Transmission Networks of Drug Resistance Acquired in Primary/Early Stage HIV Infection

Abstract

OBJECTIVES

Population-based sequencing of primary/recent HIV infections (PHI) can provide a framework for understanding transmission dynamics of local epidemics. In Quebec, half of PHI represent clustered transmission events. This study ascertained the cumulative implications of clustering on onward transmission of drug resistance.

METHODS

HIV-1 pol sequence datasets were available for all genotyped PHI (<6 months post-seroconversion (n=848 subtype B infections, 1997-2007). Phylogenetic analysis established clustered transmission events, based on maximum likelihood topologies having high bootstrap values (>98%) and short genetic distances. The distributions of resistance to nucleoside and non-nucleoside RT inhibitors (NRTIs and NNRTIs) and protease inhibitors (PIs) in unique and clustered transmissions were ascertained.

Results

Episodic clustering was observed in half of recent/early stage infections from 1997-2008. Overall, 29% and 28% of new infections segregated into small (<5 PHI/cluster, n=242/848) and large transmission chains (≥5 PHI/cluster, n=239/848), averaging 2.8 ± 0.1 PHI/cluster and 10.3 ± 1.0 PHI/cluster, respectively. The transmission of nucleoside analogue mutations and 215 resistant variants (T215C/D/I/F/N/S/Y) declined with clustering (7.9% vs. 3.4% vs. 1.2% and 5.8% vs. 1.7% vs. 1.1% for unique, small and large clustered transmissions, respectively). In contrast, clustering was associated with the increased transmission of viruses harbouring resistance to NNRTIs (6.6% vs. 6.0% vs. 15.5%, respectively).

CONCLUSIONS

Clustering in early/PHI stage infection differentially affects transmission of drug resistance to different drug classes. Public health, prevention and diagnostic strategies, targeting PHI, afford a unique opportunity to curb the spread of transmitted drug resistance.

Introduction

The incidence and prevalence of HIV infections in Western settings have risen in the era of highly active antiretroviral therapy (HAART), particularly among male-sex-male (MSM) populations [1-3]. Comprehensive surveillance strategies are required to monitor emerging trends in HIV transmission and assess the effectiveness of intervention strategies [4-6].

Sequence data acquired from systematic antiretroviral resistance testing programs of primary HIV infections can provide a framework for studying transmission dynamics of local epidemics [7-14]. Accumulated HIV pol sequences allow for phylogenetic reconstruction of possible transmission events. Phylogenetic, mathematical and epidemiological modelling can be combined to better understand factors leading to amplified transmission risk at early/acute disease stages [5-16].

In Quebec, half of early stage infections (n=481/848) are phylogenetically linked to other primary infections, with 28% (n=239) forming large clusters, averaging 10 primary HIV infections (PHI) per cluster [8]. The present study was designed to evaluate the cumulative effects of clustering on forward transmission of viruses containing drug resistance mutations related to nucleoside and non-nucleoside reverse transcriptase inhibitors (NRTIs and NNRTIs), as well as protease inhibitors (PIs).

Methods

Study Population

Our PHI study population was drawn from the provincial genotyping program, at either of two Quebec reference laboratories (2001-2007, n=645) and the Quebec PHI cohort study (1997-2007, n=249). The inclusion criterion for the genotyping program, (PHI< 6 months), was based on laboratory requisitions completed by prescribing physicians. These patients have been infected for an average of 4.9 months [17]. The Quebec PHI cohort study provides demographic data and a serologic testing algorithm for recent seroconversions (STARHS) [2,8].

Sequence data were compiled using non-nominative identifiers and cross-identifiers to ensure confidentiality while controlling for repeat sampling. In all, 848 unique PHI (subtype B infections) were identified. Ethics approval was obtained from individual study sites, the Laboratoire de santé publique du Québec, and the Quebec Ministry of Health committee on confidentiality and access of information.

