Published ahead of print on July 17, 2008, doi:10.1164/rccm.200802-336OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200802-336OC
The Pittsburgh Lung Screening Study (PLuSS)Outcomes within 3 Years of a First Computed Tomography Scan1 Division of Pulmonary, Allergy and Critical Care Medicine, Department of Medicine; 2 Department of Epidemiology, 3 Department of Radiology, 4 Heart, Lung, Esophageal Surgery Institute, Department of Surgery, and 5 Department of Pharmacology, University of Pittsburgh, Pittsburgh, Pennsylvania Correspondence and requests for reprints should be addressed to David O. Wilson, M.D., M.P.H., UPMC-Shadyside Place, 580 S. Aiken Ave., Suite 400, Pittsburgh, PA 15232. E-mail: wilsondo{at}upmc.edu
Rationale: The role of computed tomography (CT) screening for lung cancer is controversial, currently under study, and not yet fully elucidated. Objectives: To report findings from initial and 1-year repeat screening low-radiation-dose CT of the chest and 3-year outcomes for 50- to 79-year-old current and ex-smokers in the Pittsburgh Lung Screening Study (PLuSS). Methods: Notified of findings on screening CT, subjects received diagnostic advice from both study and personal physicians. Tracking subjects for up to three years since initial screening, we obtained medical records to document diagnostic procedures, lung cancer diagnoses, and deaths. Measurements and Main Results: 3,642 and 3,423 subjects had initial and repeat screening. A total of 1,477 (40.6% of 3,624) were told about noncalcified lung nodules on the initial screening and, before repeat screening, 821 (55.6% of 1,477, 22.5% of 3,642) obtained one or more subsequent diagnostic imaging studies (CT, positron emission tomography [PET], or PET-CT). Tracking identified 80 subjects with lung cancer, including 53 subjects with tumor seen at initial screening. In all, 36 subjects (1.0% of the 3,642 screened), referred for abnormalities on either the initial or repeat screening, had a major thoracic surgical procedure (thoracotomy, video-assisted thoracoscopic surgery [VATS], median sternotomy, or mediastinoscopy) leading to a noncancer final diagnosis. Out of 82 subjects with thoracotomy or VATS to exclude malignancy in a lung nodule, 28 (34.1%) received a noncancer final diagnosis. Forty of 69 (58%) subjects with non–small cell lung cancer had stage I disease at diagnosis. Conclusions: Though leading to the discovery of early stage lung cancer, CT screening also led to many diagnostic follow-up procedures, including major thoracic surgical procedures with noncancer outcomes.
For 2008 in the United States, the American Cancer Society predicted 215,020 new lung cancer diagnoses and 161,840 lung cancer deaths (1). From 1996 to 2003, only 16% of new lung cancer cases were localized stage at diagnosis and amenable to surgical cure (1). In the United States, lung cancer death is more common than breast, prostate, and colorectal cancer death combined (1). Improvements in 5-year relative survival have been small for lung cancer (13% and 16% for cases diagnosed in 1975–1977 and 1996–2003, respectively), particularly when compared with the substantial improvements seen for breast, prostate, and colorectal cancer (1). Screening for early detection is currently recommended for breast, prostate, and colorectal, but not lung cancer (2). New technologies, primarily low-radiation-dose imaging with multi-detector helical computed tomography (CT), has revived interest in lung cancer screening. CT clearly detects smaller lung tumors at earlier stages than conventional chest X-ray (3–8). Because of lead-time bias, length-time bias, and over-diagnosis (9), conclusions regarding the effects of screening on lung cancer mortality must await results from randomized controlled trials ongoing in the United States (10) and Europe (11–13). Two recent analyses of single-arm cohort studies (14, 15) point to the uncertainty and associated controversy created by the absence of results from randomized control trials. In one analysis, an international collaboration, 85% of CT-detected lung tumors were clinical stage I (15). Moreover, persons with clinical stage I CT-detected lung cancer had favorable 10-year survival (15). From these observations, the investigators concluded that annual CT screening can detect curable lung cancer in a manner capable of preventing 80% of lung cancer deaths (15). In contrast, comparing outcomes observed in three large studies of CT screening to predictions from two validated risk models, the second analysis concluded that CT screening increases lung cancer diagnosis and lung cancer resection without decreasing advanced lung cancer or lung cancer death (14). In addition to over-diagnosis (screen-detection and treatment of clinically insignificant or indolent lung cancer), CT screening frequently detects benign abnormalities (generally, nonmalignant noncalcified lung nodules) that can lead to anxiety, costs, and unnecessary procedures (16, 17). With a focus on major thoracic surgical procedures with noncancer outcomes, we report findings from initial and 1-year repeat screening computed tomography (CT) of the chest and 3-year outcomes for 50- to 79-year-old current and ex-smokers in the Pittsburgh Lung Screening Study (PLuSS), the largest single-institution CT lung cancer screening study reporting to date.
