## Abstract

The greatest news of the past year in this field was the first large-scale early detection trial that could prove a 20% reduction in lung cancer-related mortality by screening high-risk individuals with low-dose computed tomography (LDCT). Several expert groups and medical societies have assessed the data and concluded that LDCT screening for lung cancer is, however, not ready for large-scale population-based implementation. Too many open questions remain, such as definition of the at-risk population, timing and intervals of screening, optimal method of acquisition and interpretation of the images, how to handle (false) positive findings, and especially cost-effectiveness in relation to other lung cancer prevention strategies, mainly smoking cessation. Further analyses and several ongoing European trials are eagerly awaited. Much hope also resides in the use of biomarkers, as their use in, e.g., blood or exhaled air may provide more easy-to-use tests to better stratify high-risk populations for screening studies. While exciting research is ongoing in this domain—e.g. with microRNAs—none of the tests has yet reached sufficient validation for clinical use. Early central lung cancers are more difficult to visualise by CT. For these patients, standard bronchoscopy, complemented by autofluoresence endoscopy, has been studied in different screening and follow-up settings.

## introduction

Despite advances in treatment, lung cancer remains incurable in many patients because of its advanced stage at the time of diagnosis. Banning smoking would have the largest impact on lung cancer mortality, but even then the problem would persist for several decades, as 70%–80% of all lung cancers in the Western world occur in former smokers. Several strategies aim to detect lung lesions at an earlier stage: better public awareness of the ‘alarm’ symptoms of locoregional stage lung cancer (beyond the topic of this review), screening with imaging methods such as chest X-ray or chest CT, screening with biomarkers in blood, lung lavage fluid or exhaled air, and bronchoscopy in individuals at high risk to develop lung cancer. An overview of the most important developments in each of these fields is given.

## CT scan

After chest radiography had been unsuccessful for lung cancer screening in several randomised, controlled trials (RCTs) reported in the 1980s and 1990s, low-dose multidetector computed tomography (LDCT) of the chest has been investigated intensively in the last two decades [1]. Chest CT is more sensitive than chest radiography for the detection of early lung cancers presenting as small, non-calcified, solitary pulmonary nodules (SPNs). In several large non-randomised studies with LDCT, it was shown that higher numbers of lung cancer could be detected with LDCT compared with chest radiography, and that most of these lesions were detected in an early and thus resectable stage (stages IA to IB) [2]. In some of these non-randomised LDCT studies—like the I-ELCAP trial—very high 5- and even 10-year survival rates of diagnosed stage I lung cancers were observed, but the influence of certain biases on these results could not completely be excluded [3]. The question whether lung cancer screening with LDCT could really result in a reduction of lung cancer-specific mortality could only be addressed in large RCTs, several of which were set up in the last decade. The first RCT of LDCT to reveal its results on lung cancer mortality was the North American National Lung Cancer Screening Trial (NLST). After the release of a preliminary report in November 2010, announcing a 20% reduction in lung cancer-specific mortality in the LDCT arm, it became the first published positive lung cancer screening trial ever [4]. Whether these NLST results will result in large-scale implementation, thereby changing the pattern of care for lung cancer, is at present uncertain, as several important problems related to LDCT screening need to be addressed first [5].

### the National Lung Cancer Screening Trial

Table 1:

Overview of the randomised controlled trials with CT screening.

NLST NELSON DLST ITALUNG DANTE
Country USA NL/Belgium Denmark Italy Italy
Number of sites 33
Number controls 26.732 7.907 2.052 1.593 1.196
Number screened 26.722 7.557 2.052 1.613 1.276
Recruitment period 2002-2004 2003-2006 2004-2006 2004-2006 2001-2006
Age range (year) 55-74 50-75 50-70 55-69 60-74
Smoking history 1 ≥30/<15 >15/<10 ≥20/<10 ≥20/<10 ≥20/<10
Male/Female M/F M/F M/F M/F
Control arm Chest X-ray Usual care Usual care Usual care Usual care 2
Screening rounds
Interval (years) 3 1,2,3 1,2,4,6.5 1,2,3,4,5 1,2,3,4 1,2,3,4,5
Nodule evaluation 2D 2D,3D 2D,3D 2D 2D
Prevalence detection (%) NR 0.9 0.8 1.5 2.19
Incidence detection (%) NR 0.5 0.67 0.4 4.7
False positives (%) 4 96.4 1.7 7.9 NR NR
Mortality reduction 5 20% (2016) (2016) NR NR
NLST NELSON DLST ITALUNG DANTE
Country USA NL/Belgium Denmark Italy Italy
Number of sites 33
Number controls 26.732 7.907 2.052 1.593 1.196
Number screened 26.722 7.557 2.052 1.613 1.276
Recruitment period 2002-2004 2003-2006 2004-2006 2004-2006 2001-2006
Age range (year) 55-74 50-75 50-70 55-69 60-74
Smoking history 1 ≥30/<15 >15/<10 ≥20/<10 ≥20/<10 ≥20/<10
Male/Female M/F M/F M/F M/F
Control arm Chest X-ray Usual care Usual care Usual care Usual care 2
Screening rounds
Interval (years) 3 1,2,3 1,2,4,6.5 1,2,3,4,5 1,2,3,4 1,2,3,4,5
Nodule evaluation 2D 2D,3D 2D,3D 2D 2D
Prevalence detection (%) NR 0.9 0.8 1.5 2.19
Incidence detection (%) NR 0.5 0.67 0.4 4.7
False positives (%) 4 96.4 1.7 7.9 NR NR
Mortality reduction 5 20% (2016) (2016) NR NR

