ESTRO 2024 - Abstract Book

S5267 ESTRO 2024 This study received approval from the local Ethics Committee of the Canton of Bern, Switzerland (2021-00728). We included treatment-naïve patients with locoregionally advanced HNSCC who underwent upfront neck dissection, followed by RT (+/- chemotherapy) with curative intent. Genetic material was sourced from formalin fixed, paraffin-embedded biopic or surgical specimens. Normal tissue (salivary gland) was identified microscopically, correlated with the corresponding paraffin block, and automatically cored out. DNA extraction was performed with QiaCube (Qiagen) from 2-10 randomly selected regions per block, followed by target capture (Human SureSelect All Exon V5/V7, Agilent) and whole-exome sequencing (NovaSeq 6000/HiSeq2500, Illumina) with 50x depth. Genetic variants were identified with DeepVariant (v1.3.0) and joint variant calling was performed with GLnexus (v1.4.1). Genotype imputation was accomplished with Beagle 5.4, whereas PLINK 2.0 and R (v4.1.1) was used for quality control, correlation of geno- and phenotype and data analysis. Clinical data was collected from electronic health records, including early and late adverse events according to the CTCAE v5.0. Staging was based on AJCC/UICC, 7th edition. Missing values were imputed with the mice (v3.16.0) package. Large salivary glands and oral cavity were re-contoured manually and RT doses were extracted from dose-volume histograms. Multivariable models included as covariates: doses to the parotid and submandibular glands (V50Gy), and oral cavity (Dmean), surgical resection of salivary glands, and concomitant chemotherapy. The dependent variables mucositis, xerostomia, weight loss, pain, dysphagia, and dermatitis were investigated individually in separate modes. The Bonferroni method was used for multiple testing correction of genetic variants. Ninety-two patients were analyzed, comprising 77% men and 23% women, with a median age of 58 years (range 20 – 88 years). All patients presented with locally advanced HNSCC: 14% in stage III, 73% in stage IVA and 7% in stage IVB.One third of patients had human papillomavirus-associated carcinoma. Oral cavity and oropharynx were the most common primary tumor sites, accounting for 45% and 46%, respectively. Only 5% of patients presented with supraglottic larynx carcinoma, and 4% with hypopharyngeal carcinoma. Median follow-up was 48 months, ranging from 2 to 173 months. Multivariable models identified 10 statistically-significantly associated variants (Table 1) with the endpoint late dysphagia (3-4 months after RT). Of these, 90% were situated in non-coding regions, whereas only one was a synonymous variant in a coding sequence (rs11626122). All were single nucleotide variants and are present in our study cohort with a variant allele frequency between 11-47%. Pathway analysis using Reactome revealed enrichment of genes in pathways affecting signal transduction, neuronal system, cell cycle, and transcription. Regarding covariates, dose to oral cavity was statistically significant in all, dose to large salivary glands in several, and chemotherapy in no multivariable models investigating the selected ten variants. No other toxicities or time points were associated with genetic variants. Radiobiology - Normal tissue radiobiology Results:

Table 1: Statistically-significant genetic variants with endpoint late dysphagia.

Chromosome

Position

REF Allele

ALT Allele (IUPAC)

dbSNP

Overlapping/nearest genes

Variant Allele Frequency (study cohort)

Corrected p-value (assoc. with late dysphagia)

1 1 8 9 9

16201723

G T T C C A G A C G

T

rs6676143 rs2297170 rs57638937 rs2281999 rs2274593 rs2278342 rs11626122 rs11555426

ARHGEF19

0.47 0.23 0.16 0.29 0.25 0.22 0.13 0.11 0.36 0.39

0.01 0.02 0.02

100353717 86477167 35381507 35398409 70555053 73502906 30874833 36141362

C C T T C A G

CDC14A

WWP1, RMDN1

UNC13B UNC13B

0.004 0.006 0.048

12 14 14 18

PTPRB

HEATR4

0.03

COCH

0.045

T

rs3819174

ELP2

0.01 0.02

X

2502830

C

N/A

DHRSX, MIR6089