COMBINE Manuscript Proposal Form
For Administrative Use OnlyMS # CM09 / Date Received: 5-19-06 / Date Reviewed: 5-19-06
Priority Set: / Scheduled for Review: / Action: approved
Title: Health Effects of Medications
Proposed Writing Group Members: Domenic A. Ciraulo MD, Robert Swift MD, Barbara Mason Ph.D., J. C. Garbutt MD, and Marston Youngblood
Proposal:
The primary purpose of the health effects paper is to examine the safety and tolerability of the active drug treatments compared each other and to placebo. The secondary goal is to determine if there is improvement in physical health during the course of treatment, and whether there are differences between active drug treatments and placebo.
Introduction:
Therapeutic doses of acamprosate and naltrexone have good safety profiles, both in clinical practice and clinical trials (Croop et al 1997; Mason 2005; Anton and Swift 2003). A recent meta-analysis (Carmen et al 2004) found that both drugs were well tolerated, with acamprosate most often associated with gastrointestinal effects (diarrhea) and naltrexone with nausea, dizziness, and headaches (although headaches were common in both naltrexone and comparison groups). None of the studies in that meta-analysis used the doses of drugs studied in COMBINE. We are aware of only two published clinical trials examining the combination of naltrexone and acamprosate, again at lower doses than used in COMBINE. In the first (Kiefer et al 2003), found that only diarrhea and nausea were more common in the combination therapy group compared to the monotherapy groups. Laboratory values did not differ among the drug groups. In the second study (Feeney et al 2006), “medication complaints were uncommon” and usually gastrointestinal in nature, but detailed data are not reported.
The COMBINE trial offers the opportunity to examine drug safety and tolerability at high therapeutic doses of acamprosate, naltrexone, and the combination. The data set provides an opportunity to examine the adverse effects and health consequences of naltrexone 100 mg, acamprosate 3000 mg, and the combination in a 16-week clinical trial, where safety was assessed by the SAFTEE-GI, clinical chemistry analyses, and vital signs. Physical health was assessed using the WHO QOL and the SF-12 QOL Physical Health components.
Hypotheses:
1. Naltrexone alone (N), acamprosate alone (A), and the combination (AN) will have more adverse effects than placebo over the 16 week period of active treatment. The pattern of adverse effects will differ between groups over the 16 week treatment period. Adverse effects of interest are: 1) Vital Signs 2) Laboratory Measurements (see November 2004 COMBINE Outcomes Report for references to Tables): a) mean values ALAT, ASAT, GGT, creatinine, BUN, Bilirubin (Table 48) b) Percentage of renal and hepatic measures three times and five times above normal (Tables 49 and 50) c) Study Specific Adverse Events (Table 51) d) Serious Adverse Events (Tables 52-55)
2. Physical Health as reported at Week 16 and 26 on the WHO QOL Physical Health and SF-12 QOL Physical Health will be improved from baseline in all active drug treatments compared to placebo
3. The longitudinal pattern of adverse effects will be described over 16, 26, and 52 weeks for each of the four groups to determine if specific adverse effect patterns are present. We hypothesize that adverse effects will be greatest early in treatment and improve over the 16 week course of active treatment.
4. To determine if the rReporting of adverse effects by SAFTEE-GI will be greater in the MM groups compared to the MM/CBI groupsdiffers between psychosocial intervention cells (i.e, are there different rates of adverse effects reporting in the CBI plus MM groups compared to the MM alone groups, with comparisons made within drug groups?) on active medications.
Approach:
Sample and Procedures:
The subset of patients randomized to study medications (N=1226) is the group of interest. Since CBI is considered a priori to have no influence on number and frequency of adverse experiences, the analysis will collapse across the CBI therapy condition to focus on medication related effects.
Assessments:
The structured instrument for reporting adverse experiences, SAFTEE-GI, was routinely collected at each of the 9 MM visits. The general inquiry format elicited self report of adverse experiences without prompts and allowed the MM practitioner to record severity of event, relationship to study medication, and note a course of action. Vital signs were also recorded at each MM visit. Clinical chemistry analyses were run on specimens collected monthly during treatment and repeated as needed when indicated by LFT monitoring.
Data Analytic Plan:
Inclusion / Exclusion Rules:
The CBI only group (n=157) will be excluded from the analysis of randomized patients because that arm received no study medication and as a consequence was not systematically evaluated for emergent adverse experiences like the groups receiving the MM intervention. The remaining 1226 patients will be analyzed on an intent to treat basis by combining like medication groups across the CBI intervention condition.
Description of Data Sources (forms) Used:
The chemistry panel results for specimens collected prior to and during treatment analyzed at the core clinical chemistry laboratory for the study will be used. Laboratory results for patients who did not withdraw from MM or MA is available at the time point weeks 0 (baseline), 4, 8, 12, and 16 (end of treatment). The SAFTEE is available for all patients who received MM during the treatment course and was administered during each visit. Physical health will be obtained from the WHO and SF-12 Physical Health components.
Proposed Analysis:
Objectives:
The active medications will be compared to placebo for evaluation in improvement over time in measures of physical health and in reduction in adverse experiences. The associations between active medications and specific patterns of adverse experiences will be explored.
Hypotheses to Test:
1) The hypothesis that Naltrexone alone (N), acamprosate alone (A), and the combination (AN) will have more adverse effects than placebo during treatment will be tested across the different dimensions of adverse experiences (physical signs and symptoms, clinical chemistries, LFT alert levels). The association of medications, singly and in combination, with a specific pattern of adverse effects will be tested by mixed linear and logistic models that control for baseline PDA, baseline lab value, and center. To control for the effects of active drug exposure in relation to side effects we will adjust the models for amount of study medication comsumedconsumed reported at the same time point as the adverse experience report. Drinking status can similarly be contolledcontrolled by use of the corresponding average drinks per day consumed during the assessment period for AEs.
2) Mixed linear models controlling for baseline PDA and center will evaluate differences in the domains of WHO QOL Physical Health and SF-12 QOL Physical Health from baseline to Week 16 and 26 for all active drug treatments compared to placebo by combining like medication groups across the CBI intervention condition.
3) Time trends will be evaluated for of proportion of participants reporting specific adverse effects over treatment using GEE or generalized linear mixed models methods to account for the repeated measures nature of the AE data. The influence of missing data can be reduced by using a more robust analytic approach using mixed model methods for repeated binary outcomes (GLIMMIX) which we will utilize and compare to the GEE results which are more sensitive to the problem of missing data points. The physical component summary score (PCS) of the SF-12 provides one global measure of health effects that can be described over time.
4) The contribution of the main effects of CBI therapy in explaining reported adverse experiences will be evaluated in the logistic models by controlling for all 3 factors (acamprosate , naltrexone and CBI) and interactions. The effect of CBI therapy will be tested by comparing the results of the reduced analytic model to the more fully saturated ones.
1