Table of Contents  

Krajc, Marzluf, and Mueller: Principles of lung cancer screening – low-dose computerized tomography

Introduction

The prognosis of patients with advanced-stage lung cancer is dismal despite developments in targeted therapy. Lack of symptoms delays diagnosis, and thus more than 80% of patients present at an inoperable stage. Incidental discovery of an early-stage tumour, and adequate and timely treatment, greatly increase the chance of cure.

In spite of the measurable effects of smoking cessation campaigns and a declining prevalence of tobacco addiction in the western world, there is still a large population of (ex-)smokers at risk. Other factors (asbestos and radon exposure) will continue to contribute to the leading position of lung cancer in statistical reports on cancer-related mortality.

Now that computerized tomography (CT) is widely available, detecting the disease while it is still in its early stages should not represent a significant problem, should it? After all, screening for other cancers has been successfully implemented and is generally known to be effective in saving lives (colon, cervix, etc.).

Screening starts with the identification of individuals at risk of developing lung cancer, and then progresses to application of the actual test, interpretation of test results and appropriate response (observe/intervene). Implementing such a complex and possibly expensive strategy doubtlessly requires solid scientific proof that the strategy is beneficial in terms of saving patients’ lives.

The following is an attempt to intelligibly and concisely summarize the ideas presented in the plethora of original reports on lung cancer screening, their analyses, and emotional comments and heated replies.

Characteristics of an efficient screening regimen

Mortality reduction

In a successfully screened population of subjects at risk of developing a potentially lethal condition, the condition-specific mortality should be lower in the screened population than in a non-screened population. This parameter is considered by many to be the least biased measure of screening effectiveness. The detection of early-stage cancer, followed by successful treatment, should result in a reduction in mortality density ratio at a particular point in time when the cancer would have caused death had it progressed and remained undetected. As speed of progression exhibits interindividual variability, successful treatment will result in a depression of variable width in mortality density ratio several years after screening. The deepest point of the depression in mortality density ratio represents the maximum benefit of screening for and treating early cancer.1

However, the difference in cumulative mortality obscures the effect of screening. If mortality is measured too soon after one-time screening, it might be found to have increased because of complications of further diagnostic workup and treatment of positively screened patients, and because the positive effects of screening on mortality will show only after a significant, albeit expected, delay.

In the case of one-time screening, and provided the incidence of cancer in the given population is stable, new cases start accumulating and the mortality density ratio returns to normal values. Observing the population for too long after the effect of one-time screening has diminished, as well as including the initial period in evaluation of mortality, may dilute the difference and skew the results towards the erroneous conclusion that there is no or minimal difference in mortality of screened versus non-screened subjects at risk.

To keep the depression of mortality density ratio at a stable level, repeat screening rounds have to be added to detect newly occurring cases of early-stage cancer. The dimple will naturally be deepest in the first baseline screening round, as early-stage cancers, assuming that the incidence is unchanged and that disease progression stabilizes at a relatively slow rate, can accumulate in the population over a period of time. Repeat rounds of screening should identify disease that has been missed in the baseline round, or newly occurring early-stage tumours. Adding these successive and overlapping dimples to the dimple caused by baseline screening should stabilize the reduction in mortality density and prolong the interval during which the greatest benefit of screening is identifiable (Figure 1).

FIGURE 1

Effect of persistent uninterrupted annual screening rounds on estimation of reduction of mortality density ratio (based on Hanley1 and Henschke et al.2).

9-1-8-fig1.png

The consequences of stopping the screening process (annual rounds) have been demonstrated by comparing the International Early Lung Cancer Action Project (IELCAP) cohort and two other independent unscreened cohorts [the American Cancer Society Study (CPS-II) and the Carotene and Retinol Efficacy Trial (CARET)]. The cumulative lung cancer mortality rate curves, practically identical in the first 3 years, started separating as the effect of two screening rounds became evident, and they remained on divergent curves until the ninth year, when the screened cohort mortality started paralleling the unscreened cohort mortality again.2

Identical phenomena could be observed in a colon cancer screening study (interrupted because of funding issues)3 and in breast cancer screening trials, as well as in the National Lung Screening Trial (NLST), which observed a decrease in the frequency of detection of stage I lung cancer from 63% to 31% after screening was stopped.4

The mortality reduction observed in the NLST,5 reported as the overall lung cancer-related mortality reduction of 20%, would probably have been significantly greater had the duration of screening been longer and the follow-up period had allowed for lengthier observation of the effects of screening. Owing to the lead time of 4–5 years (simply put, the lag between screening and the time when deaths would have occurred) the difference between the cumulative and annual mortality rates would be achieved only 7 years after baseline screening. The NLST was powered to detect a 50% reduction in cancer-specific mortality after 2 years, and a 20% reduction after 5 years. For ethical and funding reasons the trial was stopped after detecting the required cumulative mortality difference; however, the reported results underestimate the efficiency of screening with regard to saving lives. Mortality reduction demonstrated by the NLST depends on the study design and is not an inherent parameter that reflects the curability of screen-diagnosed lung cancers.6 The same misinterpretation led to disappointing conclusions when analysing the early lung cancer screening trials.

Stage shift

Another phenomenon to expect with efficient screening is stage shift, i.e. increased detection of early-stage disease and reduced detection of disease at later stages, as the detected and screened tumours do not progress to the later stages. Assuming an unchanging distribution of stages and stable progression from early to advanced stages in every subject, an easily detectable staging shift would need to occur after six or seven annual screening rounds.