NLST: National Lung Screening Trial; NELSON: Nederlands Leuvens Screening Onderzoek; DLST: Danish Lung Screening Trial

1 Smoking history expressed as total number of pack years and maximal number of quit years allowed as former smoker

2Chest X-ray was performed at baseline in both trial arms

3 Screening Interval expressed as years from randomisation

4 False positive scans in the LDCT group at baseline

5 Lung cancer mortality reduction, either result in % relative to the control arm or the expected year of primary endpoint result

### European randomised low-dose CT screening trials

All other running or planned screening RCTs with LDCT are European initiatives. They differ on several aspects of recruitment and demographics of screened participants, CT protocol used for screening, total number of screening rounds, as well as on nodule management protocols for screen detected nodules (Table 1). All these aspects will become very important in the comparison of these eagerly awaited European trial results with the NLST results. Several of these studies are carried out with the idea of future pooling of their results with other RCT lung cancer screening trials.

Some trials are not listed in the table and will not be further commented upon as they are either small feasibility RCTs [DEPISCAN, a smaller French pilot trial of 1000 subjects [7], or still in their recruitment phase with little available published data—the Multicentric Italian Lung Detection (MILD) trial and the German LUSI trial, both aiming for about 4000 subjects for LDCT versus observation].

The Dutch–Belgian NELSON (NEderlands Leuvens Screening ONderzoek) trial started recruiting participants in 2003 [8]. About 15 500 participants were randomised between the LDCT and the usual care arm (no active screening test). LDCT screening was carried out in four rounds, starting at baseline, and thereafter with intervals of 1, 3 and 6.5 years. Non-calcified pulmonary nodules detected by LDCT were classified as positive screening results depending on their measured volume and volume doubling time by using specific software for semi-automated volume measurement [8]. Based on nodule characteristics, a trial-specific nodule management algorithm with the use of intermediate repeat scans or referral to a pulmonologist for further work-up was strictly adhered to in all four participating screening centres (Figure 1). In 2009, results on the first and second screening rounds of the LDCT study group were published [9]. At the first round, only 196 scans (2.6%) had a positive test result. In total, 70 lung cancers (of 7557 examined subjects) were detected, of which 64% in pathological stage I, giving a lung cancer prevalence of 0.9%. Sensitivity, specificity, positive and negative predictive value of round one CT screening were 94.6%, 98.3%, 35.7% and 99.9%, respectively. At round two, a total of 128 positive scans (1.8%) were found, while a diagnosis of lung cancer was made in 54 of 7289 participants (of which 74% in stage I), giving a lung cancer detection rate of 0.5%. Importantly, the rate of invasive diagnostic procedures was 1.2% in round one and 0.8% in round two. Results of the further screening rounds as well as the mature data on lung cancer-specific mortality in both study groups must be awaited (expected 2016).

Figure 1

A patient in the screening arm of the NELSON trial. Small nodule in the left upper lobe on CT of the first year (A) and the second year (B). On the third year CT (C), the nodule has grown: the volume is 127 mm³ and the volume doubling time is <400 days. Nodule growth category is C. Lesion-resected left superior lobectomy: well-differentiated invasive adenocarcinoma pT1N0 (image courtesy KN).

Figure 1

A patient in the screening arm of the NELSON trial. Small nodule in the left upper lobe on CT of the first year (A) and the second year (B). On the third year CT (C), the nodule has grown: the volume is 127 mm³ and the volume doubling time is <400 days. Nodule growth category is C. Lesion-resected left superior lobectomy: well-differentiated invasive adenocarcinoma pT1N0 (image courtesy KN).

The Danish Lung Screening Trial (DLST) started in 2004 and was planned in close collaboration with the NELSON trial. A total of 4104 participants were included and randomised to either five annual rounds of LDCT screening or control (no screening). The nodule management was the same as in NELSON. LDCT screening results of all five screening rounds have been recently published [10]. The lung cancer detection rate at baseline was 0.83% and the mean annual detection rate was 0.67% in all incidence rounds. More lung cancers (predominantly adenocarcinoma) were detected in the LDCT group and 70% of them were in earlier stages (IA to IIB). In absolute numbers of detected cases, however, there was no difference of late-stage lung cancers detected in both study groups, suggesting that the higher amount of early-stage cancers detected may be partly due to overdiagnosis. At the end of the screening period, no differences in lung cancer-specific mortality were found between both study groups. All these findings, however, may be premature, since the effect of lung cancer cases in the control group may be underestimated at this time, because of the short follow-up. Mature data are expected for 2016.

