1. Long-Term Comparison of Nevirapine Versus Efavirenz When Combined with Other Antiretroviral Drugs in HIV-1 Positive Antiretroviral-Naïve Persons- The NNRTI Substudy of the CPCRA 058 FIRST Study van den Berg-Wolf M 1, Peng G 2, Xiang Y 2, Huppler Hullsiek K 2, Chen L 2, MacArthur RD 3, Novak RM 4, Kozal M 5, Schmetter B 6, Henely C 7, Dehlinger M 8 1. Temple University, Philadelphia, Pa, USA 2. University of Minnesota, Minneapolis, Mn, USA, 3. Wayne State University, Detroit, Mi, USA 4. University of Illinois, Chicago, Il, USA 5. Yale University School of Medicine, New Haven, Ct, USA 6. Social and Scientific Systems, Silver Springs, Md, USA 7. Meridian Health System, Neptune, N J, USA 8. Division of AIDS, NIAID, Bethesda, Md, USA BACKGROUND CPCRA 058 (FIRST) was a randomized open-label trial comparing three initial antiretroviral treatment strategies in 1,397 HIV-infected antiretroviral-naïve persons. Treatment strategies are defined by classes of drugs, not specific antiretrovirals (ARVs). Enrollment was from March 1999 to January 2002 at 17 clinical units in the United States. Before randomization in the main study, participants were given the option of additional substudy randomizations or pre-selecting the drugs within each strategy arm. Participants consenting to the NNRTI substudy were further randomized to efavirenz or nevirapine (1:1 allocation), taken together with other classes of drugs within the strategy (the randomized cohort; Figure 1). In addition, we report on participants who were randomized to the NNRTI containing arms of FIRST, but chose to select the NNRTI used in combination with other ARVs (the non-randomized cohort; Figure 1). METHODS Population: participants naïve to active combination antiretroviral treatment. Primary Endpoint: time to HIV-RNA > 50 copies/mL (at or after month 8 visit) or death. Secondary Endpoints: progression to AIDS or death, changes in CD4+ cell count, percent of participants with HIV-RNA 1000 copies/mL (at or after month 4 visit), safety and tolerability of the drugs, and grade 4 adverse events. ARV therapy management: participants enrolled into the FIRST study could change ARVs for any reason; however, they were encouraged to stay within the assigned strategy if possible. Sample size: 200 primary events were required for 80% power to detect a hazard ratio of 0.67 at the 0.05 level of significance (2-sided). When the FIRST study closed, 187 primary events had occurred within the NNRTI substudy. Statistical Methods (intent to treat): Time to event analyses: Kaplan-Meier life table summaries and Cox’s proportional hazards models. The models were stratified by randomized strategy in the main FIRST study (2-class strategy with no PI or 3-class strategy). Hazard ratios compare efavirenz to nevirapine. Rates are per 100 person years of follow-up. Changes in CD4+ cell count: longitudinal regression models adjusted for baseline values. Proportion with HIV-RNA < 50 copies/mL: models for longitudinal binary data. Sensitivity Analyses: For both the randomized cohort and the non-randomized cohort, analyses were repeated with models adjusting for baseline factors considered to be potential confounding variables (age, race, gender, prior AIDS, history of injection drug use, baseline CD4 cell count and HIV RNA level). Those models were repeated for the combined data set, including an indicator for randomization status and the interaction between randomization status and NNRTI drug. *Resistance was defined as definite drug resistance by genotype (TRUGENE HIV-1 Genotyping Kit and CPCRA interpretive algorithm v4.0). RESULTS for Randomized Cohort The NNRTI substudy enrolled 228 participants: 111 to the efavirenz group and 117 to the nevirapine group (Figure 1). An additional 416 participants were prescribed efavirenz, and 261 were prescribed nevirapine. Median follow-up time was 