Emergence of Multiclass Drug–Resistance in HIV-2 in Antiretroviral-Treated Individuals in Senegal: Implications for HIV-2 Treatment in Resouce-Limited West Africa

Abstract

Background

The efficacy of various antiretroviral (ARV) therapy regimens for human immunodeficiency virus type 2 (HIV-2) infection remains unclear. HIV-2 is intrinsically resistant to the nonnucleoside reverse-transcriptase inhibitors and to enfuvirtide and may also be less susceptible than HIV-1 to some protease inhibitors (PIs). However, the mutations in HIV-2 that confer ARV resistance are not well characterized.

Methods

Twenty-three patients were studied as part of an ongoing prospective longitudinal cohort study of ARV therapy for HIV-2 infection in Senegal. Patients were treated with nucleoside reverse-transcriptase inhibitor (NRTI)– and PI (indinavir)–based regimens. HIV-2 pol genes from these patients were genotyped, and the mutations predictive of resistance in HIV-2 were assessed. Correlates of ARV resistance were analyzed.

Results

Multiclass drug–resistance mutations (NRTI and PI) were detected in strains in 30% of patients; 52% had evidence of resistance to at least 1 ARV class. The reverse-transcriptase mutations M184V and K65R, which confer high-level resistance to lamivudine and emtricitabine in HIV-2, were found in strains from 43% and 9% of patients, respectively. The Q151M mutation, which confers multinucleoside resistance in HIV-2, emerged in strains from 9% of patients. HIV-1–associated thymidine analogue mutations (M41L, D67N, K70R, L210W, and T215Y/F) were not observed, with the exception of K70R, which was present together with K65R and Q151M in a strain from 1 patient. Eight patients had HIV-2 with PI mutations associated with indinavir resistance, including K7R, I54M, V62A, I82F, L90M, L99F; 4 patients had strains with multiple PI resistance–associated mutations. The duration of ARV therapy was positively associated with the development of drug resistance (P = .02). Nine (82%) of 11 patients with HIV-2 with detectable ARV resistance had undetectable plasma HIV-2 RNA loads (<1.4 log₁₀ copies/mL), compared with 3 (25%) of 12 patients with HIV-2 with detectable ARV resistance (P = .009). Patients with ARV-resistant virus had higher plasma HIV-2 RNA loads, compared with those with non–ARV-resistant virus (median, 1.7 log₁₀ copies/mL [range, <1.4 to 2.6 log₁₀ copies/mL] vs. <1.4 log₁₀ copies/mL [range, <1.4 to 1.6 log₁₀ copies/mL]; P = .003).

Conclusions

HIV-2–infected individuals treated with ARV therapy in Senegal commonly have HIV-2 mutations consistent with multiclass drug resistance. Additional clinical studies are required to improve the efficacy of primary and salvage treatment regimens for treating HIV-2 infection.

HIV-2 infection is endemic in West Africa, but unlike HIV-1 infection, it has had limited spread worldwide [1]. Compared with HIV-1 infection, HIV-2 infection is characterized by a much longer asymptomatic stage, lower plasma viral loads, slower decrease in CD4 cell count, decreased mortality rate associated with AIDS, and lower rates of genital tract shedding, mother-to--child transmission, and sexual transmission [1–5]. Nonetheless, a significant proportion of HIV-2 infections eventually progress to AIDS, and HIV-2–infected individuals may benefit from antiretroviral (ARV) therapy [3, 6].

ARV therapy is becoming increasingly available in West Africa, where HIV-2 infects up to 1–2 million people [7]. As ARV therapy “scale-up” programs proliferate in West Africa, significant numbers of HIV-2–infected individuals will have access to and will be treated with ARV drugs developed against HIV-1 infection [8]. However, HIV-2 is intrinsically resistant to the nonnucleoside reverse-transcriptase inhibitors and to enfuvirtide, and reports suggest that HIV-2 may be partially resistant to some protease inhibitors (PIs) and has a low genetic barrier to nucleoside reverse-transcriptase inhibitor (NRTI) resistance [9–12].

To date, there have been no randomized clinical trials of ARV therapy for HIV-2 infection [13]. However, several small observational cohort studies in developed countries have shown poor outcomes of ARV therapy for HIV-2 infection [14–16]; similar poor results were reported in 3 small studies from resource-limited settings in Senegal, The Gambia, and Cote d'Ivoire, West Africa [17–19].

