Secondary Logo

Journal Logo


HIV-1 second-line failure and drug resistance at high-level and low-level viremia in Western Kenya

Kantor, Ramia; DeLong, Allisonb; Schreier, Leeanna; Reitsma, Marissaa; Kemboi, Emanuelc; Orido, Millicentc; Obonge, Salomec; Boinett, Robertc; Rono, Maryc; Emonyi, Wilfredc; Brooks, Katiea; Coetzer, Miaa; Buziba, Nathanc,d; Hogan, Josephb; Diero, Lameckc,d

Author Information
doi: 10.1097/QAD.0000000000001964



More than 35 million people are infected with HIV worldwide, most in resource-limited settings (RLS), predominantly sub-Saharan Africa [1]. Access and eligibility to antiretroviral therapy (ART) are increasing globally; however, only 46% of ART-eligible individuals receive it, a vast treatment gap [2,3]. Treatment failure in RLS upon nonnucleoside reverse transcriptase inhibitor (NNRTI)-based first-line ART is estimated at 24% within 12 months and 33% within 24 months [4]; and upon protease inhibitor-based second-line ART at 23% within 12 months and 26% within 24 months [5]. As the treatment gap narrows, more individuals in settings where routine sustainable virological monitoring capacity and third-line salvage regimens are limited will fail therapy with potential rises in transmitted and acquired drug resistance [3,6]. This concerning scenario, which may result in poor clinical outcomes, mandates strategic planning to minimize resistance evolution and transmission [7,8].

Adult HIV prevalence in Kenya (5.6–6.1%; 2012) is the 12th highest worldwide, representing a high health burden [9,10] of an epidemic with multiple circulating subtypes [11]. ART access has significantly increased in Kenya since 2001, with positive clinical outcomes [1,12]. As of June 2017, 70% of eligible adults were receiving ART, 93% on first-line ( Standard first-line regimens included zidovudine (ZDV) or stavudine (d4T); lamivudine (3TC); and nevirapine (NVP) or efavirenz (EFV), with tenofovir (TDF) substituting d4T in 2013, per WHO guidelines [3]. Data on transmitted [11,13–16] and first-line [17–21] resistance are limited, and nonexistent for second-line, which mostly includes ZDV or TDF, 3TC, and lopinavir/ritonavir (LPV/r) [22], with atazanavir available since 2014.

At the Academic Model Providing Access to Healthcare (AMPATH), a large HIV program in sub-Saharan Africa [23], patients are suspected to fail ART with two consecutive viral load tests more than 1000 copies/ml, despite adherence counseling. Patients failing first-line are switched to second-line, the use of which has consistently increased, with 3943 patients receiving it in 2010, and 6968 in 2016. Third-line options (e.g. darunavir, etravirine, raltegravir) have been available since 2015, with a recent addition of dolutegravir; however, this process is still restricted, and data on the extent and impact of resistance upon continued second-line, particularly at viral load below the WHO-defined failure threshold of 1000 copies/ml [3], are limited [24].

To address the lack of data on failure and resistance in experienced individuals failing second-line in Kenya and other RLS, we determined the prevalence and correlates of second-line virologic failure in a large AMPATH adult cohort, examined resistance patterns among those with detectable viral load above and below 1000 copies/ml, and ascertained their accumulation and longitudinal effect on future drug options. We hypothesized substantial resistance accumulation over time with impact on third-line salvage regimen options at both high-level and low-level viremia (LLV).


Study setting

At the time of this study, 165 461 HIV-infected persons had received care at AMPATH. Of those, 91 548 were actively in care, 70 224 (77%) of whom received ART: 65 577 NNRTI-based first-line and 4647 LPV/r based second-line. Study enrollment was at the Moi Teaching and Referral Hospital (MTRH) clinic, AMPATH's largest, where 18 282 adults (≥18 years) were followed, 76% (n = 13 983) of whom started ART; 12 741 first-line and 1242 second-line.

AMPATH patients are managed with an electronic medical record [25] according to locally developed protocols per WHO guidelines. At the time of the study CD4+s were done at HIV diagnosis and every 6 months, and viral load access was limited. Individuals suspected as failing first-line by consecutive viral load more than 1000 copies/ml or immunological failure were switched to a standard second-line regimen. Individuals failing second-line remained on this regimen for lack of third-line options.

Participant enrollment

Inclusion criteria were: HIV-positive; at least 18 years; on at least 24 weeks LPV/r-based second-line; at least 6 months prior first-line (ZDV/d4T+3TC+NVP/EFV); and more than 50% self-reported adherence in the prior month and 7 days. Between June 2011 and May 2012, participants meeting inclusion criteria were offered enrollment, consenting participants were enrolled sequentially and interviewed, charts were reviewed for demographic, clinical and laboratory characteristics, and blood samples were obtained for CD4+, viral load and pol genotyping. Participants with detectable viral load (>40 copies/ml) at enrollment (first visit) were invited for a second visit 1 month later for repeat pol genotyping, after verification of unchanged ART and adherence. Lifespan and Moi University ethics committees approved the study.

Laboratory methods

CD4+ (FACSCaliber system; Becton Dickenson, San Jose, California, USA) and HIV viral load (Amplicor, Version 1.5; Roche Molecular, Pleasanton, California, USA) testing was done at the AMPATH laboratory, where they are routinely performed for clinical and research purposes. Plasma samples from both visits were frozen (−80 °C), batched and shipped to the Kantor lab for pol genotyping, as described [21]. Briefly, viral RNA was extracted from 400 to 1000 μl plasma via Biomerieux's MiniMAG (Durham, North Carolina, USA), followed by reverse transcription and PCR using SuperScript III First-Strand Synthesis System and Platinum Taq DNA Polymerase High Fidelity (ThermoFisher Scientific, Waltham, Massachusetts, USA). A 1.3 kb fragment spanning the pol gene (2147–3503, HXB2) was Sanger-sequenced and assembled with Sequencher v4.10.1.

