Lung cancer screening with low-dose CT

https://doi.org/10.1016/S0720-048X(02)00302-9Get rights and content

Abstract

Screening for lung cancer is hoped to reduce mortality from this common tumour, which is characterised by a dismal overall survival, relatively well defined risk groups (mainly heavy cigarette smokers and workers exposed to asbestos) and a lack of early symptoms. In the past studies using sputum cytology and chest radiography have failed to demonstrate any reduction in lung cancer mortality through screening. One of the reasons is probably the relatively poor sensitivity of both these tests in early tumours. Low radiation dose computed tomography (CT) has been shown to have a much higher sensitivity for small pulmonary nodules, which are believed to be the most common presentation of early lung cancer. As, however, small pulmonary nodules are common and most are not malignant, non-invasive diagnostic algorithms are required to correctly classify the detected lesions and avoid invasive procedures in benign nodules. Nodule density, size and the demonstration of growth at follow-up have been shown to be useful in this respect and may in the future be supplemented by contrast-enhanced CT and positron emission tomography. Based on these diagnostic algorithms preliminary studies of low-dose CT in heavy smokers have demonstrated a high proportion of asymptomatic, early, resectable cancers with good survival. As, however, several biases could explain these findings in the absence of the ultimate goal of cancer screening, i.e. mortality reduction, most researchers believe that randomised controlled trials including several 10 000 subjects are required to demonstrate a possible mortality reduction. Only then general recommendations to screen individuals at risk of lung cancer with low-dose CT should be made. It can be hoped that international cooperation will succeed in providing results as early as possible.

Introduction

The aim of this review is to describe the background for lung cancer screening with low-dose computed tomography (CT), give an overview about recent CT screening studies, analyse their results and describe concepts for future research in this field.

Section snippets

Epidemiology of lung cancer

Lung cancer is the most lethal malignant tumour worldwide with 1.3 million estimated deaths per year [1]. It currently affects 3–4 times more men than women. The probability to develop lung cancer is strongly related to certain risk factors. The most important one is cigarette smoking which is estimated to cause >85% of lung cancer cases [2]. Exposure to asbestos is another important risk factor and—although the industrial use of asbestos has been abandoned in many industrialized countries

Histologic types of lung cancer

For clinical purposes, lung cancer is classified as small-cell-lung cancer (SCLC) and non-small cell lung cancer (NSCLC: squamous cell carcinoma, adenocarcinoma, large cell carcinoma) due to the different biologic behaviour of these entities. SCLC represents approximately 20% of all lung cancer cases and is characterised by rapid growth and usually early hematogenous metastases. The different types of NSCLC represent approximately 80% of lung cancer cases and typically exhibit slower growth and

Prognosis in lung cancer

The prognosis of lung cancer patients is strongly related to tumour stage at diagnosis particularly in NSCLC.

When the tumour is diagnosed at stage IA (tumour<3 cm, surrounded by lung, no lymphatic or hematogenous metastases) prognosis is favourable with a 5-year survival of >60% [4].

However, when diagnosed at more advanced stages 5-year survival drops markedly with almost no cure at stage IV (distant metastases) [5] (Table 1).

Unfortunately, if diagnosed because of symptoms only 20–25% of NSCLC

Symptoms of lung cancer

Lung cancer confined to the pulmonary parenchyma usually does not cause symptoms. Rarely hemoptysis may be an early sign in localised tumours. In most cases symptoms are due to invasion of adjacent structures (stages III or IV), e.g. oesophagus, spine, superior sulcus, superior vena cava, other mediastinal vessels or distant metastases (stage IV), e.g. seizures due to brain metastases or pathologic fractures due to skeletal metastases. Therefore, almost all symptomatic patients present at

Localisation of early lung cancer

Lung cancer may either present as a soft-tissue or ground-glass attenuation pulmonary nodule surrounded by aerated lung or as an endobronchial lesion with or without extrabronchial component. At chest radiography the proportion of cancers presenting as pulmonary nodules and hilar masses, respectively, has been reported as equal. It is believed that in small cancers, which may only be detected at CT, pulmonary nodules surrounded by lung parenchyma are far more common than purely endobronchial

Lung cancer screening with chest radiography and sputum cytology

Considerations to use diagnostic tests for early diagnosis of lung cancer are not new as the tumour is common, prognosis is markedly better at early than at advanced stages, and there are usually no early symptoms to allow the diagnosis.

In the 1970s sputum cytology with or without chest radiography has been used in two large studies for early detection of lung cancer [7], [8], [9], [10]. They demonstrated more asymptomatic, early, resectable tumours in the screened subjects than in a control

Detection of lung cancer with computed tomography

Since its introduction into clinical routine in the early 1980s chest CT has been shown to have a much higher sensitivity for small pulmonary nodules, which are likely to be the most common presentation of early lung cancer, than chest radiography [17], [18], [19], [20].

Compared to an operative and histologic gold standard single-slice spiral CT reached a sensitivity of 95% for nodules larger than 5 mm [21].

