Adjuvant programmed death-1 inhibition combined with chemotherapy in resected pancreatic cancer: A real-world cohort study

Abstract

BACKGROUND

Pancreatic ductal adenocarcinoma (PDAC) carries a high risk of early recurrence even after curative resection. Although programmed death-1 (PD-1) inhibitors have shown limited efficacy in advanced PDAC-primarily due to its profoundly immunosuppressive tumor microenvironment-the adjuvant setting may provide an immunologically favorable window for PD-1 blockade. In this context, this real-world cohort study aimed to explore the efficacy and safety of PD-1 inhibitors combined with adjuvant chemotherapy in patients with resected PDAC.

AIM

To evaluate the efficacy and safety of PD-1 inhibitor–based adjuvant therapy combined with chemotherapy in patients with resected PDAC.

METHODS

We conducted a retrospective, single-center, real-world cohort study of 57 patients who underwent R0 or R1 resection for PDAC between 2021 and 2023. Patients received either adjuvant chemotherapy alone (n = 31) or chemotherapy combined with a PD-1 inhibitor (n = 26). The primary endpoint was recurrence-free survival (RFS); secondary endpoints included distant metastasis-free survival (DMFS), overall survival (OS), and treatment-related adverse events (TRAEs). Survival outcomes were analyzed using Kaplan-Meier estimates and multivariable Cox proportional hazards models.

RESULTS

Compared with chemotherapy alone, the addition of a PD-1 inhibitor significantly prolonged median RFS [21.0 months vs 9.0 months; hazard ratio (HR) = 0.37, 95% confidence interval (CI): 0.19-0.72; P = 0.003] and DMFS (21.0 months vs 11.0 months; HR = 0.33, 95%CI: 0.16-0.69; P = 0.003). Although OS was numerically longer in the combination group (27.0 months vs 21.0 months), the difference was not statistically significant (P = 0.293). Multivariable analysis identified PD-1 therapy, younger age, lower baseline creatinine, and early-stage disease as independent predictors of longer RFS. Subgroup analyses suggested a generally consistent RFS benefit across predefined clinical and biomarker-defined strata. Exploratory analyses indicated that dynamic reductions in carbohydrate antigen 19-9 levels and neutrophil-to-lymphocyte ratio were more frequently observed in the combination group. Grade ≥ 3 TRAEs occurred in 19.2% of the combination group and 25.8% of the chemotherapy group.

CONCLUSION

Adjuvant PD-1 blockade combined with chemotherapy was associated with improved recurrence-related outcomes in patients with resected PDAC without an increase in severe toxicity. These real-world findings provide clinical rationale for further prospective evaluation of immunotherapy-based adjuvant strategies and highlight the need for biomarker-informed randomized trials to validate these observations and optimize patient selection.

Key Words: Pancreatic ductal adenocarcinoma; Programmed death-1 inhibitor; Adjuvant therapy; Recurrence free survival; Immunotherapy; Real world study

Core Tip: This real-world retrospective cohort study investigated the potential role of programmed death-1 (PD-1) inhibitor-based adjuvant therapy in patients with resected pancreatic ductal adenocarcinoma (PDAC), a population characterized by high recurrence risk after surgery. By comparing adjuvant chemotherapy alone with chemotherapy combined with PD-1 inhibition, this study demonstrated improved recurrence-free outcomes without an increase in severe treatment-related adverse events. These findings provide real-world clinical evidence supporting the feasibility of incorporating immunotherapy into postoperative adjuvant treatment strategies for resected PDAC.

INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) is among the most aggressive and lethal malignancies globally, with a steadily rising incidence and a 5-year survival rate below 12%, even after curative resection[1,2]. Although surgery remains the only potentially curative option, over 80% of patients experience disease recurrence within 18 months, primarily due to distant metastases[3,4]. Current adjuvant chemotherapy regimens, such as modified FOLFIRINOX and gemcitabine-based combinations, offer only modest survival benefits and fail to sufficiently prevent early relapse[4-6]. These limitations underscore the urgent need for more effective adjuvant strategies to improve long-term disease control in resected PDAC.

