100P - Standard serum biomarkers to help predict a cancer diagnosis in patients with non-specific symptoms: Data from Guy´s rapid diagnostic clinic

Background

Rapid diagnostic services (RDS) are a two-week wait pathway for patients with non-specific symptoms (e.g., weight loss and fatigue), concerning of malignancy, that do not fulfil criteria for a site-specific cancer referral. Notoriously, patients without clear red flag symptoms have longer time to diagnosis and worse outcomes. RDS are being expanded across England to provide a new referral route for this cohort of patients to improve time to diagnosis and the patient journey. This study aimed to create a cancer risk stratification tool based on levels of standard serum biomarkers to help clinicians triage patients attending Guy’s RDC.

Methods

We analysed a panel of standard serum biomarkers assessed at referral for patients seen at Guy´s RDC between December 2016 and April 2022. Latent class analysis (LCA) was applied to investigate classes of individuals characterized based on their biomarker values (according to medical cut-offs), which were then examined as predictors of cancer using multivariate logistic regression. The abnormal biomarkers identified within LCA class with higher risk of cancer were used to create a scoring system, which was then developed into a cancer stratification tool.

Results

LCA identified four biomarker profiles: 1) normal values for all markers (51% of population); 2) abnormal values for haemoglobin, creatinine, and inflammatory markers (28%); 3) abnormal values for neutrophil to lymphocyte ratio, neutrophils, and inflammatory markers (12%); 4) abnormal values for multiple (n=10) markers (8%). Compared with class 1, class 4 had an increased risk of cancer (Odds ratio (OR) 6.33; 95% Confidence Interval (CI) 5.92-6.77). Based on the biomarkers identified in class 4 we created a score giving one point to each abnormal biomarker. A score of >=2 was associated with an increased risk of cancer (OR 3.09; 95%CI 2.91-3.27), as compared to a score of <2 (Area under the curve for test-training split: 0.68, C-index: 0.7).

Conclusions

Our innovative approach to triage patients with non-specific symptoms in the RDC for cancer diagnosis suggests that a combination of 10 standard biomarkers could be used to help facilitate a more efficient diagnosis of cancer when presenting with vague symptoms.

Editorial acknowledgement

Clinical trial identification

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.