Radiogenomics in Aerodigestive Tract Cancers

Overview

Aerodigestive tract cancers are common malignancies. These cancers were ranked to be top-ten cancer-related deaths in Taiwan. Although many new target therapies and immunotherapies have emerged, many of the treatment eventually fail. For example, a 30-40% failure rate has been reported for target therapy, and, even higher for immune checkpoint inhibitors. A reliable model to more accurately predict treatment response and survival is warranted. The radiomic features extracted from F-18 fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) can be used to figure tumor biology such as metabolome and heterogeneity. It can therefore be used to predict treatment response and individual survival. On the other hand, genomic data derived from next-generation sequencing (NGS) can interrogate the genetic alteration of cancer cells. It can be used to feature genetic identification of the tumor and can also be used to identify target genes. However, both modalities have their weakness; a combination of the two may devise a more powerful predictive model for more precise clinical decision. The investigators plan to recruit patients aged at least 20-year with the diagnosis of aerodigestive tract cancers for radiogenomic study. Our previous studies have found that radiomic features derived from 18F-FDG PET can predict treatment response and survival in patients with esophageal cancer treated with tri-modality method. The investigators also discovered that radiomics could predict survival in patients with EGFR-mutated lung adenocarcinoma treated with target therapy. In addition, our study results showed that the level of PD-L1 expression is associated with radiomics as well. The investigators plan to add genomic data into radiomics and interrogate cancers from different aspects. The investigators seek to devise a more precise model to predict the treatment response and survival in patients with aerodigestive tract cancers.

This is a prospective study and The investigators use the routine pathological specimens for next generation sequencing (NGS), microsatellite instability (MSI) and immunohistochemical stains. Pathological examinations including PD-L1, EGFR status, ALK and ROS-1. NGS (Next Generation Sequencing)： Total DNA were extracted from EDTA-peripheral venous blood and paraffin-embedded tumor specimens with the QIAamp® DNA Blood Mini Kit and QIAamp DNA FFPE Tissue Kit (QIAGEN GmbH, Hilden, Germany), respectively. For DNA whole exome sequence, briefly, tumor and blood DNA were sonicated by Covaris M220 sonicator (Life Technologies Europe, Gent, Belgium) and then ligated to adaptor for further amplification (Illumina® TruSeq Exome Library Prep, USA). All of library preparation were performed in the Cancer Translational Core Facility of Taipei Medical University. After library preparation, all samples were sequenced using the NextSeq500 system according to the manufacturer's instructions (Illumina, San Diego, USA). After sequencing performance, quality of reads file (fastq) was assessed by FastQC and then mapped using human Hg19 as the reference. Bam files were used as input for the Varscan algorithm to identify germline and somatic mutations. Variants annotated and filtered were manually checked using IGV (Integrative Genomics Viewer), then confirmed by Sanger sequence. To calculate the TMB (total mutation burden) per megabase, the total number of mutations counted is divided by the size of the coding region of the targeted territory. To calculate MATH (mutant-allele tumor heterogeneity), The investigators will first obtain the MAF (the fraction of DNA that shows the mutated allele at a gene locus) of each tumor specimen. The MAF distribution will be used to calculate the median (center of distribution) and the MD (median deviation) of MAFs in a tumor. The MD is determined by obtaining the absolute differences of all MAFs from the median MAF. Then the median of the absolute differences is multiplied by a factor of 1.4826 to obtain the MD. The MATH value is calculated as the percentage ratio of the MD to the median: MATH = (MD/median)×100. MSI (Microsatellite instability) Microsatellite instability polymerase chain reaction (MSI-PCR) MSI-PCR testing was performed by the Cancer Translational Core Facility of Taipei Medical University using Promega MSI analysis kit (Promega). The MSI analysis consists of five nearly monomorphic mononucleotide markers (BAT-25, BAT-26, NR- 21, NR-24, and MONO-27) for MSI determination. MSI analysis was performed according to the manufacturer's directions (Promega). Products were analyzed by capillary electrophoresis and the investigators interpreted microsatellite instability at 2 or more of the 5 mononucleotide loci as MSI-high, microsatellite instability at a single mononucleotide locus as MSI-low, and no instability at any of the loci as microsatellite stable (MSS). The image features of FDG PET the investigators extracted as followed: The traditional image parameters include SUVmax, metabolic tumor volume (MTV) and total lesion glycolysis (TLG) of the primary tumor. The traditional FDG PET parameters were calculated using commercialized software (PBAS, PMOD 4.0). Radiomics (texture analysis) will be calculated only for pre-treatment FDG PET. The matrices of radiomic analysis include histogram analysis, Gray-level co-occurrence matrix (GLCM)、texture feature coding co-occurrence matrix (TFCCM)、gray-level run-length matrix (GLRLM)、gray-level size zone matrix (GLSZM)、neighborhood gray-tone difference matrix (NGTDM)、Texture Feature Coding Matrix (TFCM)、Texture Feature Coding Co-Occurrence Matrix (TFCCM) and Neighbouring Gray Level Dependence Matrix (NGLD).