Buccal spectral markers for lung cancer risk stratification.

Abstract

Lung cancer remains the leading cause of cancer deaths in the US with >150,000 deaths per year. In order to more effectively reduce lung cancer mortality, more sophisticated screening paradigms are needed. Previously, our group demonstrated the use of low-coherence enhanced backscattering (LEBS) spectroscopy to detect and quantify the micro/nano-architectural correlates of colorectal and pancreatic field carcinogenesis. In the lung, the buccal (cheek) mucosa has been suggested as an excellent surrogate site in the "field of injury". We, therefore, wanted to assess whether LEBS could similarly sense the presence of lung. To this end, we applied a fiber-optic LEBS probe to a dataset of 27 smokers without diagnosed lung cancer (controls) and 46 with lung cancer (cases), which was divided into a training and a blinded validation set (32 and 41 subjects, respectively). LEBS readings of the buccal mucosa were taken from the oral cavity applying gentle contact. The diagnostic LEBS marker was notably altered in patients harboring lung cancer compared to smoking controls. The prediction rule developed on training set data provided excellent diagnostics with 94% sensitivity, 80% specificity, and 95% accuracy. Applying the same threshold to the blinded validation set yielded 79% sensitivity and 83% specificity. These results were not confounded by patient demographics or impacted by cancer type or location. Moreover, the prediction rule was robust across all stages of cancer including stage I. We envision the use of LEBS as the first part of a two-step paradigm shift in lung cancer screening in which patients with high LEBS risk markers are funnelled into more invasive screening for confirmation.