Statistical analysis

Demographic and clinical data included age, sex, ART history including prevention of mother-to-child transmission (pMTCT) and treatment interruptions, prior pregnancy, tuberculosis (TB) treatment and previous CD4+s and viral loads. Sequence quality control was performed with SQUAT [26]. Resistance interpretation and predicted susceptibilities were evaluated using Stanford Database tools [27]. Measures of resistance at visits 1 and 2 included number of mutations and number and type of drug classes with associated resistance. Patients were classified as having resistance to current regimens if at least one medication had low-level or higher predicted resistance; and to future regimens if they had low-level or higher predicted resistance to at least one future medication (etravirine, rilpivirine, tenofovir, didanosine, abacavir, atazanavir, darunavir). Counts of resistance to future medications omitted currently prescribed drugs; counts of susceptibility to future medications included current drugs if susceptible. Accumulated resistance was defined as new mutations, seen in the second but not first visit. Visit 2 susceptibility was taken as the worse of the two visits. Pol subtyping was derived with REGA [28].

Viral load groups, based on visit 1 viral load, were defined as ‘Undetectable’ – viral load below lower limit of detection (≤40 copies/ml); ‘LLV’ – detectable viral load 1000 copies/ml or less; and ‘Failure’ – viral load more than 1000 copies/ml. We assessed relationships between viral load group and age, sex, WHO stage, TB treatment, time on first-line, time on second-line, adherence, CD4+, treatment interruptions, unique regimens (nontraditional first-line/second-line), number of first-line/second-line regimens, previous pregnancies, pMTCT treatment, and subtype using unadjusted logistic regression. One multivariable logistic regression was fit for each comparison using the covariates found to be most highly correlated with viral load group. Odds ratios (ORs) compare failure vs. nonfailure, detectable vs. undetectable, and failure vs. LLV. To examine representativeness of genotype data, demographic and clinical measures for those with and without genotypes were compared using Fisher Exact and Wilcoxon Rank Sum tests, stratified by viral load group.

Multiple statistical methods were used to examine associations between failure group (LLV vs. failure) at visit 1 as the predictor and various resistance measures as outcomes. At visit 1, dichotomous resistance mutation outcomes (e.g. any resistance) were examined using unadjusted logistic regression; categorical measures (e.g. resistance category) using Fisher Exact tests; and mutation counts using unadjusted Poisson regression. Tests for mutation outcomes at visit 2 were adjusted for the value of the measure at visit 1. At both visits, current and future medications (finite quantitative measures) were compared using t tests. 95% confidence intervals that exclude the value of the null hypothesis (i.e. 0 for t-tests and 1 for logistic and Poisson regression) were considered statistically significantly different.


Study population

A total of 394 eligible participants were enrolled (Table 1), lower than the planned 438 because of slow enrollment. The median age at enrollment was 42 years (IQR 35–48), 60% were women, and 70% had WHO stages 3 and 4. Before enrollment, participants were on second-line for median 1.9 years (IQR 0.8–4.1). Most common regimens included LPV/r with abacavir and didanosine with or without lamivudine (52%), LPV/r with ZDV or abacavir with didanosine with or without lamivudine (23%) and LPV/r with tenofovir and lamivudine with or without abacavir (19%; other regimens ≤1%). Median time on first-line was 2.9 years (IQR 1.9–4.1), most commonly lamivudine, zidovudine/stavudine and nevirapine (78%). Twenty-four percent were exposed to several second-line regimens and 39% to several first-line regimens. Treatment interruption on second-line was experienced by 3% (median 47 days, IQR 40–63), 4% between first-line/second-line (median 69 days, IQR 32–125), and 11% on first-line (median 49 days, IQR 28–70). Most participants reported complete adherence in the prior month (97%) and 7 days (94%). Eight percent received anti-TB medications on second-line. Among women, 38% had previous pregnancies, 70% once, and 16% received pMTCT ART. Median CD4+ cell counts were 282 cells/μl (IQR 182–419) and 17% (IQR 12–23), 28% with AIDS-defining CD4+ cell count less than 200 cells/μl.

Table 1:
Demographic, clinical, and laboratory characteristics of the study cohort according to viral load threshold.a

Virologic treatment failure

Detectable viral load was seen in 48% (n = 191; median 521 copies/ml, IQR 116–3,876), and 21% had viral failure (viral load >1000 copies/ml; n = 82; median 5424 copies/ml, IQR 1826–93 660; Table 1). Higher likelihood of viral load more than 1000 copies/ml vs. 1000 copies/ml or less was associated with younger age, concurrent TB treatment, shorter time on second-line, lower CD4+ cell count and percentage, and longer treatment interruption while on first-line, with similar results in the univariable and multivariable analyses for age, concurrent TB treatment, time on second-line and CD4+ count. Among women, prior pregnancy was significantly associated with viral load more than 1000 copies/ml (Supplementary Table 1,

Participants with detectable viral load were similar to those with undetectable viral load on most variables, except less time on second-line in univariable analysis and lower CD4+ values in both analyses. In univariable analysis, compared with participants with LLV, those with viral failure were younger, less likely with complete 1-week adherence, and had lower CD4+ values. CD4+ cell count remained significant in the multivariable model. Among women, those with failure were more likely to have ever been pregnant, but if pregnant, less likely to have taken ART for pMTCT (Supplementary Table 1,

