- Email: [email protected]
Commentary on “Screening for lung cancer” by Sateia et al
Seminars in Oncology 44 (2017) 83–84
Contents lists available at ScienceDirect
Seminars in Oncology journal homepage: www.elsevier.com/locate/ysonc
Commentary on “Screening for lung cancer” by Sateia et al
In their review article, Satiea et al discuss the burden of lung cancer and the potential benefits and harms of low-dose computed tomographic (LDCT) lung cancer screening [1]. As lung cancer is the most common cause of cancer mortality in the developing world, the promise of a mortality reduction on the order of 15%– 20% among those undergoing screening, as observed in the National Lung Screening Trial (NLST), is alluring. However, for the potential of LDCT screening to be optimized, the harms must be minimized and screening must be effectively targeted to persons both at high risk and also able to benefit from curative treatments. A major concern of LDCT screening, based in part on the results of the NLST, was the high false positive rate (ie, 100 minus specificity) of about 23% overall [2]. As pointed out by Satiea et al, a retrospective analysis applying the new Lung-RADS scoring system substantially decreased the false positive rate, by about 50% on baseline and almost 75% on post-baseline screens [1,3]. Note that the larger decreases on post-baseline screens reflect the availability of prior scans for comparison, such that nodules not showing growth are classified as negative on Lung-RADS. In addition, the threshold nodule size is larger in Lung-RADS than in the NLST. Going forward in population screening, three questions with respect to the use of Lung-RADS will be critical to answer. First, will the actual false positive rates obtained using Lung-RADS be similar to those observed in the retrospective analysis of NLST, or will they be higher? In a setting where clinical decisions with actual patient impact have to be made, there may be a tendency to bend the rules in favor of increased positive calls for borderline nodules. Second, what will be the effect on sensitivity, and will it reflect the retrospective NLST findings, where sensitivity decreased from 93% to 78% on post-baseline screens? Third, and most difficult to assess, if the sensitivity is indeed reduced, how will this affect the mortality benefit of screening? As described in Satiea et al, over-diagnosis is another potential harm of LDCT screening, since almost all diagnosed lung cancers are aggressively treated. In addition to NLST, where an estimated over-diagnosis rate (defined as the percentage of screen-detected cancers that were over-diagnosed) of 18.5% based on a 12% higher lung cancer incidence rate in the LDCT compared to chest radiograph arm [4], other trials have also demonstrated over-diagnosis. In fact, both the Danish LDCT Trial and the Italian DANTE trial, though much smaller, demonstrated substantially higher rates of over-diagnosis, with increased incidence rates of 88% and 44%, respectively, for their screening compared to control arms [5,6]. Though not on the scale of over-diagnosis associated with prostate-specific antigen (PSA) screening for prostate cancer or
http://dx.doi.org/10.1053/j.seminoncol.2017.02.005 0093-7754/Published by Elsevier Inc.
mammography screening, over-diagnosis is still a harm of LDCT screening that needs to be accounted for and monitored going forward. Another issue raised by Satiea et al concerns who should get screened. Several researchers have proposed risk models than can be used to determine who should undergo screening. These models can partition risk more finely than cruder instruments that only utilize cutoffs for minimum pack-years (generally 30) and years since quitting. However, the downside is more complexity, with the possibility of confusion of patients and healthcare providers as to who exactly is eligible for screening. Additionally, since smoking history is entirely based on self-report, it is possible that individuals who desire to be screened could artificially inflate their reported pack-year values. In their vignette of a 62-year-old current smoker with a 50 pack-year history, Satiea et al note that the patient is relatively healthy, with well-controlled diabetes and hypertension [1]. This raises the important point that since those eligible for LDCT screening by definition will have a substantial history of smoking, as a group they are at high risk for comorbidities, especially respiratory and cardiovascular ones. As such, many may not be candidates for surgical resection of their lung tumors, indicating they may not be in a position to benefit from screening but still could suffer harms. Therefore, it will be important to monitor who is receiving screening, not only from a cancer risk perspective but also from a comorbidity perspective. When the Center for Medicare and Medicaid Services (CMS) approved Medicare coverage for LDCT screening, they also mandated that in order to be reimbursed for such screening patients would have to be enrolled in a CMS-approved registry. Currently, the American College of Radiology (ACR) operates the only approved LDCT screening registry. The mandated data elements of an approved registry include patient demographics and smoking history, CT technical parameters, and the findings on the LDCT screen. Hopefully, the registry will be useful in monitoring various aspects of screening, including who is getting screened and the positivity rates of baseline and subsequent LDCT scans. Additionally, linkage of the registry to other standardized data sources, such as the Surveillance, Epidemiology and End Results (SEER) cancer registries (in regions covered by SEER), could substantially increase the utility of the registry. As LDCT lung cancer screening disseminates and moves from the research into the population setting, it will be critical to monitor its implementation and performance. With a favorable, but not overwhelming balance of benefits to harm, care and vigilance is needed to insure a net-positive effect.
