The NEW ENGLAND JOURNAL Of MEDICINE baseline alanine aminotransferase levels, and base- bers vouch for the validity and completeness of line Child-Pugh and fibrosis scores. The data from the data and the veracity of the data analysis patients withoutend points were censored as ofthe date that treatment was stopped (if the data and RESULTS safety monitoring board terminated the trial)or at the last date of available follow-up after treatment CHARACTERISTICS OF THE PATIENTS (if the trial was terminated for other reasons). Be- The intention-to-treat population consisted of 651 cause the study was stopped at the second interim patients who were randomly assigned to treatment alysis with strict stopping criteria applied at the at 41 sites across Australia, China, Hong Ko first interim analysis, adjustments that had to be laysia, New Zealand, the Philippines, Singapore, Tai- made to the final P values and estimates were neg- wan, and Thailand; 436 patients were assigned to ligible(an increase in the P value of <0.001 and an receive lamivudine and 215 to receive placebo. In y o The study was conducted in accordance with were male and 98 percent were Asian. The treat- rease of <0.002 for the hazard ratio ach treatment group, 85 percent of the patients good clinical practice and all applicable regula- ment groups were also well matched in terms of tions, including the Declaration of Helsinki(mod- age, laboratory results at baseline, and Ishak fibro- ified in 1996). Each investigator ensured that the sis scores (Table 1). The median Child-Pugh score protocol was reviewed and approved by the local at baseline was 5(range, 5 to 9), and no patient had ethics committee. Written informed consent was evidence of hepatocellular carcinoma, renal insuf. obtained from each patient before enrollment in ficiency, bleeding varices, or spontaneous bacterial dy entry. The study was designed by the academic inves- tigators in conjunction with medical staff from STUDY TERMINATION AND END POINTS GlaxoSmithKline. The data were collected by the At the recommendation of the data and safety investigators and analyzed by glaxoSmithKline. monitoring board, the double-blind phase of the Each author had access to the data. This article study was terminated at the second interim analy was written by a committee consisting of seven sis, because results had crossed the predefined authors(Drs. Liaw, Sung, Chow, and Farrell; and boundary for showing efficacy. At this time, 67 pa Mrs Shue, Mr. Keene, and Dr. Dixon, who are tients had achieved HBeAg seroconversion, 52 had GlaxoSmithKlineemployees) The committee mem- stopped therapy for other reasons, and 68 end Table 2. Disease Progression during Double-Blind Treatment and Follow-up after Treatment. Lamivudine Placebo P Variable (N=215 c noof patients(%) Overall disease progression 34(7.8) 045(0.28-0.73 0.001 Increase in Child-Pugh score 15(34) 19(8.8) 045(022-0.90 Hepatocellular car 17 16(74) 049(025-0.99) Bleeding varices Spontaneous bacterial peritonitis Liver-related death atient reached an end point during follow-up before the termination of the study. Dashes denote not gh score, and Ishak fibrosis score. Cl de fidencountry, sex, baseline alanine aminotransferase level, Child- When five cases of hepatocellul ed during the first cluded the hazard ratio was 0.47 (95 percent confidence interval, 0. 22 to 1.00; P-0.052) N englj Med 35715 Www.nejm.org OctobeR 7, 2004 第17页n engl j med 351;15 www.nejm.org october 7, 2004 The new england journal of medicine 1526 baseline alanine aminotransferase levels, and base￾line Child–Pugh and fibrosis scores. The data from patients without end points were censored as of the date that treatment was stopped (if the data and safety monitoring board terminated the trial) or at the last date of available follow-up after treatment (if the trial was terminated for other reasons). Be￾cause the study was stopped at the second interim analysis with strict stopping criteria applied at the first interim analysis, adjustments that had to be made to the final P values and estimates were neg￾ligible (an increase in the P value of <0.001 and an increase of <0.002 for the hazard ratio). The study was conducted in accordance with good clinical practice and all applicable regula￾tions, including the Declaration of Helsinki (mod￾ified in 1996). Each investigator ensured that the protocol was reviewed and approved by the local ethics committee. Written informed consent was obtained from each patient before enrollment in the study. The study was designed by the academic inves￾tigators in conjunction with medical staff from GlaxoSmithKline. The data were collected by the investigators and analyzed by GlaxoSmithKline. Each author had access to the data. This article was written by a committee consisting of seven authors (Drs. Liaw, Sung, Chow, and Farrell; and Mrs. Shue, Mr. Keene, and Dr. Dixon, who are GlaxoSmithKline employees). The committee mem￾bers vouch for the validity and completeness of the data and the veracity of the data analysis. characteristics of the patients The intention-to-treat population consisted of 651 patients who were randomly assigned to treatment at 41 sites across Australia, China, Hong Kong, Ma￾laysia, New Zealand, the Philippines, Singapore, Tai￾wan, and Thailand; 436 patients were assigned to receive lamivudine and 215 to receive placebo. In each treatment group, 85 percent of the patients were male and 98 percent were Asian. The treat￾ment groups were also well matched in terms of age, laboratory results at baseline, and Ishak fibro￾sis scores (Table 1). The median Child–Pugh score at baseline was 5 (range, 5 to 9), and no patient had evidence of hepatocellular carcinoma, renal insuf￾ficiency, bleeding varices, or spontaneous bacterial peritonitis at study entry. study termination and end points At the recommendation of the data and safety monitoring board, the double-blind phase of the study was terminated at the second interim analy￾sis, because results had crossed the predefined boundary for showing efficacy. At this time, 67 pa￾tients had achieved HBeAg seroconversion, 52 had stopped therapy for other reasons, and 68 end results * Only one patient reached an end point during follow-up before the termination of the study. Dashes denote not applicable. † Hazard ratios were derived from a Cox model adjusted for country, sex, baseline alanine aminotransferase level, Child– Pugh score, and Ishak fibrosis score. CI denotes confidence interval, unadjusted for interim analyses. ‡ Two patients fulfilled two criteria simultaneously at end-point confirmation. § When five cases of hepatocellular carcinoma diagnosed during the first year were excluded, the hazard ratio was 0.47 (95 percent confidence interval, 0.22 to 1.00; P=0.052). Table 2. Disease Progression during Double-Blind Treatment and Follow-up after Treatment.* Variable Lamivudine Group (N=436) Placebo Group (N=215) Hazard Ratio (95% CI)† P Value no. of patients (%) Overall disease progression 34 (7.8)‡ 38 (17.7) 0.45 (0.28–0.73) 0.001 Increase in Child–Pugh score 15 (3.4) 19 (8.8) 0.45 (0.22–0.90) 0.02 Hepatocellular carcinoma§ 17 (3.9) 16 (7.4) 0.49 (0.25–0.99) 0.047 Renal insufficiency 2 (0.5) 0 — — Bleeding varices 2 (0.5) 3 (1.4) — — Spontaneous bacterial peritonitis 0 0 — — Liver-related death 0 0 — — 第 17 页