Hepatic Arterial Infusion Chemotherapy With Folfirinox or Oxaliplatin Alone in Metastatic Colorectal Cancer

Background Hepatic arterial infusion (HAI) of chemotherapy is an option for the treatment of patients with liver metastases from colorectal cancer (LMCRC). Though HAI with oxaliplatin (HAI-Ox) is generally used, intravenous (IV) 5-fluoro-uracil (5FU)-oxaliplatin-irinotecan HAI (HAI-Folfirinox) is feasible and leads to curative-intent surgery in 30% of pretreated patients. We compared the efficacy and safety of HAI-Ox and HAI-Folfirinox. Methods Patients who underwent HAI chemotherapy for LMCRC were retrospectively included from 2008 to 2019 from six French expert centers. Results Data were collected from 273 previously treated patients with LMCRC. Patients received HAI-Folfirinox (n = 52) or HAI-Ox (n = 221) combined with IV chemotherapy. The objective response rate (ORR) was 43.2% in patients with HAI-Folfirinox and 45.9% (ns) in patients with HAI-Ox. Median overall survival (OS) was 17 months (95% CI: 15–32.3) with HAI-Folfirinox and 26.2 months (95% CI: 19.4–34.4; p = 0.1) with HAI-Ox. Median progression-free survival (PFS) was 7.9 months (95% CI: 4.9–10.3) with HAI-Folfirinox and 6.4 months (95% CI: 6.0–7.7; p = 0.6) with HAI-Ox. The secondary liver resection rate was 35.6% with HAI-Folfirinox and 16.7% with HAI-Ox (p = 0.007). Grade 2 and above toxicities were significantly more frequent with HAI-Folfirinox. In the global population, only 2 factors were prognostic for OS in multivariable analyses: liver-only disease [hazard ratio (HR): 0.4; 95% CI 0.20–0.83; p = 0.013] and local complications of the catheter (HR: 3.8; 95% CI 1.6–9.0; p = 0.002). Conclusion Hepatic arterial infusion results in high response rates, secondary resections, and long survival in pretreated patients with LMCRC.


INTRODUCTION
Colorectal cancer (CRC) management raises new challenges as its incidence increases worldwide. As the first metastatic site, hepatic progression contributes to death in half of the patients with metastatic disease (1). This trend has prompted the medical community to elaborate more aggressive systemic and locoregional approaches even after several lines of treatment (2,3). Although hepatic arterial infusion (HAI) chemotherapy relies on a robust biological rationale, its development has been initially restricted by two limitations: first, the complexity of surgical removal of the catheter with a significant local complications rate and, second, extra-hepatic progressive lesions excluding patients from an HAI loco-regional treatment approach (4). With technical advances in catheter insertion through interventional radiology procedures (5), HAI chemotherapy has become an effective option for the treatment of patients with liverdominant/exclusive disease. Moreover, the global strategy for the management of these patients has evolved, and HAI is now mostly used in combination with active systemic treatment, allowing better control of extrahepatic disease when limited (4).
The multicenter phase II trial, OPTILIV (11) has evaluated an intensified regimen with 5FU, oxaliplatin, and irinotecan administered intra-arterially combined with IV cetuximab in RAS wild-type patients with liver metastases from CRC after the failure of IV chemotherapy (12). About 30% of these heavily pretreated patients finally underwent hepatic resection. This regimen was associated with non-negligible grade 3-4 toxicities, in particular neutropenia (42.6%). To our knowledge, there is no study comparing HAI with oxaliplatin (HAI-Ox) and HAI with Folfirinox (HAI-Folfirinox) in patients with nonresectable LMCRC. In this multicenter retrospective study, we looked at HAI performance in pretreated patients with liverdominant/exclusive mCRC and compared IV chemotherapy or targeted therapy combined with HAI-Ox or with HAI-Folfirinox. We assessed the efficacy and safety of these administration modalities in patients with disease refractory to standard therapies.

