A Comparative Cost-Effectiveness Analysis of Lung Cancer Screening by CT and Chest X-ray
Until NLST, lung cancer screening was widely considered to be ineffective. As NLST showed a reduction in lung cancer mortality [1], lung cancer screening is now a real consideration for general medical practice. However, almost 40% of the patients in the CT arm in NLST had at least one positive screening test compared to 16% in the CXR arm. Additionally, about 24% of the total CTs were considered positive compared to 7% of the chest x-rays. The vast majority of the positive screening tests did not lead to a lung cancer diagnosis and are considered false positives (about 95% of the positive screening tests in both arms). Due largely to the higher cost of CT compared to CXR and the higher false-positive rate, population-based CT screening will be a more expensive intervention than CXR screening. Thus, it is uncertain if CT will be more cost effective compared to CXR screening. Accordingly, we propose to conduct a formal cost-effectiveness analysis comparing the two screening methods.
The Mayo Lung Project was the most influential of several previous lung cancer screening trials [2,3]. Though it has generally been interpreted as a negative trial in which CXR led to substantial overdiagnosis of lung cancer, an independent re-analysis suggested that CXR screening was effective in the trial. A survival benefit was found amongst the screened group, that could not be attributed to overdiagnosis or other conventional screening biases [4]. Accordingly, CXR screening may have a role in population-based screening, particularly in parts of the world where cost and availability limit access to CT screening [5].
PLCO also assessed CXR screening for lung cancer, though in the context of a general cancer screening trial not explicitly targeted to a population at high-risk because of smoking history [6]. However, it is the only large-scale screening trial to randomize to an arm without screening, and a substantial subset of patients would have met NLST criteria, thus making it of interest to study in this context.
We would like to utilize NLST data and PLCO trial data from the NLST-eligible subgroup to perform this cost-effectiveness analysis. The PLCO data will provide for a comparison with an unscreened control arm. A detailed statistical analysis plan will be developed and implemented. As no lung cancer mortality benefit was seen in the PLCO or Mayo Lung Project, we would like to explore lung cancer survival and assess the extent to which overdiagnosis, lead time, and length biases may have confounded survival comparisons. Effectiveness endpoints to be explored include lung cancer mortality, cases found, number of indolent or overdiagnosed cases, case survival, life-years gained, QALY and Q-TWiST.
REFERENCES
1. Aberle, DR, Adams, AM, Berg, CD, et al., Reduced lung-cancer mortality with low-dose computed tomographic screening. New England Journal of Medicine, 2011. 365: p. 395-409.
2. Fontana, R, Sanderson, DR, Woolner, LB, Taylor, WF, Miller, WE, Muhm, JR, Lung cancer screening: the Mayo Program. Journal of Occupational Medicine, 1986. 28: p. 746-750.
3. Fontana, R, Sanderson, DR, Woolner, LB, et al., Screening for lung cancer: A critique of the Mayo Lung Project. Cancer, 1991. 67: p. 1155-1164.
4. Strauss, GM, The Mayo Lung Cohort: a regression analysis focusing on lung cancer incidence and mortality. Journal of Clinical Oncology, 2002. 20: p. 1973-83.
5. Strauss, GM, Dominioni, L, Chest X-ray screening for lung cancer: Overdiagnosis, endpoints, and randomized population trials. Journal of Surgical Oncology, 2013.
6. Oken, MM, Hocking, WG, Kvale, PA, et al., Screening by chest radiograph and lung cancer mortality: The Prostate, Lung, Colorectal, and Ovarian (PLCO) randomized trial. Journal of the American Medical Association, 2011. 306: p. 1865-73.
John Paul Flores (Tufts Medical Center)
Jaime Caro (McGill University)
Alejandro Moreno-Koehler (Tufts Medical Center)
Matthew Finkelman (Tufts Medical Center)