The NHS-Galleri trial, a significant research effort aimed at evaluating a multi-cancer early detection (MCED) test, has reported its prespecified secondary test-performance endpoints. The study has provided detailed data on how the test performed across three annual screening rounds.

The trial involved 142,250 participants aged between 50 and 77 years, who were randomised into two groups: one receiving the MCED test alongside usual care and the other receiving usual care alone. Participants in the intervention group who received positive test results were referred to NHS standard-of-care pathways for further diagnostic evaluation, with referrals informed by the predicted cancer signal origin (CSO). Participants in both groups continued to be encouraged to attend routine NHS cancer screening for which they were eligible.

Across the three rounds of screening, the trial reported 722, 518 and 561 positive results respectively. These represented 1.03%, 0.80% and 0.90% of participants screened in rounds one, two and three. In aggregate, 937 participants had MCED-detected primary cancers. The cancer detection rates were 0.60%, 0.40% and 0.41% across the three rounds respectively.

The findings also highlighted the test’s predictive performance. The positive predictive values for each round were 58.0%, 50.4% and 45.8%, indicating the proportion of positive results associated with a primary cancer diagnosis within the prespecified follow-up period. Negative predictive values were 98.98%, 98.90% and 98.86% respectively. These figures indicate that most participants who received a negative result did not receive a cancer diagnosis within the defined follow-up period, although predictive values are influenced by the prevalence of cancer in the screened population.

Specificity, which measures the proportion of participants without the relevant cancer who received a negative test result, remained consistently high across the three screening rounds, ranging from 99.50% to 99.60%. However, episode sensitivity varied more substantially. For all cancer types, it ranged from 26.7% to 37.2% across the three rounds. For the 12 prespecified cancer types, episode sensitivity was higher, ranging from 47.6% to 63.4%. These figures reflect the test’s ability to detect cancers that were present during each screening episode and should not be interpreted as evidence that the test detected all cancers occurring in the screened population.

The study also assessed how accurately the test predicted the cancer signal origin. Correct first-or-second CSO prediction accuracy ranged from 91.1% to 93.6% across the three screening rounds. This measure reflects how often the predicted origin of a cancer signal corresponded with the eventual cancer diagnosis among relevant positive cases.

The results provide additional information on the performance of the MCED test within an NHS population-screening setting. The researchers noted that several performance measures, including cancer detection rate, positive predictive value and episode sensitivity, decreased across the second and third screening rounds. The paper discusses the possibility that this may partly reflect changes in the underlying cancer prevalence among participants undergoing repeated screening.

While the trial’s primary endpoint of reducing stage III/IV cancer incidence was not met, these secondary analyses provide detailed evidence on the test’s performance across repeated screening rounds. The findings may contribute to ongoing research into the potential role of MCED tests in population screening, while further trial outcomes, including other clinical and healthcare-related endpoints, remain important for assessing their broader impact.

‍