New developments in the imaging of lung cancer

Epidemiology of lung cancer

According to GLOBOCAN 2020 (global cancer statistics), it is estimated that 2.2 million new cases of lung cancer were diagnosed worldwide in the year 2020. It is the most common cancer in men (14.3% of all cancers), and the third most common cancer in women (after breast cancer and colorectal cancer) with an incidence of about 8.4% of all cancers [1]. Moreover, lung cancer is the leading cause of cancer death worldwide, with 1.8 million deaths in 2020, which was estimated as 18% of all cancer deaths (the most common cancer death in men (21%) and the second in women (13.7%)) [1]. The age-standardised incidence rates per 100 000 vary by sex and by region of the world, ranging from 2.8 in West Africa to 41.7 in Western Europe and 51.6 in Micronesia/Polynesia for men, and from 1.8 in West Africa to 25 in Western Europe and 30.1 in North America for women [1].

Introduction to the imaging methods in lung cancer diagnostics

Imaging plays a fundamental role in the diagnosis and staging of patients with lung cancer. In everyday practice, combination of several imaging studies, namely conventional chest radiography, chest computed tomography (CT), magnetic resonance imaging (MRI) and nuclear medicine techniques, mainly positron emission tomography (PET), is needed to ensure the detection, characterisation, staging and follow-up of lung cancer [2]. In addition, there is increasing evidence for the role of low-dose CT (LDCT) screening of the chest in the early detection of pulmonary malignancies, beyond the routine chest radiographs. In addition, radiomics and radiogenomics are emerging fields, reliant on artificial intelligence (AI), which offer great potential to improve diagnosis and personalised risk stratification, thus having a major impact on diagnostic and treatment pathways for lung cancer [3]. Ultrasound- and CT-guided interventions are minimally invasive and established methods for the diagnosis of pulmonary lesions. In this review, the use of these imaging techniques in lung cancer management will be discussed.

The role of chest radiography in diagnosis and follow-up

Chest radiography is the first and most commonly performed imaging method during the workup of suspected lung cancer in symptomatic patients, and it plays a role in the follow-up as well. Its advantages include its widespread availability, technical feasibility, simplicity, low radiation, low risk and low cost [4]. Routine chest radiography includes bilateral views and postero-anterior and lateral projections, with the patient in a standing position. Chest radiography is performed with the hard beam technique (120–130 kV, 5 mAs) from a focus-film distance of 2 m, with deep inhalation. Fluoroscopy is an additional imaging examination if the diagnosis is questionable on the chest radiograph. The radiation exposure of the fluoroscopy is high, its spatial resolution is poor, and its assessment is subjective and it may be insufficiently documented. An antero-posterior or a Frimann-Dahl scan taken in the side-lying position may be necessary for a patient in poor condition.

As soon as chest radiography has been performed, post-processing techniques can be used to improve image quality on digital images, giving the potential for dose reduction [5, 6]. Magnification, windowing and greyscale inversion might be helpful techniques for better identification. Other post-processing techniques include temporal subtraction and dual-energy subtraction, in order to increase the accuracy of detection of lung nodules, especially on lung cancer screening [4, 7]. A comparison of conventional and digital chest radiography is shown in table 1.

Chest radiography has not been shown to decrease mortality when used for lung cancer screening, and it is less effective compared to LDCT, although chest radiography is one of the most commonly utilised diagnostic tools for chest diseases in clinical practice [2]. The sensitivity and specificity of chest radiography for the detection of lung cancer are 78.3% and 97.0%, respectively [8], whereas for LDCT the sensitivity and specificity are 87.7% and 99.3%, respectively [2]. Chest radiography only detects lung cancer in ∼77–80% of cases in the year before diagnosis [9]. Therefore, LDCT is used as the primary method of cancer screening, and CT is subsequently performed if abnormalities are detected on chest radiographs.

Solitary pulmonary nodules (SPNs) are increasingly detected with the widespread use of chest radiography and chest CT scans. According to the National Institute for Health and Care Excellence (NICE) guidance, chest radiography is recommended for several symptoms in patients aged >40 years [10]. Findings on chest radiography that are suspicious for lung cancer warrant an urgent referral within 28 days, and additional workup [10].

The “blind zones” on postero-anterior chest radiographs (apices, hila, retro-cardiac and sub-diaphragmatic spaces) and the mediastinal lines and stripes (lines) should always be checked carefully, since subtle parenchymal nodules can be overlooked here due to overlying and distracting anatomical structures (figure 1). Computer-aided diagnosis (CAD) chest radiography and AI-based software have improved the detection of actionable lung nodules on chest radiographs [11, 12]. The management of patients with SPNs must include estimation of the probability of cancer-suspicious lesions, based on the patient's clinical risk factors and CT characteristics.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Images of a 73-year-old female patient admitted to hospital for secondary hypertension workup and management. a) A routine chest radiograph showed a right basal pulmonary nodule, which was misdiagnosed as the right nipple, but b) computed tomography (CT) and c) CT-guided biopsy confirmed the presence of a carcinoid tumour. In addition, a) chest radiography and CT (not shown) confirmed an aberrant right subclavian artery, known clinically as arteria lusoria, which is the most common embryological abnormality of the aortic arch. Images courtesy of the Medical Imaging Centre, Semmelweis University, Budapest, Hungary.

