PLCOm2012 models have higher sensitivity than USPSTF criteria, with no difference by Indigenous classification
Two versions of the PLCOm2012 risk‐prediction model (based on the 2012 Prostate, Lung, Colorectal, and Ovarian [PLCO] Cancer Screening Trial), with and without a predictor for race, have high sensitivities for lung cancer, according to a study published online Oct. 9 in Cancer.
Martin Carl Tammemägi Ph.D., from Brock University in St. Catharines, Canada, and colleagues compared the sensitivity of screening eligibility criteria for self‐reported Indigenous race and evaluated the need for screening at younger ages in a study of lung cancer in South Dakota. U.S. Preventive Services Task Force (USPSTF) 2013 and 2021 (USPSTF2013 and USPSTF2021) criteria and PLCOm2012 with and without a predictor for race were applied at USPSTF-equivalent thresholds of ≥1.7 and ≥1.0 percent in six years to 1,565 individuals who were sequentially diagnosed with lung cancer (12.7 percent self-reported as Indigenous).
The researchers observed no significant difference in the eligibility sensitivities of USPSTF criteria between individuals who self-reported their race as Indigenous and those who did not. Compared with USPSTF criteria, the sensitivities of both PLCOm2012 models were significantly higher, with sensitivity of 66.1, 90.7, and 89.6 percent, respectively, for USPSTF2021 criteria and comparable PLCOm2012 models with and without race. Overall, 1.4 percent of individuals were younger than 50 years, with no difference noted in the proportion based on Indigenous classification.
“Determining screening eligibility using risk prediction models that consider more individualized lung cancer risk factors has been shown in several studies, including this one, to do a better job in selecting people for screening as compared with USPSTF age and smoking history criteria,” Tammemägi said in a statement.
Several authors disclosed ties to the pharmaceutical industry; one author developed the PLCOm2012 and PLCOm2012noRace lung cancer risk‐prediction models discussed in this article.