Prevalence of genotypic and phenotypic resistance to anti-retroviral drugs in a cohort of therapy-naive HIV-1 infected US military personnel

Objective: While transmission of drug-resistant HIV-1 has been reported, estimates of prevalence of resistance in drug-naïve populations are incomplete. We investigated the prevalence of genotypic mutations and phenotypic antiretroviral resistance in a cohort of HIV-1 infected U.S. military personnel prior to the institution of antiretroviral therapy.

Cited Here...: Cross-sectional cohort study.

Cited Here...: Plasma was obtained from 114 recently HIV-1 infected subjects enrolled in an epidemiological study. Genotypic resistance was determined by consensus sequencing of a PCR product from the HIV-1 pol gene. Sequences were interpreted by a phenotypic-genotypic correlative database. Resistance phenotypes were determined by a recombinant virus cell culture assay.

Cited Here...: Genotypic mutations and phenotypic resistance were found at a higher than expected frequency. Resistance to non-nucleoside reverse transcriptase inhibitors was most common, with a prevalence of 15% of 95 subjects by genotype and 26% of 91 subjects by phenotype. Genotypic and phenotypic resistance respectively were found in 4% and 8% of subjects for nucleoside reverse transcriptase inhibitors and in 10% and 1% for protease inhibitors. One subject harbored virus with resistance to all three drug classes.

Conclusions: A substantial frequency of resistance to antiretroviral drugs was identified in a therapy-naïve U.S. cohort. In most cases, the genotypic and phenotypic assays yielded similar results, although the genotypic assay could detect some protease inhibitor resistance-associated mutations in the absence of phenotypic resistance. These data suggest the need for optimization of treatment guidelines based on current estimates of the prevalence of drug resistance in HIV-1 seroconverters.

Introduction

Successful anti-HIV-1 chemotherapy depends on the ability to effectively suppress viral replication ^[1-3]. Current highly active antiretroviral therapy regimens contain drugs with more than one mechanism of action, and non-overlapping patterns of drug resistance. If viral replication is not completely suppressed, the high mutation rate of HIV-1 combined with selection pressure of drugs can lead to mutations in either the viral protease or reverse transcriptase genes that confer drug resistance ^[2,4]. Complete suppression of HIV-1 could be compromised if a therapy-naïve patient already harbors virus with mutations conferring resistance to antiretroviral drugs used for initial therapy. Incomplete suppression of viral replication with an initial drug regimen would diminish the clinical benefit to the patient and could promote the development of broader drug resistance that may cause subsequent treatment regimens to be ineffective ^[5-8].

The widespread use of multi-drug regimens and the increasing prevalence of resistance among HIV-1 infected patients have increased the probability that HIV-1 resistant to one or more classes of antiretroviral drugs will be transmitted ^[9-12]. A substantial prevalence of resistance in untreated patients would necessitate new strategies to optimize initial therapy. Treatment guidelines, based on ongoing estimates of resistance prevalence and pre-therapy resistance testing are the standard for other infectious diseases such as tuberculosis and gonorrhea ^[13,14]. However, the frequency of resistance in HIV-1 infected therapy naïve patients has not been well characterized. We report the finding of a substantial prevalence of drug resistance in a geographically dispersed, ethnically diverse cohort of HIV-1 infected subjects. The results reported here are similar to other recently published reports ^[15-17], and highlight the potentially pervasive nature of antiretroviral resistance in recently infected patients.

Methods

Study design

Study subjects were active duty U.S. military personnel with documented HIV-1 seroconversion within the preceding 3 years that were referred to the HIV clinics of medical centers that participate in the U.S. Military HIV Research Program. Eligible subjects were counseled by study staff, and those who agreed to participate were administered an informed consent document according to the guidelines of the Tri-service Human Subjects Research Review Board. Between February 1997 and October 1998, 265 HIV-1 seroconverters agreed to participate, of which 114 were drug naïve and included in this study. Estimated date of infection was defined as the midpoint between the last negative and first positive HIV-1 enzyme immunoassay. The volunteers answered an extensive questionnaire including information about sexual behavior, including geographic location of sexual contacts during the seroconversion window. An extensive medical evaluation including a detailed drug history was also obtained, and the volunteers donated blood for the performance of resistance assays. Extreme care was taken to protect the confidentiality of sensitive risk factor information, and multiple levels of review were used to insure the accuracy of the drug naïve designation of this cohort. This study represents an extension of a previously published cohort pilot study, which presented epidemiologic risk factor data, mutations associated with drug resistance and viral subtypes for the first 30 patients enrolled at Naval Medical Center, San Diego ^[18]. The previous study did not include other sites, and did not include resistance phenotypes or Vircogen interpretation of the genotypes.

