Current Oncology Reports

, Volume 12, Issue 4, pp 226–234

Lung Cancer Screening: An Update, Discussion, and Look Ahead

Authors

    • Respiratory Institute, Cleveland Clinic
Article

DOI: 10.1007/s11912-010-0111-6

Cite this article as:
Mazzone, P.J. Curr Oncol Rep (2010) 12: 226. doi:10.1007/s11912-010-0111-6

Abstract

Over the past few years there has been a great deal of debate about the status of lung cancer screening. The debate has focused on at least three areas: the unmet need to prove a mortality reduction from the screening tests being studied, the potential for these screening tests to produce harm, and the possible cost-effectiveness of an image-based screening program. In this manuscript, I review the chest imaging cohort and controlled trials that have been added to the evidence base over the past few years. I then discuss the evidence related to the areas that are currently debated, describe the ongoing trials that will help to clarify these issues, and speculate about the future.

Keywords

Lung cancerScreeningBiasOver-diagnosisLung nodulesCost-effectiveness

Introduction

In 2007, we wrote a review of lung cancer screening in this journal [1]. In that review, we outlined the evidence base for lung cancer screening, then discussed that evidence in the context of the criteria that must be met to have a successful screening program. Over the past 3 years, lung cancer screening has received a great deal of attention, both within the medical and lay literature. The aim of this review is to update the evidence base, provide an evidence-based discussion on the issues that continue to generate debate, and provide thoughts about the potential evolution of this topic.

Evidence Base

Since our last review [1], additional cohort trials of chest CT screening have been published, as well as early results from ongoing randomized, controlled trials.

Cohort Trials (Table 1)

  • The New York Early Lung Cancer Action Project reported prevalence screens on 6295 high-risk individuals and a yearly incidence screen in 5134. A total of 101 prevalence cancers, three interval cancers, and 20 incidence cancers were identified. Eighty-one of these were proven to be pathologic stage I cancers [2].

  • The Canadian I-ELCAP site reported 1000 baseline screens in high-risk individuals. They identified 20 lung cancers, 19 of which were non–small cell carcinomas, and 15 were stage I [3].

  • The Cosmos trial reported baseline screening results from 5201 high-risk subjects and an annual screen from 4815. Ninety-one lung cancers were found, 70% of which were stage I. Six nodules seen at baseline were found to grow at the annual scan. All six proved to be advanced-stage cancer. Surgical biopsy occurred in 15 subjects with benign lesions (14%; 7/62 at baseline and 8/42 at annual follow-up). Positron emission tomography (PET) imaging was more accurate at baseline (88% sensitivity, 94% specificity) compared to follow-up (70% and 87%, respectively). The median doubling time of tumors was 202 days (97–343; 25th–75th percentiles) [4].

  • The Pittsburgh lung screening study reported 3642 baseline and 3423 first annual screens. A total of 1477 (40.6%) had a nodule, of which 821 (55.6%) received an additional imaging test. Eighty lung cancers were diagnosed, including 69 non–small cell carcinomas of which 40 were in stage I (58%). Eighty-two had surgical biopsies for diagnosis, of which 28 (34.1%) were benign. Thirty-six subjects (1% of all) had a major thoracic surgical procedure for benign disease [5].

  • The ATOM 002 trial reported on prevalence screening of 1045 subjects with known asbestos exposure and variable smoking histories. Forty-four percent had a lung nodule, 10 total cancers were identified, and 89% of lung cancers were stage I. Half of all major thoracic surgical procedures were for benign disease [6].

Randomized, Controlled Trials (Table 1)

  • A French randomized pilot study of low-dose CT scan screening versus chest x-ray (CXR) screening randomized 621 subjects to CT versus CXR. Nodules were found in 45% of the CT group. All positive findings were discussed by a multidisciplinary specialty group. Eight lung cancers were identified, of which three were stage I [7].

  • A Danish randomized trial (part of the NELSON trial) enrolled 4104 high-risk subjects (20+ pack-years of smoking). All subjects had lung function tests and completed quality-of-life surveys yearly. The screened group had yearly CT scans for 5 years. On the prevalence screen, 22% of screened subjects had at least one noncalcified nodule. Seventeen lung cancers were diagnosed. Forty invasive procedures were performed on 25 patients [8].

