The benefits of modified FOLFIRINOX for advanced pancreatic cancer and its induced adverse events: a systematic review and meta-analysis

FOLFIRINOX has been one of the first-line options for advanced pancreatic cancer, even though it induces significant adverse effects. Several institutions have begun using modified FOLFIRINOX to decrease its side effects and increase its tolerability. We systematically investigated the outcome from patients who initially received modified FOLFIRINOX as a chemotherapy regimen for advanced pancreatic cancer. We used the random-model generic inverse variance method to analyse the binary data with 95% confidence intervals (CIs). Eleven studies were included in the meta-analysis with 563 total patients. The 6-month and 1-year overall survival (OS) rates of locally advanced pancreatic cancer (LAPC) were 90.9% and 76.2%. The 6-month and 1-year progression-free survival (PFS) rates of LAPC were 81.5% and 48.5%. The 6-month and 1-year OS rates of metastatic pancreatic cancer (MPC) were 79.7% and 47.6%. The 6-month and 1-year PFS rates of MPC were 56.3% and 20.6%. The following rates were also calculated: complete response rate (CR): 2.9%; partial response rate (PR): 35.9%; stable disease rate (SD): 41.2%; overall response rate (OR): 34.6%; disease control rate (DCR): 76.7%; progressive disease: 23.1%; and grade III/IV adverse events (AEs): neutropenia 23.1%, febrile neutropenia 4.8%, thrombocytopenia 4.8%, anaemia 5.7%, fatigue 11.5%, nausea 9.1%, diarrhoea 10.1%, vomiting 5.7%, neuropathy 3.8%, and increased ALT 5.7%. In conclusion, modified FOLFIRINOX could provide comparative survival benefits with fewer adverse events compared to the conventional dosage.

regimen should be used in clinical practice as a first-line option for advanced pancreatic cancer patients 14 . Shortly thereafter, a regimen of gemcitabine and albumin-bound paclitaxel was shown to have statistically significant survival benefits in OS and PFS, thus providing another choice for treating advanced pancreatic cancer 15 . However, FOLFIRINOX appears to be more effective than GEM/NAB-P 16 . Although FOLFIRINOX is a first-line option for patients with metastatic pancreatic cancer, there is a controversy about whether the survival benefits of the four-drug combination regimen outweigh the associated toxicities 17 . The significant adverse effects induced by this regimen include neutropenia, thrombocytopenia, febrile, diarrhoea neutropenia, febrile neutropenia, thrombocytopenia, diarrhoea, and neuropathy, which limit its usage and require stopping chemotherapy during treatment. Therefore, FOLFIRINOX is usually prescribed for patients ≤76 years old who have a good performance status (ECOG 0 or 1) 14 . To decrease the side effects and increase its tolerability, several institutions have used modified FOLFIRINOX. We conducted a systematic review and meta-analysis to assess the effectiveness and toxicities of modified FOLFIRINOX for patients with advanced pancreatic cancer compared to the conventional dosage.

Methods
Literature search. A systematic search was conducted to find eligible articles. Two investigators independently searched for prospective or retrospective studies (phase I-III trials, cohort studies, or case series) using Embase, PubMed, Web of Science, Scopus, and Cochrane without an upper-limit date until December 31, 2017. The search criteria included studies of advanced pancreatic cancer patients at any age who received any type of modified FOLFIRINOX in initial chemotherapy without language restrictions and no consideration of subsequent treatment. The preceding original regimen of FOLFIRINOX contained oxaliplatin 85 mg/m 2 , leucovorin 400 mg/m 2 , irinotecan 180 mg/m 2 , 5-fluorouracil (5-FU) bolus 400 mg/m 2 and 5-fluorouracil (5-FU) 2400 mg/m 2 . Modified FOLFIRINOX was defined as at least one of the drugs was reduced and/or the removal of 5-FU bolus in FOLFIRINOX 14 .
After removing duplicate articles, two investigators independently reviewed the abstracts. Studies were excluded if the study type was a review/meta-analysis, case report, comment, letter to the editor, or irrelevant literature. Differences between the investigators were resolved by a third-party investigator's opinion. Full articles were then selected for further assessment, and articles with only abstracts were excluded. Other exclusion criteria included studies that used a regimen other than modified FOLFIRINOX, did not include the initial usage of modified FOLFIRINOX or dose adjusted by physician's judgement without a specific time or presented the same patient cohort in another study. For details of the excluded articles, see the Supplement. Data extraction. General information was extracted from the foregoing selected publications and included the name of article, first authors, the name of journal, year of publication, study design, participating centres, country, observation sites, beginning and ending time, tumour stage, the composition of the modified FOLFIRINOX, its usage, number of patients, the ratio of males and females, average age, duration of follow-up, and performance status.
Survival was evaluated by the OS (6-month and 1-year) and PFS (6-month and 1-year) for the LAPC and MPC groups, which were extracted from the selected publications. If the survival rates were not directly available from the articles or authors, Engauge Digitizer was used to pool survival data from the Kaplan-Meier survival curve in each selected publication, especially for advanced pancreatic cancer reports for which the OS and PFS rates were not provided 18 . We chose the complete response (CR), partial response (PR), overall response (OR), stable disease (SD), disease control ratio (DCR), and progressive rate to evaluate the objective response to chemotherapy. The adverse events were calculated when they achieved grade III/IV. Grade III/IV toxicity includes neutropenia, febrile neutropenia, thrombocytopenia, anaemia, fatigue, vomiting, nausea, diarrhoea, neuropathy, and increased ALT.
Statistical analysis. First, we used the Critical Appraisal Skill Program (CASP) to evaluate each study (supplement). The CASP is a critical appraisal tool that is used in observational studies to assess the methodological quality of the individual studies. Binary data were meta-analysed with the random-model generic inverse variance method. We used random-effects rather than fixed-effects models because of the heterogeneity in the initial treatment of advanced pancreatic cancer. We used the odds ratio as the effect measure method and then changed it to probability. The I 2 statistics reflected the heterogeneity: I 2 = 0% indicated no heterogeneity, I 2 = (0%,25%) indicated low heterogeneity, I 2 = [25%,50%) indicated mild heterogeneity, I 2 = [50%,75%) indicated moderate heterogeneity, and I 2 = [75%,100%] indicated high heterogeneity 19 . All analyses were performed in Review Manager version 5.3 and Excel 2010. Figure 1 is a flow diagram that shows the selection process for the searched studies. We searched all databases that are available. There were 4772 related studies identified from the initial literature search; 2541 studies were eliminated because of duplications. Only 70 studies were eligible upon abstract screening. After full-text screening, only 11 studies remained, and they were included in the final analysis [20][21][22][23][24][25][26][27][28][29][30] .

