Resolution of adefovir-related nephrotoxicity by adefovir dose-reduction in patients with chronic hepatitis B

Authors


Correspondence to:

Prof. S. G. Lim, Department of Gastroenterology and Hepatology, National University Hospital, 1E Lower Kent Ridge Road, Singapore (119228).

E-mail: mdclimsg@nus.edu.sg

Summary

Background

Chronic hepatitis B patients (CHB) treated with adefovir were followed up to evaluate nephrotoxicity and its outcome.

Aim

To assess the incidence of renal dysfunction during adefovir therapy in Asian patients and factors associated with it, and evaluate strategies to improve adefovir-related renal dysfunction and their impact on viral suppression.

Methods

Chronic hepatitis B clinic patients from a tertiary hospital on adefovir treatment, with their clinical and laboratory parameters were extracted from the hospital electronic clinical database in an observational study design. Patients were excluded if they had liver/renal transplant, baseline renal impairment or were on dialysis. Adefovir-related renal dysfunction was defined as adefovir-related abnormal serum creatinine (ARASC) > 125 μmol/L (males), >90 μmol/L (females); adefovir-related abnormal GFR <60 mL/min; and adefovir-related increased serum creatinine >0.5 mg/dL, without other known causes of nephrotoxicity.

Results

A total of 271/383 adefovir-treated patients were suitable for analysis and 33(12%) patients developed abnormal serum creatinine. Cumulative increase in proportion of patients with ARASC was 33.8% and GFR ≤60 mL/min was 38.3% by 6 years, while serum creatinine increase ≥0.5 mg/dL was 21.48% by 5 years. Using multivariate analysis, the only independent baseline predictor of ARASC was GFR ≤76.1 mL/min. Patients who had ARASC had similar levels of viral suppression to those who did not have ARASC. Those who had ARASC either continued adefovir (24%), switched therapy (24%) or had adefovir dose reduction (52%). ARASC resolved and GFR normalised in almost all patients after either switching therapy or reducing adefovir dose, with no difference between the two strategies (P = 0.737). Those with adefovir dose reduction had no significant increase in HBV DNA (P = 0.170).

Conclusions

Adefovir-related renal dysfunction occurred in a significant number of adefovir-treated patients, but reduction of the dose led to renal improvement without compromising treatment efficacy.

Introduction

Chronic Hepatitis B is the most common cause of chronic viral liver disease worldwide afflicting 400 million persons, leading to significant morbidity and mortality due to liver disease and HCC in 20–40% of infected persons. Some 500 000–1 million persons die annually from HBV-related liver disease.[1] Treatment of chronic hepatitis B with nucleoside analogues have been shown to reduce progression of liver disease and reduce HCC, particularly in cirrhotics,[2] and possibly in noncirrhotic patients.[3]

Despite the numerous choices now available for treatment of chronic hepatitis B, lamivudine remains the most commonly prescribed drug globally for chronic hepatitis B,[4] especially in Asia where cost considerations are important. A market report in 2011 acknowledged that adefovir and lamivudine were the undisputed market leaders in China.[5] As lamivudine resistance develops rapidly, in 71% after 5 years of treatment,[6] there is a large burden of patients with lamivudine resistance in Asia, and adefovir is still the mainstay of rescue therapy as both branded and generic adefovir are widely used. Although tenofovir is superior to adefovir for lamivudine resistance,[7, 8] it is not yet available in Asia especially in Japan and China for treatment of chronic hepatitis B.

Adefovir is a nucleotide analogue, which has been shown to lead to Hepatitis B viral suppression, HBeAg seroconversion and normalisation of ALT level.[9, 10] In the suggested ‘roadmap concept’, adefovir can be used in combination with other nucleoside analogue in the event of drug resistance.[11] In the clinical trials of adefovir for chronic hepatitis B, nephrotoxicity at 48 weeks was found in 13%, 27% and 50% for adefovir at doses of 30 mg, 60 mg and 120 mg, respectively,[12] hence the dose of adefovir subsequently licensed for therapy was 10 mg daily. Although this lower dose was associated with significantly lower nephrotoxicity,[13] the patients in the follow-up cohort of HBeAg negative adefovir-treated patients were relatively small. The incidence of renal dysfunction with prolonged adefovir treatment in a real-life setting is still scanty and rather controversial. A series of 45 patients treated with adefovir in Greece showed no increase in nephrotoxicity.[14] However, two case series have shown a significant reduction in renal function in patients treated with adefovir. None of these comes from Asia, a region where adefovir usage is likely to continue to be high. In a recent small series from the NIH,[15] only 14% developed renal tubular dysfunction over a mean of 7.4 years, whereas a case–control study[16] of 145 adefovir exposed patients showed an Odds ratio of 3.98 for a reduction in eGFR >20%. In this study,[16] 24% of patients developed a higher degree of renal impairment, 11.7% had moderate renal impairment and 6.9% had to switch therapy as a result of renal impairment.

