Effect of oxaliplatin‐based chemotherapy on chemosensitivity in patients with peritoneal metastasis from colorectal cancer treated with cytoreductive surgery and hyperthermic intraperitoneal chemotherapy: proof‐of‐concept study

Abstract Background Chemosensitivity testing, including collagen gel droplet‐embedded culture drug sensitivity test, has proven to be a useful tool in therapeutic decision‐making. This retrospective analysis investigated chemosensitivity testing of peritoneal metastases collected during cytoreductive surgery (CRS), and its impact on survival in patients with colorectal cancer. Methods All patients with peritoneal metastasis from colorectal cancer who underwent CRS with or without hyperthermic intraperitoneal chemotherapy (HIPEC) between November 2008 and October 2014 were included. The growth inhibition rate was expressed as the ratio between the image density after treatment (T) and that before treatment (control, C). Tumours with a reduction in T/C ratio of less than 20 per cent were defined as resistant and those with a reduction of 20 per cent or more as sensitive. Groups were compared for overall (OS) and disease‐free (DFS) survival. Results Of 84 eligible patients, 81 received neoadjuvant chemotherapy (NACT), including 56 patients with an oxaliplatin‐based regimen. Mean(s.d.) follow‐up was 23·4(22·9) months. The median overall survival of all patients was 19·0 (i.q.r. 5·7–36·1) months, with a progression‐free survival time of 10·1 (4·5–17·0) months. Patients who received oxaliplatin‐based NACT had significantly altered chemosensitivity to oxaliplatin; only 20 of 51 such patients showed chemosensitivity to oxaliplatin compared with 16 of 24 who did not undergo oxaliplatin‐based NACT (P = 0·046). However, patients who showed chemoresistance to oxaliplatin had similar OS to those with chemosensitivity (18·8 versus 18·1 months; P = 0·835). The choice of HIPEC agents in patients who received oxaliplatin‐based NACT did not significantly influence survival (oxaliplatin versus mitomycin C: median OS 20·6 (10·9–24·8) versus 19·0 (10·5–34·6) months, P = 0·811; DFS 6·6 (2·8–25·7) versus 9·3 (4·1–13·9) months, P = 0·191). Conclusion Patients who had oxaliplatin‐based NACT showed a higher rate of chemoresistance to oxaliplatin at the time of CRS and HIPEC. The impact of chemosensitivity testing on OS remains unclear and needs further investigation.

However, several aspects in the selection of treatment regimens, dose, dialysate, temperatures and duration of HIPEC have yet to be standardized. Treatment protocols differ between groups and a large number of patients needs to be included in clinical studies for a statistically significant difference to be shown; thus, there are various confounding factors making it difficult to draw firm conclusions.
Chemosensitivity testing has been used to guide treatment for several cancers, including colorectal cancer. The collagen gel droplet-embedded culture drug sensitivity test (CD-DST) is an in vitro anticancer drug sensitivity test in which even small quantities of tumour cells can be cultured successfully, while maintaining the original growth characteristics of the cancer cells. This technique has the ability to evaluate chemosensitivity to various chemotherapeutic drugs at physiological concentrations 10 .
A previous study 11 reported on the clinical efficacy and feasibility of CD-DST in guiding systemic treatment in patients with stage IV colorectal cancer. Reported radiological response rates were 85⋅7 per cent in patients treated with drugs that showed sensitivity on in vitro chemosensitivity testing, compared with 41⋅6 per cent in those treated with drugs that were non-sensitive in vitro (P < 0⋅001). Patients benefiting from optimization of the chemotherapy regimen by means of in vitro drug sensitivity assessments experienced an improved median progression-free survival (PFS) of 23 months and a median OS of 34 months, compared with 10 months (P = 0⋅03) and 17 months (P = 0⋅001) respectively in patients who did not undergo chemosensitivity testing. Other reports [12][13][14][15] have described the value of CD-DST-guided choice of chemotherapeutic agents in gastric, ovarian, lung and breast cancers, leading to a survival benefit for patients.
The aim of this retrospective analysis was to evaluate the chemosensitivity of PMs collected during CRS in patients with colorectal cancer, and to assess its association with survival.