Starting in 2004, the Italian ITALUNG study recruited 3206 participants referred to three screening centres by general practitioners in the Tuscany region [11]. Four annual LDCT scans were carried out, and the control arm received usual care. Screen-detected nodules were classified and evaluated according to their size. In the work-up of positive nodules, repeat CT scan was carried out but also fluorodeoxyglucose positron emission tomography (FDG-PET) and fine-needle aspiration (FNA). At baseline, the LDCT was estimated positive in 30.3% of 1406 subjects. A total of 21 lung cancers were detected in the LDCT arm, or a prevalence detection rate of 1.5%, with 47.6% of these cancers in stage I, the majority of adenocarcinoma type. The other Italian study DANTE started in 2001 and recruited 1276 male subjects in the LDCT arm with five yearly screening CT scans, while 1196 male control subjects received a yearly clinical examination. Before participation in the trial, all subjects first had a chest radiography and a sputum cytology test. The 3-year results were published in 2009 [12]. In the LDCT arm, the lung cancer detection rate was 2.29% at prevalence screening and 4.70% at 3 years, compared with 0.67% and 2.84% in the control arm. Stage I lung cancer was mostly detected in the LDCT arm (54%); however, the absolute number of advanced lung cancer cases was identical in both screening arms, as in the DLST Trial. In both screening arms, 20 participants had died of lung cancer at the time of this preliminary analysis.

In the UK, a pilot LDCT trial started to recruit 4000 subjects for one round of LDCT versus no screening [13]. Particular to this RCT is the fact that reduction in lung cancer mortality is searched for by only one LDCT screening round. This is based on the fact that study participants are selected for this trial only when they present a higher risk for lung cancer development (5% over 5 years of observation) according to the validated Liverpool Lung Project risk model [13]. Using this patient selection, it is estimated that a lung cancer detection rate of ∼1.5% could be reached. Nodule categorisation and follow-up is carried out according to the NELSON nodule management. When this pilot trial succeeds in its end points (also regarding cost-effectiveness), a further 28 000 subjects will be recruited.

### current problems with CT screening

Following last year's announcement of the innovative NLST CT screening results, different medical societies, like the International Association for the Study of Lung Cancer (IASLC), immediately released statements not only to admit the true relevance of these results for the early detection of lung cancer, but also to openly invite all lung cancer clinicians and researchers to keep on studying CT-based lung cancer screening in current or planned RCTs [14]. Indeed, as of today, many questions remain unanswered on the benefits and potential risks of CT screening, and it will be necessary to answer these questions first before large-scale implementation of LDCT screening. In a recently published report of an IASLC 2011 CT screening workshop, the problems and need for standardisation of multiple aspects of the lung cancer CT screening process were addressed [15]. Important ones are as follows:

• Variability in the choice of the at-risk populations in the different RCT screening protocols (Table 1). As more lung cancers will be detected by inviting higher risk subjects, possible implementation of LDCT screening will have to depend on age, gender, smoking history, socioeconomic class, presence of chronic obstructive pulmonary disease (COPD). The starting UKLS trial uses the Liverpool Lung Project risk model to recruit subjects [13, 16] and will hopefully help us to properly select the most appropriate screening participants.

• Variability in radiological standards for LDCT screening technology, image acquisition and use of computer-aided interpretation (Table 1).

• Variability in the number and time intervals of the screening rounds, important to minimise possible harms by radiation exposure risk for screening participants. Here as well, the result of the UK trial with one single screen may deliver important information (Table 1).

• In particular, the volumetric versus diametric assessment of detected nodules and the precise work-up and follow-up of suspect nodules. The definition when a detected lung nodule is a true-positive finding is crucial to minimise the false-negative (sensitivity) and false-positive results (specificity). In the NELSON trial, and later adapted by the DLST and the UKLS trial, the concept of ‘indeterminate lung’ nodules was used during screening, and for these indeterminate nodules a management algorithm with the use of repeat LDCT scanning was created (Table 2, see [17] for more details). By doing so, the number of false-positive LDCT findings, and the number of invasive investigations needed to correctly classify screen-detected lung nodules, could be drastically reduced [9].

• The role of bronchoscopy and FDG-PET scan in screen-detected lung nodules. From the NELSON experience, it was concluded that a conventional white-light bronchoscopy should not routinely be carried out [18]. The effect of a preoperative FDG-PET scan was different in patients with either a conclusive or a non-conclusive non-surgical work-up (which was physical examination, standard contract-enhanced CT and bronchoscopy). In the former, it was useful to reduce the amount of resections for benign disease by 72%; in the latter it is not recommended, because it had a low negative predictive value of 47% (24 cancers would have been missed) [19]. Surveillance of the total amount of extra-invasive procedures and (recall) CT scans carried out will also be very important for safety reasons in order to minimise radiation exposure risks for screening participants.

• Overdiagnosis bias. Given the results of identical absolute numbers of detected late-stage lung cancers between the screen and no screen arms of several screening RCTs, it remains unclear how many of the extra-LDCT-detected early-lung cancers were indolent, slow-growing cancers [20]. This bias should be as low as possible since it provokes an excess of unnecessary risky interventions for healthy screening participants.

• Psychosocial consequences and acceptability of LDCT screening. In one study, compared with non-smokers, smokers had less belief in a better survival when lung cancer would be detected more early and the latter were less willing to consider participation in an LDCT screening programme [21].

• Role of smoking cessation practice. Very relevant, as implementation of LDCT screening could give a false feeling of safety to smokers, and make them less likely to quit. Therefore, any LDCT screening programme implementation without attention for smoking cessation is not a good idea [22]. Whether CT screening for lung cancer offers a teachable moment for smoking cessation is not entirely clear. In the DLST trial, smoking quit rates after 1 year of screening were similar in the LDCT and control arm [23]. In a random sample of the screen and the control arms of the NELSON trial—evaluated 2 years after randomisation—quit rates were high in both arms (control arm 18.7%, screen arm 13.9%, non-significant difference) [24].