As adefovir causes a range of renal dysfunction from mild to severe, how should we define adefovir nephrotoxicity? Although serum creatinine is an insensitive marker of renal dysfunction, in the clinical setting, it is often a trigger for a change in therapy, and would be the one of the more severe forms of renal dysfunction documented. In adefovir clinical trials, a rise in serum creatinine of >0.5 mg/dL was used to define adefovir nephrotoxicity and would be an important benchmark to compare our study findings. However, glomerular filtration rate (GFR) is still the most sensitive method to evaluate renal dysfunction and to assess its severity. For purposes of our study, we have used all three methods to characterise adefovir-related renal dysfunction.

Thus, our study objective was not only to assess the incidence of renal dysfunction during adefovir therapy in Asian patients and factors associated with it but also to evaluate strategies to improve adefovir-related renal dysfunction and their impact on viral suppression.

Materials And Methods

Observational study design

Patients with chronic hepatitis B attending the University Digestive Centre, National University Health System, Singapore treated with adefovir alone or in combination were extracted from the hospital Computerised Patient Support System (CPSS) from the year 2002 to 2010. CPSS is a comprehensive electronic clinical database for clinical management of all inpatients and outpatients attending the National University Hospital, Singapore, and links pharmacy prescriptions, laboratory results and radiology results. Prescriptions of adefovir were documented in CPSS, and were crosschecked with pharmacy records of adefovir prescriptions. Data on baseline characteristics, duration of therapy, HBeAg status, ALT, LDH, creatinine, ultrasound findings and HBV DNA during follow-up were extracted from CPSS. Patients seen in the University Digestive Centre were seen every 3–6 monthly and had LFT, HBV DNA, serum creatinine and hepatitis serology performed routinely. Patients were excluded if they had undergone liver transplant, had hepatocellular carcinoma or other malignancy, had abnormal creatinine before adefovir therapy, had renal transplantation or were on renal dialysis, on adefovir treatment <6 months, had malignancy or acute hepatitis B, before starting therapy. This study was approved by the National Healthcare Group Institutional Review Board and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki. Waiver of consent was granted subject to anonymisation of patient identity.

Evaluation of renal dysfunction

Adefovir-related abnormal serum creatinine was defined as serum creatinine level above upper limit of normal based on our local laboratory reference value (>125 μmol/L for males and >90 μmol/L for females) after starting adefovir therapy and without other known causes of nephrotoxicity, when baseline values were within the normal range. Discontinuation or dose adjustment of adefovir were based on development of abnormal serum creatinine. Adefovir-related increased in serum creatinine >0.5 mg/dL was defined as a rise in serum creatinine of >0.5 mg/dL during adefovir therapy. Adefovir-related abnormal GFR was defined as GFR <60 mL/min based on National Kidney Foundation as calculated by MDRD (Modification of Diet in Renal Disease) formula.[17]

Laboratory testing

Liver Function Tests, renal chemistry and full blood count were performed using Advia Chemistry (Advia Centaur Systems, Siemens Medical Solutions Diagnostics Pty Ltd, Bayswater, Australia); HBsAg, HBeAg and anti-HBe were tested using Roche Diagnostics kits (Roche Diagnostics GmbH, Mannheim, Germany). Serum HBV-DNA was measured with the Hybrid Capture II HBVDNA Test (Digene Corporation, Gaithersburg, MD, USA) with a lower detection range of 1.4 × 105 copies/mL to upper detection range of 1.7 × 109 copies/mL until April 2006. Thereafter, HBV-DNA levels were measured with the Artus HBV RT PCR (Real-time PCR) kit (Qiagen Diagnostics, Hamberg, Germany), with lower detection range of 100 copies/mL to upper detection range of 1 × 109 copies/mL. HBVDNA results were standardised by converting to IU/mL. The conversion factor for Hybrid Capture was 2.3 copies/mL,[18] while that for Artus RT PCR was 5.8 copies/mL equivalent to 1 IU/mL.[19]

Statistical analysis

Overall incidence rate of abnormal serum creatinine was analysed by Kaplan–Meier survival analysis. Comparison of categorical variables was performed using Fisher's exact test, whereas comparison of continuous variables was performed using student's t-test. Two-tailed significance tests were used and P-value of <0.05 was considered significant. Multiple logistic regression was performed to determine independent factors for raised creatinine and undetectable HBVDNA (<100 copies/mL). Log-rank test and Multivariate Cox regression analysis were performed where necessary.

