Predicting response to anti-tumour necrosis factor-alpha therapy-alpha in fibrostenotic Crohn's disease using infra-red microspectroscopy.

Abstract

Background: Fibrostenotic strictures are common in Crohn's disease, have limited treatment options and lack validated biomarkers to predict treatment response. Fourier-transform infra-red (FTIR) spectroscopy provides reproducible biochemical information from biospecimens using a rapid, label-free, non-destructive technique via molecular vibrations. We previously completed a proof-of-concept study that identified spectral biomarkers for inflammation in a model of colitis.1 The aim of this study was to investigate whether the tissue biochemical 'fingerprint' using FTIR spectroscopy can predict patient clinical response to anti-tumour necrosis-factor-alpha (anti-TNFalpha) therapy, and identify spectral biomarkers for treatment response. Method(s): Hyperspectral images were acquired using the Agilent Cary 670 FPA-IR coupled with the Agilent Cary 620 microscope in transflection mode in the wavenumber region 1800 - 800 cm-1 from fixed ileo-colonoscopic biopsies from 62 patients with Crohn's disease and 12 healthy controls. A subset of patients (n = 24) were included from STRIDENT2, a randomised study assessing adalimumab therapy for intestinal Crohn's strictures. Thousands of spectra were obtained per sample, pre-processed via transformation to the second derivative, normalised and reduced to 25 spectra per sample prior to analysis using partial least squares discriminant analysis (PLSDA) with internal cross-validation. Spectral biomarkers were identified by the most contributive infra-red bands from the latent variables (LV) and significant variable importance in projection (VIP) scores of >1. Adjacent histological sections were stained with H&E, Masson's trichrome and alpha-smooth muscle actin with inflammation and fibrosis scored by a clinical pathologist. Result(s): Spectra from baseline biopsies from the 24 patients in the STRIDENT study demonstrated excellent performance using PLSDA in discriminating between responders and non-responders in relation to the 12 month clinical outcomes, defined by area under the receiver operating characteristic curve (AUC) >0.9 for MRI and IUS responses, normalisation of faecal calprotectin and CRP, CDAI and pain responses (Table 1). Spectral biomarkers of interest for stricture response to anti-TNFalpha are shown in Figure 1 and include the amide I and II protein bands (peaks at 1651 and 1543 cm-1, respectively), the carboxylate group of proteins (1454 cm-1), lipid bands (1750 and 1395 cm-1), collagen band (1236 cm-1), glycoprotein bands (1081 and 1030 cm-1), and nucleic acid bands (1236, 1081 and 966 cm-1). Conclusion(s): The novel technique of FTIR spectroscopy detects tissue biochemical 'fingerprints' in fibrostenotic Crohn's disease that demonstrate excellent discrimination for predicting clinical response to adalimumab therapy.