This multicenter international retrospective analysis from 18 centers included 234 patients diagnosed with PSM from EOC, irrespective of the histologic subtype, between July 2015 and March 2020. Eligibility criteria were as follows: adult patients having palliative treatment with PIPAC, recurrent EOC, tumor board approval for PIPAC, and signed surgical informed consent. Patients with extraperitoneal metastases were excluded from this study. Recurrence was defined according to the timing of recurrence. Patients were described as “platinum-sensitive” (PS) if recurrence occurred more than 6 months after the completion of the initial treatment. Early recurrence before 6 months was considered “platinum-resistant” (PR) (13).

Treatment strategies were defined and regularly reassessed during multidisciplinary team (MDT) meetings. Following 3 or at least 2 PIPAC, the morphological and pathological responses were confirmed during the MDT meeting, based on expert radiologists’ and pathologists’ reviews. Morphological response was described according standard and objective radiological response criteria described using the Response Evaluation Criteria In Solid Tumors (RECIST) version 1.1 (14). The types of response described were: complete response, partial response, progressive disease, and stable disease. Pathological response was described according the peritoneal regression grading score (PRGS); no residual cancer cells in all specimens (PRGS 1: complete response), 1 to 49% residual cancer cells (PRGS 2: major response), ≥ 50% (PRGS 3: minor response) and finally no response (PRGS 4) (15).

Eligibility for PIPAC was confirmed after a systematic exploratory laparoscopy done with a peritoneal cancer index (PCI); a sample of ascitis or peritoneal washing for cytology; peritoneal biopsies for histopathological examination; and the sufficient work space for aerosolization of the intraperitoneal chemotherapy. All PIPAC procedures were performed by expert surgeons dedicated to peritoneal metastases management. Every surgeon was specifically trained in PIPAC procedures following published standard practice and safety protocols (8, 16–18). Drugs administered during early experience were cisplatin at 7.5 mg/m2 dosage and doxorubicin at 1.5 mg/m2. Furtherly those dosages were upgraded to respectively 10.5 and 2.1 mg/m2 with supporting safety and encouraging data (12) Postoperative morbidity and mortality were recorded according to Dindo-Clavien classification (19).

Student’s t-test was used for continuous variables, as a parametric test, and the McNemar test for categorical variables, as a non-parametric test. Fisher’s exact test was used for comparisons between the groups. The Mann-Whitney U test was used as a non-parametric test for comparisons between independent variables without a Gaussian distribution and the equality of variance assumption. Univariate and multivariate survival analyses were conducted using Cox model regression. Missing data was handled without imputation. Survival endpoints were defined as the time between the PSM diagnosis date and first PIPAC until death from any cause for overall survival (OS), and disease progression (PFS) expressed by radiological recurrence, symptomatic disease progression or death. The potential impact of PIPAC on OS is further supported by the fact that our study focused on patients with previously controlled disease through systemic chemotherapy, without the presence of extraperitoneal disease. By selecting patients with controlled disease, we aimed to evaluate the additional benefits of PIPAC in a specific subset where the peritoneal cavity remained a significant site of disease burden. The assumption underlying our study is that by targeting and controlling the spread of cancer within the peritoneal cavity, PIPAC may further contribute to improved survival outcomes in this particular context. The hazard ratios (HR) for PIPAC, clinical symptoms, and PCI before PIPAC, and the confounding factors were estimated with 95% confidence interval (95% CI) through the Cox regression multivariate model. The assumption of hazard proportionality over time was confirmed in the selected model. The best regression model was chosen with the literature based known prognostic factors, with the “stats” R package. Survival rates were estimated using the Kaplan-Meier method and compared using the log-rank test. Analysis was performed using RStudio Software (RStudio: Integrated Development for R. PBC, Boston, MA, 2020). Statistical significance was set at a two-sided p-value of < 0.05.

This study was conducted in compliance with international standards for research practice and reporting. Written informed consent was obtained from all included patients. All data were de-identified and anonymized prior to analysis. A retrospective analysis was approved by the local institutional review board of each participating center and was conducted in compliance with the STROBE criteria (www.strobe-statement.org).

A total of 234 consecutive patients were treated with palliative intent for OC. Patients without sufficient data were excluded (n = 20, 9%). After excluding early recurrence, 192 patients (82%) had recurrence after receiving initial treatment, including chemotherapy ± surgery, and 22 patients (9%) were treated frontline after initial chemotherapy and unrespectability. A flow chart of the included patients is shown in Figure S1 ( Supplementary Materials ). Baseline patient characteristics are shown in a comparative cross-table by platinum sensitivity group, with the majority of patients in the platinum-sensitive group (116 patients, 60%). Patients were comparable in terms of comorbidities, performance status, and delay of management between PSM diagnosis and 1st PIPAC cycle. The patients differed in age and primary tumor subtype. Patients in the PS group were older (median 64 vs. 60 years, p = 0.024) and more heterogeneous regarding histologic subtypes compared to the PR group. Further details are provided in Table 1 .

Regarding the number of previous chemotherapy lines or maintenance treatments (bevacizumab) received, PR group (n=28/48, 58%) had more bidirectional treatment in combination with PIPAC cycles compared to PS group (n=27/79, 34%, p = 0.008). About 2/3 and 1/3 of patients had undergone prior PIPAC initiation respectively 3 and 2 lines of chemotherapy. Further details are presented in Table 2 .

