Development of biomarkers for the early detection of cancer by integrating novel ultra-rapid microfluidics
John Timms, Geraint Thomas and Genetic Microdevices
Cancer is a major cause of mortality worldwide and a huge burden on society. It is clear that earlier diagnosis reduces cancer mortality and in turn, societal and economic burden. However, there is currently a lack of early detection biomarkers to accelerate diagnosis or to screen asymptomatic or high-risk populations. To address this issue, our group is employing novel proteomic technologies applied to pre-diagnosis serum samples sourced from the UKCTOCS Longitudinal Women’s Cohort (UKLWC) (1), with the aim of identifying improved early detection biomarkers. We have identified numerous potential candidates and by applying novel machine learning approaches to the serial, multi-analyte data generated, we have derived models that outperform current markers (2-7).
Biomarker verification and validation is heavily reliant on antibody (Ab)-based detection and whilst commercial reagents and tests are often available, they can lack sensitivity, specificity, reproducibility or be expensive and time-consuming to perform. This creates a bottleneck in discovery and improved methods are needed for validation and assay development for clinical translation. Our hypothesis is that antibody-capture mass spectrometry (MS) and/or antibody-detection in microfluidic devices will improve assay sensitivity, specificity and speed. The project thus aims to develop assays for a set of promising biomarkers, collect data from relevant case control samples using these assays and build and test multi-marker models capable of accurate, early detection of ovarian, pancreatic and colorectal cancers.
The first objective is to screen commercial Abs for immunocapture of a selection of promising biomarker candidates from sera using small-scale columns with detection of derived tryptic peptides by MS; so-called immunocapture MS (IMS) (8). If robustly detectable, synthetic, isotope-labelled versions of peptides will be used as standards to develop assays for these proteins. Detection limits, linearity, reproducibility, normal ranges and protocols for the assays will be established.
Our industrial partner Genetic Microdevices, have developed CycloChip, a novel microfluidic, electrophoretic, fluorescence detection device, allowing separation and detection of biomolecules at unprecedented resolution (100-500x narrower bands than capillary electrophoresis), attogram/μL sensitivity and seconds to minutes per analysis. The second objective therefore is to develop CycloChip for ultra-rapid, miniaturised protein assays. Firstly, conditions will be optimised for on-chip detection of the purified recombinant target proteins using Alexa488-tagged Abs. We have feasibility data showing on-chip resolution and detection of an antibody-antigen complex at the pg/uL level within minutes. Pre-mixed targets and Ab will then be spiked into serum and resolving power, specificity, limits of detection and reproducibility assessed, using Cy3 and Cy5-labelled recombinant proteins for calibration. Detection of multiple targets on the same chip and use of multi-channel CycloChips for parallelisation will also be explored. Testing will then be extended to the endogenous proteins in test sera, comparing CycloChip assay performance with commercial ELISAs and the developed IMS assays.
Once fit-for-purpose assays have been established, they will be applied to serial serum samples taken prior to diagnosis of ovarian, pancreatic or colorectal cancer and matched serial control samples. Data will be analysed using standard univariate tests and predictive performance calculated for different time-to-diagnosis groups. Data will also be added to existing serial data for these samples and used to build multi-marker longitudinal and network models using novel machine learning approaches and performance and lead times of detection determined. The project thus aims to deliver assays for the rapid, accurate measurement of a set of promising biomarkers and novel predictive models for the early detection of ovarian, pancreatic and colorectal cancers. The work will also accelerate the development of novel diagnostic devices by our industrial partner.
2) Blyuss O, et al. Comparison of longitudinal CA125 algorithms as a first line screen for ovarian cancer in the general population. Clin Cancer Res 2018 July 3
3) Whitwell HJ, et al. Parenclitic networks for predicting ovarian cancer. Oncotarget 2018;9(32):22717-22726
4) Krishnan S, et al. Evidence of altered glycosylation of serum proteins in pre-diagnosis pancreatic cancer cases. Int J Mol Sci. 2017;18(12)
5) Mariño I, et al. Change-point of multiple biomarkers in women with ovarian cancer. Biomedical Signal Processing and Control 2016; 33:169-177
6) Thomas DS, et al. Evaluation of CEA, CYFRA21-1 and CA125 for the early detection of colorectal cancer using longitudinal preclinical serum samples. British J. of Cancer 2015 113, 268–274
7) O’Brien DP, et al. Serum CA19-9 levels are significantly up-regulated up to 2 years prior to diagnosis with pancreatic cancer: implications for early disease detection. Clinical Cancer Research 2015 Feb; 21(3): 622-31