India has one of the highest incidences of oral cavity squamous cell carcinoma (OCSCC) in the world. However, unfortunately, OCSCC is typically detected during later stages, which drastically drops the long-term survival rate of patients (from about 80% to 20–60%). Traditionally, detection is via physical exam and a tissue biopsy — invasive tests, that require the presence of a lab and trained medical practitioner
Strand Life Sciences, in collaboration with the University of Chicago and HCG, has developed an ultrasensitive liquid biopsy panel that utilises a simple oral rinse to look for the presence of 7 key genes commonly seen in patients with OCSCC.
As explained in our last post, the key challenge in developing a liquid biopsy test is the low levels of DNA present in the samples, thereby requiring a low detection threshold, or lower limit of detection (LLOD), which will require the sample to be sequenced at very high depth, typically 10,000 reads per genetic locus (10,000x) or higher. Data from the high depth sequencing is put through a bioinformatics pipeline to identify variants at very low concentrations. For example, at 10,000x, variants occurring in 1 out of the 10,000 reads, i.e., at 0.1% VAF (!), can be detected by such an approach.
However, due to the deep sequencing the data tends to also carry high levels of ‘noise’.i.e., false positives’. Sophisticated noise profiles are used to model those variants that are likely artifacts due to various errors in the laboratory or sequencing processes, so that those that remain are most likely due to the underlying biological condition, i.e., the tumor.
Our team at Strand has been able to fine tune the OCSCC panel to detect variants with 100% analytical sensitivity (see image 1) and 99.97% analytical specificity, i.e. synthetically injected variants were picked up at an LLOD of 0.1% 100% of the time, while the high specificity was achieved with a noise model that removed false positives. Further, we saw that the panel matches the detection in solid tumour (FFPE) samples 93.4% of the time overall — or has 93.4% clinical concordance with FFPE blocks, with 87.5% concordance in early stage tumors. The high concordance in the early stage is especially promising, indicating the possibility of early-stage OCSCC detection.
We are very excited about these early results, and we’re exploring a number of possibilities to develop this test so it can be truly field-deployable and help detect oral cancer as early as technologically possible.