Lung cancer: Diagnostic techniques

Chest roentgenogram (chest radiograph)

X-ray imaging of the chest can be used as the first step to diagnosis of lung cancer ( Fig. 2.1 ); however, the rate of false-negative results is significantly high (approximately 25% of missed cases). ^, Studies investigating the utility of chest radiographs for the screening of lung cancer have not shown any mortality benefit, and associations with increased survival have largely been attributed to lead-time bias. ^, For these reasons, their use for reliable diagnosis and staging of lung cancer is limited.

Computed tomography scans

CT of the chest is commonly used for the initial evaluation of known or suspected lung nodules, as well as to direct the initial staging assessment. Additionally, many lung cancers are initially noted as incidental findings on CT examinations obtained for other purposes ( Fig. 2.2 ). Asymptomatic patients diagnosed with lung cancer incidentally noted on CT have been found to have an earlier stage at the time of diagnosis (19.7% stage IV if asymptomatic vs. 43% if symptomatic; P < 0.001) and better overall survival (38.9 months vs. 16.1 months; P < 0.001).

Standard CT scans of the chest enable staging assessment by a number of means. First, they allow for detailed measurement of tumor size, number, and position, defining all aspects of the “T” component of the TNM staging classification except for more subtle findings such as superficial endoluminal involvement (which requires direct bronchoscopic examination and biopsy) or microscopic metastases ( Fig. 2.3 ). Standard thin-section CT also allows for preliminary assessment of hilar and mediastinal lymphadenopathy which is discussed in more detail below. Initial CT evaluation plays an important role in the preliminary assessment of extrathoracic metastasis. Most standard CT examinations of the chest are designed to image structures from the vocal cords to the upper abdomen—in terms of defining sites of disease, this allows visualization of many structures for which lung cancers have a predilection for metastasis. Among patients with metastatic disease, the most common sites of involvement include the pleura (39%), contralateral lung (32%), adrenal glands (17%), and liver (13%), all of which are visualized on most CT examinations of the chest.

CT scanning is also instrumental in the preliminary assessment of mediastinal and hilar lymph nodes, though detailed noninvasive assessment of lymph node structures may require contrast enhancement to better define lymph nodes relative to adjacent vasculature, particularly at the hila. Lymph nodes abnormalities include abnormal shape (rounded rather than ovoid), enlargement that is defined as a node >1 cm in the short axis (measured perpendicular to the longest diameter of the node), and/or complex internal attenuation ( Fig. 2.4 ). A large prospective study on the efficacy of clinical noninvasive staging defined the sensitivity of CT for mediastinal metastasis as 86% (95% CI, 70%–93%) and specificity of 67% (95% CI, 56%–75%) with a positive predictive value of 49% (95% CI, 36%–62%) and negative predictive value of 93% (95% CI, 84%–98%). The 2013 American College of Chest Physicians guidelines on the mediastinal staging of lung cancer define four major radiographic patterns on CT of the chest proposed to help guide patient selection for invasive mediastinal staging workup ( Fig. 2.5 ):

• 1.

  Radiographic group A: patients with direct tumor invasion of the mediastinum (“radiographic group A”), obscuring assessment of discrete lymph node enlargement.

• 2.

  Radiographic group B: includes patients with definite enlargement of mediastinal lymph nodes.

• 3.

  Radiographic group C: is patients with either central tumor, involving the central third of the thorax or with N1 lymph node involvement. Occult involvement of N2 or N3 nodes occurs in approximately 20%−25% of patients with this pattern.

• 4.

  Radiographic group D: patients with peripheral nodules otherwise classified as stage I tumors without any lymph node enlargement—with primary lesions <3 cm and no evidence of metastasis on CT or positron emission tomography (PET) (e.g., patients clinically staged as cT1N0M0).

The first three radiographic patterns generally warrant invasive mediastinal assessment for definitive nodal staging. The fourth group does not have a strong indication for adding invasive mediastinal staging, but it may be considered in patients not undergoing definitive surgical resection with lymph node dissection.

Noninvasive staging clearly plays an important role in the initial clinical staging of lung cancer; however, there is a considerable rate of false-negative and false-positive results; it is estimated that 40% of malignant-appearing lesions on CT scans ultimately prove to be benign. Thus CT imaging cannot be used in isolation to stage lung cancer. ^{, ,}

Positron emission tomography

PET scanning detects lung cancer by identifying tissue with a high glycolysis rate and hence higher uptake of glucose at the cellular level. For this, it uses radiolabeled glucose analog 2-[ ¹⁸ F]fluoro-2-deoxy-d-glucose (FDG). PET has emerged as a key diagnostic study in staging and treatment planning for non−small cell lung cancer (NSCLC). PET is a functional study, measuring the distribution of a parenterally administered radioisotope (for NSCLC, most commonly fluorodeoxyglucose-F-18), which collects in tissues with high metabolic activity, including neoplasms; images obtained can be formatted as stand-alone PET images or combined with CT examinations to aid anatomical localization (often termed “integrated PET-CT”). The addition of concurrent CT images to PET significantly increases the accuracy of noninvasive mediastinal staging ( Fig. 2.6 ). One series found that PET-CT improved the specificity for malignancy over PET scans from 89% to 94%. A meta-analysis incorporating data from 39 studies comparing the diagnostic accuracy of PET and CT for mediastinal staging found that CT had a median sensitivity and specificity of 61% (IQR, 50%−71%) and 79% (IQR, 66%−89%), respectively, compared with PET, which had a median sensitivity and specificity of 85% (IQR, 67%−91%) and 90% (IQR, 82%−96%), respectively. The sensitivity of PET was higher when CT showed enlarged mediastinal lymph nodes; however, PET was less specific in this circumstance, potentially reflecting other confounding etiologies of mediastinal lymphadenopathy (infection, inflammation, edema, etc.).

PET is recommended for patients without obvious extrathoracic metastasis as part of staging evaluation, and since its adoption, it has also demonstrated value in both simplifying diagnostic workup and as a prognostic tool. In a randomized controlled trial that evaluated workup at the discretion of treating physicians compared with upfront PET, the use of early PET in staging NSCLC reduced the number of patients requiring ≥1 invasive test for nodal staging (39% vs. 22% in the PET arm; P < 0.0001).

Integrated PET-CT has also been found to add prognostic information for patients with early-stage lung cancer. A retrospective cohort study of 75 patients with stage IA NSCLC who underwent resection found that survival was higher if the preoperatively standardized uptake value (SUV) was <5 (87% vs. 67%; P = 0.04) with an adjusted hazard ratio for death of 1.21 for each 1 unit increment in SUVmax.

Although the negative predictive value of PET-CT (89%−94% for T1-T2 lesions) is high, there remains an incidence of occult mediastinal nodal metastasis in approximately 11% of patients, with half of these patients showing multistation involvement. ^, Multivariate analysis demonstrated that risk factors for nodal metastasis included tumor size >3 cm, upper- or middle-lobe location, primary tumor SUVmax >4, and adenocarcinoma histology. Factors potentially contributing to false-negative PET-CT may include concurrent lung disease and diabetes, with false positives observed in the setting of age >65 years, low primary tumor SUVmax (<4), and well-differentiated primary tumors. These observations taken together suggest that even in the absence of FDG-avid mediastinal lymph nodes, patients with larger (>T1) or more metabolically active primary lesions may benefit from invasive mediastinal staging before definitive resection.

Obtaining tissue for staging and diagnosis still remains the gold standard. The evolution of bronchoscopic techniques in recent years has offered minimally invasive alternatives to open thoracotomy and mediastinoscopy. An overview of some common techniques is presented next.

Risk stratification of nodules

Determining the pretest probability of malignancy once a nodule is detected is helpful in guiding the next steps in workup and management. This practice can assist in promptly intervening on malignant nodules, as well as avoiding unnecessary procedures for benign lesions. Several validated quantitative models exist to help in risk stratification: low (<5%), intermediate (5%–65%), and high (>65%) risk of malignancy.

The Mayo Clinic model is the most extensively validated model. It was developed using multiple logistic regression analyses to determine six independent predictors of malignancy in 419 patients with noncalcified nodules (4–30 mm in diameter). Criteria include age (odds ratio [OR], 1.04 for each year), history of smoking (OR, 2.2), prior malignancy at an extrathoracic site (OR, 3.8), nodule size (OR, 1.14 for each mm), and location and presence of spiculation in the nodule on CT (OR, 2.8). ^, The Mayo model can be used in patients with or without a history of smoking with lung nodules over a fairly wide range in size.

Other models available include VA, Brock University Cancer Prediction Equation, Herder, Bayesian Inference Malignancy Calculator, and Gurney. The patient being assessed should guide model selection. For example, the Brock model was developed in a screened population, including smokers and former smokers who had a prevalence of malignancy of 5.5%. This model performs well in smaller nodules (<10 mm). The VA model consisted of a population of older male smokers with nodules ranging from 7 to 30 mm and had a prevalence of malignancy of 54%. The Herder model used 106 patients referred for PET evaluation of an indeterminate lung nodule. This model performs most accurately in nonscreened populations who have undergone PET-CT. ^,

Expert opinion evaluating the radiological characteristics of lung nodules on imaging in the context of patient factors is comparable with the quantitative models described earlier. The 2013 American College of Chest Physicians guidelines on lung cancer diagnosis and staging suggest that in patients with a solid, indeterminate lung nodule, measuring >8 mm in greatest dimension, the clinician should estimate the pretest probability of malignancy either qualitatively by using clinical judgment or quantitatively by using one of the validated pretest probability prediction models (grade 2C recommendation).

Risk stratification of nodules quantitatively or qualitatively helps to guide further management, which could entail serial observation with imaging, functioning imaging (PET), biopsy, or surgical resection. The risks and benefits of each strategy along with patient preference should be considered as an integral part of formulating the plan (grade 1C recommendation).

Diagnostic and surveillance options

Surveillance CT scan is recommended for patients in several scenarios. First is when the clinical probability of malignancy is very low (<5%) (grade 2C recommendation). Second is when the clinical probability is less than 30%−40% and the results of functional imaging and/or needle biopsy are negative. Finally, it is recommended if a patient defers further invasive testing after being well informed about his or her risk of malignancy and alternative management strategies.

Surgical resection is recommended in patients who have a solid, indeterminate pulmonary nodule, >8 mm in size when the pretest probability of malignancy is high (>65%), when the nodule is active on functional imaging, or when the biopsy is positive for malignancy. Some patients may prefer a surgical diagnostic procedure to afford simultaneous diagnostic and therapeutic intervention, and in this scenario a surgical approach should also be considered (grade 2C recommendation).

In patients with an intermediate probability of malignancy (5%−65%), further functional imaging (PET scan) is recommended. Alternatively, a biopsy can be considered to aid in treatment planning. Lung biopsy may be performed nonsurgically or surgically. A less invasive route for tissue diagnosis should be recommended if there is discordance between imaging and pretest probability, if the patient strongly desires a known diagnosis before surgical intervention, or if the risk of surgery is high (grade 2C recommendation).

In patients with multiple suspicious lesions, the choice of diagnostic technique depends on the safest, least invasive method starting with the most distal site from the suspected primary. Biopsy, for example, of a metastatic site can help confirm diagnosis and metastasis in a single procedure. If the presence of distal metastasis is reasonably well confirmed on imaging but this site is technically difficult to biopsy, the biopsy of the primary site can be considered for histological diagnosis while relying on noninvasive imaging for staging. A joint multidisciplinary tumor board discussion is recommended for cases in which definitive pathological staging is incomplete or unclear.