Impact of an integrated classifier using biomarkers, clinical and imaging factors on clinical decisions making for lung nodules.

Abstract

An integrated classifier that utilizes plasma proteomic biomarker along with five clinical and imaging factors was previously shown to be potentially useful in lung nodule evaluation. This study evaluated the impact of the integrated proteomic classifier on management decisions in patients with a pretest probability of cancer (pCA) ≤50% in "real-world" clinical setting. Retrospective study examining patients with lung nodules who were evaluated using the integrated classifier as compared to standard clinical care during the same period, with at least 1-year follow-up. A total of 995 patients were evaluated for lung nodules over 1 year following the implementation of the integrated classifier with 17.3% prevalence of lung cancer. 231 patients met the study eligibility criteria; 102 (44.2%) were tested with the integrated classifier, while 129 (55.8%) did not. The median number of chest imaging studies was 2 [interquartile range (IQR), 1-2] in the integrated classifier arm and 2 [IQR, 1-3] in the non-integrated classifier arm (P=0.09). The median outpatient clinic visit was 2.00 (IQR, 1.00-3.00) in the integrated classifier arm and 2.00 (IQR, 2.00-3.00) in the non-integrated classifier (P=0.004). Fewer invasive procedures were pursued in the integrated classifier arm as compared to non-integrated classifier respectively (26.5% vs. 79.1%, P<0.001). All patients in the integrated classifier arm with post-pCA (likely benign n=39) had designated benign diagnosis at 1-year follow-up. In patients with lung nodules with a pCA ≤50%, use of the integrated classifier was associated with fewer invasive procedures and clinic visits without misclassifying patients with likely benign lung nodules results at 1-year follow-up.