Immune checkpoint inhibitors (ICIs) targeting the programmed death-1 (PD-1)/programmed death-ligand 1 axis have transformed the management of several solid tumors, including melanoma, non-small cell lung cancer, and renal cell carcinoma[7-9]. However, in PDAC, clinical responses to ICIs have been disappointingly low, with objective response rates below 5% in unselected advanced-stage populations[10]. This resistance is largely attributed to the uniquely immunosuppressive tumor microenvironment (TME) of PDAC, characterized by a dense desmoplastic stroma, low tumor mutational burden (TMB), sparse CD8⁺ T-cell infiltration, and the enrichment of regulatory T cells and myeloid-derived suppressor cells[11]. Consequently, PD-1 inhibitors are currently recommended only for the rare subset of patients with microsatellite instability-high tumors, which constitute fewer than 1% of PDAC cases[12].

Emerging preclinical and translational data suggest that the postoperative adjuvant setting may provide a more permissive immunologic niche for checkpoint inhibition[13]. Surgical debulking can reduce tumor burden and transiently remodel the TME, exposing residual tumor antigens and enabling improved immune priming[14]. Moreover, chemotherapy-induced immunogenic cell death (ICD) may act synergistically with PD-1 blockade by promoting dendritic cell maturation, antigen presentation, and effector T-cell recruitment[15]. These mechanisms support the concept of an “immune reset window” during the minimal residual disease (MRD) phase, when immune surveillance may be more readily restored.

Despite this strong biological rationale, clinical evidence for PD-1–based adjuvant immunotherapy in PDAC remains limited. To date, no prospective randomized studies have specifically evaluated this strategy, and real-world data are scarce. Against this background, we conducted a retrospective, real-world cohort study to explore whether the addition of PD-1 inhibitors to standard adjuvant chemotherapy is associated with prolonged recurrence-free survival (RFS) and other clinically relevant outcomes in patients with resected PDAC.

MATERIALS AND METHODS

Study design and patient population

This was a retrospective, single-center cohort study conducted at the First Affiliated Hospital of Chongqing Medical University. Consecutive patients who underwent curative-intent R0 or R1 resection for histologically confirmed PDAC between January 2021 and December 2023 were included. Patients were eligible if they received at least one cycle of postoperative adjuvant therapy. Exclusion criteria included prior neoadjuvant treatment, synchronous malignancies, receipt of adjuvant radiotherapy or targeted therapy, and incomplete clinical or follow-up data. No sampling or case selection beyond the prespecified criteria was performed. Treatment decisions were made by a multidisciplinary team based on clinical factors and patient preferences. The patient selection process is illustrated in Figure 1.

Figure 1 Flowchart of patient selection and cohort allocation. Among 83 patients who underwent surgical resection for pancreatic cancer between January 2021 and December 2023, 26 were excluded due to prior neoadjuvant therapy, synchronous malignancies, receipt of adjuvant radiotherapy or targeted agents, or incomplete data. The final cohort comprised 57 patients with histologically confirmed pancreatic ductal adenocarcinoma and R0/R1 resection. Of these, 26 received adjuvant chemotherapy plus programmed death-1 inhibitors and 31 received chemotherapy alone. PD-1: Programmed death-1; PDAC: Pancreatic ductal adenocarcinoma.

Treatment regimens

Adjuvant chemotherapy consisted of one of the following regimens: Modified FOLFIRINOX (oxaliplatin, reduced-dose irinotecan, leucovorin, and continuous-infusion 5-fluorouracil), gemcitabine plus capecitabine, or gemcitabine monotherapy. Regimen selection was based on performance status, comorbidities, and clinician judgment. In the Chemo + PD-1 group, patients received either tislelizumab or camrelizumab at a fixed dose of 200 mg intravenously every 3 weeks, administered according to routine clinical practice, either concurrently or sequentially with chemotherapy. All patients completed at least one full cycle of the assigned regimen. Details regarding PD-1 initiation timing, treatment duration, and cumulative cycles are provided in Supplementary Table 1.

Follow-up and endpoints

Patients were followed from the date of surgery until disease recurrence, death, or the cutoff date (December 31, 2024). Surveillance was performed using contrast-enhanced computed tomography or magnetic resonance imaging every 2-3 months during the first postoperative year and every 3-6 months thereafter, in accordance with institutional follow-up protocols. Laboratory evaluations included complete blood count, liver and renal function panels, and tumor markers such as carbohydrate antigen 19-9 (CA19-9) and carcinoembryonic antigen (CEA).