To assess the emergence of reverse-transcriptase (RT) and protease (PR) resistance mutations that occur in virus strains during ARV therapy among a cohort of HIV-2–infected individuals in Senegal, West Africa, we sequenced pol genes from plasma samples and PBMCs obtained from a cohort of ARV-treated persons participating in the Senegalese Initiative for Access to Antiretrovirals (ISAARV) program.

Patients, Materials, and Methods

Patients were studied as part of an ongoing prospective longitudinal cohort study of ARV therapy for HIV-2 infection in Senegal, West Africa; enrollment began in October 2005. HIV-2–infected individuals with clinical AIDS, with CD4 cell counts <200 cells/μL, or with CD4 cell counts <350 cells/μL and clinical symptoms were treated with ARVs as part of the ISAARV program at the University of Dakar, Fann Hospital Infectious Diseases Clinic (Dakar, Senegal). HIV-2–infected patients initiating ARV therapy (and those already receiving ARV therapy) in the ISAARV program were referred to participate in this study. This study was conducted according to procedures approved by the Institutional Review Boards of the Universities of Washington and Dakar and the Senegalese National AIDS Committee. All patients provided informed consent for study participation.

Participants were screened for HIV-1 and HIV-2 infection by serologic testing. Serum samples were tested using a microwell plate HIV-1/HIV-2 EIA (Genetic Systems); serum sample positivity was confirmed using a rapid synthetic peptide-based membrane immunoassay (Multispot; Sanofi Pasteur), which classified patients as seropositive for HIV-1 or HIV-2 or as dually seropositive. At enrollment and subsequent follow-up visits (at 1 month and, then, every 4 months), patients underwent a physical examination and completed an interview with questions about demographic characteristics and sexual and other behaviors. A routine medical history was recorded, and examination was performed; the information was recorded on a standardized form by clinicians, as described elsewhere [5]. Peripheral blood samples were collected in tubes containing EDTA and were analyzed using the FACSCount analyzer (Becton Dickinson Biosciences) to determine the CD4, CD8, and CD3 cell counts (as cells/μL of blood); the samples were also used for the quantitative HIV-2 RNA load assays and pol genotyping. The quantitative real-time RT-PCR assay that was used for testing plasma HIV-2 RNA levels, HIV-2 pol genotyping, and phylogenetic analysis is described in the Appendix (online only).

Statistical analysis

HIV-2 load analyses were performed on log-transformed values with use of the Wilcoxon rank-sum test with a limit of detection of 25 copies/mL (1.4 log₁₀ copies/mL) and with an undetectable level set to 0. A sensitivity analysis of using a limit of detection set to 10 or 50 copies/mL did not change the result that viral loads were significantly higher in patients with virus strains with evidence of ARV resistance than in patients with virus strains without evidence of ARV resistance. JMP software, version 5.1.2 (SAS Institute), was used for statistical calculations.

Results

The baseline demographic, clinical, and virological characteristics of the 23 patients reported in this study are shown in table 1. The majority of patients were treated with zidovudine (AZT), lamivudine (3TC), and indinavir, which is the first-line regimen for HIV-2 infection in Senegal. Eleven patients were receiving ARV therapy at study entry (median duration of therapy, 474 days; range, 111–1534 days); 12 were ARV therapy naive at enrollment and began ARV therapy during the follow-up period. Genotyping of HIV-2 pol RT and PR was performed at study entry and then prospectively every 4 months, irrespective of plasma HIV-2 load, because emergence of drug resistance in HIV-2 may occur at low viral loads and could potentially affect viral fitness. Eleven (48%) of the 23 patients had virus strains with no evidence of drug-resistance mutations in PR or RT at study entry or during the follow-up period. Eleven patients were receiving ARV therapy at study entry; 7 had virus strains with evidence of drug resistance at study entry, 1 had a virus strain that developed drug resistance during the follow-up period, and 3 had virus strains that did not develop drug resistance during the follow-up period. Twelve patients were ARV therapy naive at study entry and began ARV therapy during the follow-up period; none of these patients had drug-resistant virus at study entry, and 4 had virus strains that developed drug resistance during the follow-up period.

Table 1.

Demographic, clinical, and virological characteristics of HIV-2–infected patients receiving antiretroviral (ARV) therapy at study entry.

