First-line gemcitabine plus nab-paclitaxel versus FOLFIRINOX for metastatic pancreatic cancer in a real-world population

Abstract

Aim

To compare the overall survival (OS) among patients who received first-line modified gemcitabine plus nab-paclitaxel (G/nab-P) or 5-fluorouracil, leucovorin, irinotecan and oxaliplatin (mFOLFIRINOX) for metastatic pancreatic cancer.

Methods

A single-center, retrospective, real-world study was conducted.

Results

The median OS was 9.4 months versus 7.5 months in the mFOLFIRINOX and modified G/Nab-P groups, respectively (p = 0.16). An exploratory subgroup analysis excluding patients who received one infusion and had an Eastern Cooperative Oncology Group performance score of 2 demonstrated similar OS of 11.3 months and 8.9 months, respectively. Median progression-free survival and time-to-treatment failure were not significantly different. Higher rates of adverse events were noted with mFOLFIRINOX.

Conclusion

mFOLFIRINOX did not significantly prolong OS compared with modified G/nab-P and was associated with increased toxicities.

Plain language summary

Pancreatic cancer that has spread to other parts of the body cannot be cured and patients usually have a limited time to live after diagnosis. It is therefore important to choose an effective treatment while minimizing the side effects of chemotherapy and maintaining patients' quality of life. Initial chemotherapy regimens include a combination of gemcitabine plus nab-paclitaxel (G/nab-P) or 5-fluorouracil, leucovorin, irinotecan and oxaliplatin (FOLFIRINOX). While some studies have found that patients treated with FOLFIRINOX may live longer, they also experience more side effects. This single-center study showed that patients treated with modified G/nab-P and FOLFIRINOX regimens had similar survival outcomes but FOLFIRINOX was associated with increased toxicities.

Pancreatic cancer is the third leading cause of cancer-related deaths in the USA and its incidence continues to rise with 60,430 new diagnoses anticipated in 2021 [1]. De novo metastatic disease is diagnosed in approximately 50% of patients and is associated with significant morbidity and mortality with a 5-year overall survival (OS) of less than 5% [2].

For more than a decade, gemcitabine was the long-standing frontline treatment for patients with advanced pancreatic cancer. Pivotal trials in 2011 and 2013 changed the treatment landscape when the combination regimen of 5-fluorouracil (5-FU), leucovorin, irinotecan and oxaliplatin (FOLFIRINOX) and gemcitabine plus ALB-bound paclitaxel (G/nab-P), respectively, each demonstrated improved OS and progression-free survival (PFS) compared with gemcitabine monotherapy [3,4]. While the two regimens have not been compared prospectively in the metastatic setting, the results of the PRODIGE-4/ACCORD-11 [3] and MPACT [4] trials showed that patients treated with FOLFIRINOX benefited from a numerically longer median OS and PFS compared with those who received G/nab-P. However, these benefits need to be weighed against the significant toxicities associated with FOLFIRINOX, including high rates of fatigue, nausea, diarrhea and neutropenia.

Current National Comprehensive Cancer Network (NCCN) guidelines designate both FOLFIRINOX and G/nab-P as category 1 recommendations for the initial treatment of metastatic pancreatic cancer [5]. Similarly, the American Society of Clinical Oncology (ASCO) Metastatic Pancreatic Cancer Guideline endorses both regimens as first-line (1L) treatment options [6]. However, it is recommended that FOLFIRINOX be restricted to patients with a favorable comorbidity profile compared with a relatively favorable comorbidity profile for G/nab-P. The ASCO guideline also addresses practical considerations for FOLFIRINOX, including access to a chemotherapy port and 5-FU infusion pump management services. These additional requirements, as well as the use of prophylactic G-CSF, distinguish FOLFIRINOX as potentially more resource-intensive than G/nab-P.

Retrospective, real-world studies have been conducted in an attempt to determine the superior efficacy of one regimen over the other. A systematic review and meta-analysis of 16 studies comparing FOLFIRINOX and G/nab-P in patients with advanced or metastatic pancreatic cancer, 14 retrospective case series and two cohort studies, showed the overall risk of death, progression and overall response rates were similar [7]. The toxicity profiles, however, were notably different: patients who received G/nab-P experienced higher rates of grade 3–4 neuropathy and anemia while FOLFIRINOX led to worse grade 3–4 nausea, neutropenia and febrile neutropenia.

If a patient experiences disease progression and maintains a good performance status, guidelines recommend G/nab-P as a second-line (2L) treatment after FOLFIRINOX. Following 1L G/nab-P, liposomal irinotecan with 5-FU and leucovorin (nal-iri/5-FU) is distinguished as the preferred 2L regimen. In contrast to the NCCN guideline, the ASCO guideline does not recommend FOLFIRINOX as a 2L option [5,6]. While there are limited existing studies to suggest the ideal treatment sequence, some data indicate that either G/nab-P or FOLFIRINOX could be given in the 1L or 2L setting [8].

