High Frequency of Clinically Significant Mutations after First-Line Generic Highly Active Antiretroviral Therapy Failure: Implications for Second-Line Options in Resource-Limited Settings

Abstract

Continuation of failed highly active antiretroviral therapy regimens can lead to the accumulation of mutations that may limit options for second-line treatment. We studied the pattern of drug resistance mutations among 138 Indian patients who experienced failure of nonnucleotide reverse-transcriptase–containing first-line highly active antiretroviral therapy. This study demonstrates a high frequency of drug resistance mutations in human immunodeficiency virus–infected Indians who experience immunologic treatment failure and suggests the need for viral load monitoring.

Currently, the World Health Organization (WHO) recommends monitoring of CD4 cell counts every 6 months after initiation of highly active antiretroviral therapy (HAART) because of the high cost and technological infrastructure required for human immunodeficiency virus (HIV) RNA testing [1]. Consequent immunologic failure and clinical events after initiation of HAART generally occur 6 months to 2 years after virologic failure [2]. Studies have documented that continuation of failed HAART regimens can lead to the accumulation of genotypic mutations, which may limit options for second-line treatment [3, 4]. Understanding patterns of mutations among patients who are experiencing failure of first-line HAART with use of immunologic monitoring can assist clinicians in selecting second-line regimens in resource-limited settings with already constrained second-line treatment options. Therefore, the present study was undertaken to examine the pattern and severity of genotypic mutations among HIV subtype C–infected South Indian patients experiencing failure of first-line HAART.

Patients and methods.

YRG Centre for AIDS Research and Education (CARE) is a nonprofit medical and research institution in Chennai, India, that provides medical care to 111,000 HIV-infected individuals. All patients were treated according to WHO treatment guidelines [1]. Patients were seen every 3 months or as clinically indicated. CD4 cell count monitoring was performed every 3–6 months. Plasma viral load monitoring was not standard of care. Data were collected under the approval of YRG CARE’s free-standing institutional review board.

Patients naive to antiretroviral therapy before initiation of HAART who later underwent genotyping after immunologic or clinical failure of first-line therapy were included in this analysis. Of the 11,528 HIV-infected individuals who registered at YRG CARE from February 1996 through March 2008, 4848 initiated first-line HAART containing zidovudine or stavudine, lamivudine, and nevirapine or efavirenz. Samples for genotyping were obtained at the same time as diagnosis of treatment failure. According to WHO treatment guidelines, immunologic failure was defined as a decrease in CD4 cell count to baseline levels (or less than baseline levels) after 6 months of therapy, a persistent CD4 cell count <100 cells/μL after 6 months of therapy, or a 50% decrease in CD4 cell count from the peak value during treatment [1]. The clinical criterion for treatment failure was the development of an AIDS-defining illness after ⩾3 months of HAART. HIV subtype C was the predominant strain in this patient population [5]. None of the patients received a diagnosis of treatment failure on the basis of plasma viral load.

Genotyping was performed using Home-brew Assay (reverse-transcriptase nested polymerase chain reaction) [6, 7]. Interpretation of the genotype in terms of drug resistance was based on an algorithm established by the Stanford HIV Reverse Transcriptase and Protease Sequence Database (http://hivdb.stanford.edu/).

Descriptive statistics were calculated with mean values and standard deviation for variables that were normally distributed; median values and interquartile range (IQR) were calculated for variables influenced by extreme values. Statistical analyses were performed using SPSS, version 13.0 (SPSS). A P value <.05 was considered to be statistically significant.

Results.

The present data include the first 138 patients who underwent genotyping at the time of treatment failure before switching to second-line regimens, and only these patients are included in the subsequent analysis; 37% were female (table 1). The median CD4 cell count at the time of HAART initiation was 69 cells/μL (IQR, 40–125 cells/μL) and at the time of failure of first-line treatment was 144 cells/μL (IQR, 90–199 cells/μL). The mean duration of treatment with a first-line regimen was 4.2 years. Forty-six percent of patients experienced clinical failure and developed a new WHO-defined opportunistic infection.