Results

Of the total 383 patients identified on adefovir treatment, 112 patients were excluded (liver transplant = 11, hepatocellular carcinoma = 42, end stage renal failure = 11, less than 6 month of adefovir = 42, malignancies = 4, acute hepatitis B = 3), leaving 271 patients available for analysis. Of the 271 patients, 33 developed renal dysfunction and 25 had a change in adefovir therapy, while 8 continued adefovir (see Figure 1 for patient disposition). Baseline characteristics of patients are summarised in Table 1. The mean age of patients was 47.03 (95% CI: 45.38–48.68) years, the mean duration of adefovir treatment was 28.98 (95% CI: 26.46–31.50) months and 99 (36.5%) patients had liver cirrhosis.

Table 1. Baseline characteristics of study patients (n = 271)
BaselinePatients with adefovir-related abnormal serum creatinine, total = 33 (%)*Patients without adefovir-related abnormal serum creatinine, total = 238 (%)*P-value
  1. AST, aspartate transaminase; ALT, alanine transaminase; ALP, alkaline phosphatase; GFR, glomerular filtration rate.

  2. Data shown as numbers of patients (%) or mean (95% CI). Patients were divided into those with (n = 33) or without adefovir-related abnormal serum creatinine (n = 238). Significant differences (P < 0.05) were found in age, GFR <85.5 mL/min and mean GFR.

Male23 (69.7)177 (74.4)0.535
Chinese31 (93.9)213 (89.5)0.550
Treatment naive12 (36.4)93 (39.1)0.850
Child's A28 (84.8)222 (93.3)0.236
HBeAg positive16 (48.5)135 (56.7)0.455
Cirrhosis16 (48.5)83 (34.9)0.176
Diabetes mellitus5 (15.2%)36 (15.1%)0.997
Hypertension10 (30.3%)44 (18.5%)0.111
NSAID use3 (9.1%)9 (3.8%)0.165
Age (years)54.5 (50.0–59.1)45.9 (44.2–47.7)0.001
Albumin (g/L)40.2 (38.5–41.8)41.7 (41.1–42.4)0.086
Bilirubin (μmol/L)17.0 (13.4–20.6)15.6 (14.4–16.9)0.462
AST (U/L)109.6 (46.2–172.9)68.4 (56.6–80.2)0.203
ALT (U/L)152.5 (67.1–237.9)98.8 (81.2–116.4)0.218
ALP (U/L)81.4 (71.3–91.4)80.8 (76.9–84.8)0.933
HBV DNA (log IU/mL)5.9 (5.2–6.7)5.2 (4.8–5.4)0.065
Creatinine (μmol/L)91.1 (84.8–97.2)74.8 (72.7–76.8)<0.001
GFR < 85.5 mL/min25 (75.8)56 (23.5)<0.001
Mean GFR (mL/min)76.1 (71.4–80.7)102.2 (98.9–105.4)<0.001
Figure 1.

Flowchart of subjects in the study. ADV, adefovir; ETV, entecavir; LAM, lamivudine; Cr, creatinine; LT, liver transplantation; HCC, hepatocellular carcinoma; ESRF, end stage renal failure; RTP, renal transplant patients; LOF, lost to follow-up.

Renal dysfunction during adefovir treatment

There were 33 of 271 patients who developed adefovir-related abnormal serum creatinine. The baseline GFR in these patients was 76.1 (95% CI: 71.4–-80.7) mL/min, which deteriorated to 54.9 (95% CI: 51.8–58.0) mL/min during adefovir treatment. The cumulative proportions of patients who developed abnormal serum creatinine was 5.3% in year 1, rising to 33.8% by year 6 (Figure 2). Using GFR of less than 60 mL/min as the cut-off, cumulative renal impairment was found in 7.7% in year 1, rising to 38.3% by year 6 (Figure 1). Overall, 27/271 (9.9%) had an increase in serum creatinine of ≥0.5 mg/dL, and when analysed by Kaplan–Meier graph, there was a cumulative increase over time, reaching 21.48% at 5 years (Figure 2).

