"Can we detect multiple cancers early with just one blood draw?"
As interest grows in diagnostic tests that quickly identify various types of cancer, the combination of Whole Genome Sequencing (WGS) and Artificial Intelligence (AI) is being highlighted as a key technology. The principle is simple. It broadly scans DNA fragments in the blood at the whole genome level, and AI identifies patterns related to cancer within that data.
DNA is the blueprint that contains our body's genetic information. Unlike tests that only look at a few specific genes, WGS reads the entire DNA (genome) to create a 'complete map'. However, the data obtained from reading the entire genome is vast and complex. This is where AI comes into play. It identifies subtle changes or recurring signals that are easily missed by the human eye and classifies them as potential cancer indicators.
The primary material used for multi-cancer early detection is 'liquid biopsy'. In the blood, there are fragments of DNA known as 'cell-free DNA (cfDNA)' that are released during the process of cell death and regeneration. If cancer is present, cfDNA is also released from cancer cells, but in the early stages, the amount is so small that it is difficult to distinguish.
Recently, there has been an increasing approach to read indirect signals shown by cfDNA rather than directly capturing cancer DNA. For example, utilizing characteristics such as the length and distribution of cfDNA fragments, or broadly scanning with 'low-density whole genome sequencing (lcWGS)' to lower costs while still securing signals. AI then combines various clues to filter out cancer signals. In simple terms, while WGS paints the overall picture, AI highlights suspicious traces within it.
There are also attempts to implement this approach into actual commercial tests. GC Genome, a company specializing in liquid biopsy and clinical genome analysis, recently announced the performance verification results of its AI-based multi-cancer early screening test 'ai-CANCERCH' at a Japanese liquid biopsy conference.
According to the company, ai-CANCERCH is a test that detects multi-cancer signals with just one blood tube, combining AI algorithms with whole genome analysis. It reported a specificity of 95.5%, overall sensitivity of 79.7%, and stage-weighted sensitivity of 80.2% in an external validation cohort. It showed high sensitivity for pancreatic cancer and biliary tract cancer, and also demonstrated a certain level of sensitivity for lung cancer, colorectal cancer, and breast cancer.
Rather than Replacing Standard Screening, “Filling Gaps”… Application Targets are also a Challenge
The reason these technologies are referred to as the 'key to multi-cancer early detection' is clear. They have a greater potential to capture overall patterns compared to methods that only look at specific genes, and can be designed as screening tools that are not limited to specific cancer types. Expectations are particularly high for cancers like pancreatic cancer and biliary tract cancer, where existing standard screenings are relatively weak.
Additionally, strategies like lcWGS that reduce cost burdens increase the likelihood of commercialization. Instead of conducting high-cost precise analyses of the entire genome, obtaining signals through broad scanning and interpreting them with AI means finding a compromise between realistic costs and efficiency.
However, experts emphasize that “good technology does not necessarily mean good screening.” Rather than immediately replacing well-established standard screenings like colonoscopy or mammography, it should establish its role as a tool to complement areas that still have significant gaps. Ultimately, for this technology to be effective in real-world settings, it must also be validated regarding 'who will receive it and when' and 'how to confirm after a positive result.'
