The detection of small lesions confirmed to be malignant, but which do not grow, spread, or cause death is referred to as overdiagnosis. This includes patients who are destined to die from another cause, e.g., comorbidity or an unexpected event, in addition to slow growing/non-spreading cancers [
10]. Overdiagnosis represents an important potential harm of screening, since it incurs additional cost, anxiety and morbidity associated with the cancer treatment. During earlier screening trials using chest radiographs in the Mayo and the Czechoslovakian randomised trials, substantially more cancers (20 %) were detected in the screened than in the unscreened group [
30,
31]. Nearly all of the excess cancers detected in the screened group in the Mayo clinical trial were early stage cancers. However, the failure to detect early stage cancers in the control group was without apparent ill effect: the control group experienced no excess number of lung cancer deaths [
30]. The results were generally confirmed by the Czechoslovakian study. Both studies suggest that screening is detecting “excess” lesions, which probably would not progress to advanced/lethal disease [
30,
31]. The PLCO trial [
17] examined 155,000 subjects in the general population and found 18 excess lung cancers in the chest radiography group (compared with no chest radiography group) after 6 years of follow-up (2 years after screening ended) and 76 lung cancers after 13 years of follow-up. Data from the same trial, evaluating overdiagnosis among a high-risk population only, showed a cumulative incidence of lung cancer of 606 per 100,000 person-years in the chest radiography group and 608 per 100,000 person-years in the usual care group after 6 years of follow-up.
The overdiagnosis rate for LDCT screening cannot yet be estimated [
24]. The NLST data shows a persistent gap of about 120 excess lung cancers in the LDCT versus the chest radiography arm, but further follow-up is needed [
32]. In both groups, the percentage of stage IA and stage IB lung cancers was high. Relative to the issue of overdiagnosis, fewer stage IV cancers were detected in the LDCT group than in the chest radiography group at the second and third screening rounds in the DANTE trial, where 2472 subjects were screened with chest radiography and sputum cytology at baseline and randomised afterwards to yearly LDCT or clinical follow-up. Lung cancer prevalence in the control chest radiography arm was 0.67 % (n = 8) and 50 % of these patients had stage I cancer, while the prevalence in the CT group was 2.19 % (n = 28) with 57 % stage I cancer, respectively. It has to be noted that 13 of the 28 LDCT lung cancer cases already had abnormal chest radiography findings at baseline [
33].
Still, while most lung cancer prevention experts think lung cancer screening leads to overdiagnosis, many clinicians believe it does not [
34]. Death rates from lung cancer imply that essentially all histological foci of lung cancer pose a threat to health, irrespective of their CT phenotype or how they are discovered. In the NLST, the size of the nodule and whether it is solid or sub-solid mattered. However, whether this appearance is linked to higher overdiagnosis probability remains to be concluded. Based on the Pan-Canadian early Detection of Lung Cancer Study (PanCan), McWilliams et al. [
35] presented a model to predict a cancerous pulmonary nodule (versus benign). Predictors for cancer were older age, female sex, family history of lung cancer, emphysema and larger nodule size, location of the nodule in the upper lobe, part-solid nodule type, lower nodule count and spiculation. Adopting such a model may direct the clinicians in their follow-up management.
Risk models
Risk models help to increase pre-test probability and reduce overdiagnosis. They improve the patient selection in order to define populations with higher pre-test probabilities: the Liverpool Lung Project (LLP) risk prediction model is used in the UKLS screening trial; the PLCO2012 (Prostate, Lung, Colorectal, and Ovarian) randomised trial and the NLST trial. The former two studies predict lung cancer detection, while the latter predicts death by lung cancer (Table
3).
