Droplet Biosciences, a diagnostics company developing lymph-based liquid biopsy tests, announced on Tuesday a collaboration with NVIDIA Corp (NASDAQ:NVDA) to detect residual cancer within 24 hours of surgery — weeks earlier than traditional blood-based monitoring.
By pairing lymphatic fluid analysis with GPU-accelerated genomics tools from Nvidia, the company has also sharply reduced computational turnaround times.
•NVIDIA shares are experiencing downward pressure. Why is NVDA stock trading lower?
A New Timeline For Residual Cancer Detection
Unlike conventional minimal residual disease (MRD) tests that rely on blood plasma weeks after surgery, Droplet collects lymphatic fluid just 24 hours after tumor removal.
According to Wendy Winckler, chief scientific officer at Droplet Biosciences, the fluid contains significantly higher concentrations of tumor-derived DNA because it is sourced near the tumor site rather than diluted in circulating blood.
The company reports that post-surgical lymphatic fluid holds 130 times more tumor-derived DNA molecules than time-matched blood plasma, translating to roughly double the sensitivity in detecting rare residual cancer cells.
Earlier detection may allow clinicians to adjust treatment plans sooner, including escalating therapy for patients with remaining disease.
Droplet's first Clinical Laboratory Improvements Amendments-validated test targets HPV-negative head and neck cancer. Late-stage proof-of-concept programs are also underway in lung and bladder cancers, with potential applicability across solid tumors.
Addressing Turnaround and Compute Bottlenecks
Although next-generation sequencing requires about five days in the lab, Droplet sought to compress bioinformatics timelines. Its prior CPU-based pipelines created bottlenecks, particularly given the high sequencing depth — often exceeding 10,000x raw coverage — required to identify rare variants.
Leveraging NVIDIA Parabricks For Cancer Detection
To accelerate processing, Droplet implemented Nvidia Parabricks, a GPU-accelerated genomic analysis suite.
Zhuosheng Gu, senior director of informatics, R&D, said alignment times dropped from as long as 36 hours per sample to under three hours using Parabricks' FQ2BAM workflow. Variant calling decreased from over 10 hours to about one hour.
The company is also working to implement GPU-accelerated tools for unique molecular identifier processing, targeting a reduction from more than 15 hours to under two hours.
Overall, Droplet expects to shrink its analysis timeline from 10 days to fewer than five, with a goal of under two days.
Despite higher hourly GPU costs, the shorter runtimes have produced an estimated 30% reduction in total computational expenses, according to the company.
On Monday, under the nonexclusive agreement, Nvidia said it would invest $2 billion in Coherent Corp. (NYSE:COHR) to support research and development, future capacity expansion and operations as the company strengthens its U.S.-based manufacturing footprint.
NVDA Price Action: Nvidia shares are down 1.55% at $179.54 at the time of publication on Tuesday, according to Benzinga Pro data.