Given the absence of a prospective head-to-head trial and literature variability, it remains unclear whether FOLFIRINOX or G/nab-P can be designated the preferred regimen. Additionally, much of the prior literature fails to address how, if at all, 1L treatment choices have been impacted by the US FDA approval of 2L nal-iri/5-FU. The predominant conclusion among the real-world analyses suggests there is no meaningful difference in clinical outcomes in patients who receive FOLFIRINOX compared with G/nab-P as 1L treatment. This study was designed to replicate the results from previous indirect comparison studies to determine if a similar trend in efficacy exists in a targeted population of patients with metastatic pancreatic cancer at an academic medical center.

Patients & methods

This was a single-center, retrospective, cohort study to assess the comparative effectiveness of 1L G/nab-P versus FOLFIRINOX at The Ohio State University Comprehensive Cancer Center, Arthur G. James Cancer Hospital and Richard J. Solove Research Institute (OSUCCC–James). The Ohio State University Institutional Review Board approved this study with a waiver of consent documentation due to the retrospective study design.

Patients for whom G/nab-P and/or FOLFIRINOX treatment plans were started during the predetermined study period were reviewed. Duplicate patient entries, as well as patients with treatment plans associated with a diagnosis other than pancreatic cancer, a curative treatment goal and 2L of treatment or later, were eliminated. The resulting patients comprised the screening population and were evaluated for inclusion in the study. Patients were included if they were at least 18 years old with a histologically or cytologically confirmed diagnosis of pancreatic adenocarcinoma, whose staging was consistent with metastatic disease and who received one or more infusions of G/nab-P or FOLFIRINOX between 1 November 2015 and 1 August 2020. Protected patient populations, including pregnant women and prisoners, as well as patients who received cytotoxic doses of chemotherapy ≤6 months prior to the diagnosis of metastatic disease were excluded. Patients treated with G/nab-P or FOLFIRINOX while enrolled in a clinical trial were also excluded from the analysis.

Retrospective data were obtained via chart review using the electronic medical record system. Baseline demographics and characteristics included age, gender, ethnicity, BMI, Eastern Cooperative Oncology Group (ECOG) performance status, presence of liver metastasis and peritoneal carcinomatosis, absolute neutrophil-to-lymphocyte (ANC/ALC) ratio, serum carbohydrate antigen (CA) 19-9, ALB and homologous recombinant deficiency (HRD) status. The ANC/ALC ratio and ALB cut points were chosen based on the Memorial Sloan Kettering Prognostic Score, which demonstrated the prognostic utility of these values in metastatic pancreatic cancer [9]. While there is heterogeneity in cutoff values, a review of the literature showed that approximately 18 cohorts used an ANC/ALC cutoff above and below 4 [10]. Thus, the Memorial Sloan Kettering system was chosen for a value that seemed to be centric. CA 19-9 levels were grouped according to cutoff values of normal (0–37 U/ml), <59× the upper limit of the normal (ULN) range and ≥59× the ULN due to the wide range of numeric CA 19-9 levels. These cutoffs were selected to mirror those reported in the PRODIGE-4/ACCORD-11 and MPACT trials [3,4]. HRD testing was performed using any of the commercially available assays. Due to the small number of patients with HRD, subgroup analyses of this population were not performed. Information about patients' treatment histories as well as 1L metastatic treatment dose reductions or omissions, cycle delays, total number of infusions administered, subsequent chemotherapy regimens and adverse events were collected. In addition, dates of diagnosis, treatment initiation, treatment discontinuation, disease progression and death were obtained.

Treatment

Patients received G/nab-P or FOLFIRINOX regimens and schedules per common practice at OSUCC–James, herein referred to as modified G/nab-P or modified FOLFIRINOX (mFOLFIRINOX). In the metastatic setting, modified G/nab-P was given on days 1 and 15 only and included ALB-bound paclitaxel 125 mg/m² infusion and gemcitabine 1000 mg/m² infusion, both administered over 30 min [11]. Similarly, mFOLFIRINOX was given on days 1 and 15 as conventional irinotecan 150–165 mg/m² over 90 min and oxaliplatin 85 mg/m² over 2 h followed by continuous 5-FU 2400 mg/m² infused via central line through a continuous ambulatory delivery device (CADD) pump over 46 h [12]. Leucovorin was omitted based on the theoretical, limited role of its use in the absence of the 5-FU bolus. As chemotherapy was utilized with palliative intent, regimen modifications were made to balance efficacy and tolerability. mFOLFIRINOX without leucovorin and/or the 5-FU bolus, as well as dose-reduced irinotecan, has been published in the literature [12–14]. Upfront dose reductions and modifications due to chemotherapy-related toxicities were applied at the physician's discretion. G-CSF was routinely used as primary prophylaxis with the mFOLFIRINOX regimen.

Outcomes

The primary outcome of this study was OS. Secondary outcomes included PFS, time-to-treatment failure (TTF), the proportion of patients who received 2L chemotherapy and the frequency of adverse events. All adverse effects were graded according to the National Cancer Institute's Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. OS was defined as the time from first treatment to death or censoring and PFS was defined as the time from first treatment to disease progression, death or censoring. TTF was calculated as the time from first treatment to treatment discontinuation or last contact. Patients who were alive without disease progression were censored at the data cutoff date of 1 December 2020.

