Virologic suppression in response to antiretroviral therapy despite extensive resistance within HIV-1 reverse transcriptase after the first virologic failure

Approximately 21 million people are living with HIV (PLHIV) and receiving antiretroviral therapy [1], which is responsible for a significant decrease in their morbidity and mortality as well as in the risk of transmission [2, 3]. However, the importance of this therapy to global health has been threatened by an increased prevalence of resistance to antiretrovirals (ARVs), which has increased from 11 to 29% since 2001 [4]. Of the individuals under ARV treatment, 20% will have to switch ARVs due to virologic failure [5, 6]. Thus, drug resistance is a real obstacle to viral suppression and disease control [1].

In Brazil, approximately 60% of PLHIV receive ARVs [7] through a program by the Brazilian Ministry of Health. From 2004 to 2017, the first-line therapy in the country was based on regimens involving efavirenz (EFV), a non-nucleoside reverse transcriptase inhibitor (NNRTI). In 2013, tenofovir (TDF) became the preferred nucleos(t)ide reverse transcriptase inhibitor (NRTI(t)), and it was given in a single dose combined with lamivudine (3TC) and EFV [8]. In 2017, dolutegravir (DTG), an integrase inhibitor that was administered with TDF/3TC, became the first-line therapy [9].

Genotypic and immunovirologic data from patients following the first virologic failure are scarce in Brazil, and they are limited to certain regions of the country. A recent study from the city of São Paulo¹⁰ on patients who failed first-line therapy between 2013 and 2015 found a higher prevalence of the M184 V/I mutation (74.3%) and K103 codon (56.7%), similar to that found in other Brazilian states [11]. Among the 205 subjects who started the second-line therapy with NRTI(t)s combined with protease inhibitor/ritonavir (PI/r), 76.6% experienced virologic suppression (< 200 copies/ml), despite the extensive resistance of the virus to NRTI(t)s [10].

Low- and middle-income countries in sub-Saharan Africa, where 70% of PLHIV live [1], access 2 NRTI(t) + NNRTI as a first-line therapy [12]. This regimen has a low genetic barrier and easily results in cross-resistance. Thus, the sensitivity of NRTI(t)s may be compromised by multiple mutations that arise during the first virologic failure. In addition, NRTI(t)s have recognized risks of toxic effects. These factors are concerning and have led to the decision to retain this drug class as a second-line therapy, and they have motivated the development of randomized clinical trials to evaluate the efficacy of the salvage regimen with NRTI(t)s plus PI/r.

Mutations are associated with high reduced susceptibility or virologic response to relevant NRTIs. Mutations reduce NRTI susceptibility or virologic response, which contributes to reducing susceptibility in combination with other NRTI-resistant mutations. The impact of mutations associated with NRTI(t)s during the second-line regimen was evaluated through the SECOND-LINE, EARNEST and SELECT studies [13‐15] and they were recently reevaluated in a meta-analysis [20], which showed the success of salvage regimens based on NRTI(t)s combined with PI/r. In this study, we analyzed the genotypic profile of HIV-1 in patients during the first virologic failure and the predictors of virologic success 48 weeks after switching to a salvage regimen.

This work was a cross-sectional, retrospective, multicenter study evaluating genotypic resistance tests of PLHIV during the first virologic failure in the city of Recife, the capital of the state of Pernambuco, an important medical center in Northeast Brazil. The patients were enrolled in three large centers specializing in the care of PLHIV. We reviewed tests that were performed from January 2006 to December 2016 and included patients over 18 years of age who had used ARVs for more than six months and presented two consecutive viral load measurements, with an interval of at least 30 days, and values greater than 1000 copies/ml. Individuals who underwent ARV switching due to virologic failure without genotyping tests, those who used ARVs before receiving the first-line therapy, those showing an absence of mutations associated with resistance in genotyping, and those with insufficient data were excluded.