Drug resistance and diversity

Pol sequences were available for 105 of 191 (55%) participants with detectable viral load, 35 of 109 (32%) with viral load 1000 copies/ml or less and 70 of 82 (85%) with viral load more than 1000 copies/ml. Clinical and demographic characteristics of participants with/without genotypes did not differ, other than association with increased viral load in participants with available genotypes (Supplementary Tables 2 and 3,

Seventy-nine percent of participants with genotypes had at least one resistance mutation; 67% NRTIs, 74% NNRTIs, 9% PIs, 57% dual-class, and 7% triple-class resistance. Figure 1 demonstrates resistance prevalence by viral load threshold, drug class category and subtype. Of genotyped viruses, 58% were subtype A, 22% subtype D, 8% subtype C and 12% others (4% subtype G, 2% subtype AC, 6% subtype AD, 1% subtype DC). The most prevalent (>20%) RT mutations included NRTI-associated M184V, T215Y/F/N/V, D67N/G, M41L and K219Q/R/N/E; and NNRTI-associated K103N/S, Y181C/V and G190A/S. The two most common protease inhibitor-associated mutations were M46L/I and V82A/L (Supplementary Figure 1, Higher viral load was associated with having fewer mutations (rate ratio 0.84 per 1-log10 higher viral load, CI = 0.74–0.97, P < 0.05). Table 2 provides further information on resistance measures according to viral load threshold. Other than the (very low) number of protease inhibitor-associated mutations, the LLV group had worse resistance in every examined measure. Statistical significance was observed for more overall as well as NRTI-associated mutations and more predicted resistance to next-regimen drugs. Demographic, clinical and laboratory characteristics of genotyped participants with/without resistance to current/future medications were not different in either viral load category. Participants with viral load greater than 1000 copies/ml and subtype D were significantly less likely to have resistance mutations compared with participants with viral load greater than 1000 copies/ml and subtype A (OR 0.14, CI = 0.03–0.61, P < 0.05).

Fig. 1:
Prevalence of drug resistance by drug-class, viral load category, and HIV-1 subtype.Demonstration of the proportion of patients with sequences (y-axis) that contained resistance mutations associated with different combinations of drug resistance categories (x-axis). For each category, left bars represent viral load 1000 copies/ml or less (represented by ‘≤103’) and right bars represent viral load greater than 1000 copies/ml (represented by ‘>103’). Major subtypes are also shown (A – black, C – dark gray, D – light gray, other – white; per the legend). NNRTI, nonnucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.
Table 2:
Various measures of resistance, overall and stratified by viral load group.a

Drug resistance accumulation

Of 105 genotyped participants, 48 also had second-visit sequences, median 55 days after the first visit (range 14–243 days). All sequence-pairs clustered well phylogenetically, with high (>85%) bootstrap values. Of these, 19 (40%) had 49 resistance mutations detected in the second visit, not in the first; eight of 48 (17%) with LLV (Table 3). Ten patients (53% of 19) accumulated NRTI mutations, 16 (84%) NNRTI mutations, and two (11%) protease inhibitor mutations. The most common position where mutations accumulated was RT 184 (six patients). Of 49 accumulated mutations, 26 (53%) were mixtures. In 11 of 48 (23%) participants, 28 resistance mutations detected in the first visit were not detected in the second.

Table 3:
Drug resistance mutations in participants with resistance accumulation.a

At the second visit, those with LLV were on average resistant to 1.65 medications in their current regimen compared with 1.00 medication among those with viral load more than 1000 copies/ml (P = 0.049, Table 2). Additionally, those with LLV were susceptible to fewer of the medications in their current regimen at visit 1 and 2 (P = 0.053). Those with LLV were susceptible to fewer future medications at visit 1 (CI = −1.81 to −0.11, P < 0.028) and visit 2 (CI = −2.23 to 0.2, P = 0.100) than those with viral load more than 1000 copies/ml. Those with LLV accumulated significantly more mutations (rate ratio = 2.31, CI = 1.27–4.17, P < 0.05) than those with viral load >1000, specifically NRTI mutations (rate ratio = 3.40, CI = 1.49–7.77, P < 0.05). Although not statistically significant, those with LLV also accumulated more NNRTI (rate ratio = 1.39, CI = 0.54–3.57, P = 0.50) and protease inhibitor (rate ratio = 4.59, CI = 0.48–44.11, P = 0.187) mutations.

Older participants were at increased risk of accumulating mutations (OR 2.35 per 10 years older, CI = 1.06–5.20, P < 0.05), whereas women were at decreased risk (OR 0.2, CI = 0.05–0.84, P < 0.05, data not shown). There was a nonsignificant trend towards increased risk of accumulation in participants with higher CD4+ cell counts (OR 1.65 per 100 cells higher, CI = 0.99–2.77, P = 0.064). Resistance accumulation occurred in 36% (10/28) of participants with subtype A, 25% (1/4) subtype C, 54% (7/13) subtype D, and 33% (1/3) other subtypes (P > 0.05).

Predicted susceptibilities to future options

At enrollment, 46% (48/105) of genotyped participants, in both viral load categories, had predicted intermediate–high level resistance to at least one future drug option: 26% etravirine, 29% rilpivirine, 18% tenofovir, 30% didanosine, 31% abacavir, 6% atazanavir, and 1% darunavir. Of the 48 patients with two available genotypes, 21% (10/48) had increased levels of predicted resistance at the second visit to intermediate–high levels, to at least one future drug option: 4% etravirine, 4% rilpivirine, 10% tenofovir, 13% didanosine, 15% abacavir and 2% atazanavir, more pronounced at LLV (Fig. 2).