• Finally, as resources for health care become increasingly sparse, LDCT screening—as each new medical technology—will need rigorous assessment of cost-effectiveness. These data are eagerly awaited from the large NLST trial and later from the European RCTs. A recent North American modelling study pointed at incremental cost-effectiveness ratios varying between $110 000/QALY and$280 000/QALY for annual screening [25], far above the level often used for new cancer drug reimbursements, and more expensive than other cancer screening programmes. When LDCT screening would go along with successful smoking cessation in very selected groups of patients, this could be more cost-effective ($73 000/QALY for males,$40 000/QALY for females) than screening alone.

Table 2:

NELSON classification of non-calcified nodules at baseline screening. Adapted from [17].

Nodule type NODCAT I NODCAT II NODCAT III NODCAT IV GROWCAT C
Solid Negative test Negative test Indeterminate test Positive test Positive test
Annual CT  Annual CT 3 mo follow-up CT Refer for work-up Histology needed
Partial solid Negative test Negative test Indeterminate test Positive test Positive test
Annual CT  Annual CT 3 mo follow-up CT Refer for work-up Histology needed
Solid Negative test Negative test Indeterminate test Positive test Positive test
pleural based  Annual CT  Annual CT 3 mo follow-up CT Refer for work-up Histology needed
Non-solid Negative test Negative test Indeterminate test Positive test
Annual CT  Annual CT 3 mo follow-up CT  Histology needed
Nodule type NODCAT I NODCAT II NODCAT III NODCAT IV GROWCAT C
Solid Negative test Negative test Indeterminate test Positive test Positive test
Annual CT  Annual CT 3 mo follow-up CT Refer for work-up Histology needed
Partial solid Negative test Negative test Indeterminate test Positive test Positive test
Annual CT  Annual CT 3 mo follow-up CT Refer for work-up Histology needed
Solid Negative test Negative test Indeterminate test Positive test Positive test
pleural based  Annual CT  Annual CT 3 mo follow-up CT Refer for work-up Histology needed
Non-solid Negative test Negative test Indeterminate test Positive test
Annual CT  Annual CT 3 mo follow-up CT  Histology needed

NODCAT: nodule category; GROWCAT: growth category.

## biomarkers

In the above-mentioned NSLT, the false-positive rate of CT scan findings was 94%, which led to unnecessary additional testing in many individuals [4]. There is a double-expectation from biomarkers in this setting: (i) identify individuals with screen-detected abnormalities needing further invasive investigations; (ii) help to better stratify high-risk populations for screening studies. Molecular biology and gene technologies offer some of the most promising tools to identify early detection biomarkers for lung cancer, which may then become applicable to large-scale cost-effective screening in high-risk groups.

A large number of biomarkers have been studied for the early detection of lung cancer, including DNA, promoter hypermethylation, microsatellite instability, loss of heterozygosity, chromosomal aneusomy, tumour-associated antibodies, tumour-associated antigens, proteomic profiles, messenger RNA (mRNA), micro-RNA (miRNA) and volatile organic compounds (VOCs). These biomarkers have in turn been investigated in different specimens obtained by more or less invasive procedures: bronchial biopsies or bronchiolo-alveolar lavage fluid, induced sputum, buccal/nasal swabs, blood (plasma, serum, circulating tumour cells and peripheral blood mononuclear cells) and exhaled breath [26].

An ideal early detection biomarker for large-scale screening should be applicable on easily accessible specimens through non-invasive procedures, have an easy and reproducible quantification, a high sensitivity and specificity and a low cost. Many studies have been published on biomarkers for the early detection of lung cancer. While several molecular abnormalities in different clinically accessible specimens showed sensitivity and specificity up to 100% in feasibility studies, none of these biomarker-based tests can at present be recommended as screening procedures [26].

Much recent investigation concentrated on circulating miRNAs. miRNAs are post-transcriptional regulators that bind to complementary sequences on the target mRNA, resulting in translational repression and gene silencing. miRNAs are particularly interesting because of their tissue specificity and incredible stability, which make them easily detectable and quantifiable in body fluids [27]. In several solid tumours, patient's plasma miRNA has been shown to reflect, at least in part, the miRNA classes characterising that specific tumour, leading to the suggestion that miRNA levels in body fluids could be innovative non-invasive biomarkers [28]. Two recent studies identified miRNAs as promising biomarkers in the work-up of LDCT-detected nodules [29, 30]. The overlap between different studies is low, which indicates the need for more collaborative validation research between groups. The Early Detection Research Network (EDRN)—developed by the National Cancer Institute (NCI)—is such an effort to define guidelines for early detection biomarker development [31]. Furthermore, integration of biomarkers and clinical parameters in a model may improve results, e.g. one group reported superior results with the combination of genomic signatures in bronchial brushing in normal lung area and clinical features of high-risk individuals [32].

Recent efforts also tried to validate a panel of antibodies against six tumour-related antigens (p53, NY-ESO-1, CAGE, CBU4–5, Annexin 1 and SOX2). After technical validation, the assays have been standardised in order to obtain high reproducibility, precision and linearity [33]. The same team further carried out a clinical validation on three cohorts of patients with newly diagnosed lung cancer of 145, 241 and 269 patients, respectively, and obtained consistent results with sensitivity and specificity ranging between 36%–39% and 89%–91%, respectively [34]. The same group ran their test (EarlyCDT-Lung) on four new and independent datasets of 122, 249, 122 and 81 patients, respectively, and confirmed their results and the applicability of this test to detect lung cancer, independently of stage, in high-risk patients [35].