The immunovirologic and genotypic data and the history of ARV use were collected from genotyping tests, medical charts and through the Laboratory Test Control System (Sistema de Controle de Exames Laboratoriais - SISCEL- online platform). To identify the mutations associated with resistance, the ViroSeqTM HIV-1 Genotyping System kit (Celera Diagnostics, Alameda, CA, USA) was used from 2006 to 2008 and the TRUGENE System (Siemens, Munich, Germany) was used from 2009 to 2016 by the laboratories that served the National Genotyping Network (Rede Nacional de Genotipagem - Renageno). The viral load (VL) was measured until July 2013 using the Versant HIV-1 RNA 3.0 assay (bDNA, Siemens, USA); real-time qPCR was used after this date (Abbott Laboratories, USA). The CD4 T-cell counts (CD4) were performed by flow cytometry (BD, USA).

The Stanford HIVdb program version 8.4 (https://​hivdb.​stanford.​edu/​hivdb/​by-mutations) was used to interpret the genotype resistance. The mutations were analyzed individually using a list provided by the International AIDS Society (IAS) in 2017, [16] and the Stanford HIVdb program and categorized as mutations associated with NRTI(t), NNRTI and major and accessory mutations for protease inhibitors (PI). Forty-eight weeks after switching to second-line therapy, virologic suppression was considered successful when an undetectable viral load was found by the given method. Viral load values made available 60 days before or after the date of completion (for the 48 weeks) were considered. The time on ARVs was the period between the beginning of the first ARV regimen and the performance of the genotyping test. The time in virologic failure was the period between virologic failure detection (two consecutive viral loads above 1000 copies/ml) and the switch to the salvage regimen.

The genotypic sensitivity score (GSS) was calculated individually only for the drugs used by each patient in the salvage regimen. A value of 1 was attributed to drugs with full activity, 0.75 for a low potential resistance level, 0.5 for a resistance level, 0.25 for an intermediate resistance level and zero if the drug showed a lack of activity.

SPSS 13.0 (Statistical Package for the Social Sciences) for Windows and Excel 2010 were used for the statistical analysis. The results are presented in the tables with their respective absolute and relative frequencies. The mean or median values for the quantitative variables were calculated according to the normality of their distribution. The normality of the quantitative variables was tested using the Kolmogorov-Smirnov test.

A total of 534 medical records were evaluated, of which 270 were excluded due to a lack of genotyping tests when patients switched ARVs after the first virologic failure. In 59 medical records, there was no selective pressure during genotyping. Therefore, the absence of mutation does not represent reality. Thus, 59 patients were excluded. Non-adherence to ARV use at the time of genotyping was recorded in the medical charts of the 59 excluded patients with absence of resistance-associated mutations. Furthermore, 21 patients were not either illegible or had insufficient data.

A total of 184 patients were included in the first virologic failure for analysis. The majority of the participants were male (73.4%) and the mean age at virologic failure was 41 (±9.4) years. The median CD4 (n = 155) and VL (n = 141) pre-ARV were 176 cells/m3 and 130,000 copies/ml, respectively.

This sample is representative considering the limited resources scenario. It represents a decade of genotyping in the first virologic failure in Pernambuco, Northeast Brazil. In 2016, we were able to access genotyping easily. Between 2006 and 2016, there was therapeutic rescue without genotyping due to the logistic difficulty in this region of Brazil. Therefore, more than 50% of the sample was excluded by changing the scheme without genotyping.

After evaluating 184 genotyping tests from patients during the first virologic failure, we found a higher prevalence of subtype B, of the M184 V/I and K103 N mutations, as well as a high frequency of NRTI(t) and NNRTI-associated mutations, with no impact on virologic suppression. We observed that the salvage therapy regimen was predominantly composed of PI/r and NRTI(t)s, with virologic success in most cases. Subtype B remains the most common in Pernambuco [17‐19] and in Brazil [11], except in the south, where subtype C [21] is predominant. There has been an increase in the proportion of recombinant forms in Rio de Janeiro [22] and subtype F in Minas Gerais [40].

The elevated presence of M184 codon mutations is expected and arises as a consequence of the use of lamivudine as part of all the first-line regimens in our study. This drug confers a high level of resistance to cytosine analogs (lamivudine and emtricitabine), a low level of resistance to abacavir, and the increased susceptibility of zidovudine and TDF. In addition, it decreases the replication capacity of HIV-1 [23, 24]. Its presence has been associated with virologic success [10], but we did not observe this success in the present study.

