SESSION TITLE: Lung Cancer Posters
SESSION TYPE: Original Investigation Poster
PRESENTED ON: Wednesday, October 28, 2015 at 01:30 PM - 02:30 PM
PURPOSE: NLST reported that CT screening reduces LC mort compared to CXR. In RPTs on LC screening, LC mort is assumed to provide an unbiased measure of screening efficacy. This assumes randomization produces comparison groups with equal risk of death from the target cancer unless the intervention reduces risk. However, these assumptions are often violated in RPTs, leading to uncertainty about the effectiveness of cancer screening. We aim to assess whether LC mort provides definitive evidence of screening efficacy in MLP and NLST, the most influential LC screening RPTs.
METHODS: In MLP, after a normal prevalence CXR, 9,192 smokers were randomized to an experimental group (EG) undergoing CXR every 4 moX6 yrs and 3 yrs observation, or a control group (CG), observed for 9 yrs. In NLST, 53,454 smokers were randomized to an EG undergoing CT or CG undergoing CXR. Subjects had 3 annual screens.
RESULTS: In MLP, non-small cell LC (NSCLC) mort was 13% higher in EG (p=0.48), while 5-yr surv was superior in EG (41% vs. 17%; p=0.0095). NSCLC incidence (inc) was 37% higher in EG (p=0.0098), implying overdiagnosis. However, the data are inconsistent with overdiagnosis (JCO 20:1973;2002). Rather imbalances in randomization led to higher NSCLC inc in EG, confounding the ability of mort to accurately reflect CXR efficacy. Surv, not mort, provided the most accurate measure of benefit in MLP. In NLST (NEJM: 365:395, 2011), there was a 20% reduction in LC mort in EG (p=0.0022). LC inc was 12% higher in EG (p=0.0067). Because CT is more sensitive than CXR, higher LC inc is predictable based upon lead-time bias, and conceivably overdiagnosis. However, inc of the most virulent LCs, including small-cell LC and NSCLC-not otherwise specified was 16% lower in EG(p=0.046). These differences suggest imbalances in randomization. Surv has not been reported in NLST.
CONCLUSIONS: In MLP, improved surv provided a surrogate for cure, while mort was a biased measure of efficacy. In NLST, lower incidence of lethal LC subtypes in EG predicts for lower LC mort, independent of CT efficacy. Accordingly, LC mort did not represent an unbiased measure of efficacy. While stage differences support that CT is superior to CXR, further analysis, including surv, is needed to judge the true effectiveness of CT screening.
CLINICAL IMPLICATIONS: Population-based screening is mandatory to reduce the global burden of LC mort among 1.3 billion smokers worldwide. Our findings underscore the need to reconsider how to best measure screening efficacy in the RPT setting.
DISCLOSURE: The following authors have nothing to disclose: Gary Strauss, John Paul Flores, Lorenzo Dominioni
No Product/Research Disclosure Information