Fig. 2:
Predicted susceptibility to future drug options upon second-line failure.Demonstration of the proportion of patients (y-axis) that had one of five levels of predicted susceptibility (legend) to antiretroviral medications that might be considered for future treatment options (x-axis). Results are provided for 105 patients with genotypes at the first visit (all; left bars for each drug); timepoint 1 (T1 = first visit; center bars for each drug; for 48 of 105 participants with genotypes at both visits); and timepoint 2 (T2 = second visit; right bars for each drug; for 48 of 105 participants with genotypes at both visits). Data are provided for all participants (top graph); those with low-level viremia (middle graph); and those with viral load greater than 1000 copies/ml (lower graph).


The goal of this study, the first of its kind in Kenya, was to guide management of HIV-infected persons failing second-line ART. In an analysis of 394 adults on second-line, 48% had detectable viral load and 21% had viral load more than 1000 copies/ml. Participants had low-level (9%) protease inhibitor resistance, but extensive NRTI (67%), NNRTI (74%) and dual/triple-class (64%) resistance, at viral load levels above and below the WHO threshold of 1000 copies/ml. Moreover, 40% of participants with follow-up sequences after an average 59 days accumulated resistance mutations, even at low-level viral loads, compromising potential future treatment options.

To our knowledge, this is the first report of resistance development at LLV in RLS. Almost one-third of participants in this study had LLV, consistent with current literature [29]. Although we and others identified correlates of persistent LLV, these vary widely and no specific risk factors have been identified and broadly accepted [29]. In RLS, resistance mutations at LLV upon second-line failure have not yet been investigated, mostly as individuals at this viral load range are not considered as failing ART per current WHO guidelines and their genotyping is more challenging. LLV studies in developed settings report resistance in 17–72%, suggesting detrimental outcomes of allowing patients to remain on therapy with low-level viral replication [24,30,31]. Participants with LLV in our study had 79% resistance and 72% dual-class resistance, higher than reported in non-RLS, with an even significantly higher prevalence of NRTI mutations compared with participants with viral load more than 1000 copies/ml. This is concerning, since these individuals are not considered as failing ART in most RLS, even with the now recommended annual viral load monitoring [10,32]. The 42% of participants with LLV who demonstrated resistance mutation accumulation further emphasize the potential impact of replication at this level. Additional investigation is needed to examine the impact of LLV resistance and the need to consider lowering the viral load threshold definition for ART failure and the sensitivity requirements for drug resistance testing assays.

In this study, the first to longitudinally examine intra-patient resistance accumulation upon second-line failure in RLS, 40% of participants accumulated mutations. A prior study in Nigeria demonstrated a similar concept, but using cross-sectional data [33]. Despite low overall protease inhibitor resistance, two participants accumulated new protease inhibitor mutations in the short time period between study visits. Accumulation of protease inhibitor mutations can lead to darunavir resistance [34], and indeed four patients (4%) in our cohort were found to have some resistance to darunavir, and one (1%) had intermediate resistance to this third-line drug. Interestingly, most participants who accumulated resistance had new NNRTI mutations in the second visit, despite not being on this ART class as part of second-line. This may be explained, though not examined here, by selection of more fit viral quasispecies with linked NNRTI mutations on the same viral genome [35], and the low-impact NNRTI mutations may have on viral fitness [36]. The distinction between mutation accumulation and re-emergence must, therefore, be further examined.

Virological failure rates of second-line ART in RLS vary widely (5–72%) across studies, settings and methodologies [5,37–39]. Our findings, from a large HIV program in western Kenya, fall within this wide spectrum. Prior reports associate second-line failure with poor adherence [5,38,39], longer time on first-line [5] and shorter time on second-line ART [39], lower CD4+ cell counts and higher viral load at second-line initiation [39,40], being women [41] and delayed switch to second-line [41]. We confirm some of these observations and extend them with other correlations of second-line failure: First, longer treatment interruptions while on first-line, consistent with reports of its association with first-line failure, poor adherence and/or resistance mutation accumulation [21,42]. Second, concurrent TB treatment, possibly resulting from medication interactions, particularly rifampicin and its induction of the CYP 450 enzymes, and the ensuing lower protease inhibitor concentrations [43]. This is an important finding in light of the high TB prevalence in RLS, which was previously shown in children [44], and recently in South-African adults [45]. Third, prior pregnancies among women, and if pregnant, less pMTCT exposure, possibly related to low-engagement in care. The finding that younger age is associated with second-line failure, contrasts prior reports [39,40] and may be related to specific settings. Altogether, these highly variable findings point to potential interventions to improve treatment monitoring and sustain second-line ART.

The low (9%) protease inhibitor resistance upon second-line failure is consistent with reports from RLS, ranging from 6 to 25% [34,38,46–48], as well as earlier reports, mostly from developed countries, with a slightly wider (0–33%), though still low, range [49–54]. As expected, considering future ART options, almost all participants with LPV/r resistance were also resistant to atazanavir, and darunavir seems an appropriate third-line option, as suggested by current Kenya ART guidelines [10]. The implications of low-protease inhibitor resistance upon second-line failure on subsequent ART susceptibilities and ability to recycle protease inhibitors, and the impact of postulated causes to such low resistance, including adherence [48,55], minority resistance variants [56], and newer resistance mechanisms [57–59], are under investigation. Higher rates (63–87%) of protease inhibitor resistance have been reported from India, Vietnam and Nigeria [33,37,60]; and reasons for this discordance, possibly associated with past ART exposure, differential treatment monitoring, viral subtype and host factors, should be further examined. The high reverse transcriptase resistance (65–95%) found here was associated with almost 30% predicted resistance to rilpivirine and etravirine, the newer-generation NNRTIs being considered in third-line regimens [10]. This augments similar reports from India (29%) [37], Mali (38%) [46], and Nigeria (at least 26%, based on mutational analysis of published genotypes) [33]. Though high, these data indicate that some patients may still be susceptible to these medications, if genotyping was available in this setting. Importantly, following the report of the first case of a multidrug resistant isolate in sub-Saharan Africa [61], two patients in this study had intermediate–high-level resistance to all first-line/second-line drugs available in the public sector (3TC, FTC, TDF, ABC, DDI, ZDV, d4T, EFV, NVP, ATV and LPV), substantiating and supporting the strategic need for third-line ART in RLS [10,62,63].

