ARTHRITIS & RHEUMATISMVol. 64, No. 8, August 2012, pp 25292536DOI 10.1002/art.34488 2012, American College of Rheumatology

Starting Dose Is a Risk Factor forAllopurinol Hypersensitivity Syndrome

A Proposed Safe Starting Dose of Allopurinol

Lisa K. Stamp,1 William J. Taylor,2 Peter B. Jones,3 Jo L. Dockerty,4 Jill Drake,1

Christopher Frampton,1 and Nicola Dalbeth5

Objective. Allopurinol is the most commonly usedurate-lowering therapy in gout. Allopurinol hypersensi-tivity syndrome (AHS) is a rare but potentially fataladverse event. Dosing guidelines based on creatinineclearance have been proposed based on the recognitionthat dosages of >300 mg/day may be associated withAHS, particularly in patients with renal impairment.However, the relationship between the allopurinol start-ing dose and AHS is unknown. This study was under-taken to determine the relationship between allopurinoldosing and AHS.

Methods. A retrospective casecontrol study ofpatients with gout who developed AHS between January1998 and September 2010 was undertaken. For eachcase, 3 controls with gout who were receiving allopurinolbut did not develop AHS were identified. Controls werematched with cases for sex, diuretic use at the time ofinitiating allopurinol, age (10 years), and estimatedglomerular filtration rate (estimated GFR). Startingdose and dose at the time of the reaction in cases werecompared between cases and controls.

Results. Fifty-four AHS cases and 157 controls

were identified. There was an increase in the risk ofAHS as the starting dose of allopurinol corrected for theestimated GFR increased. For the highest quintile ofstarting dose per estimated GFR, the odds ratio was23.2 (P < 0.01). Receiver operating characteristic ana-lysis indicated that 91% of AHS cases and 36% ofcontrols received a starting dose of allopurinol of >1.5mg per unit of estimated GFR (mg/ml/minute).

Conclusion. Our findings indicate that startingallopurinol at a dose of 1.5 mg per unit of estimatedGFR may be associated with a reduced risk of AHS. Inpatients who tolerate allopurinol, the dose can be grad-ually increased to achieve the target serum urate level.

Sustained reduction of the serum urate level to6 mg/dl is the mainstay of gout treatment (13).Indications for urate-lowering therapy include 2 goutattacks per year, chronic gouty arthropathy, tophi, ra-diographic changes in gout, and urate nephropathy (4,5),as well as polyarticular gout. Allopurinol, the mostcommonly used urate-lowering therapy, is generally welltolerated; approximately 2% of patients develop a mildrash (6) and up to 5% of patients stop allopurinolbecause of any adverse event.

A rare but potentially fatal adverse event isallopurinol hypersensitivity syndrome (AHS). AHS ischaracterized by rash (e.g., Stevens-Johnson syndrome,toxic epidermal necrolysis), eosinophilia, leukocytosis,fever, hepatitis, and renal failure. The mortality associ-ated with AHS is reported to be as high as 27% (7,8).There is no cure; early diagnosis and allopurinol with-drawal are important. Supportive care is the mainstay oftreatment (7,9).

Risk factors for the development of AHS includefemale sex, age, renal impairment, diuretic use, recent

1Lisa K. Stamp, MBChB, PhD, FRACP, Jill Drake, BN,Christopher Frampton, PhD: University of Otago, Christchurch, NewZealand; 2William J. Taylor, MBChB, PhD, FRACP: University ofOtago, Wellington, New Zealand; 3Peter B. Jones, MBChB, PhD,FRACP: University of Auckland, Hamilton, New Zealand; 4Jo L.Dockerty, MBChB, FRACP: Dunedin Hospital, Dunedin, New Zea-land; 5Nicola Dalbeth, MD, FRACP: University of Auckland, Auck-land, New Zealand.

Dr. Dalbeth has received consulting fees, speaking fees,and/or honoraria from Takeda, Ardea, and Novartis (less than $10,000each) and holds a patent for Fonterra milk products for gout.

Address correspondence to Lisa K. Stamp, MBChB, PhD,FRACP, University of Otago, Christchurch, PO Box 4345,Christchurch 8140, New Zealand. E-mail: lisa.stamp@cdhb.govt.nz.

Submitted for publication November 29, 2011; accepted inrevised form March 27, 2012.