Analysis of exhaled breath is—because of its non-invasive nature and repeatability—another interesting approach. It is based on the presence of VOCs in low concentrations in exhaled breath. The exhaled mixture reflects metabolic activity in the body, which leads to different VOC signatures between normal individuals and across different types of cancer. Effective cancer detection has historically been described with sniffing dogs [36], but better standardisation is expected from ‘electronic noses’ based on mass spectrometry techniques [37], other sensor arrays [38] or gold nanoparticle sensors [39]. In one recent study using a colorimetric sensor array, there was a moderate accuracy in distinguishing lung cancer from control subjects (C-statistic 0.811), but this improved when clinical risk factors such as age, sex, smoking history and COPD were taken into account [40].

## bronchoscopy

Multidetector spiral CT technology is unable to detect pre-invasive endobronchial lesions and very limited early-stage invasive lung carcinoma in the central airways. Standard white light videobronchoscopy (WLB), autofluorescence bronchoscopy (AFB) and narrow-band bronchoscopy are endoscopic techniques that may detect these endobronchial central airway lesions.

An AFB added to WLB has been extensively investigated in the last decade with respect to (i) primary screening in patients at risk for early intraepithelial pre-invasive or invasive lesions; (ii) secondary screening, i.e. search for other synchronous lesions in patients with radiologically visible lung cancer, or for metachronous pulmonary lesions during the follow-up of patients with a curatively treated lung cancer; or (iii) surveillance, i.e. follow-up of patients known with central preinvasive lesions (Figure 2).

Figure 2

Patient with a previous left upper lobectomy for squamous cell lung cancer. Follow-up white-light bronchoscopy reveals no abnormalities (A), but autofluorescence bronchoscopy shows loss of green fluorescent activity at the apical segment of the right lower lobe (B). Biopsy revealed second primary early invasive squamous cancer (image courtesy CD).

Figure 2

Patient with a previous left upper lobectomy for squamous cell lung cancer. Follow-up white-light bronchoscopy reveals no abnormalities (A), but autofluorescence bronchoscopy shows loss of green fluorescent activity at the apical segment of the right lower lobe (B). Biopsy revealed second primary early invasive squamous cancer (image courtesy CD).

### primary screening

In primary screening, AFB is added to WLB for the detection of pre-invasive intraepithelial lesions and radio-occult invasive lesions in high-risk patients. The value of AFB in detecting these lesions in the large airways of high-risk patients has been evaluated in different clinical and research settings since the late 1990s (e.g. [41]), in RCTs [42, 43], and in a recent literature-based meta-analysis [44]. The addition of AFB to WLB resulted in the detection of a substantial number of intraepithelial pre-invasive or early invasive lesions that would have been missed with WLB alone; as a result, the ratio of sensitivity of (WLB + AFB)/WLB was consistently >1 in all published trials. When considering pre-invasive intraepithelial lesions [dysplasia or carcinoma in situ (CIS)] only, the pooled relative sensitivity of AFB + WLB versus WLB was 2.04 (95% CI 1.72–2.42), when considering invasive cancer only, it was 1.15 (95% CI 1.05–1.26) [44]. If one is screening for invasive cancer, then probably modern WLB is sufficient to detect these lesions, whereas screening for pre-invasive intraepithelial lesions will be more successful with autofluorescence videobronchoscopy. The disadvantage of AFB is its suboptimal specificity compared with WLB, which is responsible for a high number of ‘false-positive’ biopsies and subsequently important impact on the cost-effectiveness of this technique. The pooled relative specificity of AFB + WLB was only 65% of WLB alone in the meta-analysis [44].

### secondary screening

AFB may be used to detect synchronous pre-invasive or radio-occult invasive lesions in patients diagnosed with radiologically visible invasive lung cancer. Several series have shown that AFB added to WLB reveals such lesions in up to 10% of the patients [45, 46], thereby confirming the concept of field cancerisation in smokers with central airway lesions. AFB may play a role in the detection of metachronous second primary lung cancer in patients curatively treated for previous upper airway cancer, e.g. lung [47] or head and neck cancer [48]. The data on the role of AFB in patients with previous lung cancer are limited. The incidence of subsequent primary lung cancers is ∼3%–4% per patient per year. With standard follow-up, the diagnosis is made too late in the natural course of the disease such that the lesion is no longer resectable in half of these patients [49]. Several studies that assessed the yield of AFB in the detection of pre-invasive or invasive lesions have included previous resection for lung cancer as a particular risk factor in addition to other inclusion criteria such as smoking history or COPD. As an example, in an RCT assessing the potential of AFB to detect precancerous lesions in 1173 patients, 359 of them had known lung cancer or previous surgical resection of lung cancer [42]. In this particular subgroup, the prevalence of pre-invasive lesions was 6.7% in the patients assessed with WLB + AFB, not really higher than in those investigated with WLB alone (5.0%).

