Recai Aktay

MD

CWRU

Graduate student

NLST (Learn more about this study)

NLST-11

Feb 20, 2013

Association Between the Severity of Smoking and the Results of Screening Low-Dose Chest CT in Heavy Smokers: Does Smoking Matter Beyond 30 Pack-years?

Lung cancer is the leading cause of cancer deaths in the United States and in the world. It is often diagnosed in an advanced stage (40% in stage IV and 30% in stage III). Screening low-dose chest CT (LDCT) can detect lung cancer at early stages (50% stage I and 7% stage II) and reduce the lung cancer-specific mortality by as much as 20% in high-risk patients, compared to chest X-Ray. In the National Lung Screening Trial (NLST), over 95% of the nodules detected by LDCT were benign and 73% of those who underwent nonsurgical invasive procedures did not have cancer. The high prevalence of false positive (FP) LDCT leads to follow-up imaging tests and unnecessary invasive procedures that increase the costs and expose patients to risks from procedure complications, radiation and psychological stress. More stringent criteria are needed for patient selection in order to reduce the FP LDCT rate in lung cancer screening.

We propose a research protocol to investigate whether or not the context of smoking behavior has an association with lung cancer diagnosis in NLST participants who were all high-risk patients with a remote or current smoking history of ≥30 pack-years. In this research study, we would use the data in the CT and chest X-Ray arms of the NLST. We would assign the patients in to two groups according to the final diagnosis: those with lung cancer and those without. We would then perform propensity score-matching analysis to create uniquely paired matches who have similar likelihood of developing lung cancer. The propensity score-matching analysis would include all the relevant and available clinical, demographic and socioeconomic covariates except for the smoking behavior.

The severity of smoking (in pack-years) and the time elapsed (in years) since the cessation of smoking (in those with remote history of smoking) would be used as measures of smoking in statistical analysis. We would compare the measures of smoking as continuous parameters in the two propensity score-matched groups of the LDCT cohort and the entire NLST cohort. Subsequently, these measures would be categorized and placed in contingency tables. The categories of these measures would then be tested for association with the diagnosis of lung cancer in the current smokers and remote, as they are relevant.

If a meaningful relationship between the quantity of smoking and the diagnosis of lung cancer following LDCT exam can be demonstrated in high-risk patients who smoked ≥30 pack-years, then some changes become necessary in patient selection criteria for lung cancer screening with LDCT. These changes might consequently, improve the yield of screening LDCT and reduce the cost, patient radiation exposure and psychological stress associated with it.

Using the NLST data which includes high-risk patients with smoking history of ≥30 pack-years, we like accomplish the following goals:

1) To demonstrate if a meaningful association exists between the severity level of smoking and the final diagnosis of lung cancer among heavy smokers (≥30 pack-years)
2) To explore the context of such association after categorizing the data by smoking severity level and its temporal relationship to the diagnosis of lung cancer (current versus remote smoking)
3. To demonstrate the added risk for lung cancer, if any, that each smoking category imposes.
4. To determine a practical selection criterion based on the levels of heavy smoking that is able to discriminate the high-risk patients with low false positive rate of LDCT from those with high false positive rate.
5. To evaluate the relationship between lung cancer survival and pack-year history.