Similar to our results, the high prevalence of M184 V/I mutations was reported in several regions of Brazil [11, 25, 40], in Sub-Saharan Africa [26] and in Asia [27], but to a lesser extent in western Europe [28]. This difference can be explained by the use of emtricitabine in European countries and by the use of lamivudine in low- and middle-income settings. However, in a recent meta-analysis [29], lamivudine and emtricitabine were clinically equivalent. All the genotype sequences of the non-B subtype (F and BF) had the M184 V/I mutation, probably due to the high prevalence of this mutation and the very low frequency of non-B subtypes in our study.

We found no association between the number of NRTI(t)-associated mutations and the ARVs used at the time of genotyping, including ARV regimens with or without TDF. There are studies showing a greater number of resistance-associated mutations among AZT [30] and TDF users [10, 31]. However, those studies had populations with different characteristics, especially with regard to subtype prevalence. A higher number of mutations with a TDF-based regimen was reported for both the B subtype¹⁰ as well as with the C subtype [33]. The association between a high number of NRTI(t)-associated mutations with CD4 pre-ARV and at the time of genotyping of > 200 cells/mm3 suggests a late diagnosis of PLHIV in our population, with an impact on the genotypic profile. These factors did not remain significant after the analysis with the Poisson regression model.

The K103 N/S mutation was documented in most genotypic sequences, due to the frequent use of EFV as the primary NNRTI in the first-line therapy for more than a decade in Brazil. These findings are similar to the national data [10‐12] and data from low- and middle-income countries [14, 15, 26, 27]. The susceptibility of etravirine, a second-generation NNRTI, was not influenced by experiencing a long period of time in virologic failure while using EFV or NVP, or by the number of NNRTI-associated mutations.

This analysis was hampered by the small number of NVP users at the time of virologic failure. However, some studies [32‐34] have shown that NVP-containing ARVs have repercussions on the response to etravirine for selecting the Y181C and G190A mutations, whereas the K103 mutation associated with EFV does not interfere with etravirine susceptibility.

The median time from ARV therapy to genotyping was similar to that of recent studies in Brazil [10], 10 Africa and Europe [36]. After the Poisson regression model analysis, remaining on ARV for more than 36 months until genotyping was the only factor that impacted the virologic response 48 weeks after the onset of salvage therapy, appearing as a protective factor for a detectable viral load. To understand the impact of time on ARV on virologic success, factors that could influence treatment adherence would have to be identified and controlled, because the major cause of acquired virologic failure is poor adherence to ARVs [42].

A shorter time on ARV is used as the best scenario for adherence in the calculation of odds ratio in studies that seek to understand the risk factors for treatment adherence. However, after association testing, its impact on adherence is not always confirmed [43]. This finding is probably observed because adherence is a dynamic behavior and varies over time for the same individual, and it is affected by diverse factors such as psychosocial-, ARV- and clinical scenario-related factors. This finding explains why the time on ARV may be a risk or protective factor for poor adherence, with an impact on virologic success [42].

By contrast, more than 36 months on ARV increased the risk for the selection of three or more NRTI(t)-associated mutations, although a longer time in virologic failure did not generate the same outcome. According to the virologic failure criterion of this study, individuals with low VL (> <1000 copies/ml) were not considered to be in virologic failure, because the genotyping tests available in Brazil during the data collection period were not able to expand the genetic material when the VL was < 1000 copies/ml. However, there is evidence that having a VL of > 200 copies/ml increases the risk for virologic failure and may lead to mutation accumulation [44]. Therefore, it is possible that many individuals who are on ARV for a long period of time (> 36 months) had low VL but were not diagnosed with virologic failure, accumulating mutations.