### surveillance

There is no standard algorithm for the follow-up of detected pre-invasive lesions or the management of high-grade pre-invasive lesions. Moreover, the risk and rate of progression of pre-invasive lesions to invasive squamous cell carcinoma, as well as the mechanism of progression or regression, are poorly understood. While evidence for a stepwise progression is weak, the concept of field cancerisation is strongly supported [50]. In an attempt to clarify the natural history of pre-invasive lesions, longitudinal studies using serial AFB and biopsies have been carried out in patients with dysplasia or CIS. These studies illustrated that almost no low-grade pre-invasive lesions (metaplasia to moderate dysplasia) progressed to a higher grade or invasive carcinoma, and even a large number of the severe dysplasia lesions did not progress to invasive carcinoma (Table 3). CIS, on the other hand, is a strong predictor of progression to invasive squamous cell carcinoma. One possible reason as to why regression occurs in up to two-thirds of dysplasia is complete removal of the pre-invasive lesion after an initial bronchial biopsy [51]. On the other hand, these longitudinal studies showed that patients with known pre-invasive lesions were at high risk of developing invasive lung cancer at any site in the lung, even if the known high-grade pre-invasive lesions regressed [50, 52–54]. The cumulative risk of development of CIS or invasive lung cancer was 7% at 1 year, 20% at 3 years and 44% at 5 years, and 85% of the detected invasive lung cancer developed at sites different from the follow-up sites [54]. This is consistent with the ‘field cancerisation’ concept, in which the entire bronchial epithelium is exposed to carcinogens and therefore at risk of developing invasive carcinoma. Therefore, AFB and CT surveillance in patients with known high-grade pre-invasive lesions is indicated for early detection and treatment with curative intent of CIS or invasive carcinoma arising at any site in the bronchial epithelium. At present, however, there are not yet randomised, controlled data that prove decrease of lung cancer-specific mortality rate with this approach.

Table 3.

Natural course of pre-invasive central airway lesions [51]

Pre-invasive lesion Regression Persistence Progression to CIS/INV
Metaplasia 37%–42% 29% 0%–9%
Mild/moderate dysplasia 64% 22% 0%–11%
Severe dysplasia 52%–63% 16% 11%–56%
CIS 12% 70% 21%–67%
Pre-invasive lesion Regression Persistence Progression to CIS/INV
Metaplasia 37%–42% 29% 0%–9%
Mild/moderate dysplasia 64% 22% 0%–11%
Severe dysplasia 52%–63% 16% 11%–56%
CIS 12% 70% 21%–67%

CIS, carcinoma in situ; INV, invasive cancer.

## disclosure

The authors have declared no conflicts of interest.

## references

1
Sox
HC
Screening for lung cancer with chest radiographs
JAMA
,
2011
, vol.
306
(pg.
1916
-
1918
)
2
Manser
RL
Irving
LB
de Campo
MP
, et al.  .
Overview of observational studies of low-dose helical computed tomography screening for lung cancer
Respirology
,
2005
, vol.
10
(pg.
97
-
104
)
3
Henschke
CI
Yankelevitz
DF
Libby
DM
, et al.  .
Survival of patients with stage I lung cancer detected on CT screening
N Engl J Med
,
2006
, vol.
355
(pg.
1763
-
1771
)
4
Aberle
DR
AM
Berg
CD
, et al.  .
Reduced lung-cancer mortality with low-dose computed tomographic screening
N Engl J Med
,
2011
, vol.
365
(pg.
395
-
409
)
5
Spiro
SG
Navani
N
Screening for lung cancer: is this the way forward?
Respirology
,
2012
, vol.
17
(pg.
237
-
246
)
6
Gohagan
JK
Marcus
PM
Fagerstrom
RM
, et al.  .
Final results of the Lung Screening Study, a randomized feasibility study of spiral CT versus chest X-ray screening for lung cancer
Lung Cancer
,
2005
, vol.
47
(pg.
9
-
15
)
7
Blanchon
T
Brechot
JM
Grenier
PA
, et al.  .
Baseline results of the Depiscan study: a French randomized pilot trial of lung cancer screening comparing low dose CT scan (LDCT) and chest X-ray (CXR)
Lung Cancer
,
2007
, vol.
58
(pg.
50
-
58
)
8
Van Iersel
CA
De Koning
HJ
Draisma
G
, et al.  .
Risk-based selection from the general population in a screening trial: selection criteria, recruitment and power for the Dutch-Belgian randomised lung cancer multi-slice CT screening trial (NELSON)
Int J Cancer
,
2007
, vol.
120
(pg.
868
-
874
)
9
Van Klaveren
RJ
Oudkerk
M
Prokop
M
, et al.  .
Management of lung nodules detected by volume CT scanning
N Engl J Med
,
2009
, vol.
361
(pg.
2221
-
2229
)
10
Saghir
Z
Dirksen
A
Ashraf
H
, et al.  .
CT screening for lung cancer brings forward early disease. The randomised Danish Lung Cancer Screening Trial: status after five annual screening rounds with low-dose CT
Thorax
,
2012
, vol.
67
(pg.
296
-
301
)
11
Lopes-Pegna
A
Picozzi
G
Mascalchi
M
, et al.  .
Design, recruitment and baseline results of the ITALUNG trial for lung cancer screening with low-dose CT
Lung Cancer
,
2009
, vol.
64
(pg.
34
-
40
)
12
Infante
M
Cavuto
S
Lutman
FR
, et al.  .
A randomized study of lung cancer screening with spiral CT (the Dante trial): three-year results
Am J Respir Crit Care Med
,
2009
, vol.
180
(pg.
445
-
453
)
13
Duffy
SW
Raji
OY
Agbaje
OF
, et al.  .
Use of lung cancer risk models in planning research and service programs in CT screening for lung cancer
Expert Rev Anticancer Ther
,
2009
, vol.
9
(pg.
1467
-
1472
)
14
2011

