Cytoreductive surgery plus hyperthermic intraperitoneal chemotherapy improves survival for peritoneal carcinomatosis from colorectal cancer: a systematic review and meta-analysis of current evidence

Objectives The therapeutic efficacy of cytoreductive surgery (CRS) plus hyperthermic intraperitoneal chemotherapy (HIPEC) for patients with peritoneal carcinomatosis (PC) from colorectal cancer (CRC) is still under debate. This meta-analysis and systematic review of published literature on this comprehensive strategy aims to evaluate its efficacy on CRC patients with PC. Methods A systemic review with meta-analysis of published literatures on treatment of CRS plus HIPEC for patients with PC from CRC was performed. In addition, a summary of study results of published literatures concerning CRS plus HIPEC treating patients with PC from CRC was also conducted. Results A total of 76 studies were selected, including 1 randomized controlled trial, 14 non-randomized controlled studies, and 61 non-controlled studies. The pooled hazard ratios (HRs) for overall survival (OS) in the 15 researches for meta-analysis was 2.67 (95% CI, 2.21-3.23, I2= 0%, P < 0.00001), and no significant evidence of publication bias was found. The difference of chemotherapy regimens of HIPEC was not associated with OS and DFS (disease-free survival) after CRS and HIPEC, with no significant difference of heterogeneity (P = 0.27, I2 = 24.1%). In both groups of mitomycin C based HIPEC group and oxaliplatin group, patients received HIPEC had significant better survival (P < 0.00001). The mean mortality and morbidity for HIPEC program were 2.8% and 33.0%, respectively. Conclusions This meta-analysis revealed that comprehensive therapeutic strategy of CRS plus HIPEC could bring survival benefit for selected patients with PC from CRC with acceptable safety.


INTRODUCTION
Peritoneal carcinomatosis (PC), as a lethal regional progression for patients with colorectal cancer (CRC), has long been considered as a terminal condition with few effective treatments. In the past, the median overall survival (OS) of PC from colorectal cancer is 4 to 7 months after palliative surgery or 5-FU-based systemic Review Oncotarget 55658 www.impactjournals.com/oncotarget chemotherapy with best supportive care [1][2][3]. Current systemic chemotherapy focusing on new chemotherapeutic agents such as oxaliplatin and irinotecan, along with antiangiogenesis molecular targeting agents cetuximab and bevacizumab [4][5][6][7], could extend the median OS up to about 12 months [5]. However, long-term survival is still hard to be achieved by systemic chemotherapy alone.
Nevertheless, therapeutic efficacy of this comprehensive treatment strategy for CRC PC patient remains controversial due to insufficient convincing evidence. Therefore, we conducted this meta-analysis of published clinical studies to verify the efficacy of this strategy against CRC PC.
Sensitivity analysis of summarized HR and 95% CI showed no difference after choosing random effects model and fixed effects model. In terms of sample size difference, 15 researches were divided into three subgroups (sample size <50, 50-100, >100) by a sensitivity study for a stratified meta-analysis. The summarized HR and 95% CI showed no difference, with no betweensubgroup heterogeneity (P = 0.48, I 2 = 0%) ( Figure  2b). In a sensitivity analysis, four studies with potential heterogeneity was removed due to small sample size [34] [14,26,40,99], but the summary HR was 2.81 (95%CI, 2.28-3.48, I 2 = 0%, P heterogeneity = 0.56).

Analysis of chemotherapy regimens
Regarding the effect of different chemotherapy regimens in HIPEC procedure on the efficacy on OS or DFS, 15 researches were divided into 3 subgroups: group of MMC based chemotherapy, group of L-OHP based chemotherapy, and group of other regimens. The heterogeneity showed no significant difference (P = 0.27, I 2 = 24.1%), which revealed that difference of chemotherapy regimens of HIPEC was not associated with OS and DFS after CRS and HIPEC in this meta-analysis ( Figure 3a). A further analysis of difference in median year survival rate between group of CRS plus HIPEC and group of traditional treatment was conducted by independentsamples T test stratified by MMC and L-OHP subgroups (Figure 3b and Figure 3c).