Lung tumours may present as central or peripheral masses. The central neoplasm may have hilar lymph node enlargement, mediastinal invasion or bronchial obstruction, with partial or total lung collapse; a parenchymal consolidation or super-infection may exist, which may mask or be the first sign of a possible underlying neoplasm. Retrospective comparison with previous chest radiographs is helpful to determine if a lesion has changed its size and appearance. Unchanged size for 2 years is a sign of benignity.

Prior to CT imaging, chest radiography may play roles in assessing lesion size and demonstrating features of local (for instance bony) invasion, as well as assessing post-obstructive collapse and pleural effusion. Chest radiography can demonstrate obvious chest wall and mediastinal invasion in large tumours; however, it has a limited sensitivity to predict T3 and T4 disease [13]. Accordingly, chest radiography cannot detect invasion of the chest wall, diaphragm or mediastinum, or nodal involvement [14].

Limitations of chest radiography must be emphasised. 45–81% of missed lung cancers occurred in the upper parts of the lung [15]. The detection rate is 29% for tumours measuring a diameter ≤10 mm [16]. A 2019 review of 21 studies reported that 20–23% of chest radiographs in patients with lung cancer symptoms were falsely negative for lung cancer [9].

In summary, chest radiography remains the first-line investigation for lung cancer in primary care and the most common radiological route to diagnosis. NICE lung cancer guidelines recommend chest radiography for initial evaluation in all patients, aside from those aged >40 years who have unexplained haemoptysis [10].

The role of CT in diagnosis and follow-up

CT is the most important imaging modality in the imaging of lung cancer. When the chest radiograph raises the suspicion for malignancy, CT with contrast should be performed for complete staging. Administration of intravenous contrast is necessary for the evaluation of contrast enhancement in the lesion, and for the better differentiation of the surrounding tissues (e.g. in the case of atelectasis), mediastinal and hilar regions, especially the lymph nodes. Multidetector CT (MDCT) scanners are capable of examining the whole lung within a single breath-hold, which increases the accuracy of the contrast material administration. Additionally, MDCT takes thin slices, which improves the spatial resolution of multiplanar reconstruction images and increases the quality of maximum intensity projection images and 3-dimensional (3D) reconstructions [2]. The CT parameters are most commonly 120 kVp, 20–80 mAs and 3–5-mm slice thickness.

64-slice to 128-slice CT scanners generally offer shorter examination time than the previous systems, such as 16-slice CT scanners (table 2), which further increases the advantages of MDCT, for example by reducing motion artefacts [17]. The comparison of dose efficiency of scanners with different numbers of slices is an interesting question in the dosimetry of radiology. A study compared 4-, 8-, 16- and 64-slice scanners by measuring dose in anthropomorphic phantoms: they concluded that the organ dose in the middle of the scan was the same, while organs at the ends of the scan received a higher dose in the case of the 64-slice scanner, because of the larger over-ranging [18]. Another study compared a prototype 256-slice scanner with a 16-slice commercial CT: in the case of a chest examination they found that the 256-slice scanner used only 72.1% of the dose used by the traditional one [19]. A third study found that MDCT with a larger number of slices uses a higher dose but delivers better image quality scores in the case of chest examinations [20]. Several international radiology societies have published guidelines on the CT requirements for lung cancer screening; for example, the American College of Radiology (ACR) recommends using at least 16-slice MDCT, whereas the European Society of Thoracic Imaging (ESTI) sets the lower limit at 32-slice and recommends 64-slice MDCT [21].

The slice thickness, considering the radiation dose, is 1.0-mm slice thickness or smaller, with the LDCT protocol for lung cancer screening [22]. For lung cancer staging and pulmonary nodule assessment, thin-section axial chest CT data with slice thickness of 1–1.25 mm are recommended [23–26]. This slice count category is a bit more expensive than those with a lower number of slices.

According to the Fleischner Society's recommendations, initial CT for the assessment of SPNs should be intravenously contrast enhanced to exclude additional intrathoracic abnormalities, and subsequent surveillance CT scans should be non-contrast, thin section and ideally low radiation dose when follow-up of a pulmonary nodule is the only indication for CT [27, 28]. Intravenous contrast administration helps to differentiate the tumour from the surrounding atelectasis and to characterise the necrotic parts of the cancer.

Multiplanar reconstructions create multiplanar reformation images of the malignancy, making the assessment of its relationship to surrounding structures (including fissures and the diaphragm) possible [2, 29]. Maximum intensity projection images increase the detection rate of small pulmonary nodules, especially in the central lung areas, which helps to reduce the number of overlooked small cancers [2, 30]. 3D reconstructions also help the assessment of the lung cancer [2, 31] by improving spatial orientation and can identify surgically relevant structures and their interindividual variance [32]. Virtual bronchoscopy, a CT-based internal 3D rendering technique of the airways, allows a noninvasive intraluminal evaluation of the tracheobronchial tree for the evaluation of cancer-related airway stenosis [2].

