Objective
We sought to perform a quantitative analysis on operative outcomes of laparoscopic staging surgery in patients with presumed early-stage ovarian cancer using a metaanalysis.
Study Design
Electronic searches for studies of laparoscopic staging surgery in patients with ovarian cancer were performed within 3 electronic databases (Medline, Embase, and the Cochrane Library) using the key words “ovarian cancer,” “early stage,” “laparoscopy,” “staging surgery,” “staging laparoscopy,” and “recurrence.” Two authors independently screened articles, and those meeting the defined inclusion/exclusion criteria were included in the metaanalysis.
Results
We identified 11 observational studies. The combined results of 3 retrospective studies showed that the estimated blood loss in laparoscopy was significantly lower than that for laparotomy ( P < .001). The overall upstaging rate after laparoscopic surgery was 22.6% (95% confidence interval [CI], 18.1–27.9%) without significant heterogeneity among all study results. The overall incidence of conversion from laparoscopy to laparotomy was 3.7% (95% CI, 2.0–6.9%). The overall rate of recurrence in studies with a median follow-up period of ≥19 months was 9.9% (95% CI, 6.7–14.4%).
Conclusion
Through our quantitative analysis, we concluded that the operative outcomes of a laparoscopic approach in patients with early-stage ovarian cancer could be compatible with those of laparotomy. In the future, further randomized controlled trials may be needed.
The laparoscopic approach is currently applied to complicated surgeries in the field of gynecologic cancer and has been associated with quick recovery times, lower morbidity, and shorter hospital stays compared with laparotomy. However, in ovarian cancer, the resection area is broader than in other gynecologic cancers, so dissemination occurring from exfoliation of tumor cells and a larger risk of intraoperative tumor rupture remain limitations of laparoscopic staging surgery.
The traditional approach for staging of clinical early-stage ovarian cancer (EOC) is through laparotomy with an extended midline incision that exposes the whole peritoneal cavity. However, due to recent advances in laparoscopic techniques and instruments, it is possible to perform the standard staging procedure for ovarian cancer laparoscopically.
In a Cochrane systematic review of studies that compared the operative outcomes of laparoscopy and laparotomy carried out in patients with EOC through November 2007, a quantitative metaanalysis was impossible, and only a qualitative review could be conducted due to the low quantity of studies in the literature. Since then, many studies have applied laparoscopic staging surgery in patients with EOC. The purpose of our study was to perform a quantitative metaanalysis on operative outcomes of laparoscopic staging surgery in patients with presumed EOC using a metaanalysis of single-armed studies and the laparoscopic arms of comparative studies.
Materials and Methods
Literature search
A literature search was performed using the key words, “ovarian cancer,” “early stage,” “laparoscopy,” “staging surgery,” “staging laparoscopy,” and “recurrence” in Medline (from December 1969), Embase (from September 1974), and the Cochrane Library (from February 1990) for articles published through Aug. 24, 2012.
Participants included in this metaanalysis are as follows. Among cases with presumed EOC (International Federation of Gynecology and Obstetrics stage I-II) prior to surgery based on a baseline study, patients who either received laparoscopic staging surgery or who were referred to the department of gynecologic oncology for laparoscopic staging surgery after pathologic diagnosis of cancer following surgery for a benign-looking ovarian mass at an outside hospital were selected as participants.
Study selection
Inclusion/exclusion criteria of studies subject to our metaanalysis were as follows: (1) among manuscripts for which the original full text was found, only those specifically providing useful operative outcomes were included; (2) studies written in languages other than English were excluded; (3) abstracts, comments, reviews, and editorials were excluded; (4) case reports and case series with a sample size of ≤10 were excluded; (5) as for laparoscopic, full-staging procedures, studies that did not clearly mention the performance of lymphadenectomy or where <10 cases of lymphadenectomy were carried out were excluded; (6) studies that did not include invasive epithelial-origin carcinoma in patients were excluded; (7) studies in which laparoscopic surgery was performed for the purpose of diagnostic biopsy instead of radical treatment were excluded; and (8) the publication year, authors, study centers, and study periods were investigated, and overlapping articles were excluded. In cases of overlapping study populations, only the larger study was included in our analysis.
