Visit Website →
← Back to blog

Encode DPYD Dosing in EHRs: The 8 Fields Every Lab Must Store

October 5, 2026
Encode DPYD Dosing in EHRs: The 8 Fields Every Lab Must Store

A DPYD dosing guidance deliverable in a laboratory PGx report is a discrete, machine-readable record: observed variants, normalized allele or haplotype, activity score, predicted phenotype, an evidence-graded dosing recommendation, the guideline source and version, and a documented reviewer status. A PDF alone cannot do this job. Your system needs structured fields that an EHR or CDS engine can consume, with assay validation kept separate from clinical-rule logic and every rule version tracked over time.

TL;DR:

  • Discrete, structured DPYD result records are essential for automation, including variant details, activity score, phenotype, guideline version, and review status, rather than relying on PDFs.
  • The activity score mapping to phenotype can change over time, requiring versioned rules and timestamped updates to ensure accurate dosing guidance.
  • Validating the assay and clinical rules separately ensures proper allele coverage and consistent guideline application, with distinct checklists for each validation step.
  • Reports should clearly communicate that DPYD dosing recommendations are starting points, accounting for other clinical factors and potential guideline updates.
  • Living reanalysis processes are necessary to identify and review previously issued reports when guidelines or allele interpretations are revised.
SignalPGx
signalpgx.com
Bring Structured PGx Into EHR Workflows
SignalPGx turns genotype, medication, and clinical evidence into physician-reviewed reports with seamless HL7 and FHIR integration.
Visit SignalPGx

Table of Contents

How do CPIC and DPWG map genotype to dosing recommendations?

Guideline bodies assign a numeric activity value, typically 0, 0.5, or 1, to each DPYD allele, and the summed activity score across both alleles determines the predicted phenotype. A normal metabolizer generally receives standard fluoropyrimidine dosing, an intermediate metabolizer a reduced starting dose, and a poor metabolizer either a substantial reduction or an alternative recommendation, following the CPIC guideline for DPYD and fluoropyrimidines.

This mapping is not static. CPIC has revised allele value assignments for haplotypes like HapB3, which changes which patients fall into which phenotype bucket and what starting dose a report should suggest. Your rules engine has to version these mappings and timestamp when each version went live, because a report issued last year under one rule set may no longer match current guidance.

A few principles should guide how your software presents this:

  • Treat every dosing recommendation as a starting point, not a fixed prescription, since clinicians titrate based on tolerance and response.
  • Record which guideline source and version produced each recommendation, down to the effective date.
  • Flag haplotypes with known normalization complexity, such as HapB3, for extra scrutiny during rule updates.

What data model and FHIR resources represent DPYD results?

A computable DPYD result needs a defined minimum field set, and the structured record, not the rendered PDF, should be your system-of-record. Labs that treat the PDF as the primary artifact lose the ability to drive CDS, because downstream systems cannot reliably parse free text.

The minimum computable fields include:

  1. Observed variant(s), transcript, and genome build.
  2. Normalized allele or haplotype designation used for scoring.
  3. Zygosity and activity score.
  4. Predicted phenotype (normal, intermediate, or poor metabolizer).
  5. Structured dosing recommendation, including any titration guidance.
  6. Guideline source, version, and effective date.
  7. Internal rule identifier tied to the recommendation.
  8. Reviewer identity and review status.

The HL7 Genomics Reporting Implementation Guide describes patterns for encoding these elements using TherapeuticImplication, MedicationRecommendation and PlanDefinition, RiskAssessment, and Observation components, so that a receiving EHR can act on the result rather than merely display it.

Pro Tip: Store the FHIR/JSON payload as your authoritative record and generate the clinician-facing PDF from it, never the reverse.

Validating the assay separately from the clinical dosing rules

Assay validation and clinical-rule validation answer different questions, and conflating them is a common source of downstream errors. The AMP consensus on DPYD clinical testing specifies allele selection and assay guidance but explicitly excludes genotype to phenotype correlation and medication-management recommendations, so your team needs separate validation evidence for each layer.