Characteristic	Patients (n = 23)
Sex
Male	11 (48)
Female	12 (52)
Age, median years (range)	49 (31–60)
WHO stage
1	2 (9)
2	0 (0)
3	13 (57)
4	8 (34)
Prior ARV therapy at study entry	11 (48)
Duration of ARV therapy prior to study entry, mean days (range)	474 (111–1534)
Plasma HIV-2 RNA loada
All patients,
Median log10 copies/mL (range)	2.0 (<1.4 to 4.3)
Undetectable	7/23 (30)
ARV therapy–experienced patients
Median log10 copies/mL (range)	1.9 (<1.4 to 2.9)
Undetectable	4/11 (36)
ARV therapy–naive patients
Median log10 copies/mL (range)	2.3 (<1.4 to 4.3)
Undetectable	3/12 (25)
CD4 cell count, median cells/mm3 (range)	200 (12–562)
Initial ARV regimen
AZT, 3TC, and IDV	19 (83)
AZT, DDI, and IDV	1 (4)
DDI, 3TC, and IDV	1 (4)
D4T, 3TC, and IDV	1 (4)
AZT, 3TC, and NVPb	1 (4)

Note. AZT, zidovudine; DDI, didanosine; D4T, stavudine; IDV, indinavir; NVP, nevirapine, 3TC, lamivudine; WHO, World Health Organization.

^aUndetectable was defined as an HIV-2 RNA load <25 copies/mL (<1.4 log₁₀ copies/mL).

^bThe patient initially received a misdiagnosis of HIV-1 infection and was treated with AZT, 3TC, and NVP prior to study entry.

RT mutations associated with NRTI resistance in HIV-2

Mutations K65R, Q151M, and M184V in RT have previously been shown to be associated with phenotypic NRTI resistance in HIV-2 [20]. The RT mutation M184V, which confers high-level resistance to 3TC and emtricitabine (FTC) in HIV-2, was found in 10 (43%) of 23 patients (table 2). Five patients had virus strains with the M184V mutation at the initial genotyping at study entry, and 5 had virus strains that developed the mutation during the follow-up period. The HIV-2 RT Q151M mutation, which confers multinucleoside resistance (to AZT, didanosine, stavudine, and FTC) was found in virus strains from 2 (9%) of 23 patients (both mutations were present at study entry). The RT K65R mutation, which confers resistance to 3TC, FTC, and didanosine, was also found in strains from 2 patients (1 patient also had virus with the Q151M mutation, and both mutations were present at study entry; the other patients had virus with a mixed K65K/R population emerge during the follow-up period).

Table 2.

HIV-2 reverse-transcriptase (RT) and protease (PR) resistance mutations observed in the cohort.

				ARV resistance			RT mutation										PR mutation
Patient	ARV therapy regimen	Duration of ARV therapy, days	Plasma HIV-2 RNA load, copies/ mL	NRTI only	PI only	NRTI and PI	M41L	A62V	K65R	69-IC	K70R	Q151M	M184V	L210W	T215YF	K219QE	K7R	D17N	K45R	M46I	V47A	I54M	V62A	164 V	I82F/L	I84V	L90M	L99F
H2A17	AZT 3TC, and IDV	111a	181	…	+	…	M	A	K	N	K	Q	M	N	S	E	R	D	K	I	V	I	V	I	I	I	L	F
H2A01	AZT 3TC, and IDV	128	UD	+	…	…	M	A	K	N	K	Q	M/V/I	N	S	E	K	D/G	K	I	V	I	V	I	I	I	L	L
H2A14	AZT 3TC, and IDV	150	118	+	…	…	M	A	K	N	K	Q	M/V/I	N	S	E	K	G	K	I	V	I	V	I	I	I	L	L
H2A10	AZT 3TC, and IDV	206	UD	…	…	+	M	A	K/R	N	K	Q	M/V	N	S	D	K	G	K	I	V	I	V	I	I	I	L/M	L
H2A21	AZT 3TC, and IDV	256	42	+	…	…	M	A	K	N	K	Q	M/V	N	S	E	K	D/G	K	I	V	I	V	I	I	I	L	L
H2A04	AZT DDI, and IDV	474a	74	…	…	+	M	A	K	N	K	Q	V	N	S	E	K	G	K	V	V	I	V	I	I	I	L	F
H2A20	DDI, 3TC, and IDV	492a	52	…	…	+	M	A	K	N	K	Q	V	N	S	E	N	G	K	I	V	M	V	I	I	I	L	L
H2A15	AZT, 3TC, and IDV	521a	72	…	…	+	M	A	K	N	K	Q	V	N	S	E	K/R/G	G	K	I	V	M	V	I	l/F	I	L	L
H2A19	AZT, 3TC, and IDV	555a	UD	…	…	+	M	A	K	N	K	Q	M/V	N	S	E/G	K/N	G	K	I	V	I	V	I	l/F	I	L	L
H2A03	AZT, 3TC, and IDV	724	40	…	…	+	M	A	K	N	K	Q	M/V	N	S	E	K	G	K	I	V	I	V	I	l/F	I	L	L/F
H2A02	AZT->D4T 3TC, and IDV	910b	53	…	…	+	M	A	K	N	K	M	V	N	S	E	K	G	K	I	V	I	A	I	I	I	L	F
H2A24	AZT 3TC, and NVPb	1534a	427	+	…	…	M	A	R	N	R	M/I	M	N	S	E	K	D/G	K	I	V	I	V/l	I	I	I	L	L
Total (%)	…	…	…	4 (33)	1 (8)	7 (58)	…	…	2 (17)	…	1 (8)	2 (17)	10(83)	…	…	…	2 (17)	…	…	…	…	2 (17)	1 (8)	…	3 (25)	…	1 (8)	4 (33)