Data extraction
After analyzing each study, variables showing operative outcomes (“A” variables) and those with unique demographic characteristics of each study (“B” variables) were examined.
“A” variables are as follows: (1) operation time (mean ± SD, min); (2) estimated blood loss (EBL) (mean ± SD, mL); (3) perioperative complications (including intraoperative and postoperative complications associated with surgery) during the postoperative and follow-up periods; (4) upstaging rate after staging surgery; (5) rate of conversion to laparotomy; (6) rate of intraoperative tumor rupture; and (7) recurrence rate during the follow-up period after laparoscopic staging surgery. In the examination of “A” variables, data presented as a median value and a range were converted to a mean value and SD using the formula proposed by Hozo et al. The recurrence rate was investigated only in studies with follow-up periods of ≥19 months. This time length was chosen because a sufficient observation period must be required to evaluate recurrence. The median recurrence-free interval should be regarded as ≤19 months when referring to the literature on ovarian cancer.
“B” variables were as follows: (1) age (mean ± SD); (2) proportion of incomplete staging procedures at the initial surgery (the proportion of patients referred to the department of gynecologic oncology for laparoscopic staging surgery due to incomplete staging at the initial surgery); (3) proportion of patients with invasive epithelial-origin carcinoma; (4) conducting rate of adjuvant chemotherapy after staging surgery; (5) total harvested number of lymph nodes (which was classed as a “B” variable because lymphadenectomy can be performed in a number of different ways according to each surgeon’s individual protocol); and (6) proportion of fertility-sparing surgeries.
Studies were selected and data were extracted by 2 reviewers (H.J.P. and Y.T.K.), and any discrepancy between reviewers was resolved through discussion.
Data analysis
Data were analyzed using software (Comprehensive Meta-Analysis, version 2.0; Biostat, Englewood, NJ). To control for differences in study designs among studies, data provided by each study were divided into retrospective and prospective categories according to the method of data collection and then summed separately. In addition, among the retrospective studies, 3 comparative studies that contrasted the operative outcomes of laparoscopy with those of laparotomy were separated to express the outcomes as pooled odds ratios and 95% confidence intervals (CIs).
Study-to-study variation was assessed using the Higgins I 2 test, which measured the proportion of the total variation across studies. To provide an approximation of the overall rate of each “A” variable, the pooled rate of all studies was calculated as the weighted average rate by using the random effect model or the fixed effect model according to the result of the Higgins I 2 test. When significant heterogeneity (I 2 ≥50%) was not observed between the subgroups in the metaanalysis, the fixed effects model was used. When significant heterogeneity was observed, the random effects model was employed.
If I 2 ≥75%, data were considered to have considerable heterogeneity and could not be combined. If I 2 ≥50%, a metaregression analysis was performed to determine factors that had an influence on heterogeneity. Here, “B” variables showing different demographic characteristics according to each study were applied in our metaregression analysis. Outcomes were given as the exponentiated slope coefficient and 95% CIs. Variables with P < .05 were regarded as significant influential factors on heterogeneity.
A potential publication bias was estimated using Egger linear regression test and P < .05 was considered significant. When there was a substantial publication bias, a Duval and Tweedie nonparametric trim and fill procedure was also performed to assess the possible effects of the publication bias and to suggest the adjusted overall values.