Phylogenetic Analysis

Genotyping was carried out as previously described to generate sequences spanning the protease and reverse transcriptase regions [8]. All sequences were aligned to consensus HXB2 sequences, removing gaps and cutting to identical sequence lengths using BioEdit software [18]. Phylogenetic interrelationships among viral sequences were estimated using Neighbour Joining (NJ) trees and Maximum Likelihood methods using BioEdit and MEGA2 integrated analysis software [18,19]. Clustering was based on the robust statistical criterion of high bootstrap values (>98%) and short genetic distances [8,11,12].

Sequence data identified minor and major resistance mutations [20]. For this study, the standardized list of the HIV-1 PR and RT mutations established by the WHO HIV Drug Resistance Surveillance Programme (V. 07-08-05) was used for comparative analysis of the epidemiology of transmitted resistance [21]. Differences in drug resistance motifs among clustered and non-clustered primary transmissions and chronically infected groups were ascertained using contingency tests [22].

Cell culture-based phenotypic assays, were used to assess drug susceptibilities of select transmitted resistant variants [23-25]. Four G190A isolates from a large transmission cluster (n=27), (GenBank accession numbers EU375798-EU375801), were assessed for baseline susceptibility to nevirapine (NVP), efavirenz (EFV), etravirine (ETV) and TMC-120.

Results

PHI Study Population

Between January 1998 and July 2007, sequence data was available from 848 persons, harbouring subtype B infections having clinical indications of PHI <6 months. Temporal changes in baseline demographics, clinical features, cluster dynamics and transmitted resistance within our PHI cohort are summarized in Table 1. As illustrated, the extent of PHI surveillance improved in 2001 with the introduction of the provincial genotyping program. Our cohort is representative of infections in Quebec [26].

Table 1.

Baseline clinical features, viral load, clustering subgroupings and acquired drug resistance of primary/recent HIV infections (PHI) diagnosed between 1997-2007.

Baseline Characteristics	Sub Group	Year of PHI diagnosisa
		1997-2001 (n=112)	2002-2003 (n=233)	2004-2005 (n=275)	2006-2007 (n=228)
Gender (% PHI)	Male	92.9	87.9	93.0	92.9
	Female	7.0	12.0	7.0	7.1
Age (yr)	Unique	38.4 ± 1.0	38.4 ± 0.9	37.4 ± 0.9	39.9 ± 1.1
	Cluster	36.3 ± 1.3	37.1 ± 0.8	36.9 ± 0.7	36.1 ± 0.8
Viral load (log copies/ml)	Unique	4.6 ± 0.1	4.7 ± 0.1	4.5 ± 0.1	4.7 ± 0.1
	Cluster	4.5 ± 0.2	4.7 ± 0.1	4.7 ± 0.1	4.8 ± 0.1
Risk group ( % PHI)b	MSM	47.0	77.0	82.5	88.8
	IDU	45.0	11.0	6.3	7.4
	Heterosexual	7.8	11.1	11.1	3.7
Cluster subgroup A c All PHI (% PHI)	Unique	50.4	46.4	42.8	43.0
	Small clusters  (>5 PHI)	30.9	24.6	28.0	28.5
	Large clusters  (≥5 PHI)	18.5	28.8	28.4	28.1
Cluster subgroup B c Male only, non- IDU  (% PHI))	Unique PHI	50.0	49.3	49.8	50.0
	Small Clusters (mean PHI ± sem)	26.9 (2.9 ± 0.3)	23.7 (2.9 ± 0.1)	24.2 (2.9 ± 0.1)	25.6 (2.8 ± 0.1)
	Large clusters (mean PHI ± sem)	23.0	26.9	25.9	24.3
		(11.9 ± 1.6)	(10.6 ± 0.7)	(14.3 ± 0.9)	(13.7 ± 1)
Drug Resistance (% PHI)	Unique	15.7	16.5	21.2	13.1
	Clustered	33.9	16.8	13.0	18.5

^aBaseline clinical characteristics of all genotyped PHI have been stratified according to year of diagnosis.

^bRisk group for participants in the PHI cohort study (n=249).

