Diagnostic characteristics of a serum biomarker in patients with positron emission tomography scans

Abstract

Background: Surgery for pulmonary nodules results in a benign diagnosis in 10% to 30% of cases. Computed tomography and fluorodeoxyglucose-positron emission tomography (FDG-PET) are highly sensitive but less specific. High-risk patients (age > 55 years and smoke > 15 pack-years) for lung cancer with negative FDG-PET scans, or low-risk patients (age < 55 years or smoke < 15 pack-years) with FDG-PET-avid lesions may have higher rates of benign nodules. We hypothesized that our serum biomarker improves diagnostic accuracy by providing greater specificity.

Methods: Fifty-eight patients with pulmonary nodules (< or = 3 cm) were prospectively enrolled. We tested the accuracy of our proteomic biomarker in the serum by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Malignancy rates, contingency tables, sensitivity, and specificity analyses were calculated for the entire group and in a subset of patients at high risk for benign disease.

Results: We identified 46 (79%) lung cancers and 12 (21%) benign lesions. Forty-five nodules were FDG-PET-avid. In 36 high-risk patients with FDG-PET-avid lesions, 32 (89%) had cancer. Of the remaining 22 lower-risk patients, 14 (64%) had cancer (p = 0.02). The serum biomarker sensitivity was 26.1%, specificity was 91.7%, positive predictive value was 92%, negative predictive value was 24%, and overall accuracy was 40%. The serum signature accurately predicted all eight benign nodules in this 22-patient subset.

Conclusions: The serum protein biomarker has a high specificity. This biomarker has a high positive predictive value but low negative predictive value and may improve noninvasive evaluation of lung nodules. Validation in a larger population is warranted.