We observed that the presence of NRTI(t)-associated mutations or TAMs at the first virologic failure did not interfere with virologic suppression 48 weeks after the switch to a second-line therapy with NRTI(t)s combined with PI/r. This finding is not only due to the high potency of PI/r or to the direct activity of NRTI(t)s, and it may reflect an effect of the NRTI(t)s on viral fitness [35]. This finding is consistent with the results of three large randomized controlled trials in which patients with extensive resistance to reverse transcriptase evolved with virologic success after salvage therapy based on NRTI(t) with PI/r [13‐15].

We found no difference in the virologic response between regimens with an individual GSS of less than two and those greater than or equal to two, confirming the hypothesis that there is no difference between active, partially active or inactive NRTI(t)s in terms of virologic success [35]. In the EARNEST study [14], the group receiving inactive or partially active NRTI(t)s exhibited viral load suppression that was similar to if not better than that exhibited by the group receiving the regimen containing a fully active drug from a new class, and it was superior to that exhibited by the group using only a protease inhibitor.

It is possible that pharmacokinetic characteristics optimize the benefits of NRTI(t)s even if inactive. Thus, ARVs with a long intracellular half-life, such as TDF and lamivudine, may help the PI/r to maintain viral suppression. This effect is independent of the drug activity [41].

We were able to define the time in virologic failure until the switch to second-line therapy. Surprisingly, most of the patients with undetected viral loads were in virologic failure for more than 12 months. Because the time under virologic failure did not interfere with the number of NRTI(t)-associated mutations or TAMs in our study, we can suggest that the resistance-associated mutations arose during the first months of virologic failure. In the EUROSIDA study [37], a high number of TAM-1 mutations was observed within one year of failure, with a lower accumulation rate than what was predicted among those who stayed on the failing regimen.

The total absence of ARV selective pressure during part of the virologic failure period is another way to explain why the longer virologic failure time did not influence the number of NRTI(t)-associated mutations or TAMs. The complete withdrawal of ARVs generates fewer resistance-associated mutations than does the maintenance of sub-therapeutic doses [9]. It was not possible to determine whether the study participants were kept on sub-therapeutic doses or if they completely abandoned the treatment when they were in virologic failure.

When we analyzed the virologic response and ARVs used in salvage therapy, we observed that there was no difference between the patients who used AZT or TDF or both. The use of LPV/r was significant among those who experienced virologic success, probably because it was the most frequently prescribed PI during the study period, as directed by the Brazilian Ministry of Health. Among users of DRV/r or RAL, the majority achieved virologic success, although they were smaller in absolute numbers.

DRV/r is the newest PI and has a high genetic barrier, in addition to fewer adverse effects than LPV/r, possessing activity even with a protease mutation [38]. When the protease is not intact, adding a new ARV class is critical. This class has often been the integrase inhibitor [39]. It should be noted that this study was not designed to assess the individual power of each regimen. Among those who reached an undetectable VL, a CD4 gain above 100 cells/mm3 was significant in subjects with CD4 at the time of genotyping > 200 cells/mm3, demonstrating that when there is adherence, a significant immunologic gain is possible even in patients with a low immunologic reserve and extensive resistance in reverse transcriptase.

Our primary limitation was the difficulty in establishing cause-and-effect relationships, because this was a cross-sectional study. Our results may have been partially affected by data we were unable to collect, particularly those related to adherence to ARV therapy, and by the loss of information in some variables. However, this study has relevant points because it updates the genotypic data of an important region of Brazil with limited resources. In addition, we were able to establish the time in virologic failure and the ARV regimen before and after genotyping in the first virologic failure as well as the factors associated with the number of NRTI(t)-associated mutations and virologic success.

Patients in their first virologic failure who were seen at referral centers in the city of Recife, Pernambuco, Northeastern Brazil presented a high frequency of mutations pertaining to secondary resistance to the use of NRTI(t)s and NNRTIs. However, extensive resistance in reverse transcriptase did not impact the second-line virologic suppression after 48 weeks of the salvage therapy with PI/r combined with active, partially active or inactive NRTI(t)s. The time on ARVs to genotyping was an independent protective factor for a detectable viral load. Long-term follow-up is needed to support the use of partially active NRTI(t)s in salvage therapy regimens while controlling for adherence-related factors. Studies assessing the actual need for genotyping after the first virologic failure with NNRTIs are also needed.