Patients
Patients who underwent HAI chemotherapy for liver metastases from pathologically proven colorectal cancer refractory to at least one standard IV regimen were retrospectively included from 2008 to 2019. Data were collected from routinely treated patients with HAI chemotherapy of six expert centers.
The study was performed according to the principles of the Declaration of Helsinki.

Treatments
Hepatic arterial infusion treatment was delivered through a catheter, which was implanted percutaneously by an interventional radiologist. This catheter was linked to an implantable port, intra-arterial catheter patency and hepatic artery skeletonization or occlusion were checked by contrast medium opacification before treatment initiation and every two courses of treatment. HAI-Ox was delivered at the dose of 85 mg/m 2 (86.5%) or 100 mg/m 2 (13.5%) in 2 h and IV chemotherapy with or without targeted therapy according to RAS status (bevacizumab or cetuximab). HAI-Folfirinox was based on the OPTILIV study protocol (11) as follows: 5-FU 2,800 mg/m 2 , irinotecan 180 mg/m 2 , and oxaliplatin 85 mg/m 2 . HAI-Folfirinox was administered alone or in association with IV bevacizumab or cetuximab according to RAS status.

Data Collected
Data were collected retrospectively from electronic medical records. Included were demographic data (age and gender), disease characteristics (date of diagnosis of colorectal cancer and metastatic status, sites of metastases, number and distribution of liver metastases, and WHO PS), molecular data (KRAS, RAS, BRAF, and microsatellite instability status), laboratory parameters (bilirubin, ACE CA19-9 before treatment and during follow-up), previous treatments (surgical resection, radioablation, and previous systemic treatment lines), concomitant anti-cancer treatment (IV chemotherapy and targeted therapies), tolerability data (treatment-related toxicities, dose reductions, and treatment stop), and response to treatment and survival (objective hepatic and extra-hepatic response, date of hepatic and extra-hepatic progressions, last date of follow-up, and status at last follow-up).

Statistical Analysis
Progression-free survival (PFS) was defined as the time elapsed from the beginning of HAI treatment until progression (any site for global-PFS, hepatic for liver-PFS) or death (all causes). Overall survival (OS) was defined as the time elapsed from the beginning of HAI treatment until death (all causes).
Survival curves were estimated using the Kaplan-Meier method. Median follow-up and the 95% confidence intervals (CI) were calculated using the reverse Kaplan-Meier method.
The statistical association of patients with tumor-related characteristics and treatment with survival was first assessed by the univariate Cox proportional hazards model. Parameters with values of p < 0.10 in univariate analysis and/or clinically relevant variables were entered into the multivariable Cox-regression model. Values of p < 0.05 were considered statistically significant. All analyses were performed using R software version 2.15.2 (R Development Core Team, Vienna, Austria).

Patient Characteristics
We included 273 patients previously treated by one or more chemotherapy protocols (Supplementary Table 1). Among them, 52 (19.0%) received HAI-Folfirinox and 221 (81.0%) received HAI-Ox. The tumor molecular profile was only available for a subset of patients (109/273; 39.9%) according to the study period. KRAS status was available in 39.2% of patients. KRAS mutations were similar in the HAI-Folfirinox and HAI-Ox groups (31.6% vs. 43.0%; p = 0.124). BRAF mutations represented 7.4% of the 122 tumors analyzed and microsatellite instability represented 7.3% of the 69 tumors analyzed. The disease was liver-limited in 69.5% of patients and liver-dominant in the remaining 30.5%. Peritoneal, lung, and lymph node metastatic lesions were described in 5.6, 20.7, and 4.2% of patients, respectively, with no significant difference between the two groups ( Table 1). All patients were previously treated with 5-FU. Most patients had previously received irinotecan (231/273; 84.9%) and oxaliplatin (231/273; 84.9%). HAI chemotherapy was proposed as a second-line or third-line treatment in 42.9 and 32.2% of cases, respectively.