Data management

Study data were collected and managed using Research Electronic Data Capture (REDCap) electronic data capture tools hosted at The Ohio State University [15,16]. REDCap is a secure, web-based software platform designed to support data capture for research studies, providing an intuitive interface for validated data capture; audit trails for tracking data manipulation and export procedures; automated export procedures for seamless data downloads to common statistical packages; and procedures for data integration and interoperability with external sources.

Statistical analysis

Potential differences in patient characteristics were accounted for by means of propensity score analysis using inverse probability of treatment weighting (IPTW). Treatment probabilities of patients were estimated from a logistic regression model using the treatment regimen as the dependent variable and patient characteristics and potential confounding variables as the independent variables. Baseline characteristics used in the model were weighted by propensity scores that included age, sex, race, BMI, ECOG score, tumor site, liver metastasis, peritoneal carcinomatosis, biliary stent, prior surgery, prior radiation and the receipt of chemotherapy plus radiation or cytotoxic chemotherapy at least 6 months prior to the diagnosis of metastatic disease. Propensity score models were reviewed graphically to confirm improvement in covariate balance after weighting.

Kaplan–Meier curves were used to compare chemotherapy type for three outcomes: OS, PFS and TTF. The curves were used to report the median survival times and TTF. Hazard ratios (HRs) and 95% CIs were estimated from Cox proportional hazards models, with and without the application of IPTW, with modified G/nab-P as the reference group. Because the results changed very little when weighted for the inverse probability of receiving FOLIRINOX (Supplementary Table 1), only the unweighted survival data are discussed here. Multivariable Cox regression analyses were conducted to evaluate the associations between patient characteristics and OS in the full cohort, including chemotherapy regimen and all significant variables from the univariate analysis.

The proportional hazards assumption of the chemotherapy variable was graphically reviewed using the observed cumulative martingale residual process for the variables along with 20 simulated paths from the null distribution. In addition, Kolmogorov-type supremum tests with 1000 simulated patterns were performed and showed insignificant results (OS: p = 0.54; PFS: p = 0.58). While there were some points in the survival curves where separation in the curves was not constant, based on the other tests, the proportional hazards assumption was not believed to be violated to the point that it needed to be accounted for in the model. Statistical significance was indicated by a p-value less than 0.05. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc., NC, USA).

Results

Patient characteristics

A total of 406 patients were screened for inclusion. Of these, 227 patients were excluded: 94 patients did not have metastatic pancreatic cancer, 67 patients did not receive at least one infusion of modified G/nab-P or mFOLFIRINOX, 40 patients were treated outside of OSUCCC–James, two were prisoners and 24 patients received cytotoxic chemotherapy less than six months before a diagnosis of metastatic disease. Among the remaining 179 patients, 128 patients received modified G/nab-P and 51 patients were treated with mFOLFIRINOX. Baseline characteristics between the groups prior to the application of IPTW were well balanced with the exception of age and baseline ECOG (Table 1). The majority of patients (78.4%) who received mFOLFIRINOX were ≤65 years old at a significantly younger median age of 57.9 years compared with 66.9 years in the modified G/nab-P group. In addition, patients in the mFOLFIRINOX arm had significantly better performance status; 92.2% of these patients had an ECOG score of 0 or 1 compared with 76.6% of patients who received modified G/nab-P.

Table 1. . Select baseline patient demographics and characteristics^†.

Characteristic	Modified G/nab-P (n = 128)	mFOLFIRINOX (n = 51)	Overall (n = 179)	p-value
Age, years   ≤65   >65	51 (39.8) 77 (60.2)	40 (78.4) 11 (21.6)	91 (50.8) 88 (49.2)	<0.001
Gender, male	77 (60.2)	27 (52.9)	104 (58.1)	0.40
Race‡   White   Black   Other	111 (86.7) 14 (10.9) 3 (2.3)	45 (88.2) 5 (9.8) 1 (2.0)	156 (87.2) 19 (10.6) 4 (2.2)	1.0
BMI, kg/m2   ≤25   >25	52 (40.6) 76 (59.4)	26 (51.0) 25 (49.0)	78 (43.6) 101 (56.4)	0.24
ECOG score   0   1   2	32 (25.0) 66 (51.6) 30 (23.4)	23 (45.1) 24 (47.1) 4 (7.8)	55 (30.7) 90 (50.3) 34 (19.0)	0.007
Liver metastasis	92 (71.9)	37 (72.6)	129 (72.1)	1.0
Peritoneal carcinomatosis	29 (22.7)	16 (31.4)	45 (25.1)	0.25
Tumor site   Head   Body   Tail	62 (48.4) 34 (26.6) 32 (25.0)	26 (51.0) 11 (21.6) 14 (27.5)	88 (49.2) 45 (25.1) 46 (25.7)	0.82
Biliary stent	40 (31.3)	13 (25.5)	53 (29.6)	0.47
Prior surgery	21 (16.4)	7 (13.7)	28 (15.6)	0.82
Prior radiation	4 (3.1)	2 (3.9)	6 (3.4)	1.0
Chemotherapy plus radiation >6 months prior	12 (9.4)	2 (3.9)	14 (7.8)	0.36
Cytotoxic chemotherapy >6 months prior	18 (14.1)	4 (7.8)	22 (12.3)	0.32
Homologous recombinant deficient	2/30 (6.7)	4/36 (11.1)	6/66 (9.1)	0.68
ANC/ALC ratio   ≤4   >4	62 (48.4) 66 (51.6)	26 (51.0) 25 (49.0)	88 (49.2) 91 (50.8)	0.87
ALB, g/dL   <4   ≥4	88 (68.8) 40 (31.3)	32 (62.8) 19 (37.3)	120 (67.0) 59 (33.0)	0.48
Level of CA 19-9   Normal§   <59× ULN   ≥59× ULN	20 (15.6) 41 (32.0) 67 (52.3)	3 (5.9) 17 (33.3) 31 (60.8)	23 (12.9) 58 (32.4) 98 (54.8)	0.20