Table 3
Risk prediction models used in different lung cancer screening trials
LLP (detection) | Age | 5 years | |
Sex |
Years of smoking |
Family history of lung cancer by age of affected relatives |
History of a previous cancer |
History of pneumonia |
History of exposure to asbestos |
PLCO (detection) | Age | 6 years | |
Race/ethnicity |
Education |
Body mass index |
Chronic obstructive pulmonary disease |
Personal history of cancer |
Family history of lung cancer |
Smoking status (current versus former) |
Smoking intensity (average cigarettes/day) |
Smoking duration |
Smoking quit time |
NLST (death) | Age | 5 years | |
Sex |
Ethnicity |
Body-mass index |
Pack-years of smoking |
Years since smoking cessation |
Presence of emphysema |
First-degree relative with lung cancer |
Recently, de Koning et al. [
39] published a study estimating the harms and benefits of lung cancer screening for efficient lung cancer screening policies. They used five separately developed micro-simulation models calibrated to the two largest randomised, controlled trials on lung cancer screening [
17,
39]. Those models were independently developed at five institutions: Erasmus Medical Center (Rotterdam), Fred Hutchinson Cancer Research Center (Seattle), Massachusetts General Hospital (Boston), Stanford University (Stanford), and University of Michigan (Ann Arbor). All account for the individual’s age-specific, smoking-related risk for lung cancer, date and stage of lung cancer diagnosis, the corresponding lung cancer mortality and the individual’s life expectancy in the presence and absence of screening. The most advantageous strategy identified is the annual screening from ages 55 through 80 years for ever-smokers with a smoking history of at least 30 pack-years and ex-smokers with less than 15 years since quitting. That approach would lead to 50 % of cases of cancer being detected at an early stage (stage I/II), 575 screening examinations per lung cancer death averted, a 14 % reduction in lung cancer mortality, 497 lung cancer deaths averted, and 5250 life-years gained per the 100,000-member cohort. Harms would include 67,550 false-positive test results, 910 biopsies or surgeries for benign lesions, and 190 overdiagnosed cases of cancer (3.7 % of all cases of lung cancer).
Thus far, there are no good risk predictors for nonsmokers and no convincing data to recommend screening. Lung cancer in never smokers is the seventh leading cause of cancer mortality, and therefore is a significant cause of death worldwide. The main risk factors include age, environmental tobacco exposure, cooking fumes, inherited genetic susceptibility, occupational and environmental exposure to carcinogens, hormonal factors, pre-existing lung disease and oncogenic viruses [
40]. Nonsmall cell lung cancer (NSCLC) in never smokers is clinically characterised by an increased incidence in females and a higher occurrence of adenocarcinoma in comparison to NSCLC in ever smokers in both surgical patients and non-resectable advanced stage patients [
41]. Even though those factors are known, there is no beneficial screening programme for lung cancer among this population.
False positives and complications during work-up
With modern multidetector CT, pulmonary nodules are detectable at a size of less than 2 mm. Small nodules are extremely common, but the vast majority of these nodules are benign. Given this fact, the definition of a positive screening result determines the number of false-positive results. On average, about 25 % of the thoracic surgical procedures performed during the various randomised controlled lung cancer screening trials were done for benign nodules [
21]. If there are fewer false-positive nodules, there is less need for further workup and lower risk of complications, especially from invasive diagnostic examinations including surgery.
The definition of a positive screening result differed substantially between the NLST and most European trials. The NLST defined any non-calcified nodule with a maximum diameter ≥ 4 mm as a positive screening result [
6]. As a consequence, the number of false-positive scans was high: 27 % of scans in the first two screening rounds, of which 96 % were false-positive. According to the NLST nodule management algorithm, these suspicious nodules needed further work-up: either a follow-up LDCT for nodules of 4–10 mm, or a referral to a pulmonologist for nodules > 10 mm in maximum diameter [
6].
The NELSON and some other European trials used a threshold of approximately 10 mm diameter (50 mm
3 volume) for a positive screening result, but also established an indeterminate group of nodules measuring 5–10 mm in diameter (50–500 mm
3 volume) that required earlier follow-up than the yearly screening interval [
42]. Only if significant growth (> 25 % volume change) was found, were these nodules considered a positive screening result. By using this approach, the number of scans with positive screening results was reduced from 27 % in the NLST to 2.7 % in the NELSON, and the false-positives could be reduced substantially from > 95 % in the NLST to approximately 50 % in the NELSON [
8,
43].
