Journal of Pharmacokinetics and Pharmacodynamics

, Volume 38, Issue 6, pp 727–742

Non-linear mixed effects modeling of antiretroviral drug response after administration of lopinavir, atazanavir and efavirenz containing regimens to treatment-naïve HIV-1 infected patients

  • Daniel Röshammar
  • Ulrika S. H. Simonsson
  • Håkan Ekvall
  • Leo Flamholc
  • Vidar Ormaasen
  • Jan Vesterbacka
  • Eva Wallmark
  • Michael Ashton
  • Magnus Gisslén
Article

DOI: 10.1007/s10928-011-9217-1

Cite this article as:
Röshammar, D., Simonsson, U.S.H., Ekvall, H. et al. J Pharmacokinet Pharmacodyn (2011) 38: 727. doi:10.1007/s10928-011-9217-1

Abstract

The objective of this analysis was to compare three methods of handling HIV-RNA data below the limit of quantification (LOQ) when describing the time-course of antiretroviral drug response using a drug-disease model. Treatment naïve Scandinavian HIV-positive patients (n = 242) were randomized to one of three study arms. Two nucleoside reverse transcriptase inhibitors were administrated in combination with 400/100 mg lopinavir/ritonavir twice daily, 300/100 mg atazanavir/ritonavir once a day or 600 mg efavirenz once a day. The viral response was monitored at screening, baseline and at 1, 2, 3, 4, 12, 24, 48, 96, 120, and 144 weeks after study initiation. Data up to 400 days was fitted using a viral dynamics non-linear mixed effects drug-disease model in NONMEM. HIV-RNA data below LOQ of 50 copies/ml plasma (39%) was omitted, replaced by LOQ/2 or included in the analysis using a likelihood-based method (M3 method). Including data below LOQ using the M3 method substantially improved the model fit. The drug response parameter expressing the fractional inhibition of viral replication was on average (95% CI) estimated to 0.787 (0.721–0.864) for lopinavir and atazanavir treatment arms and 0.868 (0.796–0.923) for the efavirenz containing regimen. At 400 days after treatment initiation 90% (76–100) of the lopinavir and atazanavir treated patients were predicted to have undetectable viral levels and 96% (89–100%) for the efavirenz containing treatment. Including viral data below the LOQ rather than omitting or replacing data provides advantages such as better model predictions and less biased parameter estimates which are of importance when quantifying antiretroviral drug response.

Keywords

Antiretroviral HIV-RNA Limit of quantification NONMEM 

Copyright information

© Springer Science+Business Media, LLC 2011

Authors and Affiliations

  • Daniel Röshammar
    • 1
    • 8
  • Ulrika S. H. Simonsson
    • 2
  • Håkan Ekvall
    • 3
  • Leo Flamholc
    • 4
  • Vidar Ormaasen
    • 5
  • Jan Vesterbacka
    • 6
  • Eva Wallmark
    • 4
  • Michael Ashton
    • 1
  • Magnus Gisslén
    • 7
  1. 1.Department of PharmacologyThe Sahlgrenska Academy at University of GothenburgGöteborgSweden
  2. 2.Department of Pharmaceutical BiosciencesUppsala UniversityUppsalaSweden
  3. 3.Department of Infectious DiseasesSundsvall HospitalSundsvallSweden
  4. 4.Department of Infectious DiseasesMalmö University HospitalMalmöSweden
  5. 5.Department of Infectious DiseasesUllevål University HospitalOsloNorway
  6. 6.Department of Infectious DiseasesKarolinska Institutet, Karolinska University HospitalStockholmSweden
  7. 7.Department of Infectious DiseasesThe Sahlgrenska Academy at University of GothenburgGöteborgSweden
  8. 8.AstraZeneca R&DMölndalSweden

Personalised recommendations