Study search.
In these 11 studies, there were 563 patients, including 333 MPC and 230 LAPC. The number of patients who were treated with modified FOLFIRINOX ranged from 10 to 137. The average age in each study ranged from 60 to 65 years old (Table 1). Most patients' performance status was 0 or 1, and a small portion had a score of 2 23 . Most of the studies removed the 5-FU bolus, but two studies reduced the dose from 400 mg/m 2 to 300 mg/m 2  was an overlap of population in one study 26,31 . The most usage of continuous infusion 5-FU was 2400 mg/m 2 , but one study increased it to 2800 mg/m 2 or 3200 mg/m 2 and eliminated the 5-FU bolus 28 . The most frequently used dose of oxaliplatin was the same as the normal FOLFIRINOX regimen, but two studies used 63.75 mg/m 2 and 68 mg/m 2 29,31 . The dosage of irinotecan ranged from 135 mg/m 2 to 180 mg/m 2 . For the detailed modified FOLFIRINOX regimens, see Table 2.

Discussion
Our systematic review and meta-analysis considered 11 studies, which contained 563 patients with advanced pancreatic cancer treated with modified FOLFIRINOX. Previously, FOLFIRINOX was used to treat advanced pancreatic adenocarcinoma and demonstrated a better therapeutic benefit than gemcitabine (GEM) 32 . Although the dosage of FOLFIRINOX was reduced, the 12-month survival rate was still much higher than those of gemcitabine    14,20,37,38 . Similar to the data obtained for OS and PFS, as mentioned above, the response rate of modified FOLFIRINOX was also comparable to that of the original regimen 14,20,37,38 . Nevertheless, the favourable overall survival after modified FOLFIRINOX might be partly attributable to patient selection from many non-randomized studies. For the adverse events, the pooled rates of grade III/IV adverse events were lower than those of the FOLFIRINOX group; some were even lower than the GEM group 14,39,40 , such as anaemia, fatigue and vomiting. Concomitantly, a prospective phase II study of dose-attenuated treatment found that modified FOLFIRINOX could significantly reduce the occurrence of vomiting and fatigue 41 . As we know, in practice, when patients experience serious adverse events during continuous FOLFIRINOX chemotherapy, the strategy for physicians is to reduce the dosage or even stop the chemotherapy. Therefore, modified FOLFIRINOX is a good choice at the beginning of therapy, particularly for those with poor performance status. Modified FOLFIRINOX provides a relatively mild intervention and thus induces lower adverse events, thereby ensuring the continuity of chemotherapy.  . This may be due to different genetic traits between the ethnic groups. In general, the modified FOLFIRINOX regimen could provide good survival benefits for patients with advanced pancreatic cancer by increasing the OS and PFS and causing fewer adverse events. Our findings suggest that the dosage attenuation of initial FOLFIRINOX improves its tolerability without compromising its efficacy. Compared to the original regimen of FOLFIRINOX, modified FOLFIRINOX may be more applicable for patients with poor performance status. However, there were multiple combinations of the four drugs in which the 5-FU bolus was removed; which combination is the best for different ethnic groups or different healthy conditions remains a significant question. Clinical trials are still needed to justify the best combination for modified FOLFIRINOX. At last, although most of the studies that we chose were non-randomized and some even had a retrospective design that might bring bias, the current meta-analysis could provide constructive information for clinicians and patients.