Colchicine

Colchicine has been used as a treatment for gout since ancient times.¹ Six prior systematic reviews^{44, 47-51} collectively identified 5 RCTs investigating the efficacy (pain reduction on VAS, number of acute gout attacks, and severity of attacks in terms of pain) and safety (total number of adverse events) of colchicine. Two of these studies were placebo-controlled trials of treatment for acute gout,^{59, 84} two were placebo-controlled studies of prophylaxis against gout flare when initiating urate lowering therapy,^{63, 74} and one study compared the addition of ice to colchicine and prednisone.⁷⁷ All reviews found that the proportion of colchicine-treated patients who reported a greater than 50 percent pain reduction was greater than that for placebo, especially if the treatment was administered within the first 12 hours of an acute attack.^{44, 47-51} Low-dose colchicine (1.2mg initially followed by 0.6mg one hour later) was found to be as effective as high-dose colchicine (1.2mg initially followed by 0.6mg each hour for the next six hours) in terms of pain relief, but had a better tolerability profile in terms of gastrointestinal adverse events: 77 percent of participants who received high-dose colchicine developed diarrhea versus 23 percent in the low-dose group versus 14 percent in the placebo group.⁸⁴

Systemic Corticosteroids

Identified systematic reviews did not find any placebo-controlled trials of systemic corticosteroids. Active-controlled trials of corticosteroids identified in the SRs are discussed in the section on comparative effectiveness.

NSAIDS

Two prior systematic reviews^{51, 52} found one low-quality trial that compared the NSAID tenoxicam (40mg once a day) against placebo in 30 patients with gout. This study reported a significant between-group difference in the fraction of patients reporting greater than 50 percent pain relief at 24 hours, no between-group differences in joint swelling at 24 hours (11/15 in the tenoxicam group vs. 4/15 in the placebo group), and no overall between-group differences at day 4.⁸⁹ No difference in adverse events was reported among patients taking NSAIDs versus those taking the placebo.

Intra-articular Glucocorticoids

One prior systematic review⁴³ on intra-articular glucocorticoids identified no randomized trials for inclusion.

Systemic Corticosteroids Versus ACTH

Three prior systematic reviews^{45, 47, 51} identified one RCT comparing systemic corticosteroids against adrenocorticotropic hormone (ACTH).⁸² In this trial 31 male patients with acute gout were randomized to receive either 40 IU of ACTH or 60mg triamcinolone intramuscularly. The study is not described as double-blinded. The duration of the acute attack and the number of joints involved were not significantly different between the two groups, although the number of reinjections for continued symptoms were fewer in the triamcinolone group (14 vs. 6 p=0.075). No mention was made of side effects. We judged this trial as being at high risk of bias.

Systemic Corticosteroids Versus NSAIDs

Four prior systematic reviews^{45, 47, 51, 52} identified three trials that compared the effectiveness of systemic corticosteroids against that of NSAIDs. None of the reviews found differences in terms of time-to-resolution of symptoms, clinical joint status at follow-up, reduction of pain at rest per hour during the first two hours and at rest per day after two weeks, and reduction of pain with activity per day after two weeks. Gastrointestinal, non-gastrointestinal, and severe adverse events were more common in the NSAID than in the systemic glucocorticoid group.⁵¹

NSAIDs Versus Selective COX-2 Inhibitors (COX-2)

Three prior systematic reviews^{47, 51, 52} identified four controlled trials that compared NSAIDs against COX-2 inhibitors. COX-2 inhibitors were as effective as NSAIDs in terms of pain, joint swelling, global improvement, and health-related quality of life, but fewer withdrawals due to adverse events were observed among those treated with selective COX-2 selective inhibitors (3 percent) versus NSAIDs (8 percent) and fewer total adverse events were observed among the recipients of selective COX-2 inhibitors (38 percent) versus recipients of NSAIDs (60 percent). Low doses of selective COX-2 inhibitors were less effective in reducing pain than high doses, and NSAIDs were as effective as high-dose COX-2 inhibitors.⁴⁷

NSAIDs Versus ACTH

Three prior systematic reviews^{46, 47, 52} identified one trial comparing the efficacy of NSAIDs to ACTH for the treatment of acute gout.⁶² In this randomized comparison of 40 IU intramuscular ACTH to 50 mg of indomethacin four times a day, among 76 (out of an initial sample of 100) men who completed 1 year of followup, the time to pain relief during an episode of acute gout was a mean of 3 hours in the ACTH-treated patients versus 24 hours in the NSAID-treated patients. No side effects were reported in the ACTH group, whereas 55 percent of patients in the NSAID group reported abdominal discomfort or dyspepsia, and 38 percent reported headaches. We judged this trial as being at high risk of bias.

NSAIDs Versus NSAIDs

We identified 16 RCTs that compared the efficacy of one NSAID versus another NSAID in patients with acute gout.^{53, 61, 64, 65, 67, 68, 70-72, 75, 76, 78, 79, 81, 83, 87} Fifteen of these 16 studies were included in prior SRs. One new trial is described below.⁵³ Most of the studies were small and therefore underpowered to detect differences. Half of the studies enrolled fewer than 30 participants; only two studies enrolled more than 100 participants. Many of the NSAIDs studied are either no longer on the market or are not FDA-approved. Ibuprofen, which is one of the most-used NSAIDs in the US, was not assessed in any study. Most of the studies reported no statistically or clinically important differences between NSAIDs in effectiveness outcomes. These data do not support a hypothesis of clinically important differences between equipotent doses of NSAIDs in terms of relief of symptoms from acute gout, a conclusion that is compatible with how NSAIDs are viewed for most other conditions, i.e., that their effectiveness is a class effect (see Table 5).

Table 5

Randomized controlled trials of NSAID versus NSAID for treatment of acute gout.