2529

initiation of allopurinol therapy, and, in some ethnicgroups, the HLAB*5801 genotype (1014). The rela-tionship between allopurinol dose and AHS is a subjectof controversy (10,15). Dose reduction in renal impair-ment is based on a reported relationship between full-dose allopurinol (300 mg/day) in patients with renalimpairment and the development of AHS (16). Thisobservation, along with recognition that excretion of theactive metabolite oxypurinol is significantly reduced inpatients with impaired renal function, led to the sugges-tion that allopurinol dose should be determined accord-ing to creatinine clearance (16). However, there is noclear evidence that creatinine clearancebased allopuri-nol dosing reduces the incidence of AHS. In a largecasecontrol study of AHS, there was a trend towardlower allopurinol doses in the AHS group compared toallopurinol-tolerant controls (10). In another study,AHS did not occur more frequently in those takinghigher than creatinine clearancebased doses comparedwith those receiving the creatinine clearancebased dose(15). Furthermore, the current allopurinol dosing guide-lines do not differentiate between starting dose andmaintenance dose.

The aim of this study was to determine therelationship of the starting dose and the dose of allo-purinol at the time of the reaction to the occurrence ofAHS in patients with gout.

PATIENTS AND METHODS

Study design. A retrospective casecontrol study de-sign was used to determine the role of the allopurinol dose inAHS. Ethics approval was obtained from the Multi-RegionEthics Committee, New Zealand. Cases were identified in the5 major regions of New Zealand (Auckland, Waikato, Wel-lington, Christchurch, and Dunedin), representing 3 millionpeople (three-fourths of the New Zealand population).

Cases. Patients with gout who developed AHS betweenJanuary 1, 1998 and September 30, 2010 were identified byphysician recall, International Classification of Diseases (ICD)code searches, local area database searches, and via the Centrefor Adverse Reactions Monitoring (CARM). We searched forthe following ICD codes: from ICD-9, codes 693.0 dermatitisdue to drugs and medicines taken internally, 695.1 erythemamultiforme, and 974.7 poisoning by uric acid metabolismdrugs; and from ICD-10, codes L27.0 general skin eruptionsdue to drugs, L27.1 local skin eruption due to drugs, L51.0nonbullous erythema multiforme, L51.1 bullous erythemamultiforme, L51.2 toxic epidermal necrolysis, and T50.4 drugsaffecting uric acid metabolism.

Clinical notes were reviewed to identify patients inwhom a diagnosis of AHS could potentially be made. Patientswho were prescribed allopurinol for indications other than gout

were excluded. Standardized data were collected for eachpotential case, including age, sex, ethnicity, height, weight,date of gout diagnosis, starting date and dose of allopurinol,diuretic use and dose at the time of starting allopurinol, timefrom initiation of allopurinol treatment until reaction, doseof allopurinol at the time of the reaction, clinical features,treatment given, and patient outcome. Laboratory results, in-cluding serum urate, creatinine, alanine transaminase (ALT),and aspartate aminotransferase (AST) levels, and eosinophilcount prior to starting allopurinol and at the time of thereaction were collected. Estimated glomerular filtration rate(estimated GFR) was determined using the Modification ofDiet in Renal Disease study equation (17).

Potential AHS cases were adjudicated by a singleinvestigator (LKS), who was blinded with regard to the allo-purinol dose, using the Gutierrez-Macias criteria (18). Thesecriteria consist of a clear history of exposure to allopurinol;lack of exposure to another drug potentially causing thereaction; and 2 of the major criteria (worsening renal function,acute hepatocellular injury, or rash) or 1 of the major criteriaand 1 of the minor criteria (fever, eosinophilia, or leukocytosis)(18). While these criteria do not state exact values for deteri-oration in renal and liver function, we used an increase increatinine level of 2025% from baseline and an increase inALT and/or AST levels of 1.5 times the upper limit ofnormal.

Controls. For each AHS case, 3 control subjects whowere receiving allopurinol for gout but did not develop AHSwere identified. Controls were matched with the cases forage (10 years) and for the following reported risk associa-tions for AHS: sex, diuretic use at the time of initiatingallopurinol treatment, and renal function. Renal function wasassessed by estimated GFR according to the following bands:5, 515, 1530, 3050, 5070, 7090, 90110, 110130, and130 ml/minute/1.73 m2.

Standardized data were collected for each controlsubject, including ethnicity, date of diagnosis, starting dose ofallopurinol, diuretic use at the time allopurinol treatment wasstarted, dose of allopurinol at the matched time from initiationof allopurinol treatment that the reaction occurred in the case,and the maximum allopurinol dose reached.

Statistical analysis. Analysis of variance and chi-square tests were used to compare demographic and clinicalfeatures in cases and controls. Conditional logistic regressionanalysis was used to determine the association between ethni-city, tophi, and quintile of starting dose per estimated GFRand the development of AHS, allowing for multiple matchingof controls to cases (3:1). These associations were summa-rized as odds ratios (ORs) and 95% confidence intervals (95%CIs). Analysis of the dose as a function of the estimated GFRshowed that differences in renal function did not explain theobserved difference in dose between cases and controls. Theoptimal allopurinol starting dose for predicting the develop-ment of AHS was determined using a receiver operatingcharacteristic (ROC) curve. Expert opinion of the authors wasused to determine an appropriate dose for starting allopurinolbased on the ROC curve and taking the standard tablet sizesfor allopurinol into consideration, to ensure that the proposeddoses were clinically practical.

2530 STAMP ET AL

RESULTS

Identification of cases and controls. A total of7,551 potential cases were identified for case notereview. All 7,551 notes were reviewed, and 70 possibleAHS cases were identified. The main reasons for exclu-sion were that the patient was not receiving allopurinolor allopurinol was used for a nongout indication. Allcases identified by physician recall or by the CARMsearch were also captured in the ICD code searches.After adjudication, 54 cases of AHS were confirmed.The remainder had a reaction to allopurinol but did notfulfill the criteria for AHS. For 49 of the 54 cases, 3well-matched controls were identified. In the remaining5 cases, 2 well-matched controls were identified, result-ing in a total of 157 controls.

Demographic characteristics of the cases andcontrols. The demographic and clinical features areoutlined in Table 1, showing good matching betweencases and controls. Thirty-six of 54 cases (66.6%) and108 of 157 controls (68.9%) had an estimated GFR of60 ml/minute.

Cases and controls were not matched for thepresence of tophi or for ethnicity, and there was noassociation between these 2 factors. Multivariate analy-sis allowing for matching between cases and controlsshowed that the presence of tophi was associated with areduced risk of AHS (OR 0.29 [95% CI 0.010.83], P 0.021). Ethnicity was associated with risk of AHS (P 0.001). Compared with New Zealand Europeans, there

was a decreased risk of AHS in patients of Maori orPacific Island descent (OR 0.24, P 0.02) and anincreased risk of AHS in those of Chinese descent (OR70.8, P 0.005).

Clinical features and outcomes of AHS cases.The median time from starting allopurinol to the occur-rence of AHS was 30 days (range 11,080 days), and90% of AHS cases occurred within the first 180 days(Figure 1). The clinical features of AHS and whichspecific criteria were fulfilled by the cases are outlined in

Figure 1. Days from starting allopurinol treatment to the occurrenceof allopurinol hypersensitivity syndrome in patients with gout.

Table 1. Demographic characteristics of the AHS cases and matched controls*

Cases(n 54)

Controls(n 157) P

Age, mean (range) years 64.8 (2487) 64.1 (2392) 0.79Sex, no. (%) male 30 (55.6) 87 (55.4) 1.0No. (%) receiving diuretic at the time allopurinol was started 26 (48.1) 77 (49.0) 0.91Baseline creatinine, mean (range) mmoles/liter 145 (68850) 134 (70510) 0.37Baseline estimated GFR, mean (range) ml/minute 50.2 (6112) 51.3 (11115) 0.73% with baseline estimated GFR of 60 ml/minute 33.3 30.3 0.68Ethnicity, no. (%) 0.001

New Zealand European 26 (48.1) 63 (40.1)Maori and Pacific Islander 16 (29.6) 75 (47.8)Chinese 10 (18.5) 1 (0.64)Other 2 (3.7) 18 (11.5)

No. (%) with tophi 15 (27.8) 78 (49.6) 0.021Serum urate level at diagnosis, mean (range) mg/dl 9.9 (7.118.2) 10.1 (3.515.9) 0.62BMI, mean SD kg/m2 31.3 6.1 34.6 6.5 0.064Weight, mean SD kg 88.6 29.5 99.9 24.9 0.04

* AHS allopurinol hypersensitivity syndrome; estimated GFR estimated glomerular filtration rate; BMI body mass index. Data were available for 39 cases and 137 controls. Data were available for 19 cases and 46 controls. Data were available for 50 cases and 67 controls.

ALLOPURINOL HYPERSENSITIVITY 2531

Table 2. Among the patients with renal deterioration,the median increase in creatinine level was 55% frombaseline. Among the patients with acute hepatocellularinjury, the median increase in ALT level was 2.4 timesthe upper limit of normal.

Of the 54 cases, 43 (79.6%) required hospitaliza-tion. Six of these 43 cases were admitted to the intensivecare unit. Twenty-five patients received corticosteroids,2 received cyclosporin A, and 4 received antihistamines.Three of the 54 patients (6%) died, 45 recovered, andthe outcome for 6 patients could not be determined.

Relationship between the allopurinol startingdose and AHS. AHS cases started allopurinol at asignificantly higher dosage than controls (mean SEMstarting dosage 183.5 14.0 mg/day versus 112.2 6.3mg/day) (P 0.001). Nineteen percent of the controlsand 19% of the cases began allopurinol at the creatinineclearancebased dose, according to the Hande guide-lines (16). Cases were more likely to start allopurinol at

doses that were higher than the creatinine clearancebased dose than were controls (OR 16.7 [95% CI5.747.6], P 0.001) (Table 3) and controls were morelikely to start allopurinol at doses that were lower thanthe creatinine clearancebased dose than were cases(OR 6.9 [95% CI 2.916.5], P 0.001).

There was a significant increase in the percentageof patients developing AHS as the allopurinol dosecorrected for the estimated GFR increased. The propor-tion of cases for each quintile of starting allopurinol dosecorrected for the estimated GFR is shown in Figure 2.Multivariate analysis, allowing for the effects of eth-nicity and tophi with matching between cases and con-trols, revealed a strong dose-response relationship be-tween the starting dose of allopurinol adjusted forestimated GFR and the risk of AHS (overall dose effectP 0.001). The risk of AHS in the 2 highest quintiles ofallopurinol dose corrected for the estimated GFR re-mained statistically significant in multivariate analysis.

There was no relationship between allopurinolstarting dose and serum urate level in the whole group

Table 2. Clinical features of the 54 AHS cases*

Major criteriaWorsening renal impairment 29 (53.7)Acute hepatocellular injury 17 (31.5)Rash 52 (96.3)

Toxic epidermal necrolysis 10 (18.5)Erythema multiforme 4 (7.4)Diffuse maculopapular 27 (50)Exfoliative 13 (24.1)

All 3 major criteria present 9 (16.7)Rash plus renal deterioration 18 (33.3)Rash plus hepatocellular injury 8 (14.8)Rash only 17 (31.5)Renal deterioration only 2 (3.7)

Minor criteriaFever 32 (59.3)Eosinophilia 26 (48.2)Leukocytosis 24 (44.4)3 minor criteria present 14 (25.9)2 minor criteria present 10 (18.5)1 minor criterion present 20 (37)0 minor criteria present 10 (18.5)

Criteria fulfillmentAt least 2 major criteria 35 (64.8)1 major and at least 1 minor criteria 19 (35.2)Rash only 17

Plus 3 minor 2Plus 2 minor 6Plus 1 minor 9

Fever 5Eosinophilia 3Leukocytosis 1

Renal only 2Plus 3 minor 1Plus 1 minor (leukocytosis) 1

* Values are the number (%) of cases. AHS allopurinol hypersen-sitivity syndrome. Some patients had more than one type of rash.

Table 3. Allopurinol starting doses in cases and controls, stratifiedby dose recommendation based on creatinine clearance*

Cases(n 53)

Controls(n 152) P

Allopurinol starting dose higherthan the creatinine clearancebased dose

23 (43.4) 18 (11.8) 0.001

Allopurinol starting dose thesame as or lower than thecreatinine clearancebased dose

30 (56.6) 134 (88.2) 0.001

* Values are the number (%) of patients.

Figure 2. Percentage of patients who developed allopurinol hypersen-sitivity syndrome (AHS) in each quintile of the starting dose ofallopurinol corrected for the estimated glomerular filtration rate(estimated GFR). Numbers above the bars are the odds ratio. P 0.05 versus the lowest quintile.

2532 STAMP ET AL

(cases and controls) or in the cases alone (P 0.05for both), but there was a negative correlation for thecontrols (r 0.06, P 0.04). This suggests that thehigher starting doses of allopurinol in the cases were nota result of a higher serum urate level.

Repeating the analyses with exclusion of the 20%of cases and controls of Chinese ethnicity resulted in nochange in the statistical significance levels in any of theanalyses.

Allopurinol dose can be increased in patientswho tolerate it. Cases were receiving a significantlyhigher allopurinol dosage than controls at the time AHSoccurred (mean SEM 217.9 9.0 mg/day versus150.3 5.5 mg/day) (P 0.001). At the time of thereaction, 11.5% of the cases and 26% of the controlswere receiving the creatinine clearancebased allopuri-nol dose. Cases were more likely to be receiving dosesthat were higher than the creatinine clearancebasedallopurinol dose (29 of 52 cases [55.8%] versus 27 of 150controls [18%]) (OR 28.5 [95% CI 6.047.6], P 0.001).Controls were more likely than cases to be receivingdoses that were lower than the creatinine clearancebased allopurinol dose (84 of 150 controls [56%] versus17 of 52 cases [32.7%]) (OR 5.3 [95% CI 2.113.7], P 0.001). Only 10 of 53 cases (18.8%) had the dosage ofallopurinol increased between the time allopurinol wasstarted and the time the reaction occurred, as com-pared to 50 of 150 controls (33.3%). In the 10 cases inwhom the dosage was increased, the mean increase was197.5 mg/day between the starting dosage and the dos-age at the time of the reaction. In the 50 controls, themean increase in allopurinol was 110.5 mg/day. Allowingfor matching between cases and controls, the meanincrease in dosage was significantly larger in the cases(P 0.002). All controls subsequently had increases inthe dosage of allopurinol after the matched time (time atwhich AHS occurred in cases). The mean increase inallopurinol dosage was 83 mg/day, and the mean SEMmaximum dosage was 229.3 8.8 mg/day.

Can a safe starting dose of allopurinol be deter-mined? ROC analysis indicated that the starting dose ofallopurinol in 91% of AHS cases and 36% of controlswas 1.5 mg per unit of estimated GFR (mg/ml/minute). In comparison, 79% of AHS cases and 53% ofcontrols started at a dose of 2.0 mg of allopurinol perunit of estimated GFR (mg/ml/minute). A starting dosebased on an estimated GFR of 1.5 mg/ml/minute wasselected as a reasonable tradeoff between a clinicallypracticable dose and the absolute risk of AHS (Table 4).

DISCUSSION

We have shown that the starting dose of allopuri-nol is a significant risk factor for AHS. However, inthose who tolerate allopurinol, the dose can be increasedsafely, a finding confirmed in a recent dose-escalationstudy (19).

The mechanisms leading to AHS remain unclear.Three potential factors are genetic background, drugaccumulation, and immunologic responses. Recent stud-ies have highlighted a significant association betweenHLAB*5801 and AHS in the Han Chinese (10) andThai (12) populations and a weaker effect in Europeans(13). There is some controversy regarding the effect ofHLAB*5801 in the Japanese population (11,20). Inaddition, the incidence of severe cutaneous adversereactions to allopurinol have been reported to be higherin Korean patients with stage IIIV chronic kidneydisease and HLAB*5801 (21). Ethnicity was a strongrisk factor for AHS in the present study, where those ofChinese descent were overrepresented among cases,while patients of Maori/Pacific Island ethnicity were lesslikely to develop AHS compared to Europeans. Part ofthis ethnic influence may relate to genetic background,although environmental risk factors may also be in-volved.

Interestingly, patients with tophi were less likelyto develop AHS than those without, but the reason forthis is unclear, especially since no relationship betweenthe presence of tophi and ethnicity was observed. Of the10 cases who were of Chinese ethnicity, 8 did not havetophi, which may have contributed to some of theobserved negative association. However, our multivari-ate analysis, which simultaneously included ethnicity andtophi, indicated that each of these was independentlyassociated with the risk of AHS.

Table 4. Proposed starting dosage of allopurinol based on 1.5 mg perestimated GFR*

Estimated GFR,ml/minute/1.73 m2

Allopurinolstarting dosage

5 50 mg/week515 50 mg twice weekly1630 50 mg every 2 days3145 50 mg/day4660 50 mg and 100 mg on alternate days6190 100 mg/day91130 150 mg/day130 200 mg/day

* Consideration should be given to starting allopurinol at even lowerdoses in patients at high risk of developing allopurinol hypersensitivitysyndrome, such as those with HLAB*5801. Estimated GFR estimated glomerular filtration rate.

ALLOPURINOL HYPERSENSITIVITY 2533

There is no clear relationship between plasmaoxypurinol level and AHS. There are case reports ofAHS occurring well within the reported therapeuticrange for plasma oxypurinol levels (30100 moles/liter)(22). Conversely, many patients without AHS haveplasma oxypurinol levels of 100 moles/liter (22,23).Immunologic mechanisms seem likely to play an impor-tant role. The onset of AHS within weeks after startingallopurinol suggests a delayed-type hypersensitivity re-action. Deposition of IgM in the dermalepidermaljunction of the skin from a patient with severe AHS hasbeen reported (24). Liver biopsy specimens from pa-tients with AHS have shown infiltration of T lympho-cytes and neutrophils (25,26). Lymphocyte studies frompatients with AHS have shown that oxypurinol andallopurinol can induce cellular proliferation (25,27).How the starting dose of allopurinol could influencethese factors is unclear. One explanation is that lowerstarting doses of allopurinol might lead to desensitiza-tion in those who are genetically susceptible or at risk ofdrug accumulation, thereby preventing AHS.

The higher starting dose of allopurinol observedin the cases compared to the controls in the presentstudy was not related to differences in weight, body massindex, or use of furosemide, although it may be asurrogate marker for other biologic factors. In addition,because cases and controls were matched for renalfunction and age, the relationship of these factors cannotbe directly evaluated in this study. A protocol for startinga patient on allopurinol treatment has not been deter-mined, leading to variations in clinical practice. Ourresults indicate that allopurinol should be started at alow dose in order to reduce the risk of AHS. We suggesta starting dose of 1.5 mg per unit of estimated GFR. Thisis based on 90% sensitivity for predicting AHS from theROC analysis. Despite such dosing, AHS may still occur,and patients beginning allopurinol treatment should beadvised to discontinue allopurinol and seek medicalattention should they develop a rash or fever. Further-more, in high-risk groups, such as the Han Chinese orpatients with the HLAB*5801 genotype, even the start-ing doses suggested may be considered too high; clini-cians need to consider this when choosing a starting doseof allopurinol.

Effective long-term management of gout requiresa treat-to-target serum urate level approach. TheBritish Society for Rheumatology guidelines suggest atarget serum urate level of 5 mg/dl (28), while theEuropean League Against Rheumatism recommenda-tions suggest a serum urate level of 6 mg/dl, perhapslower in patients with severe or tophaceous gout (5). An

important consequence of creatinine clearancebasedallopurinol dosing is failure to achieve the target serumurate level in the majority of patients (29). In the presentstudy, we demonstrated that in patients who tolerateallopurinol, the dose can be increased, supporting up-titration to achieve the target serum urate level ratherthan a creatinine clearancebased maximum doseapproach. Although the mean maximal dose was rela-tively low in the present study, a recent study has shownthat in patients in whom the creatinine clearancebasedallopurinol dose fails to achieve the target serum uratelevel, higher doses are well tolerated and are effective inlowering the serum urate level to 6 mg/dl (19).

The median time to occurrence of AHS in ourcohort was 30 days, with 90% of cases occurring within180 days after starting allopurinol. The rate of increaseof allopurinol has not been formally studied, but goutflares commonly occur during the first 612 months oftherapy. The risk of gout flare is reduced by startingallopurinol at a low dose with a gradual up-titration andby administration of low-dose colchicine or a nonsteroi-dal antiinflammatory drug during the first 6 months ofurate-lowering therapy (30). Based on the first 30 daysbeing the period of highest risk for AHS and the resultsof our previous dose-escalation study (19), we advocateincreasing the allopurinol dose at monthly intervals untilthe target serum urate level is achieved. We did notidentify an upper limit of allopurinol dose where anincreased risk of AHS occurred.

A starting dose of allopurinol of 1.5 mg per unitof estimated GFR with monthly up-titration of the doseis very conservative compared with usual practice, andmay seem overly cautious given the low absolute risk ofAHS. When treating a chronic condition, it is moreimportant to achieve long-term adherence to therapythan to achieve the target in the shortest time. A startlow, go slow approach probably results in fewer epi-sodes of acute gout during treatment initiation andimproves compliance. For patients in whom allopurinoltreatment fails to achieve the target serum urate leveldue to intolerance or treatment resistance, newer urate-lowering therapies are available. These are more expen-sive and lack long-term safety and efficacy data. Anunderstanding of the relationship between the allopuri-nol starting dose and AHS as observed in this study mayimprove the safe and cost-effective long-term manage-ment of gout with allopurinol.

This study has a number of limitations. Althoughit represents a large study for a rare condition, thenumber of cases included was small. There is no specificICD code for AHS; thus, we used the only published

2534 STAMP ET AL

definition of AHS and carefully selected cases to fit thisdefinition. Importantly, we excluded 28 potential caseswho had presumed adverse drug reactions to allopurinolbut did not meet these strict criteria for AHS. Weacknowledge the potential pitfalls of case definition,particularly in such a heterogeneous syndrome, butbelieve that this methodology is robust and that usingwidely accepted criteria will allow comparison with otherstudies of AHS.

