Secondary cytoreductive surgery and hyperthermic intraperitoneal chemotherapy for recurrent colorectal peritoneal metastases

Background Secondary treatment of recurrent colorectal peritoneal metastases after previous cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) is poorly investigated. Objectives To evaluate the overall survival outcome of secondary (repeat) CRS + HIPEC compared to palliative treatment in recurrent peritoneal disease. Methods Patients with colorectal peritoneal metastases treated with an index CRS + HIPEC and subsequently having recurrent peritoneal disease were identified from the prospective Swedish national HIPEC registry. Patients were divided into interventional group (secondary CRS + HIPEC) or palliative group. Multivariable logistic regression, propensity-score matching, and survival outcomes were calculated. Results Among 575 patients who underwent complete CRS between 2010 and 2021, 208 (36 %) were diagnosed with a subsequent recurrent peritoneal disease. Forty-two patients (20 %) were offered secondary CRS + HIPEC. Propensity-score matching of secondary interventional cases with palliative cases succeeded in 88 % (n = 37) in which female sex, lower peritoneal cancer index at index surgery, longer disease-free interval, and absence of extra-peritoneal metastases were identified as the most relevant matching covariates. Median OS from date of recurrence was 38 months (95%CI 30–58) in the interventional group and 19 months (95%CI: 15–24) in the palliative group (HR 0.35 95%CI: 0.20–0.63, p = 0.0004). Sensitivity analyses confirmed the results. As reference, the median OS from index CRS + HIPEC in the whole colorectal registry (n = 575) was 41 months (95%CI: 38–45). Conclusion After matching for relevant factors, the hazard ratio for death was significantly reduced in patients who were offered a secondary CRS + HIPEC procedure for recurrent peritoneal disease. Selection bias is inherent, but survival outcomes were comparable to those achieved after the initial procedure.


Introduction
Cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC) is a standard treatment option for patients with resectable colorectal peritoneal metastases (PM), although the HIPEC component is under some scrutiny [1].A recent trial with mitomycin C in the prophylactic setting has suggested a locoregional effect on the peritoneum [2].Regardless of the use of HIPEC, CRS is a treatment cornerstone for resectable colorectal PM [3,4].
It is still unclear whether a second CRS and HIPEC treatment after peritoneal recurrence following initial CRS and HIPEC provides any survival advantage over systemic chemotherapy.There are previous studies evaluating repeat CRS and HIPEC, but without any relevant comparative group [5][6][7][8][9][10][11][12][13].Selecting patients for a secondary procedure is a difficult decision.Many factors must be taken into account in this situation.It may be that the patients chosen for a second procedure would have done just as well on systemic chemotherapy, but any relevant comparison groups are missing in these studies except for one [7].
The aim of this study was to evaluate the outcome of secondary (first repeat) CRS and HIPEC in patients who had already undergone the treatment once.The primary outcome was overall survival.Compared between the group of patients who were offered secondary surgery with HIPEC and the group who went for palliative treatment only.This was a Swedish Peritoneal Oncology Group study.

Materials and methods
All patients in the Swedish HIPEC Registry who underwent complete CRS and HIPEC due to colorectal PM between 2010 and 2021 were extracted.From this group, patients who had a subsequent peritoneal recurrence were selected for the decision-making analysis, i.e. analysis to identify predictive factors related to receiving secondary CRS.Whether a patient was offered secondary CRS and HIPEC or not was modelled by uni-and multivariable logistic regression with secondary intervention as endpoint, where baseline variables that differed between groups were included.For prognostic evaluation, a univariate Cox regression analysis on overall survival was also performed on these baseline variables.Variables with a p-value <0.1 in the logistic regression analysis and variables considered prognostically important in the Cox regression (p < 0.1) were defined as relevant criteria in the decisionmaking analysis.Subsequently, the following variables were included in the propensity score matching process: sex, peritoneal cancer index (PCI) at index CRS and HIPEC, disease-free interval, and the absence of extra-peritoneal metastases.Operating time was also statistically significant, but due to significant collinearity between PCI and operating time, the PCI was chosen instead.This study was approved by the Ethical Review Authority (Dnr 2023-03743-01).

