PANoptosis is a relatively new concept in cell biology. Unlike classical apoptosis, the quiet and orderly dismantling of a cell, or necroptosis and pyroptosis, which are more explosive and inflammatory forms of cell death, PANoptosis involves a coordinated interplay among all of these pathways at once. The process is orchestrated by a molecular complex called the PANoptosome, which assembles as part of the innate immune response. Because dysregulated cell death and inflammation are intimately linked to tumor development, scientists have begun asking whether the genes governing PANoptosis might serve as markers of cancer behavior. Previous work has implicated the protein CASP8 as a central hub in tumor PANoptosis signaling, and related signatures have been explored in colon, gastric, liver, and pancreatic cancers.

The research team, led by investigators from Renmin Hospital of Wuhan University and collaborating institutions, began with a curated list of 65 PANoptosis-related genes drawn from prior literature, spanning pyroptosis, apoptosis, and necroptosis. They retrieved transcriptome and clinical data from The Cancer Genome Atlas and the Gene Expression Omnibus, ultimately analyzing 183 esophageal cancer samples from TCGA and 179 esophageal squamous cell carcinoma samples from the GEO dataset GSE53625. After correcting for batch effects between the two databases, the researchers compared gene expression between tumor and normal tissue, identifying 42 PANoptosis-related genes that were differentially expressed in tumors. Strikingly, all 42 were upregulated in cancerous tissue, and 19 of them showed statistically significant associations with patient survival.

To understand how these genes shape disease, the team applied consensus clustering, an algorithm that groups patients according to shared expression patterns. Two distinct molecular subtypes emerged, designated PANcluster A and PANcluster B. Patients in cluster A lived significantly longer than those in cluster B, with the difference reaching statistical significance. Principal component analysis confirmed that the two groups were molecularly separable. When the researchers examined the tumor immune microenvironment using single-sample gene set enrichment analysis, cluster B showed higher infiltration by activated B cells, CD4-positive and CD8-positive T cells, dendritic cells, macrophages, mast cells, monocytes, natural killer T cells, and neutrophils. Gene set variation analysis revealed that cluster B was enriched in immune-related pathways, including natural killer cell-mediated cytotoxicity and antigen processing and presentation.

The next step was to distill this complexity into a practical prognostic tool. The researchers identified more than 1,800 genes that differed between the two clusters, then used least absolute shrinkage and selection operator regression together with multivariate Cox regression to narrow the field to five genes: IRF1, NEUROG3, TMIGD2, SHOX2, and TRBJ1-2. Each gene was assigned a coefficient, and a risk score was calculated for every patient as a weighted sum of expression levels. Patients whose scores exceeded the median were classified as high risk, while the remainder were classified as low risk. Across both the TCGA and GEO cohorts, high-risk patients fared significantly worse than their low-risk counterparts, and the model retained its predictive power in subgroup analyses of patients with lymph node involvement, stage IIB to IIIA disease, and T3 to T4 tumors—suggesting particular value for locally advanced esophageal cancer, where treatment decisions are most difficult.

The signature’s accuracy was quantified using receiver operating characteristic analysis. The area under the curve for predicting one-year, two-year, and three-year survival reached 0.730, 0.768, and 0.725 respectively, outperforming conventional clinical features. Multivariate Cox regression confirmed that the risk score remained an independent prognostic factor even after accounting for clinical stage and nodal status. The team then integrated the risk score with clinicopathologic variables into a nomogram—a graphical calculator that clinicians can use to estimate individual survival probabilities at one, two, and three years. Calibration plots showed close agreement between predicted and observed survival, and a concordance index comparison indicated that the risk score outperformed other clinical factors as a predictor of outcome.

Beyond prognosis, the risk score illuminated the immunological character of each patient’s tumor. High-risk patients showed elevated levels of activated mast cells, while low-risk patients displayed distinct stromal and immune score profiles. Tumor mutation burden, an established predictor of immunotherapy response, correlated significantly with the risk score. Expression of immune checkpoint genes including CTLA4, TNFRSF18, CD44, TNFSF18, CD276, and PDCD1LG2 was higher in the low-risk group. Tumor Immune Dysfunction and Exclusion scores, which estimate the likelihood of immune escape, were also higher among low-risk patients, a pattern the authors interpreted as suggesting a potentially more favorable response to PD-1 and CTLA-4 blockade in that group—though they caution that the TCGA patients in the analysis were treated before immunotherapy became standard, which may explain why high-risk patients nonetheless did worse.

The model also pointed toward differences in chemotherapy sensitivity. Using the OncoPredict platform, which estimates drug response from cancer cell line data, the researchers found that high-risk tumors were predicted to be more sensitive to oxaliplatin, gemcitabine, crizotinib, and camptothecin, whereas low-risk tumors showed greater sensitivity to paclitaxel. This alignment with existing clinical practice, in which paclitaxel is a mainstay for esophageal cancer, lends credibility to the computational predictions and suggests the signature could eventually help guide drug selection, including for patients who develop resistance to first-line agents or cannot tolerate standard chemotherapy.

Several of the five signature genes have independent biological interest. IRF1 is a transcription factor activated by viral infection and interferon signaling, and it has been shown to regulate PD-L1 expression through phase-separated condensates with KAT8, making it a potential lever for boosting antitumor immunity. TMIGD2, a costimulatory receptor on T cells and natural killer cells that shares sequence homology with CD28 and PD-1, correlates negatively with angiogenesis and epithelial-to-mesenchymal transition and has been proposed as a target for bispecific antibodies in tumors resistant to PD-1 blockade. SHOX2 methylation is already being studied as a biomarker for lung cancer screening, and the gene cooperates with STAT3 to drive metastasis in breast cancer.

The authors acknowledge important limitations. The analysis relied entirely on retrospective public datasets, which introduces potential selection bias, and the cohort was predominantly composed of esophageal squamous cell carcinoma, the dominant histology in China, so generalizability to adenocarcinoma populations remains uncertain. Laboratory experiments to validate the mechanisms were limited, and some clinically relevant variables, such as surgical technique and specific chemotherapy regimens, were not consistently available. Even so, the study adds esophageal cancer to a growing list of malignancies for which PANoptosis-based signatures show prognostic promise, and it offers a concrete framework—one molecular subtype, one five-gene risk score, and one nomogram—for translating the biology of inflammatory cell death into individualized predictions. Prospective clinical validation will determine whether this computational approach can move from research datasets into the oncology clinic, but for a disease where five-year survival remains stubbornly low, any tool that sharpens prognosis and therapy selection is a welcome advance.

Subject of Research: PANoptosis-based prognostic modeling and immune microenvironment analysis in esophageal cancer

Article Title: PANoptosis-based molecular subtype and prognostic model predict survival and immune landscape in esophageal cancer

Article References: Liu, Z., Liu, J., Liao, F., Li, W., Wang, J., & Zhang, C. (2024). PANoptosis-based molecular subtype and prognostic model predict survival and immune landscape in esophageal cancer. Clinical Cancer Bulletin, 3(1), Article 16. https://doi.org/10.1007/s44272-024-00021-z

Image Credits: AI Generated

DOI: 10.1007/s44272-024-00021-z

Keywords: esophageal cancer, PANoptosis, prognostic model, tumor microenvironment, immunotherapy, nomogram, LASSO regression, TCGA, immune checkpoint, biomarkers, cell death, chemotherapy sensitivity