Note. Codon sites of HIV-2 RT or PR resistance are shown in boldface type. Patient H2A17 had an HIV-2 group B virus strain (7R and 99F observed in ARV therapy-naive patients). Genotyping was performed on isolates from plasma samples and PBMCs from all patients except H2A17 (plasma sample only) and H2A03 (PBMCs only). ARV, antiretroviral; AZT, zidovudine; DDI, didanosine; D4T, stavudine; IDV, indinavir; NRTI, nucleoside reverse-transcriptase inhibitor; NVP nevirapine; PI, protease inhibitor; 3TC, lamivudine; UD, undetectable (<25 copies/mL; <1.4 log₁₀ copies/mL).

^aPatients who had virus strains with evidence of ARV resistance at initial genotyping.

^bBefore study entry, patient H2A24 received a misdiagnosis of HIV-1 infection and was treated with AZT, 3TC, and NVP.

RT mutations associated with NRTI resistance in HIV-1 but with unclear effect in HIV-2

The M184I mutation (together with M184V) emerged in a virus strain from 1 patient. In HIV-1, the M184I mutation produces high-level in vitro resistance to 3TC and FTC and low-level resistance to didanosine and abacavir [21], has been reported in virus strains in both ARV therapy–naive and ARV-treated HIV-2–infected patients, and appears to be under selective pressure to 3TC in HIV-2 in vitro [22–24]. HIV-1–associated thymidine analogue mutations (M41L, D67N, K70R, L210W, T215Y/F, and K219Q/E) were not found, with the exception of K70R, which was present in a strain from 1 patient (in conjunction with K65R, Q151M, and 219E, the latter of which is commonly present in wild-type HIV-2 in ARV therapy naive patients).

HIV-2 RT polymorphisms of uncertain significance

Polymorphisms of unclear significance for drug resistance in HIV-2 were also found in HIV-2 RT at known resistance codon positions: Q151I (in a strain from 1 patient) and Q151R (in a strain from 1 patient). van der Ende et al. [25] reported the appearance of codon change Q151I (together with M41I and M184V) in an HIV-2 strain from a patient treated with AZT and 3TC who had virus with evidence of phenotypic resistance. The V111I polymorphism, which is commonly found in RT in strains in ARV therapy–naive patients but has been reported [26] to be associated with the Q151M resistance mutation, was found in virus strains from both patients with Q151M mutations (patients H2A2 and H2A24), in a strain from 1 patient with only an M184V mutation (patient H2A15), and in a strain from 1 patient with no RT mutations (wild-type; patient H2A5). The canonical HIV-1 nonnucleoside reverse-transcriptase inhibitor mutations Y181I and Y188L were found in virus strains from all HIV-2– infected patients. Numerous polymorphic sites of unclear significance for HIV-2 NTRI resistance were observed in our cohort; none of these sites appeared to emerge in strains from ≥1 patient, which suggests that the strains were unlikely to be under selective pressure of NRTIs (data not shown; supplemental figure 1 available at http://ubik.mullins.washington.edu/publications/gottlieb _2009_CID).

