Patients with PC treated with HIPEC at Changzhou Second People’s Hospital affiliated with Nanjing Medical University in Jiangsu Province, China, between November 2018 and January 2020 were eligible for study participation. The inclusion criteria were as follows: PC confirmed by cytology and pathology; Eastern Cooperative Oncology Group (ECOG) score 0-2; no treatment with systemic chemotherapy or local radiotherapy in the previous month; and no history of HIPEC treatment. The main exclusion criteria were as follows: presence of any serious cardiopulmonary disease; presence of significant liver and kidney insufficiency; presence of acute intestinal obstruction; and presence of other contraindications for HIPEC. Based on these criteria, 19 patients were included in the final study population. All patients signed a written consent form before the study began. The research protocol was approved by the institutional Ethics Committee of Changzhou Second People’s Hospital.

Each of the study patients was treated with HIPEC 4 times within a 2-week time period, with each treatment session lasting 90 minutes. The perfusion drugs used were raltitrexed (2.5 mg/m²; 1 treatment session), docetaxel (25 mg/m²; 1 treatment session), and oxaliplatin (40 mg/m²; 2 treatment sessions). The agents were infused at an intraperitoneal temperature of 43°C.

Peripheral blood and ascites samples were collected from each patient within 1 week before HIPEC, and peripheral blood was collected within 1 week after HIPEC (all patients’ ascites resolved after HIPEC). Ascitic fluid (≥10 mL) was centrifuged at 600 g for 10 min to separate the supernatant and the precipitate. Peripheral blood (>10 mL) was collected in cell-free DNA (cfDNA) blood collection tubes (Streck, Omaha, NE, USA) at room temperature. Within 4 hours, plasma was separated from the blood samples by centrifugation at 1,600 × g at 4°C for 10 min; the supernatants were then centrifuged at 16,000 × g at 4°C for 10 min. Plasma, peripheral blood lymphocytes, ascites supernatant, and ascites precipitate were retained and stored at –80°C before extraction of cfDNA and genomic DNA (gDNA).

cfDNA from plasma and ascites supernatants was purified using the Circulating Nucleic Acid Kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol. The DNA concentration was measured using the Qubit dsDNA HS (High Sensitivity) assay kit in the Qubit fluorometer (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA). To test the DNA integrity, 200 ng of extracted DNA was loaded onto the 1% agarose gel with λ-Hind III digest DNA marker (Takara Biotechnology Co., Ltd., Dalian, China). The DNA samples that were longer than the second largest bonds (9,416 bp) of λ-Hind III digest DNA marker were considered integrated samples and used for subsequent analysis.

Tumor DNA indexed NGS libraries were prepared using the DNA Library Preparation Kit for MGISeq-2000 (BGI, Shenzhen, China). All libraries were hybridized to custom-designed biotinylated oligonucleotide probes (IDT, Coralville, IA, USA) covering 382 genes. DNA sequencing was performed using the MGISeq-2000 Sequencing System (BGI, Shenzhen, China) per the manufacturer’s guideline; this generated 3 GB of data from tumor DNA.

PyClone was used to analyze the population structure of ctDNA clones collected longitudinally from each patient. Information about single nucleotide variants (SNV) and copy number variations (CNVs) was used as input for PyClone analysis (16, 17), which was performed using 20,000 iterations and default parameters. The cluster with the greatest mean variant allele frequency (VAF) value was identified as the dominant clonal cluster, and the mutations in this cluster were considered clonal mutations. Other clusters and mutations were considered subclones. The sum-VAF was calculated as the total of VAF in each patient. Change in the VAF of the dominant clone mutation before and after HIPEC was calculated as follows: △VAF = baseline VAF – post-HIPEC VAF.

The molecular tumor burden index (mTBI) was determined using the mean VAF of clonal mutations. The change in mTBI was calculated as follows: △mTBI = baseline mTBI – post-HIPEC mTBI. This value was used to describe the degree to which the mutations of a patient were cleared after treatment.

The curative efficacy of HIPEC was evaluated according to the 1981 WHO evaluation standard of curative effect, with brightness mode ultrasound (B-US) or computed tomography (CT) performed every 6 to 8 weeks during follow-up to detect the amount of ascites. Complete response (CR) was defined as the complete resolution of ascites lasting for >4 weeks; partial response (PR), a decrease of >50% in ascites lasting for >4 weeks; stable disease (SD), a decrease of <50% in ascites or no change in ascites; and progressive disease (PD), an increase in ascites. The objective response rate was defined as complete response + partial response.

Ascites progression-free survival (ascites PFS) and overall survival (OS) were the clinical endpoints in this study. Ascites PFS and OS were defined as the time from the start of treatment to ascites progression or death, respectively, or the time to last follow-up. The cutoff date for the analysis was December 31, 2020; all patients who had no progression in ascites or were alive as of then were censored on that date unless their date of last follow-up was earlier, in which case that date was used for censoring.

Data were reported as median and 95% confidence intervals (95% CIs) or ranges. The Wilcoxon rank-sum test was used to detect differences in mTBI changes before and after HIPEC treatment. Spearman correlation analysis was used to test the correlation between two variables (eg, △mTBI and ascites PFS). Ascites PFS and OS were analyzed using the Kaplan–Meier method. IBM SPSS software (V23.0) and GraphPad Prism (V8.2) were used for analysis. P < 0.05 was defined as statistically significant.

