The use of CT imaging for both diagnostic evaluation and screening of lung cancer has grown significantly in recent years and, as a consequence, the healthcare system is increasingly challenged to develop safer and more effective methods to determine the clinical significance of pulmonary lesions. A recent, retrospective cohort study concluded that the prevalence of incidental pulmonary nodules increased steadily from 2006 to 2012. Over this time, the annual number of chest CT scans increased by 53 percent, and the annual number of scans with a lung nodule increased by 95 percent (Am J of Resp and Critical Care Med 2015;192:10,1208-14).
As always, clinical history provides the foundational framework for evaluating pulmonary lesions. However, within this framework there is significant uncertainty, and practicing clinicians and their patients often fear missing a lung cancer diagnosis. Consequently, physicians frequently pursue the most aggressive approaches to secure a definitive result. Published literature suggests that 40 percent or more of patients with a CT-identified pulmonary lesion may undergo a risky, costly, invasive procedure for what turns out to be a benign diagnosis (Chest 2015;148(6):1405-14).
Evidence-based guidelines have identified clinical and radiographic characteristics that influence individual risk of malignancy and thus inform how aggressively to pursue a diagnostic result. These characteristics include age, smoking history, size of the lesion, spiculated shape, location in the lungs, and solid versus cystic appearance. Additionally, a number of validated algorithms (nodule risk calculators) aggregate these characteristics into risk stratifications for malignancy and have been reliably validated. However, their adoption and subsequent incorporation into clinical practice has been inconsistent and modest at best. Using all available tools, physician assessment ultimately becomes the most common method of assigning malignancy risk, which is generally grouped into three actionable categories: low risk (<10%); intermediate risk (10-60%); and high risk (>60%) (N Engl J Med 2015;373:243-51).
High risk of malignancy (>60%) is generally associated with a clinical mandate to pursue diagnostic testing to resolution, although mitigating clinical circumstances may modify the aggressiveness of the approach. Less clarity exists about the approach or level of diagnostic aggressiveness for lesions classified as intermediate and low risk. Choices for next-step diagnostics in these two risk categories range from conservative management with follow-up serial CT scans to surgical intervention. Consequently, there is much greater practice variation in the management decisions among these groups of patients, and justifications for those decisions may hinge on factors that do not represent best practice.
Bronchoscopy is frequently selected as a reasonable option because of its relatively low complication rate as compared with transthoracic needle aspiration/biopsy or surgical lung biopsy. Diagnostic yields for bronchoscopy will vary widely by pre-test risk category, size of the pulmonary lesion, and operator experience. In the low and intermediate pre-test risk population, it is not uncommon to see more than half of diagnostic bronchoscopies return an “inconclusive” result (Am J Respir Crit Care Med 2016;193(1):68-77). This result can be disconcerting for physicians and patients alike, given that additional diagnostic steps change the risk/benefit ratio significantly.
Finding Answers With Molecular Testing
New molecular testing is helping to reduce this diagnostic uncertainty. A 23-gene bronchial gene expression classifier (GEC) has been designed to improve the diagnostic performance of bronchoscopy in patients undergoing a workup for lung cancer. The classifier utilizes bronchial epithelial cells obtained from standard cytology brushings of the proximal right or left mainstem. Cells obtained for the classifier are often distant from the primary index lesion, negating the need and risk to sample the lesion directly. This bronchial GEC uses messenger RNA on a microarray platform to measure the ‘field of injury’ in cells—genomic changes induced by tobacco and lung cancer. The assay is performed only if bronchoscopy is cytologically “inconclusive” and is requested from the procedural physician. Once obtained and submitted the bronchial GEC is performed in a CLIA-certified laboratory.
Two, prospective multi-center clinical-validation studies (Airway Epithelial Gene Expression in the Diagnosis of Lung Cancer Trials, AEGIS I and II Trials) evaluated this classifier among former and current smokers undergoing bronchoscopy for the workup of lung cancer. Findings published in The New England Journal of Medicine demonstrated that the bronchial GEC improved the overall sensitivity of bronchoscopy— from 75 percent with bronchoscopy alone to 97 percent when paired with the bronchial GEC (p<0.001).
More importantly, among the low and intermediate pre-test risk groups, for whom the consequences of diagnostic uncertainty from an inconclusive bronchoscopy are the greatest, the addition of the bronchial GEC demonstrated the greatest improvement in sensitivity. Combined in both the low and intermediate pre-test risk groups the sensitivity of bronchoscopy alone significantly improved from 41 percent to 93 percent with the classifier and bronchoscopy (p<0.0001). Notably, performance of the bronchial GEC in the AEGIS trials remained constant regardless of lesion size, location, histology, and cancer stage (N Engl J Med 2015;373:243-51). The high sensitivity of the bronchial GEC corresponds with a high negative predictive value—91percent across the intermediate and low pre-test risk groups. In these two specific “intended-use” groups, a utility modeling study published by Vachani, et al suggested the classifier would identify up to 50 percent of patients who could be managed with conservative serial CT imaging as opposed to more invasive diagnostic procedures.
Since the classifier became commercially available in the U.S. in July 2015, it has been evaluated in a prospective multi-center registry study across 43 sites (an equal mix of community- and academic-based practices). At the recent annual meeting of the American College of Chest Physicians in Los Angeles, researchers presented data from a planned interim analysis highlighting the clinical utility of the test (Chest 2016;150(4_S):1026A). The analysis examined both clinical adoption and test performance as well as the decision impact of the test.
At the time of the analysis, 290 of 420 (69%) eligible patients had an “inconclusive” bronchoscopy during a workup for lung cancer. Of these, 230 patients received bronchial GEC results with sufficient follow-up to assess decision impact. Review of how the test was utilized by physicians in a real-world clinical setting demonstrated that 77 percent of patients in whom the test was ordered were within the intended use pre-test intermediate (67%) and low (10%) risk groups. The bronchial GEC re-classified 44 percent of the intended-use group to a post-test risk of less than 10 percent. The decision impact on post-test diagnostic management revealed that 41 percent of reclassified patients either avoided a subsequent invasive procedure or had a reduction in their follow-up scanning frequency by greater than two-fold. Notably this ‘real world’ clinical utility performance closely mirrors the predicted utility of the test in modeling studies (BMC Pulm Med 2016;16:66).
On the aggregate strength of the medical evidence documenting the bronchial GEC's performance, several Medicare contractors recently issued a draft local coverage determination for the test. Once finalized, this determination will provide coverage for the test to more than half of the U.S. Medicare population.
The clinically validated bronchial GEC is a valuable tool for bronchoscopists in the age of lung cancer screening and incidental lung nodules. Its performance addresses the Institute of Medicine's recent call for “better identification, analysis, and implementation of approaches to improve diagnosis and reduce diagnostic error.” Furthermore, this tool diminishes physician and patient uncertainty while improving diagnostic management of pulmonary nodules identified by CT scan.
TRAVIS DOTSON, MD, and CHRISTINA BELLINGER, MD, are from the Department of Medicine, Section on Pulmonary & Critical Care Medicine, Wake Forest School of Medicine, Winston-Salem, N.C.