Dataset Information

Drug-drug interactions in subjects enrolled in SWOG trials of oral chemotherapy.

ABSTRACT:

Background

Patients with cancer are at increased risk of drug-drug interactions (DDI), which can increase treatment toxicity or decrease efficacy. It is especially important to thoroughly screen DDI in oncology clinical trial subjects to ensure trial subject safety and data accuracy. This study determined the prevalence of potential DDI involving oral anti-cancer trial agents in subjects enrolled in two SWOG clinical trials.

Methods

Completed SWOG clinical trials of commercially available agents with possible DDI that had complete concomitant medication information available at enrollment were included. Screening for DDI was conducted through three methods: protocol-guided screening, Lexicomp® screening, and pharmacist determination of clinical relevance. Descriptive statistics were calculated.

Results

SWOG trials S0711 (dasatinib, n = 83) and S0528 (everolimus/lapatinib, n = 84) were included. Subjects received an average of 6.6 medications (standard deviation = 4.9, range 0-29) at enrollment. Based on the clinical trial protocols, at enrollment 18.6% (31/167) of subjects had a DDI and 12.0% (20/167) had a DDI that violated a protocol exclusion criterion. According to Lexicomp®, 28.7% of subjects (48/167) had a DDI classified as moderate or worse, whereas pharmacist review indicated that 7.2% of subjects (12/167) had a clinically relevant interaction. The majority of clinically relevant DDI identified were due to the coadministration of acid suppression therapies with dasatinib (83.3%, 10/12).

Conclusions

The high DDI prevalence in subjects enrolled on SWOG clinical trials, including a high prevalence that violate trial exclusion criteria, support the need for improved processes for DDI screening to ensure trial subject safety and trial data accuracy.

SUBMITTER: Marcath LA

PROVIDER: S-EPMC7995697 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Purpose: muscle transcriptomics of subjects supplemented with Urolithin A at different doses or placebo for 4 month. Method: RNA-seq was performed using Illumina HiSeq 4000 sequencing; single read 1 x 50 bp . The quantification of mRNA from the RNA-seq FASTQ files was performed using Salmon. Sample-wise quant.sf files containing raw transcript-level read estimates were read into R, v. 4.0.3 and were combined into a data matrix. Transcripts with very low total counts (< 10) across all samples were filtered out. The data was transformed using the variance stabilizing transformation (VST) method of R package DESeq2, v. 1.30.0. Top 10,000 transcripts with the highest variance across all samples were used for principal component analysis (PCA) using DESeq2. Data transformation and PCA was also done separately for each treatment group. Based on the PCAs, probable outlier samples were excluded and new PCAs were plotted without these samples. The raw transcript-level read count estimates were read in R and summarized to gene-level counts based on the provided transcript and gene ID annotations using summarizeToGene function of R package tximport, v. 1.18.0. DESeqDataSetFromTximport function of DESeq2 was then used for constructing a DESeqDataSet object for DE analysis. Pre-filtering was applied before the DE analysis by excluding genes with < 10 total counts across samples. Subset DE analysis was performed, contrasting Visit time D120 with baseline (BL) and by adjusting for the subject effect. The normalization and DE analysis was done separately for the three different treatment groups. Independent filtering option of DESeq2 was enabled (default), filtering out genes with very low counts and thus unlikely to show significant evidence. R package biomaRt. v. 2.46.0 was used for annotating the results with HGNC gene symbols, gene descriptions and gene biotypes. DESeq2-normalised expression values of all the samples in the given comparison were added to the result tables. Non-adjusted p-value 0.05 was used to filter the results by statistical significance. Results were also generated using DESeq2 function lfcShrink that allows for the shrinkage of the log2 fold change (LFC) estimates toward zero when the information for a gene is low (such as in those cases with low counts or high dispersion values) but has little effect on genes with high counts. The shrinked log2FC values were subsequently used for visualisation and ranking the genes.