Discussion
The debate about the efficacy of systematic lymphadenectomy in ovarian
cancer in the literature remains unsettled. In this meta-analysis, we
observed marginal significance in OS among sysLA in all stage ovarian
cancer. In addition, sysLA was associated with improved PFS in all stage
ovarian cancer. However, sysLA was associated with significantly higher
intraoperative, but not postoperative, complications rate.
Early stage (Stage I-IIA) ovarian cancer should be distinguished from
advanced stage (Stage IIB-IV) as the prognosis of the two diseases is
quite different. Early stage ovarian cancer has a 10-year survival of
more than 80%, while advanced stage ovarian cancer has a 5-year
survival rate less than 40% (35-38). In early stage ovarian cancer,
sysLA allows complete staging by confirming no distant microscopic
disease and provides prognostic information that can guide treatment.
Also, accurate staging may prevent unnecessary adjuvant chemotherapy.
While the role of sysLA in advanced stage is still controversial. It was
hypothesized that radical lymphadenectomy may benefit those who may have
extensive lymph node metastasis since the surgery goal is to achieve
optimal debulking. Another hypothesis pushed the case for the
therapeutic role of lymphadenectomy in advanced disease is the
pharmacologic sanctuary hypothesis. This hypothesis assumed that nodal
metastasis of ovarian cancer may be less sensitive to systemic
chemotherapy due to diminished blood supply, hence sysLA may be
therapeutic in advanced disease to remove the occult lymph node
metastasis and improve the survival (39-41).
There are 6 meta-analyses in the literature that addressed this debate.
Those conducted before the Lymphadenectomy in Ovarian Neoplasm (LION)
RCT provided a survival benefit of sysLA in all stage disease. Kim et al
concluded that sysLA is efficient in improving OS in all stage disease
compared to unsystematic lymphadenectomy (USL). However, this study did
not clearly define follow up period for survival analysis and there were
no data regarding PFS or recurrence rate (42). Similarly, Gao et al
concluded that sysLA was efficient in improving 5-year OS in all stage
disease and advanced ovarian cancer compared to USL. This study is
limited by inconsistency of definition of USL, lack of data on impact of
residual tumor status, PFS or recurrence rate (43). Zhou et al reported
that SysLA was efficient in improving 5-year OS in all stage, early and
advanced disease compared to USL in addition to improving PFS in
advanced disease. Similarly, definition of unsystematic lymphadenectomy
was not consistent among the included studies, and impact of residual
tumor status was not considered (44).
When the LION trial was published, the results tipped the balance in
favor of abandoning sysLA in advanced ovarian cancer because of no
survival benefit in addition to higher incidence of postoperative
complications. The LION trial results weighted in heavily in the
meta-analyses that were conducted this year. Lin et al concluded that
SysLA did not improve OS or PFS in optimally cytoreduction all stage
ovarian cancer patients. However, definition of unsystematic
lymphadenectomy was not consistent and no subgroup analysis was
conducted according to cancer stage (45). Xu et al reported that
analysis of RCT demonstrated that sysLA cannot improve OS or PFS in
advanced ovarian cancer which is quite the opposite of his analysis of
observational studies (46). Wang et al revealed that sysLA may improve
OS but not PFS in optimally debulked advanced ovarian cancer (47).
Our metanalysis included 22 studies with 6,825 patients with ovarian
cancer. The meta-analysis pooled results show that sysLA did not improve
OS in all stage disease. Since the studies included a range of study
population questioning whether pooled data may present a mixed effect of
sysLA, we performed a thorough subgroup analysis. While results were
close to significance, subgroup analysis of OS by splitting studies into
RCTs and retrospective demonstrated that RCTs showed no significance,
but retrospective studies were close to significance. However, subgroup
analysis by RCT demonstrated statistical significance of sysLA regarding
PFS in both RCT and observational studies. The LION trial was a
prospectively randomized, well powered, multicenter international trial
with large sample size. While the LION weight in the meta-analyses that
included RCTs will affect the results heavily, the controversy between
observational studies and the LION questions whether this discrepancy is
a result of inherited pitfalls in observational studies. However,
another explanation may be related to LION study design. First, the LION
assessed the participating centers and deemed them to be proficient in
performing sysLA and the patients who participated in the trial were of
median age 60 years and had good performance score. However, morbidity
and mortality figures in the lymphadenectomy group were relatively high.