Assay validation should confirm:

  • Allele coverage against the panel's intended use, with documented limitations.
  • Reference material testing across known genotypes, including rare alleles.
  • Concordance testing against an orthogonal method or reference laboratory.
  • Alignment with ISO 15189 requirements for method validation and documentation.

Clinical-rule validation should confirm separately:

  • Rule provenance, tracing each dosing output to a specific guideline version.
  • Unit tests covering every diplotype combination, including compound heterozygotes.
  • Correct haplotype normalization, particularly for HapB3-type cases where multiple assay variants represent one allele.
  • End-to-end verification: simulated cases pushed through EHR ingestion with confirmation that CDS triggers fire as designed.

Running these as two distinct checklists, rather than one combined sign-off, gives your medical director a clearer audit trail when a guideline changes and only the clinical-rule layer needs revalidation.

Writing the clinician-facing report so recommendations stay actionable

The clinician-facing report has to compress a complex rule chain into something a prescriber can act on in seconds, while still preserving the full computable record behind it. Genomics England's reporting guidance requires that DPYD reports include the genotype, the predicted effect on DPD production, and a dosing recommendation, with report production treated as part of laboratory quality assurance under ISO 15189:2012 reporting guidance.

A minimal clinician-facing layout should include:

  • A genotype table showing the variants detected and the normalized diplotype.
  • Activity score and predicted phenotype, stated plainly.
  • A concise dosing recommendation with titration guidance where relevant.
  • The guideline citation and version used to generate the recommendation.
  • Laboratory contact information and the review status of the report.

Activity scores derived from CPIC's allele value system map directly to the normal, intermediate, and poor metabolizer categories that drive these dosing statements, according to the CPIC guideline for DPYD and fluoropyrimidines. Phrasing should frame each recommendation as a governed starting point, reference therapeutic drug monitoring or clinician-directed titration where appropriate, and show the effective date of the guideline version used, so a prescriber reading the report six months from now understands exactly how current the recommendation is.

Governance, versioning, and living reanalysis

Dosing rules change, and your governance model has to assume that from day one rather than treat updates as exceptions. Every rule artifact needs a version number, an effective date, a change log entry, and a medical-director approval before it goes live, which gives you an auditable record of exactly which rule set produced any given report.

Living reanalysis takes this further: when a guideline body updates an allele value or a dosing threshold, your system should identify every previously issued report affected by that change and route them into a review queue, ideally with some scoring of likely patient impact rather than a blanket reprocessing of everything at once.

  • Version every rule artifact with an effective date and a readable change log.
  • Require medical-director sign-off before a rule version moves to production.
  • Route previously issued, now-affected reports into a prioritized review queue.
  • Record full provenance on every recommendation so a clinician can trace it back to the exact guideline and rule version used.

Pro Tip: Build your change log so a reviewer can answer "what changed and who did it affect" in minutes, not days.

A phased plan for standing up DPYD dosing guidance

Implementation goes smoother when ownership is explicit from the start. Assign a molecular lead for assay matters, a clinical lead or medical director for dosing-rule sign-off, an informatics or IT lead for integration, and a QA lead for validation records. Overlap between these roles is where errors tend to hide.

  1. Plan the project and complete assay validation, including reference materials and concordance testing.
  2. Encode the dosing rules as versioned artifacts and write unit tests for every diplotype, including compound heterozygotes.
  3. Run integration tests against your EHR and confirm CDS triggers fire as intended.
  4. Pilot with a limited case set and review outcomes with your medical director.
  5. Deploy at scale with ongoing monitoring and a living reanalysis process in place.

Acceptance criteria should be concrete before you call any phase complete: documented pass and fail logs for sample cases, confirmed CDS trigger behavior in a test EHR environment, and report templates that meet ISO 15189 documentation requirements. The implementation guide for DPYD genotyping in clinical practice recommends storing results discretely in the EHR specifically so post-test CDS can run without manual transcription, and designing alerts around activity score rather than raw genotype, which keeps the trigger logic simpler to maintain.