Treatment
Patients in the HAI-Folfirinox group received a median of 4 cycles (range: 1-27) and patients in the HAI-Ox group received a median of 5 cycles (range: 1-17). HAI treatment was stopped for progression, unacceptable toxicity, failure of HAI, or upon patient decision. More patients in the HAI-Folfirinox group received only a short course of HAI chemotherapy (< 4 infusions in 37.2% as compared to 29.6% in the HAI-ox group; Table 2).
Concomitant IV cytotoxic chemotherapy was administered in 96.3% of patients in the HAI-Ox group and in no patients in the HAI-Folfirinox group. IV-targeted therapies were more frequently administered with HAI-Folfirinox (86.5%) than with HAI-Ox (37.3%), in particular for anti-epidermal growth factor receptor (EGFR; Table 2).

Tolerability
During treatment, a dose reduction was performed in 62.5% of patients in the HAI-Folfirinox group and in 34.1% in the HAI-Ox group (p = 0.001). Before HAI treatment, 91% of patients presented a grade of 0 or 1 neuropathy. During treatment, grade ≥ 2 neuropathy appeared in 20.4% of patients in the HAI-Folfirinox group and in 29.3% of patients in the HAI-Ox group (ns).

Prognostic Factors in All Patients on Hepatic Arterial Infusion Treatments
In univariable analysis, parameters associated with poor OS were WHO-PS 2-3, peritoneal metastases, previous administration of irinotecan, bevacizumab treatment before HAI treatment, more than two previous treatment lines, and local complications related to the HAI catheter. Parameters associated with prolonged OS were liver metastases only and secondary hepatic resection (Supplementary Table 3).
In multivariable analysis, only 2 factors were associated with OS (Supplementary Table 3): the liver-only disease was associated with a prolonged OS (HR: 0.40; 95% CI 0.20-0.83; p = 0.013) and local complications related to HAI administration of the treatment were associated with a reduced OS (HR: 3.8; 95% CI 1.60-9.02; p = 0.002).

DISCUSSION
This study retrospectively compared HAI-Folfirinox to HAI-Ox in 273 patients with non-resectable liver metastases of colorectal cancer. Treatment was feasible and well tolerated in these patients with disease refractory to at least one line of treatment. Most of them had previously received IV oxaliplatin, irinotecan, and targeted therapies (bevacizumab or cetuximab). Treatment duration, PFS, and OS were similar between the two treatment modalities. HAI-Folfirinox was associated with significantly more secondary hepatic resections and more local complications linked to the catheter and treatment-related grade ≥ 2 toxicities, such as neutropenia, diarrhea, and vomiting. In multivariable analysis in the whole population receiving HAI, liver limited disease was associated with a prolonged OS whereas local complications were associated with shorter OS.

Hepatic Arterial Infusion Results in Longer Overall Survival
In our study, patients on HAI chemotherapy had a median OS of 15.9 months in second-line treatment and 11.2 months beyond the second line, as compared to reported OS of 13 months (13, 14) and 6 months, respectively, for IV treatment in the literature (15,16). As for the ORR, we report a lower ORR (45.3%) than Ducreux et al. (6) (64%; 95% CI, 44-81%) and Boige et al. (9) (55%; 95% CI, 40-69%). In both studies, patients in second-line therapy were treated with HAI-Ox combined with IV 5-FU with a median of 8 or 9 cycles received. In Ducreux et al., they had not previously received oxaliplatin. It is worth noting that we analyzed a much larger population of non-selected patients with the extrahepatic disease (30.5%), pretreated with targeted therapies, and cytotoxic chemotherapies, which may account for the lower ORR than expected. Altogether, our results promote HAI as an interesting option in LMCRC beyond first-line treatment. HAI treatment appeared well tolerated, especially when administered with oxaliplatin only. A randomized trial specifically designed to compare HAI treatment to IV chemotherapy in patients with non-resectable LMCRC after induction treatment is currently ongoing (17).

