Background
Small cell lung cancer (SCLC) accounts for 13% to 20% of lung cancer, and 60% to 70% of SCLC patients have extensive disease at the time of initial diagnosis [
1,
2]. Platinum-based chemotherapy is one of the main treatments for extensive-stage SCLC [
2]. The prognosis is poor, with a median survival of less than 1 year [
3]. The propensity of brain metastasis (BM) is one major difference between SCLC and other lung cancers. About 10% of SCLC patients experience BM at diagnosis and it rises to 50% or even higher after 2-year survival [
4]. It is demonstrated that there is a positive role of prophylactic cranial irradiation (PCI) on survival in resected [
5] and limited-stage SCLC patients [
6]. Whether PCI should be used for extensive-stage SCLC is still in debate.
In 2007, a randomized trial reported the role of PCI in extensive-stage SCLC patients who responded to chemotherapy [
7]. The primary endpoint was symptomatic brain metastases time. The results showed that the risk of symptomatic brain metastases in PCI group was as low as 0.27 times compared to that of the control group, along with a prolonged overall survival (OS) from 5.4 months to 6.7 months [
7]. These findings suggested PCI was useful in reducing BM risk and prolonging survival [
7]. Since then, several retrospective studies [
8,
9] also evaluated the effects of PCI on survival and incidence of BM and showed similar results. However, Toshiaki et al. [
10] showed that PCI did not significantly improve OS in extensive-stage SCLC. They reported that the median OS of PCI versus control was 11.6 months versus 13.7 months (HR = 1.27,
p = 0.094). Therefore, they concluded that PCI was not essential for extensive-stage SCLC, regardless of prior response to chemotherapy [
10]. These conflicting results contradict previous findings that highlighted a positive effect of PCI on extensive-stage SCLC.
To answer whether PCI could have a beneficial effect on OS in extensive-stage SCLC, we collected clinical studies assessing the role of PCI versus control in patients with extensive-stage SCLC. Survival data were combined to determine the impact of PCI versus control on the survival of extensive-stage SCLC. The pooled results suggest that PCI could benefit these patients by prolonging survival.
Methods
This study was performed based on the PRISMA and the Cochrane handbook guidelines [
11].
Systematic search
Electronic databases including PubMed, Embase, the Cochrane library, and the Chinese Biomedical Literature Database were systematically searched with deadline on March 2018. Search terms were “prophylactic cranial irradiation or PCI”, “lung carcinoma or lung cancer or lung tumor or lung neoplasm”, “extensive or advanced”. These terms were used in different combinations without language restrictions.
Inclusion and exclusion criteria
Inclusion criteria: (1) Patients were diagnosed by pathology or cytology to confirm SCLC, and had imaging exams to determine extensive-stage disease; (2) OS data of extensive-stage SCLC were reported; (3) If results from the same patient population were published in different journals, the most complete or up-to-date study was selected; (4) Eligible studies were not only published full-text but also abstract, conference meeting presentation and unpublished literature.
Exclusion criteria: Studies were excluded if they met the following (1) Insufficient data on survival; (2) Limited-stage SCLC, non-small cell lung cancer, or other metastatic tumor; (3) Review, animal experiments, comments, survey, and guidelines.
Two researchers conducted the screening of eligible studies, independently. When there was an inconsistency, it was solved through discussion. The following information was extracted: first author, time of publication, region, participant number, sex, age, received chemotherapy or not before PCI, PCI administration timing, thoracic radiotherapy, confirmed absence of brain metastases before PCI (pre-PCI BM status), and survival and efficacy data.
Quality assessment
The Newcastle-Ottawa scale (NOS) method [
12] was used to assess the quality of the included retrospective studies. Quality evaluation of randomized studies was performed using the literature quality assessment method provided by the Cochrane Handbook 5.1. [
11] The NOS focuses on three aspects during evaluation: the selection of the cohort, the comparability of the cohort, and the evaluation of the results. A maximum of 4 stars can be obtained in the selection, and the comparability can be obtained by up to 2 stars. The evaluation of the results can be up to 3 stars. The highest quality is assigned the highest number of stars. In the Cochrane Handbook approach, each randomized study is evaluated in five major aspects: Random, Allocation, Blinding, Selective reporting, and Other bias.