The CT scan of the primary lung cancer may demonstrate a wide spectrum of appearances. Nonsmall cell lung cancers (NSCLCs) can be either centrally or peripherally located masses, invading the mediastinal structures and the chest wall, which usually demonstrate a solid appearance as a solitary nodule or a consolidation-like mass (>3 cm) [13]. Tumour margins can be smooth, lobulated, irregular and spiculated. Squamous cell carcinomas tend to be centrally located, might contain central necrosis or cavitation, and are associated with post-obstruction atelectasis, inflammation and infection. Adenocarcinomas, usually the in situ forms, tend to have a peripheral location; therefore, these lesions are associated with less obstruction, inflammation and necrosis, and usually demonstrate a ground-glass or part-solid appearance (figure 2) [33]. Larger adenocarcinomas may demonstrate a consolidation with air bronchogram. Small cell lung cancers are often centrally located, show a rapid growth with invasion of the mediastinal and hilar structures, causing bronchial stenosis/obstruction and nodal involvement [34]. Most cases of peripheral small cell lung cancer manifest as a lobulated mass rather than a spiculated mass [34]. Finally, carcinoid tumours usually appear as a solid, lobulated nodule [35].

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2

Computed tomography (CT) scans of a growing part-solid nodule. a) In 2014. b) Axial and c) coronal views from 3 years later, in 2017. The 70-year-old female smoker patient was investigated due to cough. Positron emission tomography/CT did not show fluorodeoxyglucose uptake in the lesion. Biopsy confirmed adenocarcinoma. Images courtesy of the Medical Imaging Centre, Semmelweis University, Budapest, Hungary.

Malignant SPNs have been shown to have significantly higher mean attenuations of enhancement (<15 Hounsfield units (HU): benign; >15 HU: malignant) and higher washout on a 15-min delay series than all benign SPNs [36]. Nodule detection can be improved by CAD and AI-based programs, especially for centrally located nodules compared to peripheral ones [37, 38]. In cases of suspected lung cancer, bronchoscopic or image-guided biopsy is necessary to obtain histological diagnosis [13].

For staging, currently the eighth edition of the TNM classification for lung cancer staging is recommended by the International Association for the Study of Lung Cancer (IASLC) [39]. In recent years, new patient data have been collected to develop the ninth edition, which is expected to be completed in 2024 [40, 41]. Briefly, the analysis of the T component is based on the tumour size, endobronchial location, atelectasis/pneumonitis, and invasion of the diaphragm [41]. Tumour size must be measured on lung window images without measuring the spiculations [23, 27]. For part-solid nodules, suggestive of adenocarcinoma, measuring size involves measuring the longest diameter of the solid part, as well as the non-solid part beyond it on the CT scan, to determine the overall size of the tumour [23]. CT scans play an important role in T-staging, including showing proximity of the tumour to the main pulmonary artery or its involvement, involvement of the fissures, endobronchial tumour involvement and its distance from the carina, invasion of the chest wall, and invasion (lysis/destruction or definite sclerosis) of ribs and chest wall musculature as well as the mediastinum [13]. Moreover, CT with intravenous contrast is essential in N- and M-staging. Mediastinal and hilar nodal disease is evaluated using CT scans with intravenous contrast and, more recently, PET/CT has been used for its evaluation, due to its higher sensitivity and specificity [13]. A lymph node with short axis diameter of >1 cm is generally considered to be malignant. However, metastatic lymph nodes (usually between 4 and 10 mm) can be more sensitively analysed with PET/CT than with CT [42].

Intra- and extrathoracic metastases can be detected on CT scans with a degree of certainty, including pleural and pericardial effusions and metastatic lung nodules. Adrenal and liver metastases are common in lung cancer; therefore, it is recommended that the scan coverage should include the upper abdomen [13].

Follow-up CT scans help in the monitoring of the response to treatment, in the detection of recurrent tumours and in the planning of radiation therapy [2]. During the follow-up of treated/resected NSCLC patients, the use of thoracic CT scans enables the detection of more cases of early recurrence and second primary lung cancer, which are more amenable to curative-intent treatment, according to the IFCT-0302 study, a prospective randomised trial, which is a major study and the largest in this field [43].

In recent decades, tremendous improvements in scan speed, volume coverage, spatial and soft tissue resolution and dose reduction have been achieved, mainly involving tube current modulation, automated exposure control, anatomy-based tube voltage (kV) selection, advanced X-ray beam filtration, and iterative image reconstruction techniques, which improved image quality and decreased radiation exposure [44]. The emergence of MDCT and use of advanced iterative reconstruction algorithms have enabled volumetric lung images to be achieved at high spatial resolution and low radiation doses. In addition, AI is increasingly used in patient positioning, protocol adjustment, image reconstruction, image pre-processing or post-processing [44].

In 2021, the first clinical photon-counting detector system, referred to as photon-counting CT (PCCT), was approved, which provides ultrahigh resolution (0.2 mm and 0.4 mm) images and is considered an innovation in CT technology [45, 46]. PCCT relies on the direct conversion of incoming photons into electrical charges that are subsequently transferred into a current that is proportional to the energy of the incoming photons (energy-integrated detectors). Due to the absence of electronic noise, improved radiation dose efficiency, increased iodine signal, lower doses of iodinated contrast material administration, and better spatial resolution can be achieved. In addition, spectral imaging can be performed and quantitative functional data such as lung perfusion data can be obtained [47]. PCCT also allows K-edge imaging, which, combined with new contrast materials specifically designed for PCCT, opens the door to new applications for imaging in the future [47].

