Modeling Of Risks And Benefits Of Lung Cancer Screening Strategies Using Low-Dose Helical Ct (Ldct) Technology In Canada.

Figure 1: Effec2ve radia2on doses and ranges for standard chest CT, low-­‐dose CT (LDCT), and chest x-­‐ray (CXR) Objec2ves Lung cancer is the most common cause of death from cancer worldwide. Yet, screening for lung cancer remains controversial. We compared the risks and benefits of lung cancer screening with low-­‐dose helical CT (LDCT) rela2ve to chest x-­‐ray (CXR). Concerns have been raised about recommending LDCT as a rou2ne screening tool because of the poten2al harms, including cumula2ve radia2on risks. Methods We developed a decision analy2c Markov model to compare LDCT and CXR under alterna2ve screening scenarios, and es2mate the number of radia2on-­‐induced cancers due to LDCT and CXR exposure. The age specific model was calibrated for Canada, using Na2onal Lung Screening Trial (NLST) and Prostate, Lung, Colorectal and Ovarian trial (PLCO) data. Parameters include the sensi2vity and specificity of LDCT and CXR for lung cancer detec2on, as well as the effec2ve radia2on dose, screening interval and frequency, and pa2ent characteris2cs such as age at screening, smoking status, history of COPD, and family history. Results We compared radia2on risks to es2mated cancers prevented, using age and gender-­‐specific projec2ons. The average effec2ve radia2on dose for LDCT is only 22% (1.4 mSv) of the standard chest CT (7 mSv). The average effec2ve dose for CXR ranges from 0.013 mSv -­‐ 0.136 mSv. LDCT showed a 20% reduc2on in lung cancer mortality compared to CXR. This translated into 4 fewer deaths from lung cancer per 1,000 high-­‐ risk individuals screened with LDCT. However, more than 90% of posi2ve screened results were false posi2ve findings. Over 97% of new lung cancer cases are in high-­‐risk adults aged 50 years and older. Conclusions Concerns have been raised about recommending LDCT as a rou2ne screening tool for lung cancer. Our model addresses a number of public policy ques2ons regarding who should be targeted for screening and what the trade-­‐offs are in terms of poten2al harms and benefits. According to the Canadian Cancer Sta0s0cs, almost half of all Canadians will develop cancer over their life0me, and a quarter of all Canadians are expected to die of cancer. Cancer is the leading cause of death followed by cardiovascular disease. Lung cancer alone accounts for over 13% of all newly diagnosed cancers, and is the leading cause of cancer death, accoun0ng for over 26%. CigareEe smoking is the main risk factor for developing lung cancer, and is associated with over 85% of the cases of this disease in Canada. Almost all (97%) of the es0mated new cases of lung cancer in 2015 are expected to be iden0fied in adults aged 50 years and older. Preven0on efforts, such as screening programs could reduce incidence rates, and result in earlier diagnosis and treatments, thereby improving survival rates. There has been much debate over safe and effec0ve screening tests to iden0fy lung cancer during its preclinical phase, when it is presumed to be more amenable to cura0ve treatment. Early clinical trials showed that annual screening with chest x-­‐ray (CXR) did not reduce lung-­‐cancer mortality. Results showed the viability of emerging techniques that used low doses of radia0on, with effec0ve radia0on doses approximately 25% that of a standard diagnos0c chest computed tomography (CT), but on average s0ll 25 0mes higher than chest radiography. Newly published mortality results from The Na0onal Lung Screening Trial (NLST) showed that screening with low-­‐dose helical CT (LDCT) as compared to conven0onal chest radiography reduced lung-­‐cancer mortality. The US Preven0ve Services Task Force’s (USPSTF) endorses annual screening using LDCT for older adults who are current or former smokers. The Canadian Task Force on Preven0ve Health Care (CTFPHC) is currently reviewing the guidelines. Despite numerous studies, screening for lung cancer remains controversial. Limi0ng screening to high risk persons represents the most effec0ve approach to screening. Extending eligibility criteria to include those at lesser risk might result in overdiagnosis and eventually inducement of harm. The implicit trade-­‐offs between poten0al benefits and harms are becoming important issues to address, especially in the assessment of new technologies such as LDCT. LDCT is a promising new screening technology. Concerns have been raised about recommending LDCT as a rou0ne screening tool because of the poten0al