HIV-1 subtype may be relevant to clinical care and to resistance development [64–67]. Few studies have assessed second-line resistance in non-B subtypes [33,37,46,47]. Grossman et al. found no significant differences in second-line resistance development between subtypes B and C [52], and other studies have not included enough subtype diversity to allow inter-subtype comparisons. In this study, allowing resistance evaluation upon second-line failure in a RLS with diverse circulating subtypes, we found that at viral load greater than 1000 copies/ml, subtype D viruses had significantly less resistance compared with subtype A. These findings contrast data from Uganda demonstrating more resistance in subtype D vs. subtype A, though not solely upon second-line failure [68]. A larger study assessing differences in second-line failure and resistance between subtypes could elucidate the cause of such findings, and help guide clinical care in areas with diverse subtypes.

The major limitations of the present study are primarily its cross-sectional nature and the lack of continuous virologic monitoring and confirmed testing for viral load failure, limiting the ability to rule out viral blips, estimate time to virologic failure and determine duration of time on a failing regimen. This study design, however, allowed for accrual of a large number of patients, and unfortunately the limited monitoring is representative of routine circumstances in many RLS, even today [69]. Additionally, adherence was based on self-report; pre-ART and pre-second-line resistance testing were not available; and resistance testing sensitivity was limited to conventional, population-based pol genotyping, with no consideration of gag/env mutations or effects of minority variants. Lastly, treatment and monitoring guidelines evolve rapidly and new and improved ART as well as more routine viral load testing are available in some RLS, different than the study's settings. However, concepts presented here hold across guidelines and treatments, and will still be relevant in many settings for years to come.

In conclusion, in this large study in Kenyans on second-line ART, we found high rates of failure, extensive overall but low protease inhibitor resistance and resistance accumulation at viral load levels above and below the WHO treatment-failure threshold, that impact treatment options. Considering increased global ART access, the aging HIV-infected population and existing cost constraints in RLS, optimization of adherence and treatment monitoring (e.g. viral load/resistance testing) are required to support current local and global HIV cascade goals [70–72]. Continued investigation into resistance mechanisms, in particular when protease inhibitor resistance is absent and good adherence is documented, is needed to improve resistance detection and ART strategies [73]. As in resource-rich settings, treatment goals in RLS should be geared towards achieving virologic suppression in ART-experienced patients with resistance, making ART available to all and strategically planning for third-line regimens.


Funding: This work was supported by National Institutes of Health grants R01AI108441 and P30AI042853.

R.K. conceptualized the study; R.K. and A.D. curated and managed the data; R.K., A.D., L.S., M.Re., K.B. and M.C. analyzed the data, with specific development of methods and statistical analysis by A.D. and J.H.; E.K., S.O. and R.B. consented, enrolled and interviewed participants, supervised by L.D.; L.S., M.O. and M.Ro. conducted laboratory experiments, supervised by R.K., W.E., M.C. and N.B.; R.K., M.Re. and K.B. wrote the initial draft; all authors critically reviewed the article.

Conflicts of interest

There are no conflicts of interest.


