Part 3 · Lowering urate
Safety of urate-lowering drugs
Allopurinol is safe for most people, but a rare skin reaction can kill, and a gene test lowers that risk in some ancestries. Explore the numbers behind the reaction, the two big febuxostat heart trials, the interactions that matter and pegloticase's infusion rules.
This is a study aid, not a safety check for any person. Interactions and contraindications are reported as the US labels, UK and EU SmPCs, CPIC and ACR 2020 publish them. Decisions about a particular medicine belong with the prescriber. 64,107,127,129,135,370
Warning signs, as the labels state them. Labels say to stop allopurinol at the first sign of rash, and a person who has had a severe reaction must not take it again. 129,370 About 90% of hypersensitivity reactions begin within 60 days of starting. 326
1 · Allopurinol hypersensitivity and HLA-B*58:01
Schematic grid. 1972 dots, because 1 in 1972 people starting allopurinol in US Medicaid was hospitalised with a severe cutaneous reaction. 200 Carrier shares are the prevalences ACR 2020 quotes (Taiwan: Ko 2015). The one case is placed among carriers because 99% of tested cases in a 901-case literature series carried HLA-B*58:01. 135,214,326
| Population | HLA-B*58:01 allele frequency (CPIC weighted average) | Other published figures |
|---|---|---|
| East Asian | 6.0% (range 0.4–17%) | ACR: 7.4% prevalence in Han Chinese, Korean and Thai people; UK SmPC: up to 20% in Han Chinese, 8–15% Thai, about 12% Korean; 19.6% of Taiwanese allopurinol candidates carried it 135,214,370 |
| Sub-Saharan African | 4.4% | Not quoted separately by ACR |
| Central/South Asian | 4.2% | Not quoted separately by ACR |
| African American / Afro-Caribbean | 3.8% | ACR: 3.8% in African Americans 135 |
| Near Eastern | 1.9% | Not quoted separately by ACR |
| Latino | 1.4% | ACR: 0.7% in white and Hispanic people 135 |
| European | 0.77% | UK SmPC: 1–2% in Japanese or European people 370 |
CPIC frequencies are allele frequencies from its database. 64 ACR and the SmPC quote prevalence, which is not the same measure, so the columns do not match exactly.
How strong the gene link is
OR 580.3In Han Chinese, HLA-B*58:01 was present in all 51 people with allopurinol severe cutaneous reactions, against 15% of tolerant users. 173
In Thailand: 100% (27/27) of Stevens–Johnson/TEN cases vs 12.96% of tolerant controls (OR 348.3). Sensitivity 100%, specificity 87%; at 0.2% prevalence the positive predictive value is only 1.52% and the negative predictive value 100%. 398
What CPIC and ACR say
- CPIC: allopurinol is contraindicated in HLA-B*58:01 carriers (strong); non-carriers follow standard dosing (strong). The 2015 update left this unchanged. 64,161,347
- ACR 2020: test before allopurinol in people of Southeast Asian descent (for example Han Chinese, Korean, Thai) and African American people (conditional), and not universally in others; testing was cost-effective (ICER <$109,000/QALY). 135
- After a prior allergic reaction, desensitisation is conditionally recommended only when no other oral drug is possible (very low certainty). 135
Other risk factors
In US Medicaid, relative risk against white or Hispanic people was 3.00 in Black, 3.03 in Asian and 6.68 in Native Hawaiian/Pacific Islander people. Female sex (2.5-fold), age ≥60 (1.7-fold), CKD (2.3-fold) and a starting dose >100 mg/day (1.9-fold) added risk; older women of high-risk ancestry had RR 12.25. 200
In Europe, allopurinol was the commonest cause of SJS/TEN (adjusted OR 18), with risk confined to the first 8 weeks and far higher at ≥200 mg/day (OR 36 vs 3.0). 152
Who gets it, and how bad
21%of those hospitalised in US Medicaid died; the mean stay was 9.6 days. 200
In 901 published cases (1950–2012), all-cause mortality was 14%, 48% had renal impairment and 45% took diuretics. 326
2 · The starting-dose rule
eGFR in mL/min. The rule is the one Stamp 2012 proposed; the mg figure is its arithmetic, not a dose for anyone. 379
3 · Febuxostat and the heart: CARES vs FAST
Real trial results. CARES rates come from the US febuxostat label's account of the trial; FAST rates are on-treatment. The two trials measured different endpoints in different people, so the bars sit side by side, not on one scale of risk. 127,241,431
CARES (US-led)
6190 people with gout and major cardiovascular disease, median 32 months, non-inferiority margin HR 1.3. Major cardiovascular events: 10.8% vs 10.4% (HR 1.03), non-inferior. Cardiovascular death HR 1.34 (95% CI 1.03–1.73); all-cause death HR 1.22 (1.01–1.47). 56.6% stopped the trial drug and 45.0% stopped follow-up. 431
FAST (UK, Denmark, Sweden)
6128 people aged ≥60 with ≥1 cardiovascular risk factor, allopurinol first optimised to target, open label, median follow-up 1467 days. On-treatment primary endpoint 1.72 vs 2.05 per 100 patient-years (adjusted HR 0.85, 95% CI 0.70–1.03), non-inferior. Deaths 7.2% vs 8.6%. Only 6.2% vs 5.5% were lost from all follow-up. 241
What changed after CARES
