Abstract 42P
Background
DPD (dihydropyrimidine dehydrogenase, encoded by DPYD) is the rate-limiting enzyme for fluoropyrimidine catabolism and a key enzyme for the metabolism of chemotherapy drugs such as fluorouracil and capecitabin. Testing for certain DPYD variants is recommended by international clinical guidelines to predict toxicity in patients and adjust treatment accordingly. A consensus has been reached to screen for four DPYD variants, namely c.1905+1G>A, c.1679T>G, c.2846A>T and c.1236G>A. Considering all four variants combined, around 7% of Europeans carry at least one decreased-function DPYD variant. Testing for DPYD variants is done using germline DNA, however, considering the high frequency in which capecitabine is used in breast cancer treatment and the high number of primary tumors for which RNAseq is performed for diagnostic purposes, we hypothesised that DPYD status could be read from tumor RNAseq and evaluated sensitivity and specificity compared to blood-DNA genotyping.
Methods
We studied a SCAN-B cohort of over 6000 breast tumors with RNAseq data, for which over 1700 patients also had available OncoArray germline genotyping. Since c.1905+1G>A is located in an intronic splice-site, an RNAseq model was developed to infer this SNP using exon-skipping events and ratios of supporting reads.
Results
In preliminary results, there was no case with c.1679T>G variant in either RNAseq data or OncoArray. Variant c.2846A>T was not included in OncoArray but 20 cases were identified by RNAseq. For variant c.1236G>A, RNAseq data analysis showed 67 true positive (TP) cases, 1660 true negative (TN) cases, 1 false positive case (FP), and 5 false negative (FN) cases. For the variant call c.1905+1G>A, RNAseq data analysis showed 28 TP, 1692 TN, 0 FP and 7 FN cases. Overall, using an estimated DPYD SNP prevalence of 7%, the positive predictive value (PPV) and negative predictive value (NPV) were both preliminarily >99%.
Conclusions
DPYD variant calling is feasible using RNAseq reads. If RNAseq from breast tumor tissue is available, it may be possible to accurately predict toxicity to 5-FU-containing regimes and thus forgo the need for further specific genomic assays, thus optimizing the diagnostic techniques.
Legal entity responsible for the study
The authors.
Funding
Mrs. Berta Kamprad Foundation.
Disclosure
All authors have declared no conflicts of interest.