The NLST identified more advanced-stage patients in the cohort receiving chest radiography after only three rounds of screening, with most LDCT-detected cancers being stage IA (47.5%) and most radiography-detected cancers being stage III or IV (59.1%).5 Predictions based on several natural history and screening models calibrated to NLST and PLCO (Prostate, Lung, Colorectal and Ovarian cancer screening trial) data were compliant with the observed stage distributions.7 The IELCAP study implies an analogous stage shift (85–92% of screen-detected tumours with no metastatic disease).8

Improved survival

It appears to be intuitively simple and logical to assume that improved survival of patients with cancers detected by screening is sufficient proof of screening efficiency. Knowing for a fact that patients with early-stage lung cancer who are adequately and timeously treated survive significantly longer and that survival rates (depending on tumour size) will be incredibly high, doubts seems entirely unjustified.

However, a simple comparison of annual survival rates is not without flaws (discussed in more detail below). Therefore, a quasi-experimental approach relying on Kaplan–Meier methodology to balance potential confounders must be used to compare screened and unscreened cohorts.

Several biasing phenomena can negatively influence the validity of survival analysis of screened patients: lead time bias, length bias and overdiagnosis.

Lead time bias

If cancer is discovered in the early stages, time to death will naturally be longer than for advanced-stage cancers, even if no intervention takes place. The actual effect of treatment applied to early-stage disease thus appears to be devalued by such lead time bias. However, this happens only if annual survival rates are used. Estimation of cure rate without lead time bias is possible using the cure rates of screened and unscreened cohorts when survival curves have reached their asymptotic value.6,9 For the Kaplan–Meier methodology with cure models to work for screening, a baseline screen and at least one annual screening round are necessary. Common sense tells us that lead time is the very purpose of screening and presents the opportunity to intervene before the disease becomes incurable.10

Length bias

Tumours that grow slowly relative to screening frequency are more prone to detection by screening than fast-growing tumours, which may be missed and become clinically manifest in between screening rounds. Thus, the screening cohort will involve more slowly growing than rapidly growing tumours and appear to have better survival rates as a result of lead time bias.11

Faster-growing tumours are more likely to be identified during repeat screening rounds, whereas the slower ones should show up during the initial screening. To eliminate lead time bias, a separate analysis of baseline and repeat screening rounds can be carried out, thus not only addressing the bias problem but also gaining information on aggressiveness of various cell types.6,12

Overdiagnosis

Overdiagnosis refers to the assumption that some lung cancers will not cause symptoms or death during a patient’s lifetime because of their slow growth or because of ‘competing’ causes of death, i.e. the patient succumbs to another disease before their cancer becomes fatal. If a portion of cancers detected by screening were overdiagnosed, survival of the screened cohort would appear to be better because of the inclusion of these ‘harmless’ cancers. Overdiagnosis is different from the false-positive rate of screening (assigning cancer where there is none) – as a histological diagnosis of cancer is already confirmed. The extent of overdiagnosis can be estimated from the survival curve of untreated subjects after it reaches its asymptote; a non-zero asymptote indicates overdiagnosis (lead time method).9,13

Another, more common, method of quantifying overdiagnosis compares the numbers of cancer cases in screened and unscreened cohorts and, assuming an equal distribution of cancers in both cohorts, attributes the excess of cases in the screened cohort to overdiagnosis (excess incidence method).14 Longer duration of screening improves the accuracy of this approach. A modelling approach based on convolution models and mean sojourn time was used to estimate overdiagnosis in the screened cohort of the NLST.15,16 As the entire concept of overdiagnosis is somewhat elusive, any estimate of overdiagnosis should be regarded with caution and attention to methodology.8 As a result of variability in screening duration (3–5 years), follow-up duration (5–7 years vs. lifetime) and histological type of cancer (‘bronchioloalveolar’ vs. other), the published estimates overdiagnosis to vary from 2.6% to > 80%.17

It is unlikely that patients with early-stage lung cancer would voluntarily take part in a trial that analyses their survival but does not offer treatment. Information on case fatality rate and length of survival of such patients comes from early screening trials with 45 pooled cases18 and analysis of data from 1432 cases from the California Cancer Registry.19 In a heated debate over whether or not overdiagnosis truly exists, these two papers are often presented as a proof that lung cancer, if left untreated, is lethal in 100% of cases (definition of cancer). On taking a closer look, we find non-zero, albeit small, cancer-specific survival rates at 5 years (0–8%18 and 22% with a median of 20 months).19

This categorical, binary view (lethal vs. non-cancer) of lung cancers has prevented advocates of different approaches to evaluating overdiagnosis from reaching a consensus.

The new lung adenocarcinoma classification20 reflects the fact that some tumours and their precursors (formerly labelled bronchioloalveolar carcinoma) are associated with a relatively favourable prognosis and progress very slowly, with lengthy volume doubling times, do not progress at all or even spontaneously regress. Thus, a continuous spectrum of aggressiveness may be present.

The existence of such indolent lesions (it has been suggested that the term ‘cancer’ is avoided;21 instead, low-risk lesions could be labelled as indolent lesions of epithelial origin) explains a substantial proportion of overdiagnosed cases, which can now be identified based on correlation of the lesion’s appearance and growth dynamics on CT2225 with underlying histology.2628

Overdiagnosis due to expected (severe comorbidity) or unexpected (trauma) competing causes can be diminished either by selecting screenees based on their general health status and life expectancy or by using the Kaplan–Meier modelling approach, in which subjects dying of non-cancer causes do not influence the actual survival curve because of censoring.

