Efavirenz Use in HIV Infected Patients and Neurocognitive Impairment: Study Findings

1. www.aids2014.org Use of efavirenz is not associated to an increased risk of neurocognitive impairment in HIV-infected patients Pinnetti C 1 , Balestra P 1 , Libertone R 1 , Lorenzini P 1 , Cozzi-Lepri A 2 , Ammassari A 1 , Ricottini M 1 , Menichetti S 1 , Tozzi V 1 , Antinori A 1 1 Clinical Department, National Institute for Infectious Diseases Lazzaro Spallanzani IRCCS, Roma, Italy; 2 Research Department of Infection & Population Health, University College London, London, United Kingdom.

2. www.aids2014.org Disclosures Personal fees for consultancy and lectures from Abbvie, Bristol Myers Squibb, Gilead, Janssen, Merck, ViiV. Travel grants from Abbvie. Research grants from Bristol Myers Squibb, Gilead, Janssen, ViiV.

3. www.aids2014.org Background Efavirenz (EFV) is currently recommended as a preferred component of ARV regimens for treatment-nave patients in most of ART guidelines. Due to patent expiry in 2013, EFV is now available as a generic drug. EFV has been associated to a well recognized neuropsychiatric side effect pattern 1 , but an association with neurocognitive impairment (NCI) was not originally confirmed in a randomized trial in ARV naive 2 . More recently, EFV has been related to a worse neurocognitive function 3-4 , but this association remains controversial. 1 Mills A, et al. HIV Med 2013;14:391-400; Clifford DB, et al. Ann Intern Med 2005;143:714-721; 3 Ciccarelli N, et al. Neurology, 2011;76:1403-1409; 4 Letendre S, et al. 20 th CROI, Atlanta, GA, 2013; Poster #407.

4. www.aids2014.org Objectives Aim of the study was: to evaluate if patients receiving an EFV-based regimen had a worse neurocognitive performance compared to patients currently receiving an ARV regimen not including EFV; to estimate the risk of NCI independently associated to EFV current use. The prevalence and severity of HIV-associated neurocognitive disorders (HAND) according to EFV current use was also investigated.

5. www.aids2014.org Methods Single-centre, retrospective, cross-sectional analysis of cART-treated HIV- infected patients undergoing neuropsychological assessment (NPA). NPA was carried out by a set of 14 standardized and comprehensive tests on 5 different domains. People were classified as having NCI if they scored > 1 standard deviation (SD) below the normal mean in at least 2 tests, or > 2 SD below in 1 test. As additional binary outcome, HIV-associated neurocognitive disorders (HAND) were classified according to Frascatis criteria (Neurology, 2007). Clinical depression and other psychiatric disorders has been controlled as confounders. Z-Scores (neg values if performance was below the mean) were used as continuous outcome: Global NPZ-8 Deficit Score Z-Scores for each Cognitive Domain

6. www.aids2014.org NP Battery and Domains Concentration and Speed of Mental Processing Trail Making A WAIS-R Digit Span (forward) WAIS-R Digit Span (backward) WAIS R Digit Symbol Stroop Test (word and color) Corsi's Block-Tapping Test Mental Flexibility Trail Making B Stroop test (color and word) Controlled Oral Association Test Memory Rey Auditory Verbal Learning (immediate recall) Rey Auditory Verbal Learning (delayed recall) Rey Complex Figure (delayed recall) Fine Motor Functioning Lafayette Grooved Pegboard (dominant hand) Lafayette Grooved Pegboard (non dominant hand) Visuospatial and Constructional Abilities Rey Complex Figure (copy)

7. www.aids2014.org Statistical analyses Comparisons between currently EFV-exposed and currently EFV-not exposed Mann-Whitney test and Spearmans Rho for continuous variables Chi-square test for categorical variables Univariable and multivariable logistic and linear regression models were fitted. Variables were retained in multivariable analysis if they had a p-value <0.05.

