Inflammatory transcriptome dominates Indian AML: High iscore predicts poor outcomes even in genomically favorable subtypes and reveals regional immune-molecular divergence

Abstract Inflammation is increasingly recognized as a key modifier of AML outcomes, independent of genomic risk. A validated inflammation gene-expression score (iScore) predicts relapse and inferior survival in multiple cohorts, particularly in African ancestry AML. Its role in Indian AML—a genetically and environmentally distinct population—remains unexplored. We performed the first Indian immune transcriptomic profiling of AML to quantify inflammation, assess prognostic value, and explore genomic and regional associations. We prospectively analyzed 32 newly diagnosed Indian AML patients (median age 41 years; range 22–68; 56% male), using the Oncomine Immune Response Research Assay, a 395-gene panel covering 36 immune pathways. Genomic classification per ELN 2022 identified 8 core-binding factor (CBF) AML, 9 NPM1-mutant AML, 4 CEBPA double-mutant, 7 AML with myelodysplasia-related changes (MRC), and 4 other/unclassified. Patients were regionally distributed as North (n=12), South (n=11), East (n=5), and West (n=4). An iScore was calculated using expression of 25 canonical inflammatory genes (e.g., IL1B, TNF, CXCL8, CD274), normalized to healthy controls. Unsupervised k-means clustering (k=2) revealed natural immune subtypes: High-inflammation (n=19, 59%) and Low-inflammation (n=13, 41%). Median iScore across the cohort was 1.84 (range: 0.83–3.72), significantly higher than prior non-Indian cohorts (reference median ~1.2, p<0.01). Of 32 patients, 21 (65.6%) were above the international iScore high-risk threshold. CR rates after standard 7+3 induction were 68.4% (13/19) in high-iScore patients vs. 92.3% (12/13) in low-iScore patients (p=0.12). At 12-month follow-up, 1-year overall survival (OS) was 52.6% in the high-iScore group vs. 84.6% in the low-iScore group (p=0.008, log-rank). Median relapse-free survival (RFS) was not reached in low-iScore patients, while it was 8.6 months in high-iScore patients. On multivariable Cox regression adjusting for age and ELN risk, high iScore remained independently predictive of inferior OS (HR 3.1, 95% CI 1.2–7.9, p=0.016). Among ELN 2022 favorable-risk cases (n=14), 9 patients had high iScore, of whom 5 relapsed within 1 year (55.6%). In contrast, all 5 patients with favorable genetics and low iScore remained in continuous remission (0 relapses). In CBF-AML specifically, 4 of 5 high-iScore cases relapsed, compared to 0 of 3 low-iScore CBF-AML. Median iScore for CBF-AML was 2.47, significantly higher than NPM1-mutant AML (median 1.76) and CEBPA double-mutant AML (median 1.28; p=0.04). Gene set enrichment revealed overexpression of TLR signaling, IL-1, TNF, and NF-κB pathways in high-iScore CBF-AML. NPM1-mutant AML showed immune heterogeneity: 4/9 were high-iScore (2 relapsed), 5/9 were low-iScore (0 relapses). Among 7 MRC-AML cases, 3 clustered as high-inflammation and expressed PD-L1, CTLA4, and TIGIT, suggesting an immune-suppressed phenotype. Regional immune profiling revealed striking variation: North Indian patients had the highest mean iScore (2.36), driven by CBF predominance (6/12 cases). Southern patients showed greater MRC frequency (5/11) with checkpoint upregulation but lower IL1B/TNF signaling. Western and Eastern cases were fewer (n=4, n=5) but trended toward mixed immune states. These findings reflect a north–south divergence in AML biology, with younger, high-iScore CBF AML in North India and older, immune-exhausted MRC AML in the South. Our study reveals that high inflammatory signatures are common in Indian AML and independently predict poor outcomes even in genomically favorable subsets. iScore unmasked early relapses in 64% of favorable risk patients with high inflammation and identified 100% remission in low-iScore favorable-risk AML. These findings suggest that inflammation modifies AML trajectory regardless of genotype. Importantly, high-iScore patients overexpressed T-cell exhaustion and checkpoint genes, raising therapeutic opportunities with corticosteroids or immune checkpoint inhibitors. This is the first Indian AML study integrating genomics with immune transcriptomics. It provides evidence for an inflammation-driven AML subtype, prevalent in India, with adverse prognosis and region-specific patterns. Larger validation cohorts are underway to refine iScore thresholds and evaluate immunomodulatory therapy for high-iScore patients. These findings advocate for integrating immune biomarkers into AML risk stratification especially in underrepresented global populations.