The concept of overdiagnosis is far more important in the scientific setting; in clinical practice, overdiagnosis is quite harmless as long as it does not lead to overtreatment.

Different approaches to lung cancer screening research

The dogma of randomized controlled trial

In evaluating screening effectiveness, the dogma of paramount superiority of the randomized control trial (RCT) in obtaining evidence-based, valid information has been causing persistent disagreement in interpreting existing data. RCTs of screening versus no screening can also suffer from biases (dilution by including periods when no effect can have yet occurred, non-/adherence bias, bias of false attribution and false non-attribution of cancer to death, arbitrarily short duration of screening).10 Furthermore, such trials are expensive and take a very long time to complete, with further delays until an analysis of the data is available.29 Extensive application of RCT principles to evaluating efficacy of screening for breast cancer led to ‘spectacular failure’ of eight trials to detect a screening benefit.10

International Early Lung Cancer Action Project approach

The IELCAP has been prospectively building its huge cohort of LDCT-screened patients and accumulating and analysing data in a smart, innovative, logical, methodical and standardized way while at the same time allowing for flexibility in updating the screening regimen based on new knowledge. Across a variety of journals, countless editorials, letters to the editor and replies have been published in attempts to attack, defend or clarify the IELCAP methodology, thus providing some high-quality entertainment to clinicians seeking answers within the monotonous world of epidemiology.29

In a (in)famous (in part because of a missing disclosure and suspicious ethics of funding) 2006 analysis published in the New England Journal of Medicine,30 screening of 31 567 asymptomatic at-risk subjects yielded 302 patients with stage I cancers who underwent resection within 1 month after diagnosis; the estimated 10-year survival rate reached 92%. The apparent large discrepancy between the reduction in case fatality rate (almost 70%) and the reduction in cancer-specific mortality (20%) detected by the NLST stems from different statistical and modelling approaches to efficacy evaluation mentioned earlier. The IELCAP approach has, of course, been criticized for not providing a control group and underestimating the very existence of notorious biases.31,32 Such criticisms arose from persistent adherence to the tenets of RCT in statistical thinking.10

The IELCAP team gradually provided an impressive body of analyses on mortality reduction,2 behaviour of non-calcified pulmonary nodules2325,33 and the importance of baseline screening, as well as ongoing repeat rounds of screening.24,27,28

A direct comparison to NLST data4 identified a higher prevalence of stage I lung cancer, as well as significantly higher 5-year survival rates in the IELCAP cohort (83% vs. 62%), which can mainly be explained by differences in regimen applied, i.e. ongoing annual screening rounds in IELCAP cohort and only three annual rounds in the NLST. The importance of continuing annual screening is also supported by results from the Italian COSMOS (Continuous Observation of SMOking Subjects) study.34

National Lung Screening Trial

The only RCT that compared lung cancer screening using LDCT with screening by chest radiography that successfully refuted the null hypothesis (of no cancer-related mortality benefit of LDCT screening) was the NLST.5,35,36

A total of 53 454 participants (aged 55–74 years with a smoking history of at least 30 pack-years, including quitters in the preceding 15 years) were randomized to three annual screenings with either LDCT or single-view posteroanterior chest radiography and subsequent follow-up. The lung cancer detection rate was 24.2% in the LDCT group and 6.9% in the chest radiography group. The frequency of false-positive results was similar in both arms (96.4% vs. 94.5%). The number of deaths from lung cancer was 247 per 100 000 person-years in the LDCT arm, compared with 309 per 100 000 person-years in the radiography arm, resulting in a relative reduction in lung cancer mortality associated with LDCT screening of 20.0%. In the LDCT arm, overall mortality was reduced as well: by 6.7%. For a typical trial participant, screening 256 persons annually for 3 years would prevent one lung cancer death every 6 years. An incorrect interpretation of this result is that screening a high-risk population prevents 20% of deaths from lung cancer. The fallacy of using overall mortality rate as the outcome for screening trials has already been discussed. Nevertheless, this trial rightfully served as the single most important reference for all influential organizations in support of implementation of LDCT lung cancer screening.

Other RCTs performed in Europe failed to reject their null hypotheses [DANTE (Detection And screening of early lung cancer with Novel imaging TEchnology),37 DLSCT (Danish randomized controlled lung cancer screening trial)38], and the results of the NELSON (Nederlan-Leuvens Longkanker Screenings Onderzoek, Dutch Belgian randomized lung cancer screening trial) trial3941 are eagerly awaited (this trial was due to finish in December 2015), as well as the outcome of the UKLS (UK Lung Cancer Screening Trial) trial,42,43 by design focusing on higher-risk subjects. A pooled analysis of eligible European trials is also planned (European randomized lung cancer CT screening, EUCT).44

Risk models as a means of improving efficacy and reducing the false-positive rate

NLST inclusion criteria (asymptomatic individuals aged 55–74 years with a smoking history of at least 30 pack-years of smoking and no more than 15 years since quitting) worked for the trial, but the resulting false-positive rate (96.4%) was rather high. Reduction of the high false-positive rate of LDCT screening regimen and elimination of overdiagnosis due to expected competing causes of death in order to maximize the benefit–harm ratio can be achieved by risk model-based preselection of patients eligible for screening.

