DIAGNOSTIC PERFORMANCE OF NOVEL MULTIMODAL BIOMARKERS FOR EARLY CANCER DETECTION IN INDIA: AN EXPLORATORY DIAGNOSTIC ACCURACY STUDY

Abstract

Background: Cancer screening coverage remains low in India, while blood- and body-fluid-based biomarkers could potentially complement organ-specific screening. This study evaluated the diagnostic performance of multiple novel biomarker domains for distinguishing cancer from non-cancer states in an Indian cohort, with a specific exploratory analysis of stage I-II disease. Materials and Methods: An exploratory diagnostic accuracy study included 60 participants from 14 Indian states. Department of Pathology, Shri Sathya Sai Medical College & Research Institute, Sri Balaji Vidyapeeth University (DU), Chennai Campus, Ammapettai, Nellikuppam Post, Chengalpattu District, Tamil Nadu, India. Study period was between July 2024- June 2026 Final diagnostic categories were cancer (n=15), high-risk (n=7), benign (n=22), and no evidence of cancer (n=16). Peripheral venous blood, saliva, and first-void urine were collected under a standardized pre-analytical SOP with barcoding and time-temperature monitoring. Nine systemic biomarker domains were evaluated: DNA methylation, fragmentomics, cell-free RNA (cfRNA), extracellular vesicles (EV), circulating tumour cells (CTC), tumour-educated platelets (TEP), protein biomarkers, autoantibodies, and metabolomics. Salivary and urinary HPV markers were assessed in relevant site-specific subsets. Molecular and cellular assays incorporated sequencing, RT-qPCR, immunoassay, microfluidic/cell-based enrichment, mass spectrometry, and methylation-based platforms. Receiver operating characteristic analysis compared cancer with all non-cancer categories; a separate exploratory analysis compared stage I-II cancers with non-cancer participants. Results: The cohort had a mean age of 55.0 ± 7.3 years; 32 (53.3%) were male, and all 60 participants (100%) had no recorded prior cancer-screening history. Six of 15 cancers (40.0%) were stage I-II. Biomarker distributions differed significantly across diagnostic categories for all nine systemic domains (Kruskal-Wallis p<0.001). For cancer versus non-cancer, AUCs ranged from 0.919 for TEP to 0.933 for CTC. Methylation, CTC, and metabolomics each achieved 100% specificity at their exploratory Youden-derived thresholds, with 86.7% sensitivity. In the stage I-II analysis, AUCs were lower, ranging from 0.798 to 0.833, with wide confidence intervals because only six early-stage cancers were available. All nine systemic biomarkers showed strong positive correlations with cancer stage (Spearman rho 0.963-0.979, all p<0.001). Tissue-of-origin classification was concordant with the recorded primary site category in all 15 cancer cases. Conclusion: Multiple circulating biomarker domains showed strong discrimination between cancer and non cancer states in this exploratory Indian cohort, but performance attenuated for stage I-II disease. The findings support multimodal biomarker development while also demonstrating the need for larger prospective, screening based Indian validation cohorts with standardized assay methods, blinded reference standards, and external validation before clinical implementation.