The primary endpoint was RFS, defined as the interval from surgery to the date of documented recurrence, death from any cause, or last follow-up. Secondary endpoints included distant metastasis-free survival (DMFS), defined as the time from surgery to the occurrence of distant metastases, death, or last follow-up, and overall survival (OS), defined as the time from surgery to death from any cause.

Exploratory endpoints were prespecified and included dynamic trends in CA19-9 and neutrophil-to-lymphocyte ratio (NLR) during surveillance. These biomarkers were assessed longitudinally for descriptive and hypothesis-generating purposes to explore their association with recurrence and treatment response. Another exploratory outcome was the incidence and type of grade ≥ 3 treatment-related adverse events (TRAEs), including both chemotherapy-related and immune-related adverse events (irAEs). Adverse events were graded according to the CTCAE version 5.0.

Statistical analysis

Baseline characteristics were summarized using descriptive statistics. Categorical variables were compared using the χ² test or Fisher’s exact test, and continuous variables were analyzed using the student’s t-test or Mann-Whitney U test, depending on data distribution. RFS, DMFS, and OS were estimated using the Kaplan-Meier method and compared using the log-rank test.

Univariable Cox proportional hazards models were used to identify variables associated with RFS. Variables with P < 0.10 in univariable analysis or deemed clinically relevant a priori were included in the multivariable Cox proportional hazards model. The proportional hazards assumption was assessed using Schoenfeld residuals. Hazard ratios (HR) and 95% confidence intervals (CI) were reported.

Predefined subgroup analyses were exploratory and conducted for RFS based on clinical and biomarker-defined strata, including age, sex, tumor location, American Joint Committee on Cancer (AJCC) stage, baseline CA19-9 level, baseline NLR, and the interval between surgery and initiation of adjuvant therapy. Longitudinal changes in CA19-9 and NLR were visualized using spider plots. All analyses were performed using SPSS version 27.0 (IBM Corp., Armonk, NY, United States) and GraphPad Prism version 9.5. Given the exploratory nature of subgroup analyses, no adjustments for multiple comparisons were applied. A two-sided P value < 0.05 was considered statistically significant.

RESULTS

Patient characteristics and treatment overview

A total of 57 patients with resected PDAC were included in the final analysis, with 31 receiving adjuvant chemotherapy alone (Chemo group) and 26 receiving chemotherapy combined with a PD-1 inhibitor (Chemo + PD-1 group). The median follow-up duration was 22.4 months (range, 15.6-31.2 months). An overview of individual treatment exposure and clinical outcomes is illustrated using a swimmer plot (Figure 2).

Figure 2 Swimmer plot depicting individual treatment courses and outcomes. Each horizontal bar represents one patient’s clinical course. Gray bars indicate total follow-up duration; green segments represent chemotherapy alone; blue segments denote chemo-immunotherapy. Red crosses (x) indicate recurrence, black stars (★) indicate death, gray arrows (►) indicate survival at last follow-up, and downward blue triangles (▼) indicate loss to follow-up. The X-axis reflects months from surgery. PD-1: Programmed death-1.

Baseline demographic and clinicopathological characteristics were well balanced between the two groups (Table 1). No significant differences were observed in age, sex, body mass index, smoking or alcohol use, diabetes, or Eastern Cooperative Oncology Group performance status. Tumor location, histologic grade, T and N stage, and AJCC stage were also comparable. Molecular characteristics are summarized in Supplementary Table 2. Although not statistically significant, a higher proportion of stage I cases was noted in the Chemo + PD-1 group (80.8% vs 54.8%; P = 0.058), while stage II disease was more common in the Chemo group (38.7% vs 11.5%). AJCC stage was subsequently adjusted for in multivariable analyses. The median interval from surgery to adjuvant therapy initiation was similar between groups (45 days vs 39 days; P = 0.718). Baseline serum cholesterol levels differed between the two groups (Supplementary Table 3); however, cholesterol was not significantly associated with RFS in univariable Cox regression analysis and was therefore not included in the multivariable model.

Table 1 Baseline characteristics of patients in the two treatment groups, n (%).

