Delays in cancer diagnosis for patients with non-specific symptoms (NSSs) lead to poorer outcomes. Rapid Diagnostic Clinics (RDCs) expedite care, but most NSS patients do not have cancer, highlighting the need for better risk stratification. This study aimed to develop biomarker-based clinical prediction scores to differentiate high-risk and low-risk NSS patients, enabling more targeted diagnostics.

Retrospective and prospective cohort study.

Secondary care RDC in London.

Adult patients attending an RDC between December 2016 and September 2023 were included. External validation used data from another RDC.

The primary outcome was a cancer diagnosis. Biomarker-based risk scores were developed using Latent Class Analysis (LCA) and Least Absolute Shrinkage and Selection Operator (LASSO). Model performance was assessed using logistic regression, receiver operating characteristic curves (AUROC) and decision curve analysis.

Among 5821 RDC patients, LCA identified high white cell count, low haemoglobin, low albumin, high serum lambda light chain, high neutrophil-to-lymphocyte ratio, high serum kappa light chain (SKLC), high erythrocyte sedimentation rate (ESR), high C-reactive protein (CRP) and high neutrophils as cancer risk markers. LASSO selected high platelets, ESR, CRP, SKLC, alkaline phosphatase and lactate dehydrogenase. Each one-point increase in score predicted higher odds of cancer (LCA: AOR 1.19, 95% CI 1.16 to 1.23; LASSO: AOR 1.29, 95% CI 1.25 to 1.34). Scores ≥2 predicted significantly higher cancer odds (LCA: AOR 3.79, 95% CI 2.91 to 4.95; LASSO: AOR 3.44, 95% CI 2.66 to 4.44). Discrimination was good (AUROC: LCA 0.74; LASSO 0.73). External validation in 573 patients confirmed predicted increases in cancer risk per one-point LASSO score rise (AOR 1.28, 95% CI 1.15 to 1.42), with a borderline increase for LCA (AOR 1.16, 95% CI 1.06 to 1.27).

Biomarker-based scores effectively identified NSS patients at higher cancer risk. LCA captured a broader biomarker range, offering higher sensitivity, while LASSO achieved higher specificity with fewer markers. These scores may also help detect severe benign conditions, improving RDC triage. Further validation is needed before broader clinical implementation.