Lung cancer is most often detected at late stages where treatment options are limited and outcomes are poor. Furthermore, common chest imaging tests detect significant numbers of lesions, of which most are found to be benign. As a result, there is a significant demand for a blood test for lung cancer diagnosis.
We analyzed a cohort of serum samples from patients with biopsy-confirmed non-small cell lung cancer, along with control samples from cancer-free patients, matched for gender, age, smoking status (active, ex-smoker, never-smoker) and smoking history. Serum samples were processed to remove high abundance proteins, and analyzed by capillary electrophoresis-electrospray ionization mass spectrometry using a proprietary microfluidic chip-based platform and an ultra-high sensitivity mass spectrometer. Samples were blinded and randomized during sample preparation and analysis in order to remove bias in the measurement. After signal pre-processing of the data, the resulting intensities of ∼1000 molecular species in the MS profiles were analyzed using pattern recognition methods.
Preliminary results for 93 samples show that a pattern of 24 molecular components yields an error rate of 18% (76% sensitivity, 87% specificity) for distinguishing cancer from non-cancer.
This system for proteomic analysis provides increased sensitivity and reliability for protein peak identification, substantially increasing the number of proteins observed and reducing inter- and intra-sample variability.
This approach holds promise as a new method for diagnosing lung cancer. This advanced system facilitates discovery of molecular signatures, and will lead to the roll-out of clinically practical, high-throughput cancer detection methodologies.
Jonathan Heller,Employee Vice President, Information and Project Planning