^†Data are presented as n (%).

^‡Other races included American Indian/Alaskan Native, Asian and Middle Eastern/North African. The Fisher exact test for the race variable compared white versus nonwhite.

^§The normal range was 0–37 U/ml.

ALC: Absolute lymphocyte count; ANC: Absolute neutrophil count; ECOG: Eastern Cooperative Oncology Group; G/Nab-P: Gemcitabine plus nab-paclitaxel; mFOLFIRINOX: Modified 5-fluorouracil, leucovorin, irinotecan and oxaliplatin; ULN: Upper limit of the normal range.

Overall survival

The OS analysis was based on 102 deaths (79.7%) in the modified G/nab-P group and 40 (78.4%) in the mFOLFIRINOX group. The median OS was 7.5 months (95% CI: 5.9–8.5) in the modified G/nab-P group compared with 9.4 months (95% CI: 7.5–12.3) in patients treated with mFOLFIRINOX. The short-term risk of death was 23% lower in the mFOLFIRINOX group compared with the modified G/nab-P group, although this was not statistically significant (HR: 0.77; 95% CI: 0.53–1.12; p = 0.16; Figure 1A). In a multivariate analysis, ECOG performance score, liver metastasis, peritoneal carcinomatosis, ANC/ALC ratio >4 and CA 19-9 levels ≥59-times the ULN were identified as independent adverse prognostic determinants of OS (Supplementary Table 2).

Figure 1. . Kaplan–Meier curves for overall survival, progression-free survival and time-to-treatment failure.

(A) Median overall survival of the entire population. (B) Median progression-free survival of the entire population. (C) Median time to treatment failure of the entire population. (D) Median overall survival of the subgroup excluding patients who received only one infusion or had an Eastern Cooperative Group performance score of 2. (E) Median progression-free survival of the subgroup excluding patients who received only one infusion or had an Eastern Cooperative Group performance score of 2.

FOLFIRINOX: 5-fluorouracil, leucovorin, irinotecan and oxaliplatin; G/Nab-P: Gemcitabine plus nab-paclitaxel.

Progression-free survival

The analysis of PFS was based on 116 events (90.6%) in the modified G/nab-P group and 44 events (86.3%) in the mFOLFIRINOX group. The median PFS was 3.9 months (95% CI: 3.1–5.3) in the modified G/nab-P group versus 5.3 months (95% CI: 2.4–7.2) in the mFOLFIRINOX group (HR: 0.81; 95% CI: 0.57–1.15; p = 0.24; Figure 1B).

Time-to-treatment failure

The median TTF was 0.6 months longer in the mFOLFIRINOX cohort at 4.1 months (95% CI: 2.4–6.4) versus 3.5 months (95% CI: 2.5–4.1) in the modified G/nab-P group, although this difference was not statistically significant (HR: 0.78, 95% CI: 0.55–1.10; p = 0.15; Figure 1C).

Subgroup analysis

An exploratory subgroup analysis of survival outcomes was conducted excluding patients who received only one infusion or had an ECOG performance score of 2. Patients who received modified G/nab-P achieved a median OS and PFS of 8.9 months (95% CI: 7.8–10.6) and 5.3 months (95% CI: 4.0–6.1) compared with 11.3 months (95% CI: 8.6–13.6) and 6.1 months (95% CI: 3.3–8.5) in the mFOLFIRINOX group, respectively (Figure 1D & 1E). However, the OS (HR: 0.80; 95% CI: 0.53–1.23; p = 0.31) and PFS (HR: 0.87; 95% CI: 0.59–1.30; p = 0.50) between the two groups were not significantly different.

Subsequent therapy

A higher proportion of patients in the mFOLFIRINOX group received subsequent cytotoxic chemotherapy. In total, 73 patients received additional lines of therapy: 46 patients (35.9%) in the modified G/nab-P group compared with 27 patients (52.9%) in the mFOLFIRINOX group received 2L treatment (Table 2). The most common 2L regimens in the modified G/nab-P group were conventional irinotecan plus 5-FU (39.1%) and nal-iri/5-FU (37%) while the majority of patients (85.2%) in the mFOLFIRINOX group received 2L modified G/nab-P. In addition, 23 patients (18%) in the modified G/nab-P group versus 10 patients (19.6%) in the mFOLFIRINOX group received third-line (3L) treatment.