Methods
Chemosensitivity testing (CD-DST) of tumour tissue collected during CRS and before HIPEC has been done routinely since November 2008 in the peritoneal surface malignancy unit of Tokushukai Hospital, Kishiwada, and Kusatsu General Hospital, Shiga, Japan. The unit is a national referral centre for peritoneal malignancies, performing more than 200 cytoreductive procedures per year, and is part of the Japanese/Asian School of Peritoneal Surface Oncology.
All patients with histologically proven PM from colorectal cancer treated with CRS with or without HIPEC, between November 2008 and October 2014 were reviewed. Those who had undergone CD-DST of tumour tissue were included in this study. CD-DST was undertaken for patients whose treatment-related expenses were covered by their health insurance or if it was paid out of pocket. Patients who underwent standard perioperative chemotherapy, and CRS with or without HIPEC, who did not have CD-DST, were excluded from the present analysis.
The database contained demographic, operative (Peritoneal Carcinomatosis Index, blood loss, duration of surgery, completeness of cytoreduction (CC) score) and histopathological data. Neoadjuvant, intraoperative and adjuvant chemotherapy regimens were noted, including dose, number of cycles and therapeutic agents used. Postoperative complications were recorded up to a minimum of 30 days or until the day of discharge. Complications were graded according to the Clavien-Dindo classification 16 . Major postoperative morbidity was defined as grade III and grade IV complications. Follow-up was carried out routinely at the outpatient clinic in the authors' institute, including clinical examination, analysis of tumour markers (carcinoembryonic antigen, carbohydrate antigen (CA) 125 and CA-19⋅9) and CT annually, or earlier if indicated clinically. Clinical follow-up was undertaken every 2 months in the first 2 years and 6-monthly thereafter.
The following drugs and concentrations were used during HIPEC: cisplatin 40 mg/m 2 , oxaliplatin 200 mg/m 2 , mitomycin C 20 mg/m 2 and 5-FU 500 mg/m 2 . HIPEC was administered by an open technique and for a duration  Values in parentheses are percentages. *Five patients in the oxaliplatinand four in the non-oxaliplatin-based neoadjuvant chemotherapy (NACT) group did not undergo chemosensitivity testing for oxaliplatin. †Sixteen patients in the oxaliplatin-and ten patients in the non-oxaliplatin-based NACT group did not undergo chemosensitivity testing for mitomycin C (MMC). ‡Nine patients with and 12 without chemosensitivity to oxaliplatin did not undergo chemosensitivity testing for MMC; three patients with and one without chemosensitivity to MMC did not undergo chemosensitivity testing for oxaliplatin. §χ 2 test.
of 40 min at a temperature of 43 ∘ C. Drugs were chosen at the surgeon's discretion or according to the treatment policy being followed at the time. HIPEC was avoided in patients with a difficult intraoperative course, such as excessive blood loss, or inability to maintain adequate urine output or mean arterial pressures. Bowel anastomosis was performed after completion of HIPEC.

Collagen gel droplet-embedded culture drug sensitivity test
Tumour tissue obtained during CRS and before HIPEC was minced finely with a scalpel and digested using collagenase (Kurabo Industries, Osaka, Japan) 17 . The tumour cells were incubated in a collagen gel-coated flask (CG flask; Kurabo Industries) at 37 ∘ C for 24 h. Viable cells adhering to the collagen gel were used for chemosensitivity testing. A cancer cell suspension was created by adding these viable cells to reconstructed type I collagen solution (Cellmatrix ® Type CD; Kurabo Industries) and three drops of this cell suspension were then added to each well of a six-well plate. Oxaliplatin (0⋅5 μg/ml), mitomycin C (MMC) (2 μg/ml), 5-FU (1 μg/ml) and irinotecan (0⋅03 μg/ml) were added   to separate wells and incubated for 24 h. Not all patients underwent CD-DST for all four drugs, as the number of tumour cells harvested before culture differed significantly between patients. If there were not enough cells, fewer drugs were tested, with the choice depending on the numbers of cells and type of NACT. After 24 h, the medium containing anticancer drug was removed, and each well was rinsed with 3 ml Hank's balanced salt solution and further incubated in PCM-2 medium (Kurabo Industries) for 7 days. Neutral red was added to each well after 7 days of incubation to obtain a final concentration of 50 μg/ml. The colonies in the collagen gel droplet were stained for 2 h, fixed in 10 per cent buffered formalin, washed in water, air-dried and quantified by image analysis [18][19][20] . Image density was also analysed on day 0, before the chemotherapeutic agent had been added.

Analysis of growth inhibition rate
The growth inhibition rate was calculated using image density (total volume of cells observed) on day 7 versus day 0. The growth inhibition rate or the in vitro chemosensitivity effect was expressed as the T/C ratio, where T represents the image density of the treated group and C the image density of the cells on day 0 (considered as the control group). Tumours with a reduction in T/C ratio of below 20 per cent were considered resistant and those with a value of 20 per cent or more as sensitive. Although T/C ratios of 60 or 50 per cent were taken to indicate sensitivity in a previous study 11 of metastatic colorectal cancer, these tests were carried out on chemotherapy-naive cells, whereas patients in the present study received preoperative chemotherapy.

