Abstract Background: Pancreatic cancer is the third leading cause of cancer-related death in the United States, and current chemotherapy options provide limited benefit. Recent studies have shown that a ketogenic diet (KD) can exert anti-tumor effects by reprogramming tumor metabolism and exposing new therapeutic vulnerabilities1. Efforts to target glutamine metabolism—an essential pathway in many cancers—have shown promise in preclinical models but have not translated into significant clinical success. Methods: A ketogenic diet was administered to mice, along with a normal diet in control mice. Pancreatic cancer xenografts were monitored for growth and harvested for biochemical analyses. Tumors were analyzed for metabolites and histone modification by LC/GC-MS and for oxidative stress by measuring the levels. Result: Here, we show that a KD increases tricarboxylic acid (TCA) cycle activity and elevates reliance on glutamine-related metabolites in murine pancreatic cancer models and in vitro under KD-mimicking conditions. This metabolic adaptation is in response to increased tumor dependence on glutamine-mediated anaplerosis to compensate for reduced glucose availability. Additionally, we examined the impact of a ketogenic diet on redox homeostasis and found that tumors from ketogenic-diet-fed mice exhibited a marked increase in ROS levels. This redox imbalance provides further rationale for combining TCA-cycle inhibition with ROS-inducing agents to amplify metabolic and oxidative stress in PDAC tumors. We demonstrate that combining glutamine metabolism inhibitors, such as CB839 or 6-diazo-5-oxo-L-norleucine (DON), with a KD leads to robust anti-tumor effects in preclinical models of pancreatic cancer. Moreover, the combination of ivosidenib and a ketogenic diet caused a synergistic rise in intratumoral ROS and yielded a 50% survival benefit in treated mice. Conclusion: Together, these findings demonstrate that a ketogenic diet exposes a metabolic and redox vulnerability in PDAC by increasing reliance on glutamine-driven anaplerosis and elevating intratumoral ROS. Exploiting this state with targeted metabolic inhibitors—alone or in combination with ROS-inducing agents—produces potent anti-tumor responses and highlights a promising therapeutic strategy for pancreatic cancer. Citation Format: Omid Hajihassani, Asael Roichman, Jacob A. Boyer, Michael MacArthur, Ricardo Cordova, Priyashree Sunita, Goutam Dey, Parsa Rezvani, Alexander Loftus, Christina Boutros, Jonathan Hue, Parnian Naji, Nimat Manzoor, Anika Buch, Hallie J Graor, Joshua D. Rabinowitz, Jordan M. Winter. A ketogenic diet sensitizes pancreatic cancer to metabolic therapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 6805.

Supplemental Figure 11 - Durable response to immunotherapy in a patient with diabetes who developed metastatic disease after a pancreatectomy and progressed on chemotherapy. CA 19-9 (U/mL) values are depicted by the red line. Peripheral glucose levels (mg/dL) are depicted by the black line. The patient’s median glucose level over the course of treatment is depicted with the horizontal gray dashed line (190 mg/dL).Supplemental Figure 11 - Durable response to immunotherapy in a patient with diabetes who developed metastatic disease after a pancreatectomy and progressed on chemotherapy. CA 19-9 (U/mL) values are depicted by the red line. Peripheral glucose levels (mg/dL) are depicted by the black line. The patient’s median glucose level over the course of treatment is depicted with the horizontal gray dashed line (190 mg/dL).

Hyperglycemia affects the immune profile of mice receiving macrophage-modulating immunotherapies. GO analysis of tumors from mice receiving standard water or D30 water with or without PLX3397 (A–D). Representative flow cytometry plots and quantitative analyses showing the effects of vehicle control, D30, PLX3397, PLX3397 + D30, PF-4136309, and PF-4136309 + D30 on tumor-associated macrophages and their functional subsets. Tumor-derived single-cell suspensions were gated sequentially on viable cells based on FSC-A and SSC-A properties, followed by the identification of F4/80+ macrophages (E). Representative contour plots illustrate the frequency of total F4/80+ macrophages across treatment groups. Macrophage polarization was further assessed by delineating M1-like macrophages (iNOS+; F) and M2-like macrophages (arginase-1+; G). PD-L1 expression on macrophages (H) was analyzed within the F4/80+ compartment. Bar graphs summarize the percentage of total macrophages, M1-like, M2-like, and PD-L1+ macrophage populations for each treatment group. Data are presented as the mean ± SD from independent biological replicates. Statistical significance was assessed using one-way ANOVA with multiple-comparison correction. ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. All gates were defined using fluorescence-minus-one (FMO) controls and applied uniformly across all samples.

