Model may ‘serve as a starting point’ to identify patients with MASLD at high risk for HCC
Click Here to Manage Email Alerts
Key takeaways:
- Researchers used EHR data from more than 1.8 million patients with MASLD to develop a prediction model for HCC.
- The model had a sensitivity of 78.4% and specificity of 90.1% at the high-risk threshold.
A prediction model could be a “starting point” to identify patients with metabolic dysfunction-associated steatotic liver disease at high risk for hepatocellular carcinoma who may require intervention or surveillance, researchers wrote.
“Despite known MASLD-related HCC risk factors, few prediction models have been developed in MASLD populations, primarily using genetic risk scores, limiting their applicability in routine clinical settings, or in racially and ethnically homogeneous populations,” Luis A. Rodriguez, PhD, of the division of research at Kaiser Permanente Northern California, and colleagues wrote in JAMA Network Open. “Thus, there is a need to develop risk stratification tools using routinely collected demographic and clinical variables from diverse populations in clinical settings to identify a subgroup of patients with high-risk MASLD with and without cirrhosis in whom HCC surveillance can be prioritized.”
Using data from the electronic health records of 1,811,461 patients with MASLD (median age, 52 years; 54.2% women) from Kaiser Permanente Northern California, researchers conducted a prognostic study to develop a prediction model for HCC incidence. Patients were enrolled between 2009 and 2018 and monitored until development of HCC, death or termination of the study on Sept. 30, 2021.
Researchers divided the study population 70:30 into derivation and internal validation cohorts and used an extreme gradient boosting algorithm to model the risk for developing HCC, which was classified as low, if the cumulative estimated probability of HCC was 0.05% or less; medium, 0.05% to 0.09%; or high, at least 0.1%. Researchers also performed stratified analyses based on cirrhosis, race and ethnicity groups, and age categories.
Over a median follow-up of 9.3 years, 946 patients developed HCC, which corresponded with an incidence rate of 0.065 per 1,000 person-years. The 5-year cumulative incidences were 0.003%, 0.018% and 0.242% in the low-risk, medium-risk and high-risk groups, respectively.
According to stratified analyses, the HCC incidence rate in those with cirrhosis was significantly greater (1.216 events per 1,000 person-years) vs. those without cirrhosis (0.057 events per 1,000 person-years).
The rate also was higher among Asian patients (0.088 events per 1,000 person-years) compared with Black patients (0.03 per 1,000 person-years), Hispanic patients (0.054 per 1,000 person-years) and white patients (0.061 per 1,000 person-years), as well as patients aged 76 years and older (0.204 events per 1,000 person-years). The lowest incidence rate was reported in those aged 40 years or younger (0.005 events per 1,000 person-years).
The model achieved an area under the curve of 0.899 (95% CI, 0.882-0.916) in the validation set and 0.941 (95% CI, 0.933-0.95) in the derivation set. In addition, at medium-risk and high-risk thresholds the model identified 18.6% and 9.9% of the overall sample, respectively, with sensitivities of 87.5% and 78.4% and specificities of 81.4% and 90.1. Researchers reported numbers needed to screen of 406 and 241, respectively.
“This prognostic study presents the first HCC risk algorithm for racially and ethnically diverse patients with MASLD in the U.S. using routinely collected EHR variables that adequately discriminated among patients at low-, medium- and high-risk for developing HCC,” Rodriquez and colleagues wrote. “This model can serve as a starting point to identify patients with MASLD and guide decision-making about risk-stratifying patients at high risk of HCC, whether for prevention efforts or higher-intensity HCC surveillance.”
Collapse
Click Here to Manage Email Alerts