Results
In this study, we ultimately enrolled 11 observational studies ( Figure 1 ). The characteristics of patients and the design of each study are provided in Table 1 .
| Study (period) | Total patients, mean age, y (SD) [range] | Median follow-up, mo [range] | Method of data collection | Diagnosis of disease stage | Fertility-sparing surgery, n/total (%) | Incomplete staging at initial surgery, a n/total (%) | Invasive epithelial carcinoma, n/total (%) | Conducting rate of AC, n/total (%) |
|---|---|---|---|---|---|---|---|---|
| Leblanc et al, 2004 (1991 through 2001) | n = 53, 41.3 (13.9) [18–63] | 54 [8–116] | Retrospective | Clinical b | 9/53 (17.0) | 53/53 (100) | 44/53 (83.0) | 19/53 (35.8) |
| Chi et al, 2005 (2000 through 2003) | n = 20, 47.3 (11.2) | Not reported | LSARC | Clinical b | Not reported | 13/20 (65.0) | 17/20 (85.0) | Not reported |
| Park, 2008 (2001 through 2006) | n = 17, 43.2 (12.3) | 19 [5–56] | LSARC | Clinical b | Not reported | 6/17 (35.3) | 17/17 (100.0) | 10/17 (58.8) |
| Park, 2008 (2004 through 2007) | n = 19, 43.9 (9.8) | 17 [2–40] | LSARC | Clinical b | 3/19 (15.8) | 7/19 (36.8) | 19/19 (100.0) | 15/19 (78.9) |
| Nezhat et al, 2009 (1995 through 2007) | n = 36, 47.8 [17–89] | 55.9 | Retrospective | Clinical b | 11/36 (30.6) | 9/36 (25.0) | 20/36 (55.6) | 10/36 (27.8) |
| Lee et al, 2011 (2005 through 2010) | n = 26, 42.2 (10.8) | 12 [1–42] | Retrospective | Clinical b | Not reported | 9/26 (34.6) | 22/26 (84.6) | 17/26 (65.4) |
| Schreuder et al, 2012 (2001 through 2009) | n = 25, 49.7 [18–79] | 43 [1–116] | Retrospective | Clinical b | Not reported | 24/25 (96.0) | 20/25 (80.0) | 14/25 (56.0) |
| Tozzi et al, 2004 (1996 through 2003) | n = 24, 36.8 [19–76] | 46.4 [2–72] | Prospective | Pathologic c | 10/24 (41.7) | 11/24 (45.8) | 18/24 (75.0) | 5/24 (20.8) |
| Colomer et al, 2008 (2003 through 2008) | n = 20, 42.8 [16–67] | 24.7 [1–61] | Prospective | Clinical b | 8/20 (40.0) | 17/20 (85.0) | 11/20 (55.0) | 12/20 (60.0) |
| Jung et al, 2009 (2004 through 2007) | n = 24, 52.8 (11.3) | 10 [2–39] | Prospective | Clinical b | 1/24 (4.2) | 5/24 (20.8) | 16/24 (66.7) | 21/24 (87.5) |
| Ghezzi et al, 2012 (not suggested) | n = 82, 56 [13–80] | 28.5 [3–86] | Prospective | Clinical b | 14/82 (17.1) | 19/82 (23.2) | 75/82 (91.5) | 64/82 (78.0) |
a Proportion of patients referred to oncologic department for laparoscopic staging due to incomplete staging operation at initial surgery
b Diagnosis by clinical findings such as results of image studies (computed tomography/magnetic resonance imaging/positron emission tomography) before staging surgery
Results on the operative time could not be summed because of considerable heterogeneity ( Table 2 ). As a result of the metaregression on our data, operation time was not associated with a lower proportion of incomplete staging at the initial surgery and a greater harvested number of lymph nodes ( P = .714 and P = .209, respectively). From the overall data and prospective data, operative time was significantly reduced with a higher proportion of fertility-sparing surgery ( P = .013 and P = .042, respectively).