8. www.aids2014.org Patients characteristics ( 1020 consecutive NPA over 859 patients) overall EFV (n=324) no EFV (n=696) p Male gender, n (%) 795 (77.9%) 265 (81.8%) 530 (76.1%) 0.043 Age, yrs, median (IQR) 46 (40-53) 46 (40-53) 46 (39-54) 0.874 Mode of HIV transmission, n (%) 0.002 MSM 400 (39.2%) 154 (47.5) 246 (35.3) IVDU 182 (17.8%) 44 (13.6) 138 (19.8) heterosexual 393 (38.5%) 112 (34.6) 281 (40.4) other/unkown 45 (4.4%) 14 (4.3) 31 (4.4) Years of education, median (IQR) 13 (8-13) 13 (8-17) 13 (8-13) <0.001 CDC C stage, n (%) 245 (24.0%) 55 (17.0) 190 (27.3) <0.001 HCV-Ab positivity, n (%) 285 (27.9%) 64 (19.7) 221 (31.8) <0.001 Nadir CD4 cell/mm 3 , median (IQR) 190 (82-300) 225 (112-323) 165 (73-282) <0.001 Current CD4 cell/mm 3 , median (IQR) 483 (287-689) 532 (354-714) 470 (261-671) <0.001 HIV-RNA cp/mL at NPA, n (%) <0.001 undetectable 279 (27.5%) 120 (37.1%) 159 (22.9%) detectable <40 423 (41.6%) 133 (41.2%) 290 (41.8%%) >=40 314 (30.9%) 70 (21.7%) 244 (35.2%) Haemoglobin <=12 mg/dL, n (%) 99 (9.7%) 22 (6.8%) 77 (11.1%) 0.023 Calendar year of NPA 0.001 2000-2004 256 (26.0%) 80 (24.7%) 185 (26.6%) 2005-2009 338 (33.1%) 86 (26.5%) 252 (36.2%) 2010-2013 417 (40.9%) 158 (48.8%) 259 (37.2%)

9. www.aids2014.org NCI impairment by EFV current use (NPZ-8) P=0.291 at univariable analysis (Mann-Whitney test) Spearman's rho=-0.029; P=0.525 Difference in slope for NPZ-8 from fitting multivariable linear regression models for patients receiving EFV vs those not receiving the drug [Beta -0.04 (95%CI -0.18; 0.09) p=0.514] NPZ-8 by months of drug exposure for EFV current regimen Median (IQR) exposure to current EFV regimen: 30 months (12-40).

10. www.aids2014.org EFV current use and NCI risk (Unadjusted and adjusted risk by logistic regression) * Subjects are defined as neurocognitivelly impaired if they scored > 1 standard deviation (SD) below the normal mean in at least 2 tests, or > 2 SD below in 1 test. ^ Sensitivity analysis including only the 859 subjects at the first NP assessment. # Risk of n eurocognitive impairment for EFV current use. Adjusted for age, mode of HIV transmission, years of education, CDC C stage, HCV-Ab positivity, nadir CD4, HIV-RNA and CD4 at NPA, haemoglobin and BMI at NPA, NRTI combination in the current regimen, calendar year of NPA) Older age, IVDU, HCV-Ab+, CDC-C stage, were related to an increased risk of NCI; current CD4 >500 and years of schooling to a decreased risk. Gender, CD4 nadir, HIV-RNA undetectable at NPA, anemia, BMI and CPE 2010 not associated.

11. www.aids2014.org Factors likely to have confounded the association (crude and adjusted OR of NCI by EFV exposure) Current EFV use was associated with a significantly decreased risk of NCI at univariable analysis (OR: 0.71 vs. no EFV; 95%CI 0.54-0.94) , but no longer significant after controlling for potential confounding (OR: 1.02; 95%CI 0.74-1.41) .

12. www.aids2014.org EFV current exposure and HAND occurrence *EFV-exposed individuals had a significant lower risk to have a symptomatic HAND (MND/HAD) by univariate analysis.

13. www.aids2014.org Beta and 95%CI of Z-score for cognitive areas from fitting multivariable linear regression models for patients receiving EFV vs those not receiving the drug Fully models were adjusted for gender, age, years of education, mode of HIV transmission, CDC-C stage, HCVAb positivity, nadir CD4, CD4 cell count/mmc and HIV-RNA cp/mL at NPA, haemoglobine value, BMI, NRTI backbone in the current regimen.

14. www.aids2014.org Limitations Observational nature of design channeling bias (if EFV preferably allocated in individuals at low- risk of developing NCI); unmeasured confounders (e.g. drug adherence). Cross-sectional analysis reverse causality (if EFV was avoided in patients with NCI); not considered the effects of previous regimens exposures. Major depression controlled as a confounder, but a systematical assessment of depression (e.g. BDI-II, PHQ-9 or CES-D) was not performed in all subjects.

15. www.aids2014.org Conclusions In this large case series, EFV exposure was not associated with an increased risk of NCI. Current EFV use was associated with a 30% decreased risk of NCI at univariable analysis, but no longer significant after controlling for schooling and other potential confounding. In subjects receiving EFV, no increased risk of HAND occurred. The duration of EFV exposure was not related to worse neurocognitive performance. No increased risk of impairment according to specific domains was observed in EFV-exposed. Even though confounding by indication may play a role, and reverse causality cannot be ruled out, our results suggest that presence of NCI among persons treated with EFV-based cART is not more common than in people not treated with EFV