rs75017182 — DPYD HapB3 (c.1129-5923C>G)

DPYD HapB3 — The Deep Intronic Splice Variant That Drives a Decreased-Function Haplotype

DPYD encodes dihydropyrimidine dehydrogenase (DPD), the rate-limiting enzyme¹¹ rate-limiting enzyme
DPD catabolizes 80-90% of administered 5-fluorouracil into inactive metabolites, preventing toxic accumulation in the bone marrow and gastrointestinal tract that breaks down fluoropyrimidine chemotherapy drugs. DPYD HapB3 is one of only four DPYD variants that CPIC designates as clinically actionable²² CPIC designates as clinically actionable
Alongside DPYD*2A (rs3918290), DPYD*13 (rs55886062), and c.2846A>T (rs67376798) — together these four variants form the standard pre-chemotherapy screening panel mandated by the European Medicines Agency since 2020 before fluoropyrimidine-based cancer treatment. Unlike the other three, HapB3 is not a single coding-sequence variant — it is a haplotype defined by several linked polymorphisms, of which the functional driver is the deep intronic variant c.1129-5923C>G tagged by rs75017182.

The Mechanism

rs75017182 sits 5,923 base pairs deep inside intron 10 of the DPYD gene, far from any exon boundary. At first glance, a deep intronic variant shouldn't matter — introns are spliced out of the mature mRNA and don't end up in the protein. What makes HapB3 extraordinary is that the C>G substitution creates a new, cryptic 5' splice donor site³³ the C>G substitution creates a new, cryptic 5' splice donor site
van Kuilenburg et al. used RT-PCR on patient-derived cells to show that this cryptic site captures splicing machinery about 25% of the time, producing an aberrant 44-nucleotide insertion between exons 10 and 11 inside the intron. Some of the time, the spliceosome uses this cryptic donor instead of the authentic one, and a 44-nucleotide fragment of intron 10 gets spliced into the mature mRNA as a new pseudo-exon. That insertion is not a multiple of three, so it shifts the reading frame and introduces a premature stop codon a few amino acids later. The result is a truncated, catalytically dead DPD protein from the affected allele.

Because DPYD sits on the minus strand of chromosome 1, the coding-strand C>G change that papers describe corresponds to a G>C change on the GRCh38 plus strand — which is how consumer genotyping chips and whole-genome sequencing files report it. The plus-strand reference allele at chr1:97579893 is G (the wild-type, non-HapB3 allele), and the C is the risk allele. This strand convention is the same as for the other DPYD variants in the GeneOps catalog.

The HapB3 haplotype as originally characterised by Amstutz et al. in 2009⁴⁴ Amstutz et al. in 2009
Swiss cancer patients sequenced across the full DPYD coding region; haplotype structure inferred from a dozen linked variants, of which c.1129-5923C>G and c.1236G>A were the two strongest toxicity predictors includes several additional linked SNPs (c.483+18G>A, c.680+139G>A, c.959-51T>G, and the synonymous c.1236G>A tag), but functional work by van Kuilenburg et al.⁵⁵ functional work by van Kuilenburg et al.
Minigene splicing assays, RT-PCR on patient RNA, and western blots of DPD protein all converged on rs75017182 as the single causal variant has consistently pointed to the deep intronic splice variant as the sole functional driver. The other haplotype variants are benign passengers — they just happen to travel together in European populations because of shared ancestry.

CPIC classifies HapB3 as a decreased-function allele with an activity score contribution of 0.5, reflecting the fact that the aberrant splicing is partial rather than complete. Heterozygous HapB3 carriers retain roughly 70-80% of normal DPD activity (one fully normal allele plus an allele producing a mix of functional and aberrantly-spliced transcripts), while homozygotes drop to roughly 50-60% — sufficient to cause severe toxicity on standard fluoropyrimidine dosing but not complete DPD deficiency.