Statistics
The process of choosing the variables for propensity score matching (PSM) is described above.The nearest neighbor method without replacement and with a caliper of 0.1 was used to define the propensity score matching process.Both the recurrent cohort (n = 208) and the PSM cohort (n = 74) was compared with descriptive statistics.The overall survival (OS) was compared between groups using Kaplan-Meier curves and log-rank test.Cox regression analysis was used to calculate hazard ratios.Signet-cell histology was not included in Cox regression due to few cases in the interventional group.A p-value of <0.05 was defined as statistically significant.

Sensitivity analyses statistics
Several sensitivity analyses were performed.First, an unadjusted Kaplan-Meier curve in the recurrent cohort was rendered.Second, a multivariable Cox regression analyses using the most often identified prognostic factors from the literature was performed for the recurrent cohort.Third, the same multivariable Cox regression was performed for the propensity score matched cohort.Fourth, several propensity-score matched methods were tested to investigate whether propensity score methodology affected the results.Lastly, an analysis of survival from index surgery was performed in the propensity score matched cohort.

Results
Among 575 patients who underwent complete CRS for colorectal peritoneal metastases 2010-2021 (Fig. 1), 208 (36 %) were diagnosed with a peritoneal recurrencethe recurrent cohort (Fig. 1).In Table 1, the patient and tumor characteristics and treatment are described in the two groups.Forty-two patients (20 %) were offered secondary CRS + HIPEC, whereas 166 patients were selected for palliative chemotherapy or best supportive care.Female sex, PCI at index surgery, disease-free interval, and extra-peritoneal metastases were identified as relevant matching factors according to the relevancy criteria described in the methods section (Table 2).The baseline factors were well-balanced in the propensity score matched cohort except for four factors (age, tumor sidedness, neoadjuvant therapy, operating time), Table 3. Propensityscore matching using the variables above successfully matched 37 cases in the interventional group with 37 palliative cases  (Supplementary Figs.S1-S2).OS in the matched cohort from date of peritoneal recurrence was 38 months (95%CI 30-58) in the secondary intervention group and 19 months (95%CI: 15-24) in the palliative group (HR 0.35 95%CI: 0.20-0.63,p = 0.0004, Fig. 2).The median OS from index surgery in the whole cohort (n = 575) was 41 months (95% CI: 38-45).Four patients in the interventional group were unable to receive the planned intervention (Table 3).

