Genetic markers linked to early discontinuation of three antiretrovirals

Mark Mascolini,

Pharmacogenetic markers that purportedly signal antiretroviral side effects predicted discontinuation of atazanavir, efavirenz, and tenofovir (but paradoxically not abacavir) within the first year of treatment in the Swiss HIV Cohort Study (SHCS). [1]

Sara Colombo and colleagues cautioned that gender and ethnicity also correlated with stopping antiretrovirals and complicate interpretation of their results. But they plan a randomised study to see if monitoring patients for toxicity-related genetic shifts can improve antiretroviral care.

SHCS investigators used Veracode technology to search for 13 pharmacogenetic markers thought to signal antiretroviral toxicity on 9 genes. They targeted markers linked to (1) central nervous system toxicity with efavirenz, (2) Gilbert syndrome with atazanavir, (3) cardiovascular disease with lopinavir, (4) renal proximal tubulopathy with tenofovir, and (5) hypersensitivity to abacavir. The abacavir hypersensitivity marker they focused on was not HLA-B*5701 [2], but the HCP5 gene shift designated rs2395029.

The study involved 577 treatment-naive people starting their first antiretroviral regimen from 2004 through 2008. This group was three quarters male (73%) and white (79%), though 13% were black Africans. Within the first year of treatment, 190 people (33%) stopped one of more antiretrovirals because of toxicity. For tenofovir, efavirenz, lopinavir, and atazanavir, people with risky genes stopped each of these drugs substantially more often than people without the genetic risk markers:

  • Tenofovir: 500 patients, 28.6% at risk stopped versus 13.6% not at risk
  • Efavirenz: 272 patients, 67.6% at risk stopped versus 27.4% not at risk
  • Lopinavir: 184 patients, 51.4% at risk stopped versus 32.6% not at risk
  • Atazanavir: 121 patients, 62.5% at risk stopped versus 18.9% not at risk

The HCP5 genetic marker was not a good predictor of quitting abacavir, perhaps because the study considered stopping abacavir for any reason, not just the hypersensitivity reaction, and because people in Switzerland started getting HLA-B*5701 screening before beginning abacavir during this period.

For the other four drugs, hazard ratios adjusted for other risk factors found that genetic markers independently predicted stopping efavirenz (adjusted hazard ratio [aHR] 3.10, 95% confidence interval [CI] 1.48 to 6.46, p=0.0259 and atazanavir (aHR 7.31, 95% CI 2.86 to 18.72, p<0.0001. There was a strong trend toward an independent effect of genetic markers on quitting tenofovir (aHR 2.30, 95% CI 0.99 to 5.31, p=0.052) but not lopinavir (aHR 1.42, 95% CI 0.62 to 3.25, p=0.41).

Gender had a big impact on genetic risk for efavirenz-related central nervous system toxicity. Among women, 80% of those with a risk marker versus 42.5% of those without a risk marker stopped efavirenz. Respective rates for men were 50% and 24.3%. The SHCS investigators noted that certain genetic markers they analysed are more common in nonwhites and that most of the nonwhites in this group were women.

Colombo and colleagues believe their study “highlights the interest of conducting a prospective clinical trial of pharmacogenetics-driven choice of first-line antiretroviral therapy.” They plan such a trial in which clinicians planning a first-line regimen are randomised to receive or not receive antiretroviral advice based on analysis of each first-line patient’s toxicity risk markers. At this point, though, it is not clear which markers have a negative predictive value high enough to reliably warn physicians away from using a certain drug. HLA-B*5701 has a 100% negative predictive value for hypersensitivity to abacavir. [2]


  1. Colombo S et al. Association of pharmacogenetic markers with premature discontinuation of first-line ART.. 10th International Workshop on Clinical Pharmacology of HIV Therapy, April 15-17, 2009, Amsterdam. Abstract O03.
  2. Mallal S et al. HLA-B*5701 screening for hypersensitivity to abacavir. N Engl J Med. 2008;358:568-579.

Links to other websites are current at date of posting but not maintained.