Table 2. . Subsequent lines of therapy^†.

	Modified G/nab-P (n = 128)	mFOLFIRINOX (n = 51)
Second-line treatment   mFOLFIRINOX   FOLFIRI   FOLFOX   nal-Iri/5-FU   5-FU   Capecitabine   Modified G/nab-P   Gemcitabine   Gemcitabine/cisplatin   Other	46 (35.9) 1 (2.2) 18 (39.1) 5 (10.9) 17 (37.0) 2 (4.4) 1 (2.2) 0 (0) 0 (0) 0 (0) 2 (4.4)	27 (52.9) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 23 (85.2) 1 (3.7) 1 (3.7) 2 (7.4)
Third-line treatment   mFOLFIRINOX   FOLFOX   FOLFIRI   5-FU/mitomycin   Modified G/nab-P   Irinotecan   Gemcitabine/cisplatin   Other	23 (18) 0 (0) 14 (60.9) 2 (8.7) 2 (8.7) 1 (4.4) 0 (0) 0 (0) 4 (17.4)	10 (19.6) 1 (10.0) 0 (0) 0 (0) 1 (10.0) 1 (10.0) 2 (20.0) 1 (10.0) 4 (40.0)

^†Data are presented as n (%).

FOLFIRI: 5-fluorouracil, leucovorin and irinotecan; FOLFOX: 5-fluorouracil, leucovorin and oxaliplatin; G/Nab-P: Gemcitabine plus nab-paclitaxel; mFOLFIRINOX: Modified 5-fluorouracil, leucovorin, irinotecan and oxaliplatin; nal-Iri/5-FU: Liposomal irinotecan plus 5-fluorouracil.

Safety

Overall, the frequency of treatment discontinuation was comparable between the cohorts. Treatment was discontinued in 122 patients (95.3%) in the modified G/nab-P group and 46 patients (90.2%) in the mFOLFIRINOX group. Patients treated with 1L modified G/nab-P or mFOLFIRINOX most commonly discontinued therapy due to disease progression (52.5 vs 63%), performance status (23 vs 13%) and death (11.5 vs 13%). Patients in the mFOLFIRINOX group experienced significantly higher rates of all-grade thrombocytopenia (76.5%), peripheral neuropathy (76.5%), diarrhea (60.8%) and mucositis (33.3%) compared with patients who received modified G/nab-P (Table 3). The frequency of adverse events, including all-grade neutropenia, anemia, fatigue, nausea and skin toxicities, were not significantly different between the two groups. An analysis of grade 3 and 4 toxicities showed patients treated with modified G/nab-P had numerically higher rates of thrombocytopenia and peripheral neuropathy while patients who received mFOLFIRINOX had numerically higher rates of diarrhea.

Table 3. . Adverse events^†.

Toxicity	Grade	Treatment			p-value (any grade)
		Modified G/nab-P (n = 128)	mFOLFIRINOX (n = 51)	Overall (n = 179)
Neutropenia	Any	28 (21.9)	13 (25.5)	41 (22.9)	0.60
	3–4	12 (9.4)	6 (11.8)	18 (10.1)
Anemia	Any	114 (89.1)	49 (96.1)	163 (91.1)	0.24
	3–4	15 (11.7)	11 (21.6)	26 (14.5)
Thrombocytopenia	Any	64 (50.0)	39 (76.5)	103 (57.5)	0.001
	3–4	7 (5.5)	2 (3.9)	9 (5.0)
Peripheral neuropathy	Any	70 (55.1)	39 (76.5)	109 (61.2)	0.01
	3–4	2 (1.6)	0 (0)	2 (1.1)
Diarrhea	Any	49 (38.6)	31 (60.8)	80 (44.9)	0.008
	3–4	4 (3.2)	3 (5.9)	7 (3.9)
Mucositis	Any	23 (18.1)	17 (33.3)	40 (22.5)	0.046
	3–4	0 (0)	0 (0)	0 (0)
Fatigue	Any	121 (95.3)	46 (90.2)	167 (93.8)	0.30
	3–4	11 (8.7)	4 (7.8)	15 (8.4)
Nausea	Any	84 (66.1)	35 (68.6)	119 (66.9)	0.86
	3–4	7 (5.5)	6 (11.8)	13 (7.3)
Skin toxicities	Any	32 (25.2)	12 (23.5)	44 (24.7)	1.0
	3–4	1 (0.8)	0 (0)	1 (0.6)

^†Data are presented as n (%).

G/Nab-P: Gemcitabine plus nab-paclitaxel; mFOLFIRINOX: Modified 5-fluorouracil, leucovorin, irinotecan and oxaliplatin.

