Treatment interruptions to decrease risk of resistance emerging during therapy switching in HIV treatment. Ryan Zurakowski* Dominik Wodarz Department of Electrical and Computer Engineering Department of Ecology and Evolutionary Biology 140 Evans Hall Steinhaus Hall University of Delaware University of California Newark, DE 19713 Irvine, CA 92697 ryanz@udel.edu dwodarz@uci.edu Abstract— The development of multi-drug resistant strains of HIV remains the primary reason for treatment failure and progression to AIDS in the United States. The failure of a particular multi-drug regimen necessitates a switch to a new multi-drug regimen. We use a simple model of the interaction of resistant strains to show that the transition to the new regimen involves a significant risk of strains resistant to the new regimen emerging, and that treatment interruptions using the failing regimen can be used to decrease this risk. I. INTRODUCTION The development of resistance to a particular drug regimen necessitates a change in regimen, and there are, for every ap- plication, only a limited number of drug regimens available. If resistance develops to every available drug regimen, the patient will die. It is critical, therefore, to switch regimens in a manner that minimizes the chance of strains resistant to the new regimen emerging. Human Immunodeficiency Virus (HIV) is a widespread retrovirus which infects the CD4+ T-Cell. Long-term un- treated infection leads to severe immunodeficiency and death from opportunistic infections. The reverse-transcriptase me- diated life-cycle of HIV is highly error-prone, resulting in a high mutation rate and fast viral evolution. Coupled with a very high replication rate, this means that HIV will rapidly evolve resistance to any single-drug antiviral regimen [1]. The solution is the use of Highly Active Anti-Retroviral Therapy (HAART), which uses three drugs which attack dif- ferent viral epitopes. This increases the mutational barrier to a degree which makes drug resistance theoretically unlikely to emerge. Although adherence to a successful HAART regimen makes the emergence of resistant strains of the virus unlikely, it happens nevertheless. The likely causes of this are either pre-existance or poor adherence to the treatment regimen [2], [3]. The development of resistance to a particular drug regimen necessitates a change in regimen, and the new regimen must consist entirely of drugs for which there is no cross-resistance with the current regimen. There are, for every application, only a limited number of drug regimens available [4]. If resistance develops to every available drug regimen, the patient will die. It is critical, therefore, to switch *Corresponding Author regimens in a manner that minimizes the chance of strains resistant to the new regimen emerging. The importance of maintaining the availability of effective drug regimens has lead to a number of studies in reducing the risk of resistance emerging during therapy, usually by attempting to improve adherence. These include the two arms of the STACCATO study [5], [6], one of which proposed to reduce drug toxicities by pursuing an 1-week-on 1-week- off drug suppression schedule, supposedly reducing drug- related toxicities and improving overall adherence. However, this arm was discontinued due to a dramatic increase in resistance mutations in the experimental group. The other arm involved longer interruptions, driven by viral load and CD4+ T-Cell counts. This arm saw no increase in resistance mutations and an increase in overall adherence corresponding to a decrease in drug-related toxicities. The studies reported in [7], [8], [9], [10], [11] attempted to improve the virological response to therapy in patients with extensive multi-drug resistance through various types of therapy interruptions, with little to no effect. An interesting study involving period- ically switching HAART regimens at fixed intervals yielded promising results in reducing the rate of resistance emerging (the SWATCH study [12]). This study was prompted by a stochastic model which predicted this reduced risk of resis- tance [13]. However, the long term benefit of this approach can only be verified with a longer and larger study. In the case where a particular regimen has failed, signifi- cant research has been done on how to choose a new regimen, based on the probability of cross resistance between the new regimen and the failing regimen. However, no research has been done on the effect of the timing of the therapy switch on the possibility of resistance emerging to the new regimen. Current switching recommendations either suggest switching regimens as soon as resistant virus is detected, or waiting until a particular disease marker (either viral load increase or CD4 T-Cell count decrease) is reached [4]. In this paper, we show that a pattern of structured treatment interruptions using the failing regimen preceding the introduction of the new regimen can significantly decrease the risk of resistance emerging to the new regimen. This paper is organized as follows: In Section II, we introduce a basic model of competition between wild-type Proceedings of the 46th IEEE Conference on Decision and Control New Orleans, LA, USA, Dec. 12-14, 2007 FrB01.3 1-4244-1498-9/07/$25.00 ©2007 IEEE. 5174