6533b838fe1ef96bd12a4f83

RESEARCH PRODUCT

Development of a gene panel for 18F-FDG PET positivity prediction in gastric cancer.

subject

description

420 Background: 18F-FDG PET is widely used in clinical cancer diagnostics. However, 18F-FDG PET scan in gastric cancer (GC) is still controversial because of its lower sensitivity in diagnosis and staging compared to other imaging modalities. The purpose of this study was to establish a gene panel for 18F-FDG PET positivity in GC by using patient-derived xenografts (PDXs). Methods: BALB/c nude mice were subcutaneously implanted with 30 cases of GC PDX tissues and underwent a simultaneous PET/MRI scanner. Using RNA-seq data of the 30 GC PDXs for training set, we constructed a gene co-expression network which was correlated with the maximal standardized uptake values (SUVmax). The least absolute shrinkage and selection operator (LASSO) was used for identification of genomic signature for the PET positivity and a prediction model was established. By using qRT-PCR, a gene panel (PredictionScore) based on the gene signature was developed. Results: We found that the PDXs could recapitulate FDG avidity of those parental tumors between 15 Patient-PDX pairs (Spearman r = 0.54, p-value = 0.04). The prediction model with the identified five genes ( PLS1, PYY, HBQ1, SLC6A5, NAT16) provided excellent prediction values compared with actual SUVmax for 15 patients as a validation set (Spearman r = 0.56, p-value = 0.03) and for 8 patients as a test set (Spearman r = 0.90, p-value = 0.005). The PredictionScore showed significant positive correlation with the actual SUVmax for 7 patients as an external validation set (Spearman r = 0.82, p-value = 0.03). Conclusions: PDX can be used to develop a gene panel for the PET positivity prediction in GC. Our results showed that the scoring system can be clinically applicable for developing a predicted stratification model. Future studies will aim to evaluate the panel for a higher number of PET-scanned GC patients to establish a rational patient selection for PET scan in clinical settings.