1. UNAIDS. Available at: [Accessed 15 August 2017]
2. UNAIDS. Global AIDS Update 2016. Available at: [Accessed 15 August 2017]
3. WHO. Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection. Recommendations for a public health approach. Available at: [Accessed 15 August 2017]
4. Barth RE, van der Loeff MF, Schuurman R, Hoepelman AI, Wensing AM. Virological follow-up of adult patients in antiretroviral treatment programmes in sub-Saharan Africa: a systematic review. Lancet Infect Dis 2010; 10:155–166.
5. Ajose O, Mookerjee S, Mills EJ, Boulle A, Ford N. Treatment outcomes of patients on second-line antiretroviral therapy in resource-limited settings. AIDS 2012; 26:929–938.
6. WHO. HIV drug resistance report 2017. Available at: [Accessed 16 October 2017]
7. WHO. Guidelines on the public health response to pretreatment HIV drug resistance, July 2017. Available at: [Accessed 16 October 2017]
8. WHO. Global action plan on HIV drug resistance 2017–2021. Available at: [Accessed 16 October 2017]
9. Maina WK, Kim AA, Rutherford GW, Harper M, K’Oyugi BO, Sharif S, et al. KAIS Study Group. Kenya AIDS Indicator Surveys 2007 and 2012: implications for public health policies for HIV prevention and treatment. J Acquir Immune Defic Syndr 2014; 66 (suppl 1):S130–S137.
10. NASCOP. Guidelines on use of antiretroviral drugs for treating and preventing HIV infection in Kenya. 2016 Edition. Available at: [Accessed 3 October 2017]
11. Kantor R, DeLong A, Balamane M, Schreier L, Lloyd RM Jr, Injera W, et al. HIV diversity and drug resistance from plasma and nonplasma analytes in a large treatment programme in western Kenya. J Int AIDS Soc 2014; 17:19262.
12. Wools-Kaloustian K, Kimaiyo S, Diero L, Siika A, Sidle J, Yiannoutsos CT, et al. Viability and effectiveness of large-scale HIV treatment initiatives in sub-Saharan Africa: experience from western Kenya. AIDS 2006; 20:41–48.
13. Hassan AS, Mwaringa SM, Obonyo CA, Nabwera HM, Sanders EJ, Rinke de Wit TF, et al. Low prevalence of transmitted HIV type 1 drug resistance among antiretroviral-naive adults in a rural HIV clinic in Kenya. AIDS Res Hum Retroviruses 2013; 29:129–135.
14. Kiptoo M, Brooks J, Lihana RW, Sandstrom P, Ng’ang’a Z, Kinyua J, et al. HIV-1 drug resistance-associated mutations among HIV-1 infected drug-naive antenatal clinic attendees in rural Kenya. BMC Infect Dis 2013; 13:517.
15. Lihana RW, Khamadi SA, Lubano K, Lwembe R, Kiptoo MK, Lagat N, et al. HIV type 1 subtype diversity and drug resistance among HIV type 1-infected Kenyan patients initiating antiretroviral therapy. AIDS Res Hum Retroviruses 2009; 25:1211–1217.
16. Nyamache AK, Waihenya R, Ng’ang’a ZW, Muigai AW, Khamadi SA. Reverse transcriptase inhibitors drug resistance mutations in drug-naive HIV type 1 positive Kenyans. East Afr Med J 2011; 88:4–8.
17. Crawford KW, Njeru D, Maswai J, Omondi M, Apollo D, Kimetto J, et al. Occurrence of etravirine/rilpivirine-specific resistance mutations selected by efavirenz and nevirapine in Kenyan patients with non-B HIV-1 subtypes failing antiretroviral therapy. AIDS 2014; 28:442–445.
18. Hassan AS, Nabwera HM, Mwaringa SM, Obonyo CA, Sanders EJ, Rinke de Wit TF, et al. HIV-1 virologic failure and acquired drug resistance among first-line antiretroviral experienced adults at a rural HIV clinic in coastal Kenya: a cross-sectional study. AIDS Res Ther 2014; 11:9.
19. Osman S, Lihana RW, Kibaya RM, Ishizaki A, Bi X, Okoth FA, et al. Diversity of HIV type 1 and drug resistance mutations among injecting drug users in Kenya. AIDS Res Hum Retroviruses 2013; 29:187–190.
20. Steegen K, Luchters S, Dauwe K, Reynaerts J, Mandaliya K, Jaoko W, et al. Effectiveness of antiretroviral therapy and development of drug resistance in HIV-1 infected patients in Mombasa, Kenya. AIDS Res Ther 2009; 6:12.
21. Mann M, Diero L, Kemboi E, Mambo F, Rono M, Injera W, et al. Antiretroviral treatment interruptions induced by the Kenyan postelection crisis are associated with virological failure. J Acquir Immune Defic Syndr 2013; 64:220–224.
22. UNAIDS. Office of the President National AIDS Control Council. The Kenya AIDS Epidemic. Available at: [Accessed 16 October 2017]
23. AMPATH. Academic Model Providing Access to Healthcare. Available at: [Accessed 16 October 2017]
24. Swenson LC, Min JE, Woods CK, Cai E, Li JZ, Montaner JS, et al. HIV drug resistance detected during low-level viraemia is associated with subsequent virologic failure. AIDS 2014; 28:1125–1134.
25. Tierney WM, Rotich JK, Hannan TJ, Siika AM, Biondich PG, Mamlin BW, et al. The AMPATH medical record system: creating, implementing, and sustaining an electronic medical record system to support HIV/AIDS care in western Kenya. Stud Health Technol Inform 2007; 129:372–376.
26. DeLong AK, Wu M, Bennett D, Parkin N, Wu Z, Hogan JW, et al. Sequence quality analysis tool for HIV type 1 protease and reverse rranscriptase. AIDS Res Hum Retroviruses 2012; 28:894–901.
27. Shafer R. Stanford HIV Sequence Database. Available at: [Accessed 16 October 2017]
28. Pineda-Pena AC, Faria NR, Imbrechts S, Libin P, Abecasis AB, Deforche K, et al. Automated subtyping of HIV-1 genetic sequences for clinical and surveillance purposes: performance evaluation of the new REGA version 3 and seven other tools. Infect Genet Evol 2013; 19:337–348.
29. Ryscavage P, Kelly S, Li JZ, Harrigan PR, Taiwo B. Significance and clinical management of persistent low-level viremia and very-low-level viremia in HIV-1-infected patients. Antimicrob Agents Chemother 2014; 58:3585–3598.