The role of MRI in diagnosis and follow-up

MRI of the chest is still considered a supplementary imaging tool in cases of lung cancer, used for the determination of tumour invasion in the chest wall and the mediastinal structures (especially in cases of pancoast tumour), differentiation between solid and vascular hilar masses (using ECG-triggered magnetic resonance angiography to reduce cardiac motion- and breathing-related artefacts in pulmonary vessels), assessment of diaphragmatic abnormalities and in cases of mediastinal lymphoma (table 3) [2].

It is not possible using MRI to assess the airway and lung parenchyma, due to low proton density and artefacts due to respiratory and cardiac motion [48]. However, with recent advances related to MRI pulse sequences, post-processing software and analysis methods, multi-coil parallel imaging and acceleration methods, and the administration of contrast media, pulmonary MRI has become a considerable imaging modality in lung cancer [49, 50]. On the one hand, higher lesion contrast can be expected at 3 T compared to 1.5 T machines [49]. On the other hand, the performance of low-field (0.55 T) MRI systems equipped with high-performance image technology can reduce distortion by reducing susceptibility and deliver excellent image quality because of improved field homogeneity (figure 3) [49, 51]. Due to their lack of signal, the blood vessels are well separated from the soft tissues and tumours. Therefore, MRI is more sensitive in terms of vascular infiltration.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3

Low-field magnetic resonance imaging (0.55 T) images (T2 FBLADE sequence). a) Axial and b) coronal views of a metastatic lung nodule in the right upper lobe (S3 segment), and c) axial view of another metastatic pulmonary lesion in the right upper lobe and a growing mediastinal lymph node, in a young female pregnant patient with metastatic breast cancer. Images courtesy of the National Institute of Oncology, Budapest, Hungary.

Whole-body MRI may detect distant metastatic disease [2, 52]. Diffusion weighted imaging (DWI) is a powerful imaging tool that provides unique information related to tumour cellularity and the integrity of the cellular membrane; therefore, it can be applied for tumour detection, characterisation of different subtypes (based on ADC values) and N- and M-stage evaluation, as well as for the monitoring of response to treatment [2, 53, 54]. Dynamic contrast-enhanced (DCE) MRI indexes can differentiate potentially malignant and benign SPNs [55].

Several studies have demonstrated that lung MRI could be a potentially effective screening tool, with a performance comparable to that of LDCT, but with a lower false-positive rate and no radiation exposure [49, 56]. MRI can differentiate lung cancer from progressive massive fibrosis, tuberculomas and obstructive pneumonia as well as from surrounding post-obstructive atelectasis [49]. In addition, MRI has a high diagnostic performance for N-staging in NSCLC using short tau inversion recovery (STIR) turbo-spin echo (TSE) sequence and DWI [57, 58], and screening for brain metastases by contrast-enhanced MRI is useful in patients with clinical stage III lung cancer considered for curative therapy [59].

Recently, various MRI techniques, for instance 3D DCE-perfusion sequence, have been shown to be useful in the prediction of post-operative lung function in lung cancer patients with underlying COPD [49, 60]. Additionally, magnetic resonance lymphangiography has been shown to be promising in the detection of post-operative chylothorax after pneumonectomy and lobectomy [49]. Finally, MRI might play a role in treatment response evaluation with DWI sequence, such as in differentiating between radiation pneumonitis and true progression [49, 61].

Despite the developments in MRI technique described here, access to MRI is a major issue for patients and clinicians even in European countries, greatly restraining its involvement in lung cancer patient management except in specific cases such as lung apex cancer, involvement of vertebra (and spine), or brain MRI for staging.

The role of nuclear medicine in diagnosis and follow-up

Nuclear imaging techniques visualise molecular processes in vivo. Together with radiological imaging, multimodal functional and morphological imaging addresses two main topics in respiratory medicine: the individualised workup of lung cancer and pulmonary thromboembolic diseases. In lung cancer, PET with integrated CT currently serves as a first-line staging tool, provides essential information for radiotherapy planning, and monitors response to therapy [62–64].

An integral part of nuclear medical imaging is the radiotracer, which consists of a chemical compound and a radionuclide. Since malignant tumours in particular express glucose transporters (GLUT 1 and 3) on their outer cell membranes, glucose serves routinely as a suitable chemical compound in proliferating cells. Increased glucose uptake is based on the Warburg effect, describing that proliferating neoplastic cells increasingly use cytoplasmic glycolysis for nucleic acid and lipid biosynthesis instead of mitochondrial glucose oxidation, despite the presence of oxygen [65]. Labelling with the positron emitter fluorine-18 (¹⁸F) produces fluorodeoxyglucose (FDG), which is used as a radiotracer for temporal and spatial imaging of glucose metabolism. FDG cannot be further metabolised after phosphorylation and accumulates especially in tumour tissue (known as “metabolic trapping”). The semi-quantitative assessment of cellular FDG uptake from a target region is usually carried out by calculating the standardised uptake value (SUV), measuring the uptake in a tumour normalised to the distribution volume. An SUV mean cutoff >2.5 has been used to differentiate between benign and malignant nodules [66]. However, other factors, such as the time of SUV evaluation, the shape of the region of interest (ROI), attenuation correction and body size, influence the calculation of SUV [67]. In addition, the SUV is also influenced by biological factors of the pulmonary nodules, e.g. slowly growing and well-differentiated tumours generally have lower SUVs than rapidly growing and undifferentiated ones [68]. Differential diagnoses of FDG uptake by means of PET/CT have an impact on further workup, which is why knowledge of false-positive and false-negative findings is essential. Table 4 gives an overview of the differential diagnoses of cellular FDG uptake.