Comparability of the existing plethora of models is limited, as their optimal performance is naturally observed within the cohort that the given model is based on, but some models have been externally validated.45

Identifying high-risk cohort in lung cancer screening

The best AUC ROC (area under curve of the receiver operating characteristic) values were reported for PLCO (0.859), PLCOm2012 (0.803) and Hoggart (0.843) models. PLCOm201246,47 is the best model for capturing a high proportion of cases, whereas the Liverpool Lung Project (LLP) model48 (AUC 0.67–0.82) delivers the best results in capturing a lower proportion of cases in a high-risk population and thus improving cost-effectiveness.49

In most models the crucial parameter is smoking history and smoking status. However, some have been designed to include never-smokers as well (LLP, PLCO). Highly predictive models (AUC 0.92) can be efficiently used for identifying high-risk and already oligosymptomatic patients in primary care.50

Risk calculators

An uncomplicated way to get a grasp of the various risk models is provided by a few instructive online risk calculators:

Sex, age, education level, body mass index (BMI), number of past radiographs, relative with lung cancer of early or late onset, smoking status, smoking duration, smoking intensity, age at start and quitting smoking, exposure to asbestos, previous malignancy, previous pneumonia and presence of emphysema are entered and subsequently evaluated according to four different models (Spitz,54 LLP,48 Hoggart55 and PLCO47).

The USPSTF (United States Preventative Services Task Force) website (http://www.shouldiscreen.com/lung-cancer-risk-calculator/) presents selection criteria and assesses risk based on age; current smoking status, duration and intensity; sex; education; race; BMI; previous malignancy; relatives with lung cancer; and presence of chronic obstructive pulmonary disease. This calculator evaluates eligibility for screening and also provides intelligible information on the risk of developing lung cancer in the following 6 years, as well as an illustration of the number of deaths prevented by screening individuals with similar characteristics. In addition, potential screenees can give feedback on their perception of net benefit.56

MyLungRisk is based on the LLP risk model.

Other means of preselecting high-risk patients currently under development include fluorescence bronchoscopy, genomic and proteomic analysis of bronchoscopic samples, serum protein microarrays and microRNA analysis, as well as the very attractive and entirely non-invasive exhaled breath analysis, which is dealt with in the second part of this review (‘Lung cancer screening – exhaled breath analysis’).

Harms of screening

Radiation exposure

Although LDCT scanning delivers only approximately 1.3–1.5 millisievert (mSv) of radiation (yearly background dose being around 3 mSv), it is the use of diagnostic CT to follow up nodules, with doses reaching 8 mSv, that gives rise to concerns about excess cumulative exposure. Estimation based on nuclear worker cohort studies and atomic bomb survivor studies designates exposure from (unrealistically long, 20–30 years) CT screening with non-trivial levels of risk;58 long-term protocols can independently increase the risk of lung cancer beyond cigarette smoking. Nevertheless, one (quite delayed) cancer death caused by radiation from imaging per 2500 subjects undergoing screening is negligible compared with the benefit of preventing lung cancer deaths.59 Carcinogenicity of low-dose radiation has been challenged in a recent review.60 Updated data from atomic bomb survivors no longer correspond to the linear threshold model used to estimate radiation-associated risk of developing cancer. Similar re-evaluation of radiation workers and an analysis of DNA damage by low-dose radiation negated the original assumption of carcinogenicity; in fact, a completely opposite, protective, effect has been suggested. Efforts to further reduce the dose delivered by LDCT may actually harm the patient by providing imaging of inferior quality and thus suboptimal input for subsequent assessment of suspicious nodules.

Unnecessary diagnostic procedures and unnecessary treatment

In the NLST, 1.2% of patients with benign final diagnosis underwent an invasive procedure and 0.7% of such patients had a surgical diagnostic procedure. Limited data from this trial suggest that eight deaths per 10 000 subjects screened by LDCT are caused by complications of diagnostic workup (bronchoscopy, needle biopsy or surgical biopsy) of detected (benign or malignant) abnormalities. The 60-day mortality of subjects who turned out to have benign lesions and underwent a diagnostic or (mostly) a therapeutic intervention was as low as 0.06%, which can be regarded as a small price to pay for the 20% mortality reduction found in the NLST.

Other harms

There have been conflicting reports on psychological harms induced by LDCT screening.59 A taxonomy including physical, psychological effects, financial strain and opportunity costs has been suggested to enable further study of this subject with potential social and economic implications.61

Computerized tomography screening algorithms – screening regimen

Low-dose requirements for computerized tomography screening

By definition, low-dose CT uses settings that allow the radiation dose to be ‘as low as reasonably achievable’ (ALARA). In NLST, lower tube currents (40 mA) and tube voltages (120–140 kVp) and a pitch of 1.5 reduced the mean effective dose to 1.4 mSv. Other trials have used similar settings, sometimes with adjustment for BMI (higher doses for higher BMI). Increased image noise due to further dose lowering can impair evaluation, especially for ground-glass nodules (GGNs).62

Morphology, size and invasiveness of screen-detected nodule

Non-calcified pulmonary nodules (NCNs) detected by CT can be classified as solid or subsolid.

Solid nodules represent a focal area of high attenuation completely obscuring the lung parenchyma, leaving no normal structures visible. In terms of adenocarcinoma invasiveness, they represent invasive cancers. Solid nodules > 10 mm require an immediate diagnostic workup because of the high risk of malignancy.

Subsolid nodules (any nodule containing elements of less than solid density) occur either as pure GGNs or as part-solid nodules.

Pure GGNs are areas of high attenuation that do not obscure the structure of underlying parenchyma, bronchi and vessels. Representing preinvasive lesions, they correspond either to atypical adenomatous hyperplasia (< 5 mm) or to adenocarcinoma in situ (> 5 mm). The majority of these lesions resolve fully (especially when < 5 mm). However 10 mm (sometimes 8 mm) is regarded as the threshold of invasiveness/malignancy. Therefore, long-term follow-up of lesions > 5 mm is recommended.