Figure 2.

Proportion of patients with adefovir-related renal dysfunction over time. Adefovir-related renal dysfunction was defined as either abnormal serum creatinine, GFR <60 mL/min or increase in serum creatinine >0.5 mg/dL. In all cases, Kaplan–Meier graphs show progressive and cumulative development of renal dysfunction while on adefovir therapy. GFR, glomerular filtration rate; ULN, upper limit of normal. Data shown in number (%), mean (95% CI), *Multivariate Cox regression.

Impact of baseline renal function

To determine if baseline renal function predicted future development of adefovir-related abnormal serum creatinine, the area under the reporting operator characteristics (AUROC) was performed. The GFR that best predicted subsequent abnormal serum creatinine during adefovir therapy was GFR <85.5 mL/min. In univariate analysis, factors associated with development of adefovir-related abnormal serum creatinine were GFR ≤85.5 mL/min, older age (P = 0.001) and GFR ≤76.1 mL/min (the mean GFR in those with adefovir-related abnormal serum creatinine) (P < 0.001). The presence of diabetes mellitus, hypertension, or nonsteroidal anti-inflammatory drugs (NSAID) usage did not appear to be significant risk factors for renal impairment, although hypertension and NSAID use were numerically more frequent, but not statistically significant, in those who developed renal impairment compared with those who maintained normal renal function (30.3% vs. 18.5%, OR: 1.63, 95% CI: 0.423–2.37; and 9.1% vs. 3.8%, OR: 2.404, 95% CI: 0.685–8.432 respectively). Multivariate analysis showed that only baseline GFR ≤76.1 mL/min (P < 0.001, OR: 1.09, 95% CI: 1.03–1.14) was associated with abnormal serum creatinine with adefovir.

Efficacy of adefovir in patients with and without abnormal serum creatinine

To determine if adefovir was more efficacious in viral suppression in patients with adefovir-related abnormal serum creatinine, the proportion of patients with undetectable HBV DNA after adefovir treatment (from baseline positive HBV DNA) were compared between those with abnormal serum creatinine and without. The proportion of patients on adefovir-lamivudine combination therapy was similar between patients with adefovir-related abnormal serum creatinine (28.1%) and those without abnormal serum creatinine (31.8%). Overall, 90.6% of patients had undetectable HBV DNA in those with adefovir-related abnormal serum creatinine compared to 66.3% in without abnormal serum creatinine (P = 0.004). However, after adjusting for baseline differences, no significant difference between the two groups was found using Cox Regression (P = 0.586, data not shown). It appears that the apparent better efficacy of adefovir in those with adefovir-related abnormal serum creatinine may have been due to baseline differences between the two groups, although only 9.4% of patients with adefovir-related abnormal serum creatinine remained HBV DNA positive, while 33.7% of those with normal serum creatinine remained HBV DNA positive. In univariate analysis, factors associated with undetectable HBV DNA were: HBeAg negative status (P < 0.001), cirrhosis (P = 0.008), raised creatinine (P = 0.004), older age (P < 0.001), higher AST (P = 0.010), lower HBV DNA (P < 0.001) and GFR (P = 0.005). In multivariate analysis (using Cox regression), independent factors associated with undetectable HBV DNA were HBeAg seronegative state (P = 0.001), higher AST (P < 0.001) and lower HBV DNA (P < 0.001).

Outcome in renal function in those with adefovir-related abnormal serum creatinine

Among patients who developed adefovir-related abnormal serum creatinine (n = 33), 8 patients continued with the same dose of adefovir while 25 patients changed therapy (dose reduction, change in antiviral therapy or cessation in therapy). The outcome of these patients is listed below.

Continuing adefovir in patients with adefovir-related abnormal serum creatinine

In the 8 patients with adefovir-related abnormal serum creatinine who continued therapy, normalisation of creatinine occurred in 3 patients (LSC, NSH, LTS), while serum creatinine remained abnormal in the other 3 patients (TSH, LWC, CYL) (Figure 3). In the three patients who normalised their serum creatinine despite continuing adefovir, there was a significant increase in serum creatinine and reduction in GFR after starting adefovir (Figure 2), although in patient LTS, the reduction in GFR was not as pronounced. In all patients, GFR continued to remain abnormal but stable, despite mild improvement in serum creatinine. In the three patients who continued to have abnormal serum creatinine during continued adefovir therapy, one patient (TSH) continued to have abnormal serum creatinine and abnormal GFR which stabilised, while the other two (LWC, CTL) had increasing levels of serum creatinine and continued worsening of GFR (Figure 3). In summary, those who remained on adefovir had persistently abnormal GFR. Of the remaining two patients, one patient was lost to follow-up and the other one died from peritonitis secondary to perforated sigmoid diverticulitis.