Representative IHC sections of pancreatic tumors from mice treated with vehicle control, D30, PLX3397, PLX3397 + D30, PF-4136309, or PF-4136309 + D30. A, Panels depict IHC staining for Ki-67 and cleaved caspase-3 across treatment groups. Brown chromogenic signal indicate positive staining with hematoxylin counterstaining of nuclei. Quantitative analysis of Ki-67 (B) and cleaved caspase-3 (C). D, Panels depict IHC staining for iNOS and CD163 across treatment groups. Magnified views are shows to better demonstrate differences. Pink chromogenic signal indicates positive staining with hematoxylin counterstaining of nuclei. Across treatment groups, qualitative differences in proliferative index, apoptotic activity, and macrophage polarization are evident. Scale bars, 100 μm. All images are representative of multiple tumors examined per group and were acquired using identical staining and imaging conditions. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

Supplemental Figure 7 - Flow cytometric analysis of T cell infiltration following single and combination treatments in pancreatic tumors. Representative flow cytometry plots and quantitative analyses showing the effects of vehicle control, D30, PLX3397, PLX3397 + D30, PF-4136309, and PF-4136309 + D30 on intratumoral T cell populations. Tumor-derived single-cell suspensions were first gated on viable cells based on forward- and side-scatter properties, followed by identification of CD3+ T cells. CD3+ cells were subsequently subdivided into CD4+ and CD8+ T cell subsets as indicated. Representative contour plots illustrate the gating strategy and frequencies of each population, with arrows denoting the sequential gating steps. Bar graphs summarize the percentage of CD3+, CD4+, and CD8+ T cells across treatment groups. Data are presented as mean ± S.D. from independent biological replicates. Statistical significance was determined using one-way ANOVA with multiple-comparison correction. ns, not significant; * indicated p<0.05; ** indicated p<0.01; *** indicated p<0.001; **** indicates p<0.0001. All gates were defined using FMO controls and applied consistently across all samples.

BACKGROUND:This study analyzed cure probabilities and the time required to achieve statistical cure in patients with pancreatic ductal adenocarcinoma in the United States. METHODS:Patients diagnosed with pancreatic ductal adenocarcinoma between 2000 and 2021 were identified from the Surveillance, Epidemiology, and End Results database. Nonmixture cure models were utilized to compare the overall survival with the survival expected for a matched general population based on age and gender. RESULTS:A total of 125,706 patients were included; 25,724 (20.5%) underwent surgical resection. In the entire pancreatic ductal adenocarcinoma cohort, the overall cure fraction was 5.70% (95% confidence interval, 5.55%-5.89%). Cure probabilities decreased with older age and advancing clinical stage. However, there was no clinically meaningful difference in time to achieve statistical cure based on age or clinical stage. Patients who underwent surgical resection demonstrated higher probabilities of cure (25.24% vs 6.82%) and a shorter time to statistical cure (10.1 vs 12.3 years) than those who did not. Among patients with resected pancreatic ductal adenocarcinoma, cure probabilities decreased with older age, advancing clinical stage, and nodal positivity. Patients treated with neoadjuvant chemotherapy followed by surgery showed higher cure probabilities than those treated with upfront surgery (30.0% vs 17.05%). CONCLUSION:This study suggests that patients undergoing resection for pancreatic ductal adenocarcinoma may achieve a statistical cure. Earlier clinical stage, neoadjuvant chemotherapy, and negative nodal status seem to be associated with improved cure probabilities. These findings highlight the importance of early detection and multimodal treatment approaches to improve long-term outcomes in pancreatic ductal adenocarcinoma.

