Are the Proposed env Mutations Actually Associated With Resistance to Maraviroc?

To the Editors: To initiate infection, HIV-1 requires an interaction with a chemokine co-receptor, which, in addition to CD4 receptor, permits entry into target cells. The 2 major co-receptors are CCR5 and CXCR4, and according to their differential use, HIV-1 isolates are classified as R5, X4, and R5X4 (dual/mixed, D/M), which utilize both co-receptors. Maraviroc,1 the first commercially available co-receptor antagonist, is an extracellular HIV-1 inhibitor designed to block the entry of virions into the target cell by binding the CCR5 co-receptor and, as such, is believed to be ineffective in patients harbouring X4 viruses while its effectiveness on D/M carriers is still controversial.2 Resistance to maraviroc can develop with either of 2 mechanisms: selection of a X4 or R5/X4 virus or development of mutations in the gp120 V3 loop, permitting HIV-1 to continue using CCR5, despite the maraviroc blockade. At present, a signature mutation for maraviroc resistance has yet to be identified. However, 2 mutations, A316T and I323V (HXB2 numbering, corresponding to positions 19 and 26 within the 35-amino acid-long V3 loop), seem to be selected by maraviroc in vitro,3 whereas a larger number of variously combined amino acid substitutions were found to be associated with resistance to anti-CCR5 inhibitors in patients enrolled in phase 3 MOTIVATE-1 and -2 trials.4,5 Due to the high genetic variability of the V3 loop region, the possibility that these mutations can occur in maraviroc-naive patients has been suggested.6 However, the extent of this eventuality, which would be relevant for the correct selection of candidates for anti-CCR5-based therapy, has not been quantified. To further investigate this issue, 239 sequences from 189 HIV-infected maraviroc-naive subjects were analyzed to detect substitutions reported as associated to maraviroc resistance both in vitro and in vivo. Patients were randomly enrolled among those with detectable HIV-1 viral load (>1000 copies/mL), irrespective of clinical/treatment status. At sampling, mean CD4 count and plasma viral load were 243 ± 200 cells per microliter and 4.6 ± 1.1 log10 copies per milliliter, respectively; 60% of patients was HAART naive. Phylogenetic analysis (neighbor-joining method using Kimura 2-parameter distances and Simplot analysis) of pol and env genes identified 118 of 239 sequences (49%) as either pure non-B subtypes (21C, 20F, 12G, 6A, and 1D) or recombinant forms (42 CRF_02 AG,8 CRF12_BF, 7 CRF01_AE, and 1 CRF06_cpx). For V3 loop sequencing and subtyping, a 947-bp fragment encompassing almost the entire gp120 from codon 43 of C1 to codon 2 of V4 was amplified from HIV-1 RNA with a nested polymerase chain reaction protocol. Polymerase chain reaction products were sequenced in both 5′ and 3′ directions using the BigDye dye terminator cycle sequencing kit (Applied Biosystems, Foster City, CA) and run on an ABI 310 automated sequencer.7 A multiple alignment of patient sequences (GenBank accession numbers: GQ175591-GQ175784, DQ984220-DQ98476, and FJ798215-FJ798302) with the respective reference strain of V3 region for X4 and R5 viruses, HXB2 (K03455) and ADA (AF004394), was obtained with ClustalW2 (http://www.ebi.ac.uk/Tools/clustalw2/). WebPSSM (http://ubik.microbiol.washington.edu/computing/pssm/) was arbitrarily chosen for classifying sequences as R5 (200 of 239, 84%) and X4 or D/M (39 of 239, 16%) strains. Pearson χ2 test (Fisher exact test when appropriate) was used to compare categorical variables; a P value of <0.05 was considered statistically significant. The majority (97.5%) of sequences harboured at least 1 substitution among those previously described as associated with maraviroc resistance.3-6 In decreasing frequency, they included 20F (82.4%), 11S (74.5%), 25D (50.2%), 13H (39.3%), 22T (28.5%), 19T (20.1%), 26V (8.4%), 13S (7.9%), 18G (2.5%), 19S (3.3%), and 21I (0.4%). The substitutions 19S (P = 0.03), 19T (P < 0.001), and 20F (P = 0.006) were significantly more frequent in non-B than in B subtypes, whereas substitutions 13H (P < 0.001), 13S (P = 0.002), and 22T (P = 0.002) were less common in non-B subtypes. A different distribution of these substitutions was also observed between R5 and X4 viruses; in particular, 11S and 25D were significantly more frequent in R5 (P < 0.001 and P = 0.016, respectively) (Table 1).Table 1: Distribution of Mutations Playing a Role in Maraviroc ResistanceEven though some substitutions were combined (11S + 20F, 66.1%; 11S + 20F + 25D, 40.6%), the association of mutations linked to maraviroc resistance, which have emerged in clinical trials and in vitro studies3-6 were uncommon. In particular, the 19S/T + 26V association, causing maraviroc resistance in vitro,3 was observed in 5 cases (2.1%), all of which were classified as R5 and non-B subtype strains. Conversely, the combination of 13S/H + 26V,4-6 also involved in maraviroc resistance in site-directed mutagenesis experiments,4 was limited to 9 cases (6 in B subtypes). Other associations were 11S + 26V4-6 (13 sequences, 5.4%), 20F + 25D + 26V4-6 (5 sequences, 2.1%), 18G + 22T and 20F + 21I5,6 (1 sequence each, 0.4%) (Table 1). Overall, it seems that single substitutions related to maraviroc resistance were common in our series. This result is not surprising considering the extensive V3 env polymorphism. In particular, 11S and 25D are determinant for the 11/25 genotype and are also the signature amino acids of the consensus of consensus sequence (Los Alamos database, http://www.hiv.lanl.gov/). When analyzing the first 1200 sequences downloaded from the Los Alamos database (query: only CCR5, any subtype, sample data before 2005), 11S or 25D alone accounted for 85% and 45% of sequences, respectively. The association of 11S + 25D and 11S + 25E were the most frequent patterns (44.8% and 23.1%, respectively). In our series, 48 sequences (20.1%) demonstrated the 19T substitution, 45 (93.7%) of which were non-B subtypes; in fact, this substitution, even if rare in B subtypes, is quite common in several non-B subtypes including A1 and C pure subtypes and AG and BC circulating recombinant forms.9 Likewise, 13H, which is the consensus amino acid of the B subtype, was obviously more frequent among B subtype sequences (P < 0.001), and 20F (consensus of consensus), although prevalent in non-B sequences (P = 0.006), was observed in both B and non-B groups (76% vs 89%). Based on these data, we are reluctant to attribute a key role for these mutations, either single or combined, in provoking maraviroc resistance, which actually seems to be rare even among patients on functional maraviroc monotherapy or with ongoing low-level replication,10 and caution must be taken in drawing conclusions from preliminary reports.4-6 ACKNOWLEDGMENT We thank Mrs. Paulene Butts for the review of the article. Annalisa Saracino, MD* Laura Monno, MD† Gaetano Brindicci, MD† Giovanna Trillo, MD† Maurantonio Altamura, MD† Grazia Punzi, PhD* Antonella Lagioia† Gioacchino Angarano, MD* *Clinic of Infectious Diseases, University of Foggia, Foggia, Italy †Clinic of Infectious Diseases, University of Bari, Bari, Italy