Also, the LION excluded 65% of the registered population before
randomization for different reasons, one must question the possible
survival benefit that lymphadenectomy could have provided if patients
with poor prognosis indicators were included. Nevertheless, LION still
presents the best available evidence and should be considered over other
observational studies.
Our novel MOGGE Meta-analysis Matrix (MMM) revealed interesting
findings. Subgroup analysis by stage revealed statistically significant
superiority of sysLA in OS in advanced disease (stage IIB-IV) whether
they received adjuvant or neoadjuvant chemotherapy (C0). The clear
advantage of sysLA on OS in advanced ovarian cancer was consolidated by
a further subgroup analysis in the adjuvant chemotherapy group (C3). So,
while the LION trial was among the included studies in this subgroup
analysis, the scales here were tipped in favor of sysLA in this
particular population. Survival benefit of sysLA was more prominent in
studies covering particularly stage III-IV compared to stage IIB-IV
(D3). This may be attributed to the weight of LION trial, which is
included in the first, but not the second subgroup analysis. Although
this may be apparently reflective of superiority of LION trial study
design, impact of disease stage was investigated. Approximately 78% of
study population in LION trial were staged as IIIB to IV (10).
Similarly, Du Bios el al. is another study included in the first but not
the second analysis (included population was stage IIB to IV). Of their
cohort, only 79.6% were staged as IIIB to IV. In comparison, all
patients reported by Chang et al. were staged as IIIC (23), all patients
reported by Eoh et. al. were stage IIIC-IV (16), and 94% of patients
reported by Paik et. al. were stage IIIB-IV (18). In addition, LION
trial did not present subgroup results based on disease stage.
Therefore, study design may not be the only contributing factor, and OS
benefit of sysLA may be associated with more advanced stages, including
stage IIIB or IIIC, compared to earlier stages, including stage IIB.
Unfortunately, specific studies on stage IIIB or IIIC are too few to
draw a direct conclusion. Therefore, it seems that although LION study
provides the most robust level of evidence, our current results may
warrant further assessment/subgroup analysis that narrows the spectrum
toward more advanced stages specifically stage IIIB, IIIC.
Unlike women who received adjuvant chemotherapy, there was clearly no
survival benefit of sysLA among women who received neoadjuvant
chemotherapy regardless of disease stage. Therefore, neoadjuvant
chemotherapy seems to omit need for sysLA if ever needed. This outcome
may reflect either that: (1) neoadjuvant chemotherapy provides
therapeutic benefit that deems sysLA unnecessary, or (2) selected
patients for neoadjuvant chemotherapy who responded to treatment may
still yield a poor prognosis that limits survival benefit of sysLA.
However, recent evidence reflects that prognosis among women with
advanced ovarian cancer receiving neoadjuvant chemotherapy is non
inferior to adjuvant chemotherapy after surgery (6, 48). Given surgical
complications of sysLA, these findings raise a question whether
neoadjuvant chemotherapy may be considered in women with stage III-IV to
avert sysLA if the latter is anticipated to improve survival.
Our metanalysis is the first to date to evaluate intraoperative
complications of sysLA in ovarian cancer. It is of crucial importance in
determining the benefit risk of performing a complex procedure as
lymphadenectomy. Our metanalysis overcame the common limitation that
most of the previously published metanalyses faced which is
non-consistent definition of unsystematic lymphadenectomy among the
included studies by dividing the control groups into selective
lymphadenectomy and no lymphadenectomy groups. This meta-analysis
provides a novel and comprehensive subgroup analysis to reveal specific
conclusions.
However, this meta-analysis is limited by heterogeneity of the included
studies. Since most of the studies were retrospective, they might
contain selection and confounding biases. No subgroup analysis was
conducted with regard to histological type because of no individual
patient data or aggregate level data available. Histological subtype may
present different biological tumor behaviors and hence different therapy
lines.
In conclusion, sysLA was associated with improved PFS, but not improved
OS, in all stages of ovarian cancer. Current evidence, based on a
well-designed RCT, did not endorse a prognostic role of sysLA in women
with advanced ovarian cancer. Nevertheless, future studies on a narrow
spectrum of patients with stage IIIB and IIIC may be warranted
particularly those who received adjuvant chemotherapy. On the contrary,
women who received neoadjuvant chemotherapy did not seem to benefit from
sysLA regardless of study design or disease stage.
Acknowledgement: none
Disclosure of Interests: The authors have no conflicts of
interest. No financial disclosure to declare