Discussion

The results of this single-center, retrospective cohort study demonstrated similar survival outcomes among patients with metastatic pancreatic cancer treated with 1L modified G/nab-P and mFOLFIRINOX. Patients treated with mFOLFIRINOX experienced numerically longer OS and PFS, but mFOLFIRINOX did not significantly prolong survival versus modified G/nab-P. Comparable outcomes were also achieved in the subgroup analyses excluding patients who received only one infusion or patients with an ECOG performance status of 2, which were consistent with published clinical trial results. In addition, mFOLFIRINOX was associated with more toxicities and patients experienced higher rates of all-grade thrombocytopenia, peripheral neuropathy, diarrhea and mucositis. However, no grade 5 toxicities were associated with either modified G/nab-P or mFOLFIRINOX.

To date, there are no published randomized controlled trials comparing the effectiveness of these two regimens in the 1L setting for metastatic pancreatic cancer. Prospective efforts were published comparing the regimens in earlier disease therapy spaces that are discussed later. Because of the paucity of prospective data, comparisons of the efficacy of these two regimens have been largely dependent on heterogeneous retrospective reports. Several real-world cohort studies evaluated upfront treatment selection in patients with metastatic pancreatic cancer. In a review of 654 patient charts from participating oncologists throughout the USA, the median OS was numerically (albeit not statistically) significantly longer at 13.8 months in the FOLIRINOX group versus 12.1 months in the G/nab-P group [17]. FOLFIRINOX was significantly more toxic with higher rates of mucositis, fatigue and gastrointestinal side effects: nausea, vomiting and diarrhea. Another study conducted in Korea demonstrated that patients treated with 1L FOLFIRINOX achieved a significantly longer OS of 12.5 months compared with 10.3 months in patients who received G/nab-P [18]. The OS benefit of FOLFIRINOX persisted after propensity score matching at 11.8 months versus 10.3 months in the G/nab-P group, respectively. In addition, a retrospective analysis assessed patients from community-based oncology practices and confirmed similar outcomes in patients treated with FOLFIRINOX and G/nab-P [19]. Survival was measured using database persistence, a proxy for OS that was defined as the time between the initial dose of 1L therapy and the last available date of patient data due to the absence of dates of death. Database persistence was 8.6 months for both regimens. FOLFIRINOX was associated with a significantly higher incidence of any-grade adverse events and percentage of patients requiring G-CSF compared with G/nab-P. In a review of prospectively collected Korean registry data of patients with metastatic or recurrent pancreatic cancer, no difference was reported in major outcomes measures [20]. The median OS was 10.7 months compared with 12.1 months and PFS was 8 months versus 8.4 months in patients who received FOLFIRINOX and G/nab-P, respectively. Treatment-related anemia and thrombocytopenia were similar, but there was a higher incidence of grade 3 and higher neutropenia as well as gastrointestinal side effects in the FOLFIRINOX group.

Although a prospective comparison of the two regimens in the metastatic setting is not yet available, a phase II trial in Japan compared the two regimens in patients with locally advanced disease [14]. A total of 126 patients were allocated to the FOLFIRINOX (n = 62) or G/nab-P arm (n = 64). This trial, despite limitations of small size and its phase II nature, gave insight into the activity of G/nab-P in FOLFIRINOX-fit patients. The primary end point of 1-year survival was 77.4% (95% CI: 64.9–86.0) for the FOLFIRINOX arm compared with 82.5% (95% CI: 70.7–89.9) for the G/nab-P arm. Secondary end point analyses showed that the FOLFIRINOX arm had a 2-year OS of 48.2%, median OS of two years and a response rate of 30.9%. The secondary end point analyses for the G/nab-P arm revealed a 2-year OS of 39.7%, median OS of 1.8 years and a response rate of 41.4%. In addition, a separate trial, SWOG 1505, compared the two regimens in the resectable pancreatic cancer space [21]. Perioperative use of mFOLFIRINOX and G/nab-P resulted in similar 2-year OS rates with the majority of secondary end points also indicative of comparable activity.

In line with the results of previously published real-world studies, this analysis demonstrated that patients treated with 1L mFOLFIRINOX benefit from numerically longer survival outcomes over patients who received modified G/nab-P. However, due to the retrospective design, these results are likely driven by selection bias in the clinical setting. This is in contrast to prospective studies that select for populations of FOLFIRINOX-fit patients and whose results reveal similar outcomes between the two regimens [14,20,21]. The differences in outcomes underscore the idea that in equivalent patient populations, G/nab-P and FOLFIRINOX can be considered comparable regimens. Furthermore, studies investigating the effect of the treatment sequence of G/nab-P and FOLFIRINOX in 1L and 2L settings have been published. A cohort study of 83 patients found the median OS in patients who received FOLFIRINOX then G/nab-P and the reverse sequence were 13.7 months and 13.8 months, respectively [22]. A second study conducted in France found that patients who received 1L FOLFIRINOX had a significantly longer OS of 14 months compared with 9 months in the G/nab-P group [8]. However, in the matched population there was a nonsignificant trend toward a longer OS of 14 months versus 9 months in the FOLFIRINOX and G/nab-P groups. When looking at the sequencing of specific regimens, a higher percentage of patients who received FOLFIRINOX as 1L therapy were able to go on to receive G/nab-P in the 2L (46.9%) compared with the opposite sequence (20.4%). Additionally, in both the whole and matched populations, the median OS of patients who received FOLFIRINOX followed by G/nab-P was numerically but not significantly longer than for patients who received the reverse.