Variable	Chemo + PD-1 (n = 26)	Chemo (n = 31)	P value
Age (years), mean ± SD	57.3 ± 9.5	60.7 ± 8.5	0.169
Sex			0.059
Male	13 (50.0)	23 (74.2)
Female	13 (50.0)	8 (25.8)
BMI (kg/m²), mean ± SD	22.7 ± 3.0	22.1 ± 2.4	0.307
Smoking history	10 (38.5)	15 (48.4)	0.452
Alcohol use	8 (30.8)	9 (29.0)	0.886
Diabetes	5 (19.2)	4 (12.9)	0.721
Family history of cancer	2 (7.7)	0 (0.0)	0.291
ECOG performance status			0.217
0	22 (84.6)	21 (67.7)
1	4 (15.4)	10 (32.3)
Differentiation			0.884
High	1 (3.8)	1 (3.2)
Moderate	11 (42.3)	10 (32.3)
Poor	12 (46.2)	15 (48.4)
Tumor location			0.498
Head	19 (73.1)	25 (80.6)
Body/tail	7 (26.9)	6 (19.4)
T stage			0.673
T1	7 (26.9)	6 (19.4)
T2	18 (69.2)	22 (71.0)
T3	1 (3.8)	3 (9.7)
N stage			0.212
N0	21 (80.8)	19 (61.3)
N1	3 (11.5)	10 (32.3)
N2	2 (7.7)	2 (6.5)
AJCC stage			0.058
I	21 (80.8)	17 (54.8)
II	3 (11.5)	12 (38.7)
III	2 (7.7)	2 (6.5)
Time to adjuvant therapy (days), median (IQR)	45 (32.25-64)	39 (30-55)	0.718

T and N stages were classified according to the American Joint Committee on Cancer 8^th edition. Data are presented as mean ± SD for normally distributed continuous variables, median (interquartile range) for non-normally distributed variables, and number (percentage) for categorical variables. BMI: Body mass index; ECOG: Eastern Cooperative Oncology Group; AJCC: American Joint Committee on Cancer; IQR: Interquartile range; PD-1: Programmed death-1.

Details of PD-1 treatment administration are provided in Supplementary Table 1. Among the 26 patients in the combination group, 73.1% initiated PD-1 therapy concurrently with chemotherapy, and 26.9% received it sequentially within 2 months. The median number of PD-1 cycles was 5 (range, 2-10), with 80.8% completing ≥ 4 cycles. Treatment discontinuation due to irAEs occurred in 2 patients (7.7%).

Molecular profiling and baseline laboratory markers are summarized in Supplementary Tables 2 and 3. All tested tumors were microsatellite stable. KRAS mutations were prevalent in both groups, with G12D being the most frequent subtype. One patient in the Chemo + PD-1 group was KRAS wild-type, whereas all tested patients in the Chemo group harbored KRAS mutations (P = 0.350). Frequencies of TP53, SMAD4, CDKN2A, and ARID1A mutations did not differ significantly. The median TMB was higher in the Chemo + PD-1 group (3.20 mutations/Mb vs 2.05 mutations/Mb), though the difference was not statistically significant (P = 0.229). Baseline levels of CA19-9, CEA, interleukin-6, and NLR were similar between groups, except for tumor necrosis factor-α, which was modestly higher in the Chemo + PD-1 group (2.71 pg/mL vs 1.71 pg/mL; P = 0.049).

Survival outcomes

The Chemo + PD-1 group demonstrated significantly improved RFS compared with the Chemo group. Median RFS was [21.0 months (95%CI: 16.8-25.2) vs 9.0 months (95%CI: 6.1-11.9)], respectively (P = 0.003; HR = 0.37, 95%CI: 0.19-0.72; Figure 3A).

Figure 3 Kaplan-Meier survival curves comparing adjuvant treatment groups. A: Recurrence-free survival was significantly longer in the Chemo + programmed death-1 (PD-1) group [hazard ratio (HR) = 0.37, 95% confidence interval (CI): 0.19-0.72; P = 0.003]; B: Distant metastasis-free survival was significantly prolonged in the Chemo + PD-1 group (HR = 0.33, 95%CI: 0.16-0.69; P = 0.003); C: Overall survival favored the Chemo + PD-1 group but was not statistically significant (HR = 0.66, 95%CI: 0.30-1.44; P = 0.293). The number of patients at risk is shown below each curve. PD-1: Programmed death-1; HR: Hazard ratio.

