Impact of R0 resection for synchronous peritoneal metastasis from colorectal cancer: A propensity score‐matched analysis of a multi‐institutional database

Abstract Aim To investigate the usefulness of resection for synchronous peritoneal metastasis from colorectal cancer. Methods The patients who underwent surgery for stage IV colorectal cancer at 16 hospitals between 1991 and 2007 were enrolled in this study. The overall survival rates of patients with synchronous peritoneal metastasis from colorectal cancer with and without R0 resection were compared using a propensity score‐matched analysis. Results Among the 3965 patients with stage IV colorectal cancer, 1169 had synchronous peritoneal metastasis (28.5%). No patients received hyperthermic intraperitoneal chemotherapy (HIPEC) in this study. Among the 1169 patients, 783 had enough clinicopathologic information and went through further analysis. Out of 783 patients, 204 underwent R0 resection. A multivariate analysis revealed that severity of peritoneal metastasis according to the Japanese classification (P < .0001) and distant metastases (P < .0001) were independently associated with non‐R0 resection. In a propensity score‐matched analysis, 118 patients who underwent R0 resection were matched with 118 patients who did not undergo R0 resection. There was no significant difference in each parameter between patients with and without R0 resection. After matching, the overall survival in patients with R0 resection was better than that without R0 resection (median survival time: 28.8 months and 15.6 months, P < .0001). Conclusion The overall survival of patients with R0 resection for synchronous peritoneal metastasis from colorectal cancer was better than that without R0 resection even without HIPEC.


| INTRODUC TI ON
Colorectal cancer is the second most common cause of cancer mortality in the United States and Japan. 1,2 Furthermore, the incidence of colorectal cancer is increasing rapidly in Japan. 2,3 Peritoneal metastasis is one of the poor prognostic factors in patients with colorectal cancer and is found in 5%-10% of primary colorectal cancer cases. 4,5 The patients with synchronous peritoneal metastasis are classified into Stage IVC in the latest American Joint Committee on Cancer (AJCC) Cancer Staging Manual. 6 There is urgent need to develop modalities to improve the prognosis of patients with colorectal cancer with synchronous peritoneal metastasis. R0 resection of both primary colorectal cancer and peritoneal metastasis has been reported to lead to better outcomes 5 .
However, preoperative detection of peritoneal metastasis from colorectal cancer is difficult. The surgeons often choose operative procedures for peritoneal metastasis during surgery.
The aim of this study was to investigate the usefulness of R0 resection for colorectal cancer with synchronous peritoneal metastasis using a multi-institutional database and a propensity scorematched analysis.

| Patients
The 16 member-hospitals of the Japanese Society for Cancer of the Colon and Rectum (JSCCR) collected the data of the 3965 patients with stage IV colorectal cancer treated by surgery between 1991 and 2007. Of these patients, 1169 had synchronous peritoneal metastasis (28.5%). Among these, 783 had detailed information, and their data were further analyzed. Neither extended peritonectomy nor hyperthermic intraperitoneal chemotherapy (HIPEC) was performed in this study. The evaluation of peritoneal metastasis was performed by surgeons during the colorectal surgery. The ethics committee of Tokyo Metropolitan Hiroo Hospital approved this study.

| Parameters
The parameters used in this study included gender, age, location of primary tumor, histologic type, depth of tumor invasion, lymph node metastasis, hematogenous metastasis, and the extent of peritoneal metastasis. The extent of peritoneal metastasis was described according to the Japanese classification (Table 1). 7 A previous report evaluated the objectivity of the Japanese classification as follows: P1, peritoneal metastases 20 mm or smaller confined to one area; P2, 10 or fewer peritoneal metastases disseminated in two or more areas, or peritoneal metastases confined to one area but the size is >20 mm; and P3, more than 10 peritoneal metastases disseminated in two or more areas. 8

| Statistical analysis
Associations between patients with and without curative resection were analyzed by the χ 2 test. The actuarial survival after surgery was determined from Kaplan-Meier curves. The log-rank test was used to compare the overall survivals. The Cox proportional hazards model was used to determine the independent prognostic factors in patients with colorectal cancer and synchronous peritoneal metastasis.
Thereafter, pairwise 1:1 propensity score matching, including logistic regression, was used to reduce the effects of non-random assignment of patients to curative resection. The propensity score matching method has been used to reduce potential confounding caused by unbalanced covariates. 9 In short, by multivariate logistic regression analysis, the propensity score for curative resection was determined. Patients with and without curative resection were matched by greedy matching without replacement.
The SPSS 22 software (IBM) was used for data analysis.
Median and its range or numbers of patients and ratios (%) were expressed as data. P value <.05 indicated statistical significance in this study. Table 2 shows the patient characteristics for the entire cohort. The median age was 63 years in patients with and without R0 resection, respectively. Among the eight parameters, location of tumor (P = .0091), distant metastasis (P < .0001), and extent of peritoneal metastasis (P < .0001) had significant differences between patients with and without R0 resection for colorectal cancer with synchronous peritoneal metastasis. The smaller extent of peritoneal metastasis (P < .0001) and the absence of distant metastasis (P < .0001) were independently associated with R0 resection. In 481 patients with distant metastasis, the number of patients with liver metastasis only, both liver and other distant metastasis, and distant metastasis other than liver was 250, 132, and 99, respectively. The R0 resection rates in the three groups were 11.6%, 2.3%, and 23.2%, respectively.

| Survival
The median follow-up period of the entire cohort was 14 There was a significant difference in overall survival between patients with and without R0 resection (P < .0001).

| Propensity score matching cohort
A binomial logistic regression model was used to calculate the propensity score. Extent of lymphadenectomy (P < .001), tumor location (P = .007), and distant metastasis (P < .0001) were selected.
The Hosmer-Lemeshow test showed a good fit for this model (P = .096). The C statistic of this model was 0.88 (95% confidence interval: 0.85-0.90). In this study, 118 patients who underwent curative resection were matched with 118 patients who did not undergo curative resection (Table 4). As for each predictive parameter, no significant difference was found between patients with and without R0 resection, which showed that these two groups were well-matched by propensity score.
The MST of patients with and without R0 resection was 28.8 months and 15.6 months, respectively, in the propensity scorematched cohort. The 5-year overall survival rates after matching were 23.2% and 2.2%, respectively. The overall survival of patients with R0 resection was better than that of those without (P < .0001, Figure 1B). Abbreviations: CI, confidence interval; Well or mod, well or moderately differentiated adenocarcinoma. including peritonectomy with HIPEC, were beneficial for patients with a peritoneal carcinoma index (PCI) score of 10 or less. 12 Although we also reported the usefulness of R0 resection without HIPEC for peritoneal metastasis from colorectal cancer, those were retrospective studies. 13,14 Therefore, we conducted a propensity score-matched analysis in this study to remove biases as much as possible.

| D ISCUSS I ON
In the present study, the prognoses of the patients with R0 resection for synchronous peritoneal metastasis from colorectal cancer were better than those without in propensity score-matched groups as well as in non-matched groups. Before matching, there were significant differences in location of primary tumor, accompanying distant metastasis, and severity of peritoneal metastasis between R0 and non-R0 resection groups. After matching, although no differences were found in these factors, 5-year survival rate in R0 resection group was 20% higher than that in non-R0 group even without HIPEC.
Currently, four randomized controlled trials have been reported regarding the efficacy of HIPEC. 11,15-17 Of these, two trials were stopped before the completion of the proposed patient recruitment. 15 Abbreviation: Well or mod, well or moderately differentiated adenocarcinoma.