65 months (inter-quartile range 57–71 months). Among participants randomized to the 3-class strategy, nelfinavir was the most frequently prescribed PI. One person in the nevirapine group was prescribed efavirenz at study entry (Table 2). About 75% discontinued the assigned NNRTI drug during follow-up (p=0.68 for difference between the two groups; Table 3). More participants in the nevirapine group either initiated a PI (among those in the 2-class strategy, p=0.08) or switched PIs (among those in the 3-class strategy, p=0.06) than the efavirenz group (Table 3). There was no significant difference between the groups for the primary endpoint (time to death or HIV RNA level > 50 copies/mL): hazard ratio (HR) = 0.92, p=0.59; Figure 2 and Table 4). The rate of AIDS or death was higher in the efavirenz group (HR = 1.67, p=0.06; Table 4); The rate of death only did not differ between the groups. The rate of grade 4 events was lower for the efavirenz group than for the nevirapine group (HR=0.55, p=0.02; Table 4). However, the percents with grade 4 elevated AST or ALT levels were similar (Table 5). The rate of discontinuation of randomized drugs due to toxicity was similar (HR=1.34, p=0.38, Table 4). The majority (86%) of the toxicities were grade 1 to 3. The rates of virologic failure (HIV RNA > 1000 copies/mL) were similar (HR = 0.80, p = 0.17; Table 4). However, the rate of virologic failure associated with any drug resistance was lower in the efavirenz group (HR = 0.60, p=0.02). The percent with HIV RNA < 50 copies/mL was not significantly different between the 2 groups (odds ratio = 1.26, p=0.24; Figure 3). CD4+ cell increases from baseline did not differ significantly over follow-up (p=0.25; Figure 4). The average CD4+ cell count increases over the entire follow-up period were 172 and 153 cells/mm³ for the efavirenz and nevirapine groups, respectively. In the NNRTI substudy, the efavirenz and nevirapine groups did not differ significantly for rate of HIV RNA > 50 copies/mL or death; however pooled results indicate that the nevirapine group has increased risk of the event. There was no differences found for CD4+ cell increases over time in the two groups. Nevirapine was associated with more drug resistance at the time of virologic failure. The different results for AIDS or death events in the randomized and non-randomized cohorts suggest residual confounding in the non-randomized cohort. Larger, longer-term randomized trials are needed to assess the clinical outcome of AIDS or death with adequate power comparing efavirenz with nevirapine as initial antiretroviral treatment options. Joint analyses of randomized and non-randomized cohorts can potentially improve power for treatment comparisons and permit assessment of residual confounding for non-randomized analyses. CONCLUSIONS TABLE 1: Baseline Characteristics* Efavirenz (N=111) Nevirapine N=117) Overall (N=228) Age (mean years) 383637 Female (%) 232223 Race/Ethnicity (%) Latino/Latina161817 African American60 White/other242223 Prior AIDS event (%) 3837 Hepatitis B (%) 555 Hepatitis C (%) 211618 History of injection drug use (%) 151817 2-class NNRTI Strategy (%) 475048 CD4+ cells (median cells/mm 3) 181196186 HIV RNA (median log 10 copies/mL) 5.05.15.0 *There were no significant differences between the two groups. Efavirenz (N=111) Nevirapine (N=117) Overall (N=228) PI(%) Nelfinavir2726 Indinavir14911 Ritonavir-boosted PI141514 No PI465048 NRTI (%) Abacavir + Lamivudine273029 Didanosine + Stavudine252927 Zidovudine + Lamivudine221619 Stavudine + Lamivudine111211 Single NRTI12911 Other343 * One person randomized to the nevirapine group was prescribed efavirenz at study entry. TABLE 2: Antiretroviral Drugs Prescribed at Study Entry* Efavirenz (N=111)Nevirapine (N=117) Overall(%) Switched to alternate NNRTI 3.62.6 Discontinued assigned NNRTI 69.472.6 2-class strategy (%) Switched to alternate NNRTI1.93.4 Discontinued assigned NNRTI69.270.7 Initiated PI28.844.8 Switched NRTI76.974.1 3-class strategy (%) Switched to alternate NNRTI5.11.7 Discontinued assigned NNRTI69.574.6 Discontinued PI62.761.0 Switched PI11.925.4 Switched NRTI74.684.7 *Categories are not mutually exclusive. There were no significant differences between the groups. TABLE 3: Antiretroviral Drug Changes* During Follow-up TABLE 4: Event Rates Efavirenz (N=111)Nevirapine (N=117) Hazard Ratio 2 95% CI)P-value No. EventRate 1 No. EventRate 1 Primary endpoint 3 8941.29842.80.92 (0.69-1.23)0.59 Death 203.8183.21.18 (0.62-2.23)0.61 HIV RNA > 50 copies/mL 8242.09445.90.89 (0.66-1.19)0.42 AIDS or death 347.4234.41.67 (0.98-2.83)0.06 HIV RNA > 1000 copies/mL 7128.78636.40.80 (0.58-1.10)0.17 With resistance to NNRTI drugs 3212.94920.60.65 (0.41-1.01)0.05 NRTI drugs 52.02510.50.20 (0.08-0.52)<0.01 Any ARV drugs 3313.35422.70.60 (0.39-0.93)0.02 Grade 4 event 245.44310.20.55 (0.33-0.9.)0.02 Discontinue randomized drug due to toxicity 204.5163.21.34 (0.70-2.59)0.38 1. Rates are per 100 person-years 2. Hazard ratios are for the comparison of efavirenz versus nevirapine 3. Time to death or to HIV RNA level > 50 copies/mL (at or after month 8 visit) TABLE 5: Hepatic Toxicity Efavirenz (N=111) Nevirapine (N=117) P- value No. Events % % >10 ULN* at least once during follow-up or grade 4 AE 76.3108.50.52 > 5 ULN* at least once during follow-up 1614.41815.40.84 >2.5 ULN* at least once during follow-up 3027.03832.50.37 * Upper limit of normal for AST or ALT, required at each follow-up visit Figure 2 Time to Death or First HIV RNA > 50 Copies/mL at or After Month 8 Visit Figure 3 Percent with HIV RNA < 50 Copies/mL Figure 4 Mean CD4+ Cell Count Change from Baseline The hazard ratios for the primary endpoint (HIV RNA > 50 copies/mL or death) favored the efavirenz group over the nevirapine group for both the randomized and non-randomized cohorts (pooled HR = 0.79, 95% CI: 0.66-0.94; Table 6). The hazard ratios for virologic failure associated with any drug resistance also favored the efavirenz group. (pooled HR = 0.74, 95% CI: 0.54-1.01; Table 6). For AIDS or death events, there was a significant interaction between NNRTI drug and substudy randomization status (Table 6). In the randomized substudy there was a significantly increased risk of AIDS or death in the efavirenz group that was not present in the non-randomized cohort. In the combined cohort, there was no difference between the groups for the rate of grade 4 events (Table 6) All results were consistent with on-treatment analyses. RESULTS of Comparison of Randomized and Non-randomized Cohorts in FIRST We would like to thank all the participants who volunteered to be part of this study. Supported by grants (5U01AI042170-10 and 5U01AI046362-03) from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, USA. Reprints: mvan@temple.edu Event Randomized to Substudy Not Randomized to Substudy 3 Combined Data 3 Interaction P-value 4 UnadjustedAdjusted 3 Primary 2 0.92 (0.69-1.23) 0.95 (0.71-1.28) 0.79 (0.67-0.95) 0.79 (0.66-0.94) 0.30 AIDS or death 1.67 (0.98-2.83) 2.21 (1.27-3.84) 0.90 (0.64-1.26) 0.88 (0.63-1.23) 0.02 HIV RNA > 1000 copies/mL with resistance 0.60 (0.39-0.93) 0.61 (0.39-0.95) 0.75 (0.55-1.02) 0.74 (0.54-1.01) 0.53 Grade 4 Event 0.55 (0.33-0.90) 0.51 (0.30-0.85) 0.89 (0.67-1.17) 0.88 (0.67-1.16) 0.08 1 Hazard ratios (95% CI) are for the comparison of efavirenz versus nevirapine 2 Time to death or to HIV RNA level > 50 copies/mL (at or after month 8 visit) 3 Adjusted for age, gender, race, prior AIDS, history of injection drug use, baseline CD4 cell count and HIV RNA level 4 From an interaction test between study drug (efavirenz or nevirapine) and substudy randomization status TABLE 6: Comparison of Hazard Ratios 1 of Randomized and Non-randomized Cohorts in FIRST