| Metaregression or metaanalysis | N a (n b ) | Heterogeneity (I 2 ) | Exponentiated slope coefficient (95% CI) | P value |
|---|---|---|---|---|
| Operation time (min) | ||||
| Mean (95% CI): laparoscopy vs laparotomy–cannot be suggested | 3 (56 vs 82) | 83.2% | .973 | |
| Metaregression by incomplete staging % at initial surgery (random c ) | ||||
| Retrospective data | 7 (196) | 88.4% | −0.15 (−1.24 to 0.94) | .791 |
| Prospective data | 4 (150) | 93.2% | −0.61 (−2.32 to 1.11) | .488 |
| Overall data | 11 (346) | 90.0% | −0.15 (−0.93 to 0.64) | .714 |
| Metaregression by harvested no. of lymph node (random c ) | ||||
| Retrospective data | 7 (196) | 88.4% | −1.56 (−6.30 to 3.18) | .519 |
| Prospective data | 4 (150) | 93.2% | −3.65 (−16.31 to 9.00) | .572 |
| Overall data | 11 (346) | 90.0% | −1.89 (−4.83 to 1.06) | .209 |
| Metaregression by fertility-sparing surgery, % | ||||
| Retrospective data (fixed d ) | 3 (108) | 0.0% | −0.38 (−2.61 to 1.86) | .740 |
| Prospective data (random c ) | 4 (150) | 93.2% | −1.79 (−3.51 to −0.06) | .042 |
| Overall data (random c ) | 7 (258) | 86.8% | −1.57 (−2.80 to −0.34) | .013 |
| Estimated blood loss (mL) | ||||
| Mean (95% CI): laparoscopy vs laparotomy (fixed d ) 233.8 (195.7–272.0) vs 466.8 (340.1–593.4) | 3 (56 vs 82) | 0.0% | < .001 | |
| Metaregression by incomplete staging % at initial surgery | ||||
| Retrospective data (fixed d ) | 6 (143) | 49.6% | 1.84 (0.39–3.30) | .013 |
| Prospective data | 2 (106) | – | Cannot be determined | – |
| Overall data (random c ) | 8 (249) | 93.3% | 1.20 (–3.23 to 5.64) | .595 |
| Metaregression by harvested no. of lymph node | ||||
| Retrospective data (fixed d ) | 6 (143) | 49.6% | −3.70 (−8.53 to 1.13) | .133 |
| Prospective data | 2 (106) | Cannot be determined | – | |
| Overall data (random c ) | 8 (249) | 93.3% | −2.99 (−15.92 to 9.93) | .650 |
| Metaregression by fertility-sparing surgery % | ||||
| Retrospective data | 2 (55) | 0.0% | Cannot be determined | – |
| Prospective data | 2 (106) | 98.1% | Cannot be determined | – |
| Overall data (random c ) | 4 (161) | 96.9% | −13.60 (−32.10 to 4.89) | .149 |
The results of EBL could not be combined ( Table 2 ). We had difficulty in obtaining significant values due to the small number of studies that provided data for EBL. Three retrospective studies that compared the EBL for laparotomy with that of laparoscopy showed homogeneity among its results (I 2 = 0.0%), and the combined findings revealed that EBL in laparoscopy was significantly lower than that in laparotomy (466.8; 95% CI, 340.1–593.4 vs 233.8; 95% CI, 195.7–272.0 mL; P < .001). From the metaregression on retrospective data, EBL significantly increased by the higher proportion of incomplete staging at the initial surgery ( P = .013).
In the analysis on perioperative complications of all studies, we could not combine the data because every study had a different definition of perioperative complication, that is to say, postoperative, intraoperative, short-term, and long-term complications. In a review of all reported complications, only 1 case with port-site metastasis was reported.
The rate of upstaging after surgery was investigated with nonsignificant heterogeneity among the results of all studies (I 2 = 43.8%), and the overall rate was found to be 22.6% (18.1-27.9%) ( Figure 2 , A). When we combined the data of the 3 comparative studies, no significant difference was found in the upstaging rate between the laparoscopy group and the laparotomy group (I 2 = 17.9%, P = .430) ( Figure 2 , B).
