Fine needle aspirates (FNAs) of suspicious breast lesions are often used to aid the diagnosis of female breast cancer. Biospectroscopy tools facilitate the acquisition of a biochemical cell fingerprint representative of chemical bonds present in a biological sample. The mid-infrared (IR; 4,000-400 cm(-1)) is absorbed by the chemical bonds present, allowing one to derive an absorbance spectrum. Complementary to IR spectroscopy, Raman spectroscopy measures the scattering by chemical bonds following excitation by a laser to generate an intensity spectrum. Our objective was to apply these methods to determine whether a biospectroscopy approach could objectively segregate different categories of FNAs. FNAs of breast tissue were collected (n = 48) in a preservative solution and graded into categories by a cytologist as C1 (non-diagnostic), C2 (benign), C3 (suspicious, probably benign) or C5 (malignant) [or C4 (suspicious, probably malignant); no samples falling within this category were identified during the collection period of the study]. Following washing, the cellular material was transferred onto BaF(2) (IR-transparent) slides for interrogation by Raman or Fourier-transform IR (FTIR) microspectroscopy. In some cases where sufficient material was obtained, this was transferred to low-E (IR-reflective) glass slides for attenuated total reflection-FTIR spectroscopy. The spectral datasets produced from these techniques required multivariate analysis for data handling. Principal component analysis followed by linear discriminant analysis was performed independently on each of the spectral datasets for only C2, C3 and C5. The resulting scores plots revealed a marked overlap of C2 with C3 and C5, although the latter pair were both significantly segregated (P < 0.001) in the Raman spectra. Good separation was observed between C3 and C5 in all three spectral datasets. Analysis performed on the average spectra showed the presence of three distinct cytological groups. Our findings suggest that biospectroscopy tools coupled with multivariate analysis may support the current FNA tests whilst increasing the sensitivity and associated reliability for improved diagnostics.

No abstract