Subjects Between January 2002 and April 2005, the PLuSS used advertisements and mass mailings to recruit volunteers with the following characteristics: (1) age 50 to 79 years; (2) no personal lung cancer history; (3) nonparticipation in concurrent lung cancer screening studies; (4) no chest computed tomography (CT) within 12 months; (5) current or ex-cigarette smoker of at least one-half pack per day for at least 25 years, and, if quit, quit for no more that 10 years before study enrollment; and (6) body weight less than 400 pounds. Individuals were not excluded because of symptoms. Baseline activities included completing a risk factor questionnaire, collecting peripheral blood samples, forced expiratory spirometry conducted and analyzed in accordance with American Thoracic Society standards (18, 19), low-radiation-dose CT screening, and physician referral for noncalcified lung nodules. Follow-up activities included repeat CT screening after 12 months and active surveillance for lung cancer–related endpoints. Occurring between March 2002 and September 2005 and between March 2003 and November 2006, respectively, subjects received the initial and 1-year repeat CT at study expense. Of 9,386 persons assessed by telephone for study eligibility, we excluded 4,352 (46.4%) for reasons related to age (n = 84), lung cancer history (n = 60), participation in other lung cancer screening studies (n = 1,043), recent chest CT (n = 120), cigarette smoking history (duration or dose; n = 1,439), duration quit (n = 1,600), and body weight (n = 6) (exclusions applied in sequential fashion). Of 5,034 eligible persons, 1,279 (25.4%) did not accept an enrollment visit and 113 (2.2%) enrolled, but did not return for CT screening. The remaining 3,642 persons, the study population, completed the initial CT screening a median 53 days (range, 2–604 d) after the initial telephone contact. Among eligible persons (n = 5,034), CT screening occurred more frequently in men than women (74.6 vs. 70.2%), more frequently in ex–cigarette smokers than current cigarette smokers (76.7 vs. 69.7%), and more frequently in younger than older persons (72.5, 73.9, and 66.5% in 50- to 59-, 60- to 69-, and 70- to 79-yr-old persons, respectively). Otherwise, CT-screened and unscreened eligible persons were similar with respect to age started smoking, duration of cigarette smoking, and duration quit (among ex-smokers).
Screening CT Imaging Protocol
Protocol for Primary Interpretation of the Initial Screening CT
Protocol for Primary Interpretation of the Repeat Screening CT A computer algorithm, programmed into our Access database application, assigned the moderate lung cancer suspicion rating to every noncalcified lung nodule that either first appeared on repeat CT or showed suspicious change from baseline. The radiologist was allowed to use subjective criteria to assign a high lung cancer suspicion rating to nodules in this category. The algorithm assigned the low lung cancer suspicion rating to nodules showing no significant change between initial and repeat CT. The repeat screening received a preliminary summary nodule interpretation (low, moderate, or high lung cancer suspicion) based on the rating of the most suspicious nodule.
Secondary Review Procedures
Follow-up Recommendations
To coincide with the postal delivery of written reports, a nurse practitioner telephoned every subject with any CT finding that generated a follow-up recommendation. The nurse practitioner summarized relevant CT findings and referral recommendations and described follow-up options, options that could include either a consultation visit with the study pulmonologist (D.O.W.) or a no-fee physician office visit with the principal investigator (JLW). Again, based on the final summary nodule interpretation, the investigators generally advised thoracoscopy with excisional biopsy or mediastinoscopy for persons with high suspicion; additional diagnostic imaging studies (thoracic CT, positron emission tomography [PET], or PET-CT) after no greater than 3-month delay for persons with moderate suspicion; and interval periodic thoracic CT for persons with low suspicion, in accordance with published guidelines. During the first year of our study, we advised repeat CT in 6 months for persons with only small (< 0.4 cm) low suspicion nodules and repeat CT in 3 to 6 months for persons with larger (
Tracking/Surveillance of Study Endpoints Focusing on clinical indications and outcomes, one author (D.O.W.) reviewed the assembled medical records of (1) every subject who received a lung cancer diagnosis and (2) every subject with a follow-up recommendation who subsequently experienced an invasive diagnostic procedure. The same author and a certified tumor registrar independently staged every lung cancer case and met in conference to reconcile disagreements. We report only lung cancer diagnoses and invasive diagnostic procedures that occurred within 2 years of the repeat CT or, for persons without repeat CT, within 3 years of the initial CT.