Other chemotherapy regimes in HIPEC procedure
Three trials [14,26,29] were identified as the subgroup of other regimen due to difficulties in identifying them as MMC subgroup or L-OHP subgroup since mixed chemotherapy regimens were used in HIPEC during the whole disease course. A significant survival benefit in HIPEC group vs. traditional group (HR = 3.90, 95% CI, 1.73-8.81, I 2 = 0%, P < 0.00001) was demonstrated ( Figure 3a).

Publication bias
Publication bias was evaluated with funnel plot analyses, as shown in Figure 4, and the funnel plot was symmetric. No apparent publication bias was found in our OS meta-analysis with Begg's test (z continuity corrected = 0.99, Pr >|z| continuity corrected = 0.32) (Figure 4a

Summary of HIPEC-related data
In 15 controlled studies and 59 single-arm studies, HIPEC-related outcomes including survival rates, median OS and 95% CI, DFS/RFS, PFS, follow-up time, morbidity, and mortality, are summarized in Table 20-25 and Figure 5

DISCUSSION
Due to the tumor biologic characteristics of colorectal cancer, about 10-13% patients have already  progressed to PC when CRC is diagnosed [3,7], which has a poor prognosis. In order to improve the efficacy, a comprehensive treatment strategy with combination of CRS plus HIPEC had been developed. With wide application of this treatment, CRS plus HIPEC has been proved capable to achieve better survival in selected patients with PC from colorectal cancer. This meta-analysis of 15 controlled studies demonstrated that CRS+HIPEC comprehensive therapeutic strategy could bring significant survival benefit for selected CRC PC patients than traditional treatment of palliative surgery alone or systemic chemotherapy (HR = 2.67, 95% CI 2.21-3.23, P < 0.00001). In addition, the summarizing analysis of these 76 studies showed that the median OS was about 29 months in HIPEC group, which is significant longer compared with median OS of 17.9 months for CRC PC patients receiving contemporary chemotherapy reported by Kerscher et al (n = 2,406) [7]. These results provide further supporting evidence that CRS+HIPEC as the principal comprehensive treatment can bring more survival benefit to selected patients with CRC PC than traditional therapy.
The different regimens used in chemotherapy may be one potential confounding factor for survival outcomes. In order to investigate the influence of chemotherapy regimens on postoperative survival, a stratification analysis between MMC based regimens and L-OHP based regimens was conducted. The results of heterogeneity showed no significant difference (P = 0.50). These results are inconsistent with the reports by Elias et al [14], which showed that OS advantage for L-OHP regimens over non-L-OHP regimens (32 vs. 25 months, P = 0.02). However, L-OHP used in HIPEC was not an independent prognostic factor for survival in the study of Elias and colleagues. A multi-center retrospective controlled study reported by Prada-Villaverde et al. [72] showed that of 539 patients undergoing CRS plus HIPEC, L-OHP based HIPEC and MMC based HIPEC achieved similar median OS (31.4 vs. 32.7 months, P = 0.925). Similarly, the study of Hompes et al. [59] yielded the same conclusion that there was not obvious benefit in OS for HIPEC with L-OHP (37.1 months) or MMC (26.5 months) (P = 0.45). Although different chemotherapy regimens in HIPEC may have an effect on stability and reliability of this meta-analysis, the result of heterogeneity analysis was in accordance with above studies. As a result, both MMC and L-OHP were the feasible chemotherapy drugs in HIPEC for CRC PC patients to achieve similar efficacy.
Moreover, there are also some doubts that different chemotherapy in intravenous or postoperative There were a few statistical flaws in this metaanalysis. For example, only one RCT [12] was included. It may be due to the difficulty of performing RCT. Therefore, we had to select meticulously current studies of best evidence level besides the only RCT. However, this meta-analysis showed acceptable outcomes of low heterogeneity and sensitivity. Regrettably, a patient-level (based on single patient data) meta-analysis as the gold standard for meta-analysis was not performed because of the difficulty in obtaining vast data from each database or institution. In addition to meta-analysis, this report provided a summary of 76 clinical studies published until today about CRS and HIPEC, which can get a review of published studies. In order to get the best evidence level results, more RCTs and prospective, multicenter, largescale clinical trials need to be performed in future studies.
With the summary of 76 studies, it is found that although HIPEC is now widely accepted and performed in most institutions, details of performing HIPEC varies among different institutions. As we noted, there are several mainly different techniques concerning HIPEC including 1) "open" or "closed" technique, 2) using MMC and/ or L-OHP, 3) mono-chemotherapy or combination of chemotherapy regimens, 4) temperature and duration of HIPEC. These can be further studied in future studies.
In conclusion, this meta-analysis showed that CRS+HIPEC comprehensive therapeutic strategy was associated with improvement of OS in CRC PC patients, and the results of the meta-analysis were proved of good reliability by low heterogeneous and robust sensitivity. Meanwhile, CRS and HIPEC can be performed with acceptable safety according to summary results of all 76 studies.