PR mutations associated with ARV resistance in HIV-2

The contribution of individual HIV-2 PR mutations that confer phenotypic resistance to PIs have not been delineated; however, the emergence of mutations (K7R, G17N, K45R, M46I, V47A, I54M, V62A, I64V, V71I, I82F/L, I84V, L90M, and L99F) in PR in HIV-2 from individuals receiving PI-based therapy have been observed in previous studies. Because patients in our cohort were exclusively treated with indinavir during the study, we focused on HIV-2 PR mutations K7R, I54M, V62A, I82F, L90M, and L99F, which have been associated with phenotypic PI resistance to indinavir in previous, albeit limited, studies. Eight of 23 patients had virus strains with PI mutations (K7R, I54M, V62A, I82F, L90M, and L99F) suggestive of indinavir resistance; 4 of these patients had virus strains that harbored multiple PI mutations (K7R-L99F, K7K/R/G-I54M-I82I/F, I82I/F-L99L/F, and V62A-L99F) (table 2). Three patients had strains with the 82I→F mutation emerge during the study period; 7K→R, 90L→M, and 99L→F also emerged in virus strains in individual patients during the study period.

PR mutations associated with ARV resistance in HIV-1 but with uncertain effect in HIV-2

Naturally occurring HIV-2 PR polymorphisms that are associated with PI resistance in HIV-1 were common; major (V32I/L, M46I/V, and I47V) and minor (L10V/I, E35G/R, Q58E, A71V/I, and G73A/T) PI resistance mutations were found in all 23 patients. These combinations of mutations in HIV-1 would predict intermediate resistance to all PIs (atazanavir, darunavir, fosamprenavir, indinavir, lopinavir, nelfinavir, saquinavir, and tipranavir) with use of the Stanford HIV Drug Resistance Database algorithm [27]. Phenotypic data on HIV-2 ROD strains [12] suggest that these natural polymorphisms confer resistance of 15-fold, compared with HIV-1 strains, to at least amprenavir, atazanavir, and tipranavir.

HIV-2 PR polymorphisms of uncertain significance

HIV-2 polymorphisms not known to be associated with ARV resistance in HIV-2 or HIV-1 have been commonly reported in virus strains from both ARV therapy–naive and –experienced patients. Numerous polymorphic sites of unclear significance for PI resistance in HIV-2 were observed in our cohort; none of these sites appeared to emerge in virus strains from ≥1 patient, which suggests that the viruses were unlikely to be under selective pressure of indinavir (data not shown; supplemental figure 2 available at http://ubik.mullins.washington.edu/publications/gottlieb_2009_CID).

Multiclass drug resistance (to an NRTI and a PI) was detected in virus strains from 7 (30%) of 23 patients and in virus strains from 7 (58%) of 12 patients with drug-resistant virus (table 2). Of the 7 patients with multiclass drug–resistant virus, 3 had virus strains in which both NRTI and PI mutations emerged during the follow-up period, and 1 patient had a virus strains with RT M184V and PR I54M mutations at study entry that developed K7K/R/G and I82I/F mutations during the follow-up period. Six of 7 patients had virus strains with evidence of linked drug-resistance mutations in RT and PR on the same viral clone, which suggests that these virions were multiclass drug resistant.

We sought to determine whether HIV-2 NRTI and/or PI resistance was associated with duration of therapy. In 11 (48%) of the 23 patients who had virus strains with no evidence of drug-resistance mutations, the median duration of ARV therapy exposure was 148 days (range, 28–491 days). Seven patients had virus strains with evidence of drug resistance at study entry (median duration of ARV therapy exposure, 521 days; range, 111–1534 days), and 5 had virus strains that developed drug resistance during the follow-up period (median duration of ARV therapy exposure, 206 days; range, 128–742 days). Detection of drug resistance was associated with longer duration of ARV therapy (median duration of ARV therapy exposure among patients with virus strains with no drug resistance vs. that among patients with virus strains with drug resistance, 148 days vs. 483 days; P = .02, by Wilcoxon rank-sum test). There appeared to be a trend toward multiclass drug resistance with increased duration until the first detection of drug resistance, although this was potentially confounded by patients with virus strains with detectable drug resistance at study entry and by 1 patient (patient H2A24) who had initially received a misdiagnosis of HIV-1 infection and, therefore, only received 2 active NRTIs (AZT and 3TC) and HIV-2–inactive nevirapine (table 2).

