Chan JK, Urban R, Hu JM, et al. The potential therapeutic role of lymph node resection in epithelial ovarian cancer: a study of 13918 patients. Br J Cancer. 2007;96:1817-1822.
7. Chan JK, Munro EG, Cheung MK, et al. Association of lymphadenectomy and survival in stage I ovarian cancer patients. Obstet Gynecol. 2007;109:12-19.
8. Maggioni A, Benedetti Panici P, Dell'Anna T, et al. Randomized study of systematic lymphadenectomy in patients with epithelial ovarian cancer macroscopically confined to the pelvis. Br J Cancer. 2006;95:699-704.
9. Panici PB, Maggioni A, Hacker N, et al. Systematic aortic and pelvic lymphadenectomy versus resection of bulky nodes only in optimally debulked advanced ovarian cancer: a randomized clinical trial. J Natl Cancer Inst. 2005;97:560-566.
10. Isonishi S, Niimi S, Sasaki H, et al. Drug sensitivity-related benefit of systematic lymphadenectomy during cytoreductive surgery in optimally debulked stages IIIc and IV ovarian cancer. Gynecol Oncol. 2004;93:647-652.
11. Kikkawa F, Ishikawa H, Tamakoshi K, et al. Prognostic evaluation of lymphadenectomy for epithelial ovarian cancer. J Surg Oncol. 1995;60:227-231.
12. Allen DG, Coulter J. Survival of patients with epithelial ovarian cancer and the effect of lymphadenectomy in those with stage 3 disease. Aust N Z J Obstet Gynaecol. 1999;39:420-424.
13. Saygili U, Guclu S, Uslu T, et al. Does systematic lymphadenectomy have a benefit on survival of suboptimally debulked patients with stage III ovarian carcinoma? A DEGOG* Study. J Surg Oncol. 2002;81:132-137.
14. Suzuki S, Kajiyama H, Shibata K, et al. Is there any association between retroperitoneal lymphadenectomy and survival benefit in ovarian clear cell carcinoma patients? Ann Oncol. 2008;19:1284-1287.
15. Tierney JF, Stewart LA, Ghersi D, et al. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials. 2007;8:16.
16. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557-560.
17. Pereira A, Magrina JF, Rey V, et al. Pelvic and aortic lymph node metastasis in epithelial ovarian cancer. Gynecol Oncol. 2007;105:604-608.
18. Eisenkop SM, Spirtos NM. The clinical significance of occult macroscopically positive retroperitoneal nodes in patients with epithelial ovarian cancer. Gynecol Oncol. 2001;82:143-149.
19. Soper JT, Johnson P, Johnson V, et al. Comprehensive restaging laparotomy in women with apparent early ovarian carcinoma. Obstet Gynecol. 1992;80:949-953.
20. Baek SJ, Park JY, Kim DY, et al. Stage IIIC epithelial ovarian cancer classified solely by lymph node metastasis has a more favorable prognosis than other types of stage IIIC epithelial ovarian cancer. J Gynecol Oncol. 2008;19:223-228.