The Evidence

HapB3 was first characterised in 2009⁶⁶ first characterised in 2009
Amstutz et al. Pharmacogenomics 2009, the first systematic DPYD haplotype assessment in cancer patients as a haplotype significantly associated with severe 5-fluorouracil toxicity in a Swiss cohort. The causal splice mechanism was nailed down by van Kuilenburg and colleagues in 2010⁷⁷ causal splice mechanism was nailed down by van Kuilenburg and colleagues in 2010
Intragenic deletions and a deep intronic mutation affecting pre-mRNA splicing in the dihydropyrimidine dehydrogenase gene as novel mechanisms causing 5-fluorouracil toxicity. Hum Genet 2010;128(5):529-38, who used RT-PCR on patient-derived cells to visualise the aberrantly spliced mRNA and quantify the fraction of transcripts carrying the 44-nt pseudo-exon insertion.

The definitive clinical evidence came from the 2015 Meulendijks meta-analysis⁸⁸ 2015 Meulendijks meta-analysis
Clinical relevance of DPYD variants c.1679T>G, c.1236G>A/HapB3, and c.1601G>A as predictors of severe fluoropyrimidine-associated toxicity: a systematic review and meta-analysis of individual patient data. Lancet Oncol 2015;16(16):1639-1650, which pooled individual patient data from 7,365 cancer patients across eight studies. HapB3 carriers (identified by the c.1236G>A tag) had a relative risk of 1.59 (95% CI 1.29-1.97, p<0.0001) for severe fluoropyrimidine toxicity, independent of the effects of DPYD*2A and c.2846A>T. The effect size is smaller than for the no-function alleles (DPYD*2A carriers have a ~25-fold mortality risk), which is why CPIC rates HapB3 as decreased-function (activity score 0.5) rather than no-function (activity score 0).

The landmark Alpe-DPD prospective trial by Henricks et al.⁹⁹ landmark Alpe-DPD prospective trial by Henricks et al.
Henricks LM, Lunenburg CATC, de Man FM, et al. DPYD genotype-guided dose individualisation of fluoropyrimidine therapy in patients with cancer: a prospective safety analysis. Lancet Oncol 2018;19(11):1459-1467 prospectively genotyped 1,103 Dutch cancer patients for all four CPIC priority variants before starting fluoropyrimidine chemotherapy and applied 50% initial dose reductions to DPYD variant carriers. HapB3 carriers treated with the reduced dose had severe toxicity rates of approximately 39% — still elevated above the ~23% non-carrier baseline, but dramatically lower than the ~73% rate observed historically in HapB3 carriers given standard doses. Critically, matched-pair survival analysis showed no loss of overall survival or progression-free survival in dose-reduced HapB3 carriers compared to non-carrier controls. This is the trial that established 50% as the evidence-based initial dose reduction for HapB3 and cemented it in the 2018 CPIC guideline update.

Homozygous HapB3 carriers were first systematically described by Meulendijks et al. in 2016¹⁰¹⁰ first systematically described by Meulendijks et al. in 2016
Meulendijks D, Henricks LM, van Kuilenburg ABP, et al. Patients homozygous for DPYD c.1129-5923C>G/haplotype B3 have partial DPD deficiency and require a dose reduction when treated with fluoropyrimidines. Cancer Chemother Pharmacol 2016;78(4):875-880. They have partial rather than complete DPD deficiency but still experience severe toxicity on standard doses and require dose reduction — at approximately 25% of standard rather than the 50% used for heterozygotes, because two aberrant alleles compound the splicing defect.

An Important Nuance — The HapB3 LD Is Not Perfect

A 2024 paper by Turner and colleagues¹¹¹¹ Turner and colleagues
Turner AJ et al. Updated DPYD HapB3 haplotype structure and implications for pharmacogenomic testing. Clin Transl Sci 2024;17(1):e13699 used large-scale population genomics data to show that the linkage between c.1236G>A (rs56038477, the easier-to-genotype synonymous tag) and c.1129-5923C>G (rs75017182, the causal splice variant) is close but not perfect. A small number of individuals in diverse populations carry the c.1236G>A tag without the underlying splice variant — and, more rarely, the reverse. Genotyping rs56038477 alone will therefore misclassify a small fraction of patients in both directions: some flagged as HapB3 carriers who are actually functionally normal, and some missed HapB3 carriers (with the causal rs75017182 splice variant but not the tag) who would develop severe toxicity if given standard doses.

