South Asian populations remain underrepresented in cancer genomics, despite elevated risk for certain malignancies and distinct clinical profiles. This gap is especially pronounced for British Bangladeshi and Pakistani communities. We analysed data from 57,416 individuals of Bangladeshi and Pakistani ancestry in the UK-based Genes & Health cohort, integrating electronic health records, cancer registry data, and whole-exome sequencing (n=43,462). Among them, 2,782 (4.8%) had cancer, with earlier onset across multiple types compared to national benchmarks. A phenome-wide case–control analysis identified 147 significant cancer–comorbidity associations, with larger effect sizes observed for topographically concordant cancer-comorbidity pairs, and stronger associations for systemic disorders such as obesity and hypertension. Exome-wide analyses revealed 39 variant-level and 31 gene-level associations, over 60% novel and enriched for rare coding variations. Notable loci included ARHGAP45, CBR1, FKBP6, NF1, and ZNF155, with several ancestry-specific associations. These findings highlight distinct cancer risk architectures in this South Asian subpopulation, emphasising the need for population-tailored risk models and screening strategies.