Prediction of prognosis in oral squamous cell carcinoma using infrared microspectroscopy.

BACKGROUND: Estimation of prognosis of oral squamous cell carcinoma (OSCC) is inaccurate prior to surgery, only being effected following subsequent pathological analysis of the primary tumour and excised lymph nodes. Consequently, a proportion of patients are overtreated, with an increase in morbidity, or undertreated, with inadequate margins and risk of recurrence. We hypothesise that it is possible to accurately characterise clinical outcomes from infrared spectra arising from diagnostic biopsies. In this first step, we correlate survival with IR spectra derived from the primary tumour. METHODS: Infrared spectra were collected from tumour tissue from 29 patients with OSCC and subject to classification modelling. RESULTS: The model had a median AUROC of 0.89 with regard to prognosis, a median specificity of 0.83, and a hazard ratio of 6.29 in univariate Cox proportional hazard modelling. CONCLUSION: The data suggest that FTIR spectra may be a useful early biomarker of prognosis in OSCC.

Tissue discrimination in head and neck cancer using image fusion of IR and optical microscopy.

A regression-based fusion algorithm has been used to merge hyperspectral Fourier transform infrared (FTIR) data with an H&E image of oral squamous cell carcinoma metastases in cervical lymphoid nodal tissue. This provides insight into the success of the ratio of FTIR absorbances at 1252 cm-1 and 1285 cm-1 in discriminating between these tissue types. The success is due to absorbances at these two wavenumbers being dominated by contributions from DNA and collagen, respectively. A pixel-by-pixel fit of the fused spectra to the FTIR spectra of collagen, DNA and cytokeratin reveals the contributions of these molecules to the tissue at high spatial resolution.

Metric-based analysis of FTIR data to discriminate tissue types in oral cancer.

A machine learning algorithm (MLA) has predicted the prognosis of oral potentially malignant lesions and discriminated between lymph node tissue and metastatic oral squamous cell carcinoma (OSCC). The MLA analyses metrics, which are ratios of Fourier transform infrared absorbances, and identifies key wavenumbers that can be associated with molecular biomarkers. The wider efficacy of the MLA is now shown in the more complex primary OSCC tumour setting, where it is able to identify seven types of tissue. Three epithelial and four non-epithelial tissue types were discriminated from each other with sensitivities between 82% and 96% and specificities between 90% and 99%. The wavenumbers involved in the five best discriminating metrics for each tissue type were tightly grouped, indicating that small changes in the spectral profiles of the different tissue types are important. The number of samples used in this study was small, but the information will provide a basis for further, larger investigations.

Prediction of malignant transformation in oral epithelial dysplasia using machine learning.

A machine learning algorithm (MLA) has been applied to a Fourier transform infrared spectroscopy (FTIR) dataset previously analysed with a principal component analysis (PCA) linear discriminant analysis (LDA) model. This comparison has confirmed the robustness of FTIR as a prognostic tool for oral epithelial dysplasia (OED). The MLA is able to predict malignancy with a sensitivity of 84 ± 3% and a specificity of 79 ± 3%. It provides key wavenumbers that will be important for the development of devices that can be used for improved prognosis of OED.

Prediction of malignant transformation in oral epithelial dysplasia using infrared absorbance spectra.

Oral epithelial dysplasia (OED) is a histopathologically-defined, potentially premalignant condition of the oral cavity. The rate of transformation to frank carcinoma is relatively low (12% within 2 years) and prediction based on histopathological grade is unreliable, leading to both over- and under-treatment. Alternative approaches include infrared (IR) spectroscopy, which is able to classify cancerous and non-cancerous tissue in a number of cancers, including oral. The aim of this study was to explore the capability of FTIR (Fourier-transform IR) microscopy and machine learning as a means of predicting malignant transformation of OED. Supervised, retrospective analysis of longitudinally-collected OED biopsy samples from 17 patients with high risk OED lesions: 10 lesions transformed and 7 did not over a follow-up period of more than 3 years. FTIR spectra were collected from routine, unstained histopathological sections and machine learning used to predict malignant transformation, irrespective of OED classification. PCA-LDA (principal component analysis followed by linear discriminant analysis) provided evidence that the subsequent transforming status of these 17 lesions could be predicted from FTIR data with a sensitivity of 79 ± 5% and a specificity of 76 ± 5%. Six key wavenumbers were identified as most important in this classification. Although this pilot study used a small cohort, the strict inclusion criteria and classification based on known outcome, rather than OED grade, make this a novel study in the field of FTIR in oral cancer and support the clinical potential of this technology in the surveillance of OED.

