Multi-Cancer Blood Test Leverages Machine Learning to Detect and Classify Underlying Disease

However, existing methods have not fully tapped into the potential of the methylome to identify subtle but crucial early disease indicators. Now, a cutting-edge genome-wide methylome enrichment platform has the potential to detect early-stage diseases, thanks to the breadth of information captured through the whole methylome approach.

Adela, Inc.’s (New Haven, CT, USA) proprietary platform specializes in isolating the highly informative (methylated) regions of the genome, allowing it to capture and preserve a wide range of genomic information during sequencing more efficiently than other platforms that rely on enzymatic or chemical treatment (bisulfite conversion). The platform's potential edge in identifying early-stage diseases lends itself to various applications in multi-cancer early detection and cancer management, including minimal residual disease detection and disease monitoring. Adela's genome-wide methylome enrichment platform gathers information from small amounts of cell-free DNA and employs machine learning to identify and categorize the underlying disease. Initially developed for cancer detection and management, the technology is expected to be applied to other conditions beyond cancer in the future.

Recent studies show that Adela's genome-wide methylome enrichment platform can effectively detect a wide range of diverse cancers in their early stages (stage I and II), when treatments are most likely to succeed. These findings highlight the potential for using the platform in a multi-cancer early detection screening context. Strong early-stage cancer detection with low circulating tumor DNA levels also holds promise for guiding treatment decisions and monitoring recurrence in patients who have completed curative-intent cancer treatments.

The analysis included bladder, breast, colorectal, endometrial, esophageal, head & neck, hepatobiliary, lung, ovarian, pancreatic, prostate, and renal cancers. The study cohort consisted of cancer cases (individuals with newly diagnosed, treatment-naïve cancer) and age- and gender-matched non-cancer controls from multiple biobanks. The data presented stems from an interim training readout. Cancer cases were differentiated from controls with an area under the curve (AUC) of 0.94, while AUCs for individual cancer types ranged from 0.89 to 0.99. AUCs by stage were as follows: 0.92 (stage I), 0.95 (stage II), 0.95 (stage III), and 0.97 (stage IV).

"The presentation of these results, demonstrating the platform's strong ability to distinguish between cancer cases and controls in multiple different cancer types is an important step forward in the development of Adela's platform," said Dr. Daniel De Carvalho, Chief Scientific Officer of Adela.

Related Links:
Adela, Inc.