Lung cancer screening: are race- and risk-aware criteria needed?

Lung cancer screening with low-dose computerized tomography was launched following the 2011 report from the National Lung Screening Trial showing a 20% reduction in lung cancer mortality for those screened with low-dose computed tomography versus chest X-ray.¹ With these findings and an evidence grade of B, the US Preventive Services Task Force (USPSTF) provided guidelines for screening eligibility in 2013, subsequently revised in 2021. With more than a decade of screening experience accrued since 2013, inequities in screening coverage across racial groups have been documented with both the original and the revised guidelines.

Manful and colleagues propose that lung cancer screening guidelines should be “race-aware” and reflect the higher risk, given smoking history, in Black/African Americans (Black/AA) compared with Whites.² Their recommendation acknowledges the epidemiological context for lung cancer risk among Black/AA individuals with a smoking history. Black/AA males have experienced much higher lung cancer rates than White males, although the excess has narrowed recently.³ The difference is not explained by smoking patterns of the 2 groups. In more than several decades of follow-up in the Multiethnic Cohort based in Hawaii and California, Black/AA and Native Hawaiian individuals who smoke, particularly men, have had substantially higher risk for lung cancer than Whites after adjustment for smoking history.^4,5 Proposed explanations for the higher risk include smoking intensity, genetics and metabolic pathways.^5-7

The guidelines for lung cancer screening advanced by the USPSTF have been risk-factor-based, but without considering race. The first set, released in 2013, followed the selection criteria of the National Lung Screening Trial for smoking history, but with an expanded age range; the criteria were ages 55-80, a minimum of 30 pack-years of cigarette smoking, and no more than 15 years following cessation for successful quitters (ie, 55-80-30-15). These criteria, supported by modeling the 1950 US birth cohort, achieved a reasonable balance of benefits to harms across the many modeling scenarios considered.⁸ In 2021, based on new modeling that included the 1950 and 1960 US birth cohorts and results of the NELSON trial, the pack-year criterion was lowered to 20, in part to increase the inclusion of Black/AA participants (ie, 50-80-20-15).^9,10 Two studies were cited as a rationale for the change, one based on the The Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) and the other using the Southern Community Cohort.¹¹ Using the PLCO study, Pinsky and Kramer showed comparable risk for lung cancer among individuals with a 20- to 29-pack-year history and currently smoking compared with eligible (30+ pack-year) individuals who formerly smoked. In the Southern Community Cohort data, the 2013 USPSTF criteria had a lower sensitivity for identifying Black/AA individuals who currently smoked with lung cancer (32%) compared with affected white individuals who currently smoked (56%).¹² The gap was minimized with reduction of the pack-year criterion to 20.

Basing screening eligibility on predicted individual lung cancer risk is an alternative approach to the USPSTF risk-factor-based criteria. In the past, as a practicing pulmonologist, one of us (JMS) occasionally diagnosed lung cancer in patients at high-risk, but not meeting those criteria, for example, people in their 30s who had smoked heavily from childhood and uranium miners at high risk because of radon exposure. Such cases would not have met USPSTF criteria, leaving them ineligible for covered screening. Risk-based screening is possible, and models of lung cancer risk are available to implement it.¹³ The modeling completed for the 2021 Guidelines did consider risk-based strategies, but without incorporating race and ethnicity into the models for technical reasons. The risk-based approaches shifted screening eligibility toward older ages and detected more lung cancers than risk-factor criteria, but with lesser gain of life expectancy and more overdiagnosis.

The article by Manful et al. seeks solutions to the persistent gap in screening eligibility for Black/AA individuals with a smoking history. The investigators assess the performance of the current USPSTF guidelines, other risk-factor-based criteria, and risk-based criteria in participants in the Southern Community Cohort, which includes 52 667 ever-smoking participants (65% Black/AA, 31% White, 4% Multiracial/Other) of whom 1689 developed lung cancer over follow-up of up to 15 years. They apply 8 different sets of screening criteria and 2 additional race-aware models incorporated modifications: for the USPSTF guidelines, the pack-year criterion was lowered to 13 pack-years for Black/AA and for the PLCOm2012 model, risk thresholds varied by race.

Comparing the 10 different sets of screening criteria, including the 2021 USPSTF criteria, they find that the 2 risk-based and race-aware criteria optimize sensitivity and specificity for lung cancer in Black/AA participants. Two other analyses based on the Multi-Ethnic Cohort also support giving consideration to risk-based lung cancer screening.^14,15

Are risk-based approaches feasible? Manful et al. present compelling evidence that warrants consideration for altering eligibility criteria to better reflect risk and that inclusion of race-aware approaches would reduce disparities in eligibility and increase the efficiency of lung cancer screening. While these novel algorithms will initially need consideration by guideline-producing entities, switching eligibility to risk-based calculations and race-aware algorithms is also likely to face a range of implementation challenges that similarly warrant consideration and research efforts to support adoption and integration. In fact, the USPSTF noted this consideration in their most recent review of the lung cancer screening criteria, expressing apprehension that the use of more complex risk prediction models could impose a barrier to uptake of a service already starkly underused, but also made a specific call for additional implementation research in this area.^16,17

One key barrier is the quality of information on the use of tobacco products in medical records, including electronic health records.¹⁸ Algorithms for screening electronic health records do not yet provide a solution.¹⁹ Although risk calculators have been integrated into electronic health records, these algorithms still rely on the needed information being in the record and sufficiently accurate to calculate pack-years.^20,21 If fully implemented some of the models require data beyond pack-years, a potential challenge to their utilization.

We note that while race-aware risk calculators may help make progress toward reducing disparities in eligibility for screening, reducing the screening barriers does not mitigate the broader historical and justified distrust that race-minoritized communities have in the healthcare system.²² The societal biases of stigma and nihilism that intersect with race and other identity-based considerations hinder engagement with lung cancer screening, including both uptake and adherence to the lung cancer screening algorithm.²³ Thus, while recommendations by Manful et al. and by others for lung cancer screening criteria that incorporate race are counter to the move away from the use of race in clinical algorithms, they broaden eligibility, unlike other algorithms that have been criticized as reflecting historical racism and misunderstanding of race as a biological risk factor.^24,25 For some diseases, race-based criteria set different bars for clinical action and compensation. One notable example is the estimated glomerular filtration rate. Before the revision of race-adjustment in 2021, race-corrected prediction equations had sweeping implications for the care of Black patients with renal disease.²⁶

The analyses by Manful et al. call for race-aware screening criteria to correct the bias of the current USPSTF screening criteria. In spite of the changes made in 2021, Manful et al. show that the current criteria do not achieve the same sensitivity in Black/AA Southern Community Cohort members as in Whites. We expand their recommendation for race-aware criteria to calling more generally for risk-aware criteria and similarly extend a call for additional implementation research to guide adoption and integration of these more equitable and efficient strategies to maximize the benefits and reduce the harms of the lung cancer screening cascade.

Author contributions

Jonathan Samet (Conceptualization, Writing—original draft, Writing—review & editing) and Jamie Studts (Writing—original draft, Writing—review & editing)

Funding

J.L.S. acknowledges grant funding from the National Cancer Institute (UG3CA305778, P30CA046934) and the Bristol Myers Squibb Foundation, and he has provided compensated consultation to Genentech and J&J related to efforts to facilitate implementation of lung cancer screening.

Conflicts of interest

J.M.S. has no disclosures. J.M.S., a JNCI Associate Editor and author of this editorial, was not involved in the editorial review of this manuscript or the decision to publish the editorial.

Data availability

No new data were generated or analyzed for this editorial.