Table 1.

Characteristics and patterns of mutations in 138 patients at the time of treatment failure.

Variable	Value
Male	63
Female	37
First-line HAART regimen
Lamivudine plus stavudine plus nevirapine	46
Zidovudine plus lamivudine plus nevirapine	29
Lamivudine plus stavudine plus efavirenz	10
Zidovudine plus lamivudine plus efavirenz	7
CD4 cell count
At the time of HAART initiation, cells/μL	69 (40–125)
At the time of first-line HAART failure, cells/μL	144 (90–199)
Reasons for treatment failure
Clinical failure	46
Decrease in CD4 cell count to less than or equal to baseline value	2
50% Decrease in CD4 cell count from peak value	37
Persistent CD4 cell count <100 cells/μL	15
>1 Major mutation with drug resistance to NRTIs	90
>1 Major mutation with drug resistance to NNRTIs	65
⩾2 major NNRTI and/or NRTI mutations	88
No NRTI or NNRTI mutations	5
Major NRTI resistance mutations
M184V	79
T69D	4
L74V	7
K65R	5
Q151M	11
TAMS1 pathway
M41L	40
T215Y/F	39
L210W	7
TAMS2 pathway
D67N	32
K70R	18
K219E/Q	12
Major NNRTI resistance mutation
Y181C	33
K103N	27
G190A	26
TAMSa	60
M184V and TAMS	50
Q151M and TAMS	6
K65R and TAMS	3
L74V and TAMS	7
K65R and clinical failure	3
L74V and clinical failure	4
M184V: duration of first-line HAART, years	4.1 (2.1–6.6)
K103N: duration of first-line HAART, years	4.5 (2.9–6.9)

NOTE. Data are percentage of patients or median value (interquartile range [IQR]). HAART, highly active antiretroviral therapy; NNRTI, nonnucleotide reverse-transcriptase inhibitor; NRTI, nucleotide reverse-transcriptase inhibitor; TAMS, thymidine analogue mutations.

^aD67N, M41L, T215Y/F, K219E/Q, K70R, and L210W.

Of the 138 patients, 6.6% had only 1 mutation, and 88% had ⩾2 major nonnucleotide reverse-transcriptase (NNRTI) and/or nucleotide reverse-transcriptase (NRTI) genotypic mutations. Of the NRTI resistance mutations, M184V was the most common (79%). Thymidine analogue mutations (TAMS) were found in 60% of patients. TAMS 1 mutations included M41L (40%), T215Y/F (39%), and L210W (7%); TAMS 2 mutations included D67N (32%), K219E/Q (12%), and K70R (18%). Eleven percent of patients had the Q151M mutation. K65R, L74V, and T69D mutations were observed in 5%, 7%, and 4% of patients, respectively. NNRTI resistance mutations included K103N (27%), Y181C (33%), and G190A (26%). Twenty-five percent of patients had ⩾3 NNRTI mutations. Forty-six percent of patients had M184V and TAMS. Eight patients (6%) had Q151M and TAMS. Three patients (2%) with K65R had previously received a stavudine-containing regimen.

Patients without TAMS had significantly higher median CD4 cell counts at treatment initiation than did patients with TAMS (122 cells/μL vs. 65 cells/μL; P = .035). Patients with and without TAMS had similar rates of opportunistic infection (48% vs. 44%) and adverse events related to therapy (74% vs. 78%), as well as similar mean durations of first-line therapy (4.3 vs. 4.4 years).

Discussion.

