Sample characteristics and summary of main study results
summarizes demographic and clinical features of the 152 participants included in the study, by treatment group. Mean age was 19.1 years, and mean duration of opioid dependence was 1.5 years. 76% used heroin, 56% used prescription opioids, and 50% injected drugs. Maximum doses for DETOX patients were as follows: 24 (31%) received 2 to 8 mg and 53 (68%) received 9 to 14 mg. For BUP patients, 20 (27%) received 2 to 8 mg, 43 (59%) received 9 to 16 mg, and 10 (14%) received 17 to 24 mg. As reported previously (Woody et al., 2008
), during-treatment substance use outcomes were much more favorable in the BUP group. At 4 and 8 weeks, 61% and 54% of DETOX patients had urines positive for opioids, vs. 26% and 23% of BUP patients, respectively. Self-reported opioid, cocaine, marijuana, and injection drug use were also significantly lower in BUP patients (Odds ratios (95% CI) = 4.30 (2.25–8.22), 16.39 (3.07–87.47), 6.15 (2.10–18.01), 3.54 (1.27–9.87), respectively).
Baseline Sample Characteristics by Treatment Assignment*
Rates of transaminase abnormalities at baseline and during treatment
Overall, 8/60 BUP participants (13.3%, 95% CI 6.3–25.1%) vs. 12/51 DETOX participants (23.5%, 95% CI 13.3–37.8%) had at least one elevated ALT value during follow-up (Chi-square n.s.). 5/60 BUP participants (8.3%, 95% CI 3.1–19.1%) vs. 11/51 DETOX participants (21.6%, 95% CI 11.8–35.7%) had at least one elevated AST value (Chi-square = 3.194, p = .048). illustrates rates of abnormal transaminase values for each time point in BUP and DETOX participants. Elevated AST values were significantly more common in the DETOX group than in the BUP group at 4 and 8 weeks. Two individuals in the DETOX group and 2 in the BUP group developed markedly elevated transaminases (greater than 5 times the upper limit of the normal range).
Percent ALT and AST abnormalities at baseline, 4 weeks, 8 weeks, and 12 weeks by treatment group (BUP vs. DETOX) and HCV status (HCV+ vs. HCV−)
shows the mean transaminase values at each time point for participants assigned to BUP vs. DETOX. When means were compared, the only significant difference between BUP and DETOX was the 8-week AST value (24.3±10.3 vs. 40.6±45.7, U = 649.5, Z = 2.49, p = .013). No significant differences were found for any of the other liver chemistries at any time point.
Mean transaminase values (± S.D.) before and during treatment, by treatment assignment and HCV status
Relationship of HCV status and treatment assignment to transaminase abnormalities
28/152 (18.4%) of participants were HCV positive at baseline. Highly significant differences were found between sites in rates of HCV, with 3/3, 12/51, 0/28, 11/39, and 2/30 participants HCV positive at baseline at each of the participating nodes (Excluding the node that had only 3 subjects, Chi Square = 11.93, p = .008). Four participants seroconverted within 12 weeks, 2 in each group. A fifth participant had newly elevated liver chemistries at 4 weeks but was lost to follow-up until 12 months, at which time he had seroconverted.
Combining across time points, 9/24 participants (37.5%, 95% CI 19.6–59.2%) who were HCV seropositive at baseline or at 12 weeks had at least on elevated ALT value, vs. 11/87 (12.6%, 95% CI 6.78–21.9% of those who were HCV-negative (p = .009, Fisher’s Exact Test). 8/24 HCV-positive participants (33.3%, 95% CI 16.4–55.3%) had an elevated AST value during follow-up, vs. 8/87 (9.2%, 95% CI 4.3–17.8%) of those who were seronegative (p = .006, Fisher’s Exact Test). illustrates the corresponding rates of transaminase abnormalities by HCV status at each time point. Again, significant differences were found for both transaminases at each time point with the exception of week 8, where the differences were not statistically significant. shows mean transaminase values of HCV+ and HCV-participants at each time point. Values for both transaminases were significantly higher in the HCV+ group for all time points except week 8.
Logistic Regression models
Logistic regression was used to assess the independent contributions of treatment assignment, HCV status, and their interaction to the presence of any abnormal AST or ALT value during treatment (n = 111). The baseline transaminase value was entered first as a covariate. In these models (), treatment assignment was a significant predictor of both ALT and AST elevation. In addition, the interaction between treatment assignment and HCV status was significantly related to ALT abnormalities, and was related to AST abnormalities at the level of a trend. After accounting for these effects, HCV status was a significant predictor of neither ALT nor AST elevations. This interaction may be interpreted by noting that among DETOX patients, 8/12 (66.7%, 95% CI 35.4–88.7%) who were HCV+ had at least one abnormal ALT value, compared to 4/39 HCV- participants (10.3%, 95% CI 3.3–25.2%) (Chi-Square = 16.229, p = .0001). Among BUP patients, 1/12 HCV+ participants (8.3%, 95% CI 0.4–40.2%) had an abnormal ALT value, not significantly different from 7/48 (14.6% 95% CI 6.5–28.4%) of the HCV-participants. Similarly, for AST, 7/12 DETOX patients (58.3% 95% CI 28.6–83.5%) who were HCV+ had at least one abnormal AST value, compared to 4/39 (10.3% 95% CI 3.3–25.2%) of HCV- participants (Chi-Square = 12.538, p = .0004). Among BUP patients, 1/12 (8.3% 95% CI 0.4–40.2) of HCV+ participants had an abnormal AST value, equivalent to 4/48 (8.3% 95% CI 2.7–20.9%) of the HCV- participants.
Logistic regression models of the effects of treatment assignment, HCV status, and their interaction on presence of transaminase abnormalities*
Relationship of transaminase abnormalities to use of hepatotoxic substances
It is plausible that the relationship between treatment assignment and transaminase abnormalities could be mediated by greater use of alcohol, acetaminophen or prescription opioid use in the DETOX group. Prescription opioids are of interest in this regard because they are frequently compounded with acetaminophen. 19/51 DETOX vs. 27/60 BUP participants reported acetaminophen use during the trial (n.s.). There were no significant relationships between reported acetaminophen use and transaminase elevations, neither for the entire sample nor the HCV-positive sub-group. Alcohol use was more frequent at 4 weeks in DETOX than in BUP participants (1.04 days per week vs. .34 days per week, p = .008), and opioid use (excluding heroin and methadone) was higher at 4 and 8 weeks (respectively .93 vs. .16 days per week, p = .010; .62 vs. .10 days per week, p = .012), However, neither was significantly correlated with transaminase abnormalities or values at any time point. Again, this was true for both the entire sample and those who were HCV-positive. Thus we found no evidence that toxicity due to these substances mediated the observed treatment effect.
Finally, because of the possibility that buprenorphine itself could be hepatotoxic, we examined the distribution of buprenorphine doses among the participants who had elevated transaminases. Among the 9 such individuals in the BUP group, maximum dose was 8 or lower for n=3, between 9 and 16 for n=4, and between 17 and 24 for n=2, a distribution very similar to that of the entire bup-treated sample. For the DETOX participants (who received buprenorphine only for the first two weeks of treatment) 3 of the 15 participants with an abnormal transaminase had a maximum dose of 8 or less, and 12/15 had a dose between 9 and 14. This is a somewhat lower proportion of low-dose participants in the group with abnormal transaminases (3/15 = 20%, vs. 31% in the whole sample), but the difference is not statistically significant (p = .366, Fisher’s exact).