harms, including cumula0ve radia0on risks. We compared the risks and benefits of lung cancer screening with low-­‐dose helical CT (LDCT) rela0ve to chest x-­‐ray (CXR). We developed a decision analy0c Markov model to evaluate a poten0al lung cancer screening program with LDCT rela0ve to CXR among high and low risk pa0ents in Canada. The model was set to follow the design of the Na0onal Lung Screening Trial (NLST) and the Prostate, Lung, Colorectal and Ovarian trial (PLCO) trial. Three screening groups were defined for ages 55-­‐74, current smokers with a 30 pack year smoking history, former smokers with a 30 pack year smoking history that have quit within the preceding 15 years, and never smokers. All pa0ents are asymptoma0c, have no prior lung cancer diagnosis and no evidence of other cancer within the preceding 5 years. Es0mates of key parameters in our model were taken from the NLST, PLCO, the Canadian Cancer Registry, Sta0s0cs Canada, the Na0onal Cancer Ins0tute's Cancer Interven0on and Surveillance Modeling Network consor0um, The Na0onal Research Council’s Biological Effects of Ionizing Radia0on studies, and the Surveillance, Epidemiology, and End Results program (SEER). We es0mated the lung cancers outcomes for current, former, and never smokers over a six year follow-­‐up period using a validated predic0ve logis0c regression model PLCOallM2014. The model included sensi0vity and specificity of LDCT and CXR for lung cancer detec0on, as well as the effec0ve radia0on doses (Figure 1), screening interval and frequency, and pa0ent characteris0cs such as age at screening, smoking status, history of COPD, and family history. The results of the model included: es0mated incidence of lung cancer, number of lung cancer cases detected, number of false posi0ve tests, number of lung cancer deaths, and the es0mated number of radia0on induced solid cancer incidences and deaths. The results are presented for a 62 year old current smoker. It is es0mated that 1,104 per 100,000 popula0on cases of lung cancer will be detected with one round of LDCT screening, as compared to 683 per 100,000 popula0on, with CXR screening (Figure 2). For LDCT and CXR, false posi0ve results account for 96% and 93% of all posi0ve results. For current smokers, screening with LDCT over a six year follow-­‐up period will detect 3,900 per 100,000 popula0on cases of lung cancer, with 1,803 per 100,000 leading to death. For current smokers, screening with CXR, the comparable numbers are 3,058 and 2,243, respec0vely (Figure 3, Figure 6 & Figure 7). Thus, for current smokers, screening with LDCT results in 440 per 100,000 fewer deaths than CXR. For former smokers, LDCT will detect an addi0onal 690 lung cancer cases per 100,000 popula0on compared to CXR screening, and will result in 360 per 100,000 fewer deaths (Figure 4, Figure 6 & Figure 7). For never smokers, screening with LDCT will only detect 120 lung cancer cases per 100,000, with 55 per 100,000 leading to death. For CXR screening, only 94 cancers are detected per 100,000 never smokers, with 69 per 100,000 leading to cancer death (Figures 5-­‐7). The rates of radia0on induced cancer are negligible: for LDCT over a six year period we es0mate 4 per 10,000 popula0on cases of radia0on induced cancer, and 3 per 10,000 cases of radia0on induced mortality. For CXR the rates of radia0on induced cancer are close to zero (Figures 3-­‐5). Overall, the es0mated six year risk of lung cancer is 4,160 per 100,000 current smokers, 3,405 per 100,000 former smokers, and 128 per 100,000 never smokers (Figure 6). Lung cancer is the most common cause of death from cancer not only in Canada but in the world. Despite numerous studies screening for lung cancer remains controversial. Using simula0on modeling techniques, the long-­‐term clinical outcomes, including the trade-­‐offs between poten0al benefits and harms of lung cancer screening, can be quan0fied. Concerns have been raised about recommending LDCT as a rou0ne screening tool for lung cancer. Key ques0ons are, who should be targeted for screening using LDCT technology, what is a pa0ent’s risk of death, and the poten0al u0lity of using LDCT rela0ve to other clinical op0ons, given the large number of false posi0ve results? Our model addresses a number of public policy ques0ons regarding who should be targeted for screening and what the trade-­‐offs are in terms of poten0al harms and benefits. The results of the model might be instrumental in discussions between providers and their pa0ents to understand the uncertain0es surrounding lung cancer screening. Background