False-positive findings have been reported in infectious diseases (mycobacterial, fungal and bacterial infection), sarcoidosis, radiation pneumonitis and post-operative surgical conditions, while tumours with low glycolytic activity such as adenomas, low-grade and in situ adenocarcinomas, carcinoid tumours, low-grade lymphomas and small-sized tumours (SPNs <8 mm) have revealed false-negative findings on PET scans [71]. Immune checkpoint inhibitors can upregulate the immune pathways, resulting in a wide variety of adverse effects, such as reactive adenopathy with increased FDG uptake. In addition, pseudoprogression has been identified in melanoma patients receiving ipilimumab therapy. Therefore, utilisation of iRECIST guidelines (Response Evaluation Criteria in Solid Tumours in cancer Immunotherapy trials) when reporting studies that demonstrate possible progression, and careful review of all components of the PET/CT, prior imaging and the patient's clinical history, are essential to avoid misinterpretation [72]. Moreover, SUV, which is normalised for body weight, overestimates metabolic activity, and this overestimation is more significant in obese patients than in patients with a normal body mass index. In this respect, SUV normalised by lean body mass is not affected by body weight and appears to produce more accurate SUV results [73].

In clinical routine, the basic workup of suspicious pulmonary nodules or masses should include contrast-enhanced CT of the thorax and the abdominal region. CT visualises localisation and radiological characteristics of suspicious morphology and assesses the therapeutic effect of treatment [74, 75]. However, information about tumour vitality and metabolism as well as on the mediastinal lymph nodes is not sufficiently provided with CT as the diagnostic tool. Multimodal FDG-PET/CT makes it possible to co-register glucose metabolism (as a surrogate parameter for tumour vitality of a nodule suspicious for lung cancer) together with morphology [69]. The T descriptor is usually reliably detected by CT, including accurate determination of the maximum longitudinal diameter of the target lesion and infiltration of pleura, vessels, mediastinum, bone or pericardium. Hybrid FDG-PET/CT uses imaging of metabolic tumour extension to better estimate infiltration into surrounding tissues, metabolically active pleural effusion, and tumour size in atelectasis, than is possible with CT alone (figure 4). PET/CT is also superior to CT alone in sensitivity and specificity for determining the N descriptor, and can guide the histological workup of mediastinal lymph nodes suspicious for malignancy by endobronchial ultrasound (EBUS) or video-assisted mediastinoscopy.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4

a) A fluorine-18 fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography (PET)/computed tomography (CT) image in three sectional planes of an 82-year-old smoker. In the left upper lobe, a pulmonary mass with a hypermetabolic border (maximum standardised uptake value 13.6) and a necrotic centre comes into view. Staging remained without evidence of mediastinal or extrathoracic metastasis. b) The maximum intensity projection (MIP) shows an overview of the entire PET scan. Information about the distribution of FDG uptake assisted CT-guided puncture, which was taken from the marginal area of the pulmonary mass and confirmed a pulmonary squamous cell carcinoma.

FDG-PET/CT provides helpful information about both thoracic (local) and extrathoracic (distant metastatic) lesions. The detection of extrathoracic lymph node metastases, adrenal, bone and bone marrow metastases on PET/CT can guide further tissue workup (such as image-guided biopsy). During radiotherapy planning, FDG-PET complements CT imaging to narrow down the radiation volumes of hypermetabolic tumour tissue (e.g. mediastinal lymph nodes). PET is also suitable for assessing therapy response by showing metabolic activity in the target lesions after local (radiotherapy) and/or systemic therapy. In addition, it is used during multidisciplinary team discussion to assess the course of suspicious CT findings in terms of progress or recurrence of disease [76].

PET tracers other than ¹⁸F-FDG may play a potential role in imaging tumour biology and heterogeneity by assessing hypoxia, proliferation, and vasculature. Neuroendocrine tumours of the lung, such as carcinoids, large cell neuroendocrine and small cell carcinomas, show both neuroendocrine morphology (organoid or palisade-like structures) and neuroendocrine differentiation (chromogranin A, synaptophysin, etc.), depending on the degree of differentiation. They can express membrane-bound somatostatin receptors (SSTRs) 1–5 on the cell surface, which indicate the presence of a neuroendocrine tumour in tumour-specific structures. For staging of neuroendocrine tumours, assessment of SSTRs by means of gallium-68 (⁶⁸Ga) dodecane tetraacetic acid (DOTA) scan is an appropriate imaging modality [77]. Here, DOTA acts as a complexing (chelating) agent for the radionuclide ⁶⁸Ga. ⁶⁸Ga-DOTA complex is conjugated with peptides, such as octreotide and octreotate (DOTATOC or DOTATATE, respectively). These ⁶⁸Ga-DOTA-conjugated peptides are used as tracers in PET/CT imaging for the identification of neuroendocrine tumours and quantification of specific tumour metabolism. Figure 5 depicts a case study in which SSTR-specific tracers and PET/CT imaging were applied for the diagnosis of a neuroendocrine tumour of the lung.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5

Positron emission tomography (PET)/computed tomography (CT) of a 56-year-old smoker who presented with a chronic dry cough. In the left lower lobe of the lung, a) CT showed a solid compaction (23 mm) close to the hilus, which in b) fluorine-18 fluorodeoxyglucose (¹⁸F-FDG) PET/CT showed only a slightly increased uptake (maximum standardised uptake value (SUV_max) 3.2) and was not considered to be malignant. c) Somatostatin receptor expression was significantly increased in the gallium-68 dodecane tetraacetic acid-conjugated octreotide (⁶⁸Ga-DOTATOC) PET/CT at this site (SUV_max 43). Left lower lobe resection revealed a typical carcinoid.