Our study from India complements evidence from Africa that the current WHO guidelines for immunologic and clinical monitoring for antiretroviral drug treatment failure fail to detect resumption of viral replication in a significant proportion of patients before selection of highly drug-resistant viruses [8]. We detected a high frequency of NRTI and NNRTI resistance mutations in patients who experienced failure of first-line NNRTI-based regimens, as defined by the WHO guidelines. The high level of NNRTI mutations was not surprising because of the low genetic barrier to drug resistance for these agents. Of greater concern is the fact that more than half of the patients who experienced treatment failure had ⩾1 TAMS. These mutations are selected for by stavudine or zidovudine and cause cross-resistance to all thymidine-based NRTIs. Because participants were monitored according to WHO treatment guidelines, our drug resistance study sample was restricted to patients who experienced immunologic or clinical failure. On the basis of the experience reported by Mee et al. [8], WHO’s combined immunologic and clinical criteria may detect as few as one-third of patients who experience resumption of viral replication. Thus, the prevalence of drug resistance mutations in our cohort should be viewed as an underestimate of the full amount of drug resistance selected by WHO’s treatment management algorithm. Maintenance of patients experiencing treatment failure has a substantial impact on long-term mortality [9].

In addition to the insensitivity of the WHO clinical and immunologic criteria for virologically defined treatment failure, evidence is emerging that these criteria are lacking in specificity. In the experience of Mee et al. [8], only 11 of 52 patients who experienced treatment failure—as defined by the clinical and immunologic criteria—at 1 year after initiation of treatment had documented virological failure. Because we did not routinely measure plasma HIV RNA levels in every patient who experiences treatment failure according to WHO’s clinical or immunological criteria, we cannot comment on this aspect of the experience in India. In the absence of viral load measurements, an even greater number of patients whose therapy is defined as failing by clinical and immunologic parameters may be prematurely administered second-line therapy, compared with patients who are missed and or are assumed to be successfully treated.

The addition of viral load monitoring should be a priority in resource-limited settings. It has been argued, on the basis of mathematical models, that cost-benefit analyses do not justify the addition of viral load testing to treatment monitoring in resource-limited settings [10]. Although these models are eloquent, they do not fully take into account the impact of transmitted drug resistance and the additional costs of premature progression to second-line regimens. Furthermore, the costs of viral load monitoring will likely decrease further over time, as has been the case with CD4 cell counts and antiretroviral drugs. Although it has become conventional in most resource-rich settings to perform viral load testing on a repetitive basis at 3–6-month intervals, it is not clear that this strategy is necessary in all patients for their entire duration of therapy. Because most treatment failures occur during the first year of treatment, the introduction of more testing during the early period of treatment to assess adherence and reduction over time is a strategy that should be evaluated. In a recently reported study from KwaZulu Natal, the introduction of HIV RNA level monitoring at 6-month intervals identified patients who were experiencing treatment failure at a time when substantially less drug resistance had accumulated than in our cohort [11].

In a Thai cohort, the rate of TAMS was significantly higher among patients with HIV RNA loads >4 log₁₀ copies/mL than among patients with HIV RNA loads [H11088]⩽4 log₁₀ copies/mL [12]. Similar observations were seen in the Development of Antiretroviral Therapy in Africa (DART) Virology study in an African setting [13]. If viral load testing was available, a number of individuals who consequently experienced immunologic and clinical failure could have been identified at an earlier date at the time of virological failure [14]. Early detection could have preserved future treatment options and might have prevented opportunistic infections in those who experienced treatment failure.

In light of the high prevalence of multiple NNRTI mutations in this study population, the use of newer NNRTIs, such as etravirine, as a second-line agent would be under jeopardy [15]. Currently, first-line antiretroviral therapy has become inexpensive, but the cost of plasma viral load testing remains high. To keep open options for second-line treatment in patients who experience treatment failure, the HIV RNA assay remains a vital assessment

The present study included the first 138 patients who underwent genotyping in our cohort. Although this is a small sample of the total patients who received care, it is unlikely to be a biased sample, because all patients were consecutively enrolled after they experienced treatment failure. This study found that >50% of patients who experience treatment failure, as defined by WHO, have multiple treatment-limiting mutations. The results of this study can assist in the development of antiretroviral therapy guidelines for patients who experience treatment failure in resource-limited settings where genotyping is not available. Studies that address operational issues, such as optimal use of treatment monitoring tools, should be a research priority.