Discussion
In this national, population-based, prospectively registered cohort, the intention to treat a peritoneal recurrence with a secondary CRS and HIPEC (interventional group) was associated with better survival than palliative treatment intent (palliative group) in the unmatched recurrence study cohort (n = 208), 38 vs. 15 months from date of recurrence (p < 0.00001).In the propensity score matched cohort, the survival benefit remained -38 vs 19 months (p = 0.00038).In fact, median survival from date of recurrence after secondary intervention was similar to the survival noted after index surgery among all colorectal patients in the HIPEC registry (38 vs 41 months).Furthermore, patients who underwent index and secondary CRS+ HIPEC had a median survival of 60 months from the index procedure compared to 36 months in patients receiving palliative treatment upon recurrence (Fig. S3).Sensitivity analyses corroborated these results with very little variation.Based on these findings, we suggest that secondary CRS + HIPEC should be considered for all patients with recurrent peritoneal disease in colorectal cancer after initial CRS + HIPEC by evaluation in an MDT.There should be some caution exercised as a secondary CRS + HIPEC should not risk greater morbidity than the index procedure.Thus, the MDT evaluation of the tumor burden and the overall biological behavior of the recurrence need to be assessed.
The inherent selection bias in our study is difficult to completely overcome.To make the groups more comparable, a propensity score matching was undertaken.The four variables used to propensity score match patients were the factors found relevant in a proforma decisionmaking analysis (Table 2).PCI at index CRS and HIPEC, disease-free interval, and absence of extra-peritoneal metastases are known from the literature to be of importance in deciding whether to recommend secondary treatment, which was also found in our study [5][6][7][8][9][10][11][12][13].Somewhat unexpected, we found a sex difference in patients being offered secondary CRS + HIPEC.The distribution was even in the palliative group, but in the interventional group 2/3 were women (Table 2).One may speculate that surgical re-intervention is considered less difficult in the female pelvis and abdomen.Among all patients with a peritoneal recurrence after CRS + HIPEC (n = 208), women had a better prognosis, although not statistically significant in the propensity score matched group (n = 74), Tables 2 and 4. It may be that women's tumor biology, for unknown reasons, is more favorable and that this might be amplified in the relapse setting with longer disease-free intervals.In summary of the decision-making analysis, it appears that patients being offered secondary CRS more often were female, of lower age (although not prognostically significant in the analysis), had lower PCI at index CRS, also lower PCI at secondary CRS compared to their own index PCI, had longer disease-free interval > 12 months, and more often isolated peritoneal recurrence (less oligometastatic).
The PSM model was successful in finding suitable matching pairs for 88 % (37/42) of patients receiving secondary CRS and HIPEC.We believe there is a reasonable comparability between the two PSM groups although there are baseline differences in age, tumor sidedness, neoadjuvant therapy, and operating time (Table 3).We have addressed each of these issues in different ways.Age and tumor sidedness were not prognostic factors in either of the cohorts (Tables 2 and 4).Neoadjuvant therapy was assessed in a separate manner (data not shown) yielding no prognostic impact which is in line with a large recent analysis on neoadjuvant therapy [14].Operating time was not evaluated due to collinearity with PCI.Furthermore, several sensitivity analyses were performed to evaluate the robustness of the propensity score matched survival results.Results were similar, which strengthens the conclusions.Although our propensity score matching was successful without too many unmatched cases (n = 5 or 12 %), there may be unidentified factors, such as co-morbidity and recurrent disease extent, that play an important role in the patient selection process, which is why a comparison with survival from the index procedure is necessary to mitigate this selection bias issue.As mentioned earlier, the second intervention appears to "reset" the survival outcome achieving almost the same median survival as the index procedure achieves (38 vs 41 months).
Overall survival in this study was superior to most previously published series, with or without comparison groups [5,6,[8][9][10][11][12][13].Probably, the main reason for this is the recent advances in palliative chemotherapy for advanced colorectal cancer.To our knowledge, only one other study has compared secondary CRS + HIPEC to palliative chemotherapy for recurrent peritoneal disease in patients who previously have undergone CRS + HIPEC for colorectal cancer PM [7].In that study, which included patients from 2004 to 2015, 8 % of patients previously treated with complete CRS + HIPEC had secondary treatment (34/414), similar to the 7 % rate in our cohort (42/575).Interestingly, OS in the interventional group was basically the same between the two studies (36 vs. 38 months), whereas survival in the palliative group was twice as high in our cohort (7.6 vs. 15 months), who received treatment between 2010 and 2021.Despite the apparent improvement in systemic chemotherapy over time, there was still a significant difference in outcome between patients selected for secondary CRS + HIPEC or palliative chemotherapy.A randomized trial will be difficult to perform (although not impossible with international efforts) and we therefore recommend at least considering patients with recurrent PM for eligibility of secondary CRS+ HIPEC at a dedicated peritoneal surface oncology MDT.Further research is needed to determine which patients may benefit the most from secondary CRS + HIPEC.
In reference to the external validity of this study, the national Swedish HIPEC registry may not contain all patients treated for peritoneal recurrence following CRS + HIPEC.For example, a small isolated peritoneal metastasis may be viewed as a local recurrence and removed

Table 2
Uni and Multivariable Logistic Regression and Cox regression -Decision-making analysis on recurrence cohort (n = 208).Results of OR > 1 meant more likely palliative group while OR < 1 meant more likely offered secondary CRS + HIPEC.HR is calculated from the date of recurrence to death of any cause.A p-value of at least <0.1 in all three categories were required to be included as factor in propensity score matching.in a non-HIPEC hospital without complete abdominal exploration or HIPEC.
When reviewing the literature on systemic chemotherapy treatment of peritoneal metastases, survival ranges from 17 to 25 months, which is similar to our results for the propensity score matched cohort (19 months) [4,[15][16][17].Unfortunately, the registry does not cover systemic chemotherapy treatment following the diagnosis of a peritoneal recurrence after CRS + HIPEC.Therefore, it is unknown how many of the matched patients in the palliative group received systemic chemotherapy, and with what drugs.This limitation means that the palliative group should be considered an intention-to-treat group, which is also the
P.H.Cashin et al.
Odds ratio, HRhazard ratio, PCIperitoneal cancer index.Due to collinearity between PCI and operating time, the PCI was chosen as the more prognostically important factor.

Table 4
Multivariable Cox regression analysis from date of first relapse to endpoint death.