Assessment of genotypic resistance

For genotypic analysis, plasma derived viral RNA was reverse transcribed into complementary DNA and amplified by PCR. Consensus sequencing was performed using an automated ABI sequencer. Results of the genotypic analysis were reported as amino acid changes at positions in the gag-protease-reverse transcriptase coding regions compared to a wild-type reference sequence. Phenotypic-genotypic correlative data analysis and mutational pattern analysis were performed using VircoGEN interpretation software that is able to recognize multiple patterns of primary and secondary resistance mutations, and interpret them as either `sensitive' (S), `resistance possible' (I), or `evidence of resistance' (R) for individual antiretroviral drugs. Major mutations considered as `evidence of resistance' are shown in Table 1. In circumstances where only secondary or minor mutations were detected, or resistance reversal mutations were present, the sample was scored as I ^[19].

Table 1

Measurement of phenotypic resistance

Phenotypic analysis was performed using the recombinant virus assay approach (Antivirogram) as described by Hertogs et al.^[20] and with modifications as described by Pauwels et al.^[21] Briefly, protease and reverse transcriptase (RT) coding sequences were amplified from plasma derived viral RNA with HIV-1 specific primers. After homologous recombination of amplicons into a protease-RT deleted proviral clone, the resulting recombinant viruses were harvested, titrated and used for in vitro susceptibility testing. The results of this analysis were expressed as fold resistance values, reflecting the fold increase in the 50% inhibitory concentration (IC₅₀; μM) of a particular drug when tested with patient derived recombinant virus isolates, relative to the IC₅₀ (μM) of the same drug obtained with a reference wild-type virus isolate (IIIB/LAI). A relative IC₅₀ increase of 5-10 fold was defined as intermediate resistance (I), and a relative IC₅₀ increase of > 10 fold was defined as high level resistance (R).

Results

Of the 114 subjects in this study, enrollment at different centers was 52 (46%) from Naval Medical Center, San Diego, California; 29 (25%) from the National Naval Medical Center and the Walter Reed Army Medical Center in Washington, District of Columbia; 17 (15%) from the Naval Medical Center, Portsmouth, Virginia; and 16 (14%) from Wilford Hall Air Force Medical Center, San Antonio, Texas.

Genotypic and phenotypic results were obtained for 95 (83%) and 91 (80%) subjects, respectively. Failure to obtain results on the remaining subjects was the result of the inability to generate sufficient quantities of PCR amplicons for the performance of the assays. Except for two cases, the successfully phenotyped group was a subset of the successfully genotyped group, resulting in the availability of resistance data on 97 subjects.

All but two of the 97 subjects were men, with a mean age of 30 years (range, 19-45 years). Forty-two (44%) of the subjects were Caucasian, 38 (40%) African-American, nine (10%) Hispanic, and eight (8%) had no identified ethnicity. Estimated duration of infection was < 12 months for 54 (56%), 12-24 months for 37 (38%), and > 24 months for six (6%). The prevalence of genotypic and phenotypic resistance to individual anti-retroviral drugs is shown in Fig. 1. Overall, 21 (22.1%) and 27 (29.7%) subjects had I or R virus for at least one antiretroviral drug by genotype and phenotype, respectively. For nucleoside reverse transcriptase inhibitors (NRTI), resistance (I or R) was seen in four (4%) cases by genotype and seven (8%) by phenotype. For non-NRTI (NNRTI), resistance was found in 14 (15%) and 24 (26%) of cases by genotype and phenotype, respectively. For protease inhibitors, genotypic resistance was observed in 10 (10%) cases, whereas phenotypic resistance was observed in one (1%) case. The relative percentages of cases with I and R determinations by genotype and phenotype for each drug class are shown in Table 2.

Fig. 1

Table 2

Table 3 lists the frequency of individual drug resistance mutations identified in this cohort. The two cases with mutations at RT codon 215 had mutation from threonine to aspartic acid (T215D) associated with partial reversion to wild-type of a transmitted drug-resistant mutant virus. Both observed mutations at RT codon 103 were mutations from lysine to arginine (K103R), which alone has not been associated with phenotypic resistance.

Table 3

Resistance to more than one class of antiretroviral drug was observed in this cohort. Three (3%) cases had resistance (I or R) to two classes of antiretroviral drugs by both genotype and phenotype. Two (2%) and one (1.1%) of cases had resistance to all three classes of antiretroviral drugs by genotype and phenotype, respectively. Figure 2 shows the genotypic and phenotypic characteristics of the case that exhibited three-drug class resistance by both assays.

Fig. 2

Discussion

We have determined the prevalence of genotypic mutations and phenotypic resistance to antiretroviral drugs in a racially diverse, geographically dispersed cohort of drug naïve HIV-1 infected persons in the United States of America.