The clinical characteristics of the 19 study patients are shown in Table 1. The mean age was 57 years, and 73.68% of the patients were female. The following primary tumors were observed: ovarian cancer, 8 (42.11%); gastric cancer, 5 (26.31%); appendix cancer, 2 (10.53%); rectal cancer, 1 (5.26%); endometrial cancer, 2 (10.53%); and pancreatic cancer, 1 (5.26%) (Table 2). Four patients (21.06%) had only peritoneal metastasis; the remaining 15 patients (78.94%) had metastases in other organs, as well.

The median follow-up among study patients was 9 months (range, 6–14 mo). At 1 month after HIPEC, the ORR was 100%. The median ascites PFS was 5.13 months (95% CI: 3.75–6.15 mo), and the median OS was 9.23 months (95% CI: 7.22–10.59 mo).

Information about the samples collected is shown in Table 2. Among the 19 patients, ascites was collected from 12 patients at baseline (samples were not collected from 5 patients, and 2 patients had no ascites at baseline). After HIPEC, ascites were resolved in all patients, so samples could not be collected. Baseline plasma samples were collected from all 19 patients, and post-HIPEC serial plasma samples were collected from 14 patients. A total of 37 functional mutations were detected in 7 of the 12 ascites samples collected at baseline (detection rate, 58.33%). A total of 51 functional mutations were detected in 10 of the 14 plasma samples collected at baseline (detection rate, 71.43%). A total of 38 functional mutations were detected in 8 of the 14 plasma samples collected after HIPEC (detection rate, 57.14%). In addition, 27 CNVs were detected in ascites samples and 4 CNVs were detected in baseline plasma samples. The most common CNVs were in MYC, RECQL4, KRAS, and EGFR.

Among the 14 patients with serial plasma samples, no gene mutations were detected in the plasma samples obtained from 3 patients (P1, P16, and P18) at baseline and after HIPEC, so the mTBI of the two nodes was 0 for these patients. Thus, a gene mutation heatmap based on serial plasma samples from the remaining 11 patients is presented (Figure 1A). For these 11 patients, the change in mTBI from baseline to after HIPEC (△mTBI) was significant (Wilcoxon, P = 0.026). In these patients, the median ascites PFS was 3.35 months (95% CI: 2.34–5.13 mo) and the median OS was 5.93 months (95% CI: 4.93–11.17 mo). There was a significant positive correlation between △mTBI and ascites PFS (Spearman r = 0.673; P = 0.023) (Figure 1B). The △mTBI was moderately positively correlated with OS (Spearman r = 0.510; P = 0.109) (Figure 1C), but the correlation was not significant.

The baseline median mTBI for the 19 study patients was 0.67. Based on this value, the patients were divided into a high mTBI group (≥0.67; n = 10) and a low mTBI group (<0.67; n = 9). Survival analysis demonstrated that ascites PFS was longer in the low mTBI group (median ascites PFS, 7.20 mo; 95% CI: 5.23–8.77 mo) than in the high mTBI group (median ascites PFS, 2.46 mo; 95% CI: 1.50-4.12 mo; HR = 0.157; 95% CI: 0.046–0.540; P = 0.003) (Figure 2A). Similar results were observed for OS, with patients in the low mTBI group having a longer OS (median OS, 11.38 mo; 95% CI: 9.23–13.13 mo) than those in the high mTBI group (median OS, 4.97 mo; 95% CI: 3.24-10.00 mo; HR = 0.296; 95% CI: 0.109–0.804; P = 0.017) (Figure 2B).

To assess the correlation between efficacy and changes in VAF and tumor markers, we examined changes in the primary clonal VAF and the most significant tumor markers (eg, CEA, CA125) in 11 patients (Figure 3A). Four patients (P5, P6, P9, and P14) were found to have inconsistent trends in VAF and tumor markers during the same period. The △VAF was moderately positively correlated with ascites PFS (Spearman r = 0.588; P = 0.057) (Figure 3B), and slightly positively correlated with OS (Spearman r = 0.386; P = 0.241) (Figure 3C), but these results were not significant. There was no correlation between the △ tumor markers and ascites PFS (Spearman r = 0.091; P = 0.790) or OS (Spearman r = 0.287; P = 0.396). Both of these factors were less correlated with efficacy than mTBI.

Because the highest VAF does not accurately represent the overall tumor burden, we also analyzed the correlation between the change in sum of VAF (△sum-VAF = baseline sum-VAF – post-HIPEC sum-VAF) and ascites PFS and OS. There was a significant positive correlation between △sum-VAF and ascites PFS (Spearman r = 0.945; P < 0.001) (Figure 4A) and between △sum-VAF and OS (Spearman r = 0.866; P = 0.001) (Figure 4B).

Among the 12 patients with baseline ascites samples and both baseline and post-HIPEC plasma samples, 3 patients (P10, P11, and P12) had negative test results (no mutation was detected) for both samples. For the remaining 9 patients, VAF values for the ascites and plasma samples were compared (Figure 5A). No mutation was detected in the plasma samples of 2 patients (P1 and P8), and no mutation was detected in the ascites samples of 2 patients (P7 and P9). A common mutated gene (Figure 5B) was observed in both samples from 5 of the 9 patients. In addition, there was a significant positive correlation between the ascites and plasma comutation gene VAF (Spearman r = 0.794; P = 0.001) (Figure 5C).

We calculated the number of comutations/the total number of mutations in plasma samples as the rate of ascites entry and assessed whether there was a correlation with ascites PFS and OS. The ascites entry rate was negatively correlated with ascites PFS (Spearman r = –0.647; P = 0.059) (Figure 5D) and OS (Spearman r = –0.638; P = 0.064) (Figure 5E), but the correlation was not significant.