Study On The Detection Of The BRAF V600E Mutation In Colorectal Cancer By Near-infrared Spectroscopy

The mutation of the B-rapidly accelerated fibrosarcoma（BRAF）gene in colorectal cancer（CRC）often make the treatment more difficult.Most of the BRAF mutations involve a substitution of valine（V）to glutamic acid（E）at codon 600,which is called BRAF V600E mutation.The typical methods for the clinical diagnosis of the BRAF V600E mutation in CRC are immunohistochemistry（IHC）in conjunction with microscopy,polymerase chain reaction（PCR）,and gene sequencing.However,these methods have their inherent disadvantages.In this paper,near-infrared spectroscopy（NIRS）in conjunction with chemometrics is used to provide a sample non-destructive,sensitive,robust,easy-to-use,rapid,and low-cost auxiliary method for the diagnosis of the BRAF V600E mutation in CRC based directly on the difference of amino acids between wild type and mutant.This work can expand the application of NIR spectroscopy in the auxiliary diagnosis of gene mutation in cancer.OBJECTIVES:To establish a sample non-destructive,sensitive,robust,easy-to-use,rapid,and low-cost NIRS for the auxiliary diagnosis of the BRAF V600E mutation in CRC.METHODS:1.Collection of the CRC tissue section samples and measurement of their NIR spectraA total of 312 CRC tissue sections of BRAF V600E wild type and mutant including 104 paraffin-embedded,104 deparaffinized,104hematoxylin-eosin（HE）-stained sections were collected.And their near-infrared transflectance spectra（NIRTFS）were measured in the range of 12000–4000 cm^-1 using the selected resolution 8 cm^-11 and the selected number of co-added scans 64.2.Establishment and validation of the models for the detection of the BRAF V600E mutation in CRC using paraffin-embedded sectionsBased on the NIRTFS of 104 paraffin-embedded sections,the counter propagation artificial neural network（CP-ANN）model for the detection of the BRAF V600E mutation in CRC was established and validated.3.Establishment and validation of the models for the detection of the BRAF V600E mutation in CRC using deparaffinized sectionsBased on the NIRTFS of 104 deparaffinized sections,the CP-ANN model for the detection of the BRAF V600E mutation in CRC was established and validated.4.Establishment and validation of the models for the detection of the BRAF V600E mutation in CRC using HE-stained sectionsBased on the NIRTFS of 104 HE-stained sections,the CP-ANN model for the detection of the BRAF V600E mutation in CRC was established and validated.5.Establishment of the model for the detection of the BRAF V600E mutation in CRC using deparaffinized and paraffin-embedded sectionsBased on the NIRTFS of 40 deparaffinized and 40 paraffin-embedded sections,the CP-ANN model for the detection of the BRAF V600E mutation in CRC was established.6.Establishment of the model for the detection of the BRAF V600E mutation in CRC using deparaffinized and HE-stained sectionsBased on the NIRTFS of 40 deparaffinized and 40 HE-stained sections,the CP-ANN model for the detection of the BRAF V600E mutation in CRC was established.RESULTS:1.The established CP-ANN model,using paraffin-embedded sections,has 98.0%of the classification accuracy of calibration（CAC）,95.0%of the classification accuracy of cross-validation（CACV）,and 94.4%of the classification accuracy of validation（CAV）;100.0%of the sensitivity,87.5%of the specificity,and 93.8%of the accuracy,in the clinical diagnosis.2.The established CP-ANN model,using deparaffinized sections,has97.0%of CAC,92.0%of CACV,and 94.4%of CAV;100.0%of the sensitivity,95.0%of the specificity,and 97.5%of the accuracy,in the clinical diagnosis.3.The established CP-ANN model,using HE-stained sections,has95.0%of CAC,88.0%of CACV,and 93.1%of CAV;100.0%of the sensitivity,82.5%of the specificity,and 91.3%of the accuracy,in the clinical diagnosis.4.The established CP-ANN model,using deparaffinized and paraffin-embedded sections,has both 97.0%of CAC and CACV;100.0%of the sensitivity,92.5%of the specificity,and 96.3%of the accuracy,in the clinical diagnosis.5.The established CP-ANN model,using deparaffinized and HE-stained sections,has 95.0%of CAC and 90.0%of CACV;100.0%of the sensitivity,85.0%of the specificity,and 92.5%of the accuracy,in the clinical diagnosis.CONCLUSION:1.The three CP-ANN models established based on the NIRTFS of paraffin-embedded,deparaffinized,HE-stained sections have appropriate model performances and clinical diagnostic performances.Since the paraffin-embedded section is the most common for the pathological specimen storage,it is the most suitable sample for the clinical auxiliary diagnosis.2.The two CP-ANN models established based on the NIRTFS of the deparaffinized and paraffin-embedded sections,the deparaffinized and HE-stained sections have appropriate model performances and clinical diagnostic performances,which prove that the NIRS is used to detect the BRAF V600E mutation in CRC based directly on the fundamental difference between wild type and mutant（that is,between V and E）,rather than the difference between the sample types.