The most commonly used hypoxia tracers are ¹⁸F-FMISO (fluoromisonidazole), ¹⁸F-HX4 (flortanidazole) and ¹⁸F-FAZA (fluoroazomycin arabinoside). ¹⁸F-FLT (fluorothymidine) can be used to study proliferation. Because hypoxia is a marker of radioresistance and areas of high proliferation, which are presumably more aggressive, such tumour areas may require dose escalation [78–85]. A prospective, phase II multicentre dose escalation study using ¹⁸F-FMISO in NSCLC patients showed the feasibility of increasing the dose up to 86 Gy in hypoxic subvolumes. After 3 years of follow-up, the authors found that in ¹⁸F-FMISO-positive patients, radiation therapy escalation increased median overall survival by 11.2 months [86, 87].

Using ⁶⁸Ga-pentixafor PET/CT, it is possible to quantify the CXCR4 receptor in vivo in different lung cancer subtypes, as well as to examine the correlation of quantity of CXCR4 receptors with tissue density by immunochemical analyses. ⁶⁸Ga-pentixafor PET/CT proved to be highly sensitive and specific for in vivo targeting of CXCR4 receptors in lung cancer, so it can be effectively used to assess response and develop CXCR4-based radioligand therapies [88].

In conclusion, results from nuclear imaging techniques give critical direction to prognosis and treatment planning. A multimodal nuclear medical imaging technique, PET/CT forms a mainstay in the workup of thoracic tumour diseases and makes an important contribution to individualised therapy in patients with lung cancer.

Lung cancer screening

Throughout Europe, LDCT screening for lung cancer is in a phase of clinical implementation. In the USA, the National Lung Screening Trial demonstrated that LDCT screening leads to a 20% reduction in lung cancer-associated mortality in high-risk populations compared to that seen with chest radiography [89]. The US Preventive Services Task Force recommends annual screening for lung cancer with LDCT in adults aged 50–80 years who have a 20-pack-year smoking history and currently smoke or have quit within the past 15 years [90]. LDCT for lung cancer screening in high-risk subjects has started to be used in routine clinical practice in some countries, since LDCT screening may facilitate early diagnosis and thus curative-intent treatment in lung cancer (figure 6) [91]. The LDCT-based Lung CT Screening Reporting and Data System (Lung-RADS) is widely used for nodule assessment and subsequent management [92].

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6

Low-dose computed tomography screening of a 77-year-old female smoker, which confirmed a tumour-suspicious speculated solid lesion in the right upper lobe. Image courtesy of the National Institute of Oncology, Budapest, Hungary.

The drawbacks of LDCT screening include the number of unnecessary invasive procedures or complications due to the high malignancy-related false-positive rate (23%) [89], which can be decreased by appropriate nodule classification using deep learning methods and radiomics [3, 93]. The longer-term COSMOS study demonstrated an intermediate benefit-to-radiation-risk ratio of approximately 23:1, assuming 20% mortality benefit [94]. In LDCT trials, volume CT dose index ranged from 0.8 to 3.81 mGy depending on the participant's body weight, with an average effective dose estimate of ≤2 mSv per CT [95]. The sensitivity and specificity of LDCT, in successfully diagnosing early-stage lung cancers were 87.7% and 99.3%, respectively [2].

The American College of Radiology Appropriateness Criteria have been recently updated, with publication in November 2023, reviewing the available literature on various imaging modalities and summarising management of incidentally detected indeterminate pulmonary nodules [96]. In the past decade, ultra-low-dose CT (ULDCT) has been gaining more attention, offering a sub-millisievert radiation dose (approximately equal to two chest radiography sessions) [97]. Although limitations, such as lack of former human experiences, hamper the comprehensive understanding of technology applicability to the clinical setting, ULDCT by spectral shaping seems a promising technique for the detection and characterisation of SPNs, with an excellent agreement with LDCT, as well as a feasible approach for lung cancer screening [98]. However, the role of ULDCT in the diagnosis and workup remains unsolved. The latest articles suggest that ULDCT can be utilised for screening and follow-up studies of primary or secondary lung cancers for the detection or exclusion of solid pulmonary nodules of size >3 mm and monitoring of the number of >3-mm-sized solid pulmonary nodules [99].

Pulmonary nodules and their management

In chest imaging, a pulmonary nodule is defined as a rounded or oval opacity measuring <3 cm in diameter. Small pulmonary nodules are defined as lesions <8 mm in diameter, while micronodules are diagnosed if the diameter is <3 mm [100, 101]. Pulmonary nodules can be classified, according to their composition, as solid or sub-solid. Sub-solid nodules can be subdivided into pure ground-glass opacities or part-solid lesions that contain both ground-glass and soft tissue elements. Pulmonary nodules are commonly found during routine radiological examinations or during lung cancer screening with LDCT scanning. They are identified in 0.2–2% of chest radiographs and in 10–40% of CT scans [101–104]. The prevalence of malignancies among pulmonary nodules diagnosed during lung cancer screening programmes is about 0.5–3.5% [101, 102].

The differential diagnosis of pulmonary nodules includes neoplasms (benign or malignant) and lesions caused by infection, inflammation, vascular abnormalities or congenital malformations. The majority of pulmonary nodules are benign; however, a few can be malignant. Differentiation of pulmonary nodule character is an important clinical problem, as malignant nodules need to be removed without delay while benign lesions only need to be monitored [101–104].

Chest imaging and clinical characteristics of the patient are key factors in predicting the probability of malignancy in pulmonary nodules. The most frequent independent clinical predictors of pulmonary nodule malignancy are older age and relevant smoking history. Chest CT without contrast is the most appropriate imaging tool in the evaluation of incidentally detected indeterminate pulmonary nodules identified on chest radiographs or CT [96]. Radiological features associated with increased risk of malignancy in pulmonary nodules include large nodule diameter and volume, spiculated margins, upper lobe location and increase in diameter or volume during follow-up. Predictors of benign character of pulmonary nodules, in turn, are some calcification patterns (diffuse, central, laminar or popcorn), perifissural or subpleural location and stability of nodule size in follow-up or previous scans [101–104].

There are a few clinical models for predicting the probability of malignant character of pulmonary nodules. Among them, the most commonly used are three models based on clinical and CT characteristics, two of which also use PET with ¹⁸F-FDG: the Mayo Clinic, Brock University and Herder models [101, 105]. Comparison of these models is presented in table 5. Separate prediction models are dedicated to pulmonary nodules detected during LDCT (PanCan model, Lung-RADS and National Comprehensive Cancer Network (NCCN) guidelines) [106, 107]. An additional criterion used in these models to assess malignant character of the nodule is the volume doubling time, which is particularly useful in small pulmonary nodules (<8 mm in diameter) [101, 102].

When the character of a pulmonary nodule is indeterminate despite chest CT, additional imaging techniques may help differentiate malignant versus benign aetiology. Among them, FDG-PET seems to be the most accurate method, with sensitivity >95% and specificity of 82–85% [101, 102, 108, 109]. For a long time, MRI has not been used for lungs, but recent progress in MRI technology now allows for its implementation in lung imaging, including analysis of pulmonary nodules. In comparison to CT, MRI enables accurate assessment of nodule diameter in those >8 mm [110]. Thus, sometimes MRI may be considered an alternative to CT for follow-up of some lung lesions, especially as it is a radiation-free procedure. Recent developments in MRI sequences for the lung, implementing ultrashort echo time on a 3 T system, proved that MRI may reach similar sensitivity and specificity to CT in diagnosing all types of pulmonary nodules (solid, sub-solid and ground-glass), although MRI underestimated pulmonary nodule diameter by approximately 1–2 mm [111].

The management of a pulmonary nodule depends on its type (solid versus part-solid versus pure ground-glass nodule) and its probability of malignancy. Two algorithms for management are presented in figure 7 [27, 102].

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7

a) Algorithm for management of solid pulmonary nodules (PNs) based on British Thoracic Society guidelines. Reproduced and modified from [102] with permission. b) Algorithm for management of sub-solid PNs based on Fleischner Society guidelines. Information from [27]. PET: positron emission tomography; CT: computed tomography; GGO: ground-glass opacity; PSN: part-solid nodules.

Evaluating complications of treatment in lung cancer

In recent years, immunological checkpoint inhibitors and molecularly targeted drugs have been added to the lung cancer treatment arsenal, in addition to chemotherapy. Immune checkpoint inhibitors are active due to the activation of the immune system to recognise and destroy tumour cells. The most common side-effect associated with immunotherapy is immune-related hypothyroidism (20%). The incidence of other side-effects is much lower: in clinical trials, immune-related adverse events occurred in <5% of patients; the most important among them were pneumonia, inflammatory bowel disease, hepatitis and cardiac toxicities [112]. However, the incidence of pneumonitis among lung cancer patients in daily practice can reach 20% and even 30% in patients after chest radiotherapy, with additional morphological features in this group of patients [113, 114]. Radiological features of pneumonitis are classified into five subtypes: cryptogenic organising pneumonia (19%), ground-glass opacities (37%), interstitial (7%), hypersensitivity (22%) and pneumonitis not otherwise specified (15%) [115]. Pneumonia after immune checkpoint inhibitors is more likely to be bilateral, involving multiple lobes and without sharp borders. In patients who receive both radiotherapy and immune checkpoint inhibitors, lesions may be confined to the lung, with acute borders similar to those seen in radiotherapy-associated pneumonia, or they may be found in a more extensive area [114].

Pulmonary complications are relatively frequent among those taking additional anti-cancer drug treatments, which can be difficult to differentiate from progressive cancer, infection, cardiopulmonary disease and other causes. Radiation therapy can be commonly associated with lung injury (pneumonitis and/or fibrosis) with a strong relationship between radiation dose and volume [116]. The differential diagnosis for pneumonia or lung cancer progression must be based on detailed information from the clinical history. It should be noted that similar interstitial changes can also occur as a complication of treatment with tyrosine kinase inhibitors (brigatinib or osimertinib). They then need to be differentiated, especially as the underlying disease progresses [117].