Hyperglycemia improves efficacy of macrophage-targeting immunotherapies in vivo. A, Overall survival of C57Bl/6J mice with KPC orthotopic tumors treated with vehicle or PLX3397, with or without D30 water. B, Overall survival of mice with varying degrees of immunodeficiency (NSG, athymic nude, and C57Bl/6J) treated with PLX3397 with or without D30. C, Overall survival of C57Bl/6J mice with KPC orthotopic tumors treated with vehicle or PF-4136309, with or without D30. An asterisk (*) seen in A and C signifies that mice were alive at the end of the experiment. Median survival was plotted using the Kaplan–Meier method and compared using the log-rank test in the preceding survival experiments.

Intentional hyperglycemia affects the TME. A, Peripheral glucose levels in tumor-bearing C57Bl/6J mice receiving standard water or D30 water for 2 weeks, measured using an Abbot glucometer. B, Relative intratumoral glucose levels from KPC orthotopic pancreatic tumors in C57Bl/6J mice consuming standard water or D30 water for 2 weeks, assessed using a Glucose-Glo assay. C, RNA expression of select tumor-suppressor and tumor-promoter genes from treatment-naïve KPC orthotopic pancreatic tumors in C57Bl/6J mice receiving either standard water or D30 water for 4 weeks assessed via bulk RNA-seq. D, RNA expression of select immune-related genes from the same tumors as C. Single outlying data points were excluded from analyses of D30 iNOS expression and D30 arginase expression (n = 4 for these genes). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Supplemental Figure 2 - Macrophages differentially affect KPC cell growth based on polarization. Co-culture clonogenic assay of KPC cells with M1- or M2-like bone marrow-derived macrophages. Macrophages without KPC cells are depicted for reference. Relative KPC survival was assessed by quantification of luminescence of luciferase-expressing KPC cells after luciferin supplementation. Background luminescence of macrophages was subtracted. *** indicates p<0.001.

Nicotinamide adenine dinucleotide (NAD+) is a central metabolic cofactor essential for cell survival and stress response in normal tissues. Its dietary precursors, commonly referred to as vitamin B3 derivatives, including nicotinamide (NAM), nicotinamide riboside (NR), and nicotinamide mononucleotide (NMN), are marketed as nutraceuticals with potential energy-boosting, cardioprotective, and neuroprotective benefits. Consequently, many cancer patients utilize NAD+ precursors to alleviate chemotherapy-induced toxicity and promote health. However, the impact of NAD+ supplements on intrinsic tumor biology and progression remains controversial and incompletely characterized. In this study, we assessed the impact of common vitamin B3 derivatives, NAM, NR, and NMN, on chemotherapy efficacy in pancreatic ductal adenocarcinoma (PDAC) using both in vitro and in vivo models. Among the compounds tested, NMN exhibited the strongest protective effect on cancer cells, enhancing resistance to oxaliplatin, 5-fluorouracil, and gemcitabine in vitro. Mechanistically, NAD+ precursors promoted mitochondrial function, reduced oxidative stress, and suppressed DNA damage and apoptosis in treated cancer cells, all contributing to chemotherapy resistance. In murine models, both immunocompetent and immunodeficient, supplementation with NAM and NMN similarly conferred resistance to standard chemotherapy and supported cancer growth. Our findings highlight a potentially concerning role for NAD+-boosting supplements in the context of an active cancer, especially when used in conjunction with chemotherapy. These data underscore the need for careful evaluation of nutraceutical use in cancer patients, particularly those with PDAC, as vitamin B3 derivatives may inadvertently promote tumor cell survival and compromise treatment efficacy.