30. Taiwo B, Gallien S, Aga E, Ribaudo H, Haubrich R, Kuritzkes DR, et al. Antiretroviral drug resistance in HIV-1-infected patients experiencing persistent low-level viremia during first-line therapy. J Infect Dis 2011; 204:515–520.
31. Gonzalez-Serna A, Min JE, Woods C, Chan D, Lima VD, Montaner JS, et al. Performance of HIV-1 drug resistance testing at low-level viremia and its ability to predict future virologic outcomes and viral evolution in treatment-naive individuals. Clin Infect Dis 2014; 58:1165–1173.
32. WHO. Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection. Available at: [Accessed 1 October 2017]
33. Rawizza HE, Chaplin B, Meloni ST, Darin KM, Olaitan O, Scarsi KK, et al. Accumulation of protease mutations among patients failing second-line antiretroviral therapy and response to salvage therapy in Nigeria. PLoS One 2013; 8:e73582.
34. Steegen K, Bronze M, Papathanasopoulos MA, van Zyl G, Goedhals D, Van Vuuren C, et al. Prevalence of antiretroviral drug resistance in patients who are not responding to protease inhibitor-based treatment: results from the first national survey in South Africa. J Infect Dis 2016; 214:1826–1830.
35. Palmer S, Kearney M, Maldarelli F, Halvas EK, Bixby CJ, Bazmi H, et al. Multiple, linked human immunodeficiency virus type 1 drug resistance mutations in treatment-experienced patients are missed by standard genotype analysis. J Clin Microbiol 2005; 43:406–413.
36. Joly V, Descamps D, Peytavin G, Touati F, Mentre F, Duval X, et al. Evolution of human immunodeficiency virus type 1 (HIV-1) resistance mutations in nonnucleoside reverse transcriptase inhibitors (NNRTIs) in HIV-1-infected patients switched to antiretroviral therapy without NNRTIs. Antimicrob Agents Chemother 2004; 48:172–175.
37. Saravanan S, Vidya M, Balakrishnan P, Kantor R, Solomon SS, Katzenstein D, et al. Viremia and HIV-1 drug resistance mutations among patients receiving second-line highly active antiretroviral therapy in Chennai, Southern India. Clin Infect Dis 2012; 54:995–1000.
38. Boender TS, Hamers RL, Ondoa P, Wellington M, Chimbetete C, Siwale M, et al. Protease inhibitor resistance in the first 3 years of second-line antiretroviral therapy for HIV-1 in sub-Saharan Africa. J Infect Dis 2016; 214:873–883.
39. Pujades-Rodriguez M, Balkan S, Arnould L, Brinkhof MA, Calmy A. Treatment failure and mortality factors in patients receiving second-line HIV therapy in resource-limited countries. JAMA 2010; 304:303–312.
40. Johnston V, Fielding K, Charalambous S, Mampho M, Churchyard G, Phillips A, et al. Second-line antiretroviral therapy in a workplace and community-based treatment programme in South Africa: determinants of virological outcome. PLoS One 2012; 7:e36997.
41. Levison JH, Orrell C, Losina E, Lu Z, Freedberg KA, Wood R. Early outcomes and the virological effect of delayed treatment switching to second-line therapy in an antiretroviral roll-out programme in South Africa. Antivir Ther 2011; 16:853–861.
42. Kranzer K, Ford N. Unstructured treatment interruption of antiretroviral therapy in clinical practice: a systematic review. Trop Med Int Health 2011; 16:1297–1313.
43. McIlleron H, Meintjes G, Burman WJ, Maartens G. Complications of antiretroviral therapy in patients with tuberculosis: drug interactions, toxicity, and immune reconstitution inflammatory syndrome. J Infect Dis 2007; 196 (suppl 1):S63–S75.
44. Walters E, Reichmuth K, Dramowski A, Marais BJ, Cotton MF, Rabie H. Antiretroviral regimens containing a single protease inhibitor increase risk of virologic failure in young HIV-infected children. Pediatr Infect Dis J 2013; 32:361–363.
45. Collier D, Iwuji C, Derache A, de Oliveira T, Okesola N, Calmy A, et al. French National Agency for AIDS and Viral Hepatitis Research (ANRS) 12249 Treatment as Prevention (TasP) Study Group. Virological outcomes of second-line protease inhibitor-based treatment for human immunodeficiency virus type 1 in a high-prevalence rural south African setting: a competing-risks prospective cohort analysis. Clin Infect Dis 2017; 64:1006–1016.
46. Maiga AI, Fofana DB, Cisse M, Diallo F, Maiga MY, Traore HA, et al. Characterization of HIV-1 antiretroviral drug resistance after second-line treatment failure in Mali, a limited-resources setting. J Antimicrob Chemother 2012; 67:2943–2948.
47. Van Zyl GU, Liu TF, Claassen M, Engelbrecht S, de Oliveira T, Preiser W, et al. Trends in genotypic HIV-1 antiretroviral resistance between 2006 and 2012 in South African patients receiving first- and second-line antiretroviral treatment regimens. PLoS One 2013; 8:e67188.
48. van Zyl GU, van Mens TE, McIlleron H, Zeier M, Nachega JB, Decloedt E, et al. Low lopinavir plasma or hair concentrations explain second-line protease inhibitor failures in a resource-limited setting. J Acquir Immune Defic Syndr 2011; 56:333–339.
49. Madruga JV, Berger D, McMurchie M, Suter F, Banhegyi D, Ruxrungtham K, et al. TITAN study group. Efficacy and safety of darunavir-ritonavir compared with that of lopinavir-ritonavir at 48 weeks in treatment-experienced, HIV-infected patients in TITAN: a randomised controlled phase III trial. Lancet 2007; 370:49–58.
50. Boyd MA, Kumarasamy N, Moore CL, Nwizu C, Losso MH, Mohapi L, et al. Ritonavir-boosted lopinavir plus nucleoside or nucleotide reverse transcriptase inhibitors versus ritonavir-boosted lopinavir plus raltegravir for treatment of HIV-1 infection in adults with virological failure of a standard first-line ART regimen (SECOND-LINE): a randomised, open-label, noninferiority study. Lancet 2013; 381:2091–2099.
