European Clinical Data on HIV-1 Coreceptor Usage and Genotypic Identification of Tropism in HIV-2 Matthias Döring Max Planck Institute for Informatics May 8, 2015
Coreceptors are crucial for HIV-1 cell entry Delhalle et al, Int. J. Mol. Sci. 2011 May 8, 2015 2
EucoHIV: European HIV coreceptor study panel doctor.madza-wordpress-premium-themes.com Goals Clinical significance of coreceptor usage Validation of geno2pheno [coreceptor] May 8, 2015 3
EucoHIV: European HIV coreceptor study panel doctor.madza-wordpress-premium-themes.com Data Twenty clinical centers in Europe Measurements from more than 450 patients: Required Optional – Viral load – CD4 cell count – Tropism – V3 sequence – Env sequence May 8, 2015 3
Criteria for selecting patient records Measurement Cutoff (viraemic) Cutoff (suppressed) V3/Tropism ≤ 12 weeks earlier Viral load ≤ 24 weeks ≤ 24 weeks CD4 count ≤ 16 weeks earlier All cutoffs are with regard to therapy start May 8, 2015 4
Overview of the data set Ethnicities Viral subtypes count 100 150 50 0 0 A A (01_AE), A (A1) A/E A1 B B,D C CRF 01_AE Subtype CRF 02_AG CRF 05_DF CRFAG D May 8, 2015 F1 Groupe M Groupe M, PR:F/RT:B U non B urf,D urf,D,CRF 5
Overview of the data set Treatment status Therapy-experienced Previous antiretroviral therapy Suppressed VL: < 50 cps/ml Detectable VL: ≥ 50 cps/ml Therapy-naive First-line treatment May 8, 2015 5
Treatment status and immune status of patients Baseline CD4 counts May 8, 2015 6
Virologic response to Maraviroc Approach Change in viral loads from baseline 1. Use geno2pheno[coreceptor] to compute FPRs 2. Classify as by applying different FPR cutoffs: German guidelines: 15 % GB guidelines: 5 . 75 % EU guidelines: 10 % / 20 % May 8, 2015 7
Virologic response to Maraviroc Status-specific change in viral loads May 8, 2015 7
Immune response Overall rate of immunologic success 0.7 0.6 Ratio of patients with CD4 > 400 cells Method Pheno/g2p_FPR=15_R5 0.5 Pheno/g2p_FPR=15_X4 g2p_FPR=15_R5 g2p_FPR=15_X4 g2p_FPR=5.75_R5 g2p_FPR=5.75_X4 0.4 g2p_FPR=EU_R5 g2p_FPR=EU_X4 0.3 0.2 0 2 4 6 8 12 16 20 24 32 40 48 Time [Weeks] May 8, 2015 8
Therapy success Success criteria Combined outcomes VL < 50 cps/ml CD4 ≥ 400 cells/ µ l May 8, 2015 9
HIV-2 coreceptor usage and MVC Background Motivation No genotypic tool for HIV-2 Fewer treatment options coreceptors available Major determinant: V3 loop, Treatment with MVC is an but V1 and V2 also important option (Armstrong-James, (Santos-Costa, 2014) 2010) Rules for V3 loop exist (Visseaux, 2012) May 8, 2015 10
Statistical learning on the HIV-2 V3 loop sequence Data Set Model Pairs of sequences and Support Vector Machine phenotypic annotation of tropism Input: V3 amino acid sequences R5 X4-capable Total Output: FPR for an X4-capable prediction 89 49 138 Results Performance: balanced accuracy of 0.897 Interpretability: FPRs and positions in V3 May 8, 2015 11
Positional weights highlight important amino acids S I K AA Y A K E 0.3 H F G R X4 V R H Q I R K K F M Q F K K K G H Y L Weight H I L 0.0 F K V N A M I R T M T S P E Q R N V P Q R5 R L S −0.3 I T V Y 18 19 23 INS(1)_24 28 24 22 13 11 27 2 10 15 INS(1)_22 INS(2)_22 14 20 8 12 Position May 8, 2015 12
Summary and conclusions EucoHIV HIV-2 Coreceptor Identification First, large European Most important region: V3 loop data set on tropism and Individual positions are highly MVC treatment predictive MVC improved both, Genotypic prediction of tropism is virologic and immune possible status considerably – Interpretable – High accuracy May 8, 2015 13
Thank you for your attention and thanks to . . . Thomas Lengauer and Nico Pfeifer The EucoHIV steering committee – Charles Boucher – Anna Maria Geretti – Rolf Kaiser – Maurizio Zazzi – Anne-Mieke Vandamme Viiv Nuno Taveira and Pedro Borrego May 8, 2015 14
Recommend
More recommend