Investigating the effects of treatment in HIV disease models

Van der Berg, Anden (2021-12)

Thesis (MSc)--Stellenbosch University, 2021.

Thesis

ENGLISH ABSTRACT: HIV/AIDS and disease response to antiretroviral (ARV ) drugs are of major importance to the developing world, and the disease remains a burden on society, as viral replication still needs to be controlled continually in persons infected with HIV. The ever-increasing prevalence of drug-resistant viral strains and the latent reservoir which harbours dor- mant virus also remain barriers to a cure. To overcome these barriers, novel ways of treating the disease and new tools for effective and efficient drug development are re- quired. The use of mathematical models of disease and drug treatment continues to grow and remains an essential tool in drug development and the search for a cure. In this study, combined HIV disease-PKPD models are created and tested for their abil- ity to simulate real world patient data. First, independent mathematical models of HIV disease dynamics and ARV pharmacology are reproduced from literature. The effect of patient variability on simulation results is tested using Monte Carlo simulations, in which parameters are varied within biologically relevant ranges. The HIV disease models are then linked to PK and PD models of the currently prescribed ARVs. Monte Carlo sim- ulations are used to examine the effect that heterogeneity, model structure, and model assumptions have in the newly linked models. The viral load and CD4+ T-cell count predictions made by the combined models are compared to clinical patient data from the Western Cape, South Africa. Analysis of the combined models show that model struc- ture has to include latently infected cells and drug-resistant viral strains to be able to accurately predict the disease progression of HIV. Models need to incorporate the mechanisms that affect disease outcome. In the context of HIV, this may include drug-resistant strains and the effect of long-lived latently infected cells. Model predictions can be improved by including these mechanisms which have an impact on disease outcome and by considering longitudinal patient datasets. Such continual improvements will aid in making models powerful diagnostic tools.

AFRIKAANSE OPSOMMING: MIV/VIGS en siekterespons op antiretroviale medikasie is van uiterste belang vir die ontwikkelende wêreld; die virussiekte plaas druk op die samelewing aangesien virale replisering in ’n MIV-positiewe persoon deurlopend beheer moet word. Die immertoen- emende algemeenheid van medikasiebestande virale stamme, sowel as die latente reser- voir wat dormante virus huisves, bly steeds ’n hindernis vir genesing. Nuwe maniere om die virussiekte te behandel en nuwe instrumente vir die ontwikkeling van effektiewe en doeltreffende medikasie is broodnodig ten einde hierdie hindernisse te oorkom. Die gebruik van wiskundige modelle van siekteverloop en behandeling groei toenemend en bly ’n onontbeerlike instrument in die ontwikkeling van medikasie en soeke na ’n ge- neesmiddel. Vir die doeleinde van hierdie studie is saamgestelde MIV PKPD-siektemodelle geskep en getoets vir hul vermoë om werklike pasiëntedata te simuleer. Eerstens is onafhank- like wiskundige modelle van MIV-siektedinamika en ARV-farmakologie uit vaklektuur geskep. Die invloed van pasiëntveranderlikheid op simulasies, waarin parameters binne biologiestoepaslike reikwydtes wissel, word deur Monte Carlo-simulering getoets. Die MIV-siektemodelle word dan aan PK- en PD-modelle van tansvoorgeskrewe ARVs gekop- pel. Monte Carlo-simulering word gebruik om die invloed van heterogeniteit, model- struktuur en modelaannames in die gekoppelde modelle te toets. Die gekombineerde modelle se virale lading- en CD4+ T-seltelling-voorspellings word dan met kliniese pasiënte- data in die Wes-Kaap van Suid-Afrika vergelyk. Ontleding van die saamgestelde modelle toon dat modelstruktuur, latent-geïnfekteerde selle en medikasiebestande virusstamme moet insluit ten einde die siekteverloop van MIV presies te voorspel. Modelle moet die meganismes wat siekte-uitkoms beïnvloed, bevat. In die geval van MIV kan dit medikasiebestande virusstamme en die invloed van langlewende latent- geïnfekteerde selle insluit. Modelvoorspellings kan verbeter word deur dié meganismes - wat ’n invloed op siekte-uitkoms het - sowel as longitudinale pasiëntdatastelle, in te sluit. Sulke deurlopende verbeteringe sal daartoe bydra om van modelle kragtige diagnostieke instrumente te maak.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/124262
This item appears in the following collections: