Cost‐effectiveness of neoadjuvant FOLFIRINOX versus gemcitabine plus nab‐paclitaxel in borderline resectable/locally advanced pancreatic cancer patients

Abstract Background The 2020 National Comprehensive Cancer Network guidelines recommend neoadjuvant FOLFIRINOX or neoadjuvant gemcitabine plus nab‐paclitaxel (G‐nP) for borderline resectable/locally advanced pancreatic ductal adenocarcinoma (BR/LA PDAC). Aim The purpose of our study was to compare treatment outcomes, toxicity profiles, costs, and quality‐of‐life measures between these two treatments to further inform clinical decision‐making. Methods and Results We developed a decision‐analytic mathematical model to compare the total cost and health outcomes of neoadjuvant FOLFIRINOX against G‐nP over 12 years. The model inputs were estimated using clinical trial data and published literature. The primary endpoint was incremental cost‐effectiveness ratios (ICERs) with a willingness‐to‐pay threshold of $100 000 per quality‐adjusted‐life‐year (QALY). Secondary endpoints included overall (OS) and progression‐free survival (PFS), total cost of care, QALYs, PDAC resection rate, and monthly treatment‐related adverse events (TRAE) costs (USD). FOLFIRINOX was the cost‐effective strategy, with an ICER of $60856.47 per QALY when compared to G‐nP. G‐nP had an ICER of $44639.71 per QALY when compared to natural history. For clinical outcomes, more patients underwent an “R0” resection with FOLFIRINOX compared to G‐nP (84.9 vs. 81.0%), but FOLFIRINOX had higher TRAE costs than G‐nP ($10905.19 vs. $4894.11). A one‐way sensitivity analysis found that the ICER of FOLFIRINOX exceeded the threshold when TRAE costs were higher or PDAC recurrence rates were lower. Conclusion Our modeling analysis suggests that FOLFIRNOX is the cost‐effective treatment compared to G‐nP for BR/LA PDAC despite having a higher cost of total care due to TRAE costs. Trial data with sufficient follow‐up are needed to confirm our findings.

In a previous study, we compared the effectiveness and costeffectiveness of neoadjuvant versus adjuvant therapies for BR/LA PDAC. 6 Neoadjuvant FOLFIRINOX was compared to adjuvant gemcitabine and adjuvant gemcitabine plus capecitabine. We found that neoadjuvant. FOLFIRINOX was the optimal strategy when compared to the two adjuvant strategies but were not able to incorporate other neoadjuvant therapies due to a lack of data.
To date, the efficacy and cost-effectiveness of neoadjuvant FOLFIRINOX and G-nP in BR/LA PDAC have not been compared in a randomized, prospective clinical trial. Because of this lack of clinical trial data, we developed a decision-analytic model incorporating the best available published data to simulate a hypothetical clinical trial between FOLFIRNOX and G-nP. The aim of this study is to compare these two neoadjuvant strategies in terms of cost-effectiveness for BR/LA PDAC treatment.

| Model overview and target population
We developed a Markov model, using Python 3.7, to follow and track hypothetical cohorts of BR/LA PDAC patients undergoing neoadjuvant therapy prior to resection through a simulated trial  Furthermore, we included a hypothetical scenario analyses to our model. This scenario explored the base analysis in a center of excellence setting. For this scenario, the dropout rates, PDAC recurrence rates, and R0 rates were adjusted to align with published literature from centers of excellence. [27][28][29] We defined a center of excellence as a highly specialized and interdisciplinary program within a healthcare institution that supplies expertise and resources to a particular medical area, in this case PDAC. We included the hypothetical scenario analysis to investigate what effect a center-of-excellence setting had on our model results. The parameters of the scenario can be found in Table S2.

| Study perspective and outcomes
We assessed the incremental cost-effectiveness of G-nP versus FOLFIRINOX from the perspective of the U.S. healthcare system. F I G U R E 1 Model schematic. Boxes represent health states, circles represent temporary transitional states. The black death state is absorbing. Arrows denote transitions The primary endpoint was the optimal neoadjuvant treatment, defined as the highest QALYs with an incremental cost-effectiveness ratio (ICER) below a willingness-to-pay threshold (WTP) of $100 000 (2021 USD). Secondary endpoints included overall survival (OS), progression-free survival (PFS), total cost of care, R0 status, PDAC resection rate, and monthly treatment-related adverse events (TRAE) costs. Costs and QALYs were discounted at 3% per year, and a half-cycle correction was applied to QALYs and unadjusted life-years.

| Health state transition probabilities, model calibration
We estimated the transition probabilities of our Markov model from published literature and clinical trial data (Tables 1 and 2). We calibrated our model by fitting the overall and progression free survival curves of each arm, including the natural history arm, to published Kaplan-Meier curves. [7][8][9][10][11]32 We extracted data from retrospective published data and preliminary clinical trial data to inform our model. In instances where BR/LA PDAC neoadjuvant treatment data was limited, we used conservative estimates from metastatic patient cohorts. We used a software called Engauge Digitizer to extract the data from the overall and progression free survival curves from published literature and used that data to calibrate cancer progression rates of our model. FOLFIRINOX clinical trials were more inclined to have younger patients, lower Charlson Comorbidity Index (CCI) scores, and higher likelihood of patients with N0 status than G-nP clinical trials. Since FOLFIRINOX clinical trials tend to enroll healthier and more robust patients, we used a more conservative estimate of overall survival and progression-free survival in the FOLFIRINOX trials to reconcile the baseline differences that like existed between patients enrolled in FOLFIRINOX and G-nP trials (Table S1). We also validated our model by comparing model outputs such as R0 rate, patient resection rate, and median OS/PFS with published clinical endpoints. 6-10,30-42 All-cause mortality was derived from the average of male and female 2016 U.S. life tables.

| Costs and health state utility values
Costs were calculated from a payer perspective and no indirect costs were included in the analysis. The costs of chemotherapy drugs were based on Centers for Medicare and Medicaid (CMS) 2020 average sale price. Capecitabine and radiation therapy costs for the neoadjuvant arms were estimated from published literature. 6

| Sensitivity analyses
We performed one-way deterministic sensitivity analyses, which involved changing each parameter individually across a plausible range of values, to determine the robustness of our base-case results (Tables 1 and 2). Probabilistic sensitivity analyses were also conducted by changing all input parameters at the same time. We sampled the parameter values on specific distributions and ran 10 000 iterations to investigate how ICER of the optimal strategy was affected.  Figure S1). The natural history arm had a median OS/PFS of 12.13/7.29 months and no survival after 10 years (Table 3).

| Center-of-excellence scenario
In the center-of-excellence scenario, FOLFIRINOX remained the opti-

| Sensitivity analyses
The model outcomes were most sensitive to monthly toxicity cost for treatment, resection surgery cost, PDAC recurrence rate, PDAC survival rate, and rate of TRAEs. Increasing monthly toxicity costs for FOLFIRINOX caused the strategy to exceed the $100 000 WTP threshold, thus making G-nP the optimal strategy. Increasing the rate of TRAEs, the dropout rate, and the PDAC recurrence rate for FOLFIRINOX all resulted in an ICER above the WTP threshold and decreasing the PDAC progression-free survival rate caused G-nP to dominate FOLFIRINOX (Figure S2). For G-nP, the ICER did not exceed the $100 000 WTP threshold for all the plausible ranges of input parameters tested ( Figure S3). Additionally, although the model was also sensitive to R0 rate, PDAC mortality rate, second-line treatment cost, and utility associated with progressive disease, changing these parameters within the pre-specified parameter ranges did not alter the model results.
The probabilistic sensitivity analysis results demonstrated that the base-case results were robust to parameter uncertainty.
FOLFIRINOX remained the optimal strategy in 76.8% of the F I G U R E 2 Base case efficiency frontier. G-nP, Gemcitabine plus nab-paclitaxel; QALYs, quality-adjusted life-years simulations, while G-nP was optimal in the remaining 23.2% with a WTP threshold of $100 000 ( Figure S4). The FOLFIRINOX strategy was the cost-effective strategy 37.2% of the simulations with a WTP threshold of $50 000 and 89.7% of the time with a WTP threshold of $150 000 ( Figure S5).

| DISCUSSION
The results of our analysis found that neoadjuvant FOLFIRINOX is the preferred treatment strategy for patients with BR/LA PDAC compared to neoadjuvant G-nP by multiple endpoints. Although FOLFIRINOX had the highest monthly TRAE cost, it was the costeffective strategy as the cost to gain a QALY was below the defined threshold. The strategy also yielded superior OS, PFS, QALYs, and R0 resection rates in our modeling projections.
As we described in the methods, we included a hypothetical scenario analysis to explore how the results of the base case changed in a center-of-excellence setting. In the hypothetical scenario where patients were only treated at a center-of-excellence, FOLFIRINOX remained the optimal treatment strategy compared to Gn-P. For the center-of-excellence analysis, we found that although fewer patients underwent resection (FOLFIRINOX 57.9 vs. 67.32%; G-nP 52.6 vs. 59.33%), and that overall survival was longer in this scenario compared to the FOLFIRINOX arm in the base-case analysis. This result may be explained by a sicker patient population at the center-ofexcellence, but improved surgical outcomes compared to patients treated in centers without such a designation. 27