Optimizing treatment switch for virologic failure during first-line antiretroviral therapy in resource-limited settings

Abstract

We evaluated adult Nigerian patients with antiretroviral switch to second-line treatment with protease inhibitor (PI/r)-based regimens due to virologic failure (confirmed HIV-1 RNA viral load >1000 copies/mL) during first-line antiretroviral therapy. Proportion of patients with viral load (VL) >400 copies/mL and characteristics associated with non-suppression during second-line treatment are described. Approximately 15% of patients (34/225) had VL >400 copies/mL at 1-year after treatment switch to PI/r-based regimens. In adjusted analyses, VL ≥5 log₁₀ copies/mL at treatment switch (OR 2.90 [CI 1.21–6.93]); duration of first-line treatment after virologic failure >180 days (OR 2.56 [CI 1.0–6.54]); and PI/r regimen adherence <90% (OR 3.27 [CI 1.39–7.68]) were associated with VL >400 copies/mL at 1-year of second-line treatment. We therefore recommend that the maximum permissible time between suspicion of virologic failure and completion of antiretroviral treatment switch should not exceed 6-months when patients develop first-line antiretroviral failure in resource-limited settings.

INTRODUCTION

As HIV treatment programs in resource-limited countries mature, a growing number of patients are failing first-line antiretroviral therapy (ART) ^[1-3]. Detection of ART failure may be delayed due to sub-optimal access to HIV-1 RNA plasma viral load (VL) testing for routine follow-up of treatment in these settings ^{[4, 5]}. Continuation of virologically non-suppresive highly active antiretroviral therapy (HAART) is associated with increased risk of morbidity, as well as accumulation of antiretroviral resistance mutations ^[6-9]. A need exists for locally generated data to inform prompt treatment switch when patients fail first-line ART within under-resourced programs ^[6]. HAART was introduced into Nigeria in 2002, and ART scale-up has been ongoing for nearly a decade. Nigeria, with a HIV sero-prevalence of 4.1% in 2010, has the second largest number (estimated at 3.1 million) of people living with HIV/AIDS (PLWHA) in sub-Saharan Africa, of which approximately 360,000 patients are on HAART ^[10]. Currently, VL testing is not routinely available within ART programs in Nigeria. Therefore, analyses of virologic outcomes of ART are few ^[11], and patients failing first-line ART are typically detected by using clinical and immunological criteria ^[12]. In the AIDS Prevention Initiative in Nigeria (APIN) PLUS program, VL testing is used for routine monitoring of ART, and patients who develop virologic failure during first-line ART (with a non-nucleoside reverse transcriptase inhibitor-based regimen) are switched to a ritonavir-boosted protease inhibitor (PI/r)-based regimen for second-line treatment. We analyzed a cohort of such patients for proportion with detectable VL at six months and one year following initiation of PI/r-based treatment, as well as characteristics associated with virologic non-suppression at these time points.

METHODS

This study was conducted at the HIV clinic of the University College Hospital (UCH), Ibadan. The clinic is an integrated facility providing care, treatment and support to over 10,000 PLWHA in Nigeria. Since 2004, UCH HIV clinic has received support from the Nigerian Government and the United States President’s Emergency Plan for AIDS Relief (PEPFAR) through the APIN PLUS program. This study was approved by the joint Ethics Committee of the University of Ibadan and UCH, and by the Northwestern University Institutional Review Board.