The ethnicity of the study population, which wasnot matched between cases and controls, may limit thegeneralizability of the results. The higher number ofChinese patients among the cases may relate to theknown association between AHS and HLAB*5801 inthose of Chinese ethnicity and the higher prevalence ofHLAB*5801 in those of Chinese descent (1520%)compared to those of European descent (16%) (13).The results, and therefore our conclusion that AHS isrelated to the starting dose of allopurinol, did notchange, however, when the analysis was limited to thoseof European, Maori, or Pacific Island descent. Due tothe retrospective study design, we were unable to deter-mine HLAB*5801 genotype or plasma oxypurinol con-centrations. The rate at which allopurinol was increased,which may influence the risk of AHS, was not standard-ized. Therefore, the question remains as to whether therate of increase in combination with the starting dose isimportant. In the present study, the estimated GFR wasused rather than the creatinine clearance, upon whichthe Hande criteria are based. These methods are notfully interchangeable for the assessment of renal func-tion. Finally, despite careful matching of cases andcontrols for recognized risk factors for AHS, with theexception of HLA genotype, there may be other uniden-tified risk factors.

The choice of ROC cutoff for the starting dosewas based on clinician opinion and consideration ofpractical dosing with the available allopurinol tabletstrengths. Validation in an independent cohort of pa-tients will be required to confirm our observation ofthe relationship between the allopurinol starting doseand AHS. To determine whether the proposed dosingstrategy results in fewer cases of AHS would require avery large randomized controlled trial to have adequatepower and would probably not be feasible.

In summary, we have shown that the starting doseof allopurinol is an important risk factor for the devel-opment of AHS. Using ROC analysis, a starting dose ofallopurinol of 1.5 mg per unit of estimated GFR isappropriate to minimize the risk of AHS. Progressiveup-titration of allopurinol is not associated with an

increased risk of AHS, and once allopurinol treatment isestablished, this strategy should be adopted to achievethe target serum urate level.

ACKNOWLEDGMENTS

We acknowledge the assistance of Karen Lindsay,Karen Pui, Angela Crowley, Karen Dentener, Sanjib Ghosh,and Philip Robinson with data collection.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising itcritically for important intellectual content, and all authors approvedthe final version to be published. Dr. Stamp had full access to all of thedata in the study and takes responsibility for the integrity of the dataand the accuracy of the data analysis.Study conception and design. Stamp, Taylor, Jones, Dockerty, Drake,Frampton, Dalbeth.Acquisition of data. Stamp, Jones, Dockerty, Drake, Dalbeth.Analysis and interpretation of data. Stamp, Taylor, Drake, Frampton,Dalbeth.

REFERENCES

1. Shoji A, Yamanaka H, Kamatani N. A retrospective study of therelationship between serum urate level and recurrent attacks ofgouty arthritis: evidence for reduction of recurrent gouty arthritiswith antihyperuricemic therapy. Arthritis Rheum 2004;51:3215.

2. McCarthy GM, Barthelemy CR, Veum JA, Wortmann RL. Influ-ence of antihyperuricemic therapy on the clinical and radiographicprogression of gout. Arthritis Rheum 1991;34:148994.

3. Perez-Ruiz F, Calabozo M, Pijoan JI, Herrero-Beites AM, RuibalA. Effect of urate-lowering therapy on the velocity of size reduc-tion of tophi in chronic gout. Arthritis Rheum 2002;47:35660.

4. Mikuls TR, MacLean CH, Oliveri J, Patino F, Allison JJ, FarrarJT, et al. Quality of care indicators for gout management. ArthritisRheum 2004;50:93743.

5. Zhang W, Doherty M, Bardin T, Pascual E, Barskova V, Con-aghan P, et al. EULAR evidence based recommendations for gout.Part II: management. Report of a task force of the EULARStanding Committee for International Clinical Studies IncludingTherapeutics (ESCISIT). Ann Rheum Dis 2006;65:131224.

6. McInnes GT, Lawson DH, Jick H. Acute adverse reactionsattributed to allopurinol in hospitalised patients. Ann Rheum Dis1981;40:2459.

7. Arellano F, Sacristan J. Allopurinol hypersensitivity syndrome: areview. Ann Pharmacother 1993;27:33743.

8. Lang PG Jr. Severe hypersensitivity reactions to allopurinol. SouthMed J 1979;72:13618.

9. Lupton GP, Odom RB. The allopurinol hypersensitivity syndrome.J Am Acad Dermatol 1979;1:36574.

10. Hung S, Chung W, Liou L, Chu C, Lin M, Huang H, et al.HLA-B*5801 allele as a genetic marker for severe cutaneousadverse reactions caused by allopurinol [published erratum ap-pears in Proc Natl Acad Sci U S A 2005;102:6237]. Proc Natl AcadSci U S A 2005;102:41349.

11. Kaniwa N, Saito Y, Aihara M, Matsunaga K, Tohkin M, Kurose K,et al. HLA-B locus in Japanese patients with anti-epileptics andallopurinol-related Stevens-Johnson syndrome and toxic epider-mal necrolysis. Pharmacogenomics 2008;9:161722.