Furthermore, significant dose reduction has been shown not to decrease sensitivity of

Lung cancer screening with CT

Several feasibility studies of low-dose CT in heavy smokers have been performed. In summary, using similar diagnostic algorithms, they all demonstrated that non-invasive classification of detected nodules based on density, size and detection of growth is possible resulting in a small proportion of invasive procedures for benign lesions [25], [26], [27], [28], [29], [30], [31], [32]. Briefly, in these studies only non-calcified pulmonary nodules were regarded as potentially malignant and biopsy

Prevalence results

The first publications on results of low-dose CT screening for lung cancer in heavy smokers described findings at baseline (prevalence) CT studies [25], [27], [29], [31], [32].

It was shown that a high proportion of asymptomatic lung cancers was found in populations at risk ranging from 0.43 and 0.48%, respectively, in two Japanese studies [25], [27] up to 2.7% in a study performed in the USA [29]. The proportion of stage I NSCLC of all detected cancers ranged from 63% [31] to 93% [25] and was

Incidence results

These results from baseline CT studies (prevalence results) were also confirmed in preliminary results at repeat CT studies (incidence results) which more likely reflect the true potential of low-dose CT screening as findings at the initial (baseline) CT are influenced by chance (study design can not control whether the first CT scan in a lung cancer patient is performed only days or rather years before symptoms occur).

Classification of nodules at follow-up CT studies was easier than at

Possible biases

The results of the studies published in the literature have shown that lung cancer screening with low-dose CT using the diagnostic algorithms described is feasible in that a high number of asymptomatic tumours was detected with a low proportion of unnecessary invasive procedures. However, they are not appropriate to conclude that low-dose CT screening can, in fact, reduce mortality from lung cancer.

Different biases could explain a high proportion of asymptomatic, small early stage lung cancers

Prospective controlled randomized trials

Although it is likely that low-dose CT screening will actually decrease mortality from lung cancer most researchers feel that unequivocal proof of lung cancer mortality reduction through screening is required before general recommendations can be made to apply this technique to the general population at risk [37]. This is particularly important in view of the enormous cost general CT screening recommendations would cause for health systems.

Also, if low-dose CT screening would not reduce lung

Conclusion

Screening for lung cancer promises reduction in mortality from this common malignancy. Low-dose CT in combination with appropriate diagnostic algorithms has been shown to detect many asymptomatic, small, early stage, resectable NSCLC with a low proportion of invasive procedures for benign lesions. Because of the huge implications of general screening recommendations definite proof of the benefit of screening by controlled randomised trials should be awaited.

References (38)

  • R.S. Fontana et al.

    Early lung cancer detection: results of the initial (prevalence) radiologic and cytologic screening in the Mayo clinic study

    Am. Rev. Resp. Dis.

    (1984)
  • J.K. Frost et al.

    Early lung cancer detection: results of the initial (prevalence) radiologic and cytologic screening in the Johns Hopkins study

    Am. Rev. Resp. Dis.

    (1984)
  • A. Kubik et al.

    Lung cancer detection: results of a randomized prospective study in Czechoslovakia

    Cancer

    (1986)
  • D.M. Eddy

    Screening for lung cancer

    Ann. Int. Med.

    (1989)
  • R.S. Fontana et al.

    Screening for lung cancer: critique of the Mayo lung project

    Cancer

    (1991)
  • P.M. Marcus et al.

    Lung cancer mortality in the Mayo lung project: impact of extended follow-up

    J. Natl. Cancer Inst.

    (2000)
  • J.R. Muhm et al.

    Lung cancer detected during a screening programme using 4-month chest radiographs

    Radiology

    (1983)
  • J.H.M. Austin et al.

    Missed bronchogenic carcinoma: radiographic findings in 27 patients with a potentially resectable lesion evident in retrospect

    Radiology

    (1992)
  • J.R. Muhm et al.

    Comparison of whole lung tomography and computed tomography for detecting pulmonary nodules

    Am. J. Roentgenol.

    (1978)
  • Cited by (45)

    • A review of patient dose and optimisation methods in adult and paediatric CT scanning

      2012, European Journal of Radiology
      Citation Excerpt :

      In the UK, the rise in the collective dose in the last 10 years can be partially attributed to the greater use of these three examinations of the trunk (chest–abdomen, abdomen–pelvis or chest–abdomen–pelvis) [4,5]. CT scanning is now frequently used for the diagnosis and management of patients with complex cardiovascular disease with a high level of detail [7,8,9], pulmonary embolism [10], as well as for screening of asymptomatic patients (primarily the heart, chest and colon, usually referred as “wholebody” screening examinations, performed only when completely justified on high risk patients e.g. smokers) [11–13] and CT guided interventional procedures (such as radiofrequency ablation, biopsies, abscess drainages, and nephrostomies) [14]. Spiral CT has also greatly improved the evaluation of children's diseases, and it is particularly useful for diagnosis and assessment of congenital heart abnormalities, lung and airway diseases, trauma and infections [15–20].

    • Interobserver-variability of lung nodule volumetry considering different segmentation algorithms and observer training levels

      2007, European Journal of Radiology
      Citation Excerpt :

      Modern radiology has to cope with an increasing number of incidentally detected subcentimeter pulmonary nodules (SCPN) of unknown dignity. One reason for this development is the introduction of multislice computed tomography (MSCT) that led to an improvement of spatial resolution in daily routine CT examinations [1,2,3,4]. Additionally, it can be expected that a growing number of undetermined SCPN is detected due to the progressive use of computer aided detection (CAD) software [5].

    View all citing articles on Scopus
    View full text