From Variant to Visit: How Genomics Actually Changes Management

Genomic testing changes care when it moves us from “treat the average” to treat this individual in a way that alters a drug, a dose, or how closely we watch for disease.

The genome is vast, but the number of variants that actually rewrite a clinical decision today is surprisingly focused. This post walks through three concrete scenarios where a genotype changes what happens in the exam room — and why that distinction matters more than the variant count on a report.

Why “actionable” really matters

Genomics is clinically meaningful when three things line up:

• The variant has a clear, reproducible effect on biology. It isn’t a variant of uncertain significance; the mechanism is understood and validated across populations.
• There is more than one reasonable management option. If every patient gets the same therapy regardless, the variant adds information but not a decision.
• Choosing differently changes outcomes we care about — events avoided, toxicity prevented, cancer caught at an earlier stage, or cost reduced.

In other words, a variant is useful when it helps you pick the right drug, the right intensity, or the right surveillance window for this person rather than their risk group. That is the bar every genomic finding should clear before it reaches a clinician’s screen.

Scenario 1 — Changing the drug, not the diagnosis

Pharmacogenomics is the clearest example of genomics directly rewriting a prescription. The diagnosis stays the same; the molecule or its dose is different because the genome says the “standard” choice is unsafe or ineffective.

CYP2C19 and clopidogrel

Roughly 30% of the general population carries at least one CYP2C19 loss-of-function allele, and 2–5% of Caucasians are poor metabolisers who barely activate clopidogrel at all. In East Asian populations the frequency is substantially higher, with poor-metaboliser rates reaching 12–23%.

The clinical signal is strong: a 2010 meta-analysis found carriers of even one loss-of-function allele had approximately 53% higher risk of major adverse cardiovascular events on clopidogrel after PCI (Mega et al., NEJM 2010). Stent thrombosis risk was 2–3-fold higher in poor metabolisers.

The POPular Genetics trial (Claassens et al., NEJM 2019) demonstrated that a genotype-guided strategy — switching carriers to ticagrelor or prasugrel — was noninferior to blanket ticagrelor use while reducing bleeding events. CPIC guidelines now give a strong recommendation: use an alternative P2Y12 inhibitor in intermediate and poor metabolisers undergoing PCI.

HLA-B*57:01 and abacavir

A single HLA allele predicts abacavir hypersensitivity with near-complete clinical sensitivity. The PREDICT-1 trial (Mallal et al., NEJM 2008) showed that prospective screening reduced immunologically confirmed hypersensitivity reactions from 2.7% to effectively zero in HLA-B*57:01-negative patients.

Today, every major HIV guideline — FDA labelling, DHHS, EACS, WHO — mandates HLA-B*57:01 testing before prescribing abacavir. It is one of the clearest success stories in pharmacogenomic implementation: a simple test, a binary decision, and a life-threatening reaction essentially eliminated.

UGT1A1*28 and irinotecan

Homozygous UGT1A1*28 carriers — roughly 10–15% of Caucasians — have reduced glucuronidation capacity, pushing them toward markedly higher rates of severe neutropenia and diarrhoea at standard irinotecan doses. CPIC and DPWG guidelines recommend a 20–30% starting-dose reduction for these patients, with protocols varying by regimen intensity.

In each case above, the diagnosis is unchanged. What differs is the molecule, or its dose, because the genome flags the “standard” choice as unsafe or ineffective for this individual.

Scenario 2 — Redefining “how closely we watch”

Inherited cancer genes turn population screening algorithms on their head.

BRCA1/2 and breast cancer surveillance

Pathogenic BRCA1 variants shift lifetime breast cancer risk to approximately 55–72%, with ovarian cancer risk of 36–44%. BRCA2 carriers face a 45–69% breast cancer risk and 11–17% ovarian cancer risk (Kuchenbaecker et al., JAMA 2017). These numbers dwarf the general-population lifetime risk of roughly 12%.

NCCN guidelines respond by building an entirely different surveillance pathway:

• Clinical breast exams every 6–12 months starting at age 25
• Annual breast MRI with contrast from age 25–29
• Annual mammography added from age 30, alternating with MRI so imaging occurs every six months
• Discussion of risk-reducing mastectomy
• Risk-reducing salpingo-oophorectomy between ages 35–40 for BRCA1 and 40–45 for BRCA2

The rationale for MRI is quantitative: mammography sensitivity in BRCA carriers drops to roughly 25–59% (younger, denser tissue; tumour biology that mimics benign lesions), while MRI sensitivity reaches 77–95%. Combined screening pushes sensitivity to 93–100%.

Here, genomic information doesn’t just label someone “high risk.” It pulls them onto a completely different surveillance pathway — different modalities, different start ages, and a qualitatively different conversation about prophylactic surgery that population-level screening never triggers.

Scenario 3 — Escalating therapy in “silent” high-risk states

Some monogenic conditions justify more aggressive treatment targets than phenotype alone would suggest.

Familial hypercholesterolaemia

Heterozygous FH affects roughly 1 in 250 people — making it one of the most common monogenic disorders — yet more than 90% of cases worldwide remain undiagnosed. Without treatment, approximately 50% of men with FH will have a coronary event by age 50.

Identifying a pathogenic LDLR, APOB, or PCSK9 variant reframes “primary prevention” as a coronary event waiting to happen. ESC/EAS guidelines set an LDL-C target below 100 mg/dL for uncomplicated FH and below 55 mg/dL when additional risk factors or established ASCVD are present — often requiring high-intensity statins, ezetimibe, and PCSK9 inhibitors.

The long game matters: a landmark 20-year Dutch follow-up (Luirink et al., NEJM 2019) showed that statin therapy initiated in childhood reduced cardiovascular events by approximately 76% compared to affected parents at the same age. Early treatment essentially normalises risk.

Cascade screening — testing first-degree relatives of a known FH proband — identifies a new case for roughly every 2–3 relatives tested, consistent with autosomal dominant inheritance. Cost-effectiveness analyses consistently show it falls well below standard willingness-to-pay thresholds. Yet most health systems still rely on phenotypic criteria that miss the majority of carriers.

Here, genomics converts a borderline-looking lipid profile into a mandate for lifelong, intensive risk-factor modification and systematic screening of an entire family tree.

From reports to decisions — what “clinical-grade” should promise

For patients and clinicians, the value of genomics lies in a small number of clear, defensible decisions — not in pages of variants. The PREPARE study (Swen et al., Lancet 2023) demonstrated this at scale: preemptive 12-gene pharmacogenomic panel testing across seven European countries reduced clinically significant adverse drug reactions by roughly 30%.

A clinically useful genomic report should:

• Distinguish variants that change a drug, dose, or surveillance plan today from background genetic noise.
• Anchor recommendations in recognised guidelines — CPIC, NCCN, ESC/EAS, regulatory labels — with explicit “if this genotype, then this action” clarity.
• Fit into real workflows so the right therapy or surveillance pathway can be chosen in minutes, not after a literature review.

This is the gap platforms like Purna aim to close: turning complex genomic data into a handful of precise, actionable management choices for the clinician sitting in front of a real patient. Not more data — better decisions.

Variant interpretation is one layer of what Purna calls molecular intelligence — the broader application of AI to understand and predict molecular behavior across genomics, proteomics, and drug discovery. On the structural biology side, advances in protein folding and generative AI are opening new therapeutic modalities that complement the pharmacogenomic approaches discussed above. Researchers exploring these workflows can apply for free MIP credits to run analyses on Purna’s platform.