- February 2019: the FDA added a boxed warning, "Cardiovascular death", and limited febuxostat to people with an inadequate response to maximally titrated allopurinol, intolerance, or when allopurinol is not advisable. 120,127
- ACR 2020 conditionally recommends switching people on febuxostat with a history of cardiovascular disease or a new event to another drug when possible. 135
- The EU SmPC cites both trials: more fatal cardiovascular events in CARES in people with major cardiovascular disease, none in FAST. 107
Why the trials disagree Early evidence
CARES lost 45% of follow-up; FAST lost about 6% but was open label and enrolled people already tolerating allopurinol. 241,431 In Japan, febuxostat mortality did not differ from placebo or allopurinol in FEATHER and FREED. 164,215
4 · Interactions
| Combination | What happens | What the labels say |
|---|---|---|
| Azathioprine or mercaptopurine + xanthine oxidase inhibitor | Both thiopurines are inactivated by xanthine oxidase, so blocking it raises their levels | US allopurinol label: with 300–600 mg/day, cut them to about one-third to one-quarter of the usual dose. UK SmPC: avoid; fatal cases reported. US febuxostat label: contraindicated. EU febuxostat: not recommended; if unavoidable, ≤20% of the previous dose 107,127,129,370 |
| Didanosine + allopurinol | Allopurinol 300 mg roughly doubles didanosine Cmax and AUC | US didanosine label: contraindicated. Neither febuxostat label lists didanosine 107,119,127,370 |
| Ampicillin or amoxicillin + allopurinol | More frequent rash; cause not established | US allopurinol label 129 |
| Theophylline + febuxostat | Mechanism not detailed in our sources | Use with caution 127 |
| Probenecid + methotrexate, penicillins or salicylates | Methotrexate and penicillin levels rise (penicillin 2- to 4-fold); salicylates block its uricosuric effect | US probenecid label; salicylates contraindicated 126 |
| Benzbromarone + warfarin | Benzbromarone inhibits CYP2C9 | Review 225 |
| Pegloticase + other PEGylated products | Anti-PEG antibodies may cross-react | US label 122 |
5 · Pegloticase: the infusion rules
G6PD first
Pegloticase is contraindicated in G6PD deficiency, because of haemolysis and methaemoglobinaemia; people at risk are screened before treatment. 122
Boxed warning and the stopping rule
Anaphylaxis and infusion reactions can happen with any infusion, usually within 2 hours. Serum urate is checked before every infusion, and treatment stops if it rises above 6 mg/dL, particularly after two consecutive values above 6. 122
Loss of the urate response came before 79% (44/56) of infusion reactions, so the pre-infusion urate works as a stand-in for the antibodies that cause them. 233
Methotrexate changed the numbers
71.0% vs 38.5%responded at month 6 with methotrexate vs placebo in MIRROR, and infusion reactions fell to 4.2% (4/96) from 30.6% (15/49). 27
Without methotrexate and without the stopping rule, anaphylaxis occurred in 6.5% on every-2-week dosing and infusion reactions in 26% vs 5% on placebo. 122
Flares at the start
74% (every 2 weeks) and 81% (every 4 weeks) flared in the first 3 months, against 51% on placebo, so prophylaxis is given for at least the first 6 months. 122
6 · The other drugs
Probenecid
"May not be effective" at GFR ≤30 mL/min; contraindicated with uric acid kidney stones. 126 ACR 2020 conditionally recommends against checking urinary uric acid or alkalinising the urine on a uricosuric. 135
Benzbromarone
Withdrawn by Sanofi-Synthélabo in 2003 after reports of liver toxicity; the estimated risk in Europe is about 1 in 17,000 patients, possibly higher in Japan. 225
Lesinurad
Boxed warning: "risk of acute renal failure, more common when used without a xanthine oxidase inhibitor". 128 In CLEAR 2, creatinine rose ≥1.5× baseline in 5.9% (200 mg) and 15.0% (400 mg) vs 3.4% on allopurinol alone. 16
Allopurinol and the kidney
Oxypurinol is cleared almost entirely by the kidney, so its levels rise in CKD. 87,154 Yet ACR 2020 still names allopurinol first line in CKD stage ≥3, started low and titrated. 135
7 · Pregnancy and breastfeeding, as labels state them
Allopurinol
UK SmPC: the safety evidence in pregnancy is "inadequate"; use only when there is no safer alternative. Breast milk contained 1.4 mg/L allopurinol and 53.7 mg/L oxipurinol at 300 mg/day, and use while breastfeeding "is not recommended". 370 The US label reports animal data and a case report of malformations, and advises caution in nursing. 129
In 31 prospectively followed first-trimester exposures the major malformation rate was 3.7%; the authors advise caution in the first trimester. 166
Febuxostat
US label: data are insufficient to inform risk, and nothing is known about human milk. 127 EU SmPC: it "should not be used during pregnancy". 107
Probenecid and pegloticase
Probenecid crosses the placenta and appears in cord blood. 126 Pegloticase has no adequate studies in pregnancy and should not be used while breastfeeding unless the benefit is clear. 122