Plasma HIV-2 RNA loads were undetectable (<25 copies/ mL; 1.4 log₁₀ copies/mL) in 9 (82%) of 11 patients with virus strains without detectable ARV resistance, compared with 3 (25%) of 12 patients virus strains with detectable ARV resistance (P < .009, by Fisher's exact test). In addition, patients infected with ARV-resistant virus had higher plasma HIV-2 RNA loads (median, 1.7 log₁₀ copies/mL; range, <1.4 to 2.6 log₁₀ copies/mL), compared with those with virus strains without evidence of ARV resistance (median, <1.4 copies/mL; range, <1.4 to 1.6 log₁₀ copies/mL; P = .003, by Wilcoxon rank-sum test) (figure 1).

Figure 1.

Plasma HIV-2 RNA loads in patients with and without antiretroviral (ARV) resistance. Viral load data were log₁₀ transformed. Raw data (circles) and box plots are shown. Nine (82%) of 11 patients with virus strains without detectable ARV resistance and 3 (25%) of 12 patients with virus strains with detectable ARV resistance had HIV-2 RNA loads that were undetectable (UD; <1.4 log₁₀ copies/mL; set to 0; P = .003, by Wilcoxon rank-sum test).

Phylogenetic analysis of pol sequences revealed that 22 of 23 patients harbored HIV-2 group A strains, and 1 patient was infected with a HIV-2 group B strain (patient H2A17) (figure 2). Two patients who were married (patients H2A14 and H2A25) were infected with closely linked strains, although only 1 partner was infected with virus with evidence of ARV resistance (M184M/V mutation) (figure 2).

Figure 2.

Neighbor-joining phylogenetic tree of amino acid sequences of the HIV-2 pol gene. Clonal sequences from virus strains from 23 patients are shown. Reverse-transcriptase (RT) and protease (PR) resistance mutations in HIV-2 are shown. Twenty-two of 23 patients were infected with HIV-2 group A strains, and 1 patient was infected with an HIV-2 group B strain. Two patients infected with closely linked viruses (patients H2A14 and H2A25 [gray]) were married. HIV-2 reference sequence HIV-2 ROD (A.SN.85.ROD accession number M15390) is shown. WT, wild type (no RT or PR mutations).

Discussion

This study reports on the emergence of RT and PR resistance mutations that occurred during ARV therapy in a cohort of HIV-2–infected individuals in Senegal, West Africa, who participated in the ISAARV program. It is the largest study to date on HIV-2 ARV resistance from West Africa, where HIV-2 infection is endemic. In the only other study of NRTI- and PI-based ARV therapy for HIV-2 infection in Africa (in Cote d'Ivoire), Adje-Toure et al. [17] found no evidence of RT or PR mutations in virus strains from 4 patients who received indinavir-based therapy during a 6–11-month study period. However, the authors did find high treatment failure rates, RT mutations M184V/I and Q151M, and the PR mutation L90M in the virus strains from 7 patients who received NRTI- and nelfinavir-based regimens [17].

We found a remarkably high level of genotypic mutations consistent with RT and/or PR resistance (52% overall), with 30% of patients having virus strains with mutations that confer multiclass drug resistance to both NRTIs and PIs. The high prevalence (43% of patients; 83% of those with strains with ARV resistance) of the RT mutation M184V, which confers high-level resistance to 3TC and FTC, likely reflects the near universal use of 3TC as a first-line ARV for HIV-2 infection in Senegal. In contrast to what has been reported for HIV-1 infection treated with regimens containing AZT, thymidine analog mutations were rare in HIV-2, with 1 patient infected with a strain harboring the K70R mutation (a mutation of unknown significance in HIV-2). However, the RT Q151M mutation, which confers multinucleoside resistance in HIV-2, occurred relatively frequently (in 17% of patients infected with ARV- resistant virus), an observation that has been noted in other cohorts [17, 28]. The K65R mutation was also observed in virus strains from 17% of patients infected with NRTI-resistant virus and has been associated with resistance to 3TC and FTC. A recent report suggested that K65R may be associated with abacavir resistance in HIV-2, but the patients in that report with virus with the K65R mutation also received AZT and 3TC [29].

The HIV-2 PR contains natural polymorphisms that are commonly associated with PI resistance in HIV-1 [30, 31]. Wild-type HIV-2 appears to be naturally less susceptible to some PIs—most remarkably, amprenavir in PBMC phenotyping assays [9, 12, 32] and in vitro with purified HIV-2 PR [33]. Susceptibility data for other PIs have been less consistent. Desbois et al. [12] reported decreased susceptibility (≥5-fold) for amprenavir, atazanavir, indinavir, nelfinavir, and tipranavir in HIV-2, compared with HIV-1. Rodes et al. [32] reported reduced susceptibility to nelfinavir (6-fold) and amprenavir (31fold) in 1 HIV-2 isolate, compared with HIV-1. Witvrouw et al. [9] reported HIV-2 group A (ROD) strains with reduced susceptibility to amprenavir and nelfinavir but HIV-2 group B (EHO) strains with reduced susceptibility to amprenavir but hypersusceptibility to saquinavir, compared with HIV-1 (III_B). Brower et al. [33] reported that lopinavir, saquinavir, tipranavir, and darunavir exhibit the highest potency based on their inhibition constant (K_i) values measured using purified HIV-2 (ROD) PR and a polypetide substrate. Adje-Toure et al. [17] reported that HIV-2 in a cohort in Cote d'Ivoire appeared to be clinically resistant to nelfinavir. Indinavir is currently the first-line PI for treatment of HIV infection in Senegal and was the only PI treatment used in our study. Phenotypic resistance to indinavir has been reported with the K7R, I54M, I82L/F, V62A-L99F, and L90M mutations [34]; our study provides an in vivo correlation for this observation. One patient (patient H2A17) in our study was infected with a group B HIV-2 strain; the majority of HIV-2 group B sequences (in ARV therapy– naive patients) have K7R and L99F mutations present naturally, although at least 1 report [12] describes 2 ARV therapy–naive patients with HIV-2 group B strains with the 99 L “mutation.” Colson et al. [35] reported that K7R and L99F mutations occur more commonly during PI treatment, and Ntemgwa et al. [34] reported selection for L99F by indinavir and nelfinavir in HIV-2 group A strains in culture. M'Barek et al. [36] reported that introducing L99F to HIV-2 ROD (group A) strains had no effect on lopinavir or saquinavir resistance in a yeast system. No data on K7R or L99F phenotypic resistance are available for HIV-2 group B strains.

Detection of HIV-2 resistance in our study was associated with longer duration of ARV therapy. Vergne et al. [37] reported that ARV resistance in an HIV-1 Senegalese cohort emerged at a median interval of 17.8 and 18.3 months after study entry in ARV therapy–experienced patients and ARV therapy–naive patients, respectively. This duration of therapy does not appear to be substantially different than the 483 days (16.1 months) that we observed overall for our cohort but may be longer than the duration for the 5 patients who were infected with ARV-susceptible HIV-2 at study entry that developed ARV-resistance mutations during the follow-up period (median duration of therapy, 206 days [6.9 months]; range, 128–742 days).

Despite the frequent presence of multiple resistance mutations, plasma HIV-2 loads were relatively low (median, 1.7 log₁₀ copies/mL), albeit higher than the plasma HIV-2 loads in patients with virus strains without ARV resistance (median, <1.4 log₁₀ copies/mL [undetectable]). This finding is consistent with the relatively low viral loads that are found in patients with untreated HIV-2 infection, even when the patients have low CD4 cell counts, but also may reflect additional replication capacity and/or fitness impairment caused by resistance mutations [2, 5, 7, 38, 39]. These findings are in contrast to what has been observed in HIV-1, in which emergence of ARV resistance is typically associated with a relatively rapid increase in plasma HIV-1 load. In addition, that ARV resistance mutations in HIV-2 are common at low viral loads suggests that monitoring viral load as a harbinger of ARV resistance and virological failure may be of limited value for HIV-2–infected individuals receiving ARV therapy. Also concerning is that initial reports suggest that immune reconstitution and CD4 cell recovery may be poor in treated individuals with HIV-2 infection [16]. Consequently, more-reliable methods for assessing the efficacy of treatment of HIV-2 infection are needed.

In summary, HIV-2 from individuals treated with NRTI and indinavir–based ARV therapy regimens in Senegal commonly develop mutations in the RT and PR that are consistent with the development of multiclass drug resistance. Effective primary and salvage regimens for HIV-2 infection remain to be determined, and well-designed randomized, controlled trials that demonstrate which ARV therapy regimens are effective for treating HIV-2 infection are urgently needed to guide patients and the clinicians who care for them.