Clinical considerations and contraindications when applying DPYD dosing guidance

A DPYD dosing recommendation is only one input into a prescribing decision, and your report should make that boundary clear rather than presenting the recommendation as a complete clinical verdict. Complete DPYD deficiency, when identified, generally points toward avoiding the fluoropyrimidine entirely rather than adjusting the dose, and your rule logic needs a distinct pathway for that scenario rather than forcing it through the same reduction math used for partial deficiency.

Reports should also account for factors outside the genotype itself: concurrent use of other medications that affect fluoropyrimidine metabolism, hepatic or renal function, and prior treatment tolerance all influence the final prescribing decision in ways a genotype-only recommendation cannot capture. Your report wording should state this explicitly, framing the genetic result as one input that a prescriber integrates with the full clinical picture rather than a standalone directive.

Rare or novel variants present a separate challenge. When an observed variant has not been characterized by CPIC or DPWG, the rule engine should default to flagging the result for manual clinical review rather than guessing at an activity value, and the report should say so plainly instead of implying a confidence level the evidence does not support. Building this uncertainty handling into the rule logic from the outset, rather than retrofitting it after a near-miss, keeps the report defensible when it reaches a tumor board or a pharmacy review.

Management strategies for intermediate and poor DPD metabolism phenotypes

Intermediate metabolizers typically warrant a reduced starting dose with close monitoring during the early treatment cycles, while poor metabolizers often call for a more substantial reduction or consideration of an alternative agent, following the activity score framework in the CPIC guideline for DPYD and fluoropyrimidines. Your software should encode these as distinct recommendation pathways rather than a single sliding-scale formula, since the clinical reasoning and monitoring intensity differ meaningfully between the two groups.

For both groups, the report should prompt for a defined monitoring plan rather than leaving titration entirely to clinical judgment with no structure. That means referencing therapeutic drug monitoring where it applies, suggesting an interval for toxicity assessment after the first cycle, and making clear that any subsequent dose change should be documented and, where your workflow supports it, fed back into the patient's record alongside the original genotype-based recommendation.

Where multiple variants or a compound heterozygous result are involved, the normalized haplotype used to compute the activity score needs to be traceable back to the individual observed variants, consistent with the variant normalization guidance Genomics England provides for DPYD reporting. This traceability matters most for intermediate and poor metabolizer cases, where a scoring error has the largest clinical consequence and where a reviewing clinician is most likely to ask how the system arrived at its number.

Management strategies for intermediate and poor DPD metabolism phenotypes — overview diagram

Dose escalation and de-escalation protocols tied to genotype

De-escalation protocols in a DPYD context generally start from the activity-score-derived phenotype and apply a defined percentage reduction or an alternative-agent recommendation, rather than a single universal cut applied regardless of genotype severity. Your rule engine should encode these as distinct tiers tied to activity score ranges, each with its own guideline citation and version, so a report can state not just the recommended dose but the specific evidence tier behind it.

Escalation protocols matter too, particularly for patients who tolerate an initial reduced dose well and whose care team wants to consider a cautious increase in later cycles. Software supporting this should distinguish clearly between the genotype-based starting recommendation and any subsequent clinician-directed titration, keeping both in the discrete record so the report never implies that a later escalation was itself a genotype-driven recommendation rather than a clinical judgment.

Every escalation or de-escalation pathway in your rules needs the same versioning discipline as the initial dosing recommendation: a rule identifier, a guideline source and version, and an effective date. When CPIC or DPWG revises a threshold, as has happened with HapB3-related allele values, your living reanalysis process should be able to identify which previously issued reports used the superseded tier and route them for review, rather than leaving that determination to manual chart review.

Versioned DPYD dosing rule review workflow

Patient counseling points regarding DPYD testing and dose adjustments

Patients benefit from understanding, in plain terms, what a DPYD result means before treatment starts: the test identifies genetic variation that affects how the body processes fluoropyrimidine chemotherapy drugs, and the result informs a starting dose rather than predicting treatment outcome. Framing it this way helps set realistic expectations and distinguishes the genetic test from a general prognosis.

