Introduction New cheap and high throughput technologies may allow earlier diagnosis and cost-effective screening programmes for lung cancer (LC). We have shown that sputum is a feasible biofluid for FTIR spectroscopy analysis1 and now further evaluate FTIR in diagnosing LC.

Methods Sputum was taken from three groups: a) 54 patients (mean age 66.6±8.7 years) with a histological diagnosis of LC (39 NSCLC, 9 small cell, 1 carcinoid, 5 clinical diagnosis). b) 24 patients (mean age 65.1±13.6 years) having bronchoscopy for possible LC but no evidence of cancer was found after 1-year follow-up (high-risk). c) 54 healthy controls (HC) (mean age 51.1±15.3 years) who had no history or symptoms of LC or known respiratory disease. Sputum was self-expectorated and frozen immediately at −80°C, thawed in batches, mucolytics were added then samples centrifuged at 3000 rpm for 10 min to form pellets. FTIR was performed using the VERTEX 70 spectrometer (Bruker Optics Ltd, Banner Lane, Coventry, UK). Median absorbance values for each wavenumber for the LC and HC cohorts were compared, then principal component analysis (Abstract S38 Figure 1) and logistical regression identified the wavenumbers that provided the greatest accuracy in differentiating the two groups; the high-risk cohort was then applied to the predictive model to see if they could be correctly identified.

• Download figure
• Open in new tab
• Download powerpoint

Abstract S38 Figure 1

Principal component analysis of cancer (c), healthy control (s) and high-risk (h) spectra.

Results 126 light absorbance wavenumbers were significantly different between the LC and HC groups (each p<0.05). Two wavenumbers, 1031.7 cm⁻¹ and 1409.7 cm⁻¹ were used to develop a predictive model providing a sensitivity of 93% and specificity of 91%. This model then predicted 17 of the 24 high-risk cohorts as LC.

Conclusion FTIR spectroscopy can distinguish LC from HC with high accuracy but had reduced specificity when applying high-risk patients, tending to over-diagnose LC. Follow-up will determine if these 17/24 people are indeed false positives or have pre-cancerous molecular changes not identifiable by current methods.