The rationale for, and design of, a lung cancer screening program
ABSTRACTWe are entering a new era in which lung cancer screening may be considered the standard of care. The National Lung Screening Trial (NLST) has shown that the number of deaths due to lung cancer can be reduced through screening with low-dose computed tomography (CT) in a high-risk population (N Engl J Med 2011; 365:395–409). Key issues—such as how to manage lung nodules, how to improve cost-effectiveness, and how to minimize radiation exposure—need to be addressed when designing a lung cancer screening program. Time and further technical advances will help to optimize the programs that are developed.
KEY POINTS
- The NLST documented a 20% reduction in the rate of death from lung cancer with low-dose CT screening compared with chest radiography screening (number needed to treat = 320). This was in a population at high risk (age 55–74 with a smoking history of at least 30 pack-years, at least some of it within the past 15 years).
- CT screening detects many lung nodules, of which only a few (3.6% in the NLST) prove to be cancer.
- In view of the positive results of the NLST, Cleveland Clinic has begun a lung cancer screening program, using the same entry criteria as those in the NLST.
- Of possibly greater impact than detecting lung cancer will be the opportunity to promote smoking cessation.
THE NATIONAL LUNG SCREENING TRIAL
The NLST aimed to determine if screening with low-dose chest CT could reduce lung cancer mortality rates.2
This controlled trial enrolled 53,454 people, who were randomized to undergo either low-dose chest CT or posteroanterior chest radiography at baseline and then yearly for 2 years.
Participants were men and women age 55 to 74 with at least 30 pack-years of cigarette smoking. If they had quit smoking, they had to have quit within the past 15 years. All were followed until study conclusion (median 6.5 years, maximum 7.4). About 41% were women, and nearly three-quarters were age 55 through 64. More than 48% were current smokers, with the rest being former smokers.
Results. Adherence to screening was 95% in the CT group and 93% in the radiography group, with a 4.3% annual rate of CT outside the study during the screening phase.
Cumulative lung cancer incidence rates were 645 per 100,000 person-years in the CT group and 572 in the radiography group.
In the CT group there were a total of 1,060 lung cancers during the entire study. Of these, 649 (61%) were detected by screening, 44 (4%) were interval cancers, and the rest were diagnosed after the screening period during follow-up.
In the chest radiography group, there were a total of 941 lung cancers during the entire study. Of these, 279 (30%) were detected by screening, 137 (15%) were interval cancers, and the rest were diagnosed after the screening period. Within the CT group, the cancers detected by screening were more likely to be adenocarcinomas and less likely to be small-cell carcinomas than those not detected by screening. Also within the CT group, the cancers detected by screening were more likely to be stage I (63%) than those not detected by screening.
The cumulative number of deaths from lung cancer was 443 in the radiography group, but only 356 in the CT group—20.0% lower (P =.004). The cumulative overall mortality rate was 6.7% lower in the CT group (P = .02).
Comments. The results of the NLST provide the first evidence that lung cancer mortality rates can be reduced by screening. Though many questions remain, the conclusions of this study are that screening a well-defined high-risk group with low-dose CT reduces the rate of death from lung cancer.
REMAINING CHALLENGES
The NLST showed that lung cancer screening with low-dose CT can meet the most important criterion for a successful screening program, ie, a reduction in the disease-specific mortality rate. Many challenges remain in meeting the other criteria for a successful or ideal screening program (low risk, few false-positive results, acceptability to the patient, and affordability). The issues with low-dose CT-based screening that challenge these ideals are outlined in this section.
Lung nodules: Benign or malignant?
A meta-analysis of CT screening studies found that for every 1,000 people screened at baseline, 9 were found to have stage I non-small-cell lung cancer, 235 had false-positive nodules, and 4 underwent thoracotomy for benign lesions.6
The NLST results were similar. In this trial, only nodules that were 4 mm or greater in diameter were reported. Using these criteria, over 27% of all study participants were found to have a lung nodule on CT at baseline and at year 1. The rate fell to nearly 17% at year 2, as nodules present from baseline were not reported. Of all the lung nodules detected, only 3.6% were ultimately proven to represent lung cancer.2
Many issues with small lung nodules need to be considered. The nodules are difficult to find, with highly variable reporting even by expert radiologists.7 They are difficult to measure accurately and thus are difficult to assess for growth.8 Adjunctive imaging and nonsurgical biopsy have a low yield for small nodules.9–11 Follow-up of these lung nodules includes additional imaging and nonsurgical and surgical biopsy procedures, adding expense to the program and risk to the patient. Finally, knowing that they have a lung nodule makes patients feel anxious and thus negatively affects their quality of life.12,13
Radiation exposure: How great is the risk?
There is a great deal of concern about radiation exposure from medical imaging, as many people receive a substantial amount of radiation each year from medical testing.14 A single low-dose scan with chest CT delivers a whole-body effective dose of about 1.5 mSv—less than one-fifth of the radiation dose of a typical diagnostic CT scan.
Many have tried to estimate the consequences of radiation exposure from low-dose CT screening. All estimates are extrapolations from unrelated radiation exposures. The increase in risk of death ranged from 0.01% to a few percent,15 and the increase in cancers was as high as 1.8% over a 25-year screening period.16 In general, the risks are felt to be very low but not negligible.
Cost-effectiveness is unknown
The cost-effectiveness of lung cancer screening is also unknown. Many highly variable estimates have been published.17–20 The studies have differed in the perspective taken, the costs of testing assumed, and the rounds of screening included. The most cost-effective estimates are in populations with the highest risk of cancer, in programs that achieve the greatest reduction in mortality rate, and in programs that lead to high rates of smoking cessation.
Screening in the real world as opposed to a clinical trial may involve different risks, benefits, and costs. Compliance with screening and with nodule management algorithms may be lower outside of a study. One study suggested that those at highest risk of developing lung cancer would be the least likely to enroll in a screening program and the least likely to accept curative-intent surgery for screening-detected cancer.21
We expect that the NLST data will be analyzed for cost-effectiveness. This should provide the most accurate estimates for the group that was studied.
