A patient-tailored protein marker blood test for monitoring treatment response and postoperative recurrence across multiple cancers.

Abstract

e15047 Background: ctDNA-based minimal residual disease (MRD) assays are increasingly used for post-treatment surveillance but remain costly and technically complex, limiting adoption, particularly in low- and middle-income countries (LMICs). Protein tumor markers (PTMs) are inexpensive and widely available but lack disease specificity and are limited by inter-patient heterogeneity. We developed OncoGraph, a patient-tailored approach that identifies abnormally elevated PTMs at baseline and tracks their longitudinal dynamics for individualized monitoring of recurrence risk and treatment response. Methods: Pretreatment blood samples were analyzed using OncoSeek, a validated multi-cancer early detection assay quantifying a standardized PTM panel (e.g., AFP, CEA). PTMs exceeding predefined thresholds were designated as patient-specific markers and serially monitored after curative surgery or during systemic therapy. Postoperative surveillance in liver cancer and lung cancer used PTM changes at one month after surgery to MRD, while treatment monitoring in advanced lung cancer and lymphoma used PTM dynamics after two treatment cycles to evaluate response and outcomes. Results: In liver cancer, 18/30 patients (60.0%) had elevated baseline PTMs. Three of 18 (16.7%) showed insufficient postoperative PTM decline (<25%) and all recurred within 95 days, with a shorter RFS than patients achieving significant PTM reduction (≥25%) (median RFS, 42 vs 385 days; HR=3.5; P<0.05). Among surgically treated lung cancer patients, 16/23 (69.6%) had elevated baseline PTMs. One patient with insufficient PTM decline recurred at 186 days, whereas patients with significant PTM reduction (15/16, 93.8%) had a median RFS of 336 days (HR=5.3; P=0.07). In advanced lung cancer, 21/23 patients (91.3%) exhibited elevated baseline PTMs. After two treatment cycles, six (28.6%) patients with insufficient PTM decline (<25%) had significantly shorter PFS than those with marked PTM reduction (15/21, 71.4%) (median PFS, 177 vs 434 days; HR = 3.6; P < 0.01), with a trend toward inferior OS (median OS, 431 vs. 922 days; HR = 2.5; P = 0.07). In lymphoma, baseline-abnormal PTMs were observed in 16/116 patients (13.8%). After two treatment cycles, four (25.0%) patients with insufficient PTM reduction (<50%) was associated with poorer PFS (HR = 6.2; P = 0.13) and inferior OS compared with greater PTM decline (12/16, 75.0%) (median OS, 506 days vs not reached; HR = 5.3; P = 0.06). Conclusions: OncoGraph enables patient-tailored PTM-based monitoring by addressing the limitations of conventional method. Early PTM dynamics correlated with postoperative recurrence risk and treatment outcomes across multiple cancers. Given its low cost, minimal technical requirements, and broad applicability, OncoGraph represents a scalable blood-based monitoring strategy, particularly suitable for LMIC.