We performed an on-treatment analysis of 225 HIV-infected adults (at least 18 years old) enrolled in UCH HIV clinic who met the following inclusion criteria: (1) initiation of first-line combination ART (an NNRTI plus two nucleoside reverse transcriptase inhibitors NRTIs) between January 2006 and December 2008; (2) diagnosis of virologic failure (defined as two consecutive VL >1,000 copies/mL after six months of first-line ART; (3) treatment switch to a PI/r-based regimen without drug resistance testing; (4) treatment with a second-line regimen (ritonavir-boosted PI plus two NRTIs and one nucleotide reverse transcriptase inhibitor) for at least one year. ART eligibility was determined according to the Nigerian National Adult ART Guidelines ^[12] (now revised) which recommended ART for PLWHA with CD4⁺ cell counts ≤200 cells/mm³, and for PLWHA with CD4⁺ cell counts between 200 – 350 cells/mm³ in the presence of WHO stage 3 or 4 disease. We excluded patients who had switched to PI/r regimen due to other reasons besides virologic non-suppression. We obtained de-identified data from UCH HIV clinic electronic database which contained the following information collected at the time of initial and follow-up clinical visits: (1) demographics; (2) CD4⁺ cell counts; (3) VL; and (4) antiretroviral drugs dispensing records. All patients had VL and CD4⁺ cell count estimations done once every six months as part of routine treatment monitoring. These tests were performed at the UCH HIV Reference Laboratory using Cobas Amplicor HIV-1 Monitor assay, version 1.5 (Roche Diagnostic systems, Inc., Branchburg, NJ) for VL testing, and flow cytometry (Partec GmbH, Munster, Germany) for CD4⁺ cell count estimations. We defined detectable VL as >400 copies/mL based on the threshold of the assay used. The primary outcome measure was proportion of patients with a detectable VL at one year following initiation of PI/r regimen. Second-line regimen adherence was determined according to pharmacy-based antiretroviral dispensing records, which defined drug adherence as percentage of actual versus expected pharmacy refills of PI/r regimen over the first year of second-line treatment. The duration of first-line treatment after virologic failure was defined as the time interval (in days) between the first VL >1000 copies/mL and treatment switch to PI/r-based ART. We analyzed virologic suppression at six months and one year of PI/r-based treatment in relation to the following characteristics: sex, age, switch VL, switch CD4⁺ cell count, nadir CD4⁺ cell count, duration of first-line treatment after virologic failure, and PI/r adherence. We dichotomized covariates as follows: (1) age ≤35 years, (2) switch VL <5 log₁₀ copies/mL, (3) switch CD4⁺ cell count ≤200 cells/mm³, (4) nadir CD4⁺ cell count ≤100 cells/mm³, (5) duration of first-line treatment after virologic failure ≤180 days, (6) PI/r adherence <90%. We performed bivariate comparisons of categorical and continuous variables using Fisher’s exact and Wilcoxon rank sum tests, respectively, and utilized logistic regression to estimate multivariate odds ratios (OR, with 95% confidence intervals) associated with detectable VL at six months and one year of PI/r-based therapy. Continuous variables are presented as median and interquartile range (IQR), and categorical variables as percentages. A p value ≤ 0.05 was considered statistically significant.

RESULTS

Out of 225 patients, 212 (94.2%) had complete data on all study variables. All patients (100%) had VL results at six months and one year of PI/r-based ART. Sixty-five percent (146/225) were women; median age was 34 years (IQR 29, 40). Seventy-three percent (164/225) and 27% (61/225) of patients received nevirapine-based and efavirenz-based first-line regimens, respectively. The median duration of first-line ART was 16.0 months (IQR 12, 23). The virologic and immunologic characteristics of study patients are depicted in Table 1. The proportion of patients with switch VL <5 log₁₀ copies/mL was 64.4% (145/225). Patients with duration of first-line treatment after virologic failure ≤180 days made up 38.2% (86/225). The median duration of first-line treatment after virologic failure was 258 days (IQR 112, 379). Ninety-nine percent of patients received ritonavir-boosted lopinavir (LPV/r)-based regimens. Concerning PI/r regimen adherence, 77.3% (174/225) had ≥90% adherence performance. At six months and one year of second-line treatment, 21.3% (48/225) and 15.1% (34/225) of patients, respectively had VL >400 copies/mL. In unadjusted analysis, patient characteristics significantly (p<0.05) associated with VL >400 copies/mL at six months and one year of second-line ART were: switch VL ≥ 5 log₁₀ copies/mL; duration of first-line treatment after virologic failure >180 days; and PI/r adherence <90%. As shown in Table 2, patients who had VL >400 copies/mL at one year of PI/r-based treatment (N=34) were significantly more likely to have had a switch VL ≥5 log₁₀ copies/mL (16/34); to have experienced duration of first-line treatment after virologic failure >180 days (27/34), and to have recorded PI/r adherence <90% (19/34) when compared with patients who achieved virologic suppression (N=191). Multivariate analyses (see Table 3) adjusting for sex, age, switch VL, duration of first-line treatment after virologic failure, PI/r adherence switch, CD4⁺ cell count, and nadir CD4⁺ cell count showed the following factors to be independently associated with VL>400 copies/mL at six months and one year of PI/r-based ART: switch VL ≥5 log₁₀ (OR 2.40 [CI 1.09–5.27]), and (OR 2.90 [CI 1.21–6.93]) respectively; duration of first-line treatment after virologic failure >180 days (OR 2.23 [CI 1.0 – 4.97]), and (OR 2.56 [CI 1.0 – 6.54]) respectively; and PI/r adherence <90% (OR 3.56 [CI 1.6 – 7.7]), and (OR 3.27 [CI 1.39 – 7.68]) respectively.

