HLA-B Genotype

Some people carry a version of one immune gene that turns an ordinary prescription into a life-threatening reaction. Allopurinol for gout. Abacavir for HIV. Carbamazepine for seizures. Dapsone for leprosy and some inflammatory skin conditions. In each case, a specific inherited version of the gene changes the risk enough that the result can alter the prescription before the first dose. For allopurinol and carbamazepine, the feared reaction can include skin blistering and detachment.

This test reads the sequence of the HLA-B gene. HLA stands for human leukocyte antigen. This is one of the most variable genes in the human genome, and the result reports the two HLA-B alleles you inherited. It's the same gene behind the HLA-B27 result people get when a doctor suspects ankylosing spondylitis, and behind the HLA-B51 result in Behçet's disease. Sequencing gives you the full HLA-B type at a level of detail older single-allele tests can't reach.

Drug Reactions That Can Be Prevented

This is where the result is strongest. Certain HLA-B variants can make a specific drug trigger severe cutaneous adverse reactions: Stevens-Johnson syndrome, toxic epidermal necrolysis, and DRESS. DRESS is a whole-body drug reaction involving rash, fever, blood changes, and organ inflammation. These are not ordinary rashes. In the worst form, the outer layer of skin detaches across large areas of the body and people are treated in burn units.

Taiwan ran the test that matters here. Of 2,926 people with an indication for allopurinol who had not taken it before, 2,910 were genotyped. The 571 who carried HLA-B*58:01, about one in five, were steered to a different plan. Among the 2,339 who tested negative and went on to take allopurinol, seven severe skin reactions were expected from historical rates. Zero occurred. That's the cleanest demonstration that knowing this genotype before treatment prevents the outcome rather than just predicting it.

Abacavir is the other well-worked example. Every immunologically confirmed case of abacavir hypersensitivity in a study of white and Black participants occurred in someone carrying HLA-B57:01, and the test correctly cleared 96 to 99 out of every 100 tolerant controls. The trade runs the other way too: roughly half of HLA-B57:01 carriers would not develop a confirmed reaction if they took the drug, so this catches every case at the cost of steering some people away who would have tolerated it. Pre-treatment screening for it is now standard practice. For dapsone, carrying HLA-B13:01 raised the odds of dapsone hypersensitivity syndrome dramatically: about 50 times in a Nepalese group of people with leprosy, and about 62 times when studies from several countries were pooled. Thai data and a meta-analysis point the same direction, with odds roughly 39 to 43 times higher. But about one quarter of dapsone cases in that Nepalese study were not carriers, so a negative result lowers risk rather than erases it. Several HLA-B variants including B39:01, B13:01, and B38:02 are also linked to severe skin reactions from sulfasalazine in Asian populations.

If you're about to start allopurinol, abacavir, dapsone, sulfasalazine, or carbamazepine, this result is useful before the first dose and too late to prevent a reaction after it has already happened. Published estimates of how many people you'd need to test to prevent one abacavir reaction vary with allele frequency and the endpoint used, from about 13 in one review to somewhere between 27 and 59 in others. For carbamazepine and HLA-B*15:02 it's 400 to 700 in Asian populations and around 60,000 in a Canadian one. Ancestry changes the math enormously, which is the single most useful thing to understand about interpreting this test.

Carbamazepine carries a second risk allele that sits outside this gene entirely. HLA-A31:01 raises the risk of carbamazepine hypersensitivity, including blistering reactions, DRESS, and ordinary drug rashes, across European, Japanese, Korean, and Chinese populations. A negative HLA-B15:02 result says nothing about it, and sequencing HLA-B does not read HLA-A. If carbamazepine is the drug in question, that second allele needs its own test.

Ankylosing Spondylitis and Spine Inflammation

HLA-B27 is the reason most people encounter this gene at all. It's the main inherited risk marker for ankylosing spondylitis and the related seronegative spondyloarthropathies, a group of conditions that inflame the spine and the joints where tendons attach to bone. Roughly 74 to 89% of people with axial spondyloarthritis carry it, and the odds of disease in carriers are more than 50 times those of non-carriers. If you've had years of inflammatory back pain that's worse in the morning and better with movement, and nobody has explained it, this is part of the workup.

Sequencing names the exact subtype. HLA-B27 is a family, not one sequence. Antibody-based flow cytometry can cross-react with close relatives such as HLA-B*07, and tag-SNP shortcut assays have missed about 2.4% of true carriers. Sequence-based or allele-specific methods sort this out more cleanly, especially when a result from an older method doesn't fit the symptoms.

Behçet's Disease and Its Look-Alikes

HLA-B51 is the primary genetic susceptibility marker for Behçet's disease, an inflammatory condition marked by recurrent mouth and genital ulcers and eye inflammation. In an Egyptian group, the B51:08 subtype was the most common susceptibility allele and carried a worse outlook for eye involvement. In Thai patients, B51:01:01 was among the associated variants.

Sequencing can also tell you when the HLA result is only part of the story. In one UK cohort of children and young adults with suspected Behçet's, broader exome sequencing found a single-gene inflammatory condition in 29%. Those conditions have different treatments and implications for relatives. A negative HLA-B51 result doesn't rule out Behçet's. In early-onset or atypical cases, it should lower the threshold to widen the search.

