Patient-derived colorectal microtumors predict response to anti-PD-1 therapy

Abstract

Immune checkpoint inhibitors have made remarkable impacts in treating various cancers, including colorectal cancer (CRC). However, CRC still remains a leading cause of cancer-related deaths. While microsatellite instability (MSI) CRC has shown positive responses to anti-PD-1 therapy, this subgroup represents a minority of all CRC patients. Extensive research has focused on identifying predictive biomarkers to understand treatment response in CRC. Interestingly, a growing number of clinical cases have reported favorable outcomes from a subtype of supposedly non-responder microsatellite stable (MSS) CRC, characterized by DNA polymerase ϵ (POLE) proofreading domain mutations with high tumor mutational burden (TMB). This subtype has shown a notable response, either partial or complete, to pembrolizumab as salvage treatment, often following significant disease progression. To improve efficiency, cost-effectiveness, and clinical outcomes, there is an essential need for a testing platform capable of promptly identifying evidence of anti-PD-1 response to inform treatment strategies. Here, we established a novel 3D ex vivo immunotherapy model using patient-derived tumor microexplants (or microtumors <1 mm) co-cultured with autologous peripheral blood mononuclear cells (PBMCs) from treatment-naïve CRC patients. We demonstrate that long-term ex vivo treatment with pembrolizumab induced a heterogeneous but appreciable interferon-gamma (IFN-γ) secretion, accompanied by infiltrating PBMCs. Intriguingly, a case study involving an MSS CRC phenotype harboring POLE mutation and associated ultrahigh TMB demonstrated a response to PD-1 blockade, potentially from the intratumoral immune cell population. Ultimately, this novel model could serve as a valuable tool in complementing clinical diagnostics and guiding personalized treatment plans for CRC patients, particularly those with specific phenotypes and mutational profiles.

Keywords: CRC; anti-PD-1; colorectal cancer; ex vivo; immune checkpoint inhibitor; immunotherapy; microsatellite instability; patient-derived explants.

Figure 1

Schematic illustrating the sequential steps in the co-culture of patient-derived microtumors and autologous PBMCs for anti-PD-1 (aPD1) treatment. (A) Isolation of PBMCs from whole blood through Ficoll density gradient centrifugation, prelabeling with CFSE. (B) Surgical removal and collection of tumor specimens for pathology diagnoses, followed by mechanical dissection into tumor microexplants (namely microtumors; approximately ≤ 1 mm). (C) Uniform suspension of PBMCs and microtumors in LLS microgels. (D) Co-culture in a perfusion platform under varied treatment conditions. Assessment of immune response involves (E) analyzing effluent media for cytokine production, (F) evaluating immune infiltration and anti-tumor function, and (G) performing immunohistology.

Figure 2

Evaluation of anti-PD-1 (aPD1) in patient-derived ex vivo models. (A) Patient-specific MMR status and other relevant information. (B) Schematic representation of the experimental workflow from blood draw and surgical resection to ex vivo co-culture of patient-derived microtumors and autologous PBMCs. (C) Illustration depicting the dynamic interaction between microtumors and immune cells within the 3D environment. (D) Typical tissue morphology (bright field) and immunofluorescence (IF) viability data using Calcein AM (live, green) and BOBO-3 Iodide (dead, red) after long-term ex vivo culture. (E) Detection of interferon-gamma (IFNγ) secretion from perfused media collected every 24 hours for 7 days in the 3D tumor-immune cells co-culture for MSI (left) and MSS (right) CRC cases. Statistical significance was shown for MOD1 only. There is no statistical significance for MOD2. Statistical analysis was performed using a two-way ANOVA followed by Tukey’s HSD post-hoc test for IFNγ level between the treatment groups at each time point. (technical replicates n=3 unless indicated otherwise, *p<0.05, **p<0.01, ***p<0.001). (F) TUNEL assay was employed to detect apoptotic cells, where the presence of CFSE+ immune infiltration correlates with regions of cell death. ROI: region of interest.

Figure 3

POLE mutations potentiate anti-tumor response to anti-PD-1 Treatment. (A) Patient information comparing case studies of CRC harboring POLE mutation with other MSS CRC cases. (B) Distribution of the POLE mutations within the POLE full-length sequence for MSS CRC MOD3 (top) and MSI CRC MOD7 (bottom). (C) Somatic mutation profiles in known cancer genes of CRC for the presented cases. (D) Mutation load reveals hypermutation status in MSS CRC with POLE mutation compared to other MSS and MSI phenotypes. (E) Typical IF viability data using Calcein AM (live, green) and BOBO-3 Iodide (dead, red) on day 4 and day 7 for the two case studies. (F) Detection of interferon-gamma (IFNγ) from effluent media collected every 24 hours for 7 days. Technical replicates, n=3 for all conditions. (G) Confocal images displaying resident immune cells in microtumors. An ROI highlights the detection of CD45+ intratumoral immune cell populations. Separate confocal channels show distinct fluorescent markers for actin (red), immune cells (green: CD45), nuclei (blue), and PD-L1 (white).

Figure 4

Multiplex immune cell profiling reveals heterogenous intratumoral immune populations in CRC tumors. (A) Histology images depict individual tumors with regions of interest (ROIs, blue box, n=10) selected for immune cell profiling. (B) IHC staining of TILs and immune checkpoint expression. (C) Representative example of cell segmentation and immune phenotyping. Tumor and stromal areas exhibit strong agreement with histology images from A. (D) Cell population densities (cells per mm²) show heterogeneous distribution among patients, even in those with similar CRC phenotypes. (E) T-cell population densities. (F) Proximity analysis reveals closer T cell distance to tumor cells in MOD1 (MSI) and MOD3 (MSS POLE) compared to other cases. (G) Cell density heat map reveals a high proportion of T cells in MOD1 compared to the rest of the samples. Statistical analysis was performed using One-Way ANOVA followed by Tukey multiple comparison test; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.