Similarly, DMFS was significantly prolonged in the combination group (21.0 months vs 11.0 months; P = 0.003; HR = 0.33, 95%CI: 0.16-0.69; Figure 3B). Although OS favored the Chemo + PD-1 group (27.0 months vs 21.0 months), the difference was not statistically significant (P = 0.293; HR = 0.66, 95%CI: 0.30-1.44; Figure 3C). The OS analysis should be interpreted cautiously given the limited number of events and the relatively short follow-up duration.

Multivariable analysis for RFS

In univariable Cox regression, age, AJCC stage, serum creatinine, and PD-1 treatment were significantly associated with RFS (Supplementary Figures 1 and 2) and were included in the multivariable model. Multivariable analysis identified PD-1 therapy as an independent predictor of improved RFS (HR = 0.335, 95%CI: 0.140-0.801; P = 0.014). Other independent predictors included younger age (HR = 1.051 per year increase, 95%CI: 1.003-1.101; P = 0.036), AJCC stage III vs I (HR = 5.104, 95%CI: 1.204-21.641; P = 0.027), and elevated baseline creatinine (HR = 1.036 per μmol/L, 95%CI: 1.008-1.065; P = 0.011; Figure 4A). Baseline creatinine was interpreted as a surrogate of overall physiological reserve rather than tumor-specific biology.

Figure 4 Forest plot of multivariable Cox regression analysis for recurrence-free survival and forest plot of subgroup analyses for recurrence-free survival. A: Independent predictors of recurrence-free survival included programmed death-1 (PD-1) treatment, younger age, advanced American Joint Committee on Cancer (AJCC) stage, and elevated baseline serum creatinine. Hazard ratio (HR) and 95% confidence interval are displayed. HR < 1 indicates a lower risk of recurrence; B: Subgroups included sex, age, tumor location, AJCC stage, interval from surgery to adjuvant therapy, baseline carbohydrate antigen 19-9, and neutrophil-to-lymphocyte ratio. Hazard ratios across subgroups consistently favored the Chemo + PD-1 group. No significant interaction effects were observed. AJCC: American Joint Committee on Cancer; CA19-9: Carbohydrate antigen 19-9; HR: Hazard ratio; CI: Confidence interval; NLR: Neutrophil-to-lymphocyte ratio.

Subgroup analysis

Prespecified subgroup analyses suggested a generally consistent RFS benefit with PD-1-based therapy across most clinical subgroups (Figure 4B). The magnitude of benefit appeared more pronounced in patients aged < 65 years (HR = 0.400, 95%CI: 0.178-0.897), those with tumors located in the pancreatic head (HR = 0.376, 95%CI: 0.175-0.810), and those who initiated adjuvant therapy within 48 days of surgery (HR = 0.242, 95%CI: 0.083-0.705). Additional subgroups with numerically favorable outcomes included patients with baseline CA19-9 ≤ 37 U/mL (HR = 0.313, 95%CI: 0.130-0.751) and NLR < 3 (HR = 0.429, 95%CI: 0.196-0.937). No significant interaction effects were detected.

Dynamic biomarker changes

As an exploratory analysis, dynamic trends in CA19-9 and NLR during follow-up are presented in Supplementary Figure 3. A greater proportion of patients in the Chemo + PD-1 group exhibited sustained declines in both markers compared with the Chemo group. Although slope differences between groups were not statistically significant, the overall trends were descriptively consistent with improved tumor control and reduced systemic inflammation in the combination group.

Safety

Grade ≥ 3 TRAEs occurred in 8 of 31 patients (25.8%) in the Chemo group and in 5 of 26 patients (19.2%) in the Chemo + PD-1 group. In the Chemo group, TRAEs included nausea/vomiting (n = 4), myelosuppression (n = 2), anemia (n = 1), and rash (n = 1). In the combination group, immune-related TRAEs included hyperthyroidism (n = 2), immune-mediated hepatitis (n = 2), and grade 3 rash (n = 1). Overall, the incidence of severe TRAEs was numerically comparable between groups, and all reported adverse events were managed according to established clinical guidelines. No treatment-related deaths were observed in either group.

