A liquid biopsy-based test for detection of early cancer in high risk cohort
Demographic characteristics are already used to select patients at high risk for cancer screening. For example, low dose CT screening use criteria of age 50 – 80, current smoking status and pack year history for patients at risk of lung cancer. By integration of demographic data, which are known to be predictive of risk and molecular insights from proteomics data, a novel approach can be developed to comprehensively understand disease dynamics and enhance early detection of cancer.
At Oxford Cancer Analytics, we use state-of-the-art mass spectrometry to provide a comprehensive understanding of previously unknown blood-based proteomic biomarkers in early cancer, analysing thousands of proteins (~3000-6000) with a focus on low-abundance proteins to discover novel biomarkers. We combine this with DECancer, our advanced machine learning pipeline tailored to proteomics liquid biopsy data to dissect complex biomarkers/pathways, which combines a novel approach of data augmentation for proteomics and feature selection for a succinct biomarker panel.1 With DECancer and a data-independent acquisition for proteomic analysis, we have shown in 600+ patients, 94% sensitivity and 99% specificity for all-stage lung cancer vs control and 86% and 99% specificity for early-stage lung cancer, with <20 biomarkers selected from hundreds, representing unprecedented early detection accuracy. We have also translated our proprietary biomarker panel into an immunoassay blood test that can be conducted affordably and routinely. This has been analytically validated for precision, analytical sensitivity, specificity, measuring range, linearity and trueness and clinically validated to 86% sensitivity and 99% specificity for lung cancer when combining the readout from our immunoassay panel with demographic characteristics. We have since expanded our test to a multi-cancer biomarker panel.
In this study, we propose a highly sensitive and specific multi-cancer screening test comprised of our blood based biomarker panel with basic demographic characteristics.
References
1. Halner A, Hankey L, Liang Z, Pozzetti F, Szulc D, Mi E, Liu G, Kessler BM, Syed J, Liu PJ. DEcancer: Machine learning framework tailored to liquid biopsy based cancer detection and biomarker signature selection. iScience. 2023;26(5):106610. doi: 10.1016/j.isci.2023.106610
1. Primary objective: apply OXcan's liquid biopsy blood-based assay to a large and independent cohort of high risk patients (patients who have been identified and screened by the PLCO trial). We will determine the sensitivity and specificity of our test for effective early cancer detection in comparison to standard of care screening methods on the PLCO cohort. Our test will be applied to plasma or serum samples most proximal to diagnosis and used to classify these into likely cancer or no cancer. If successful, we will look to extend our study to samples collected with greater lead time prior to diagnosis.
- Sensitivity and specificity will be determined by comparison to whether the patient was diagnosed with cancer or not.
- Sensitivity and specificity will be similarly calculated for standard of care screening methods and compared to the metrics from our test.
2. Secondary objective: determine the sensitivity and specificity of OXcan's liquid biopsy blood-based test combined with demographic characteristics on the PLCO cohort.
Ella Mi (Oxford Cancer Analytics)
Junetha Syed Jabarulla (Oxford Cancer Analytics)
Emma Mi (Oxford Cancer Analytics)