When compared with phase III PRODIGE-4/ACCORD-11 and MPACT trial data, the numeric OS and PFS resulting from this analysis were slightly shorter among all patients treated with modified G/nab-P or mFOLFIRINOX [3,4]. However, the patient population in this study was not analogous to patients treated in the clinical trial setting. In contrast to the MPACT trial, 23.4% of patients in the G/nab-P group had an ECOG performance score of 2 and 14.1% of patients received cytotoxic chemotherapy in the adjuvant setting [4]. Additionally, patients excluded from the original PRODIGE-4/ACCORD-11 trial were represented in this study: 7.8% of patients had an ECOG performance score of 2, 3.9% received chemotherapy plus radiation and 7.8% received cytotoxic chemotherapy at least six months prior to metastatic disease [3].

To evaluate a similar patient cohort, separate analyses were conducted excluding patients with an ECOG performance score of 2 and patients who only received one infusion due to the potential lack of therapeutic benefit after a single dose of chemotherapy. Patients in the modified G/nab-P group achieved an OS and PFS of 8.9 months and 5.3 months compared with 8.5 months and 5.5 months in the MPACT trial [4]. Similarly, the observed OS and PFS in the mFOLFIRINOX group were 11.3 months and 6.1 months while the PRODIGE-4/ACCORD-11 trial showed survival outcomes of 11.1 months and 6.4 months [3]. The results achieved in these subgroups showed almost identical OS and PFS to the published trial data and may reflect an overall high acuity and poor prognosis real-world population.

Furthermore, while patients treated with 1L mFOLFIRINOX achieved a numerically longer PFS in both the overall and subgroup populations, a higher percentage of mFOLFIRINOX patients discontinued treatment due to progression of disease compared with patients in the modified G/nab-P group. Since patients in the mFOLFIRINOX group were more likely to tolerate the intensity of the regimen by being younger and fitter, this may have allowed patients more time to develop progression and falsely lead one to conclude that treatment with mFOLFIRINOX can result in superior PFS. Additionally, a higher proportion of modified G/nab-P patients discontinued treatment due to declining performance status, despite mFOLFIRINOX being a more difficult regimen to tolerate, and again suggests that this cohort was sicker at baseline leading to shorter PFS.

The comparable outcomes in this study validate results from the published literature that demonstrate efficacy using modified G/nab-P and FOLFIRINOX despite lower dose intensities relative to the original reference regimens. Studies conducted at OSUCCC–James by Ahn et al. [11] and Blazer et al. [12] showed favorable results in treating patients with metastatic disease using biweekly G/nab-P as well as neoadjuvant mFOLFIRINOX for locally advanced and borderline resectable pancreatic cancer. The phase III PRODIGE-24 trial also utilized a modified FOLFIRINOX regimen that led to significantly longer survival in the adjuvant setting [13]. Due to the limited sample size, analyses of dose intensities were not conducted as it would likely not generate meaningful data.

The median TTF was similar in patients treated with modified G/nab-P or mFOLFIRINOX at 3.5 months and 4.1 months, respectively. However, 35.9 and 18% of modified G/nab-P patients compared with 52.9 and 19.6% of the mFOLFIRINOX patients received 2L and 3L treatment. Differences in the baseline characteristics of mFOLFIRINOX patients revealed that they were younger and fitter compared with the modified G/nab-P cohort, which may explain the higher percentages of patients who received 2L and 3L therapy. Despite the inherent bias in the selection of patients treated with mFOLFIRINOX and the ability of patients in this group to receive additional treatments, OS was not significantly extended compared with patients who received 1L modified G/nab-P. Furthermore, among the patients who were believed to have received all five classes of chemotherapy contained in these regimens, the median OS was 14.1 months in the modified G/nab-P group and 12.3 months in the mFOLFIRINOX group (p = 0.37). Survival in the modified G/nab-P subset was numerically longer than the OS reported in the overall mFOLFIRINOX cohort as well as the OS of 11.3 months excluding patients who received one infusion or with an ECOG performance score of 2. Thus, patients who were fit enough to receive multiple lines of treatment achieved comparable outcomes, suggesting similar efficacies of the two regimens, regardless of the order of administration.

In addition to the safety and efficacy of treatment, one must consider important patient factors and resource utilization that may impact the decision to treat a patient with one regimen over the other. In line with the ASCO recommendations, patients who proceed with FOLFIRINOX or mFOLFIRINOX should have strong social support to assist with managing a home infusion pump, central venous catheter or implanted port and potential chemotherapy toxicities [6]. Treatment with mFOLFIRINOX may also come at the cost of additional supportive care resources such as the use of G-CSF, antidiarrheal medications and medications to treat mucositis due to the increased toxicities seen in this study.