Part-solid nodules contain both solid and ground glass. When the solid component exceeds 5 mm, these lesions mostly correspond to minimally invasive adenocarcinoma. Owing to their potentially malignant nature, lack of growth should be proven by a follow-up CT at 3 months.20 To evaluate such lesions, the solid component should be measured on mediastinal windows, while also recording total size of the nodule with the ground-glass component. Emergence of a solid component in a pure GGN is a sign of potentially developing invasiveness. Lung-RADS™ (or lung imaging reporting and data system) criteria from the American College of Radiology63 use a somewhat more conservative measure of 6 mm based on IELCAPs analysis.25

Volumetric analysis is the basis for computation of volume doubling time during screening, thus allowing for assessment of growth rate and malignant potential of the nodule. Indolent GGNs may exhibit volume doubling times longer than 800 days, semisolid lesions around 200–400 days and solid lesions around 100–150 days. Risk of malignancy of a NCN can also be estimated using calculators based on parameters of sex, nodule size, spiculation and location in upper lobes.52

Baseline-detected versus annual-detected nodules

Slow-growing tumours are more likely to be detected during baseline screening, as they have accumulated in the population in the preceding years. Faster-growing tumours will emerge in the population de novo after baseline screening and should be detected by (annual) repeat rounds (Figure 2). This causes detection of two differing cancer populations with different risk of progression and different needs for timing of follow-up studies outside the annual screening schedule.6

FIGURE 2

Faster-growing tumours are more likely to be detected by annual screening, whereas baseline screening generally detects more slow-growing tumours.

9-1-8-fig2.png

Guidelines of scientific bodies and research institutions

Based on the results of the NLST, most professional organizations focusing on lung cancer research and management now recommend screening high-risk individuals (e.g. https://www.iaslc.org/research-education/policies, http://www.nccn.org/professionals/physician_gls/f_guidelines.asp#detection).6469 There are slight differences (Figure 3) in the definitions of nodule size thresholds and in the recommended intervals at which lesions that display potentially malignant changes should be followed up, as well as in definitions of the population at risk based on various risk models. The recommendations and guidelines of these organizations importantly influence policies of health-care providers and thereby the potential of lung cancer screening implementation.

FIGURE 3

Comparison of several guidelines for LDCT lung cancer screening. Note the differing criteria for solid and non-solid nodules at baseline and repeat screening rounds. Algorithms for endobronchial nodules are omitted on purpose. Malignant growth rate evaluation is dependent on nodule size ratio [(new diameter minus old diameter) divided by old diameter] as follows: for nodule diameter < 6 mm, % change > 50% growth, for nodule diameter 6–9 mm, % change > 30% growth, for nodule diameter > 10 mm, % change > 20% growth.64,6769

9-1-8-fig3.png

Cost-effectiveness

A model analysing the cost-effectiveness of LDCT screening based on NLST data estimated that LDCT screening would cost US$81 000 per quality-adjusted life year (QALY) gained.70 However, these estimates reflect the situation in the USA. Based on the cost of care in the UK and a trial design directed at lowering the false-positive rate (UKLS), the baseline estimate for the incremental cost-effectiveness ratio (ICER) of once-only CT screening relative to symptomatic presentation was £8466 (GBP) per QALY. CP5YS (cost per 5-year survivor) is another, perhaps more understandable and applicable, measure of evaluating cost-efficiency. In this model,71 estimated CP5YS values were higher for the symptom-detected approach (US$86 400 to US$233 300) than for the proactive CT screening approach (US$149 400 to US$282 100).

Closing remarks on low-dose computerized tomography screening

If people at high risk of developing lung cancer are told that screening will reduce their risk of death from lung cancer by only 20%, a large number will decide not to be screened. If they are made aware that perhaps three out of four otherwise incurable lung cancers could be cured if detected by the screening programme, the number of people choosing to participate would be greatly increased.6 Implementing a screening programme requires adequate adjustments of existing diagnostic and therapeutic infrastructure, which can be accompanied by further costs and unforeseen difficulties even in high-volume centres dedicated to treatment of lung cancer patients.

References

1. 

Hanley JA. Analysis of mortality data from cancer screening studies: looking in the right window. Epidemiology 2005; 16:786–90. http://dx.doi.org/10.1097/01.ede.0000181313.61948.76

2. 

Henschke CI, Boffetta P, Gorlova O, Yip R, Delancey JO, Foy M. Assessment of lung-cancer mortality reduction from CT Screening. Lung Cancer 2011; 71:328–32. http://dx.doi.org/10.1016/j.lungcan.2010.10.025

3. 

Miettinen OS, Henschke CI, Pasmantier MW, Smith JP, Libby DM, Yankelevitz DF. Mammographic screening: no reliable supporting evidence? Lancet 2002; 359:404–5.

4. 

Yip R, Henschke CI, Yankelevitz DF, Boffetta P, Smith JP. The impact of the regimen of screening on lung cancer cure: a comparison of I-ELCAP and NLST. Eur J Cancer Prev 2015; 24:201–8. http://dx.doi.org/10.1097/CEJ.0000000000000065

5. 

Aberle DR, Adams AM, Berg CD, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011; 365:395–409. http://dx.doi.org/10.1056/NEJMoa1102873

6. 

Henschke CI, Boffetta P, Yankelevitz DF, Altorki N. Computed tomography screening: the international early lung cancer action program experience. Thorac Surg Clin 2015; 25:129–43. http://dx.doi.org/10.1016/j.thorsurg.2014.12.001

7. 