^cCluster subgroup information is presented for all PHI (Cluster group A n=850), as well as for the male non-IDU subpopulation (n=730).

MSM, male-sex-male,; IDU-intravenous drug user

Clustering represented a steady ∼50% of PHI transmission events between 1997-2008. This is evident for the male-sex-male (MSM) population, i.e. the male only population that excludes intravenous drug users (Table 1). The accumulation time for clusters averaged 16.5 ± 9.8 months (mean ± sd), using STARSH for the PHI cohort. However, the majority of transmissions within clusters occurred over shorter 6-12 month periods [8, 10].

Overall, phylogenetic analysis identified 368 unique PHI, 89 small clusters (<5 PHI/cluster) and 30 large clusters (≥5 PHI/cluster) representing 43.3%, 28.5% and 28.1% of transmission events, respectively. There was an increase in the number and size of existent clusters between December 2005 and July 2007. The numbers of infections in small clusters and large clusters increased from 2.7 ± 0.2 to 3.4 ± 0.2 and from 8.8 ± 1.0 to 10.3 ± 1.2 PHI/cluster, respectively (mean ± sem).

It should be noted that non-B subtype infections, including subtypes C, A/AG, AG, G, and complex recombinant forms, introduced into Quebec (n=410) through recent immigration have been excluded from our analysis. Clustering among non-B subtypes was low (13.7%), largely restricted to heterosexual partners, and mostly involved wild-type infections.

Clustering of Transmissions Harbouring Drug Resistance

The prevalence of single drug class and multidrug resistance (MDR) in our PHI cohort was 14.9% (n=126/848) and 2.1% (n=18/848), respectively. No significant trend in the prevalence of transmitted resistance was obvious (Table 1). The overall incidence of PHI harbouring mutations conferring resistance to NRTIs (n=58), NNRTIs (n=65), and PIs (n=42) was 6.8%, 7.6%, and 5.0%, respectively.

The overall frequency of drug-resistant viruses in non-clustered and clustered transmissions was similar, 14.3% and 16.5%, respectively. However, clustering had a significant impact on the relative distribution of resistance mutations for different drug classes. This is illustrated in the phylogenetic tree in Figure 1A, that includes the 144 cases of transmitted resistance in our treatment-naïve, PHI population.

Figure 1. Distribution of sequenced primary infections harbouring drug resistance.

A. The phylogenetic tree shows clustering patterns of sequenced PHI (n=144/848) harbouring resistance to thymidine analogue mutations (TAMs), non-nucleoside RT inhibitors (NNRTIs), as well as protease inhibitors (PIs). This includes seven NNRTI clusters, one TAM, and two PI clusters. Boxes list relevant NNRTI mutations. B. The mean frequency of PHI harbouring mutations to NRTIs, NNRTIs, PIs, T215 revertants, and multidrug resistant (MDR) viruses, as well as the G190A mutation.

As shown, viral variants harbouring mutations to NRTIs, including revertants at codon 215 (T215C/D/I/N/S), thymidine analogue mutations, and M184V, as well as to PIs were less frequent in clustered transmissions. In marked contrast, seven transmission clusters (Clusters A-G) harboured mutations, e.g. K103N, V108I, G190A, associated with resistance to NNRTIs (Fig. 1A). In addition, cluster C represents an MDR transmission network, wherein all four PHIs harboured K103N and three of the four also harboured L10I, I54V, A71V, V82A/I/T, I84I/V, and L90M. These PHIs, dating from Oct 2002 to May 2003, were resistant to the PIs available at that time.

Cluster A is noteworthy in that it represents the largest in our cohort, including 24 PHI and three chronic asymptomatic patients. Clustering was episodic; 6 patients were infected in 2004, 8 patients within a 3-month interval in 2005, and 8 within a 6-month interval in 2006. All 27 patients harboured the G190A and A98S mutations. Cell-based phenotypic assays on four isolates revealed >100-fold NVP resistance, sensitivity to EFV and ETV, and hypersensitivity to TMC-120. This resistance profile was confirmed for these and two other isolates based on Virco Antivirogram phenotypic analysis.

