No survival benefit with suboptimal CA19-9 response: defining effective neoadjuvant chemotherapy in resectable or borderline resectable pancreatic cancer

Abstract

Background/purpose: Neoadjuvant chemotherapy (NAC) is gaining popularity over a surgery-first (SF) approach in treating resectable and borderline resectable pancreatic ductal adenocarcinoma (PDAC). However, what constitutes effective neoadjuvant chemotherapy is unknown.

Methods: We retrospectively analyzed resectable and borderline resectable PDAC patients who underwent pancreaticoduodenectomy (2010-2019) at a single institution. Optimal CA19-9 response was defined as normalization AND >50% reduction. We utilized Kaplan-Meier and multivariable-adjusted Cox models and competing risk subdistribution methods for statistical analysis.

Results: 586 patients were included in this study. The multivariable-adjusted analysis demonstrated OS benefit in the NAC group only when OS was calculated from diagnosis (HR = 0.72, p = 0.02), but not from surgery (HR = 0.81, p = 0.1). However, in 59 patients who achieved optimal CA19-9 response, OS is significantly longer than the 134 patients with suboptimal CA19-9 response (39.3 m vs. 21.5 m, p = 0.005) or the 117 SF patients (39.3 m vs. 19.5 m, p < 0.001). Notably, a suboptimal CA19-9 response conferred no OS advantage compared to SF patients. The accumulative incidence of liver metastases (but not other metastases) was significantly reduced only in patients with optimal CA19-9 response to NAC (multivariable-adjusted subdistribution HR = 0.26, p = 0.03).

Conclusion: CA19-9 response to NAC may serve as the marker for effective NAC. These findings warrant validation in a multi-institutional study.

Figure 1

Patient flow chart

Figure 2

Univariate and multivariable analysis on the effect of NAC on overall survival. On the left was survival measured from diagnosis, and on the right was survival measured from the time of surgery. *Adjusted for age, gender, CCI, vascular resection, resectability, neoadjuvant or adjuvant radiation therapy, tumor size, margin, LNR, LVI, PNI, Grade, Clavien classification post-op complication, adjuvant chemotherapy, and salvage therapy

Figure 3

A: Proportion of patients with liver, lung peritoneal, or local recurrence as their first site of progression in the overall cohort, divided by SF group (black) and NAC group (gray). The width of the bar indicated the sample size of each group. B: upper: survival analysis and lower: cumulative incidence with competing risk on time to progression from surgery

Figure 4

A–D: Univariate and multivariable analysis on the response of NAC and overall survival. Survival is measured from diagnosis. Unadjusted Kaplan-Meier model A) and Multivariable-adjusted Cox regression model C). Survival is measured from the time of surgery. Unadjusted Kaplan-Meier model B) and Multivariable-adjusted Cox regression model D). *Adjusted for age, gender, CCI, resectability, vascular resection, neoadjuvant or adjuvant radiation therapy, tumor size, margin, LNR, LVI, PNI, Grade, Clavien classification post-op complication, adjuvant chemotherapy, and salvage therapy. E–H: Propensity score methods on the response of NAC and overall survival. Survival was measured from surgery in the propensity-matched cohort, univariate Kaplan-Meier model E), and multivariable-adjusted inverse probability-weighted analysis G) when comparing the SF group with suboptimal responders. Survival measured from surgery, in propensity march cohort, univariate Kaplan-Meier model F) and multivariable-adjusted inverse probability-weighted analysis H) when comparing the SF group with optimal responders

Figure 5

A: Proportion of patients with liver, lung peritoneal, or local recurrence as their first site of progression in the overall cohort, divided by NAC response. The width of the bar indicated the sample size of each group. B: upper: survival analysis and lower: cumulative incidence with competing risk on time to progression from surgery