Meza R, ten Haaf K, Kong CY, et al. Comparative analysis of 5 lung cancer natural history and screening models that reproduce outcomes of the NLST and PLCO trials. Cancer 2014; 120:1713–24. http://dx.doi.org/10.1002/cncr.28623

8. 

Mortani Barbosa EJJ. Lung cancer screening overdiagnosis: reports of overdiagnosis in screening for lung cancer are grossly exaggerated. Acad Radiol 2015; 22:976–82. http://dx.doi.org/10.1016/j.acra.2014.10.011

9. 

Henschke CI, Yankelevitz DF, Kostis WJ. CT screening for lung cancer. In: Schoepf UJ (ed.) Multidetector-Row CT of the Thorax. Cham, Switzerland: Springer Link; 2004, pp. 133–44.

10. 

Miettinen OS, Yankelevitz DF, Henschke CI. Evaluation of screening for a cancer: annotated catechism of the Gold Standard creed. J Eval Clin Pract 2003; 9:145–50. http://dx.doi.org/10.1046/j.1365-2753.2003.00412.x

11. 

Sharma D, Newman TG, Aronow WS. Lung cancer screening: history, current perspectives, and future directions. Arch Med Sci 2015; 11:1033–43. http://dx.doi.org/10.5114/aoms.2015.54859

12. 

Henschke CI, McCauley DI, Yankelevitz DF, et al. Early Lung Cancer Action Project: overall design and findings from baseline screening. Lancet 1999; 354:99–105. http://dx.doi.org/10.1016/S0140-6736(99)06093-6

13. 

Etzioni R, Gulati R. Oversimplifying overdiagnosis. J Gen Intern Med 2014; 29:1218–20. http://dx.doi.org/10.1007/s11606-014-2867-0

14. 

Zahl P-H, Jørgensen KJ, Gøtzsche PC. Lead-time models should not be used to estimate overdiagnosis in cancer screening. J Gen Intern Med 2014; 29:1283–6. http://dx.doi.org/10.1007/s11606-014-2812-2

15. 

Patz EF, Pinsky P, Gatsonis C, et al. Overdiagnosis in low-dose computed tomography screening for lung cancer. JAMA Intern Med 2014; 174:269–74. http://dx.doi.org/10.1001/jamainternmed.2013.12738

16. 

Patz EF, Pinsky P, Kramer BS. Estimating overdiagnosis in lung cancer screening--reply. JAMA Intern Med 2014; 174:1198–9. http://dx.doi.org/10.1001/jamainternmed.2014.1525

17. 

Detterbeck FC. Cancer, concepts, cohorts and complexity: avoiding oversimplification of overdiagnosis. Thorax 2012; 67:842–5. http://dx.doi.org/10.1136/thoraxjnl-2012-201779

18. 

Flehinger BJ, Kimmel M, Melamed MR. The effect of surgical treatment on survival from early lung cancer. Implications for screening. Chest 1992; 101:1013–18.

19. 

Raz DJ, Zell JA, Ou S-HI, Gandara DR, Anton-Culver H, Jablons DM. Natural history of stage I non-small cell lung cancer: implications for early detection. Chest 2007; 132:193–9. http://dx.doi.org/10.1378/chest.06-3096

20. 

Gardiner N, Jogai S, Wallis A. The revised lung adenocarcinoma classification-an imaging guide. J Thorac Dis 2014; 6(Suppl. 5):537–46. http://dx.doi.org/10.3978/j.issn.2072-1439.2014.04.05

21. 

Esserman LJ, Thompson IM, Reid B, et al. Addressing overdiagnosis and overtreatment in cancer: a prescription for change. Lancet Oncol 2014; 15:e234–42. http://dx.doi.org/10.1016/S1470-2045(13)70598-9

22. 

Henschke CI, Yankelevitz DF, Naidich DP, et al. CT screening for lung cancer: suspiciousness of nodules according to size on baseline scans. Radiology 2004; 231:164–8. http://dx.doi.org/10.1148/radiol.2311030634

23. 

Henschke CI, Yankelevitz DF, Mirtcheva R, McGuinness G, McCauley D, Miettinen OS. CT screening for lung cancer: frequency and significance of part-solid and nonsolid nodules. AJR Am J Roentgenol 2002; 178:1053–7. http://dx.doi.org/10.2214/ajr.178.5.1781053

24. 

Henschke CI, Yankelevitz DF, Yip R, et al. Lung cancers diagnosed at annual CT screening: volume doubling times. Radiology 2012; 263:578–83. http://dx.doi.org/10.1148/radiol.12102489

25. 

Henschke CI, Yip R, Yankelevitz DF, Smith JP. Definition of a positive test result in computed tomography screening for lung cancer: a cohort study. Ann Intern Med 2013; 158:246–52. http://dx.doi.org/10.7326/0003-4819-158-4-201302190-00004

26. 

Zhang L, Yankelevitz DF, Carter D, Henschke CI, Yip R, Reeves AP. Internal growth of nonsolid lung nodules: radiologic-pathologic correlation. Radiology 2012; 263:279–86. http://dx.doi.org/10.1148/radiol.11101372

27. 

Yankelevitz DF, Yip R, Smith JP, et al. CT screening for lung cancer: nonsolid nodules in baseline and annual repeat rounds. Radiology 2015; 277:555–64. http://dx.doi.org/10.1148/radiol.2015142554

28. 