Counseling should also address timing. Because DPYD results typically take time to return, patients and care teams benefit from understanding upfront whether treatment will wait for results or proceed with interim dosing, and from knowing the result applies specifically to fluoropyrimidine drugs rather than treatment broadly. Clinician-facing reports that include a brief counseling-oriented summary alongside the technical data give prescribers language they can reuse directly with patients.

Finally, patients should understand that a DPYD result is not static in its clinical interpretation: guideline bodies periodically revise dosing thresholds, and a living reanalysis process means a report issued today could, in principle, be revisited if the underlying guidance changes. Communicating this possibility in plain language, without overstating how often it happens, keeps expectations grounded and gives patients a reason to maintain contact with their care team if guidance evolves.

What implementation teams consistently get wrong

The single biggest priority is discrete data and interoperability. Teams that ship PDF-only reports end up rebuilding the entire pipeline later to support CDS, because free text was never meant to drive automated triggers. A close second mistake is hard-coding dosing tables instead of building versioned, auditable rules with clinician-titration language built in. Living reanalysis is what separates a report that stays trustworthy from one that quietly goes stale.

— Tarek

How SignalPGx supports DPYD dosing guidance implementation

Building discrete, versioned, interoperable DPYD dosing logic in-house takes real engineering time away from your core laboratory work, and that is the gap our platform at SignalPGx is built to close. Our system outputs discrete genotype and phenotype data alongside compatible records, so your reports integrate into EHR and CDS workflows rather than sitting as static PDFs.

SignalPGx

Our rules run on versioned, auditable logic with living reanalysis that flags previously issued reports when a guideline update changes a recommendation, and every report carries a review workflow before it reaches a clinician. For labs evaluating a path forward, our White-Label PGx Reporting page outlines the Pilot, Standard, Scale, and Enterprise plans, and our Discovery & Strategy and Deployment & Integration services walk through technical documentation and a demo for teams ready to see the pipeline in action.

FAQ

What fields belong in a computable DPYD dosing record?

A computable record needs the observed variant and transcript build, the normalized allele or haplotype, zygosity, activity score, predicted phenotype, a structured dosing recommendation, the guideline source and version, and reviewer status. The HL7 Genomics Reporting Implementation Guide outlines FHIR resources, including MedicationRecommendation and PlanDefinition, suited to representing these fields.

How does activity score determine a DPYD dosing recommendation?

Each DPYD allele receives a numeric activity value under the CPIC guideline for DPYD and fluoropyrimidines, and the summed score across both alleles maps to a predicted phenotype that drives the starting dose recommendation. Normal metabolizers generally receive standard dosing, while intermediate and poor metabolizers receive reduced doses or alternative-agent guidance.

Why should DPYD results be stored as discrete data rather than PDF reports?

Discrete, structured results let an EHR trigger CDS alerts automatically, while PDF-only reports require manual transcription that introduces delay and error. The implementation guide for DPYD genotyping in clinical practice recommends discrete EHR storage specifically to support post-test clinical decision support.

What should a lab validate separately for DPYD assay versus dosing rules?

Assay validation covers allele coverage, reference materials, and concordance testing, while clinical-rule validation covers rule provenance, diplotype unit tests, and haplotype normalization checks. The AMP consensus on DPYD clinical testing specifies that assay guidance does not itself extend to genotype-to-phenotype correlation or dosing recommendations, so labs need distinct validation evidence for each layer.

Does SignalPGx provide DPYD dosing recommendations directly in reports?

Our platform generates evidence-graded PGx reports that include DPYD genotype, activity score, phenotype, and dosing recommendation fields, built on discrete data suited to EHR and CDS integration. Details on plan structure are available on our White-Label PGx Reporting page.

Sources

For teams building or auditing a DPYD dosing pipeline, these primary sources cover the guideline logic, reporting standards, and interoperability patterns referenced throughout this guide: the CPIC allele-to-activity-score mapping, Genomics England's validation and reporting requirements, the HL7 FHIR implementation patterns for representing pharmacogenomic recommendations, and a peer-reviewed implementation report describing an end-to-end PGx reporting pipeline.