Hepatic Arterial Infusion to Increase Secondary Resection Rate
The use of HAI-Folfirinox relied on the resection rate of 16.7% and reported for patients treated with HAI-Folfirinox after two or three lines of treatment in the prospective OPTILIV trial (11). In terms of secondary resection, we noted a significantly higher secondary resection rate in patients with HAI-Folfirinox (35.6%). Our population included 42.9% of patients in a second-line setting, but the difference in secondary resection rate achieved was even greater between HAI-Folfirinox and HAI-Ox in third and further lines when compared with the second-line setting. Patients in the HAI-Ox group more often had extra-hepatic lesions (33.0% vs. 19.6%). The most frequent extra-hepatic lesions reported were lung metastases, which are less life threatening than peritoneal metastases. The peritoneal disease is rarely reported in this series: 0% in the HAI-Folfirinox group and 6.2% in the HAI-Ox group. The low number of patients concerned, resulting in a lack of statistical power, may account for the nonsignificance of the differences observed. In contrast with the OPTILIV trial, secondary surgical resection did not emerge as a predictor of longer OS in this work. HAI-Folfirinox induced more toxicities than HAI-Ox and more than one-third of patients received less than 2 months of treatment. HAI-Folfirinox did not significantly improve median OS as compared to HAI-Ox. These results suggest that resection of hepatic metastases should not be the predominant parameter used to guide treatment decision after second-line treatment. The increasing life expectancy of patients with LMCRC justifies randomized trials with HAI to conclude with certainty that the resection rate correlates with OS (OSCAR NCT02885753, SULTAN NCT03164655).

Hepatic Arterial Infusion and Other Local Treatments
Other local treatments, such as chemoembolization with drugeluting beads loaded with irinotecan (DEBIRI) (18) or selective internal radiation therapy (SIRT) (19), showed better OS than IV chemotherapy in patients pretreated for LMCRC. First attempts at comparing SIRT and HAI chemotherapy to set up the best treatment sequences have been published. They ruled in favor of HAI chemotherapy, which appeared as a predictor of longer OS (31.2 vs. 16.3 months; p < 0.001) when compared to SIRT (20). In addition, HAI is technically difficult after internal SIRT, which may affect the liver vasculature (21). Therefore, there is a need for HAI chemotherapy optimization. In this regard, previous reports suggest that several lines of HAI chemotherapy could be proposed before switching to another therapeutic strategy (22).

Limits and Strengths
The retrospective and non-randomized nature of our study limits the strength of the conclusions that can be drawn from this work. Especially, the period of inclusion from 2008 to 2019 accounts for a significant percentage of missing data in laboratory analyses. BRAF and RAS testing in France was recommended in 2014 and implemented in 2015. We were therefore not able to adjust data for RAS and BRAF status. In addition, the heterogeneity of previous lines of treatment and concomitant treatments to HAI may have introduced bias. Especially, the concomitant administration of cetuximab was unbalanced between the HAI-Folfirinox and HAI-oxaliplatin groups (78.4% vs. 23%, respectively). Although one of the largest series reported in HAI-treated patients, it remained too small for efficient application of a propensity score. These results show the importance of conducting randomized prospective studies in the field, which is the only method to build comparable treatment groups and determine the best HAI modalities.
Nevertheless, we report here, to our knowledge, the largest series of HAI-treated patients after failure of a first-line IV chemotherapy and the first series comparing two HAI regimens. As HAI chemotherapy is of particular interest in achieving resection of liver metastases (23), this study provides evidence to help digestive oncologists and surgeons to adapt the modality of HAI treatment to their care objectives.

CONCLUSION
Hepatic arterial infusion treatment is an effective option in patients with LMCRC refractory to previous lines of treatments. Intensifying HAI treatment led to a higher rate of secondary hepatic resection with no significant impact on OS and PFS. Our study suggests that when resection is the main objective of the treatment, intensification of HAI can be proposed for selected patients.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author, upon reasonable request.

ETHICS STATEMENT
Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

AUTHOR CONTRIBUTIONS
VR completed the database, wrote, and revised the manuscript. SP initiated the project and the database. BS, DS, AL, YT, CG, AT, SJ, RG, PR, and MK reported cases to the study. EA performed the statistical analyses. JT initiated the study and supervised the project and the writing. All authors contributed to the article and approved the submitted version.