Image fusion of IR and optical microscopy for mapping of biomolecules in tissue.

It is shown that a pixel-level image fusion technique can produce images that combine the spatial resolution of optical microscopy images of haematoxylin and eosin (H&E) stained tissue with the chemical information in Fourier transform infrared (FTIR) images. The fused images show minimal distortion and the higher spatial resolution of the H&E images overcomes the diffraction limit on the spatial resolution of the FTIR images. A consideration of the FTIR spectra of nucleic acids and collagen can explain the changes in contrast between non-cancerous oral epithelium and underlying stroma within fused images formed by combining an H&E stain of oral tissue with FTIR images of the tissue obtained at a number of wavenumbers.

Insight into metastatic oral cancer tissue from novel analyses using FTIR spectroscopy and aperture IR-SNOM.

A novel machine learning algorithm is shown to accurately discriminate between oral squamous cell carcinoma (OSCC) nodal metastases and surrounding lymphoid tissue on the basis of a single metric, the ratio of Fourier transform infrared (FTIR) absorption intensities at 1252 cm-1 and 1285 cm-1. The metric yields discriminating sensitivities, specificities and precision of 98.8 ± 0.1%, 99.89 ± 0.01% and 99.78 ± 0.02% respectively, and an area under receiver operator characteristic (AUC) of 0.9935 ± 0.0006. The delineation of the OSCC and lymphoid tissue revealed by the image formed from the metric is in better agreement with an immunohistochemistry (IHC) stained image than are either of the FTIR images obtained at the individual wavenumbers. Scanning near-field optical microscopy (SNOM) images of the tissue obtained at a number of key wavenumbers, with high spatial resolution, show variations in the chemical structure of the tissue with a feature size down to ∼4 µm. The image formed from the ratio of the SNOM images obtained at 1252 cm-1 and 1285 cm-1 shows more contrast than the SNOM images obtained at these or a number of other individual wavenumbers. The discrimination between the two tissue types is dominated by the contribution from the 1252 cm-1 signal, which is representative of nucleic acids, and this shows the OSCC tissue to be accompanied by two wide arcs of tissue which are particularly low in nucleic acids. Haematoxylin and eosin (H&E) staining shows the tumour core in this specimen to be ∼40 µm wide and the SNOM topography shows that the core centre is raised by ∼1 µm compared to the surrounding tissue. Line profiles of the SNOM signal intensity taken through the highly keratinised core show that the increase in height correlates with an increase in the protein signal. SNOM line profiles show that the nucleic acids signal decreases at the centre of the tumour core between two peaks of higher intensity. All these nucleic acid features are ∼25 µm wide, roughly the width of two cancer cells.

A de-waxing methodology for scanning probe microscopy.

A de-waxing protocol that successfully removes paraffin from tissue microarray (TMA) cores of fixed tissue obtained from oral cancer is described. The success of the protocol is demonstrated by the comparison of Fourier transform infrared (FTIR) results obtained on paraffin-embedded and de-waxed tissue and the absence of any significant correlations between infrared scanning near-field optical microscopy (SNOM) images of de-waxed tissue obtained at the three main paraffin IR peaks. The success of the protocol in removing paraffin from tissue is also demonstrated by images obtained with scanning electron microscopy (SEM) and by energy dispersive spectra (EDS) of a de-waxed CaF2 disc which shows no significant contribution from carbon. The FTIR spectra of the de-waxed TMA core overlaps that obtained from OE19 oesophageal cancer cells which had never been exposed to paraffin.