The University of Houston has landed a five-year, $2.5 million grant from the National Institutes of Health to push forward a light-based technology that could one day detect liver cancer from a simple blood draw instead of a tissue biopsy. The work, led by electrical and computer engineering professor Wei-Chuan Shih, centers on reading molecular signals inside microscopic particles called exosomes that cells release into the bloodstream.
According to the University of Houston, the grant supports research under NIH award number 1R01CA316276-01 and is aimed at hepatocellular carcinoma, or HCC — the most common form of primary liver cancer, as described by JCO Oncology Advances. Shih's technology uses an engineered chip containing metasurfaces and light to detect and analyze exosomes, and it's designed to work directly on blood plasma without the extensive purification steps that typically slow such tests down.
Why Skipping the Biopsy Matters
The appeal of this approach is straightforward: liver biopsies are invasive, and catching cancer early tends to make an enormous difference in outcomes. Per the university, researchers plan to study patients with early-stage HCC after liver transplants as part of the effort to move the technology toward clinical testing, with a broader goal of evaluating it for liver cancer detection in high-risk patients.
The stakes attached to early detection are stark. The university's announcement cites figures attributed to the American Cancer Society showing liver cancer's overall five-year survival rate sits at just 22%, but that climbs to 37% when the cancer is caught early — and falls below 5% once it reaches later stages. The same announcement notes that liver cancer incidence rates have tripled over the past four decades.
A Texas Problem With National Stakes
Texas carries an outsized share of this burden. The Texas Department of State Health Services has described the state as having the highest liver and intrahepatic bile duct cancer rates in the nation, and a snippet from its 2025 report put that figure at 14.6 cases per 100,000 population — a comparison the agency's report document leaves incomplete in the portion reviewed for this story. The same state report projects that nearly 3,500 Texans will be diagnosed with HCC in 2025 and lists a five-year relative survival rate of 23%, according to the Texas Department of State Health Services. Nearly 90% of Texas liver-origin cancers, excluding bile duct cancers, are HCC, per the university's announcement, which also notes that Hispanics make up 40% of the Texas population — a demographic detail the university's release ties to the state's cancer burden.
National trends tell a more mixed story. The University of Houston's announcement states that U.S. HCC incidence peaked around 2015 even as Texas rates continued climbing or holding steady. Separately, the National Cancer Institute's SEER program reports that age-adjusted rates for new liver and intrahepatic bile duct cancer cases nationally fell by an average of 0.7% per year from 2014 to 2023, while age-adjusted death rates stayed stable from 2015 to 2024.
Building on Years of Exosome Research
Shih is no stranger to this line of work. He is involved in UH Health's Institute for Digital Healthcare Transformation, which focuses on developing and commercializing technologies that address healthcare challenges, and the university's announcement says his total research funding for this technology now totals $8 million across three NIH grants plus a Cancer Prevention and Research Institute of Texas grant. Earlier in 2026, Shih also received a separate $900,000 CPRIT grant aimed at lung cancer detection, with that project designed to count and profile exosomes from just 20 microliters of blood plasma without isolation or purification, according to the UH Cullen College of Engineering.
A 2021 University of Houston report described an earlier version of Shih's work, a platform called AGNIS, as a label-free system for molecularly specific exosome biosensing with a reported detection sensitivity between 112 and 600 exosomes per microliter. That same year, the Cullen College of Engineering reported that certain surface antigens — CD9, CD63 and CD81 — showed up more abundantly in cancer-derived exosomes than in those from normal cells, a finding that helped lay groundwork for distinguishing cancerous signals from healthy ones.
Other Shih-led efforts have reported strong early results in related cancer detection work. A 2024 account from InnovationMap described a separate blood-based method combining Shih's PANORAMA platform with fluorescent imaging, reporting 99.5% sensitivity and 97.3% specificity for telling cancer samples apart from healthy ones in that particular study. Shih's previous research in this space was published in ACS Sensors, according to the university.
Where the Science Still Needs to Catch Up
Researchers broadly see promise in blood-based liquid biopsy approaches for liver cancer. A review in JCO Oncology Advances notes that such methods, which can examine analytes including exosomes, have the potential to significantly improve early detection, prognostication and treatment monitoring — partly because current clinical genomic circulating tumor DNA profiling approaches can have poor sensitivity in early-stage disease. A separate study indexed on PubMed describes researchers isolating plasma exosomes from 72 cirrhotic patients, including 31 with HCC, for lipidomic analysis aimed at identifying candidate biomarkers.
Still, the new $2.5 million grant funds research that has not yet been clinically validated for liver cancer detection, and the university's announcement does not specify when patient enrollment in the planned post-transplant study will begin. Jonathan McCullers, UH's vice president for health affairs and dean of the Tilman J. Fertitta Family College of Medicine, said Shih's research exemplifies advances occurring beyond the university's traditional health professions colleges — a point that underscores how engineering labs, not just hospital wards, are increasingly where early cancer detection tools get built.