  • The DANTE trial (Italy) randomized 2472 high-risk men (age 60–75 years, 20+ pack-years smoking) to chest CT (yearly for 4 years) versus a yearly examination after a baseline CXR in each group. Fifteen percent were reported to have abnormal chest CT. At baseline, 28 cancers (2.2% prevalence) were diagnosed in the CT group (14 of these by CT only), 16 were stage I (57%), and 19 (68%) were resectable. Eight cancers were identified in the control arm CXR, four were stage I, and six resectable. Of 32 thoracotomies performed, six were for benign lesions. Invasive tests were performed in 52 in the CT arm and 12 in the CXR arm. At 33 months of follow-up, lung cancer was found in 4.7% of those screened and 2.8% of the control group. Fifty-four percent versus 34% were stage I, respectively. The number of advanced cases was equal. Twenty patients in each arm died of lung cancer [9•].

  • The ITALUNG trial randomized 3206 subjects aged 55 to 69 years with at least 20 pack-years of smoking to chest CT screening versus usual care. Of the 1406 subjects who received a CT (excludes 207 who dropped out), 30.3% had a positive finding on the prevalence screen. Twenty-one cancers were diagnosed, 10 of which were stage I. Only two procedures were performed for benign nodules [10].

Issues to Resolve

There is ongoing debate about whether currently studied lung cancer screening programs will be able to satisfy a few of the criteria used to judge the success of a screening program. Most importantly: Will lung cancer screening reduce lung cancer–specific and/or overall mortality in the screened group? How significant are the potential risks related to lung cancer screening? Can a lung cancer screening program be cost effective?
Table 1

Results from lung cancer CT screening studies

Lung cancer CT cohort screening studies (2007–2010)

Study

Screen type

Age, y

Smoking, pack-years

Slice thickness, mm

Number scanned

Patients with NCN, %

Cancers

Stage I NSCLC

Benign procedures

Follow-up compliance, %

       

n

%

n

%

  

NY-ELCAP [2]

Prevalence

60+

10+

1.25–10

6295

41.8

101

1.6

95

91.3

7

 
 

Incidence

   

5134 year 1, 880 later

6.0

20

0.33

17

85

2

82

 

Interval

     

3

  

65%a

24b

 

Canadian I-ELCAP [3]

Prevalence

55+

10+

1.25

1000

76

20

2.0

15

78% of NSCCa

6

 

ATOM 002, Italy [6]

Prevalence

40–75

66%c

5

1045

43.9

9

0.96

8

89

10

 

PLSS [5]d

Prevalence

50–79

12.5+

2.5

3642

40.6

53

1.5

31

60

36

 
 

Incidence

   

3423

 

27

0.8

9

53

 

95

Cosmos, Italy [4]

Prevalence

50+

20+

2.5

5201

43

91

1.7

65

70

7

 
 

Incidence

   

4815

 

7

0.1

0

0

 

93

Lung cancer CT randomized screening studies (2007–2010)

Study

Screen type

Age, y

Smoking, pack-years

Slice thickness, mm

Arm

Number scanned

Patients with NCN, %

Cancers

Stage I NSCLC

Benign procedures

        

n

%

n

%

 

Depiscan study, France [7]

Prevalence

50–75

15+

1–1.5

CT

336

45.2

8

2.4

3

37.5

3

     

CXR

285

7.4

1

0.4

1

100

 

ITALUNG, Italy [10]

Prevalence

55–69

20+

3

CT

1406

30.3

21

1.5

10

48

2

     

Usual care

1593

      

Denmark [8]

Prevalence

50–70

20+

 

CT

2052

21.8

17

0.8

9

53

10

     

Usual care

2052

      

DANTE, Italy [9•]

Prevalence

60–75

20+

5

CT

1276

15

28

2.2

16

57

24

     

Baseline CXR

1196

 

8

0.7

4

50

4

 

Incidencee

   

CT

  

60

4.7

33

55

36

     

Usual care

  

34

2.8

12

35

5

CXR chest x-ray, NCN noncalcified nodules, NSCCa non–small cell carcinoma, NSCLC non–small cell lung cancer

aPathology staging of all screen and interval cancers

bOut of protocol

cAll were asbestos exposed

dPittsburgh lung screening study

e33 months of follow-up

Reduction in Mortality

For mortality to be reduced through screening, the disease in question must be identified at a point early enough in its’ course for available therapies to cure or control the disease. These therapies must be more effective at the point the disease is identified through screening than they would have been if the disease was identified later in its course. For lung cancer screening, earlier in its course means at an earlier stage. Available treatments include surgical resection, radiotherapy and other local ablative therapies, as well as chemotherapies and targeted therapies. Assuming the pathobiology of lung cancer is such that it progresses from stage I to later stages, the detection and effective treatment of more early-stage lung cancers should lead to a reduction in the number of individuals presenting with lung cancer at a later stage (a stage shift). This stage shift will lead to both an improved survival and a reduction in lung cancer–specific mortality.