Recently, new criteria for the follow-up of pulmonary nodules, such as LungRADS and LU-RADS, have been presented in order to increase the positive predictive value in CT screening with minimum effect on sensitivity for the detection of malignancy [
44,
45].
The size of a nodule was measured in most screening trials, like the NLST, as the largest diameter of a pulmonary nodule [
6]. This approach suffers from a substantial inter-reader and intra-reader variability, which can be reduced by applying volumetric techniques, as used in the NELSON and other more recent trials. Non-actionable nodules were defined as those with benign morphology (e.g., calcification), small size (< 50 mm
3), and lack of or very slow growth of the solid component of a nodule with a volume doubling time (VDT) > 600 days. Indeterminate nodules were defined as nodules with a volume of the solid component between 50 and 500 mm
3, sub-solid nodules with a diameter of the ground glass component > 10 mm, or solid nodules with a VDT between 400 and 600 days. Actionable nodules were defined as solid components > 500 mm
3, more than 20 % growth in diameter of a ground glass component, or VDT < 400 days of a solid component [
42]. Non-actionable, reportable nodules were kept on regular (yearly) follow-up, indeterminate nodules were put on a more rapid follow-up of 3–6 months, while actionable nodules led to direct medical workup.
Increasing knowledge about the CT phenotypes of screen-detected pulmonary nodules with different biologic behaviours will lead to a better estimation of their probability of malignancy, and help to decrease the amount of additional follow-up scans and workup examinations [
46], e.g., perifissural nodules were demonstrated to have a high likelihood of being benign [
47,
48].
For the invasive diagnostic work-up of small nodules, the value of white light fibrebronchoscopy is very limited [
49], but newer diagnostic endoscopic techniques, such as endobronchial ultrasound-guided biopsy with mini probe or electromagnetic navigation bronchoscopy, might be more promising. For some peripheral nodules (> 1 cm), transthoracic CT-guided biopsy or primary resection by video-assisted thoracoscopic surgery for diagnostic and therapeutic reasons may be recommended [
50]. The risk of serious complications (pneumothorax requiring drainage, cardiorespiratory complications during anaesthesia, infection or haemorrhage) not only relates to the invasiveness of the diagnostic procedure itself, but also to the patient’s functional status [
51]. Subjects eligible for LDCT screening will present themselves mostly with a high comorbidity risk, due to COPD or chronic cardiovascular disease [
46,
52].
Adhering to a certified high quality radiology plan for LDCT screening will minimise radiation exposure for screening participants. Further, the adherence to a pulmonary nodule management plan based on nodule diameter, volume and growth rate will help to increase safety for lung cancer screening participants, mostly by decreasing the total amount of diagnostic investigations they will need to undergo in order to determine the nature of their screen-detected lung nodules. Moreover, a lower amount of false-positive lesions with a decreased number of additional diagnostic investigations may finally help to decrease participant’s anxiety and psychological stress during lung cancer screening [
53].
Radiation exposure
The vast majority of lung cancer screening trials were designed more than a decade ago. The LDCT protocols were simply achieved by reducing the fixed tube load of diagnostic CT from typically 100–300 mAs to 10–40 mAs. A CT dose index (CTDI
vol) of 2–3 mGy was used as a target for NLST [
54,
55]. Similar values were used in the NELSON and the various other European trials. The resulting effective dose is roughly 40 % of these values for males and 50 % for females, resulting in 1–1.3 mSv for a CTDI
vol of 2.5 mGy. The organ dose (mSv) to the lung or to the breast can be roughly estimated using 1.5 × CTDI
vol. Precise numbers vary depending on scanner type, and in particular on the pre-filtering of the X-ray spectrum.
With recent improvements in detector technology, automated exposure control techniques and iterative image reconstruction, a further substantial decrease in radiation exposure of 80 % to a level around 0.2 mSv is possible without impairing image quality [
56]. However, radiation exposure will always have to be higher in obese individuals than in normal weight individuals because of the difference in X-ray absorption. Excessive reduction of radiation dose will lead to image quality degradation with either high image noise or loss of image details, which will especially affect sub-solid lesions. These are the limiting factors for further dose reduction.
Radiation risk in the age range of 40 to 60 years is mainly determined by the organ dose to the lungs. Apart from the breast in premenopausal women, other organs have a much lower contribution to excess cancer risk [
57]. Radiation exposure and smoking appear to have an additive effect on cancer risk [
58]. This means that the excess risk for developing radiation-induced lung cancer may be twice as high in smokers as in never-smokers [
59].
Given an effective dose of 1.3 mSv for women and 1.0 mSv for men, the excess lifetime cancer risk was estimated to be 0.02 % in male smokers and 0.05 % in female smokers if three yearly screening rounds were performed [
60]. Risks did not change whether the starting age for screening was 30, 40 or 50 years. This implies that radiation risk becomes important only if the pre-test risk for lung cancer is small. Given a baseline cancer risk of 0.8–2.2 % in the various screening trials, the risk−benefit ratio is very favourable. Even if the number of screening examinations increases from three to 24, the excess lifetime cancer risk induced by radiation remains below the baseline cancer risk, but it increases with age [
38].
Radiation risk grows strongly if follow-up scans are performed using standard clinical protocols (old equipment 4–18 mSv, new equipment 2–4 mSv [
61]) instead of screening with LDCT settings (new equipment 0.2 mSv [
56]). For this reason, the work-ups of screen-detected nodules should remain within the screening programme as long as possible [
62].
Cost effectiveness
The cost-effectiveness of the screening intervention is one of the major considerations for those who are responsible for screening guidelines, practice measures and insurance coverage [
63]. Varying results on the cost-effectiveness of lung cancer screening have been reported [
64‐
67]. In their recent publication, the NLST reports reasonable cost-effectiveness of LDCT screening of lung cancer [
68]. LDCT screening as performed in the NLST trial costs $81,000 per quality-adjusted life year (QALY) gained (95 % CI $52,000–186,000). Screening trials that cost less than $100,000 per QALY are considered cost-effective. Incremental cost effectiveness ratio (ICER) is the ratio of the change in costs to incremental benefits of a therapeutic intervention or treatment [
69]. The NLST ICER was $52,000 (95 % CI $34,000–106,000). However, the ICER results of the NLST were highly sensitive to base-case assumptions. For example, if the reduction in mortality from causes other than lung cancer was included in the calculation, the QALY fell to $54,000. QALY increased to more than $100,000 when the cost of future care was increased. Moreover, estimated cost-effectiveness varied in the subgroup analysis. Screening with LDCT was much more cost-effective in women than in men and among the groups with a higher risk of lung cancer. Whether screening performed in different countries in Europe will be cost-effective depends on exactly how the screening will become implemented [
68], and on which respective cost structures and reimbursement policies will be used.
Expectation management
Expectation management is crucial for a successful CT screening programme. It is important for three main reasons: 1) giving participants the ability to understand the benefits and potential harms, 2) reducing anxiety in case a nodule is found and 3) reducing litigation and its chances for success. Screening is very likely to reduce a participant’s risk of dying from lung cancer. However, a substantial group of participants will still die from lung cancer. Most cancers found will be in a treatable stage (60–80 % stage I) – but not all [
8,
70]. Some cancers may grow so slowly that they will not be life-limiting and treating them may be unnecessary (overdiagnosis) [
39]. Screening is known to miss nodules present on LDCT [
71]. The annual screening programme will pick up nodules missed on earlier scans, which reduces the risk of missed nodules developing into untreatable cancer. As small nodules are extremely common, it is very likely that a nodule will be found. LDCT is not optimally suited for the detection and diagnosis of many other chest diseases. However, incidental findings leading to unnecessary workup, costs and complications may occur.
Information given to participants, clinicians not involved in screening and the public should be clearly understandable. Informed consent is important because of the dangers of undetected cancers, overdiagnosis or complications due to work-up or treatment of screen-detected lesions. The participants should be aware of the incidental finding policy of the screening programme.