Colchicine

The most common adverse effects associated with colchicine use, by far, are gastrointestinal side effects, in particular diarrhea, with reported rates of 23 percent to 77 percent. In the one placebo-controlled study of colchicine treatment for acute gout included in this review, the authors note that all patients had gastrointestinal side effects before they had relief of gout pain. Gastrointestinal side effects are dose dependent, which contributes to the popularity of “low dose” colchicine regimens.⁸⁴ Other gastrointestinal symptoms are also common, such as nausea, vomiting, cramps, and pain.⁵⁰ Fatigue and headache are reported in a small percent (1 percent-4 percent) of patients taking colchicine. Aplastic anemia has also been associated with colchicine. One analysis of VA databases calculated an adjusted hazard ratio of 3.32 (95% CI 2.32, 4.76) for aplastic anemia with colchicine use, and an incidence rate of 0.5/1000.⁵⁷ Another analyses of the FDA Adverse Event Reporting System found pancytopenia, renal failure, vomiting, and diarrhea as the most common reported adverse events, although without a denominator, no rate can be calculated.⁵⁶ According to the manufacturer, colchicine-induced neuromuscular toxicity and rhabdomyolysis have been reported with chronic treatment in therapeutic doses. Patients with renal dysfunction and elderly patients, even those with normal renal and hepatic function, are at increased risk. Dosage must be reduced in severe renal or hepatic impairment and with concomitant use of CYP3A4 inhibitors such as erythromycin and fluconazole and P-gp inhibitors like cyclosporine, and alternative therapies considered.

NSAIDs

Nonsteroidal anti-inflammatory drugs (NSAIDs) are among the most commonly used drugs in the world, and their safety profile allows for over-the-counter availability in low doses. The main harms attributed to NSAIDs are gastrointestinal side effects, both “minor” (dyspepsia) and more serious (“perforations, ulcers, and bleeds” [PUBs],” the former occurring in 10 percent or more of patients and the latter in up to 1 percent.^{90, 91} PUBs are more common in older patients.⁹² Another common adverse event associated with NSAIDs is reduced kidney function, occurring in 1 percent to 5 percent of patients, which can be acute kidney injury, worsening of hypertension, or electrolyte abnormalities. Mild-to-moderate renal impairment is a relative contraindication for NSAIDs use. NSAIDs are also reversible platelet inhibitors. Numerous other, rare, side effects have been reported, including bone marrow suppression, aseptic meningitis, and various dermatologic adverse events. NSAIDs have been associated with an increased risk of cardiac events including myocardial infarction, stroke, heart failure, and atrial fibrillation and are the subject of FDA warning labels; however, in patients without known cardiovascular disease, the increase in risk is very small.

Corticosteroids

Long term use of glucocorticoids is associated with a host of adverse reactions, affecting almost every organ system of the body. However, most of these harms are dose and duration-dependent. The effects of short courses of glucocorticoids are not as well understood but include dysphoria and mood disorders, elevation of blood glucose levels, immune suppression, and fluid retention. All of these effects are reversible on discontinuation of the glucocorticoids, but low doses have cumulative effects over time: The cumulative effects of repeated short term exposures are similar to those seen with long term use.

ACTH

Although less used and well-studied than corticosteroids the mechanism of ACTH is in part via the stimulation of cortisol production by the body, Thus although ACTH is used and studies less than are corticosteroids, the expected harms are probably very similar to those for corticosteroids. In one trial of ACTH included here, no side effects were reported among 36 treated patients. The report of the other trial stated that the 14 patients who were treated with ACTH “tolerated [it] well.”

Colchicine

We judged the strength of evidence that colchicine improves the symptom of pain in acute gout as high, because two placebo-controlled trials showed large (∼50 percent reduction) effects.

NSAIDs

Only one placebo-controlled trial of an NSAID for treating acute gout was identified. However, we nevertheless judged the strength of evidence as high that NSAIDs improve the symptom of pain. We base this assessment on the biology of gout (it is an inflammatory reaction to uric acid crystals) and the mechanism of action of NSAIDs as an anti-inflammatory. Furthermore, NSAIDs are FDA approved for the temporary relief of pain, based on dozens of placebo-controlled trials for other painful conditions. Lastly, in patients starting on ULT, which is a risk factor for acute gout attacks, the strength of evidence is high (based on observational studies) that prophylaxis with NSAIDs greatly reduces this risk of an acute attack. Therefore, the evidence from the one available placebo-controlled trial is strengthened by the biological evidence and proven benefit in other painful conditions, and the large effect on prophylaxis against acute gout attacks with ULT. For comparisons of NSAID versus NSAID, while the available studies suffer from methodologic limitations, we found no signal of a differential effectiveness between NSAIDs, and this finding is compatible with the conclusions from other painful conditions that NSAIDs do not differ in effectiveness at equipotent doses. Thus we judged the SoE for this conclusion as moderate.

Systemic Corticosteroids

While we identified no placebo-controlled RCTs of systemic corticosteroids, we judged the strength of evidence that they reduce the symptom of pain in acute gout as high. This assessment is based on the anti-inflammatory action of steroids and the equivalence in RCTs comparing systemic steroids to NSAIDs, which we judged as high strength of evidence in relieving pain.

ACTH

Although we identified no placebo-controlled RCTs of ACTH in acute gout, we judged the strength of evidence as high that it reduces the symptoms of pain in acute gout. Because a primary mechanism of action for ACTH is by increasing the body's release of corticosteroids, the reasons are the same as for corticosteroids. However, we downgraded the SoE to moderate as the only two equivalence trials were both judged as high risk of bias. In contrast, three low risk-of-bias equivalence trials assessed the efficacy of systemic corticosteroids.

Original Randomized Controlled Trials of Dietary Interventions

A trial¹⁰² by Dalbeth and colleagues with low to moderate risk of bias assessed whether skim milk powder (SMP) enriched with glycomacropeptide (GMP) and G600 milk fat extract, non-enriched SMP, or lactose powder significantly reduced the frequency of gout attacks (flares) over a three-month study period. The frequency of gout attacks (flares) decreased from baseline in all three groups, however there was no significant difference among the three arms in terms of the change in the number of gout attacks (flares) or in adverse events.

An RCT of 40 adult gout patients by Stamp and colleagues¹⁰⁴ with moderate to high risk of bias compared the effects of vitamin C supplementation to that of allopurinol.¹⁰⁴ The study found that the reduction in serum urate level over 8 weeks was significantly less in those patients receiving vitamin C than in those who started or increased their dose of allopurinol (mean reduction 0.014 mmoles/liter [0.23mg/dl] versus 0.118 mmoles /liter [1.9mg/dl]; P < 0.001). They concluded that when administered as monotherapy or in combination with allopurinol, the uric acid lowering effect of a modest dose of vitamin C seems to be small in patients with gout.¹⁰⁴

We also identified two RCTs^{103, 106} that assessed the effects of dietary advice or a specific diet on gout (Table 7).

A 2014 RCT by Holland and McGill ¹⁰⁶ compared the effects of comprehensive gout-specific dietary advice with basic advice on serum urate in gout patients. The study divided 30 gout patients (on ULT, not further specified) into an intervention group (n=14) that received comprehensive dietary advice focused on gout management, based on the British Society of Rheumatology Guidelines and a control group (N=15) that received basic advice regarding the importance of compliance with therapy and the benefit of weight loss. Two educational sessions were provided: once at baseline and once at 3 months. The study found a significant increase in knowledge and an increase in self-reported dietary modification among the group that received gout-specific advice but no differences in serum urate between the two groups at the end of 6 months (p>0.05).¹⁰⁶ The study did not implement a clinical measure of dietary compliance (including weight loss), and self-reported compliance was extremely low. The study had a high risk of bias.

A 2012 RCT¹⁰³ investigated the effects of adjusted proportional macronutrient intake on serum urate and gout attacks in overweight patients with gout. Sixty one gout patients were randomized to one of two isocaloric (1500 cal/d) diets: high protein (40% complex CHO, 30% protein, 30% unsaturated fat), or the control low purine diet (60% CHO, 10% protein, 30% fat; purine <150mg/d). The study found that frequency of gouty attacks (17 vs 28, P=0.000) and serum urate levels(420.25±36.78 vs 466.81±41.97 μmol/L, P=0.000] were significantly reduced in the high protein group compared with the low purine group.¹⁰³ This study also had a high risk of bias.

Gout-Specific Diets and Dietary Advice

We judged the strength of evidence for gout specific diets and dietary advice is insufficient to reach conclusions, as we identified only four small RCTs with three studies that had a high risk of bias.

TCM Including Herbs and Acupuncture

Although numerous RCTs of various herbal therapies or acupuncture were identified, the results of these studies are inconsistent, and the interventions all differ from study to study, making it impossible for us to draw any conclusions.

Allopurinol Versus Placebo

Our literature search identified one SR¹⁰⁹ that included data from two placebo-controlled trials of allopurinol^{58, 119} (see Table 11 for a description of the systematic reviews and Table 12 for descriptions of the two trials).

Table 11

Systematic reviews of febuxostat or allopurinol versus placebo for the management of chronic gout.

Table 12

Randomized controlled trials of allopurinol versus placebo in the management of chronic gout.

The first study, by Schumacher (2008),¹¹⁹ was a 28-week double-blind RCT (the APEX trial) that compared allopurinol, febuxostat, and placebo. Participants were adults with hyperuricemia and gout with normal or impaired renal function. One hundred thirty-four patients were assigned to the placebo group and 268 patients were in the allopurinol 300mg group (patients with renal impairment received 100 mg allopurinol daily). The study found that the proportion of patients who achieved serum urate < 6.0mg/dl was significantly higher for than the allopurinol treated group than for the placebo group, and allopurinol resulted in greater reduction in serum urate level from baseline than did placebo. No significant difference was seen in gout attacks (flares), number of tophi, reduction in median tophus size, or incidence of adverse events between the two groups. Among the small sample of patients with renal impairment (who received allopurinol 100mg) and those in the placebo group, none achieved “last 3 monthly” serum urate levels < 6.0mg/dl, or attained serum urate < 6.0mg/dl at either the week-28 or final visits.

The second study, by Taylor et al. (2012),⁵⁸ was a 10-day double-blind RCT followed by an open label study from day 11 to day 30. Participants were adult males with crystal-proven gout who were experiencing an acute gout attack. Thirty-one patients were assigned to the allopurinol 300mg group and 26 were assigned to the placebo group. No differences in VAS pain scores or the incidence of recurrent gout attacks (flares) were found between the treatment and the placebo groups during the 10-day RCT period. Subgroup analysis comparing participants having a first gout attack with those having had prior attacks also revealed insignificant differences in pain scores. During the placebo-controlled period of the study, serum urate levels in the allopurinol group decreased significantly by day 10, whereas serum urate levels remained elevated in the placebo group during this period. When open-label allopurinol was initiated in both groups on day 11, average serum urate decreased in both groups to similar levels: less than 6.0mg/dl by day 30.

Febuxostat Versus Placebo

Our literature search identified two SRs^{27, 110} that included data from two placebo-controlled trials of febuxostat (see Tables 10 and 11). In addition, we identified one new abstract of a febuxostat placebo-controlled trial¹¹¹ and one new secondary analysis of a febuxostat placebo-controlled trial already included in the systematic reviews (see Tables 13 and 14).

Table 10

Randomized controlled trials included in systematic reviews (febuxostat vs. placebo).

Table 13

Randomized controlled trials of febuxostat versus placebo in the management of chronic gout.

Table 14

Randomized controlled trials of febuxostat versus placebo for the management of chronic gout not included in existing systematic reviews.

A total of two trials that evaluated the effect of febuxostat versus placebo for gout patients were included in the four SRs. The results of one trial are supplemented by a secondary subgroup analysis. The first study, by Becker et al. (2005b),¹⁴⁹ was a 28-day double-blind RCT with 38, 37, 40, and 38 patients assigned to placebo, febuxostat 40mg, febuxostat 80mg, and febuxostat 120mg, respectively (note that febuxostat doses above 80mg are not approved for use in the US). Adult patients with gout and hyperuricemia were enrolled. No difference in the overall incidence of gout attacks (flares) were observed between the 40mg febuxostat and placebo, but the incidence increased with dosage of febuxostat (43 percent with 80mg and 55 percent with 120mg). The incidence of gout attacks (flares) was lower (8-13 percent) for all groups when colchicine was administered with febuxostat or placebo. No difference in adverse events was found between febuxostat and placebo groups. All doses of febuxostat were associated with a significantly higher proportion of patients reaching target serum urate < 6.0mg/dl and a greater reduction in serum urate from baseline, with the 120mg febuxostat being the most effective. A five-year open label extension study of this trial found that the percentage of patients who required treatment for acute gout attacks decreased to less than 5 percent after about 12 months of ULT.¹³⁴ As to treatment effect heterogeneity, significant pairwise differences in percentage reductions in serum urate between each of the febuxostat groups and the placebo group were observed regardless of baseline urinary uric acid production. Compared with either 80 or 120mg, patients with the highest baseline serum urate levels were less likely to reach a serum urate level < 6.0mg/dl when treated with 40mg/day of febuxostat on day 28. A secondary analysis by Goldfarb (2011)¹¹² concluded that the percentage change in serum urate from baseline at day 28 was similar between overproducers and underexcretors among all febuxostat groups and was significantly greater than for the placebo group.

The second study, by Schumacher et al. (2008),¹¹⁹ was a 28-week double-blind RCT (the APEX trial) with 134 patients in the placebo group and 267, 269, and 134 patients in the febuxostat 80, 120 and 240mg groups, respectively (note again that febuxostat doses above 80mg are not approved for use in the US). Adults with hyperuricemia and gout, with normal or impaired renal function, were enrolled. Patients receiving higher doses of febuxostat were more likely to require treatment for gout attacks (flares) during the first 8 weeks when gout flare prophylaxis was provided, but no differences were observed in gout flares across treatment groups after prophylaxis ended, between weeks 8 and 28. There was no substantial difference in the number of tophi, the reduction in median tophus size, or adverse event rate across groups, with the exception that febuxostat 120mg achieved a higher mean percent decrease in the number of tophi compared with placebo at week 28. All doses of febuxostat were associated with a significantly higher proportion of patients reaching serum urate < 6.0mg/dl, with the 240mg febuxostat being the most effective.

Evidence About Subgroups

We found only limited data about differences in effectiveness stratified by the prespecified subgroups:

• Chohan 2012¹¹³ conducted a meta-analysis that compared the efficacy of febuxostat or allopurinol versus placebo for female patients, pooling data from three major RCTs (the FACT trial, APEX trial and CONFIRM trial). Female patients treated with either febuxostat or allopurinol were more likely to achieve serum urate < 6.0mg/dl than those treated with placebo. No female patients in the placebo group achieved target serum urate levels. The proportion of patients with AEs was similar across placebo, febuxostat, and allopurinol groups.
• Becker (2005) ¹⁴⁹ stratified the sample by baseline serum urate levels and found that among patients with highest baseline serum urate, febuxostat 40mg was less effective in reducing serum urate levels than were 80 or 120mg.
• Schumacher (2008)¹¹⁹ compared the effectiveness of febuxostat or allopurinol versus placebo in reducing serum urate in patients with mild to moderate renal impairment. The proportion of patients with impaired renal function who achieved target serum urate levels was numerically lower than among those with normal renal function across all treatment groups. The evidence is of low quality due to the very small sample of patients with impaired renal function (ranging from 5 to 11).

Febuxostat Versus Allopurinol

Systematic Reviews Comparing the Effectiveness of Febuxostat and Allopurinol

Four high quality SRs (AMSTAR > 8) reviewed the comparative efficacy of febuxostat and allopurinol.^{27, 109, 110, 114} The results of these reviews were broadly consistent, and the results of these studies were dominated by the FACT,¹¹⁸ APEX,¹¹⁹ CONFIRMS,¹²⁰ and EXCEL¹²¹ trials (see Tables 15 and 16).

Table 15

Randomized controlled trials included in systematic reviews.

Table 16

Systematic reviews of febuxostat versus allopurinol for the management of chronic gout.

In terms of clinical outcomes, gout flare incidence was higher at high doses of febuxostat (120mg or 240mg) than with allopurinol 100-300mg. Gout flare incidence was not statistically different between febuxostat 40mg, febuxostat 80mg, and allopurinol (100-300mg). Observed changes in tophi were less consistent. One review concluded that tophus area reduction was greater in patients taking febuxostat than in those taking allopurinol, but the median reduction in the number of tophi did not differ between these groups.²⁷ Other reviews reported non-significant differences in tophi changes and resolution.^{109, 114}

Conclusions about adverse events also varied. One review found that the high-dose febuxostat (240mg) groups experienced more adverse events than patients taking allopurinol, but the allopurinol groups had more adverse events when compared with febuxostat 80mg (note that febuxostat doses above 80mg are not approved for use in the US). When all doses were analyzed together, adverse event rates did not differ between febuxostat and allopurinol.

The research biomarker outcome results for serum urate level were consistent across these reviews: Patients taking febuxostat at doses of 80mg or higher were more likely than patients taking allopurinol 100-300mg to reach target serum urate levels of less than 6.0mg/dl.

One SR¹¹⁰ assessed the comparative effectiveness and safety of febuxostat and allopurinol in patients with and without gout. Patients taking febuxostat were more likely than patients taking allopurinol to achieve target sUA level <=6.0mg/dL at the final visit (all doses analyzed together). The proportion achieving target serum urate increased with the febuxostat dose (40mg: OR 1.2, 95% CI [1.05, 1.49], 80mg: OR 3.27, 95% CI [2.14-5.00], 120mg: OR 6.67 95% CI [5.23, 8.51]. There were no significant differences in AEs between the two groups: Pooled data demonstrate that both febuxostat and allopurinol groups had similar rates of AEs, which were mostly mild or moderate in severity. In the febuxostat groups, the most common AE that led to study withdrawal was elevated liver enzymes and the most frequent serious AEs were cardiovascular in nature.

One low-quality SR was also identified whose results were broadly consistent with those of the high-quality SRs.¹¹⁵

We also identified an SR that specifically compared the safety of urate lowering drugs. This 2014 review included seven RCTs and four SRs. Two of the included studies compared allopurinol with benzbromarone, a drug not included in our scope. The other five RCTs are all described below.^{118-120, 136, 150} This review concluded that total AEs did not differ significantly between allopurinol and febuxostat (pooled relative risk =1.04, 95% CI, 0.98, 1.11) (AMSTAR of 8/11).¹¹⁶

We also identified a SR of gout treatments in patients with impaired renal function (AMSTAR of 7/9).¹¹⁷

Major RCTs Comparing Effectiveness of Febuxostat and Allopurinol

All SRs we identified that compared the efficacy of febuxostat with allopurinol included data from the FACT¹¹⁸ and APEX¹¹⁹ clinical trials, with later SRs also including CONFIRMS¹²⁰ and EXCEL.¹²¹ The results of the SRs are dominated by these studies, because of the small sample sizes of other included trials. Trials included in at least one SR were Singal (2011)¹⁵¹, Kamatani (2011)¹⁵⁰, and Naoyuki (2011).¹⁵³ We also included one abstract.¹⁵² The most important trials, FACT, APEX, CONFIRMS, and EXCEL, are summarized here. All of these trials used gout flare prophylaxis during the study period. FACT, APEX, and EXCEL withdrew prophylaxis after week eight. CONFIRMS prescribed prophylaxis for the entire study duration.

The FACT trial¹¹⁸ compared 760 patients who received either febuxostat (80 or 120mg) or allopurinol (300mg) daily for 52 weeks (note that febuxostat doses above 80mg are not approved for use in the US). No statistically significant differences in clinical outcomes were found. The overall incidence of gout attacks (flares) was similar in all groups (64 percent, 70 percent, and 64 percent, respectively) from weeks 9 to 52 (p=0.23 for febuxostat 120mg vs. allopurinol). The median reductions in tophus area were 83 percent, 66 percent, and 50 percent, respectively (p=0.08 for febuxostat 80mg vs. allopurinol, p=0.16 for febuxostat 120mg vs. allopurinol). More patients in the febuxostat 120mg group than in the allopurinol group (p=0.003) or the febuxostat 80mg group discontinued the study. Four of the 507 patients in the febuxostat groups (0.8 percent) died, compared with none in the allopurinol group (p=0.31). The outcome of achieving a target serum urate level of <6.0mg/dl was greater for the febuxostat groups than for the allopurinol group (53 percent, 62 percent, and 21 percent, respectively; p<0.001 for comparing each febuxostat group to allopurinol).

The APEX trial¹¹⁹ compared 1,072 patients over 28 weeks who received either febuxostat (80, 120, or 240mg/day), allopurinol (100mg or 300mg per day, based on renal function), or placebo. No differences were observed between the groups in the proportion of participants with gout attacks (flares) who required treatment between weeks 8 and 28. During the first 8 weeks of the study (when gout flare prophylaxis was provided), more patients in the high-dose (120 and 240mg) febuxostat groups required treatment for gout attacks (flares) (36 percent and 46 percent) compared with those in the febuxostat 80mg (28 percent) and allopurinol (23 percent) groups (p<0.05). No significant differences in number of tophi were observed between the allopurinol and febuxostat groups. Reductions in median tophus size were reported in all treatment groups, but no significant differences were seen between the allopurinol, febuxostat, or placebo groups. The only difference in the decrease in number of tophi was between the febuxostat 120mg group (-1.2) and the placebo group (-0.3) at the end of the study (p<0.05). Proportions of adverse events were similar across groups, except for dizziness and diarrhea, which were more frequent in the febuxostat 240mg group. The outcome of achieving serum urate levels <6.0mg/dl for the last three months of the study was observed in 48 percent of the febuxostat 80mg group, 65 percent of the febuxostat 120mg group, and 69 percent of the febuxostat 240mg group, which was significantly higher than the number who achieved that outcome in the allopurinol group (22 percent). In patients with impaired renal function, more patients treated with febuxostat (all doses) achieved a serum urate of <6.0mg/dl than patients taking allopurinol 100mg.

The CONFIRMS trial¹²⁰ compared 2,268 patients receiving febuxostat 40mg per day, febuxostat 80mg per day, and allopurinol 200 or 300mg per day (depending on renal function). The only clinical outcomes reported were gout flare and safety outcomes. Rates of flare requiring treatment occurred in 10 percent to 15 percent of patients in all groups during the first two months and declined during the trial. No statistically significant differences were seen between groups. Prior use of ULT was associated with lower rates of flare compared with those without prior use (p<0.001), for each comparison. Adverse events were reported by 56 percent of participants, but the rates of occurrence did not differ among the treatment groups, and most events were mild or moderate. The outcome of serum urate level <6.0mg/dl at six months was reached in 45 percent of the febuxostat 40mg group, 67 percent of the febuxostat 80mg group, and 42 percent of the allopurinol group (p<0.001 for febuxostat 80mg compared with both of the other groups). Febuxostat 80mg was similarly superior in patients with mild to moderate renal impairment, although febuxostat 40mg was superior to allopurinol in these patients as well.

The EXCEL trial¹²¹ is an open-label extension study of two phase III trials, in which 1,086 patients received febuxostat 80 or 120mg or allopurinol 300mg for up to 40 months. Gout attacks (flares) increased after prophylaxis withdrawal in week 8, but flare rates decreased over time in all treatment groups. Gout flare was reported in less than 4 percent of participants after 18 months of treatment. Participants with tophi who maintained the target serum urate level over time experienced greater reductions in the areas of index tophi, the number of tophi, and index tophi resolution. Baseline tophus resolution was achieved by 46 percent, 36 percent, and 29 percent of participants maintained on febuxostat 80mg, febuxostat 120mg, and allopurinol, respectively. Overall adverse event rates were similar among the treatment groups. After one month of initial treatment, 81 percent and 87 percent of patients receiving 80mg and 120mg febuxostat achieved serum urate < 6.0mg/dl, as compared with 46 percent for patients receiving allopurinol. To achieve a serum urate <6.0mg/dl, more participants originally assigned to allopurinol switched to febuxostat than the number who switched from febuxostat to allopurinol.

Presence of Tophi at Baseline

Becker (2010)¹²⁰ stratified results for achievement of target serum urate by presence of tophi at baseline. Overall, the presence of tophi was associated with lower rates of achieving target serum urate level. Among patients with baseline tophi,, those taking febuxostat 80mg were more likely to achieve target serum urate (57 percent) than patients taking febuxostat 40mg or allopurinol 200-300mg (35 percent and 32 percent). In comparison, patients without tophi at baseline achieved target serum urate levels at rates of 70 percent, 48 percent, and 45 percent, respectively.

Baseline Serum Urate

Becker (2005)¹¹⁸ stratified results for achievement of target serum urate by serum urate level at baseline. At all levels of baseline serum urate levels, febuxostat 80mg was more effective than allopurinol 300mg for achieving target serum urate (47 percent for those with baseline serum urate ≥10.0mg/dl and 57 percent for those with baseline serum urate <9.0mg/dl for febuxostat versus 8 percent for those with serum urate ≥10.0mg/dl and 40 percent for those with serum urate <9.0mg/dl for allopurinol). Becker (2010)¹²⁰ also stratified results for achievement of target serum urate by serum urate level at baseline. Patients with high baseline serum urate achieved target serum urate levels at lower rates than those with lower baseline serum urate. Febuxostat 80mg was more effective for reaching target serum urate among people with high baseline serum urate (>9.0mg/dl) (49 percent to70 percent) compared with febuxostat 40mg (26 percent to 47 percent) or allopurinol 200-300mg (31 percent to 40 percent).

Renal Function

Becker (2010)¹²⁰ stratified results for achievement of target serum urate by renal function at baseline. Compared with patients with normal renal function, patients with mild renal impairment taking either febuxostat or allopurinol were more likely to achieve target serum urate. Across treatment groups, about 71 percent of patients with mild or moderate renal impairment achieved target serum urate levels while taking febuxostat 80mg, compared with 43-52 percent of patients taking febuxostat 40mg or 31-46 percent of patients taking allopurinol 200-300mg. Schumacher (2008)¹¹⁹ observed similar comparative efficacy with febuxostat 80mg compared with allopurinol, but this finding was based on a small number of observations. A SR of gout treatments in patients with impaired renal function also identified only these two trials of allopurinol versus febuxostat compared with placebo (AMSTAR rating of 7 of 9).¹¹⁷

Gibson (1981)¹⁵⁸ randomized 59 patients to receive either 0.5mg colchicine (twice daily) or allopurinol (200mg) and colchicine. Patients were followed for up to two years. The mean glomerular filtration rate (GFR) was statistically significantly lower in the colchicine group and declined over the study period as compared with the allopurinol group in which GFR increased slightly. Urate clearance fell in both groups but the trend was significant only in the group that received colchicine plus allopurinol. The study monitored renal function, including blood urea concentration, serum creatinine, GFR, urine concentrating ability, number of patients with proteinuria, and severity of proteinuria. For a subgroup of patients receiving colchicine who had achieved reductions in GFR of more than 10ml/min/(1×73 m^∧2), the results were stratified by age and presence or absence of hypertension.

Age

Becker (2011)¹⁵⁴ performed a secondary analysis of the CONFIRMS trial to compare efficacy of febuxostat and allopurinol in the elderly (>65 years) with that of younger patients (<65 years). Among 374 older subjects, the efficacy of both drugs was comparable in younger and older patients and both drugs were well tolerated in spite of high comorbidity rates and renal impairment in this group. Among patients with mild renal impairment and within each treatment group, urate lowering efficacy was higher in patients aged 65 and older than in younger patients. Among patients with moderate renal impairment, older patients within the 40 and 80mg febuxostat groups were more likely to achieve target serum urate than younger patients, but this age difference was not observed for allopurinol treatment.

Jackson (2012),¹⁵⁶ another secondary analysis of the CONFIRMS data, assessed treatment efficacy in the elderly subgroup only. Febuxostat 80mg was significantly more efficacious (82 percent) than febuxostat 40mg (62 percent; p < 0.001) or allopurinol (47 percent; p < 0.001) for achieving the primary efficacy endpoint of serum urate <6.0mg/dl. Rates of AEs were low and comparable across treatments.

Race

Wells (2012)¹⁵⁷ was a secondary analysis of 228 African Americans in the CONFIRMS trial. African American patients were mostly male and obese and were more likely to have diabetes, renal impairment, and cardiovascular disease. Rates of adverse events, gout flare, and efficacy in all treatment groups, were comparable between African American and Caucasian patients, regardless of renal function. Febuxostat 80mg was more effective than febuxostat 40mg or allopurinol 200/300mg in African American patients with mild or moderate renal impairment.