Figure 3.

Serum creatinine and GFR over time in six patients who developed abnormal serum creatinine and continued adefovir treatment. ADV, adefovir; GFR, glomerular filtration rate; MDRD, modification of diet in renal disease formula.

Change in therapy in patients who developed abnormal creatinine

Adefovir dose reduction

In the remaining 25 patients who developed adefovir-related abnormal serum creatinine, 17 patients had reduction in adefovir dose to 10 mg every other day resulting in normalisation of creatinine in 13 (76.47%) patients.

Adefovir discontinuation

Adefovir was stopped in the remaining 8 patients and were switched to either entecavir (n = 3), lamivudine (n = 3), or no treatment (n = 2). All patients (100%) had improvement in GFR and normalisation of serum creatinine. The mean duration for creatinine normalisation based on Kaplan–Meier analysis was 5.601 months (95% CI: 0.00–14.11).

Using log-rank test, no significant difference was found in the proportion of patients with normalisation of GFR (Figure 4a) or creatinine normalisation (Figure 4b) between those with adefovir dose reduction compared with those with adefovir discontinuation. After adefovir discontinuation, GFR returned to normal in 75% of patients, while in those who had adefovir dose reduction, GFR returned to normal in 70.58% (P = 0.737, Log-Rank Test).

Figure 4.

Proportion of patients who normalised GFR (a) or serum creatinine (b) after adefovir therapy was changed. There was no significant difference in serum creatinine (= 0.103, log rank test) as well as GFR (P = 0.737, log rank test) in those who stopped or switched from adefovir to another antiviral agent compared with those who had adefovir dose reduction. GFR, glomerular filtration rate.

Impact of change in therapy on HBV DNA in patients with adefovir nephrotoxicity

In the 17 patients who had adefovir dose reduction due to adefovir-related abnormal serum creatinine, there was no significant HBV DNA increase after adefovir dose reduction whether they were on adefovir monotherapy or in combination with lamivudine or entecavir. Mean HBV DNA before adefovir dose reduction was 2.15 ± 2.78 and after adefovir reduction was 2.3188 ± 4.5 (P = 0.170, Student's Paired t-test).

Discussion

Our study demonstrated a spectrum of adefovir-related renal dysfunction ranging from a reduction in GFR to abnormal serum creatinine that was cumulative, reaching 33.8–38.3% after 6 years in patients who had no apparent renal impairment at baseline. Importantly, all three methods of evaluating renal dysfunction show a similar pattern, with survival curves showing progressive renal impairment over time that did not appear to plateau. In a case–control study,[16] the reported progression of renal impairment defined as >20% decrease in GFR was approximately 30% at 5 years. Similarly, a retrospective study reported 34.7% patients with >20% decrease in GFR at 5 years.[20] The different definitions of renal impairment and different measurement methods make comparison between studies difficult. The progression of renal impairment described in the Gilead clinical trials as proportion of patients with >0.5 mg/dL increase in serum creatinine had been found to be 3% at 5 years,[13] compared to 5% at 30 months in the study by Ha et al.,[16] whereas our study had an even higher rate of 21.48% at 5 years. These differences could be due to differences in baseline characteristics of patients, such as older patients with higher prevalence of liver cirrhosis in our study population. Notably, all our patients were Asians, similar to the study by Ha and colleagues[16] (98% Asians), compared with those in the Gilead follow-up study[13] in which patients were mainly Caucasians. Although it is difficult to make appropriate comparison between studies due to differences in methods and endpoints, it is an intriguing observation that Asians may be more predisposed to adefovir nephrotoxicity compared with Caucasians. Even in a study involving mainly Caucasians with advance disease awaiting liver transplantation, which is a predisposing factor for renal dysfunction, renal impairment occurred only in 6% of patients.[21] In our study, multivariate analysis showed that baseline GFR ≤76.1 mL/min was the only factor that predicted development of future adefovir-related abnormal serum creatinine, a finding that echoed that of Ha and colleagues.[16]

It was possible that the effect of adefovir on renal dysfunction could have been a result of increased metabolism of adefovir to active metabolites and consequently its increased efficacy, but when efficacy of adefovir in viral suppression was examined in those with and without renal impairment, no difference was found on Cox regression. A study[22] examining the efficacy of adefovir suspension found that the 5-mg dose resulted in numerically inferior viral suppression compared with the 10-mg dose; however, this was not in the context of patients with renal dysfunction. It is intriguing that despite dose reduction, our patients with renal dysfunction had no significant increase in HBV DNA, suggesting that the efficacy of adefovir in such patients may have been more profound, some 90% of these patients were HBV DNA-negative compared to only 66% who did not have adefovir-related abnormal serum creatinine. This observation could be important as adefovir is considered a relatively weak antiviral agent against hepatitis B virus, and poor responders have been documented.[23]

Ha et al.[16] also found that 6.9% of patients had to alter therapy due to nephrotoxicity leading to a return of renal function towards baseline levels. In our study, a similar number of patients – 8.6% had alteration in therapy due to nephrotoxicity. However, when we compared patients who switched from adefovir to other treatments and those who had dose reduction of adefovir to 10 mg every other day, we found that serum creatinine normalised in 85% and 76.5% of patients respectively. GFR returned to normal in 75% of those who switched from adefovir to other treatments, and normalised in 70.58% in those who had adefovir dose reduction. In the six patients who continued adefovir, GFR remained persistently abnormal despite the borderline improvement in their serum creatinine, hence serum creatinine cannot be used as a reliable or accurate marker of renal dysfunction. Also re-assuring was the finding that HBV DNA did not significantly increase in those patients who reduced adefovir dosage, whether they were on adefovir monotherapy or combination therapy with lamivudine.

Although the study was observational in nature, these findings are sufficient to consider dose reduction in adefovir as a therapeutic option for patients with adefovir renal dysfunction. This is particularly important as adefovir remains the main rescue therapy for lamivudine resistance as tenofovir is not readily available in some Asian countries such as China and Japan. Due to the huge burden of lamivudine resistance in Asia, the widespread use of branded and generic adefovir, this information is critical to physicians managing chronic hepatitis B. Even with the introduction of tenofovir, this problem is likely to remain as tenofovir belongs to the same class of drug and significant nephrotoxicity and acute renal failure are already a recognised feature of therapy in a meta-analysis.[24] Consequently, such a strategy of dose reduction could also benefit tenofovir patients who developed nephrotoxicity as both drugs are likely to have similar mechanisms of nephrotoxicity.[25-27]

Moreover, dose reduction while maintaining a therapeutic response is a highly cost-effective measure. The alternative of switching to entecavir is considerably more expensive and also comes with a much higher risk of entecavir resistance, particularly in those with pre-existing lamivudine resistance,[28] hence is less desirable.

Authorship

Guarantor of the article: Prof. Lim Seng Gee.

Author contributions: Seng Gee Lim was involved in study concept and design. Guan Huei Lee, Kieron Lim, How Cheng Low, Michelle Gowans, Maung Aye Thwin, Yin Mei Lee, Yock Young Dan, Poh Seng Tan and Charlene Soon were involved in acquisition of data. Seng Gee Lim, Juanda Leo Hartono and Myat Oo Aung were involved in analysis and interpretation of data. Juanda Leo Hartono and Seng Gee Lim drafted the manuscript. Seng Gee Lim critically revised the manuscript for important intellectual content. Myat Oo Aung contributed to statistical analysis. Seng Gee Lim obtained funding. Charlene Soon, Lily-Lily Chiu, Mui Joo Khoo and Evelyn Koay provided technical or material support. All authors approved the final version of the manuscript.

Acknowledgement

Declaration of personal interests: Seng Gee Lim is on the advisory board of Novartis Pharmaceuticals, Merck Sharp and Dohme Pharmaceuticals, Janssen Pharmaceuticals, Archillion Pharmaceuticals and Bristol Myers Pharmaceuticals, and is on the speaker's bureau for GlaxoSmithKline Pharmaceuticals, Novartis Pharmaceuticals and Bristol Myers Pharmaceuticals.

Declaration of funding interests: This study was funded in part by grant 02/N01 from the Health Services Development Program (HSDP), Ministry of Health. The sponsors has no role in the design, collection or analysis of data, drafting or review of the manuscript.

Ancillary