Human Studies
Subject Characteristics The 3,642 subjects with initial CT screening included 1,872 men and 1,770 women, 7.1% nonwhite race or Hispanic ethnicity, mean age 59 years, 42.7% 60+ years of age, median 47 pack-year cigarette smoking history (33–62 pack-year inter-quartile range [IQR]), 14% self-reporting any prior history of chest CT, and median 1.8% (1.0–3.4% IQR) calculated 5-year incident lung cancer risk (22). Table 1 distributes subjects according to age, cigarette smoking status, pack-years, personal history of lung disease, family cancer history, and pulmonary function.
Findings at Initial CT Screening We referred 1,477 subjects (40.6% of 3,642; Table 2) because of the finding of one or more noncalcified lung nodules at the initial CT screening. Among those screened, 856 (23.5%) had one noncalcified nodule and 621 (17.1%) more than one. Forty (1.1%) and 182 (5.0%) were referred because of high and moderate lung cancer suspicion, respectively (Table 2).
Compliance with Repeat CT Screening Within 365 days of the initial CT screening, 36 persons (17 men and 19 women) were diagnosed with primary lung cancer and an additional 14 persons died (11 men and three women). Of the remaining 3,592 subjects eligible, 3,423 (95.3%) complied with repeat screening. Compliance with repeat screening was similar among men (95.8%) and women (94.7%; P = 0.12). Median time elapsing between initial and repeat CT screening was 385 days (371–406 IQR).
Lung Nodule Findings on Repeat Screening CT We referred 1,450 subjects (42.4% of 3,423) because of new or changed nodules, 18 (0.5% of 3,423) and 206 (6.0% of 3,423) with high and moderate lung cancer suspicion, respectively (Table 2). In a subgroup of 1,928 subjects without reason for referral after initial screen, we referred 231 (12.0%) because of new nodules, including 2 (0.1%) and 92 (4.8%) with high and moderate lung cancer suspicion, respectively (see the online supplement). The online supplement contains a detailed account of repeat CT screen results in other subgroups defined according to results on the initial screen.
Diagnostic Imaging
Diagnostic Events The group of 30 subjects with thoracotomy or VATS included 28 subjects with lung resection to exclude malignancy in a lung nodule (Table 3), 1 subject with thoracotomy with emphysematous bleb resection to treat spontaneous pneumothorax, and 1 subject with VATS wedge resection to diagnose interstitial lung disease. The 28 subjects with thoracotomy or VATS lung resection to exclude malignant lung nodule constituted 0.8% (95% confidence interval [exact binomial; CI], 0.5–1.1%) of 3,642 screened subjects and 1.6% (95% CI, 1.1–2.3%) of 1,726 subjects with a referral recommendation for noncalcified nodule on either initial or repeat CT. These 28 subjects received non–lung cancer diagnoses of limited or no clinical significance (e.g., fibrosis, bronchiolitis, pneumonitis, intrapulmonary lymph node, etc.; Table 3). Notably, these 28 subjects typically underwent thoracotomy or VATS against the primary recommendation or without the advice of clinical investigators attached to the CT screening program (Table 3). Out of 80 subjects with primary lung cancer diagnosed within 2 years of the repeat screening or, for subjects without repeat screening, within 3 years of the initial screening (Table 4), we identified 54 subjects who had thoracotomy or VATS for either the diagnosis or treatment of lung cancer (Table 3). Therefore, in the setting of CT screening, we observed 28 and 54 subjects who experienced thoracotomy or VATS with benign and lung cancer outcomes, respectively (a ratio of 1 to 2).
Cancer Outcomes Eighty subjects (2.2% cumulative incidence; 95% CI, 1.7–2.7%) had primary lung cancer diagnosed within 2 years of the repeat screening or, for subjects without repeat screening, within 3 years of the initial screening. Table 4 distributes these subjects according to manner of detection (primary tumor reported or not reported on the initial CT as an abnormality in need of diagnostic follow-up), cell type (small cell or non–small cell), and diagnostic stage. Notably, 40 (58.0%; 95% CI, 45.5–69.8%) of 69 non–small cell lung cancer cases and 31 (59.6%; 95% CI, 45.1–73.0%) of 52 non–small cell lung cancer cases with manifestations visible at initial screening were early stage (stage I). Tracking of subjects referred for abnormalities seen on initial or repeat CT identified nine subjects with cancer diagnoses other than primary lung cancer (two Hodgkin's disease, two lymphoma, and five cancers metastatic to lung).
Low-radiation-dose helical CT hopes to reduce lung cancer mortality through early detection. In a high-risk cohort (n = 3,642; mean age 59 yr, median 47 pack-year cigarette smoking history, and 43% with airflow obstruction; Table 1), we report findings from initial CT followed by repeat CT 1 year later and clinical outcomes observed within 3 years of initial screening. Follow-up showed lung cancer detected at early stage (Table 4), though at the cost of frequent referral for CT-related abnormalities (Table 2).