51. Barber TJ, Harrison L, Asboe D, Williams I, Kirk S, Gilson R, et al. UK HIV Drug Resistance Database and UK Collaborative HIV Cohort (UK CHIC) Study Steering Committees. Frequency and patterns of protease gene resistance mutations in HIV-infected patients treated with lopinavir/ritonavir as their first protease inhibitor. J Antimicrob Chemother 2012; 67:995–1000.
52. Grossman Z, Schapiro JM, Levy I, Elbirt D, Chowers M, Riesenberg K, et al. Comparable long-term efficacy of Lopinavir/Ritonavir and similar drug-resistance profiles in different HIV-1 subtypes. PLoS One 2014; 9:e86239.
53. Mo H, King MS, King K, Molla A, Brun S, Kempf DJ. Selection of resistance in protease inhibitor-experienced, human immunodeficiency virus type 1-infected subjects failing lopinavir- and ritonavir-based therapy: mutation patterns and baseline correlates. J Virol 2005; 79:3329–3338.
54. Kempf DJ, King MS, Bernstein B, Cernohous P, Bauer E, Moseley J, et al. Incidence of resistance in a double-blind study comparing lopinavir/ritonavir plus stavudine and lamivudine to nelfinavir plus stavudine and lamivudine. J Infect Dis 2004; 189:51–60.
55. El-Khatib Z, Ekstrom AM, Ledwaba J, Mohapi L, Laher F, Karstaedt A, et al. Viremia and drug resistance among HIV-1 patients on antiretroviral treatment: a cross-sectional study in Soweto, South Africa. AIDS 2010; 24:1679–1687.
56. Fisher R, van Zyl GU, Travers SA, Kosakovsky Pond SL, Engelbrech S, Murrell B, et al. Deep sequencing reveals minor protease resistance mutations in patients failing a protease inhibitor regimen. J Virol 2012; 86:6231–6237.
57. Nijhuis M, van Maarseveen NM, Lastere S, Schipper P, Coakley E, Glass B, et al. A novel substrate-based HIV-1 protease inhibitor drug resistance mechanism. PLoS Med 2007; 4:e36.
58. Rabi SA, Laird GM, Durand CM, Laskey S, Shan L, Bailey JR, et al. Multistep inhibition explains HIV-1 protease inhibitor pharmacodynamics and resistance. J Clin Invest 2013; 123:3848–3860.
59. Gatanaga H, Suzuki Y, Tsang H, Yoshimura K, Kavlick MF, Nagashima K, et al. Amino acid substitutions in Gag protein at noncleavage sites are indispensable for the development of a high multitude of HIV-1 resistance against protease inhibitors. J Biol Chem 2002; 277:5952–5961.
60. Thao VP, Quang VM, Day JN, Chinh NT, Shikuma CM, Farrar J, et al. High prevalence of PI resistance in patients failing second-line ART in Vietnam. J Antimicrob Chemother 2016; 71:762–774.
61. Magambo B, Nazziwa J, Bbosa N, Gupta RK, Kaleebu P, Parry CM. The arrival of untreatable multidrug-resistant HIV-1 in sub-Saharan Africa. AIDS 2014; 28:1373–1374.
62. Ouattara EN, Ross EL, Yazdanpanah Y, Wong AY, Robine M, Losina E, et al. Clinical impact and cost-effectiveness of making third-line antiretroviral therapy available in sub-Saharan Africa: a model-based analysis in Cote d’Ivoire. J Acquir Immune Defic Syndr 2014; 66:294–302.
63. Lorenzana SB, Hughes MD, Grinsztejn B, Collier AC, Luz PM, Freedberg KA, et al. Genotype assays and third-line ART in resource-limited settings: a simulation and cost-effectiveness analysis of a planned clinical trial. AIDS 2012; 26:1083–1093.
64. Kantor R, Katzenstein D, Efron B, Carvalho AP, Wynhoven B, Soares M, et al. Impact of HIV-1 subtype and antiretroviral therapy on protease and reverse transcriptase genotypic evolution: results of a global collaboration. PLoS Med 2005; 2:e112.
65. Wainberg MA, Zaharatos GJ, Brenner BG. Development of antiretroviral drug resistance. N Engl J Med 2011; 365:637–646.
66. Taylor BS, Sobieszczyk ME, McCutchan FE, Hammer SM. The challenge of HIV-1 subtype diversity. N Engl J Med 2008; 358:1590–1602.
67. Huang A, Hogan JW, Luo X, DeLong A, Saravanan S, Wu Y, et al. Global comparison of drug resistance mutations after first-line antiretroviral therapy across Human Immunodeficiency Virus-1 subtypes. Open Forum Infect Dis 2016; 3:ofv158.
68. Kyeyune F, Nankya I, Metha S, Akao J, Ndashimye E, Tebit DM, et al. JCRC Drug Resistance Working Group. Treatment failure and drug resistance is more frequent in HIV-1 subtype D versus subtype A-infected Ugandans over a 10-year study period. AIDS 2013; 27:1899–1909.
69. Roberts T, Cohn J, Bonner K, Hargreaves S. Scale-up of routine viral load testing in resource-poor settings: current and future implementation challenges. Clin Infect Dis 2016; 62:1043–1048.
70. Genberg BL, Naanyu V, Wachira J, Hogan JW, Sang E, Nyambura M, et al. Linkage to and engagement in HIV care in western Kenya: an observational study using population-based estimates from home-based counselling and testing. Lancet HIV 2015; 2:e20–e26.
71. Hill A, Pozniak A. HIV treatment cascades: how can all countries reach the UNAIDS 90-90-90 target?. AIDS 2015; 29:2523–2525.
72. Estill J, Ford N, Salazar-Vizcaya L, Haas AD, Blaser N, Habiyambere V, et al. The need for second-line antiretroviral therapy in adults in sub-Saharan Africa up to 2030: a mathematical modelling study. Lancet HIV 2016; 3:e132–e139.
73. Coetzer M, Ledingham L, Diero L, Kemboi E, Orido M, Kantor R. Gp41 and Gag amino acids linked to HIV-1 protease inhibitor-based second-line failure in HIV-1 subtype A from Western Kenya. J Int AIDS Soc 2017; 20:e25024.

AIDS; drug resistance; HIV; Kenya; second line; treatment failure

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc.