Nondiagnostic bronchoscopies are common in the management of indeterminant pulmonary nodules, including those at high pre-bronchoscopy risk of malignancy, and decision-making in these patients can be difficult. Our data shows that a very high risk Percepta GSC result, with its PPV > 91%, increases the frequency with which Pulmonologists recommend proceeding directly to surgical resection or other ablative therapy rather than a repeat biopsy or additional CT imaging, aligning with current ACCP guidelines for high risk nodules. For patients with early stage lung cancer, this is likely to result in a reduction in the time to appropriate therapy, better outcomes[15], and reduced health care costs(3). Moreover, a Percepta GSC very high risk result increases physician confidence in this recommendation and their confidence in their assessment of a patient’s risk of malignancy. The ability to up-classify risk could also increase confidence in a physician’s decision to proceed directly to non-invasive ablative therapy in selected patients who are at prohibitive risk of complications from either surgery or transthoracic needle sampling.
Our study has some limitations that merit discussion. First, this was a decision-impact study, not Percepta GSC being used in a real-world setting, although we sought to closely model this with the pre - post format. There are several ways in which our survey format differed from real-world use. Although our survey takers had access to clinical data, including detailed radiology reports, a pulmonologist would typically review the images themselves and may have come to a different determination of risk with that data. To minimize the impact of this, all CTs were interpreted by expert thoracic radiologists and all salient features were described in detail.
Many pulmonologists utilize PET imaging in the risk assessment of indeterminate nodules(18). Given that our intent was to specifically assess use of invasive procedures, survey takers who chose PET as a first option were given an indeterminant result and asked to choose a second option. Although this was imposed, this is not an uncommon clinical scenario. This was intended to maximize diagnostic uncertainty to model real-world decision-making and to avoid confounding the primary measured outcome.
In an actual patient case, a Percepta GSC result would be returned approximately 7 days after the bronchoscopy, and the data would not be evaluated at the same time as, or immediately following a review of a case with a nondiagnostic pathology result. Our pre - post cohort most closely resembles the sequential decision-making of a real-life bronchoscopy, and it is in this cohort where we saw our largest impact on the recommendation for ablative therapy. Finally, our survey did not include the important factor of patient preference. Shared decision-making and incorporation of an individual patient’s perspectives and values is increasingly recognized as a vital part of managing pulmonary nodules, particularly those with a high risk of malignancy(19). However, the additional information provided by a Percepta GSC up classification to very high risk could provide reassurance to patients as well their health care providers, allowing for a more informed and confident decision.
It is worth noting that while the original bronchoscopist in each case and the Mayo Risk Model called all 37 cases high-risk, in many instances our survey takers did not. That reflects the wide variability in physicians’ ability to accurately assess the risk of malignancy in indeterminant lung lesions. The fact that the Percepta GSC result improved the assessment is evidence that better, more objective tools are needed for this task.
Our study has a number of important strengths. While the prior decision impact study of the Percepta bronchial genomic classifier focused on intermediate risk nodules and the utility of a down-classification of risk by a Percepta result, the current study addresses decision-making in nodules at high risk of malignancy. Arguably, decision-making in these cases following a nondiagnostic bronchoscopy is more difficult, given the potential for morbidity or even mortality from complications of additional invasive procedures or surgery. Our study used actual cases from the AEGIS trials used to develop the original classifier and from the Percepta Registry, so our survey takers were being tested on Percepta GSC’s intended use population. All of the cases had a genuine nondiagnostic bronchoscopy, and all had a real Percepta GSC result. Our study involved pulmonologists with a broad range of experience, across diverse practice settings and geographical location, and physicians who had not previously used Percepta GSC were well-represented. Finally, this study looked at decision-impact using two different models for presenting data to the survey takers, increasing the validity of the strong findings for the primary endpoint.