Extrapulmonary neuroendocrine carcinomas (EP-NECs) are a heterogeneous group of rare tumors with poor clinical outcomes. These patients have limited treatment options after progressing on first-line platinum-based chemotherapy. Although dual immune checkpoint inhibitors (ICIs) with anti-CTLA-4 and anti-PD-1 blockade have significantly improved outcomes for several solid tumors, they demonstrated modest activity for EP-NECs with 9–26% response rates and low survival rates. Preliminary data demonstrated that NP-101 (Nigella sativa formulation) enhances T-cell infiltration and is synergistic with dual ICPIs in NECs’ cellular models. This pilot study evaluated the tolerability and efficacy of NP-101 plus nivolumab and ipilimumab in patients with metastatic EP-NECs refractory to first-line platinum-based chemotherapy. This is a single-arm pilot study (NCTNCT05262556) in which patients with metastatic EP-NECs received NP-101 (oral capsules), 3,000 mg daily, plus ICPIs (intravenous nivolumab 3 mg/kg and ipilimumab 1 mg/kg) every 3 weeks for four cycles. Nonprogressors received NP-101 (3,000 mg daily), plus biweekly maintenance of nivolumab (240 mg), and then completed 24 weeks of treatment. Treatment-related adverse events (TR-AEs) were characterized according to CTCAE v4.03. The response rate was estimated according to Response Evaluation Criteria in Solid Tumors (RECIST), version 1.1. The Kaplan–Meir method was used to estimate median PFS and OS. Twelve patients received ≥1 dose of NP-101 and nivolumab plus ipilimumab. There were no dose-limiting toxicities (DLTs). Grade 1/2 TR-AEs occurred in 100% (12/12) of patients. The most common G1/2 TR-AEs included the following: fatigue (75%), nausea (41.7%), pruritus (41.7%), muscle weakness (33.3%), vomiting (25%), rash (25%), and abdominal pain (25%). Eight patients (66%) experienced grade 3/4 TR-AEs, including rash (33.3%), nausea (16.7%), vomiting (16.7%), and transaminitis (16.7%). No treatment-related grade 5 toxicities or deaths were recorded. The objective response rate was 41.7% (2/12 (16%) complete response (CR) + 3/12 (25%) partial response (PR); 95% CI: 15.2–72.3%) for all patients and 50% (2/8 CR + 2/8 PR, 95% CI: 0.16–0.84) for patients with NEC of gastrointestinal origin. The median duration of response was 7.5 months. As for the median progression-free survival, it was 5.7 months, and the median overall survival (OS) was 10.5 months with a median follow-up of 10.4 months. The combination of NP-101 plus dual ICPIs (nivolumab and ipilimumab) was safe and well tolerated with preliminary evidence of antineoplastic activity. Currently, a randomized phase II clinical trial evaluating the combination is under development.

Pancreatic cancer is the third leading cause of cancer-related death in the United States. Current chemotherapy options provide limited benefits. Emerging evidence suggests that a ketogenic diet (KD) exerts anti-tumor effects by reprogramming tumor metabolism and revealing therapeutic vulnerabilities. Efforts to target glutamine metabolism—an essential pathway in many cancers—have shown promise in preclinical models, but clinical efficacy has remained limited. Here, we show that a KD increases tricarboxylic acid (TCA) cycle activity and elevates reliance on glutamine-related metabolites in murine pancreatic cancer models and in vitro under KD-mimicking conditions. This metabolic adaptation occurs in response to reduced glucose availability. We demonstrate that combining glutamine metabolism inhibitors, such as CB-839 or 6-diazo-5-oxo-L-norleucine (DON), with a KD leads to robust anti-tumor effects in preclinical models of pancreatic cancer. Thus, metabolic vulnerability induced by dietary intervention provides a rationale for combining glutamine-targeted therapies with a ketogenic diet in future clinical studies.

Immune profiling of human metastatic PDAC tumors, stratified by glycemic status. A, Phenotypic markers used to define immune cell populations. B, Immune cell densities of liver biopsies for all 11 patients (glycemic status of each patient is denoted). C, Average of immune cell densities based on glycemic status of the patients (eight hyperglycemic and three nonhyperglycemic). D, Radar plots depict the density of cells positive for selected functional markers (HLA-II, Ki-67, and PD-L1) with the indicated myeloid cell populations comparing hyperglycemic (red) and nonhyperglycemic patients (green). E, Heatmaps depict normalized frequencies of pairwise spatial proximities between immune cell populations within 20 μm in nonhyperglycemic (left) and hyperglycemic (right) tumors. Corresponding network diagrams illustrate spatial associations. Statistically significant pairwise proximities are listed (P < 0.05). P values were calculated using a one-sided Mann–Whitney U test and corrected for multiple hypothesis testing using the Benjamini–Hochberg procedure. APC, antigen-presenting cell; DC, dendritic cell; SP, single positive.