Infection and Immune Control

Because HLA-B decides which fragments of a virus your killer T cells get to see, the version you carry shapes how well you control some infections. With HIV, HLA-B57 and HLA-B27 are associated with lower viral levels and slower progression, while HLA-B46:01 is associated with faster progression in Thai and other Asian cohorts, likely through natural killer-cell biology rather than the usual T-cell route. HLA-B07:02, B14, and B51:01 have shown statistical associations with COVID-19 severity across pooled analyses, though those associations are weaker and less consistent than the drug-reaction ones.

Transplant Matching

If you or a family member is heading toward a stem cell or kidney transplant, HLA-B is one of the loci that decides compatibility. Mismatches raise the risk of graft rejection, graft-versus-host disease, and higher transplant mortality. The gain from sequencing here is resolution: in living kidney transplantation, high-resolution genomic typing has found mismatches older methods missed, which can change donor choice and rejection monitoring.

Why a Positive Result Is Not a Diagnosis

People often get this part wrong, and it matters. Carrying HLA-B27 or HLA-B51 does not mean you have, or will get, ankylosing spondylitis or Behçet's. These are risk variants, not disease genes. Most carriers never develop the associated condition: the lifetime risk of spondyloarthritis in a B27 carrier runs about 2 to 10%, and only around 6% develop ankylosing spondylitis itself. Disease risk usually comes from a mix of genes and environment, and we can't predict the full chain from one HLA allele. There are no trials showing that screening unselected healthy adults for these autoimmune-risk alleles improves outcomes.

The drug variants are different in kind. HLA-B58:01 or B57:01 doesn't make you sick on its own either, but the trigger is known, specific, and usually avoidable: don't take that drug. That's why the same result can be decisive for prescribing and only contextual for diagnosis. Treat drug-risk alleles as prescribing warnings. Treat disease-risk alleles as context for symptoms you already have.

Where the Result Can Mislead

Genetic results fail differently from blood tests. The things that distort this one:

• Ancestry changes what the result means: these variants are common in some populations and rare in others, and the same allele can be worth testing for in one background and barely worth it in another. Carbamazepine screening prevents one reaction per few hundred people tested in Asian populations and per tens of thousands in a Canadian one.
• Older methods give softer answers: antibody-based flow cytometry can cross-react with related alleles and produce false positives for HLA-B27; tag-SNP assays miss about one in 40 true carriers. A prior HLA-B27 result from one of those methods is not equivalent to a sequenced one.
• Imputed or direct-to-consumer genotypes are not sequenced genotypes: software that infers HLA type from other genetic data has a meaningful error rate, and it's higher in people who aren't of European ancestry. A high-resolution HLA type from an imputation tool should not be treated as the same product.
• A negative is limited to what was tested: sequencing HLA-B tells you nothing about HLA-A, HLA-C, the class II genes, or non-HLA drug risks. A clean HLA-B result before carbamazepine leaves HLA-A*31:01 unchecked, which is the allele that matters most in European, Japanese, Korean, and Chinese patients. If an assay reports only selected alleles, a negative result rules out only those alleles.
• Ambiguous HLA calls can happen: rare alleles, new alleles, and phase ambiguity can leave a report less clear than it looks. When the decision is high stakes, confirm the call with the laboratory or with a different validated method.
• Sample source can matter after transplant: after an allogeneic stem cell transplant, blood DNA may reflect the donor rather than your original genotype. Tumor sequencing is also not the same as inherited HLA typing. Use the specimen type the lab recommends for the clinical question.

You Only Need This Once

Your HLA-B sequence is fixed at conception and won't change. There's no trend to track, no retest schedule, and no value in repeating it unless you have reason to doubt the call itself, such as an ambiguous result or a prior type from an older antibody, SNP, or imputation method that you want confirmed by sequencing.

What does need ongoing attention is everything downstream. If you carry a drug-risk allele, the result belongs in your pharmacy record and your own notes, permanently, because the moment it matters is when a clinician who doesn't know you is writing a prescription. If you carry B27 and have joint or back symptoms, imaging and inflammatory markers are what change over time, not the genotype.

What to Do With an Unexpected Result

A drug-risk allele calls for a short, concrete list: record it, tell your pharmacy, and know which alternatives exist for the drug in question. For allopurinol, there are other urate-lowering options; for abacavir, there are other antiretrovirals. Because these substitutions involve prescriptions, this is one of the places where a clinician genuinely has to be in the loop, ideally before you need the drug rather than in the moment.

A disease-risk allele calls for a different pathway, driven by whether you have symptoms. B27-positive with inflammatory back pain or recurrent eye inflammation warrants a rheumatologist and imaging of the sacroiliac joints, plus blood tests that sort the picture out: rheumatoid factor and anti-CCP to separate rheumatoid arthritis from spondyloarthropathy, ANA to check for lupus or scleroderma, and inflammatory markers to gauge how active things are. B51-positive with recurrent oral and genital ulcers and eye inflammation warrants the same referral plus consideration of broader sequencing, since 29% of people in one suspected Behçet's cohort had a single-gene mimic. B27-positive with no symptoms at all warrants nothing except knowing it.

A genomics result can be misread. In one randomized trial in neonatology, trained non-genetics professionals made more clinically significant disclosure errors than genetic counselors. If your result is unexpected, ambiguous, or changes a major prescription or family decision, a genetic counselor is a reasonable half hour to spend.