As shown in Table 2, we delivered a follow-up recommendation to 1,477 subjects (40.6% of 3,642) with noncalcified nodules on an initial CT. The initial CT identified 40 (1.1%) subjects with high suspicion and 182 (5.0%) subjects with moderate suspicion nodules, representing The appearance of new nodules on repeat CT is another property of screening that complicates management. In our study, 339 noncalcified nodules, not visible on the initial CT, appeared in 256 of 3,423 subjects on repeat CT, representing a 7.5% 1-year risk of new nodule. Potential harms from false-positive nodules include financial and emotional costs, as well as morbidity and mortality, associated with diagnostic intervention. We documented 737 diagnostic imaging studies (CT, PET, or PET-CT) during the first year of follow-up in 628 of 1,255 (50.0%) subjects with low suspicion nodule. In the subset with only small (< 0.5 cm) low suspicion nodules, we documented 357 studies in 309 of 768 (40.2%) subjects. Twenty-eight (34%) of 82 subjects with thoracotomy or VATS procedures for known or suspected lung cancer ended in a noncancer diagnosis, a result somewhat better than typical practice, where 50% of thoracic surgical procedures to exclude malignancy in a lung nodule do not find cancer (23, 24). Even after taking account of subjects who obtained thoracotomy or VATS against the primary recommendation of study clinicians (Table 3), the frequency of resection for benign nodules, as observed under our protocol, is higher than generally desired. Because the 3-year follow-up was not yet complete, our analysis may underestimate the true frequency of thoracotomy or VATS ending in noncancer diagnoses. In 3 of 28 instances (Table 3) of lung resection for benign nodule, a study pulmonologist (D.O.W.) recommended the procedure, triggered, in two instances, by falsely positive PET and, in the third instance, by appearance of a new nodule on repeat CT. In eight instances, subjects elected lung resection, despite the study pulmonologist's primary recommendation in favor of continued observation with serial imaging. The remaining 17 subjects did not consult with the study pulmonologist before accepting lung resection. Fourteen of 17 had lung resection after the initial screening. When characterized according to our criteria for lung cancer suspicion and the I-ELCAP diagnostic algorithm (15), we found three subjects with moderate-high lung cancer suspicion who were I-ELCAP qualified for immediate biopsy, seven subjects with moderate-high lung cancer suspicion who were I-ELCAP qualified for 3-month CT, and four subjects with low cancer suspicion who were I-ELCAP qualified for repeat CT screening in 1 year. Therefore, our results show a high frequency of thoracotomy or VATS without cancer and an apparent community bias toward aggressive intervention with biopsy for persons with indeterminate lung nodule. Observing 12% of 522 subjects having a biopsy within 1 year of a positive screening CT, Pinsky and colleagues also describe an apparent community bias toward biopsy (17). These results argue for strict adherence to diagnostic guidelines, in combination with informed decision making, when managing persons with nodules detected on screening CT. Although validated evidence-based guidelines are lacking, consensus guidelines are evolving to provide a framework for managing findings detected on screening CT (4, 8, 15, 21). Forty (50%) of 80 subjects with primary lung cancer diagnosed within 3 years of initial CT were stage I non–small cell lung cancer (Table 4). Though representing a higher percentage of early stage disease when compared with the general population (1), the stage distribution we observed is less favorable than that reported by other investigators (3–8, 14, 15), but similar to the 53% value reported by the Lung Cancer Screening Research Group (25). Potential explanations include: (1) selection factors favoring enrollment of minimally symptomatic as opposed to completely asymptomatic at-risk subjects, and (2) variable practices with respect to diagnostic intervention and/or staging. We present results from the largest single-institution study of CT lung cancer screening in current and ex–cigarette smokers. Our results include detailed information about diagnostic imaging studies and invasive procedures that emanate from CT screening.
Conclusions
Supported by the University of Pittsburgh Lung Cancer SPORE: NCI P50-CA90440, University of Pittsburgh Cancer Institute, and University of Pittsburgh Medical Center. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200802-336OC on July 17, 2008 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form February 25, 2008; accepted in final form July 17, 2008
Related articles in AJRCCM:
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
2.0 cm average diameter) with central target or lamellate calcification. Procedures further divided noncalcified nodules into low (< 0.5 cm average diameter or 0.5–0.9 cm average diameter with nonspiculated border) and moderate or high lung cancer suspicion categories (0.5–0.9 cm average diameter with spiculated border or 
1.0 cm average diameter). Finally, the interpreting radiologist used subjective criteria to classify larger noncalcified nodules (
