Is 20% of a loaf enough?

Authors

  • Robert P. Young BMedSc, MBChB, DPhil, FRACP, FRCP,

    1. Respiratory Genetics Group, Schools of Biological Science, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand
    Search for more papers by this author
  • Raewyn J. Hopkins RN, MPH

    1. Respiratory Genetics Group, Schools of Biological Science, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand
    Search for more papers by this author

Kessler raises several important points about the implementation of computed tomography (CT) screening for lung cancer, specifically the unresolved issues pertaining to cost-benefit analysis of screening, high false-positive rates, screening adherence, and who might best be screened.[1] These issues are closely intertwined, and possible solutions are beginning to emerge.

As Kessler states, assuming effectiveness is held constant, then the cost-benefit of screening can only improve if the costs of screening can be reduced. Recently, Tammemagi et al[2] demonstrated that using a multivariate lung cancer risk model (the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial [PLCO] model) to preselect individuals for screening could improve the number of lung cancers detected per individual screened (the lung cancer detection rate [LCDR]). Not only did this approach increase the LCDR about 1.8-fold (from 0.66% to 1.22% per annum), it identified 90% of all lung cancers by screening only 49% of eligible smokers while maintaining a mortality reduction of 20% across their low-risk to high-risk categories (Table 1). This illustrates that, by targeting those at greatest risk, based on several lung cancer risk variables, significant improvements in cost can be achieved without reducing effectiveness.[3] It is noteworthy that these findings strongly refute comments by Bach (see reply to letter by Young et al.[3]) who suggests that targeting smokers at greatest risk may not achieve the desired goal of maximizing the number of lives saved from screening. If our estimates above are correct, then the number needed to screen annually to detect 1 lung cancer reduces from 1 in 152 using National Lung Screening Trial eligibility criteria to 1 in 82 using a multivariate (PLCO) model. In this setting, the false-positive rate also would be expected to reduce in absolute terms.

Table 1. Calculations Based on Data From the National Lung Screening Trial Computed Tomography Arm Using the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial Multivariate Risk Model of Lung Cancer
VariableNLST Results in the CT ArmEstimates From the PLCO Model
  1. Abbreviations: CT, computed tomography; NLST, National Lung Screening Trial; PLCO, Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial.

  2. a

    This total was based on screening 48.7% of the screening participants in NLST CT arm.

  3. b

    This total was based on identifying 90% of the lung cancers detected in the NLST CT arm.

No. screened26,72213,014a
Lung cancers detected1060954b
Lung cancer detection rate (over 6 y), %3.977.33
Lung cancer detection rate (annually), %0.661.22
No. needed to screen per y to detect 1 lung cancer15282

Another issue raised by Kessler is that quitting smoking saves many more lives than those estimated from screening.[1] We concur with this view and highlight that the most immediate gains from smoking cessation in a lung cancer screening population would come from reductions in mortality from cardiovascular disease (halving the risk within 2 years) rather than from lung cancer, in which significant reductions in risk take much longer (halving the risk over 10 years). In this respect, we previously suggested that individualized risk prediction should be patient (and physician) friendly and, through increasing motivation, should have utility in improving uptake of smoking cessation and adherence to CT screening (both clinically proven interventions).[4, 5] We conclude that there are emerging lung cancer risk-assessment tools with the potential to significantly and cost effectively lower deaths from lung cancer in screening participants.

  • Robert P. Young, BMedSc, MBChB, DPhil, FRACP, FRCP

  • Raewyn J. Hopkins, RN, MPH

  • Respiratory Genetics Group Schools of Biological Science Faculty of Medical and Health Sciences University of Auckland Auckland, New Zealand

Ancillary