Table 1.

Virologic and Immunologic characteristics of study patients

Patient characteristic	Median (IQR)
Pre-ART VL (log10 copies/mL)	5.4 (4.8, 5.8)
Switch VL (log10 copies/mL)	4.6 (3.9, 5.2)
Nadir CD4+ cell count (cells/mm3)	63 (29, 132)
Switch CD4+ cell count (cells/mm3)	139 (58, 235)
CD4+ cell count at six months of PI/r follow up (cells/mm3)	264 (178, 400)
CD4+ cell count at one year of PI/r follow up (cells/mm3)	312 (214, 453)

Table 2.

Comparison of virologic outcomes at one year of PI/r-based antiretroviral therapy for 225 included patients

Patient characteristic	VL ≤400 copies/mL at one year of treatment	VL >400copies/mL at one year of treatment	p value
	Number of patients (%) N = 191	Number of patients (%) N = 34
Sex:
Male	70 (36.6%)	9 (26.5%)	0.25
Female	121 (63.4%)	25 (73.5%)
Age:
≤ 35 years	103 (53.9%)	18 (52.9%)	0.91
>35 years	88 (46.1%)	16 (47.1%)
Switch viral load:
<5 log10 copies/ml	130 (68.06%)	15 (44.12%)	0.01
≥5 log 10 copies/ml	51 (26.70%)	16 (47.06%)
Missing	10 (5.24%)	3 (8.82%)
Duration of 1st-line ART after virologic failure:
≤180 days	79 (41.4%)	7 (20.6%)	0.02
>180 days	112 (58.6%)	27 (79.4%)
PI/r regimen adherence:
<90%	36 (18.8%)	15 (44.1%)	0.001
≥90%	155 (81.2%)	19 (55.9%)
Switch CD4+ count:
≤200 cells/mm3	119 (62.30%)	19 (55.88%)	0.74
>200 cells/mm3	66 (34.56%)	12 (35.29%)
Missing	6 (3.14%)	3 (8.83%)
Nadir CD4+ cell count:
≤100 cells/mm3	122 (63.87%)	23 (67.6%)	0.72
>100 cells/mm3	67 (35.08%)	11 (32.4%)
Missing	2 (1.05%)	0 (0%)

Table 3.

Logistic regression results for undetectable viral load at six months and one year of PI/r-based antiretroviral therapy for 225 included patients

Patient characteristic	VL ≤400 copies/mL at six months of treatment		VL ≤400 copies/mL at one year of treatment
	Odds ratio (OR)	95% CI	Odds ratio (OR)	95% CI
Sex:
Male	1.87	0.80 – 4.37	1.56	0.60 – 4.04
Age:
≤35 years	0.52	0.24 -1.13	0.77	0.33 – 1.83
Switch viral load:
<5 log10 copies/ml	2.40	1.09 – 5.27	2.90	1.21 -6.93
Duration of 1st-line ART after virologic failure:
≤180 days	2.23	1.0 – 4.97	2.56	1.0 – 6.54
PI/r regimen adherence:
≥90%	3.56	1.6 – 7.7	3.27	1.39 – 7.68
Switch CD4+ cell count:
>200 cells/mm3	0.81	0.35 – 1.85	0.78	0.30 – 2.0
Nadir CD4+ cell count:
>100 cells/mm3	0.67	0.31 – 1.51	1.07	0.41 – 2.74