Xu DM, Yip R, Smith JP, Yankelevitz DF, Henschke CI. Retrospective review of lung cancers diagnosed in annual rounds of CT screening. AJR Am J Roentgenol 2014;203:1–8. http://dx.doi.org/10.2214/AJR.13.12115

29. 

Castleberry AW, Grannis FW. What is a reasonable cost to refute a preposterous hypothesis? Br J Cancer 2010; 102:627–8. http://dx.doi.org/10.1038/sj.bjc.6605523

30. 

Henschke CI, Yankelevitz DF, Libby DM, Pasmantier MW, Smith JP, Miettinen OS. Survival of patients with stage I lung cancer detected on CT screening. N Engl J Med 2006; 355:1763–71. http://dx.doi.org/10.1056/NEJMoa060476

31. 

Welch HG, Woloshin S, Schwartz LM, et al. Overstating the evidence for lung cancer screening: the International Early Lung Cancer Action Program (I-ELCAP) study. Arch Intern Med 2007;167:2289–95. http://dx.doi.org/10.1001/archinte.167.21.2289

32. 

Reich JM. A critical appraisal of overdiagnosis: estimates of its magnitude and implications for lung cancer screening. Thorax 2008; 63:377–83. http://dx.doi.org/10.1136/thx.2007.079673

33. 

Yip R, Henschke CI, Yankelevitz DF, Smith JP. CT screening for lung cancer: alternative definitions of positive test result based on the national lung screening trial and international early lung cancer action program databases. Radiology 2014; 273:591–6. http://dx.doi.org/10.1148/radiol.14132950

34. 

Veronesi G, Maisonneuve P, Rampinelli C, et al. Computed tomography screening for lung cancer: results of ten years of annual screening and validation of cosmos prediction model. Lung Cancer 2013; 82:426–30. http://dx.doi.org/10.1016/j.lungcan.2013.08.026

35. 

Aberle DR, Berg CD, Black WC, et al. The National Lung Screening Trial: overview and study design 1. Radiology 2011; 258:243–53. http://dx.doi.org/10.1148/radiol.10091808

36. 

Aberle DR, DeMello S, Berg CD, et al. Results of the two incidence screenings in the National Lung Screening Trial. N Engl J Med 2013; 369:920–31. http://dx.doi.org/10.1056/NEJMoa1208962

37. 

Infante M, Cavuto S, Lutman FR, et al. Long-term follow-up results of the DANTE trial, a randomized study of lung cancer screening with spiral computed tomography. Am J Respir Crit Care Med 2015; 191:1166–75. http://dx.doi.org/10.1164/rccm.201408-1475OC

38. 

Wille MMW, Dirksen A, Ashraf H, et al. Results of the randomized Danish lung cancer screening trial with focus on high-risk profiling [published online ahead of print 20 October 2015]. Am J Respir Crit Care Med 2015.

39. 

Horeweg N, van Rosmalen J, Heuvelmans MA, et al. Lung cancer probability in patients with CT-detected pulmonary nodules: a prespecified analysis of data from the NELSON trial of low-dose CT screening. Lancet Oncol 2014;15:1332–41. http://dx.doi.org/10.1016/S1470-2045(14)70389-4

40. 

Horeweg N, van der Aalst CM, Vliegenthart R, et al. Volumetric computed tomography screening for lung cancer: three rounds of the NELSON trial. Eur Respir J 2013; 42:1659–67. http://dx.doi.org/10.1183/09031936.00197712

41. 

Horeweg N, Scholten ET, de Jong PA, et al. Detection of lung cancer through low-dose CT screening (NELSON): a prespecified analysis of screening test performance and interval cancers. Lancet Oncol 2014; 15:1342–50. http://dx.doi.org/10.1016/S1470-2045(14)70387-0

42. 

Baldwin DR, Duffy SW, Wald NJ, Page R, Hansell DM, Field JK. UK Lung Screen (UKLS) nodule management protocol: modelling of a single screen randomized controlled trial of low-dose CT screening for lung cancer. Thorax 2011; 66:308–13. http://dx.doi.org/10.1136/thx.2010.152066

43. 

Field JK, Duffy SW, Baldwin DR, et al. UK Lung Cancer RCT Pilot Screening Trial: baseline findings from the screening arm provide evidence for the potential implementation of lung cancer screening. Thorax 2016; 71:161–70. http://dx.doi.org/10.1136/thoraxjnl-2015-207140

44. 

Field JK, van Klaveren R, Pedersen JH, et al. European randomized lung cancer screening trials: post NLST. J Surg Oncol 2013; 108:280–6. http://dx.doi.org/10.1002/jso.23383

45. 

Marcus MW, Raji OY, Field JK. Lung cancer screening: identifying the high risk cohort. J Thorac Dis 2015; 7(Suppl. 2):S156–62.

46. 

Tammemägi MC, Church TR, Hocking WG, et al. Evaluation of the lung cancer risks at which to screen ever- and never-smokers: screening rules applied to the PLCO and NLST cohorts. PLoS Med 2014; 11:e1001764. http://dx.doi.org/10.1371/journal.pmed.1001764

47. 

Tammemägi MC, Katki HA, Hocking WG, et al. Selection criteria for lung-cancer screening. N Engl J Med 2013; 368:728–36. http://dx.doi.org/10.1056/NEJMoa1211776

48. 

Cassidy A, Myles JP, van Tongeren M, et al. The LLP risk model: an individual risk prediction model for lung cancer. Br J Cancer 2008; 98:270–6. http://dx.doi.org/10.1038/sj.bjc.6604158

49. 