www.iaslc.org/policies/statement-on-ct-screening Available online, last accessed March 26, 2012.
15
Field
JK
Smith
RA
Aberle
DR
, et al.  .
International Association for the Study of Lung Cancer Computed Tomography Screening Workshop 2011 report
J Thorac Oncol
,
2012
, vol.
7
(pg.
10
-
19
)
16
Baldwin
DR
Duffy
SW
Wald
NJ
, et al.  .
UK Lung Screen (UKLS) nodule management protocol: modelling of a single screen randomised controlled trial of low-dose CT screening for lung cancer
Thorax
,
2011
, vol.
66
(pg.
308
-
313
)
17
Xu
DM
Gietema
H
De Koning
H
, et al.  .
Nodule management protocol of the NELSON randomised lung cancer screening trial
Lung Cancer
,
2006
, vol.
54
(pg.
177
-
184
)
18
Van 't Westeinde
SC
Horeweg
N
Vernhout
RM
, et al.  .
The role of conventional bronchoscopy in the work-up of suspicious CT screen detected pulmonary nodules
Chest
,
2012

February 2 [epub ahead of print]
19
Van 't Westeinde
SC
De Koning
HJ
Thunnissen
FB
, et al.  .
The role of the 18F-fluorodeoxyglucose-positron emission tomography scan in the Nederlands Leuvens Longkanker Screenings Onderzoek lung cancer screening trial
J Thorac Oncol
,
2011
, vol.
6
(pg.
1704
-
1712
)
20
Welch
HG
Black
WC
Overdiagnosis in cancer
J Natl Cancer Inst
,
2010
, vol.
102
(pg.
605
-
613
)
21
Silvestri
GA
Nietert
PJ
Zoller
J
, et al.  .
Attitudes towards screening for lung cancer among smokers and their non-smoking counterparts
Thorax
,
2007
, vol.
62
(pg.
126
-
130
)
22
Evans
WK
Wolfson
MC
Computed tomography screening for lung cancer without a smoking cessation program – not a cost-effective idea
J Thorac Oncol
,
2011
, vol.
6
(pg.
1781
-
1783
)
23
Ashraf
H
Tonnesen
P
Pedersen
JH
, et al.  .
Effect of CT screening on smoking habits at 1-year follow-up in the Danish Lung Cancer Screening Trial (DLCST)
Thorax
,
2009
, vol.
64
(pg.
388
-
392
)
24
Van Der Aalst
CM
Van den Bergh
KA
Willemsen
MC
, et al.  .
Lung cancer screening and smoking abstinence: 2 year follow-up data from the Dutch-Belgian randomised controlled lung cancer screening trial
Thorax
,
2010
, vol.
65
(pg.
600
-
605
)
25
McMahon
PM
Kong
CY
Bouzan
C
, et al.  .
Cost-effectiveness of computed tomography screening for lung cancer in the United States
J Thorac Oncol
,
2011
, vol.
6
(pg.
1841
-
1848
)
26
Mascaux
C
Peled
N
Garg
K
, et al.  .
Early detection and screening of lung cancer
Expert Rev Mol Diagn
,
2010
, vol.
10
(pg.
799
-
815
)
27
Mitchell
PS
Parkin
RK
Kroh
EM
, et al.  .
Circulating microRNAs as stable blood-based markers for cancer detection
,
2008
, vol.
105
(pg.
10513
-
10518
)
28
Chen
X
Ba
Y
Ma
L
, et al.  .
Characterization of microRNAs in serum: a novel class of biomarkers for diagnosis of cancer and other diseases
Cell Res
,
2008
, vol.
18
(pg.
997
-
1006
)
29
Boeri
M
Verri
C
Conte
D
, et al.  .
MicroRNA signatures in tissues and plasma predict development and prognosis of computed tomography detected lung cancer
,
2011
, vol.
108
(pg.
3713
-
3718
)
30
Shen
J
Liu
Z
Todd
NW
, et al.  .
Diagnosis of lung cancer in individuals with solitary pulmonary nodules by plasma microRNA biomarkers
BMC Cancer
,
2011
, vol.
11
pg.
374