Supplemental Figure 3 - Glucose is important for bone marrow-derived macrophage polarization and function in vitro. A.) Relative KPC cell survival when co-cultured with M1- or M2-like macrophages in high vs low glucose, assessed via quantification of KPC cell luminescence (similar experiment as Supplemental Figure 2). Macrophage stimulant controls (LPS and IFN- or IL-4 without BMDM) are depicted for reference. B.) Depiction of baseline KPC growth in 25 mM or 2.5 mM glucose conditions. C.) Measurement of cell culture media glucose concentrations in M1- and M2-like macrophages over a 5-day experiment. Day 0 was defined as when fresh media and macrophage stimulants were added. D.) Western blot of M1- and M2-like macrophages harvested at indicated time points. Cell culture media (initially 25 mM) was not changed during the experiment. Macrophage stimulants (LPS and IFN- or IL-4) were added at timepoint 0 hours and were not re-dosed. Naïve macrophages (M0) are shown for reference. E.) Relative mRNA iNOS and CD206 expression of M0, M1-like, and M2-like macrophages in high and low glucose after 24 hours of culture. F.) Western blot depicting iNOS and arginase protein levels of M1- and M2-like macrophages in indicated glucose concentrations for 48 hours. G.) DCFDA assay showing relative production of reactive oxygen and nitrogen species of M0, M1-, and M2-like macrophages in high and low glucose media. * indicates p<0.05, ** indicates p<0.01, *** indicates p<0.001, ns; not significant.

Supplemental Figure 1 - Immune competence is required to restrain pancreatic cancer growth in vivo. A.) Overall survival of mice with different immune backgrounds bearing identical KPC orthotopic pancreatic tumors. NSG mice lack functional T, B, and NK cells and have severely impaired innate immunity including macrophages. Athymic nude mice are deficient in T cells but retain functional innate immune cells including macrophages. C57Bl/6J mice possess an intact adaptive and innate immune system. Mice did not receive any therapy. Median survival was plotted with the Kaplan-Meier method and compared using the log-rank test. Gross photographs of spontaneous liver metastases and primary pancreatic tumor in NSG mice (B.) and normal liver with large primary pancreatic tumor in C57Bl/6J mice (C.).

Supplemental Figure 8 - Flow cytometric analysis of additional immune and stromal cell infiltration following single and combination treatments in pancreatic tumors. A.) Representative flow cytometry plots illustrating the effects of vehicle control, D30, PLX3397, PLX3397 + D30, PF-4136309, and PF-4136309 + D30 on stromal fibroblast populations within pancreatic tumors. Tumor-derived single-cell suspensions were first gated on viable cells based on forward- and side-scatter properties. The CD45− fraction was subsequently analyzed to identify CD140α+ fibroblasts, as shown by the indicated gates. Enlarged panels display representative SSC-A versus CD140α plots for each treatment group, with percentages of CD140α+ cells indicated. Bar graphs summarize the proportion of CD45+ immune cells and CD140α+ fibroblasts across treatment groups. B-C.) Representative flow cytometry plots and quantitative analyses showing the impact of vehicle control, D30, PLX3397, PLX3397 + D30, PF-4136309, and PF-4136309 + D30 on intratumoral B cells (B.) and NK1.1+ cells (C.). Representative contour plots display the frequencies of B220+ and NK1.1+ populations within the parent gates, with percentages shown in each panel. Bar graphs summarize the proportion of B220+ B cells and NK1.1+ cells across treatment groups. Data are presented as mean ± S.D. from independent biological replicates, with each dot representing an individual tumor. Statistical significance was assessed using one-way ANOVA with multiple-comparison correction. ns, not significant; * indicated p<0.05; ** indicated p<0.01; *** indicated p<0.001. All gates were defined using FMO controls and applied uniformly across all samples.