Previous reports have estimated the prevalence of primary zidovudine resistance mutations to be between 3 and 13%^[22,23]. Mutations conferring resistance to nevirapine, lamivudine, and protease inhibitors have been reported in drug naïve patients ^{[10,11,24,25]} but estimates of prevalence, especially for phenotypic resistance have been lacking. Recently, three other groups have also reported estimates of prevalence in newly infected cohorts ^[15-17]. Given the widespread use and occasional failure of combination therapy, it is likely that primary drug resistance will become an important concern that could impact a clinician's choice of initial therapy for individual patients. A recent consensus statement suggested that drug resistance testing should be considered for use in the design of initial antiretroviral regimens if there is an increased prevalence of resistance in a particular population ^[7].

In this study, primary resistance to each of the classes of antiretroviral drugs was seen. The highest prevalence of resistance was to the NNRTI. This is likely to be due to the ability of several single mutations to confer high level resistance to these drugs, resulting in clinically significant resistance after a brief period of less than completely suppressive therapy ^[6]. A variety of such mutations accounted for NNRTI resistance, including those at RT codons 98, 106, 108, 179, 181, and 189. The K103N mutation, which confers resistance to all commercially available NNRTI, including efavirenz, was not seen. While the presence of significant mutations predicted phenotypic resistance to NNRTI in every case, several cases of phenotypic resistance in the absence of known resistance conferring mutations were seen. In these cases, the phenotypic resistance was intermediate (5-10 fold). In contrast, phenotypic resistance associated with the Y181C mutation was 112-fold. The presence of phenotypic resistance in the absence of known mutations conferring NNRTI resistance could be due to an incomplete understanding of the genotypic correlates of resistance, or could point to another mechanism of resistance to this class of drugs. Alternatively, the consensus sequencing methods may not be as sensitive for the detection of minority resistant quasi-species as the phenotypic assay. It is also possible that the phenotype assay is inherently more variable for NNRTI than for other drugs, and that the intermediate decreases in NNRTI susceptibility are due to this phenomenon rather than to the actual presence of resistance. Data to support these hypotheses are lacking.

Mutations associated with resistance to zidovudine and lamivudine resistance were seen in 4% of cases. This is within the range of previous reports ^[22,23]. Two samples containing mutations at RT codon 215 were seen. In both of these cases the observed mutations were T215D and not the typical T215Y mutation which is known to confer resistance to zidovudine. The T215D mutation has been shown to be a partial reversion to wild-type that occurs in the absence of drug pressure. The T215D mutation partially restores zidovudine sensitivity in a previously resistant isolate, ^[26] and has not been reported to be a naturally occurring polymorphism. As seen for NNRTI, the presence of genotypic mutations conferring resistance to NRTI predicted phenotypic resistance in all cases, but was less sensitive than the phenotype assay.

In contrast, the genotype predicted resistance to protease inhibitors more frequently than did the phenotypic assay. Several potentially resistance conferring mutations in the protease gene were observed in this cohort. These included primary resistance mutations (in this case at codons 82 and 90) and several mutations known to occur as spontaneous genetic polymorphisms (codons 10, 36, 46, 71, and 77) ^[27-29]. The development of resistance to protease inhibitors is a genetically complex process, involving the accumulation of several mutations, driven by selective drug pressure ^[4,5]. The lack of selective pressure inherent in a drug naïve cohort may have resulted in the loss of sufficient mutations to confer phenotypic resistance and may represent an underestimate of clinically significant resistance that could rapidly develop after the initiation of therapy. Further study of the longitudinal clinical significance of genetic mutations in the absence of phenotypic resistance in drug naïve patients is clearly warranted.

The results of this study indicate that primary resistance to antiretroviral drugs could pose a substantial problem in the initial clinical management of HIV-1 infected patients. This possibility should be considered when selecting initial therapy regimens in populations where use of antiretroviral drugs is widespread. Current HIV-1 treatment guidelines recommend potent combination drug therapy to achieve maximal suppression of viral replication ^[1]. They also emphasize that ongoing viral replication will probably lead to the development of drug resistance. Primary drug resistance may compromise the initial clinical response to empiric antiretroviral therapy and further promote the replication of drug resistant virus. These data suggest the need for optimization of HIV-1 treatment guidelines based on ongoing estimates of the prevalence of drug resistance in HIV-1 seroconverters. The role and clinical benefit of pre-therapy resistance testing for individual patient management will need to be determined.

Acknowledgements

The authors thank the following for their hard work, dedication, and assistance in completing this project: the RV-117 volunteers, B. Christofferson, A. DeSoto, D. Dixon, H. Flaks, S. Garrett, J. Gilcrest, M.J. Humphries, S. Ito, K. Jewett, S. Lasater, M. Louder, J. Vita, J. Wessely, the Virco-U.K. genotyping group, and the Virco-Belgium phenotyping group. We would also like to thank D. Birx, P. Stoffels, and R. Pauwels for their encouragement and support.