1180P - An integrated metabolomics-based platform for early-stage detection of multiple cancers

Background

While early disease diagnosis currently offers the most viable strategy for reducing cancer deaths, screening tests for accurate detection of early-stage cancer are yet awaited. Although a variety of multi-cancer detection (MCD) tests are currently under development, they all suffer from a suboptimal detection sensitivity for stage I/II cancers. We, therefore, adopted a distinct approach that integrated untargeted serum metabolomics with deep learning-powered data analytics to capture cancer-specific metabolite signatures. This approach provided yielded a powerful multi-cancer detection tool that could detect a total of thirty cancers across all stages, including the early stages, with high accuracy. Furthermore, the tissue of origin could also be assigned to 21 of the cancers.

Methods

Serum samples were obtained in a prospective, observational, multi-centric trial from a total of 6500 participants. Of these, 3500 participants (male and female) were newly diagnosed, but treatment-naïve, cancer patients. This group collectively covered a total of 30 cancers. The remaining 3000 samples were from non-cancerous participants and were labeled as normal controls. After extraction of the metabolites, the serum metabolome profile was generated for each sample by high-resolution mass spectrometry coupled to ultra-pressure liquid chromatography (UPLC-HRMS/MS). The resulting data was then deconvoluted using a set of machine learning algorithms to first distinguish between cancer-positive and cancer-negative cases, followed by a further analysis of samples from the former group to identify the likely tissue of origin of the cancer.

Results

Reveal that our test system, named as OncoVeryx, could indeed detect all 30 cancers in females and males with high accuracy. The average detection sensitivity obtained was 97.5%, at a specificity of 99.2%. Importantly, the high detection sensitivity was also retained for the cancer stages I and II. Furthermore, cancer detection could also be complemented with identification of the tissue of origin for 21 cancers.

Conclusions

OncoVeryx, thus provides a promising new MCD test that is particularly suited for early-stage cancer detection.

Clinical trial identification

CL/MD/2024/000007; Dated 23 April 2024.

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.