Statistical analysis
Continuous and categorical data were compared using the t test and χ 2 test respectively. OS was calculated from the date of surgery (CRS and HIPEC) to the date of death or last follow-up. Disease-free survival (DFS) was defined as the time until tumour recurrence, and calculated for patients undergoing CC0/1 procedures, from the date of surgery (CRS and HIPEC) to the date of recurrence or death. PFS was defined as tumour progression (in patients undergoing CC2 or CC3 procedures) or tumour recurrence (in patients having CC0-1 surgery) and calculated from the date of surgery (CRS and HIPEC) to the date of tumour progression (CC2 or CC3), tumour recurrence (CC0-1) or death. Survival analysis was done using the Kaplan-Meier method, with differences between groups evaluated by means of the log rank test. P < 0⋅050 was considered statistically significant. All statistical analyses were performed using the SPSS ® software version 22 (IBM, Armonk, New York, USA).

Results
Of 190 patients who underwent CRS with or without HIPEC during the study interval, 84 (44⋅2 per cent) Baseline demographic, clinicopathological, and treatment characteristics are shown in Table 1. The most common histopathological subtype was adenocarcinoma (54 of 83, 65 per cent) and sigmoid colon was the most common site of the primary tumour (24 of 83, 29 per cent). Complete macroscopic cytoreduction (CC0-1) was achieved in 64 patients (76 per cent). The postoperative major morbidity rate was 22 per cent and the postoperative mortality rate 2 per cent.

Chemosensitivity
The mean(s.d.) reduction in tumour density was 19⋅3(11⋅5) per cent for oxaliplatin and 24⋅7(15⋅6) per cent for MMC. Analysis of the chemosensitivity of tumour tissue to oxaliplatin revealed that patients who received oxaliplatin-based NACT had significantly altered chemosensitivity; tissue from only 20 of 51 patients who received oxaliplatin-based NACT was sensitive to oxaliplatin, compared with 16 of 24 who did not have oxaliplatin-based NACT (P = 0⋅046). In contrast, patients treated with or without oxaliplatin-based NACT had similar results in terms of chemosensitivity to MMC (20 of 40 versus 11 of 18; P = 0⋅571). Analysis of chemosensitivity to both agents (MMC and oxaliplatin) showed that tumour tissue that was not chemosensitive to oxaliplatin was also likely not to be chemosensitive to MMC, although the effect was not significant (P = 0⋅414) ( Table 2). A comparison of demographic, operative and therapeutic factors in groups with chemosensitivity and chemoresistance to oxaliplatin is shown in Table 3. These groups differed significantly only in terms of the rate of complete macroscopic cytoreduction (22 of 36 patients with oxaliplatin-sensitive tumours compared with 33 of 39 in the oxaliplatin-resistant group; P = 0⋅035). Comparison of similar factors between patients who did or did not receive oxaliplatin-based NACT showed no significant differences ( Table 4).