84
P. F. Pinsky / Seminars in Oncology 44 (2017) 83–84
Paul F. Pinsky National Cancer Institute Bethesda, MD E-mail address: [email protected]
References [1] Satiea HF, Choi Y, Stewart RW, Peairs KS. Screening for lung cancer. Semin Oncol 2017 (this issue).
[2] National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365:395–409. [3] Pinsky PF, Gierada D, Black W, et al. Performance of Lung-RADS in the National Lung Screening Trial: a retrospective assessment. Ann Intern Med 2015;162:485–91. [4] Patz EF, Pinsky P, Gatsonis C, et al. Overdiagnosis in low-dose computed tomography screening for lung cancer. JAMA Intern Med 2014;174(2):269–74. [5] Wille MW, Dirksen A, Ashraf H, et al. Results of the randomized Danish Lung Cancer Screening Trial with focus of high-risk profiling. Am J Respir Crit Care Med 2016;193:542–51. [6] Infante M, Cavuto S, Lutman FR, 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;191:1166–75.
Contents lists available at ScienceDirect
Seminars in Oncology journal homepage: www.elsevier.com/locate/ysonc
Commentary on “Screening for lung cancer” by Sateia et al
In their review article, Satiea et al discuss the burden of lung cancer and the potential benefits and harms of low-dose computed tomographic (LDCT) lung cancer screening [1]. As lung cancer is the most common cause of cancer mortality in the developing world, the promise of a mortality reduction on the order of 15%– 20% among those undergoing screening, as observed in the National Lung Screening Trial (NLST), is alluring. However, for the potential of LDCT screening to be optimized, the harms must be minimized and screening must be effectively targeted to persons both at high risk and also able to benefit from curative treatments. A major concern of LDCT screening, based in part on the results of the NLST, was the high false positive rate (ie, 100 minus specificity) of about 23% overall [2]. As pointed out by Satiea et al, a retrospective analysis applying the new Lung-RADS scoring system substantially decreased the false positive rate, by about 50% on baseline and almost 75% on post-baseline screens [1,3]. Note that the larger decreases on post-baseline screens reflect the availability of prior scans for comparison, such that nodules not showing growth are classified as negative on Lung-RADS. In addition, the threshold nodule size is larger in Lung-RADS than in the NLST. Going forward in population screening, three questions with respect to the use of Lung-RADS will be critical to answer. First, will the actual false positive rates obtained using Lung-RADS be similar to those observed in the retrospective analysis of NLST, or will they be higher? In a setting where clinical decisions with actual patient impact have to be made, there may be a tendency to bend the rules in favor of increased positive calls for borderline nodules. Second, what will be the effect on sensitivity, and will it reflect the retrospective NLST findings, where sensitivity decreased from 93% to 78% on post-baseline screens? Third, and most difficult to assess, if the sensitivity is indeed reduced, how will this affect the mortality benefit of screening? As described in Satiea et al, over-diagnosis is another potential harm of LDCT screening, since almost all diagnosed lung cancers are aggressively treated. In addition to NLST, where an estimated over-diagnosis rate (defined as the percentage of screen-detected cancers that were over-diagnosed) of 18.5% based on a 12% higher lung cancer incidence rate in the LDCT compared to chest radiograph arm [4], other trials have also demonstrated over-diagnosis. In fact, both the Danish LDCT Trial and the Italian DANTE trial, though much smaller, demonstrated substantially higher rates of over-diagnosis, with increased incidence rates of 88% and 44%, respectively, for their screening compared to control arms [5,6]. Though not on the scale of over-diagnosis associated with prostate-specific antigen (PSA) screening for prostate cancer or
http://dx.doi.org/10.1053/j.seminoncol.2017.02.005 0093-7754/Published by Elsevier Inc.