Past surgical history analysis showed differences among groups, with a history of CRS greater in the PR group (80% vs. 67%, p = 0.047); however, the PS group showed more cases with a history of HIPEC (12% vs. 1.3%, p = 0.007). The peritoneal burden in the PR group was higher during the PCI evaluation at 1st PIPAC, with a higher median PCI value (16 vs. 24, p < 0.001). The PS group had significantly more PIPAC cycles (median, 3 vs. 2 cycles; p = 0.016). Follow-up after the 3rd PIPAC showed a significant decrease in initial PCI in both groups, with a median of 16 vs. 24 at 1st PIPAC (p < 0.001), and 12 vs. 20 after 3rd PIPAC (p = 0.009), respectively, for the PS vs. PR groups. The results detailed in Table 3 also showed overall surgical morbidity of 19 vs. 28% (p = 0.16), with low open laparoscopy-related morbidity (0.92% vs. 2.7%, p = 0.56), low severe postoperative complications (4.3 vs. 7.9%, p = 0.35), and in-hospital mortality (2.6 vs. 0%, p = 0.28), respectively, for the PS and PR groups. Renal parameters were closely monitored throughout the treatment course, and no instances of renal failure related to cisplatin use were observed for this cohort.

In terms of objective assessment, the morphological evaluation at the end of PIPAC cycles showed only a tendency for more complete responses in the PS group and more stable responses in the PR group (p = 0.16) with a substantial amount of missing data (49.5%). The pathological evaluation showed a significant difference with a higher rate of complete or major response in the PS group (26% and 38% versus 8 and 32% in the PR group, respectively; p = 0.016). The details of the data are presented in Table 4 .

The median follow-up was 8 months (IQR 3-17) vs. 6 months (IQR 2-14) for the PS and PR groups, respectively. The overall population follow-up rate was 86.5%. The reasons for the termination of PIPAC are listed in Table S2 . A small proportion of patients (7-8%) had to withdraw due to surgical access difficulties (multivisceral adhesions). Approximately 38% of the patients with PS and 24% of those with PR completed the planned PIPAC cycles. Between 9% and 10% of patients were eligible for CRS. Roughly 30% of patients received supportive or palliative care. The remaining 2/3 of the patients resumed systemic chemotherapy. Progression at follow-up was documented for 67% of the PS group and 81% of the PR group (see Table S2 in Supplementary Materials ).

Overall survival (OS) Overall survival analysis showed a median of 16 months (95%CI, 12-22). Subgroup OS analysis showed a median of 22 vs. 11 months (PS vs. PR, p = 0.012). The survival rates at 12, 24, and 36 months were 65% vs. 47%, 47% vs. 30%, and 36% vs. 19% for the PS and PR groups, respectively. OS analysis adjusted for the number of PIPACs performed revealed difference in platinum sensitivity, with a greater delta in the PR group (p = 0.002) ( Figure S2 ). In the subgroup analysis, patients with three or more PIPACs showed a longer OS in the PS vs. PR group (median 30 vs. 18 months, p = 0.31). Subgroup with fewer than three PIPACs had longer OS in the PS group (median 17 vs. 5 months, p = 0.051) ( Figure 1A ).

The overall population PFS analysis showed a median of 10 months (95%CI, 9-13). Subgroup PFS analysis showed median of 12 vs. 7 months (PS vs. PR, p = 0.033). Survival rates at 12, 24, and 36 months were 49% vs. 35%, 22% vs. 20%, and 16% vs. 6% for the PS vs. PR groups, respectively. Comparison of PFS between groups adjusted for the number of PIPACs performed showed a significant difference in platinum sensitivity (p = 0.007) ( Figure S3 ). Subgroup analysis with less than 3 PIPACs had median PFS 12 vs. 4 months (p = 0.12), in PS vs. PR-group, respectively. The subgroups with three or more PIPACs were comparable, regardless of platinum sensitivity (median 16 vs. 13 months, p = 0.47) ( Figure 1B ).

The multivariate overall survival analysis is summarized in Figure 2A . The OS forest plot shows the predictive factors adjusted for the key prognostic factors for survival, including platinum sensitivity. The presence of ascitis (HR = 4.02, 95% CI 1.84-8.81, p < 0.001) with positive cytology (HR = 3.91, 1.67-9.14, p = 0.002) at the 1st PIPAC was an independent OS prognostic factor. Performing three or more PIPACs treatments (HR = 0.3, 0.14-0.63, p = 0.002) showed to be an independent OS prognostic factor. The adjusted analysis of the predictive factors of PFS showed the same trends as OS ( Figure 2B ). The presence of ascitis (HR = 5.22, 2.56-10.62, p < 0.001), PCI > 15 (HR = 2.5, 1.2-5.2, p = 0.014), and cytology (HR = 1.960, 1.05-3.67, p = 0.035) were found to be independent unfavorable predictive factors for PFS. The completion of at least three PIPACs (HR = 0.48, 0.27-0.88, p = 0.017) was an independent factor for good prognosis regarding PFS. Detailed univariate and multivariate Cox regression analyses are depicted in Tables S3A , B ( Supplementary Material ).