Gender

Chohan (2012)¹¹³ was a retrospective analysis of the FACT, APEX, and CONFIRMS trials that compared the efficacy of allopurinol and febuxostat in 226 women with gout. Women enrolled in these studies were older, and were more likely to be obese and to have renal impairment, hypertension, hyperlipidemia, and diabetes than the population average. Tophus resolution and incidence of gout flare were not reported. The proportions of women achieving serum urate levels <6.0mg/dl were greater in all febuxostat dosage groups compared with the allopurinol group, with efficacy significantly greater in the 80mg (p<0.001) and 120mg (p=0.006) groups. Efficacy results were similar among women with mild renal impairment, but low-dose febuxostat (40mg) was even less efficacious than higher dose febuxostat in female patients with moderate/severe renal impairment. However the number of patients in most of the renal function subgroups was small and the evidence should be interpreted with caution. Adverse event rates were similar across groups. The most common adverse events were upper respiratory tract infections, musculoskeletal/connective tissue disorders, and diarrhea.

Diabetes

Becker (2013)¹⁵⁵ performed a secondary analysis of the CONFIRMS trial that compared the efficacy of gout drugs in 312 diabetic and 1957 non-diabetic patients. Diabetic gout patients were older, more likely to be female, and had longer gout duration than non-diabetic patients. Comorbidities were more common among diabetics, including cardiovascular disease, impaired renal function, hyperlipidemia, and obesity. The efficacy of febuxostat 80mg exceeded that of febuxostat 40mg or allopurinol (p <0.050) at all levels of renal function, achieving target serum urate levels in most diabetic and non-diabetic patients. Similar adverse events were reported by both diabetic and non-diabetic patients.

Prophylaxis Against Acute Gout Attacks (Flares) When Starting Urate Lowering Therapy

For nearly 50 years, it has been known that the initiation of urate lowering therapy is associated with an increase in the frequency of acute gout attacks (flares).¹⁵⁹ More than 30 years ago, investigators performed trials using colchicine as prophylaxis against acute attacks when starting uricosuric therapy.^{74, 160} However, it was not until 2004 that the first randomized, placebo controlled trial of colchicine prophylaxis when initiating allopurinol therapy was published.⁶³ In this study, investigators randomized 51 patients to colchicine, 0.6mg twice daily, or placebo when starting allopurinol at 100mg once a day and titrating upwards with a target serum urate of 6.5mg/dl. Eight patients dropped out before they received any study drug. Seven patients withdrew during treatment: three in the colchicine group and four in the placebo group (two in the latter group due to a high frequency of attacks or flares). The 43 patients who completed the trial averaged approximately 63 years of age, mostly male, mostly (70 percent) white, more than 60 percent had tophi, and about 10 percent had chronic renal insufficiency. Patients were followed for 6 months. The occurrence of gout flares was recorded by patient recall at 3-month and 6-month visits. The difference in the reduction in flares between treatment groups was dramatic: Flares occurred in 77 percent of placebo-treated patients and 33 percent of colchicine-treated patients (p=0.008). During the first 3 months of treatment, placebo-treated patients averaged about 2 attacks (flares) and colchicine-treated patients averaged about 0.5 flares. From months 3 to 6, this advantage diminished somewhat, with about 1 flare per patient in the placebo group and almost no flares in the colchicine group. Diarrhea was much more common in colchicine-treated patients than in placebo-treated patients (43 percent vs. approximately 4 percent).

Since that study was conducted, and even pre-dating publication of this trial, the use of prophylactic therapy concomitant with the initiation of ULT has been the standard of care according to both EULAR and ACR guidelines.^{31, 32, 161} All three of the recent large ULT trials—FACT, APEX, and CONFIRMS—used prophylaxis with colchicine or NSAIDs,^118-120 in spite of the fact that no randomized trials have assessed NSAIDs as a prophylactic therapy in this situation. In the FACT and APEX trials, where prophylaxis was given for 8 weeks, both trials showed spikes in the number of acute attacks (flares) concomitant with the discontinuation of prophylaxis (an approximate doubling of the proportion of patients reporting a flare, from 20 percent to 40 percent). CONFIRMS continued prophylaxis for the entire 6 months of the trial, and no spike in attacks (flares) occurred. Wortmann and colleagues collected the adverse event data from all three trials and pooled data for FACT and APEX.¹⁴⁸ It is important to note that in all three trials, patients were not randomized to different prophylaxis regimens; rather assignment was at the discretion of the treating physician. Hence, selection bias is potentially present. Overall adverse events were higher with colchicine prophylaxis than with naproxen prophylaxis (55 percent vs. 44 percent). Diarrhea was about three times more common with colchicine than with naproxen prophylaxis (8.4 percent vs. 2.7 percent). In CONFIRMS, no statistically significant difference was seen in overall AEs reported (about 55 percent in both colchicine and naproxen-treated patients), but gastrointestinal and abdominal pains were about three times more frequent in naproxen-treated patients (3.2 percent vs. 1.2 percent). Headache was more commonly reported in colchicine-treated patients. In all three studies, upper respiratory infection was the most frequently reported AE (8 percent-9 percent in each group, no statistically significant difference). In a 2014 SR that included the one RCT mentioned above,⁶³ plus four others that employed prophylaxis when initiating therapies not included in the scope of this review (rilonacept and canakinumab), Latourte and colleagues concluded that low-dose colchicine and low dose NSAIDs are the two first-line options for prophylaxis, and that the choice depends on comorbidities and tolerance and potential interaction with other prescribed medications (AMSTAR of 3/9).¹⁴⁶ In another 2014 SR, on preventing acute gout attacks when initiating ULT, which was conducted as part of the 3e Initiative on the Diagnosis and Management of Gout, Seth and colleagues identified four placebo-controlled RCTs: the one study described above,⁶³ one described below,⁵⁵ one study included in a SR that we included,⁷⁴ and one study that used concomitant canakinumab, a drug not included in our scope. This review (AMSTAR of 7/9) concluded, like the other review and our assessment of the original trials, that colchicine prophylaxis for at least six months, when starting ULT, reduces the risk of acute attacks.¹⁴⁷

The optimal duration of prophylaxis is unknown. Discontinuation of prophylaxis at 8 weeks is associated with a spike in attacks (flares) that does not occur when prophylaxis is continued for 6 months, but the report of the CONFIRMS trial did not describe whether flares spiked when prophylaxis was discontinued at 6 months.