DISCUSSION

Despite curative-intent resection, PDAC remains one of the most lethal malignancies, with recurrence rates exceeding 80% and distant metastases accounting for most treatment failures[5]. In this retrospective real-world study, we found that adding PD-1 inhibitors to standard adjuvant chemotherapy significantly prolonged both RFS and DMFS in patients with resected PDAC. Although OS did not differ significantly within the current follow-up period, the observed directional trends, together with multivariable analyses, suggest a potential benefit of PD-1–based adjuvant immunotherapy in this setting. These findings should be interpreted as hypothesis-generating rather than definitive evidence of clinical efficacy.

These findings align with emerging evidence that ICIs may offer greater activity in early-stage disease or MRD contexts, where tumor-driven immunosuppression is less established. PDAC has demonstrated near-universal resistance to ICIs in advanced disease, with objective response rates below 5% in unselected populations[11,16]. This resistance is largely attributed to a profoundly immunosuppressive TME, characterized by dense desmoplasia, low TMB, poor CD8⁺ T-cell infiltration, and enrichment of regulatory immune cell populations. By contrast, the postoperative period may represent a transient immunologic “reset” window, in which tumor debulking and antigen release facilitate immune reprogramming and potentially enhance responsiveness to checkpoint blockade[17].

Chemotherapy-induced ICD may further synergize with PD-1 inhibition by promoting dendritic cell activation, antigen cross-presentation, and CD8⁺ T-cell priming[18,19]. In this context, the observation that the magnitude of RFS benefit appeared more pronounced in patients with earlier-stage disease and in those initiating adjuvant therapy within 48 days postoperatively highlights the potential importance of timely immune engagement. However, these subgroup findings were exploratory and should not be interpreted as evidence of treatment effect modification. This pattern is nevertheless consistent with observations in melanoma and non–small cell lung cancer, where earlier postoperative immunotherapy has been associated with improved outcomes[20,21].

In addition to survival outcomes, we explored longitudinal dynamics of CA19-9 and NLR as exploratory biomarkers of treatment response. Patients in the combination group exhibited more frequent and sustained declines in both markers, although between-group slope differences were not statistically significant. These descriptive trends are consistent with prior studies linking CA19-9 kinetics and NLR fluctuations to prognosis and treatment response in gastrointestinal malignancies[22-24]. Given the exploratory nature of these analyses, causal inferences cannot be drawn, and future prospective studies incorporating serial immune profiling and biomarker-driven modeling are warranted to validate these observations.

Importantly, we observed a moderate but statistically significant correlation between RFS and OS (Spearman’s ρ = 0.578; P < 0.001; Supplementary Figure 4). This association supports the clinical relevance of RFS as an early endpoint in real-world or early-phase PDAC studies, particularly in settings where prolonged follow-up is required to detect OS differences. However, this correlation does not establish formal surrogate endpoint validation, and OS remains the definitive outcome for confirmatory trials.

The combination of PD-1 blockade with chemotherapy was generally manageable in this cohort. Although the incidence of grade ≥ 3 TRAEs was numerically lower in the combination group than in the chemotherapy-alone group, immune-related toxicities were observed and represent a distinct safety profile associated with PD-1-based therapy. No unexpected immune-related adverse events were identified, and adverse events were managed according to established clinical guidelines. Given the limited sample size and retrospective design, safety findings should be interpreted cautiously and warrant confirmation in larger, prospective studies.

This study has several limitations. First, the retrospective, single-center design introduces inherent risks of selection bias and treatment heterogeneity, despite multivariable adjustment. Second, the relatively small sample size-particularly in the PD-1 inhibitor group-limits statistical power for OS comparisons and the detection of rare adverse events. Third, comprehensive genomic and immune profiling was not available for all patients, precluding detailed mechanistic analyses. Finally, variability in treatment timing, adherence, and surveillance strategies, as is common in real-world studies, may have influenced clinical outcomes.

CONCLUSION

Overall, this real-world study suggests that adjuvant PD-1 inhibition combined with chemotherapy may improve recurrence-related outcomes in patients with resected PDAC without an apparent increase in severe toxicity. These findings provide a clinical rationale for prospective randomized trials to confirm efficacy, define the optimal timing and duration of PD-1 therapy, and advance biomarker-informed patient selection in the postoperative immunotherapy setting.