To date, trials using CXR (cohort and controlled) and chest CT (cohort and early controlled) as the screening tool have clearly been able to identify more early-stage lung cancers in the screened groups. They have yet to show a reduction in the number of advanced-stage lung cancers, or in lung cancer–specific mortality. A few examples:
  • The Mayo Lung Project, a controlled trial of frequent versus infrequent CXR screening, identified 240 early-stage lung cancers in the screened group compared to 212 in the control group. The numbers of advanced-stage diagnoses were 303 versus 304, respectively. The number of lung cancer deaths was 337 in the screened group and 303 in the control group [11].

  • The International Early Lung Cancer Action Program, a chest CT cohort trial, identified 484 lung cancers, 85% of which were clinical stage I at diagnosis. They reported an 80% 10-year survival [12].

  • A report compared the results from three of the CT screening cohort studies with a validated prediction model. They found that the screening studies found three times the number of lung cancers and performed 10 times the number of procedures versus what would have been predicted. The number of advanced-stage cancers diagnosed, and mortality from lung cancer, was not different than predicted [13].

  • The Lung Screening Study (a randomized feasibility study of CT vs CXR for lung cancer screening) did not show a difference in stage distribution between CT and CXR screen detected lung cancers during the first incidence scan [14].

  • As above, in the DANTE trial, a chest CT controlled trial, more lung cancers were identified in the screening arm, and the proportion of lung cancers that were stage I was higher. At 33 months of follow-up there was an equal number of advanced-stage lung cancers diagnosed and an equal number of lung cancer–related deaths [9•].

These results have led to a great deal of discussion about the potential biases that can influence the results of screening trials. A lead-time bias (identify the disease earlier so survival appears longer even if the timing of death is not altered) is inevitable with any screening test. A length-time bias (preferentially identify less aggressive disease in the screened group so survival of the group identified with disease is improved on average) is also likely to occur with any screening test. This will be most notable in the prevalence round of screening. An overdiagnosis bias (identify disease that would have never affected someone’s life) will be present in any screening program, particularly if that program targets subjects at risk for death from other causes. The nature of the disease being screened for will influence the degree of overdiagnosis that occurs. The key debate isn’t whether these biases are present in uncontrolled lung cancer screening trials. The key questions are, are the biases significant enough to affect our interpretation of dramatic results from uncontrolled trials, and are the biases influential enough to affect the success of the screening program as a whole? The evidence above suggests the answer to the first question is yes. The answer to the second question awaits further data. In the meantime, it serves as fodder for debate.

The first part of the debate to understand is a theoretical basis for the idea that these biases could impact the ability of a screening program to reduce lung cancer–specific mortality. It has been estimated that the natural history of lung cancer could include up to 40 tumor volume doublings [15]. A wide range of growth rates has been reported for malignant nodules [1618]. In one report, the median doubling time was 181 days, but 22% of the nodules had a doubling time greater than 465 days. Only 6% would have been in the stage IA category for less than 1 year. A second report suggested differences in growth rates based on the appearance of the nodule. Solid malignant nodules had a doubling time of 149 days, partial grand glass opacities of 457 days, and pure ground glass opacities of 813 days [17]. Another study suggested that only half of all stage I lung cancers would show 25% increase in volume by 8 weeks of follow-up and 5% wouldn’t reach this threshold within a year [18]. One could argue that the malignant nodules that are differentially identified by imaging tests of high sensitivity, applied intermittently, would be smaller, slower growing nodules. With this in mind, Fig. 1 illustrates the time it would take for a small lung cancer to grow to the point where it would be predicted to result in death. The figure is meant to be illustrative only—many assumptions (linear growth, size as predictor of death) are involved in its production. Recognizing that the population being screened includes individuals in an older age group who have a history of cigarette use, it is easy to see that other medical problems could result in death before the detected lung cancer would.
https://static-content.springer.com/image/art%3A10.1007%2Fs11912-010-0111-6/MediaObjects/11912_2010_111_Fig1_HTML.gif
Fig. 1