mammography screening, over-diagnosis is still a harm of LDCT screening that needs to be accounted for and monitored going forward. Another issue raised by Satiea et al concerns who should get screened. Several researchers have proposed risk models than can be used to determine who should undergo screening. These models can partition risk more finely than cruder instruments that only utilize cutoffs for minimum pack-years (generally 30) and years since quitting. However, the downside is more complexity, with the possibility of confusion of patients and healthcare providers as to who exactly is eligible for screening. Additionally, since smoking history is entirely based on self-report, it is possible that individuals who desire to be screened could artificially inflate their reported pack-year values. In their vignette of a 62-year-old current smoker with a 50 pack-year history, Satiea et al note that the patient is relatively healthy, with well-controlled diabetes and hypertension [1]. This raises the important point that since those eligible for LDCT screening by definition will have a substantial history of smoking, as a group they are at high risk for comorbidities, especially respiratory and cardiovascular ones. As such, many may not be candidates for surgical resection of their lung tumors, indicating they may not be in a position to benefit from screening but still could suffer harms. Therefore, it will be important to monitor who is receiving screening, not only from a cancer risk perspective but also from a comorbidity perspective. When the Center for Medicare and Medicaid Services (CMS) approved Medicare coverage for LDCT screening, they also mandated that in order to be reimbursed for such screening patients would have to be enrolled in a CMS-approved registry. Currently, the American College of Radiology (ACR) operates the only approved LDCT screening registry. The mandated data elements of an approved registry include patient demographics and smoking history, CT technical parameters, and the findings on the LDCT screen. Hopefully, the registry will be useful in monitoring various aspects of screening, including who is getting screened and the positivity rates of baseline and subsequent LDCT scans. Additionally, linkage of the registry to other standardized data sources, such as the Surveillance, Epidemiology and End Results (SEER) cancer registries (in regions covered by SEER), could substantially increase the utility of the registry. As LDCT lung cancer screening disseminates and moves from the research into the population setting, it will be critical to monitor its implementation and performance. With a favorable, but not overwhelming balance of benefits to harm, care and vigilance is needed to insure a net-positive effect.
84
P. F. Pinsky / Seminars in Oncology 44 (2017) 83–84
Paul F. Pinsky National Cancer Institute Bethesda, MD E-mail address: [email protected]
References [1] Satiea HF, Choi Y, Stewart RW, Peairs KS. Screening for lung cancer. Semin Oncol 2017 (this issue).
[2] National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365:395–409. [3] Pinsky PF, Gierada D, Black W, et al. Performance of Lung-RADS in the National Lung Screening Trial: a retrospective assessment. Ann Intern Med 2015;162:485–91. [4] Patz EF, Pinsky P, Gatsonis C, et al. Overdiagnosis in low-dose computed tomography screening for lung cancer. JAMA Intern Med 2014;174(2):269–74. [5] Wille MW, Dirksen A, Ashraf H, et al. Results of the randomized Danish Lung Cancer Screening Trial with focus of high-risk profiling. Am J Respir Crit Care Med 2016;193:542–51. [6] Infante M, Cavuto S, Lutman FR, 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;191:1166–75.