Research Focus

An international, multi-center study led by Stanford GI faculty Dr. Jeffrey Glenn recently published in NEJM shows that a single dose of lambda-interferon reduced hospitalization among COVID-19 outpatients. The study's senior author, Glenn, is a co-principal investigator of the trial, together with Gilmar Reis, MD, PhD, who is the lead author and an associate professor of medicine at Pontificia Universidade Catolica de Minas Gerais in Brazil. Other co- principal investigators include Edward Mills, PhD, a professor of health research methods, evidence and impact at McMaster University in Hamilton, Ontario, and Jordan Feld, MD, an associate professor of medicine at the University of Toronto. Dr. Glenn is Joseph D. Grant Professor II at Stanford, and founder of Eiger BioPharmaceuticals Inc., a biotechnology company that owns rights to lambda-interferon. Dr. Glenn worked with the TOGETHER network, to methodically conduct this international clinical trial of lambda-interferon for COVID-19.

From June 2021 to February 2022, the PEG-lambda trial was conducted at 12 sites in Brazil and five sites in Canada. Patients who tested positive for COVID-19 through a rapid antigen test and exhibited symptoms were eligible to participate. The trial included 1,950 patients, with an average age of 43, and slightly more than half were female. Approximately 95% of the patients were of mixed race, and 85% had been vaccinated against COVID-19.

During the trial, patients received either a single subcutaneous injection of PEG-lambda or a placebo within seven days of experiencing symptoms. Out of the 1,950 patients, 930 received PEG-lambda, while 1,020 received placebo. Within four weeks of receiving the single dose, 25 (2.7%) of the patients who received PEG-lambda were hospitalized or placed under observation for more than six hours in an emergency clinic, while 57 (5.6%) of the patients who received placebo were hospitalized.

The trial results indicated that PEG-lambda was effective in reducing hospitalization rates. Vaccinated patients who received PEG-lambda had a 51% reduction in hospitalization rates compared to those who received the placebo. For unvaccinated patients who received PEG- lambda within three days of symptom onset, there was an 89% reduction in hospitalization rates compared to the placebo. This reduction rate is similar to that observed with Pfizer's Paxlovid. Overall, only 11 (1.9%) of the 567 patients who received PEG-lambda within the first three days after symptom onset were hospitalized within four weeks of getting the shot, compared to 28 (3.1%) of the 590 patients who received a placebo within the same timeframe. This result indicated a relative reduction of 58%. Efficacy was seen across all virus variants including Omicron, and PEG-lambda side effects were similar to placebo.

Putting this study in the context of the ongoing pandemic, Dr. Glenn said- “There’s been a lot of talk to the effect that COVID’s over, I don’t think the virus got that memo. Meanwhile, lots of people are still unvaccinated, and this drug showed profound benefits for vaccinated and unvaccinated people alike.”

Screening substantially decreases colorectal cancer (CRC) incidence and mortality, but the linchpin of the screening process – colonoscopy as either a primary screening test, or as follow-up to an abnormal non-invasive screening test – is operator-dependent. A colonoscopist’s adenoma detection rate (ADR) is inversely associated with the risk of post- colonoscopy CRC. Artificial intelligence (AI), including computer-aided detection (CADe) of polyps, could revolutionize endoscopy by minimizing operator-dependence.

We leverage the infrastructure of our Stanford Colonoscopy Quality Assurance Program to perform a pragmatic implementation trial of the first CADe device approved in the U.S. (GI GeniusTM, Medtronic, Minneapolis MN). This device significantly increased ADR and adenomas per colonoscopy (APC) and decreased the adenoma miss rate (AMR) in randomized trials.

CADe devices were installed in our largest outpatient endoscopy unit (CADe site) for a 3- month evaluation (Implementation) period. Our system’s five other units served as Control sites. Using a difference-in-difference approach, we analyzed whether colonoscopy quality metrics, including ADR and adenomas per colonoscopy (APC), changed in the CADe site compared to Control sites, during the Implementation vs. Pre-Implementation period, matching each endoscopist’s number of colonoscopies.

CADe was used in 1,008/1,037 (97.2%) eligible colonoscopies, reflecting high uptake and enthusiasm to trial the technology.

We observed no statistically significant effect of CADe on ADR (OR 1.14 [0.83-1.56], p=0.41), APC (OR 1.08 [0.80-1.45], p=0.63) (Figure) or any other detection metric. During the Implementation Period in the CADe site, ADR was 40.1% (95% CI, 36.2-44.0%) and mean APC was 0.78 (95% CI, 0.68-0.90) with CADe, vs. 41.8% (95% CI, 37.9-45.8%) (p=0.44) and 0.89 (95% CI, 0.77-1.02) (p=0.23), respectively, during the Pre-Implementation Period without CADe. No effects of CADe on procedure times and non-neoplastic lesion resection rates were seen. CADe use did not substantially mitigate differences in performance for ADR or APC or for any other metric between top, middle and lower tertiles of endoscopist metric-specific baseline performance.

Our results contrast sharply with those of randomized trials, raising the question why. Because we were interested in the impact of a real-world, open-label implementation of CADe, we used a minimalist implementation strategy, with no additional measures to attempt to influence performance. Perhaps there truly was a higher detection rate attributable to CADe in exposed mucosa in our study, but counterbalancing factors emerged, such as dismissal of suspected adenomas that were not highlighted by CADe, errors in diagnosis and decisions about resection, dismissal of true positive CADe prompts, or an unconscious degradation in the quality of mucosal exposure due to a false sense of comfort that CADe would ensure a high quality examination. In the randomized trials, in contrast, the endoscopists knew the hypotheses, they must have known they could influence results on a novel technology, and they could not be blinded.

We remain optimistic about CADe, which clearly identifies polyps, and AI more broadly. It may take a suite of AI features to maximize impact, and to reduce post-colonoscopy CRC and mortality. Our study should prompt investigators to better understand the subtle factors at the interface of human-AI interaction.

Reference:

Computer-aided Detection of Polyps Does Not Improve Colonoscopist Performance in a Pragmatic Implementation Trial. Ladabaum U, Shepard J, Weng Y, Desai M, Singer SJ, Mannalithara A. Gastroenterology. 2022 Dec 15:S0016-5085(22)01388-9. doi: 10.1053/ j.gastro.2022.12.004. Online ahead of print. PMID: 36528131