12. Tassaneeyakul W, Jantararoungtong T, Chen P, Lin PY, TiamkaoS, Khunarkornsiri U, et al. Strong association between HLA-B*5801 and allopurinol-induced Stevens-Johnson syndrome and

ALLOPURINOL HYPERSENSITIVITY 2535

toxic epidermal necrolysis in a Thai population. PharmacogenetGenomics 2009;19:7049.

13. Lonjou C, Borot N, Sekula P, Ledger N, Thomas L, Halevy S, etal, the RegiSCAR Study Group. A European study of HLA-B inStevens-Johnson syndrome and toxic epidermal necrolysis relatedto five high risk drugs. Pharmacogenet Genomics 2008;18:99107.

14. Dalbeth N, Stamp L. Allopurinol dosing in renal impairment:walking the tightrope between adequate urate lowering and ad-verse events. Semin Dial 2007;20:3915.

15. Vazquez-Mellado J, Meono Morales E, Pacheco-Tena C, Burgos-Vargas R. Relationship between adverse events associated withallopurinol and renal function in patients with gout. Ann RheumDis 2001;60:9813.

16. Hande KR, Noone RM, Stone WJ. Severe allopurinol toxicity:description and guidelines for prevention in patients with renalinsufficiency. Am J Med 1984;76:4756.

17. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D, et al.A more accurate method to estimate glomerular filtration ratefrom serum creatinine: a new prediction equation. Ann InternMed 1999;130:46170.

18. Gutierrez-Macias A, Lizarralde-Palacios E, Martinez-Odriozola P,Miguel-De la Villa F. Fatal allopurinol hypersensitivity syndromeafter treatment of asymptomatic hyperuricaemia. BMJ 2005;331:6234.

19. Stamp LK, ODonnell JL, Zhang M, James J, Frampton C, BarclayML, et al. Using allopurinol above the dose based on creatinineclearance is effective and safe in patients with chronic gout,including those with renal impairment. Arthritis Rheum 2011;63:41221.

20. Kano Y, Hirahara K, Asano Y, Shiohara T. HLA-B allele associ-ations with certain drugs are not confirmed in Japanese patientswith severe cutaneous drug reactions. Acta Derm Venereol 2008;88:6168.

21. Jung JW, Song WJ, Kim YS, Joo KW, Lee KW, Kim SH, et al.HLA-B58 can help the clinical decision on starting allopruinol in

patients with chronic renal insufficiency. Nephrol Dial Transplant2011;26:356772.

22. Emmerson BT, Gordon RB, Cross M, Thomson DB. Plasmaoxypurinol concentrations during allopurinol therapy. Br J Rheu-matol 1987;26:4459.

23. Stamp LK, Barclay ML, ODonnell JL, Zhang M, Drake J,Frampton C, et al. Relationship between serum urate and plasmaoxypurinol in the management of gout: determination of minimumplasma oxypurinol concentration to achieve a target serum uratelevel. Clin Pharmacol Ther 2011;90:3928.

24. Utsinger PD, Yount WJ. Allopurinol hypersensitivity: granulardeposition of IgM at the dermal-epidermal junction. Am J Med1976;61:28794.

25. Braden GL, Warzynski MJ, Golightly M, Ballow M. Cell-mediatedimmunity in allopurinol-induced hypersensitivity. Clin ImmunolImmunopathol 1994;70:14551.

26. Al-Kawas FH, Seeff LB, Berendeon RA, Zimmerman HJ, IshakKG. Allopurinol hepatotoxicity: report of two cases and review ofthe literature. Ann Intern Med 1981;95:58890.

27. Emmerson BT, Hazelton RA, Frazer IH. Some adverse reactionsto allopurinol may be mediated by lymphocyte reactivity tooxypurinol. Arthritis Rheum 1988;31:43640.

28. Jordan KM, Cameron JS, Snaith M, Zhang W, Doherty M, SecklJ, et al. British Society for Rheumatology and British HealthProfessionals in Rheumatology guideline for the management ofgout. Rheumatology (Oxford) 2007;46:13724.

29. Dalbeth N, Kumar S, Stamp LK, Gow P. Dose adjustment ofallopurinol according to creatinine clearance does not provideadequate control of hyperuricaemia in patients with gout. J Rheu-matol 2006;33:164650.

30. Borstad GC, Bryant LR, Abel MP, Scroggie DA, Harris MD,Alloway JA. Colchicine for prophylaxis of acute flares wheninitiating allopurinol for chronic gouty arthritis. J Rheumatol2004;31:242932.

2536 STAMP ET AL