Discussion
This single-institution observational study of prospectively collected data from a high-volume centre focused on the impact of chemosensitivity testing on the choice of postoperative chemotherapy and the effect of intraoperative HIPEC. The results are important, especially in the context of the negative findings of the PRODIGE 7 RCT 21 , in which patients who had CRS for PM from colorectal cancer were randomized to HIPEC or no HIPEC. This French multicentre trial showed no significant difference between patients treated with CRS and HIPEC and those who had CRS alone. Despite the main shortcoming of an underpowered analysis owing to the underestimated OS in the control group (without HIPEC), one major doubt concerns the duration and dose of HIPEC. As preclinical data are limited to a few publications in experimental models 22 , clinical observations about chemosensitivity and therefore potential optimization of the chemotherapeutic drug seem promising.
The present study was able to show, as a proof of concept, that chemoresistance to oxaliplatin might be influenced by the preoperative chemotherapy regimen. Patients who had oxaliplatin-based NACT had significantly higher rates of chemoresistance to oxaliplatin than patients who did not receive oxaliplatin-based NACT. Previous investigations have reported the development of chemoresistance in cell lines after treatment with oxaliplatin, through molecular mechanisms such as upregulation of microRNA (mir 203), and downregulation of ataxia telangiectasia mutated mRNA and protein levels 23 . A recent study 24 demonstrated that the induction of circCCDC66 is dependent on treatment with oxaliplatin and is required for the establishment of chemoresistance to this agent. However, another study 25 reported on the ex vivo activity of chemotherapeutic drugs in patients with PM from various cancers,  including colorectal cancer, using a different method for drug testing sensitivity, the fluorometric microculture cytotoxic assay, and the authors demonstrated that oxaliplatin was equally active in chemotherapy-naive and previously treated patients. Data focusing on the optimization of chemotherapy regimens guided by chemosensitivity testing are seen as controversial, as increased response rates do not necessarily translate into increased survival. A recent study 26 revealed no survival benefit for patients with unresectable colorectal cancer and CD-DST-guided first-line chemotherapy compared with patients who received standard first-line chemotherapy in a cohort of 120 patients. Interestingly, the subgroup of poor responders to chemotherapy showed significantly improved survival if treated with CD-DST-guided first-line chemotherapy. Despite this limited evidence, there is a considerable lack of clinical trials, which are necessary to support a tailored chemotherapy strategy.
Although patients appeared to develop chemoresistance during oxaliplatin-based NACT, the clinical outcome, especially OS, was not affected by this finding. Interestingly, neither the choice of HIPEC (oxaliplatinversus non-oxaliplatin-based), nor receiving oxaliplatin-based NACT, had an effect on OS. These results might be explained by several factors, including small sample size. First, the chemoresistant group had a significantly higher rate of complete cytoreductions. As the completeness of cytoreduction is one of the most important factors predicting patient outcome and therefore OS 27 , inferior OS would be expected for the group showing chemosensitivity to oxaliplatin; however, this was not the case, suggesting that survival would hypothetically be better in this group after equal surgical treatment. Second, the authors' clinical practice of choosing the adjuvant chemotherapy regimen according to the results of chemosensitivity testing might have had an influence on outcomes, as patients in whom chemoresistance was detected were treated with adjuvant drugs that attacked the tumour precisely owing to its sensitivity. It is known that chemotherapy with the choice of drugs influenced by chemosensitivity testing is more effective than a rather blind choice of regimen 11 . Third, the effect of the drug distributed during HIPEC may have a minor effect on OS. This hypothesis is supported by the negative results of PRODIGE 7 21 , in which OS was similar for patients with PM from colorectal cancer treated with CRS versus CRS and HIPEC. Unfortunately, no subgroup analysis comparing patients treated with or without NACT is yet available. In contrast to this hypothesis, another RCT 28 proved the superiority of CRS and HIPEC over intravenous chemotherapy, and further large retrospective trials 4,29 have also shown increased OS among patients selected for CRS and HIPEC. A well conducted clinical study with a sample size calculation may help in understanding the effect of chemoresistance due to NACT on OS.
This study has a few limitations. The cut-off of 20 per cent reduction in cell numbers after 7 days has not been validated previously, as the majority of studies of CD-DST used either immortalized cell lines or chemotherapy-naive tumour tissue. The established cut-off values of 50-60 per cent are based on these colorectal cell types, and represent a different situation with a different opportunity for chemotherapeutic agents to act. Studies of different tumour entities, for example pancreatic cancer, revealed an optimal tumour reduction of 15 per cent indicating overall recurrence, with sensitivity and specificity of 61⋅1 and 100 per cent respectively 30 . The range of cell density in the present assay, with a mean reduction in tumour density of 19⋅3 and 24⋅7 per cent for oxaliplatin and MMC respectively, was lower than that in previous series 11 . Therefore, changing the cut-off value from 50 to 20 per cent for the pretreated tumour tissue seemed necessary, although it was not based on existing evidence. Another limitation is the CD-DST assay used and the concentration of chemotherapeutic drugs. An incubation period of 24 h before removal of the agent and culturing for an additional 6 days was used. This protocol was described previously and validated for 5-FU, irinotecan and oxaliplatin in a group with stage IV colorectal cancer 11 , but differs from other published protocols 31 . The concentrations of cytotoxic agents reported here were approximately ten times lower for MMC (2 versus 12-25 μg/ml) and about 300 times lower for oxaliplatin (0⋅5 versus 160-330 μg/ml), which is justified by the cytotoxicity associated with the longer incubation time (24 h versus 30 min). In addition, possible drug combinations in the CD-DST assay were not assessed, and only single-agent incubations were tested, as intraperitoneal MMC or oxaliplatin comprise the majority (76 per cent) of reported HIPEC regimens, and combinations of MMC with cisplatin or oxaliplatin with irinotecan are less common 32 . Furthermore, ex vivo testing has a sensitivity and specificity of 90 and 70 per cent respectively, which might also be influenced by the sample of PM chosen, as well as tumour regression after preoperative chemotherapy 33 . Finally, this study did not show a significant impact of chemosensitivity testing on OS, but the present findings provide an incentive to perform a clinical trial with a sufficient sample size to substantiate the hypothesis generated and confirm these clinical findings.

Disclosure
The authors declare no conflict of interest.