Very high levels of drug resistance seen in people experiencing first-line antiretroviral therapy failure in Kenya

Future treatment options severely limited in the quarter who fail therapy

A cross-sectional survey of viral load in patients attending the largest HIV clinic in Kenya has shown that after an average of nearly two years on first-line, tenofovir-based antiretroviral therapy (ART), 76% maintained a viral load under 40 copies/ml and 85% under 1000 copies/ml, which is the World Health Organization threshold for significant clinical and transmission consequences.

The study also found that tenofovir-based ART was more successful in patients who had been on previous ART regimens and switched to a tenofovir-based one; 93% of them had viral loads under 40 copies/ml and 99% under 1000 copies/ml.

However, the study found extremely high rates of significant drug resistance in those whose viral load was detectable. Eighty-nine per cent of first-line patients tested had mutations conferring resistance to drugs in at least one class of antiretrovirals and 83% to two or more classes. The K65R mutation, which confers resistance to all drugs of the NRTI class except zidovudine (AZT), was present in more than two-thirds of patients – considerably in excess of its prevalence in patients tested for resistance in higher-income settings.


first-line therapy

The regimen used when starting treatment for the first time.

detectable viral load

When viral load is detectable, this indicates that HIV is replicating in the body. If the person is taking HIV treatment but their viral load is detectable, the treatment is not working properly. There may still be a risk of HIV transmission to sexual partners.


In HIV, different strains which can be grouped according to their genes. HIV-1 is classified into three ‘groups,’ M, N, and O. Most HIV-1 is in group M which is further divided into subtypes, A, B, C and D etc. Subtype B is most common in Europe and North America, whilst A, C and D are most important worldwide.

drug resistance

A drug-resistant HIV strain is one which is less susceptible to the effects of one or more anti-HIV drugs because of an accumulation of HIV mutations in its genotype. Resistance can be the result of a poor adherence to treatment or of transmission of an already resistant virus.

treatment failure

Inability of a medical therapy to achieve the desired results. 

The resistance patterns seen effectively ruled out future treatment success with all available NNRTI drugs as well as all NRTI drugs except zidovudine.

The high resistance rates seen underline the importance of adopting viral load testing more widely in low-income settings, as the biggest contributor to resistance was that patients may have been on failing therapy for up to two years before failure was detected.

The fact that average CD4 cell counts were lower in patients failing ART, and even lower in patients with viral loads high enough to test for resistance, show that while CD4 monitoring may still have a part to play in flagging up treatment failure, the lack of a means to detect virological failure has real clinical consequences.

The study in detail

The study took a sample of 333 patients from the nearly 27,000 patients treated at Kenya’s largest HIV centre, the Moi Teaching and Referral Hospital in the city of Eldoret, which is itself just part of the AMPATH (Academic Model Providing Access to Healthcare) programme tjhat treats over 140,000 people.

Patients were eligible for the study if they were over 18, had been on tenofovir-based ART for more than six months, and reported more than 50% adherence.

Of the 333 patients enrolled, 217 were on their first, tenofovir-based regimen: all were also taking lamivudine (3TC). The third, NNRTI drug was efavirenz in 60% of cases, and nevirapine in 40%.

The remainder had been on prior regimens. Nearly all of them had previously taken stavudine (d4T), 80% nevirapine, a third efavirenz and a quarter zidovudine.

The average length of time on tenofovir was two years; people who had switched regimens had been on any ART for an average of six years. Not surprisingly those on prior ART were on average older than those new to it (44 versus 40 years) and had higher CD4 counts (426 versus 298 cells/mm3). Fifty-five per cent were women and this did not differ between the treatment groups.

When tested for their viral load, 59 of the 333 patients (18%) had a detectable viral load of over 40 copies/ml and 10% over 1000 copies/ml. Treatment failure was considerably more common in patients new to ART than in ones who switched – 27% of people on first-line ART had detectable viral loads and 15% viral loads over 1000 copies/ml, compared with 7% and 1% of patients who had switched from previous regimens.

Patients failing ART were somewhat younger on average (38 years) and a slightly higher proportion were women (59%). They also had considerably lower than average CD4 counts (211 cells/mm3) – not surprising, as they could have been on a failing regimen for anything between six months and two years.

Of the 59 patents failing ART, resistance tests were available from 35 of 51 patients on first-line therapy and two of eight patients who had switched. The main reason resistance tests were not performed was because of low viral load; the average in patients not tested for resistance was only 111 copies ml, though two patients who did not have resistance tests had viral loads over 35,000 copies/ml.

Of the patients that did have resistance tests, the researchers do not give figures for the two with prior ART experience, but of the 35 on first-line therapy, the mean viral load was nearly 160,000 copies/ml; only five (14%) had viral loads under 1000 copies/ml and two had viral loads over one million copies/ml. The average CD4 count in this group was only 128 cells/mm3.

Thirty-one of these 35 patients had drug resistance – 89%. All of them had resistance to the NNRTI class, and 83% to both NRTIs and NNRTIs – in other words 97% of those with any resistance had resistance to two classes of drug. Twenty-seven (77%) had the M184V resistance mutation to lamivudine and emtricitabine, and 69% the K65R mutation that confers resistance to tenofovir and to all other NRTIs except zidovudine.

A phenotypic resistance test showed that 83% of these 35 patients had clinically significant resistance to lamivudine/emtricitabine, 71% to tenofovir, 86% to efavirenz and 89% to nevirapine. In addition 77% had resistance to etravirine and rilpivirine, two second-generation NNRTI drugs that they had not even taken yet, and 77% to abacavir, a drug often used as an alternative to tenofovir. Essentially, the only drug from either of the reverse transcriptase inhibitor (RTI) classes that still worked was zidovudine, to which only 6% were resistant.


This study is a mixture of good and bad news. The high levels of viral suppression and high CD4 counts in patients who had switched therapy to tenofovir is very encouraging, and shows that there is nothing intrinsic to tenofovir-based regimens that leads to failure. The fact that 70% of switching patients had lipoatrophy and 15% either neuropathy, anaemia or both when they switched shows the necessity of newer and less toxic ART drugs. The suppression rate of 73% on patients new to ART is rather low, however. And the really bad news is the almost universal occurrence of treatment-limiting drug resistance in those who did fail on first-line ART.

The primary reason for such high resistance rates must be because patients had suboptimal adherence for long periods of time on their regimen, though because the patients had not had a viral load test before, we do not know the exact length of time they were left unmonitored with a detectable viral load. Patients with resistance on first-line therapy had been on ART on average for 15 months, and some or even most may never have achieved viral suppression.

There may be additional reasons for the high resistance levels seen. There is some evidence that HIV subtype B, the type most common in high-income settings, less easily develops the K65R mutation to tenofovir than other subtypes: in this study the predominant subtype was A. Prior resistance to thymidine analogues like zidovudine and stavudine (so called TAMs) may actually protect against the development of K65R, and heavily pre-treated patients in the first generation in high-income settings have a lot of TAMs. Conversely, in some HIV subtypes there may be a synergy between the common NNRTI mutations and K65R – together, they may add to viral fitness.

The main lesson to be drawn, however, is that even though AIDS-related deaths have been halved in Africa since their peak, without regular viral load monitoring, patients will continue to be left on suboptimal regimens or to struggle with adherence without support, and this considerably jeopardises the likelihood of their being able to benefit from second-line regimens.


Brooks K et al. Treatment failure and drug resistance in HIV-positive patients on tenofovir-based first-line antiretroviral therapy in western Kenya. JIAS 19(1):20798. See full text here.