DISCUSSION

Our study describes virologic outcomes among 225 patients on PI/r-based second-line treatment for at least one year in a Nigerian ART program. The proportion of patients (191/225) with undetectable VL at one year of follow-up in our cohort is comparable to an MSF study ^[13] which reported 81.2% (513/632) virologic suppression rate after 12 months of follow-up of African and Asian patients on second-line ART. A smaller-sized Cambodian study ^[14] reported that 92.3% (60/65) of second-line ART patients achieved undetectable VL by 24 months of follow-up. Our protocol-specified definition of virologic failure (>1000 copies/mL) is similar to that utilized in the Cambodian study by Ferradini et al ^[14]. The key finding of our study is that patients with a high switch VL (≥5 log₁₀ copies/mL) and prolonged time (>180 days) on a failing first-line regimen are twice likely to be virologically non-suppressed within the first year of second-line ART when compared to patients who had a lower level of switch VL and a shorter treatment exposure to a failing first-line regimen. Comparable to our median switch VL of 4.6 log₁₀ copies/mL, Ferradini and colleagues ^[14] reported a median VL of 4.7 log₁₀ copies/mL at switch to second-line ART in their cohort. Multiple factors contribute to VL set-point during treatment failure, including the extent of drug resistance mutations, and viral fitness. In Nigeria, genotypic resistance testing is not routinely available for management of ART patients, and was not performed in this study. Ferradini et al ^[14] reported that at ART switch, HIV drug resistance analysis in the reverse transcriptase gene showed that 100% (41/41) of those tested (41/70) had resistance mutations conferring viral resistance to both lamivudine/emtricitabine (3TC/FTC) and nevirapine/efavirenz (NVP/EFV), and 60.9% had resistance to these agents in addition to stavudine/zidovudine (d4T/ZDV), suggesting that failure to achieve virologic suppression was due to drug resistance. A Togolese study ^[15] of 188 patients on NNRTI-based first-line ART found that only 24.5% (46/188) had drug resistance mutations, of which 100% (46/46) were resistant to NNRTIs, and 54.3% (25/46) harboured the M184V mutation conferring resistance to 3TC/FTC. Our study finding that prolonged exposure to a failing first-line regimen is associated with virologic non-suppression on second-line ART supports previous reports ^[16] that accumulation of resistance mutations to first-line drugs plays a role. In our cohort, the median duration of first-line treatment after virologic failure extended well over 180 days. This observation merits closer scrutiny, and suggests clinician ambivalence to suspicion and confirmation of virologic failure. Occasions of prolonged turn-around time for VL test may contribute to this observation, with the implication that treatment switch processes are delayed for as long as it takes clinicians to receive VL test results. Economic considerations may also inform switch-delaying behaviour by patients, and physician wariness of unnecessary ART switches ^[17]. Current public health-oriented ART switch guidelines ^[18] recommend first-line to second-line treatment switch when confirmed VL exceeds 5000 copies/mL after at least six months of HAART in a person whose drug adherence is determined to be sufficient. Our study protocol exceeded these standards. The limitations of this study include being an on-treatment analysis with a small sample size, and without the benefit of HIV drug resistance analysis at any point during ART. Furthermore, due to the low proportion of patients with switch VL <4 log₁₀ copies/mL in this study, a switch VL threshold below 5 log₁₀ copies/mL was not accommodated in our analysis. Nonetheless, our findings suggest that as VL testing becomes increasingly available for ART monitoring in low and middle-income countries, a need will emerge to proactively manage patients at risk of first-line ART failure by putting in place measures which are sensitive to the time consuming aspects of the treatment switch process.

We conclude that the success of second-line ART is threatened when the processes for confirmation of virologic failure and completion of ART switch exceed six months. We therefore suggest that clinical guidelines for ART switch that are based on patient’s level of viremia as well as a maximum permissible time of six months on a failing first-line regimen may improve collective insight for prompt response to suspicion and confirmation of virologic failure during first-line ART in resource-constrained programs.