For a given tumor size at the time of detection we can estimate the time to theoretical death based on lung cancer growth rates that have been reported in the literature. For example, if a cancer is detected at 1 cm and has a doubling time (DT) of 465 days, in theory it would take 12.7 years for that cancer to lead to death. Size = size at the time of diagnosis; death is predicted based on 40 doubling times

In addition to the theoretical considerations, evidence-based arguments for and against the influence of these biases on the potential for lung cancer screening with chest imaging to be successful have been made:
  • Differences in the histology and epidemiology of screen-detected cancers have been shown. In a study from Japan, 23 cancers were identified, 16 of which were bronchoalveolar cell carcinoma (BAC) or well-differentiated adenocarcinomas. The portion of never smokers and females with cancer was higher than that in smokers and males [19]. In the Mayo CT screening trial, 17% of screen-identified lung cancers were BAC, much higher than the usual [20]. In the IELCAP, 70% were adenocarcinomas [12]. In a retrospective review of resected cancers from Japan, those that were CT screen detected had a higher incidence of cancers <2 cm than symptom-detected cancers (76.4% vs 19.6%) and adenocarcinomas (92.6% vs 58.6%) [21•].

  • Case series have described poor outcomes in those with early-stage lung cancer who do not receive treatment. As an example, 1432 patients with stage I non–small cell carcinoma who did not receive treatment had a median survival of 9 months, 13 months for patients with T1 disease, and 14 months for those who refused surgical resection [22].

  • The validity of the poor outcomes argument has been questioned. The population of cancers present in these series was different than has been identified in screening trials. In the present example, 32% of the cancers were squamous cell, 29% adenocarcinoma, 27% undifferentiated, and 60% were T2 [22]. Also, the patients in these series refused or were not eligible for curative intent therapy, and thus would be unlikely to be enrolled in a screening program. A substantial portion of the deaths reported in these series were unrelated to the lung cancer.

  • One study reported the doubling time of stage I tumors found by CXR screening and found it to be similar to values determined in non-screen detected lung cancer [23].

  • Some have argued against these growth rate estimates. Even if these estimates are accurate, one report suggests that 33% of healthy smokers of screening age would be expected to die of all causes during the progression of the cancer [24].

  • Another report showed that gene profiles of cancers detected during CT screening were found to be similar to profiles of matched cancers that were not screen detected [25].

  • However, it was recognized that nine genes involved in tumor growth were different between screen-detected and non-screen detected cancers [25]. Others have questioned the choice of controls for the gene trial, noting that matching for stage may have inadvertently led to the finding of similar profiles [24].

  • Moderate to severe comorbidity, as may be present in lung cancer patients, was seen to have a greater impact on survival of those with early-stage lung cancer than late stage. Thus, the prognosis of screen-detected lung cancers is influenced by other factors with the potential result being an overdiagnosis of some early-stage lung cancers in screening programs [26].

  • An extended follow-up report of the Mayo Lung Project found 585 individuals who received screening were diagnosed with lung cancer over the 28 years since study inception, whereas only 500 who were not screened had been diagnosed with lung cancer. Explanations for this range from random chance to concluding 17% of screen-detected cancers were overdiagnosed [27].

  • Autopsy series have reported incidental lung cancers in individuals felt to have died of “natural causes” (0.34%–1% of autopsies). The median diameter of these was 3 cm in one study [28]. Another identified a secular trend whereby fewer advanced-stage cancers were being incidentally identified at autopsy over time (1950s–1970s). This led them to speculate that “advances in technology would lead to the detection of this reservoir of smaller, less advanced lung cancers, resulting in the appearance of an increased incidence of lung cancer but a better survival for the cancers that are diagnosed” [29].

The theory and evidence base surrounding the potential for lung cancer mortality reduction from chest imaging screening supports the need to wait for results from prospective controlled trials.

Potential Risks from Lung Cancer Screening

The main risks to those undergoing imaging-based screening for lung cancer relate to finding indeterminate lung nodules, the evaluation of the nodules that are found, and the radiation received during the scan.

The detailed images provided by modern CT imaging have uncovered a large number of small lung nodules. When used in lung cancer screening programs, many indeterminate lung nodules are identified. Depending on how a study defines a positive finding, nodules may be seen in up to 75% of all participants receiving serial chest CT scans [30]. Though most of the nodules are benign, many will require additional evaluation. This can range from serial imaging surveillance to resection. Anxiety about test results has been reported [31]. PET imaging has been reported to be variably helpful in evaluating screen-detected nodules. Some have reported a lower rate of biopsy from its use, while others have reported a high false-negative rate [3234]. PET scanning has been reported to be more accurate for nodules found on prevalence screening than those found on incident scans [35]. Baseline and repeat lung nodule volume measurements are being used in the NELSON controlled CT screening trial to label a finding as positive or negative in an effort to minimize additional imaging and invasive evaluation of the many small nodules found during CT screening. Volumes <50 mm3 are considered negative, 500+ mm3 positive, and 50 to 500 mm3 are re-evaluated in 3 months. If there is no growth or the volume doubling time is >400 days during the first round of screening, they are labeled negative. With this protocol, 7361 of 7557 were determined to be negative on the baseline screen, 20 of which went on to have lung cancer after 2 years of follow-up. A total of 196 were determined to be positive, 70 of which had lung cancer, and 10 others went on to have lung cancer after 2 years of follow-up [36•]. On average, for every three cancers resected, one surgical biopsy was done for benign disease in the studies reported (Table 1). In the DANTE trial’s 3-year report, a greater number of imaging and invasive studies were performed in the screened group. Ninety-six major thoracic procedures were performed in the screened group, whereas 60 cancers were diagnosed [9•]. Lung resection can lead to reduction in cardiopulmonary reserve with the potential for increased non-lung cancer morbidity and mortality.

Radiation exposure from testing is also a concern. A large amount of radiation exposure is credited to diagnostic imaging tests already. One study reported radiation exposures from imaging procedures in 952,420 non-elderly adults (18–64 years of age) in five health care markets. They found that moderate effective doses of radiation were incurred in 193.8 enrollees per 1000 patient-years, high and very high doses in 18.6 and 1.9, respectively. The doses increased with age and were higher in women [37•]. Estimates of lifetime attributable risk of cancer from 64-slice cardiac CT angiography range from 1 in 143 for a 20-year-old woman to 1 in 3261 for an 80-year-old man [38]. Nodules found on low-dose CT screening would be followed with more detailed imaging such as this. The risk from the test needs to be considered in any CT screening program.

Cost Effectiveness

Cost effectiveness is a difficult issue to be comfortable with. To date, we cannot be sure that lung cancer screening with chest imaging is going to be clinically effective at reducing lung cancer mortality; thus, any cost-effective analysis study must make a considerable number of assumptions. For example, a well-performed cost-effectiveness analysis that questioned the potential cost effectiveness of CT screening assumed a 50% stage shift in the analysis [39]. As above, the amount of stage shift that will occur is unclear at this time. Bringing screening programs out of a study environment and into common practice will add other variables, such as patient compliance, compliance with screening algorithms, and the enrollment of appropriately at-risk individuals. For example, current smokers have been reported to be less likely to be willing to consider CT screening (71.2%) than never smokers (87.6%), and only half of current smokers would opt for surgery for a screen-diagnosed cancer [40•]. Advances in diagnostic techniques and treatment strategies will continuously affect the cost effectiveness of any lung cancer screening program.

Potential Evolution of Lung Cancer Screening

Future evidence from ongoing randomized, controlled trials will determine the near-term status of lung cancer screening. Our ability to refine the selection of screening subjects and the development of new screening tools may affect the longer-term possibilities.

Controlled Trials

There are two controlled trials of CXR screening ongoing. The PLCO is a randomized, controlled trial of CXR screening versus no lung cancer screening. Nearly 155,000 subjects were randomized to receive a baseline CXR with 3 yearly CXRs (two if a never smoker) or no chest imaging. Approximately one third of participants were over 65 years of age and nearly half were never smokers, with only 29% having a 20 pack-year or greater smoking history. Only the results of the baseline screen have been published. They found 1.9 lung cancers per 1000 prevalence screens. Nearly one half of all cancers diagnosed were in never or former smokers. Fifty-seven percent were adenocarcinomas, and 52% were in stage I [41]. The Cleveland Clinic has a randomized, controlled trial of CXR with computer-aided detection of lung nodules versus sham CXR. A total of 1300 of a planned 8000 subjects have been enrolled to date. The primary outcome that the study is assessing is a reduction in advanced-stage symptomatic lung cancer diagnoses.