We identified one RCT that compared different durations of colchicine prophylaxis when initiating allopurinol therapy in patients with gout.⁵⁵ In this study, 229 patients with gout who were beginning allopurinol therapy were randomized to receive colchicine therapy (1mg/day) for either 3-6 months, 7-9 months, or 9-12 months duration. The only clinical data presented about the patients is that they averaged 47 years of age, were overwhelmingly male, and had a mean pre-treatment serum urate level of 8.5 and an on-treatment serum urate level of 6.1mg/dl. The outcome measure was “any evidence of recurrence of gouty arthritis,” but the criteria for this clinical event were not specified. Of the enrolled patients, 190 (82 percent) contributed data to the outcome. Loss to followup by group was not specified, but almost equal numbers of patients were included in each group at followup, so loss to followup was probably similar in each group. At both 6 months and 1 year, the proportion of patients who experienced recurrence was much higher in those randomized to 3-6 months of therapy than in those randomized to longer durations of therapy (at 6 months, 46 percent vs. 11 percent vs. 6 percent; at 1 year 54 percent vs. 27.5 percent vs. 23 percent, respectively). We judged this study as being at high risk of bias; therefore we could draw no conclusions from it.

Effect of Dietary Modification in Addition to Pharmacologic Therapy

The only randomized trial of dietary modification in addition to pharmacologic therapy tested specific dietary advice compared with general dietary advice.¹⁰⁶ No difference was seen in serum urate between groups. The trial is discussed in more detail in KQ 2.

Urate Lowering Therapy and Short Term Changes in Acute Gout Attacks

We judged the strength of evidence as high that urate lowering therapy does not reduce the risk of acute gout attacks, up to about six months, based on two placebo-controlled trials that each reported no difference in that outcome between groups.

Urate Lowering Therapy and Longer Term Changes in Acute Gout Attacks

No placebo controlled RCTs examined acute gout outcomes longer than six months from initiation of therapy. Nevertheless, we judged the strength of evidence as moderate that urate lowering therapy reduces the risk of acute gout attacks, based on the RCT evidence that urate lowering therapy reduces serum urate, the primary role of elevated serum urate as a risk factor for acute gout attacks, and the results of open-label extensions of the urate lowering therapy trials that show a steadily decreasing risk for acute gout attacks and a sustained decrease in this risk after about 1 year of therapy (< 5 percent/year).

Prophylactic Therapy

Although only one placebo-controlled trial tested the efficacy of prophylactic therapy when starting urate lowering therapy, we judged the strength of evidence as high that such therapy reduces the risk of acute gout attacks. We base this assessment on the large size of the effect in this trial (of colchicine), and the evidence from three large RCTs of urate lowering therapy.

In two of these trials, prophylaxis was given for eight weeks, and discontinuation of prophylaxis was accompanied by a sudden two-fold increase in the risk of acute gout attacks. In the third trial, prophylaxis was continuous throughout the six-month trial, and no “spike” of flares occurred.

Duration of Prophylaxis

We judge the strength of evidence as moderate that a course of prophylaxis longer than eight weeks is more effective than one of eight weeks duration based primarily on a comparison of acute gout attacks in the three ULT trials described above. This assessment is also supported by one RCT with high risk of bias.

Addition of Dietary Advice

We judged the strength of evidence as insufficient that the addition of gout-specific dietary advice adds to the effectiveness of urate lowering therapy, based on the existence of only one small RCT at high risk of bias.

Monitoring Serum Urate Levels

Evidence is insufficient for an effect of monitoring serum urate levels. An argument can be made that without monitoring, treatment cannot be adjusted.

Treating to Target

The strength of evidence is low that treating to a specific serum urate level reduces the risk of acute gout attacks. While elevated serum urate is the primary risk factor for acute gout attacks, and lowering serum urate levels can be expected to reduce the risk of acute gout attacks, the concept of a specific target value, such as 6.0mg/dl, has not been tested. Different targets have been proposed (e.g., 7.0mg/dl, 6.0mg/dl, 5.0mg/dl) and trying to lower serum urate levels to a target in patients who may be asymptomatic (in that they have not had a recent acute gout attack) at higher-than-target levels will necessitate increasing use of medication. The value of that strategy has yet to be proven, and examples exist from other asymptomatic conditions, in which treating to target resulted in more side effects than benefit. Thus, despite the strong biologic appeal of such a strategy, we judged the strength of evidence as low.

Discontinuation of ULT

We judged the strength of evidence as insufficient that patients who were asymptomatic for five years with a serum uric acid of <7mg/dl could have their ULT discontinued. Only one cohort study, which enrolled more than 200 patients with gout, found evidence supporting discontinuation of medication. This strategy will need to be tested in an RCT. Selection bias is always a concern in observational studies of treatment strategies.

Prophylactic Discontinuation

A moderate strength of evidence supports the finding that prophylaxis longer than eight weeks is associated with better outcomes than prophylaxis of eight weeks duration, based on the cross-study comparison of risk of acute gout in three urate lowering therapy trials.