Genotypic drug resistance during HIV-1 primary infection in France (1996–1999): frequency and response to treatment

A key resistance mutation on the reverse transcriptase gene was observed in 8.8% of 204 patients enrolled during primary infection between 1996 and 1999 in ANRS 053 and the PRIMO cohort; the prevalence was stable over time. The prevalence of key protease gene mutations increased from 0% in 1996 to 8.3% in 1999. After the initiation of highly active antiretroviral therapy, HIV RNA in patients with one or more resistance to a prescribed drug decreased to under 20–50 copies/ml more slowly than in other patients. The widespread use of antiretroviral drugs has led to a marked reduction in mortality among HIV-1-infected patients. However, in patients with virological failure, resistant variants may be selected and transmitted. Primary infection by drug-resistant HIV-1 strains has already been documented [1–7]. The purpose of this study was to assess changes in the prevalence of primary drug-resistance mutations [reverse transcriptase (RT) and protease genes] between 1996 and 1999, by studying 205 patients with documented HIV-1 primary infection enrolled in two French multicentre studies. As most of the patients subsequently received highly active antiretroviral therapy (HAART), treatment response was also assessed. In 1996–1997, 61 patients were enrolled in the ANRS trial 053/053B. Patients had to present with symptoms within 30 days before inclusion, and with biological criteria of primary infection [8]. All patients were subsequently treated with zidovudine, lamivudine and ritonavir. Eight were also enrolled in the PRIMO Cohort. In the PRIMO cohort, 152 HIV-infected volunteers were enrolled between 1996 and 1999 with the biological criteria of HIV primary infection of less than 6 months [9]. The clinicians prescribed treatment before being informed of the genotype results. In both studies patients signed a written consent form. Genotypic analysis was performed on frozen plasma samples collected at enrolment before the initiation of HAART. The RT (codons 35–240) and protease genes (codons 1–100) were amplified by nested reverse transcription and sequenced in both directions, using the Big Dye Dideoxyterminator Cycle Sequencing kit (Applied Biosystems, France). Sequence alignment was performed using Clustal Navigator software. The amplification of both genes failed in one sample (plasma viral load below 50 copies/ml). Amplification of the RT gene failed on another sample (viral load of 3.5 log10). The French 2001 guidelines for antiretroviral drug resistance testing, revised in June 2001, were used to identify major or key resistance-conferring mutations [10]. The patients’ median age at inclusion was 32.0 years; 20% were women. The risk factor for infection was homosexual intercourse in 60% of cases, heterosexual intercourse in 34%, injecting drug use in 3% and unknown in 3%. The median plasma viral load at enrolment was 5.05 log10 copies/ml and the median CD4 cell count was 555/μl. Overall, 8.8% of the 204 samples analysed had at least one key resistance-conferring mutation, consisting of 16 on the RT gene and eight on the protease gene (Table 1). No significant change in the prevalence of primary RT mutations was noted between 1996 and 1999. However, mutations 215Y/F and 184V/I were more often associated with other key mutations when patients were infected after the first half of 1997. Key mutations conferring resistance to a protease inhibitor were first observed in September 1997; the prevalence reached 8.3% in 1999 (P trend = 0.06). In all cases but one (M46I mutation), key mutations on the protease gene were associated with key mutations on the RT gene. No significant difference in the prevalence of resistance mutations was found according to sex (8.0% in men, 12.5% in women) or HIV exposure group. The prevalence was higher in patients originating from European countries (9.7%) than from other countries (3.4%;P < 10−6).Table 1: Prevalence of the key resistance mutations on the reverse transcriptase and protease genes in 204 patients with primary HIV-1 infection (the French ANRS 053 trial and the PRIMO cohort). Nine of these patients had more than one primary resistance mutation and six had three or more. The prevalence of patients bearing more than one key mutation was higher in 1999 (10.4%) than before (2.6%;P < 0.04). The plasma viral load at enrolment was lower in patients with two or more key mutations (median 4.1 log) than in patients with no key mutations (5.1 log) or only one key mutation (5.4 log) (Kruskall–Wallis test P < 0.04). Of the 18 patients who had primary resistance mutations, 16 began treatment during primary infection and 12 had genotypic resistance to at least one drug in the regimen they were prescribed. Nine (75%) of these 12 patients reached the threshold viral load value of 20–50 copies/ml while receiving HAART, compared with 89.1% of the remainder (logrank test P = 0.06). Interestingly, all six patients who were prescribed at least two theoretically inactive drugs reached the viral load threshold of 20–50 copies/ml. This was probably the result of a lower plasma viral load at HAART initiation in these patients (median 4.2 log) than in patients with no resistance to the prescribed regimen (5.04 log) or only one (6.0 log). Indeed, after adjustment for viral load at HAART initiation in a Cox model, both patients with one and patients with more than one mutation conferring resistance to the prescribed regimen reached the viral load threshold more slowly than other patients. Similar data have been obtained in other studies, together with virological breakthrough in such patients [3,5]. In-vitro studies have shown a decreased replication capacity of strains bearing resistance mutations [11]. A longer follow-up is required to show whether a slower virological response caused by primary genotypic resistance has deleterious consequences regarding the emergence of multiple resistance and virological rebound. The reported prevalence of antiretroviral resistance mutations is generally between 5 and 10%, although some studies have yielded lower [12] or higher [5] frequencies. Discrepancies among studies can result from different prevalences in the source population, and more generally from different therapeutic strategies across countries. Different intervals between infection and sampling may also be involved, as some mutations, particularly to lamivudine and non-nucleoside reverse transcriptase inhibitors, can revert. Since 1995 the prevalence of mutation 215Y/F appears to be stable in industrialized countries, probably as a result of the frequent combination of zidovudine with other highly active drugs. As the prevalence of patients bearing more than one key mutation increased in our study in 1999, prevalence monitoring should continue. Martine Harzica Isabelle Pellegrinb Christiane Deveauc Marie-Laure Chaixe Béatrice Dubeauxa Isabelle Garrigueb Nicole Ngoe Christine Rouziouxe Cécile Goujardd Bruno Hoenf Daniel Serenig Jean-François Delfraissyd Laurence Meyerc and the PRIMO Study Group* Acknowledgements The authors acknowledge the patients and the physicians who participated in the studies, C. Capon, I. Iraqui and D. Laskri (Bicêtre Hospital) for monitoring the PRIMO cohort, M. Sinet and A. Venet for their useful comments, and D. Young for reviewing the manuscript.