The National Lung Screening trial, based in the United States, is comparing annual chest CT screening to CXR screening over a 3-year period. A total of 53,000 high-risk subjects aged 55 to 75 years have been enrolled. The study is powered to detect a 20% reduction in lung cancer mortality. The NELSON trial, based in the Netherlands, Belgium, and Denmark, is comparing chest CT at years 1, 2, and 4 with no screening. Target enrollment is 20,000, and follow-up is scheduled for 10 years after randomization. The study is powered to detect a 20% reduction in lung cancer mortality [36•]. Two smaller controlled trials of CT screening are ongoing in Italy (details in Evidence Base section [9•, 10]). Two pilot feasibility controlled trials have been completed [7, 14, 42].

The results of these trials may suggest an undeniable benefit from screening, clearly no benefit from screening, or quite likely, a smaller and more debatable benefit. If the latter is true, lung cancer screening programs may evolve to include additional means of predicting those at greatest risk of developing lung cancer, or by incorporating one of several novel tests that are under development.

Refining the Selection of Subjects to Screen

The results of ongoing lung cancer screening trials may suggest a small reduction in lung cancer–specific mortality, or that the screening program is not cost effective. If this is the case, we may be able to determine an incidence of lung cancer in a screened population that would allow the screening program to be effective. We could then develop means to enrich the population. For example, a test might be developed that can identify individuals at an increased risk for developing lung cancer. That test might be able to enrich our screened population to a point where the program would be effective (Table 2).
Table 2

The influence of test characteristics on a screening program

Baseline incidence, %

Specificity, %

Sensitivity, %

PPV of test, %

NPV of test, %

0.5

95

95

8.7

99.97

0.5

90

8.3

99.95

0.5

85

7.9

99.92

0.5

90

95

4.6

99.97

0.5

90

4.3

99.94

0.5

85

4.1

99.92

0.5

85

95

3.1

99.97

0.5

90

2.9

99.94

0.5

85

2.8

99.91

0.5

80

80

2.0

99.87

CT screening studies of high-risk individuals have found an incidence of lung cancer of approximately 0.5% (ie, 200 patients have CT scans for every 1 cancer diagnosed on yearly follow-up screens). To raise the incidence in those who would require imaging to close to 10% (ie, 10 patients would have a CT scan for every 1 cancer diagnosed on yearly follow-up screens), a test would need a specificity of greater than 95% (ie, PPV of test approaching 10%). A sensitivity of at least 90% would mean most with lung cancer will not be missed

NPV negative predictive value, PPV positive predictive value

Attempts to predict cohorts at high risk of developing lung cancer have been made. Age, smoking history, family history, obstruction on spirometry, and sputum atypia are some of the factors that have been considered [4346]. New screening and diagnostic tools are currently being investigated.

Novel Testing

Many lines of test development are being pursued by lung cancer researchers. The most useful tests for lung cancer screening would be easy to administer, not carry risks related to their performance, and be accurate enough to influence management in a positive manner. Blood testing and breath testing are two areas that may meet these criteria in time. They are both easy to perform and free of risks related to test administration. Multiple candidates are being explored. Blood tests looking at protein profiles, DNA levels, circulating tumor cells, gene expression profiles of mononuclear cells, DNA methylation, antibodies to tumor-associated antigens, and microRNAs have been reported. Breath tests are being developed, which look for unique patterns of volatile organic compounds in the breath [47], or alterations of DNA within exhaled breath condensate [48]. These tests are at varying points in their development. Some have shown considerable promise. If proven to be accurate enough, they may be studied in clinical settings, such as being included in an algorithm for lung cancer screening.

Conclusions

All involved in the care of patients with lung cancer are hopeful that a successful lung cancer screening program can be developed, so that the burden of this disease can be lessened. Trials that have been completed to date have been unable to prove that the studied tests are capable of meeting this goal. We anxiously await the conclusion of ongoing controlled trials of chest imaging, as well as the development of novel tests that could be part of a lung cancer screening program.

Disclosure

Dr. Mazzone served as a consultant at an advisory board meeting for the company Oncimmune; they are developing a blood test for lung cancer detection.

Copyright information

© Springer Science+Business Media, LLC 2010