AMIVUDINE FOR CHRONIC HEPATITIS B AND ADVANCED LIVER DISEASE Table 1.(Continued. Characteristic Aspartate aminotransferase-U/liter Bilirubin 13.7 Range 3.0-50.0 1.7-58.1 Range 35-173 35-135 H 14.7 14.6 Range 8.4-19.0 92-17.8 Median 14,000-401,000 Prothrombin time 12.5 80-23.8 9.8-27.6 White-cell count per mm3 Median 5330 1980-1,600 2200-11,500 ☆ There were no he two treatment groups. The patients' race was recorded by the investi- ations of the patients indicates good liver function and 15 poor liver function ) is a measure of the severity of liver disease f The ishak fibrosis score(range, 0 to 6)is a measure of the degree of fibrosis i topsy specimens. Scores of o to 4 indi- S All patients had detectable HBV DNA at screening: 0. 7 meq per milliliter equals approximately 7x105 copies per milliliter I To convert values for bilirubin to milligrams per deciliter, divide by 17.1. I To convert values for creatinine to milligrams per deciliter, divide by 88.4 was considered to be a clinically relevant treatment ping boundaries. At each inspection, the"Christ- effect. This difference corresponds to a hazard ra- mas tree"correction was applied to the contin cent at the 5 percent level of significance, with a ra- number and timing ofinterim analyste predictable tio of0.64. For the study to have a power of 90 per- uous boundaries to account for the ur tio of 2: 1 for the random assignment of patients The first interim analysis was scheduled for to lamivudine or placebo, 240 end points would months after the completion of patient recruit need to be observed. o Assuming a dropout rate ment, and subsequent interim analyses were to be of 25 percent during a five-year period, the num- performed between 6 and 12 months after the first ber of patients required overall was estimated to interim analysis; the aim was to have approximate- ly 35 events between interim analyses. The inten We used a sequential, asymmetric trial with the tion-to treat analysis included all patients who were triangular test to monitor the primary efficacy randomly assigned to receive either lamivudine or end point of time to clinical disease progression. placebo. Treatments were compared with the use At each interim analysis, the test statistics were of a Cox proportional-hazards model, 32 with each calculated and compared with straight-line stop- analysis allowing for the covariates of country, sex, NEnglJMed351715www.nejm.orgocTober7,2004 1525 第16页n engl j med 351;15 www.nejm.org october 7, 2004 lamivudine for chronic hepatitis b and advanced liver disease 1525 was considered to be a clinically relevant treatment effect. This difference corresponds to a hazard ra￾tio of 0.64. For the study to have a power of 90 per￾cent at the 5 percent level of significance, with a ra￾tio of 2:1 for the random assignment of patients to lamivudine or placebo, 240 end points would need to be observed.30 Assuming a dropout rate of 25 percent during a five-year period, the num￾ber of patients required overall was estimated to be 600. We used a sequential, asymmetric trial with the triangular test31 to monitor the primary efficacy end point of time to clinical disease progression. At each interim analysis, the test statistics were calculated and compared with straight-line stop￾ping boundaries. At each inspection, the “Christ￾mas tree” correction31 was applied to the contin￾uous boundaries to account for the unpredictable number and timing of interim analyses. The first interim analysis was scheduled for 18 months after the completion of patient recruit￾ment, and subsequent interim analyses were to be performed between 6 and 12 months after the first interim analysis; the aim was to have approximate￾ly 35 events between interim analyses. The inten￾tion-to-treat analysis included all patients who were randomly assigned to receive either lamivudine or placebo. Treatments were compared with the use of a Cox proportional-hazards model,32 with each analysis allowing for the covariates of country, sex, * There were no significant differences between the two treatment groups. The patients’ race was recorded by the investi￾gators, on the basis of the interviews and evaluations of the patients. † The Child–Pugh score (range, 5 to 15, where 5 indicates good liver function and 15 poor liver function) is a measure of the severity of liver disease. ‡ The Ishak fibrosis score (range, 0 to 6) is a measure of the degree of fibrosis in liver-biopsy specimens. Scores of 0 to 4 indi￾cate no or moderate fibrosis, and 5 or 6 severe fibrosis or cirrhosis. § All patients had detectable HBV DNA at screening; 0.7 meq per milliliter equals approximately 7¬105 copies per milliliter. ¶To convert values for bilirubin to milligrams per deciliter, divide by 17.1. ¿ To convert values for creatinine to milligrams per deciliter, divide by 88.4. Table 1. (Continued.) Characteristic Lamivudine Group (N=436) Placebo Group (N=215) Aspartate aminotransferase — U/liter Median 52 54 Range 14–686 17–367 Bilirubin — µmol/liter¶ Median 13.7 13.7 Range 3.0–50.0 1.7–58.1 Creatinine — µmol/liter¿ Median 88 88 Range 35–173 35–135 Hemoglobin — g/dl Median 14.7 14.6 Range 8.4–19.0 9.2–17.8 Platelet count per mm3 Median 145,000 131,000 Range 14,000–401,000 41,000–360,000 Prothrombin time — sec Median 12.5 12.8 Range 8.0–23.8 9.8–27.6 White-cell count per mm3 Median 5330 5300 Range 1980–11,600 2200–11,500 第 16 页