31
Pepe
MS
Feng
Z
Janes
H
, et al.  .
Pivotal evaluation of the accuracy of a biomarker used for classification or prediction: standards for study design
J Natl Cancer Inst
,
2008
, vol.
100
(pg.
1432
-
1438
)
32
Beane
J
Sebastiani
P
Whitfield
TH
, et al.  .
A prediction model for lung cancer diagnosis that integrates genomic and clinical features
Cancer Prev Res
,
2008
, vol.
1
(pg.
56
-
64
)
33
Murray
A
Chapman
CJ
Healey
G
, et al.  .
Technical validation of an autoantibody test for lung cancer
Ann Oncol
,
2010
, vol.
21
(pg.
1687
-
1693
)
34
Boyle
P
Chapman
CJ
Holdenrieder
S
, et al.  .
Clinical validation of an autoantibody test for lung cancer
Ann Oncol
,
2010
, vol.
22
(pg.
383
-
389
)
35
Lam
S
Boyle
P
Healey
GF
, et al.  .
EarlyCDT-Lung: an immunobiomarker test as an aid to early detection of lung cancer
Cancer Prev Res
,
2011
, vol.
4
(pg.
1126
-
1134
)
36
Ehmann
R
Boedeker
E
Friedrich
U
, et al.  .
Canine scent detection in the diagnosis of lung cancer: revisiting a puzzling phenomenon
Eur Respir J
,
2012
, vol.
39
(pg.
669
-
676
)
37
Ligor
M
Ligor
T
Bajtarevic
A
, et al.  .
Determination of volatile organic compounds in exhaled breath of patients with lung cancer using solid phase microextraction and gas chromatography mass spectrometry
Clin Chem Lab Med
,
2009
, vol.
47
(pg.
550
-
560
)
38
Dragonieri
S
Annema
JT
Schot
R
, et al.  .
An electronic nose in the discrimination of patients with non-small cell lung cancer and COPD
Lung Cancer
,
2009
, vol.
64
(pg.
166
-
170
)
39
Barash
O
Peled
N
Tisch
U
, et al.  .
Classification of lung cancer histology by gold nanoparticle sensors
Nanomedicine
,
2012
, vol.
8
(pg.
580
-
589
)
40
Mazzone
PJ
Wang
X
Xu
Y
, et al.  .
Exhaled breath analysis with a colorimetric sensor array for the identification and characterization of lung cancer
J Thorac Oncol
,
2012
, vol.
7
(pg.
137
-
142
)
41
Edell
E
Lam
S
Pass
H
, et al.  .
Detection and localization of intraepithelial neoplasia and invasive carcinoma using fluorescence-reflectance bronchoscopy: an international, multicenter clinical trial
J Thorac Oncol
,
2009
, vol.
4
(pg.
49
-
54
)
42
Haussinger
K
Becker
H
Stanzel
F
, et al.  .
Autofluorescence bronchoscopy with white light bronchoscopy compared with white light bronchoscopy alone for the detection of precancerous lesions: a European randomised controlled multicentre trial
Thorax
,
2005
, vol.
60
(pg.
496
-
503
)
43
Hirsch
FR
Prindiville
SA
Miller
YE
, et al.  .
Fluorescence versus white-light bronchoscopy for detection of preneoplastic lesions: a randomized study
J Natl Cancer Inst
,
2001
, vol.
93
(pg.
1385
-
1391
)
44
Sun
J
Garfield
DH
Lam
B
, et al.  .
The value of autofluorescence bronchoscopy combined with white light bronchoscopy compared with white light alone in the diagnosis of intraepithelial neoplasia and invasive lung cancer: a meta-analysis
J Thorac Oncol
,
2011
, vol.
6
(pg.
1336
-
1344
)
45
Pierard
P
Vermylen
P
Bosschaerts
T
, et al.  .
Synchronous roentgenographically occult lung carcinoma in patients with resectable primary lung cancer
Chest
,
2000
, vol.
117
(pg.
779
-
785
)
46
Van Rens
MT
Schramel
FJ
Elbers
JR
, et al.  .
The clinical value of lung imaging fluorescence endoscopy for detecting synchronous lung cancer
Lung Cancer
,
2001
, vol.
32
(pg.
13
-
18
)
47
Weigel
TL
Kosco
PJ
Dacic
S
, et al.  .
Postoperative fluorescence bronchoscopic surveillance in non-small cell lung cancer patients
Ann Thorac Surg
,
2001
, vol.
71
(pg.
967
-
970
)
48
Lee
P
De Bree
R
Brokx
HA
, et al.  .
Primary lung cancer after treatment of head and neck cancer without lymph node metastasis: is there a role for autofluorescence bronchoscopy?
Lung Cancer
,
2008
, vol.
62
(pg.
309
-
315
)
49
Johnson
BE
Second lung cancers in patients after treatment for an initial lung cancer
J Natl Cancer Inst
,
1998
, vol.
90
(pg.
1335
-
1345
)
50
Ishizumi
T
McWilliams
A
Macaulay
C
, et al.  .
Natural history of bronchial preinvasive lesions
Cancer Metastasis Rev
,
2010
, vol.
29
(pg.
5
-
14
)
51
Dooms
C
Seijo
L
Gasparini
S
, et al.  .
Diagnostic bronchoscopy: state of the art
Eur Respir Rev
,
2010
, vol.
19
(pg.
229
-
236
)
52
Jeremy George
P
Anerjee
AK
CA
, et al.  .
Surveillance for the detection of early lung cancer in patients with bronchial dysplasia
Thorax
,
2007
, vol.
62
(pg.
43
-
50
)
53
Pasic
A
Vonk-Noordegraaf
A
Risse
EK
, et al.  .
Multiple suspicious lesions detected by autofluorescence bronchoscopy predict malignant development in the bronchial mucosa in high risk patients
Lung Cancer
,
2003
, vol.
41
(pg.
295
-
301
)
54
Alaa
M
K
Fujiwara
T
, et al.  .
Risk of lung cancer in patients with preinvasive bronchial lesions followed by autofluorescence bronchoscopy and chest computed tomography
Lung Cancer
,
2011
, vol.
72
(pg.
303
-
308
)