Statistical method
The meta-analysis was used to combine the individual HRs and its related 95% CIs to draw a forest plot. Q test and I
2 statistic were used to test the heterogeneity across the included studies. When
p < 0.1 or I
2 > 50%, it suggested a significant heterogeneity between the included trials. Then the random effects model was used; when
p > 0.1 or I
2 < 50%, it indicated no significant heterogeneity and the results were combined using a fixed effect model [
11].
The HRs of PCI versus control on OS were extracted. If there was only a survival curve comparing PCI versus control without HRs values, the reported methods [
13,
14] were used to calculate the HRs. Combined HR < 1 indicated that PCI was a favorable factor for extensive-stage SCLC, whereas HR > 1 indicated the opposite. BM incidences after PCI versus control were also extracted to determine the influence of PCI on the risk of BM. Subgroup analyses of OS were performed in terms of study type, region, and timing of PCI administration. The funnel plot, Begg’s method, and Egger’s method were used to detect the publication bias, and
P < 0.05 suggested there was a significant publication bias [
15,
16]. Sensitivity analysis was performed to confirm the stability of the results if necessary. Meta-analysis was performed using RevMan 5.3 and Stata 11 software. For a combined estimate, a
P < 0.05 was considered as there was a statistical significance.
Discussion
In this meta-analysis, 14 clinical trials were included. The combined results showed that PCI improved OS and reduced incidences of BM in extensive-stage SCLC. Subgroup analyses of OS suggested that the benefits of PCI in improving OS were independent of region, and PCI administration timing.
The combined results found that PCI had the ability to prolong OS, while the 1-year, 2-year and 3-year survival rates (data not shown) were similar between PCI versus control. These findings were thought to be controversial. One explanation was the limited number of included studies. Although most of the studies provided OS, three studies exhibited survival rate. Another reason may be the differences in study design. The objectives of the included randomized trials were different. The primary objective of Slotman et al. [
7] was to determine whether PCI could reduce the incidence of symptomatic BM, whereas Toshiaki et al. [
10] aimed to compare the efficacy of PCI on OS in BM free patients (previously checked by imaging). This indicated selection bias probably existed among studies. A third reason may be that the HR stands for the overall hazard ratio of getting improvement in survival but not for single time-point. One time-point is not capable of estimating the overall effect of PCI on survival.
Recently, there was a meta-analysis [
27] evaluating the impact of PCI in extensive-stage SCLC. They included five studies comprising 984 cases. They found a similar role of PCI in prolonging PFS and reducing the risk of BM. This was in accordance with our findings. However, their results showed PCI did not statistically prolong OS when compared to control (HR = 0.82; 95% CI: 0.60–1.11;
p = 0.19), and the PCI group had a better 1-year survival (37.1% versus 27.1%; RR = 0.87;
p = 0.002) compared to the control group [
27]. In contrast, our analysis shows a statistical advantage of PCI on OS (HR = 0.57; 95%CI: 0.47, 0.69;
p < 0.00001). However, we also observe no significant HR after analyzing data from the included RCTs, and the previously described different aims of these two studies may be responsible for this result.
There were several limitations within our study. First, only two randomized controlled trials were included and the others were retrospective studies. Retrospective studies could introduce more selection bias and selective reporting. Secondly, the baseline characteristics along with the treatment history and PCI dose varied, which may be the origins of heterogeneity across included studies. The baseline characteristics of included studies indicated that the source of heterogeneity could be the region, chemotherapy response history, and timing of PCI. Other features like age, gender, performance status, PCI dose, and radiation technology may also contribute to the heterogeneity. These differences did not significantly affect the effect of PCI on OS as indicated by the sensitivity analysis. Thirdly, some of the HRs were indirectly collected which may underestimate or overestimate the actual role of PCI on survival. Forth, though there was no language limitation, the search language used in this study was only English and Chinese. Other language articles were not included in the study, which may cause unavoidable bias. Finally, although our study did not find obvious publication bias, it could not be completely avoided. This was because positive research results were more likely to be published, and negative research results were more likely to be rejected or not published.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (
http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.