Design: Meta-analysis of 33 studies plus haplotype analysis in two trials
Population: 17,485 patients treated with fluoropyrimidines
Primary outcome: Severe fluoropyrimidine-related toxicity
Effect: c.2194G>A OR 1.70 (1.48–1.95); c.496A>G OR 1.55 (1.19–2.02)

This study combined data from the PREPARE and Alpe-DPD trials with 31 published studies in a meta-analysis of 17,485 patients receiving fluoropyrimidines, and analysed three-locus haplotypes in 1296 trial participants. It was published in Pharmacological Research in September.

DPYD c.2194G>A (OR 1.70, 95% CI 1.48 to 1.95) and c.496A>G (OR 1.55, 1.19 to 2.02) were associated with severe toxicity; c.85T>C was not on its own but improved prediction in combination. A haplotype carrying c.496G without c.85C was consistently linked to severe and gastrointestinal toxicity, and plasma uracil levels confirmed reduced DPD function in these carriers.

Standard DPYD panels test four variants and explain only part of severe toxicity. These data support broader panels, but the added variants carry modest risk, and dose adjustment rules for them are not yet established.

Test for DPYD variants before starting 5-FU or capecitabine where testing is available.
Recognise that a negative standard panel does not exclude reduced DPD activity.
Monitor closely in the first cycle for mucositis, diarrhoea and neutropenia.
Consider phenotyping with plasma uracil where genotyping is unavailable.

Why it matters

It shows that current DPYD panels miss variants that measurably raise toxicity.

The statistics, in plain English

Odds ratios of 1.5 to 1.7 are modest; most carriers will not have severe toxicity. The haplotype results come from about 1300 patients in two trials and are wider and less certain.