The combination of oxaliplatin, irinotecan and fluorouracil, known as FOLFIRINOX, emerged as a first-line treatment option for selected patients with metastatic pancreatic cancer. The pivotal PRODIGE 4/ACCORD 11 randomized trial comparing FOLFIRINOX with single-agent gemcitabine demonstrated superior median survival and longer time without degradation of quality of life [1]. The objective response rate was 31,6% with FOLFIRINOX versus 9,4% with Gemcitabine (p<0.001) [1]. In their study, Hohla et al. suggested that female gender could positively predict response to FOLFIRINOX in patients with advanced pancreatic cancer [2]. In this secondary analysis, we explored the response to the FOLFIRINOX regimen by gender within the trial PRODIGE 4/ACCORD 11.

Demographic and clinical variables stratified by gender and treatment arms are showed in Table 1. The FOLFIRINOX group (N = 171 patients) included 106 women (62%) and 65 men. No significant differences were observed between genders regarding baseline variables, excepted for lymph nodes metastasis (17% and 35% in women and men respectively; p = 0.012). This difference was also observed in the ITT population (62 lymph node metastasis in males (29.5%) and 25 in females (19.2%); p = 0.035).

The objective response rate for females and males in the FOLFIRINOX group were of 36.9% and 28.3% respectively (p = 0.239), whereas the disease control rate was of 71.7% versus 67.7% for males and females respectively (p = 0.578). Similar results were observed in the ITT population of the study: objective response rate of 24.4% in females and 18.0% in males (p = 0.153); disease control rate of 60.7% and 60.3% in males and females respectively (p = 0.948).

Median OS was longer for females as compared to males in FOLFIRINOX group (13.1 versus 10.3 months respectively; HR = 0.73; 95% CI, 0.51–1.06) (Fig 2). Similarly, median PFS was superior (7.2 versus 5.9 months; HR = 0.79; 95% CI, 0.57–1.10). However, even if the observed differences were numerically higher, they were not statistically significant (p-values associated with HR were 0.101 and 0.169 respectively). Gender effect on OS (Fig 3) and PFS were also evaluated over the ITT population (Table 2). The HR for OS was estimated at 0.79 (95% CI 0.62–1.02; p = 0.068) in ITT population and 0.80 (95% CI (0.57–1.11; p = 0.182) in Gemcitabine arm.

The results of the univariate Cox analysis in FOLFIRINOX arm is showed in Table 3. From the initially fourteen factors under study, and, based on p-values ≤ 0.20, eight clinical variables and the three QLQ-C30 domains were considered to the inclusion in the multivariable analysis. Variables (with their associated HR and p-value at univariate Cox analysis) were the following: gender (HR = 0.73, p = 0.101), ECOG performance status (HR = 1.43, p = 0.058); primary tumor location (HR = 1.38, p = 0.086); level of albumin (HR = 1.57, p = 0.024), synchronous metastases (HR = 2.81, p = 0.002); number of metastatic sites (HR = 1.15, p = 0.125); hepatic metastases (HR = 2.19, p = 0.014); CA19-9 level (HR = 1.47, p = 0.049), physical function (HR = 0.98, p<0.001), constipation (HR = 1.01, p = 0.003), and dyspnea (HR = 1.01, p = 0.001). Additionally, in order to adjust for the presence of lymph node metastases, this variable was also considered (HR = 0.98, p = 0.900).

First, a multivariable Cox model was performed including nine clinical variables (that is, the eight clinical variables mentioned above, plus the variable: presence of lymph node metastases) as well as a term testing for the interaction between gender and lymph node metastases (Table 4). This model suggest that females would be associated with a better prognosis than males (HR = 0.53; 95% CI, 0.32–0.88; p = 0.015); moreover, a non-significant interaction between gender and lymph node metastases was observed (HR = 1.90; 95% CI, 0.66–5.49; p = 0.234).

Therefore, significant clinical variables were kept using a more stringent selection criteria (at p < 0.10) and quality of life dimensions were added, for a second multivariable Cox model (Table 4). After the addition of quality of life dimensions, the effect of gender is diminished (HR = 0.65; 95% CI, 0.41–1.04; p = 0.073). In this model, an abnormal CA19-9 level was associated with a worse prognosis (HR = 2.73; 95% CI, 1.29–5.80; p = 0.009); with respect to quality of life domains, even if the p-values associated to HR are under the significance level (0.05), HR remain around 1: a higher score of dyspnea (HR = 1.01; 95% CI, 1.00–1.02; p = 0.043) and constipation (HR = 1.01; 95% CI, 1.00–1.01; p = 0.035) were associated with a worse prognosis; whereas a higher physical function score was associated with a better prognosis (HR = 0.98; 95% CI, 0.97–0.99; p = 0.001).

In this study, despite the non-significant differences observed in the univariate analysis, the OS and PFS rates were higher for females than for males in FOLFIRINOX group (HR = 0.73 and HR = 0.79 respectively). When the ITT population was considered, a survival benefit for females was also suggested (HR = 0.79). One of the major biases is that the percentage of patients with lymph node metastasis is higher for males than for females. It could be one explanation of the differences between genders for the OS and the PFS, nevertheless a non-significant interaction between gender and lymph node metastases was showed by the model.

In the MPACT study comparing nab-paclitaxel plus gemcitabine to gemcitabine alone, age > 65 years, poorer Karnofsky performance status score, presence of liver metastasis, and higher number of metastatic sites were associated with an increased risk of death [4]. Gender was not included as a variable in the analysis, but a trend to a median superior survival in female was reported: 9.0 months (8.05–10.48) for female patients versus 8.1 months (6.93–9.30) for males in the combination arm, and 7.2 months (6.34–9.03) for gemcitabine alone in female patients versus 6.2 (5.45–6.97) for males.

It is important to highlight that all analysis conducted were not previously planned into the study protocol and were conducted only in an exploratory manner. The study has therefore several limitations inherent to unplanned subgroup analysis; moreover, multiplicity has not been taken into account. Finally, the limited sample size is one of the major limitation of this analysis by definition.

Our exploratory analysis did not permit to definitively conclude about a possible effect of gender on the prognosis of patients under FOLFIRINOX. Even if some results could numerically suggest a trend for a better prognosis in female patients with FOLFIRINOX as first-line option for patients with metastatic pancreatic cancer who are younger than 76 years and who have a good performance status (ECOG 0 or 1), no cardiac ischemia and normal or nearly normal bilirubin levels, no statistically significant differences between genders were found. This subject deserves further evaluation.