The clinical implication is that rs75017182 itself should be directly genotyped for HapB3 classification whenever possible, rather than relying on the rs56038477 c.1236G>A tag. GeneOps stores and interprets both variants separately for this reason: a user with one variant but not the other sits in a different biological category than a user with the full haplotype. The supervisor-level compound action for HapB3 only fires when both rs75017182 and rs56038477 are on the same haplotype direction — capturing true HapB3 carriers while avoiding the classification errors Turner et al. highlight.

Practical Implications

If you carry one copy of the HapB3 splice variant, you have approximately 70-80% of normal DPD enzyme activity and are at elevated risk of severe, potentially life-threatening toxicity from standard-dose fluoropyrimidine chemotherapy. CPIC recommends starting any fluoropyrimidine at 50% of the standard dose and using therapeutic drug monitoring (plasma 5-FU concentration measurement) to guide cautious dose escalation from cycle 2 onward, typically reaching 65-85% of the standard dose by cycle 3 depending on tolerability. This guidance mirrors the Alpe-DPD trial protocol.

These drugs — 5-fluorouracil (5-FU), capecitabine (Xeloda), and tegafur¹²¹² 5-fluorouracil (5-FU), capecitabine (Xeloda), and tegafur
Backbone treatments for colorectal, breast, gastric, pancreatic, head-and-neck, and several other cancers; also used in topical form for actinic keratosis but topical exposure is not contraindicated in HapB3 carriers are the backbone of chemotherapy for many common cancers, so HapB3 genotyping has broad clinical relevance. Pre-treatment DPYD panel testing is now routine in much of Europe and is cost-effective by every published analysis: preventing one case of severe toxicity saves approximately $155,000-180,000 in hospital and rescue costs, versus roughly $160-250 for a targeted DPYD panel.

If you carry two copies of HapB3 (homozygous), you have partial DPD deficiency and still require dose reduction — typically starting at 25% of standard dose — with very close monitoring and therapeutic drug level measurement. Homozygous HapB3 is rare (roughly 1 in 2,500 in European populations) but has been associated with severe toxicity even at conventional "reduced" doses, so the starting dose for homozygotes must be lower than for heterozygotes.

Interactions

HapB3 is one of four CPIC-priority DPYD variants defining the DPD metabolizer phenotype via an activity score system¹³¹³ activity score system
Activity score 2.0 = normal metaboliser; 1.0-1.5 = intermediate metaboliser (heterozygous for a no-function or decreased-function allele); 0-0.5 = poor metaboliser (homozygous or compound heterozygous for actionable variants). HapB3 contributes 0.5 per allele — heterozygotes have an activity score of 1.5 (intermediate metaboliser, 50% dose reduction), while homozygotes score 1.0 (poor metaboliser edge, still dose-reduced but approaching the fluoropyrimidine-avoidance threshold). Compound heterozygotes carrying HapB3 alongside a no-function allele (*2A/rs3918290 or *13/rs55886062) drop to activity score 0.5 — functionally poor metabolisers — and are typically treated with fluoropyrimidine avoidance rather than dose reduction. Compound heterozygotes with c.2846A>T (rs67376798, also decreased-function) score 1.0 and require more aggressive dose reduction than single-variant heterozygotes.

The HapB3 haplotype itself is defined by linkage between rs75017182 (this variant, the causal splice site) and rs56038477 (c.1236G>A, the synonymous exon 11 tag). A "true HapB3" carrier has both variants on the same chromosome. GeneOps treats the two SNPs as independent entries in the catalog because, per the Turner 2024 paper, their linkage is imperfect — isolated rs56038477 without the splice variant is a functionally normal state, and isolated rs75017182 carriers exist but are rare. The compound action for the HapB3 phenotype fires only when both variants are present, capturing the biologically real haplotype while avoiding misclassification of the LD breakdown cases.

First-degree relatives of HapB3 carriers have a ~50% prior probability of carrying the variant themselves and should be offered DPYD panel testing if they ever face fluoropyrimidine-based cancer treatment. Because HapB3 is the most common of the four CPIC-priority DPYD variants in European populations (2-4% allele frequency, versus 0.5-1% for DPYD*2A and c.2846A>T, and 0.1% for DPYD*13), it is also the variant most likely to be shared within a family.