Search strategy
The following databases were systematically searched up to July 31, 2016 including PubMed, Science Citation Index, EMBASE, and MEDLINE. The Cochrane Central Register of Controlled Trials, the National Institutes of Health trial registry, and conference proceeding articles from major oncologic and gastrointestinal cancer meetings were also sought for published results. The key words included     "colon", "rectum", "colorectal", "cancer", "peritoneal carcinomatosis", "hyperthermic intraperitoneal chemotherapy", and synonyms and related terms for these words. The MeSH terms included "colon cancer", "rectal cancer", "colorectal cancer", "peritoneal carcinomatosis", "hyperthermic chemotherapy", "hyperthermic intraperitoneal chemotherapy", "HIPEC", "intraperitoneal chemohyperthermia", and "IPCH". The combined application of "key words terms" and "MeSH terms" were conducted to improve the efficiency and accuracy of literature search.

Selection criteria
For inclusion in the meta-analysis and summarized HIPEC-related data analysis, a study had to fulfill the following criteria: (1) According to the North-England evidence-based guidelines [105,106], excluded from IV levels evidence of literatures were included; (2) All patients were diagnosed CRC PC; (3) For assessing CRS+HIPEC±SC/EPIC, the intervening measure group was CRS+HIPEC±SC/EPIC, while the control group was traditional therapy of surgery and/or SC; For systematic review of CRS+HIPEC to treat CRC PC, HIPEC-related literatures involving clinical efficacy evaluation were

Data extraction
Three authors analyzed data from a meta-analysis of 15 controlled researches of CRS plus HIPEC group vs. surgery and/or SC group and a summarized analysis of 76 researches of HIPEC group. The following data were extracted from each article: (1) Major clinicopathologic characteristics and detail HIPEC regimens; (2) Survival and advent events. All relevant text, tables, and figures were reviewed for data extraction. For equivocal literatures or discrepancies between two independently assessed reviewers, these were resolved by discussion and consensus with a third author.

Statistical methods
All meta-analysis were performed using Review Manager 5. Overall survival (OS) or disease-free survival (DFS) in all studies were extracted from original literature. If not achieved accurate data in original text, hazard ratios (HRs) for time-to-event outcomes with 95% confidence intervals (95% CI) in two groups were estimated by Tierney's methods [108]. The heterogeneity in the metaanalysis was evaluated by I 2 statistics [109] and T test [110] was calculated for each result in summarizing analysis of all HIPEC-related data from the included 76 articles. If I 2 >50%, it was defined as the unacceptable heterogeneity. If I 2 <50%, fixed effect model was used to get pooled HR and 95% CI; otherwise, random effects model was used if moderate heterogeneity. For a sensitivity analysis, we investigated the different research features of eligible trials, which included statistical methods, methodological quality, sample sizes, and clinical factors on HIPEC-related effect, after that, summarizing each subgroup data in term of Mental-Haenszel stratification analysis. According to Egger's test [111] and Begg's test [112], publication bias was considered to be inevitable when P < 0.10. The funnel plot analyses using 'STATA: Data Analysis and Statistical Software version 12.0', was to observe the results of meta-analysis whether any publication bias.