Is HIV drug resistance in decline in Europe?

Three studies from Europe presented to the Eighth Glasgow International Congress on Drug Therapy in HIV Infection suggest that HIV drug resistance, especially to several drug classes, is in decline, largely due to improved treatment efficacy.

However researchers and activists have drawn different conclusions about what this means for the future of drug development.

Jurgen Vercauteren of the Rega Institute at Leuven in Belgium analysed all 3093 HIV isolates from drug experienced patients that were included in the Portuguese Resistance Database between July 2001 and March 2006. These 3093 samples of HIV came from 2373 patients.

Vercauteren defined HIV with ‘multi-drug resistance’ (MDR) as virus whose genotype indicated susceptibility to no more than one HIV antiretroviral, with the exceptions of enfuvurtide (T-20), tipranavir and darunavir.

An approximately equal number of viral isolates was sent to the Database each year, but during the study period the proportion that were multi-drug-resistant declined from 5.7 per cent in 2001-2 (33 out of 576 isolates) to 2.1 per cent in 2005-6 (13 out of 490 isolates). The decline was steady, with an average 20% decrease per year, and highly statistically significant (p=0.003).

The two factors associated with having MDR virus were length of treatment (16% increase in likelihood per year on treatment) and having an uncertain date as to the start of therapy, which increased the chance of having MDR-virus nearly eightfold. This was taken as Vercauteren as being a ‘surrogate marker’ for having had early and suboptimal ARV treatment.

Vercauteren said that though his findings only showed that the incidence of MCR HIV had declined in one country – Portugal – he expected it to be replicated throughout Europe, telling conference delegates “it is up to you to show us that”.

He said that his findings reflected the increasing efficacy of HAART and said that in the future new HIV drugs’ ability to tackle MDR HIV might become less important compared with ease of use, tolerability and lack of toxicity.

Vercauteren’s findings were questioned by some treatment activists. It is unlikely that in a country with the highest HIV prevalence in western Europe only 3093 resistance tests have been performed since 2001 and the Database may not reflect reality. It is also important to note that the incidence of drug resistance is much easier to measure than its prevalence; resistance will remain archived in patients on successful HAART and may give trouble in the future, but will not show up as currently active resistance.

It is also the case that some of the drugs now increasingly used as first-line therapy, such as the boosted protease inhibitors, are chosen precisely because of a high resistance threshold and were initially designed as salvage therapies. This is still clearly a desirable attribute of any new therapy.

Falling resistance tracks prescribing practice

However Vercauteren’s findings were backed up by two poster presentations on resistance, also from southern Europe.

A study from the HIV clinic at the Catholic University of Rome (Di Giambenedetto) looked at the incidence of resistance in patients tested for it between 1999 and 2005.

The number of patients treated at the clinic increased from 494 in 1999 to 1429 in 2005, while the number with at least one viral load over 1,000 copies/ml declined during the same period from 59.5% to 9.4%.

Of these patients, the proportion genotyped for drug resistance increased from 35% in 1999 to 60% in 2005 – one might question why they were not all genotyped. This means that the actual number of genotypes performed declined from just over 100 in 1999 to 80 in 2005.

During this period the number of genotypes that showed any resistance mutation declined from 95% of samples to 55%, meaning that 45% of failed patients in 2005 did not fail because of drug resistance.

The percentage of treated patients regarded as having detectable viral loads due to drug resistance declined as follows:

• Any class, 57% in 1999 to 10% in 2005
• NRTI resistance, 53% in 1999 to 10% in 2005
• PI resistance, 39% in 1999 to 4% in 2005
• NNRTI resistance, 20% in 1999 to 1% in 2005.

Nearly all genotyped patients that had resistance had some nucleoside (NRTI) resistance, and this closely tracked total resistance, with 89% of genotypes showing NRTI resistance in 1999 and 50% in 2005.

The proportion of samples with resistance to non-nucleosides (NNRTIs) initially increased from 32% in 1999 to around 47% on the 2000-02 period. However it then declined to 32% in 2003-4 and 12% in 2005.

Protease inhibitor resistance showed a gentler decline from being present in 74% of samples in 2001 to 47% in 2003 and 27% in 2005.

Of note, both in terms of the total number of patients with resistance and the proportion of genotyped samples that showed evidence of resistance was the sudden and steep decline between 2004 and 2005. The proportion of genotyped samples with resistance declined from 82% in 2004 to 55% in 2005, with NRTI resistance from 81% to 50%, and NNRTI resistance from 32% to 12%.

Dr Di Giambennedeto ascribes this to changes in prescribing practice. The proportion of patients taking thymidine analogues (AZT or d4T) declined slowly from 90% in 1999 to 75% in 2005 and then suddenly to 41% in 2004.

Prescriptions for non-thymidine NRTIs (tenofovir, abacavir and ddI) increased from 22% to 49% in the same period. Non-boosted PIs were taken by 87% of patients in 1999 but declined to 16% in 2005. Concomitantly, boosted PI prescriptions increased from 1% to 71%.

Prescribing practice clearly favours boosted PIs over NNRTIs in Italy, because during the same period the proportion of patients on NNRTIs declined from 32% to 22%. This may be the reason for the generally dramatic decline in resistance, as studies from the UK, where NNRTIs are preferred for first-line therapy, indicate that although resistance in general may have undergone a modest decline, the proportion of resistance that is due to NNRTIs has increased.

In terms of specific mutations, the M184V 3TC/FTC mutation declined from being present in 53% of samples in 1999 to 22% in 2005 and thymidine analogue mutations of the kind known as ‘type one’ (which are difficult to salvage) declined from around 50% to 30%.

The L74V abacavir/ddI mutation increased from 3.8% of samples to 9.5%, but this was not statistically significant: the K65R tenofovir mutation remained unchanged at about 2-4% of samples, and the Q151M multi-nucleoside resistance mutation also stayed the same at about 1%-3% of samples.

A third study (Hernandez) did not look at resistance genotypes as such but did look at the ‘proportion of patients with therapeutic failure susceptible to resistance testing’, i.e. with viral loads over 1000. This proportion declined from 129 out of 1201 patients in 2000 (10.7%) to 41 out of 1204 patients in 2005 (3.4%). The proportion of failing patients who were new, i.e. experiencing first-line failure declined from around 60% of failures in 2000 to around 40% of failures in 2004-5.