Are changing treatment guidelines responsible for differing rates of transmitted drug resistance?

Could treatment guidelines – which have veered from the aggressive to the conservative between 1997 and 2003 – be indirectly responsible for the transmission of drug-resistant HIV? It's a "provocative hypothesis" says an editorial alongside the study which suggests this might be the case. Both appear in the online edition of the journal, Clinical Infectious Diseases, electronically published on May 31^st.

Carmen de Mendoza from Madrid and her colleagues from nine other cities in Spain collaborated on a study that wished to examine their hypothesis that changes in treatment guidelines may have influenced trends in the proportion of drug-resistant viruses in newly infected individuals.

A total of 141 individuals in ten cities in Spain were identified with evidence of recent HIV infection between January 1997 and December 2003. The researchers defined recent HIV infection as either laboratory evidence of acute primary infection or less than twelve months between a negative and positive HIV antibody test. The majority (67.4%) of exposures had occurred during sex between men, and the median estimated time between exposure and first detection of HIV infection was 7.6 months.

Overall, 14.2% (20/141) were found to have HIV with genotypic evidence of drug resistance. Some resistance to drug from all three classes was seen. However, nine of the 20 had evidence of AZT (zidovudine, Retrovir) resistance only. This suggests, say Susan Little and Davey Smith from the University of California, San Diego, in their accompanying editorial, that the individuals who transmitted these viruses were antiretroviral experienced individuals who took "suboptimal, nonpotent antiretroviral therapy, such as sequential monotherapy, during the earliest days of HIV treatment".

Five of the 20 had resistance mutations to more than one drug class: two had phenotypic resistance to a combination of nucleoside analogue and non-nucleoside reverse transcriptase inhibitors (NNRTIs); three to nucleoside analogues and protease inhibitors (PIs).

The researchers assessed the overall proportions of primary drug resistant mutations by year of infection and found the following:

• 1997 33.3%
• 1998 29.4%
• 1999 20%
• 2000 14.3%
• 2001 3.4%
• 2002 15.4%
• 2003 10.9%

They note, however, that a limitation of their study was the significant year-to-year variation of seroconverters identified in the study. The majority were identified more recently due to the Serological Testing Algorithm for Recent HIV Seroconversion (STARHS) assay which can determine the window period of HIV infection within 133 days, as well as a more active search for recent seroconverters.

Little and Smith point out that due to the "relatively small proportion of [newly infected] patients in this study (0.4%-3.5% of the total)...one must be cautious about conclusions drawn from such a small convenience sample."

These recent seroconverters were then compared with the individuals in these ten cities with undetectable blood plasma viral loads during the same period. This was determined by looking at one blood plasma viral load test per year for a minimum of three years and including all those with tests below 500 copies/ml in 1997 and 1998, and below 50 copies/ml from 1999 onwards. Of the 8388 individuals included in the study, all but 20 were on highly active antiretroviral therapy (HAART), "reflecting indirectly the extent of therapy in this population," de Mendoza and colleagues write.

The researchers assessed the proportion of chronically infected individuals with undetectable blood plasma viral load and found the following:

• 1997 33.4%
• 1998 34.6%
• 1999 39.7%
• 2000 47.5%
• 2001 52.9%
• 2002 39.7%
• 2003 58.1%

In order to make the blood plasma viral load thresholds uniform, the researchers adjusted the 1997 and 1998 proportions downwards by 25%, since prior studies have found that around 75% of individuals with less than 500 copies/ml actually have less than 50 copies/ml.

Spearman's rank correlation test was used to analyse the correlation between the rate of drug-resistance mutations seen in recent seroconverters and the proportion of individuals with undetectable blood plasma viral loads at different time points. A significant inverse correlation was found (r = -0.955; p = 0.001).

The researchers argue that "differences in the proportion of patients with chronic HIV-1 infection for whom HAART failed could explain our findings for recent HIV-1 seroconverters." In 2000, Quinn and colleagues found that no HIV transmission was observed in 90 monogamous, serodiscordant heterosexuals in Uganda when blood plasma viral load was less than 1500 copies/ml. Based on this study, de Mendoza and colleagues suggest that individuals with detectable blood plasma viral loads therefore represent potential transmitters of drug-resistant HIV. "This group includes either patients who are antiretroviral-naive or those for whom therapy is not successful," they write, adding that "the latter is a larger population and is more likely to carry drug-resistant strains."

They then go on to link the proportion of individuals with undetectable blood plasma viral load with contemporaneous treatment guidelines, which were "relatively more aggressive" between 1997-2001, but due to concerns over drug toxicity in 2002 became "more conservative" leading to "widespread use of drug holidays". In 2003, however, which saw the highest proportion of individuals with undetectable blood plasma viral loads, concerns over treatment interruptions, as well as the widespread use of lopinavir/ritonavir ( Kaletra) and tenofovir (Viread), "seemed to revert the situation."

They conclude by asserting that "antiretroviral therapy has benefits for people other than those who are receiving treatment...and this should be emphasized, particularly in the light of recent increased risk behaviours among high-risk groups that result from a misperception that HIV-1 infection is no longer a deadly illness."

However, Little and Smith state that "it has not yet been demonstrated on a population level that antiretroviral treatment reduces the risk of HIV transmission." This is due to factors such as the recent discovery that undetectable blood plasma viral load in blood does not necessarily mean undetectable viral load in genital fluids, and that drug resistance can be found in genital fluids and not in blood – and vice versa.

Little and Smith cite three separate studies that found no significant difference in mean blood plasma viral load levels between individuals on HAART and antiretroviral-naive individuals. "These data suggest," they say, "that the mean virus load amongst many different treated populations (i.e. those treated with aggressive or conservative regimens) and untreated populations are comparable.

"This dampens the optimism that antiretroviral treatment with available therapy could reduce the pool of potential transmitters in a population, and therefore, be an effective prevention strategy".

They conclude their editorial by suggesting that "the rate of transmitted drug resistance is influenced by a complex interaction of multiple variables, including changing rates of high-risk behavior, changing use and efficacy of antiretroviral therapy, and the relative transmission fitness of different viral variants."

Although de Mendoza and colleagues "offer a provocative hypothesis to explain how treatment efficacy may influence the transmission of drug resistance...it is premature to alter current treatment practices solely on the basis of the desire to limit the transmission of drug-resistant HIV."