Grey EP, Teare MD, Stevens J, Archer R. Risk prediction models for lung cancer: a systematic review. Clin Lung Cancer 2015. http://dx.doi.org/10.1016/j.cllc.2015.11.007

50. 

Wilson DO, Weissfeld J. A simple model for predicting lung cancer occurrence in a lung cancer screening program: The Pittsburgh Predictor. Lung Cancer 2015; 89:31–7. http://dx.doi.org/10.1016/j.lungcan.2015.03.021

51. 

Brock University. Lung cancer risk calculators. URL: http://www.brocku.ca/lung-cancer-risk-calculator (accessed 10 February 2015).

52. 

McWilliams A, Tammemagi MC, Mayo JR, et al. Probability of cancer in pulmonary nodules detected on first screening CT. N Engl J Med 2013; 369:910–9. http://dx.doi.org/10.1056/NEJMoa1214726

53. 

American Association for Thoracic Surgery. Lung cancer risk assessment tool. URL: http://aats.org/tools/lung-cancer/ (accessed 10 February 2015).

54. 

Spitz MR, Hong WK, Amos CI, et al. A risk model for prediction of lung cancer. J Natl Cancer Inst 2007; 99:715–26. http://dx.doi.org/10.1093/jnci/djk153

55. 

Hoggart C, Brennan P, Tjonneland A, et al. A risk model for lung cancer incidence. Cancer Prev Res 2012; 5:834–46. http://dx.doi.org/10.1158/1940-6207.CAPR-11-0237

56. 

The University of Michigan. Your lung cancer risk. URL: http://www.shouldiscreen.com/lung-cancer-risk-calculator/ (accessed 10 February 2015).

57. 

Roy Castle Lung Cancer Foundation. MyLungRisk (MLR). LLP MyLungRisk V2.0. URL: https://secure2.utlnet.co.uk/mylungrisk/welcome.aspx (accessed 10 February 2015).

58. 

McCunney RJ, Li J. Radiation risks in lung cancer screening programs: a comparison with nuclear industry workers and atomic bomb survivors. Chest 2014; 145:618–24. http://dx.doi.org/10.1378/chest.13-1420

59. 

Bach PB, Mirkin JN, Oliver TK, et al. Benefits and harms of CT screening for lung cancer: a systematic review. JAMA 2012; 307:2418–29. http://dx.doi.org/10.1001/jama.2012.5521

60. 

Doss M. COUNTERPOINT: should radiation dose from CT scans be a factor in patient care? No. Chest 2015;147:874–7. http://dx.doi.org/10.1378/chest.14-3022

61. 

Harris RP, Sheridan SL, Lewis CL, et al. The harms of screening: a proposed taxonomy and application to lung cancer screening. JAMA Intern Med 2014; 174:281–5. http://dx.doi.org/10.1001/jamainternmed.2013.12745

62. 

Chiles C. Lung cancer screening with low-dose computed tomography. Radiol Clin North Am 2014; 52:27–46. http://dx.doi.org/10.1016/j.rcl.2013.08.006

63. 

American College of Radiography. Lung CT Screening Reporting and Data System (Lung-RADS™). URL: http://www.acr.org/Quality-Safety/Resources/LungRADS (accessed 22 February 2016).

64. 

Jaklitsch MT, Jacobson FL, Austin JHM, et al. The American Association for Thoracic Surgery guidelines for lung cancer screening using low-dose computed tomography scans for lung cancer survivors and other high-risk groups. J Thorac Cardiovasc Surg 2012; 144:33–8. http://dx.doi.org/10.1016/j.jtcvs.2012.05.060

65. 

Humphrey L, Deffebach M, Pappas M, et al. Screening for Lung Cancer: Systematic Review to Update the U.S. Preventative Services Task Force Recommendation [Internet]. Rockville, MD: Agency for Healthcare Research and Quality (US); 2013.

66. 

Moyer VA. Screening for Lung Cancer: U.S. Preventative Services Task Force Recommendation Statement. Ann Intern Med 2014; 160:330–8. http://dx.doi.org/10.7326/M13-2771

67. 

International Early Lung Cancer Action Program. IELCAP Protocols, 2014. URL: http://www.ielcap.org/protocols (accessed 10 February 2016).

68. 

American College of Radiology. Lung-RADS™, Version 1.0. Assessment Categories. URL: http://www.acr.org/~/media/ACR/Documents/PDF/QualitySafety/Resources/LungRADS/AssessmentCategories.pdf (accessed 10 February 2016).

69. 

National Comprehensive Cancer Network. Lung Cancer Screening, Version I. 2016. URL: http://www.nccn.org/professionals/physician_gls/pdf/lung_screening.pdf (accessed 10 February 2016).

70. 

Black WC. Computed tomography screening for lung cancer in the National Lung Screening Trial: a cost-effectiveness analysis. J Thorac Imaging 2015; 30:79–87. http://dx.doi.org/10.1097/RTI.0000000000000136

71. 

Castleberry AW, Smith D, Anderson C, Rotter AJ, Grannis FW. Cost of a 5-year lung cancer survivor: symptomatic tumour identification vs proactive computed tomography screening. Br J Cancer 2009; 101:882–96. http://dx.doi.org/10.1038/sj.bjc.6605253




Add comment 





Home  Editorial Board  Search  Current Issue  Archive Issues  Announcements  Aims & Scope  About the Journal  How to Submit  Contact Us
Find out how to become a part of the HMJ  |   CLICK HERE >>
© Copyright 2012 - 2013 HMJ - HAMDAN Medical Journal. All Rights Reserved         Website Developed By Cedar Solutions INDIA