The current observations may be hindered by several limitations that warrant consideration. This study was conducted at a single institution, thus limiting the generalizability to other patient populations due to the standard practice of using modified dosing regimens in the palliative setting at OSUCCC–James. The absence of ethnic diversity within this patient population is also worth noting, as a recent meta-analysis illuminated the potential influence of pharmacoethnicity on survival outcomes in patients treated with 1L FOLFIRINOX versus G/nab-P [23]. Furthermore, the small sample size limited our ability to observe a difference in outcomes between patients treated with G/nab-P and mFOLFIRINOX. The estimated sample size needed for the study to have 80% power to detect an HR for death of 0.77 was 517 patients. The small sample size also precluded the possibility of conducting a matched-pairs analysis to better account for prescribing bias and the lack of randomization. While this analysis was subject to potential confounders for survival favoring the mFOLFIRINOX group, including younger age, improved performance status and higher frequency of receiving 2L chemotherapy, the IPTW was applied to minimize these biases except for the bias associated with the receipt of 2L treatment. In addition, due to the inherent limitations of the retrospective study design, outcomes and adverse event reporting were reliant on accurate documentation and subjective toxicity assessments. Of note, dermatologic adverse events, including rash, were grouped under the nursing assessment category of hand-foot syndrome and may have skewed the reporting of skin toxicities. Patients who received modified G/nab-P likely experienced rashes as opposed to a true hand-foot syndrome, which would be more expected in the mFOLFIRINOX group. Finally, patients received variable chemotherapy dose intensities due to performance status, organ dysfunction, toxicities and physician or patient preference. Despite the inability to control for different dose intensities, when limited to patients with an ECOG performance status of 0 or 1 and those who completed at least one cycle of therapy, the survival outcomes were comparable to the results achieved in phase III clinical trials.

Conclusion

In conclusion, the OS was not significantly different in patients treated with 1L mFOLFIRINOX compared with modified G/nab-P, but mFOLFIRINOX was associated with a higher incidence of adverse events. The outcomes in the overall population suggest that mFOLFIRINOX patients achieved a numerically longer OS compared with modified G/nab-P patients; however, it should be cautioned that these results may be driven by selection bias similar to that which exists in retrospective, real-world studies. While the results of this analysis are consistent with current literature, the question of whether to use G/nab-P or FOLFIRINOX is unanswered and the need for an adequately powered, noninferiority clinical trial remains. However, it is encouraging to know that the results of the ongoing NAPOLI-3 trial (NCT04083235), a global phase III study investigating 1L 5-FU, leucovorin, liposomal irinotecan and oxaliplatin versus G/nab-P, may provide clarity on the ideal sequencing of chemotherapy and could establish a new preferred regimen in patients with metastatic pancreatic cancer [24].

Summary points.

• Current National Comprehensive Cancer Network and American Society of Clinical Oncology guidelines support both gemcitabine plus ALB-bound paclitaxel (G/nab-P) and 5-fluorouracil, leucovorin, irinotecan and oxaliplatin (FOLFIRINOX) for the initial treatment of metastatic pancreatic cancer based on the PRODIGE-4/ACCORD-11 and MPACT data demonstrating significantly longer overall survival (OS) and progression-free survival (PFS) compared with single-agent gemcitabine. Trial results showed a numerically longer median OS and PFS in patients treated with FOLFIRINOX.

• G/nab-P and FOLFIRINOX have not been prospectively compared in a head-to-head trial and the preferred regimen remains unclear.

• Previously published retrospective, real-world studies suggest that G/nab-P is similar to FOLFIRINOX in its survival benefit and is associated with fewer toxicities.

• This single-center, retrospective, real-world study evaluated the comparative effectiveness of modified regimens of G/Nab-P and mFOLFIRINOX in the first-line treatment of metastatic pancreatic cancer at an academic institution.

• A total of 179 patients were included in the analysis; 128 patients received modified G/nab-P and 51 patients were treated with mFOLFIRINOX. Baseline characteristics were well balanced with the exception of younger age and lower Eastern Cooperative Oncology Group scores in the mFOLFIRINOX group.

• Median OS was not significantly different at 7.5 months (95% CI: 5.9–8.5) in the modified G/nab-P group compared with 9.4 months (95% CI: 7.5–12.3) in patients treated with mFOLFIRINOX. In a subgroup analysis excluding patients who received only one infusion and had poor performance status, the OS and PFS in each treatment group were consistent with the phase III trial data.

• Median PFS and time-to-treatment failure were similar between the treatment groups with a higher proportion of FOLFIRINOX-treated patients receiving second- and third-lines of therapy.

• Among the patients believed to have received all five classes of chemotherapy contained in the G/Nab-P and FOLFIRINOX regimens, the median OS was 14.1 months and 12.3 months in the modified G/nab-P and mFOLFIRINOX groups (p = 0.37), respectively, suggesting similar efficacies in patients fit enough to receive multiple lines of therapy.

• Higher rates of thrombocytopenia, peripheral neuropathy, diarrhea and mucositis were noted in patients who received FOLFIRINOX.

• The outcomes of this study suggest that OS was not significantly different in patients treated with first-line modified G/Nab-P versus mFOLFIRINOX, but FOLFIRINOX was associated with more adverse events.