We differ in some respects with the views expressed by Dr. Mortani Barbosa Jr. regarding the level of overdiagnosis and the consequences of resection mandated by screen-identified lung cancer . Additionally, reflecting concerns about the design limitations of the National Lung Screening Trial (NLST), we think its positive outcome is challengeable.
Overdiagnosis Detrimental
Computation of the case rate differential in the screened vs. the control cohort is the most direct means of computing an estimate of the magnitude of overdiagnosis. In the 1970’s chest radiographic (CR) screening trials, the percent was 22–24. Because many individuals in the control cohort were screened (“contamination”) and therefore equally liable to overdiagnosis, these figures underestimate the true rate . An estimate of overdiagnosis in low-dose computerized tomographic (CT) trials employing CR controls requires the addition of the CT vs. the CR differential (11%) to the CR vs. the null control differential (25%). The concern that because of delayed clinical manifestation of lung cancers in the control cohort, this differential method may overestimate the magnitude of overdiagnosis, was resolved by long-term follow-up of the Mayo Lung Project in which the differential case rate persisted . The magnitude of overdiagnosis will be influenced by trial design. Actionable nodule dimension, for example, is an additional, potent variable. The rate of volume change is a function of the square of the radius (dV/dr = f (r 2 )) . For example, assuming equal tumor volume doubling time, the rate of volume increase of a 2-mm tumor is one-ninth that of a 6-mm tumor. The far slower growth of small nodules increases their duration of exposure to competing, lethal events, thereby increasing overdiagnosis. Purely adventitious events may contribute to overdiagnosis. For example, an individual with an asymptomatic, screen-identified, stage IV lung cancer, lethally injured in a vehicular accident while leaving the imaging center, would be correctly classified as overdiagnosed. The critical issue is not precision of the estimate, but whether it is sufficiently high to be clinically relevant.
Resection Consequential
Lobectomy entails substantial morbidity and a nonnegligible mortality. The Weill-Cornell group reported an operative mortality of 0.4% for stage IA lung cancer . Pooled community data have been less favorable, with a mean operative mortality of 4.2% for lobectomy . Limited information is available about the long-term consequences of resection. Brown et al. reported that the noncancer relative hazard of death in persons with lung cancer was nearly threefold that of the noncancer death hazard in age- and gender-matched persons with nonvital organ cancers . In a review of long-term survivorship in lung cancer, Sugimura and Yang reported postresection disease-free relative survival of 60% vs. matched U.S. peers, and that one-quarter were significantly restricted in physical ability or reported significant depressive symptoms . Reich et al. reported a deficit of 41% in postlobectomy disease-free survival in surgical candidates judged to have a life expectancy more than or equal to matched U.S. population. They ascribed the deficit largely to long-term consequences of lobectomy rather than lung cancer comorbidities . The bulk of the survival deficit took place more than 6 years following lobectomy, an interval not encompassed in the 6.5-year median NLST follow-up. For individuals in whom death from lung cancer has been averted, these unfortunate consequences are acceptable; for approximately one-third of CT screen-identified individuals whose lung cancer has been overdiagnosed, these consequences constitute an unmitigated harm.
NLST Challengeable
The NLST, which employed CR-screened controls, was premised on the mortality equivalence of null and CR screening. The assumed equivalence was based on nearly identical mortality in the CR vs. the null-screened 30,000 subgroup of the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial who met the NLST criteria for age and smoking history. This comparison exhibited unexplained anomalies: (1) Fewer cases were identified in the CR-screened than the unscreened (control) cohort (518 vs. 520). In contrast, comparisons in the 1970’s CR trials showed the reverse: 22–24% more cases were identified in the CR-screened cohorts despite a far higher rate of CRs received in the controls (which would have the effect of reducing the differential); and (2) Although the annual lung cancer fatality rate was similar in the two cohorts, there were 2 years in which they markedly diverged, suggesting reportage irregularities. Furthermore, the three, small, reporting European CT trials employing unscreened controls failed to reproduce the benefit seen in the NLST; two trials reported higher fatality in the CT-screened cohorts than in the controls . The recently reported Dante trial, which provided 5 years of CT screening following a baseline CR screen of both arms, reported no reduction in lung cancer or all-cause mortality in the CT arm and no reduction in the number of advanced lung cancers. The number of lung cancers identified in the CT arm exceeded that in controls by 31%, and subjects in this arm underwent more invasive procedures, more lung cancer operations, and more thoracic operations for benign conditions. Three lung cancer patients in the CT arm vs. one lung cancer patient in the control arm died of surgical complications .
Conclusion
It is difficult to see by what means overdiagnosis can be reduced “to less than 5%.” Population screening will sample a less healthy population than the “healthy volunteers” of the NLST . Their higher morbidity will lead to higher nonlung cancer fatality, resulting in greater overdiagnosis. Reflecting contamination in the CR control cohorts, the magnitude of CR overdiagnosis is undoubtedly an underestimate. Reservations about the judiciousness of population screening is justified by the considerable harms imposed by surgery in overdiagnosed individuals, the election of surgery by some screenees with false-positive screens in whom malignancy cannot be excluded short term, and the challengeable findings of the NLST. Further follow-up of the NLST cohorts would be useful. The NELSON trial, which employed null controls, will be available shortly.
References
1. Mortani Barbosa E.J.: Lung cancer screening overdiagnosis: reports of overdiagnosis in screening for lung cancer are grossly exaggerated. Acad Radiol 2015; 22: pp. 976-982.
2. Reich J.M., Kim J.S.: Quantification and consequences of lung cancer CT overdiagnosis [Commentary]. Lung Cancer 2015; 87: pp. 96-97.
3. Marcus P.M., Bergstralh E.J., Zweig M.H., et. al.: Extended lung cancer incidence follow-up in the Mayo Lung Project and overdiagnosis. J Natl Cancer Inst 2006; 98: pp. 748-756.
4. Reich J.M., Kim J.S.: Lung cancer growth dynamics. Eur J Radiol 2011; 80: pp. e458-e461.
5. Port J.L., Kent M.S., Korst R.J., et. al.: Tumor size predicts survival within stage IA non-small cell lung cancer. Chest 2003; 124: pp. 1828-1833.
6. Romano P.S., Mark D.H.: Patient and hospital characteristics related to in-hospital mortality after lung cancer resection. Chest 1992; 101: pp. 1332-1337.
7. Brown B.W., Brauner C., Minnotte M.C.: Noncancer deaths in white adult cancer patients. J Natl Cancer Inst 1993; 85: pp. 979-987.
8. Sugimura H., Yang P.: Long-term survivorship in lung cancer: a review. Chest 2006; 129: pp. 1088-1097.
9. Reich J.M., Kim J.S., Asaph J.W.: Diminished post-lobectomy disease-free survival: screening implications. Clin Lung Cancer 2015; 16: pp. 391-397. S1525-7304(15)00076-5 [pii]
10. Reich J.M., Kim J.S.: The National Lung Screening Trial premise of null and chest radiographic control equivalence is open to question. AJR Am J Roentgenol 2015; 205: pp. 278-279.
11. Infante M., Cavuto S., Lutman F.R., et. al.: Long-term follow-up results of the DANTE trial, a randomized study of lung cancer screening with spiral computed tomography. Am J Respir Crit Care Med 2015; 191: pp. 1166-1175.
12. Howard D.H., Richards T.B., Bach P.B., et. al.: Comorbidities, smoking status, and life expectancy among individuals eligible for lung cancer screening. Cancer 2015;