Radiomics and lung cancer

Radiomics/radiogenomics is an emerging field, reliant on AI to aggregate radiology and genomics data, with the aim of predicting the macroscopic and molecular properties of tissues. In the near future, it is likely to have a major impact on diagnostic and treatment pathways for lung cancer. The pipeline for these biomarkers involves extraction of quantitative imaging features, largely from CT and PET/CT, in order to develop models capable of predicting tumour phenotypes and outcomes [118]. This requires 1) identification and segmentation of a target ROI, e.g. a nodule or proven cancer, 2) ROI feature extraction, and 3) association of these characteristics with relevant outputs [119], typically requiring deep learning AI. The ability to predict features that traditionally require invasive sampling has considerable and obvious potential for earlier and more refined personalised treatment. In NSCLC, radiomics have been studied in contexts such as cancer screening, early detection/staging [120], prediction of radiotherapy and immunotherapy response [121, 122] and prognostication [120]. The emerging evidence for neoadjuvant therapies in patients with early-stage NSCLC seems set to accelerate developments in this area further.

Several groups have recently demonstrated successful linkage of radiomic features to genetic and molecular alterations in lung cancer (“radiogenomic biomarkers”). In 2022, Wang et al. [123] reported a CT-based model for epidermal growth factor receptor status, with a receiver operator area under the curve (AUC) of 0.748–0.813. Zhang et al. [124] and Shao et al. [125] independently demonstrated the ability of a similar model to discriminate between a broader range of genotypes, with AUC values >0.85 for both eight- and 10-gene panel features.

Despite this potential, radiogenomic research also faces several challenges, including the need to standardise study methods such as image acquisition and feature extraction protocols. Perhaps the biggest challenge lies in data access and the successful colocation of deidentified imaging, genomic and clinical data with the supercomputing resource needed for deep learning. Given the challenges involved in moving large datasets around, solutions are likely to be cloud-based, mandating creation of trusted research environments with the necessary governance in place. Nevertheless, the potential impact of radiogenomic research on lung cancer practice is considerable.

Interventional radiology in lung cancer treatment

CT-guided biopsy

While central lesions in the lung parenchyma and mediastinum suspicious for malignancy are ideally targeted by bronchoscopy or EBUS transbronchial needle aspiration, peripheral pulmonary lesions can be sampled with image-guided assistance (figure 8) [135]. Image-guided biopsy can also help evaluate large mediastinal masses [101]. Image guidance can include ultrasonography or CT. CT-guided transthoracic lung biopsy is performed as needle aspiration or cutting needle biopsy [136]. Both types of CT-guided sampling show greater sensitivity in confirming lung cancer diagnosis than using transthoracic biopsy with fluoroscopy guidance [137]. Direct comparisons between aspiration cytology and cutting needle biopsy in confirming histological diagnosis have revealed that transthoracic needle core biopsy compared with needle aspiration showed similar sensitivity for malignancy (86–98% versus 92–98%) [66]. A more recent comparison showed that cutting needle biopsy improved the diagnostic accuracy (98.9% versus 93.8%) of pulmonary lesions [138]. Cutting needle biopsy is more likely to yield enough tissue for mutation analysis [101]. This diagnostic advantage for cutting needle biopsy has not been without increased rate of complications, including pneumothorax (20.6% versus 13.1%), haemorrhage (32.2% versus 13.1%) and haemoptysis (8.2% versus 3.3%) [138]. Additional guidance to improve biopsy accuracy can be provided by assessing tumour metabolism through FDG-PET/CT imaging (figure 4).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8

Computed tomography (CT)-guided transthoracic lung biopsy performed in the 82-year-old lung cancer patient shown in figure 4. Biopsy location was additionally guided by positron emission tomography/CT. In the case of this central tumour far from the chest wall, bronchoscopic tissue sampling would also have been feasible.

Emphysema is a predictor of pneumothorax and chest drainage after lung biopsy [139–143]. In addition, forced expiratory volume in 1 s [144] and diffusing capacity of the lung for carbon monoxide [145] have been reported as predictors of pneumothorax, and these should be taken into account when planning and performing biopsy procedures and evaluating risk versus benefit for the individual patient. An alternative to CT-guided transthoracic lung biopsy is bronchoscopic tissue sampling, which has a lower risk of bleeding and pneumothorax, allowing for tissue sampling in fragile patients with poor lung function [146]. However, the diagnostic yield for peripheral nodules is considerably lower, even when combined with electromagnetic navigation bronchoscopy [146, 147]. New emerging techniques such as augmented fluoroscopy, cone beam CT and robotic-assisted bronchoscopy may increase the diagnostic yield; however, studies comparing these bronchoscopic modalities to transthoracic CT-guided biopsies are still lacking [148–151].

In summary, CT-guided biopsy approaches should be chosen with consideration of the lesion's location and size. These approaches are effective to establish histological diagnosis; however, diagnostic accuracy and incidence of complications may be affected by lesion size or needle path length. The overall complication rate is acceptable, and the major complication rate is low.