Lung cancer screening

Background; Low-dose CT scan screening greatly improves the likelihood of detecting small nodules and, thus, of detecting lung cancer at a potentially more curable stage.

Methods: To evaluate the cost-effectiveness of a single baseline low-dose CT scan for lung cancer screening in high-risk individuals, data from the Early Lung Cancer Action Project (ELCAP) was incorporated into a decision analysis model comparing low-dose CT scan screening of high-risk individuals (ie, those [greater than or equal to] 60 years with at least 10 pack-years of cigarette smoking and no other malignancies) to observation without screening. Cost-effectiveness was expressed as the incremental cost per year of life saved. The analysis adopted the perspectives of the health-care system. The probability of the different outcomes following the decision either to screen or not to screen an individual at risk was based on data from ELCAP and the Surveillance, Epidemiology, and End Results Registry or published data, respectively. The cost of the screening and treatment of patients with lung cancer was established based on data from the New York Presbyterian Hospital's financial system. The base-case analysis was conducted under the assumption of similar aggressiveness of screen-detected and incidentally discovered lung cancers and then was followed by multiple sensitivity analyses to relax these assumptions.

Results: The incremental cost-effectiveness ratio of a single baseline low-dose CT scan was $2,500 per year of life saved. The base-case analysis showed that screening would be expected to increase survival by 0.1 year at an incremental cost of approximately $230. Only when the likelihood of overdiagnosis was > 50% did the cost effectiveness ratio exceed $50,000 per year of life saved. The cost-effectiveness ratios were also relatively insensitive to estimates of the potential lead-time bias.

Conclusions: A baseline low-dose CT scan for lung cancer screening is potentially highly cost-effective and compares favorably to the cost-effectiveness ratios of other screening programs.

Key words: cost-effectiveness; CT: lung cancer; screening

Abbreviations: ELCAP = Early Lung Cancer Action Project; FNA = fine needle aspiration; HRCT = high-resolution CT; SEER = Surveillance and End Results Registry TSI = Transaction System, Inc

**********

The Early Lung Cancer Action Project (ELCAP) was conducted to evaluate the usefulness of annual low-dose CT scan screening for lung cancer. The baseline and annual repeat results showed that > 80% of the screen-detected non-small cell malignancies were stage IA lung cancer and were typically < 10 mm in size on annual repeat screening. (1,2) The results of the ELCAP suggest that low-dose CT scanning can contribute substantially to the detection of early lung cancer, and thus potentially can decrease lung cancer mortality.

The purpose of this study was to evaluate the effectiveness and the economic consequences of low-dose CT scan screening compared to the current paradigm of care for lung cancer by which > 80% of lung cancers are diagnosed in the more advanced stages (ie, stages II to IV). To this end, we incorporated data from the ELCAP into a decision-analytic model to assess the cost-effectiveness of a single baseline low-dose CT scan for the early detection of lung cancer compared with usual care.

For the purpose of this study, we assumed that low-dose CT scan screening for lung cancer improves survival. Although the ELCAP data suggested that an annual CT scan screening may translate into a reduction in lung cancer mortality, at minimum the long-term follow-up of individuals with screen-diagnosed malignancies is necessary to determine the potential benefits of screening on lung cancer survival.

MATERIALS AND METHODS

Decision-Analytic Model

We evaluated the cost-effectiveness of a program consisting of a single baseline low-dose screening CT scan for the diagnosis of non-small cell lung cancer in persons aged > 60 years with at least a 10-pack-year history of smoking, fit to undergo thoracic surgery, and with no prior history of cancer (except nonmelanoma cancer of the skin). The comparison program, used to represent the alternative base case, was usual care, under which lung cancer is detected by symptoms and/or signs or is incidentally found on a chest radiograph, CT scan, or other diagnostic test that is part of a person's ongoing medical care.

A decision tree model was constructed to describe and analyze the cascade of events emanating from the decision to screen or not to screen an individual at risk (Fig 1). Under usual care, patients either develop lung cancer or die of other causes, as reflected by age-specific survival probabilities. The stage distribution of malignancies of patients diagnosed under the usual care strategy was assigned according to the distribution of lung cancers found in the National Cancer Institute Surveillance and End Results Registry (SEER).

[FIGURE 1 OMITTED

Screening CT scans results in either negative or positive findings. If the result is positive, the subjects undergo further workup, typically consisting of serial high-resolution CT (HRCT) scans of the screen-detected nodules, as long as no growth is detected, or fine-needle aspiration (FNA) biopsy of a nodule showing growth. We assumed the same stage distribution of screen-detected malignancies as found in the ELCAP study. For simplicity, we assumed that patients with stage I and II lung cancer either can be cured or will progress and they will die, while all patients with stages IIIA, IIIB, and IV lung cancer will invariably die as a consequence of lung cancer dissemination. Years of life saved were used as the measure of health outcomes. The base-case analysis was performed under the assumption of similar aggressiveness for screen-detected and incidentally discovered lung cancers, and then multiple sensitivity analysis was conducted to test the stability of the cost-effectiveness ratio when these assumptions were relaxed. Analysis was performed using a software package (TreeAge: TreeAge Software Inc; Williamston, MA).

Data Sources

The ELCAP study was used to determine the probabilities of the different events following the screening CT scan. These included the following: (1) the probability of a positive result on baseline low-dose CT scan; (2) the probability of diagnosing a non-small cell lung cancer; and (3) the stage distribution of lung cancer detected by CT scan screening (Table 1).

A screening CT scan was considered to be positive if 1 to 6 noncalcified nodules were identified. The CT scan result was considered to be negative if either no noncalcified nodules were identified, more than six noncalcified nodules were present, or features of diffuse disease (ie, diffuse bronchiectasis, ground glass opacities, or any combination of these features) were present on the CT scan. The workup following a positive test result was based on the ELCAP guidelines, which recommended a standard-dose, diagnostic CT scan of the chest with an HRCT of the nodules. Noncalcified nodules found on the diagnostic CT scan were followed according to the size of the largest nodule, as follows:

1. If the nodule was [less than or equal to] 5 mm in size, a follow up HRCT scan was performed 3 months after baseline screening, and, so long as no growth could be identified, repeat limited CT scans were performed at 6, 12, and 24 months. If no growth was noted for > 2 years, the nodule was classified as benign.

2. If the nodule was 6 to 10 mm in size, the decision about whether to perform all FNA (or other type of biopsy-video-assisted thoracotomy, bronchoscopy, or a combination of these methods) immediately or a follow-up HRCT scan was made on a case-by-case basis.

3. If the nodule was [greater than or equal to] 11 mm in size. an FNA (or other type of biopsy) was performed immediately.

Lobar resection with complete mediastinal lymph node dissection was recommended for all the patients who had a potentially resectable lung cancer diagnosed.

We used data from SEER to estimate the stage distribution of lung cancer diagnosed under usual care. (3) Tumor stage was determined according to the 1997 American Joint Committee on Cancer classification using information in the SEER registry on size, tumor extension, and nodal involvement. Lung cancer stage specific cure rates were estimated using the methodology described by Berkson and Gage (4) and later reaffirmed by Buell. (5) Thus, for each stage we constructed survival curves using SEER data, with the cure rate being defined by the point when cumulative survival no longer changed, that is, when the slope of the survival curve approached zero. The probability of lung cancer progression in each stage following initial treatment was defined as one minus the stage-specific cure rate.

Determination of Costs