Effect of Clustering on Transmissibility of Resistance

The overall frequency of viral species harbouring resistance to different classes of drugs in unique and clustered transmissions is summarized in Fig. 1B. There was a marked diminution in the incidence of viruses harbouring NRTI mutations, 215 revertants, and MDR variants in clustered transmissions (X²= 15.4, 15.6, and 4.8, p< 0.0001, 0.0001, and 0.05, respectively).

In contrast, clustering led to increased frequencies of NNRTI resistance in large clusters, compared to unique and small cluster transmissions (X²= 4.8 and 11.4, p< 0.05 and 0.001, respectively). Such forward transmission of NNRTI resistance in clusters was independent of the G190A mutation (Fig, 1).

There appeared to be no significant impact of clustering on the incidence of viral variants harbouring mutations associated with resistance to PIs. Most PI transmitted resistance was restricted to infections harbouring single mutations, e.g. L90M, V82I, that have limited impact on drug susceptibility and viral replicative capacity. The transmission of complex PI mutational patterns, conferring measureable phenotypic resistance, was lower in clustered transmissions (X²= 5.2, p< 0.05) (Figure 1A).

Discussion

Our findings are consistent with other studies showing that early/PHI infections play a key role in the spread of HIV [6-8,16]. Our new study provides additional data that confirm that clustering of early/recent infections in Quebec may contribute to ∼50% of transmissions [8]. High rates of clustering in our cohort reflect a concentrated MSM epidemic in which persons unaware of their HIV status may engage in high risk behaviours [27]. Universal access to resistance testing allows for in depth population sampling. Our results are also consistent with models that suggest that early chronic infection contributes to HIV spread [28]. Universal access to antiretroviral drugs is an obvious factor in transmission of resistant quasispecies.

This study illustrates the potential benefits of phylogenetics in understanding factors that govern HIV transmission and the spread of drug resistance. The implications of PHI/early infection and viral fitness in transmission networks of drug resistance is underscored by the distribution of resistant drug classes in early and chronic infection [23,29-33]. The frequencies of mutations among Quebec patients failing HAART are NRTI (64%) > PI (42%) > NNRTI (38%) [31]. The reverse order is observed in transmitted resistance, i.e. NNRTI (11%/7%) > PI (5%/5%) > NRTI (2%/ 8%) in clustered and nonclustered transmissions, respectively.

Viruses harbouring NRTI and MDR resistance may be replicatively unfit, creating a bottleneck for forward transmission of such variants [29,30,32]. In contrast, clustering may facilitate spread of viruses harbouring NNRTI resistance mutations. This finding is highlighted by a transmission chain of 27 G190A-containing infections and a transmission network of NNRTI/PI dual resistant infections (n=4). In general, NNRTI mutations do not impact on viral replicative fitness to the same extent as either NRTI or PI mutations [29,32,33]. The elevated frequency of acquired NNRTI resistance reported here is consistent with data from other PHI cohorts [34,35]. In contrast, clustering did not impact on the frequency of transmitted drug-resistant infections in the Swiss cohort [36].

Our findings further underscore the recommendation that all newly infected persons undergo drug resistance testing [37,38]. Resistance to antiretroviral drugs is present 10 to 25% of the time in PHI [8,34-36]. Transmitted drug resistance persists over time, can affect disease course in drug-naïve patients, and limit strategies for antiretroviral therapy [39-44]. Although MDR variants may be less replicatively fit than wild-type archival species, the generation of resistant and MDR strains that may circumvent replicative transmission barriers is of growing concern [39]. Observations of superinfection and coinfections with resistant and MDR viruses show that resistant viruses may become more virulent and transmissible over time [33,39]. Public health strategies that target early/PHI stage infection, including the introduction of rapid testing and routine genotyping, may be of significant benefit toward reducing the incidence and spread of HIV, including that of drug resistant variants [6-,8, 39,44].