Stable estimation of break-points for change in dosing regimen of fixed dose combinations (FDC) of anti-tuberculosis drugs for children
Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
Introduction: Treatment of tuberculosis (TB) is complicated by the need for many different drugs and the long duration of treatment (usually about 6 months). FDC:s are recommended by WHO since it has been shown that it among other things increases compliance. Despite long-time use for adults, are there no anti-TB FDC for children yet. Population modeling can estimate doses to be used in FDC and optimize individualization. Earlier work has found that estimations are unstable when trying to assess the breakpoints (BP) for change in dose.
Aim: To evaluate performance of two different types of functions mimicking a discontinuous break-point using the estimation methods FOCE and SAEM and to evaluate the impact on stability based on the structure of the dataset.
Methods: NONMEM 7.2 was used. Based on NHANES data, 12 different datasets with weight (BW) data for children 2-15 years were simulated. Clearances (CL) of 4 commonly used anti-TB drugs were taken from existing popPK models and allometrically scaled. The estimation methods FOCE and SAEM were used and the break point functions evaluated were Eq. 1: STEP=BWGAM/(BWGAM + BPGAM) and Eq. 2: STEP=eGAM*BW/(eGAM*BW + eGAM*BP). The value of the GAM constant was varied between 50-150 for Eq. 1 and 3-50 for Eq. 2.
Results: The stability of the estimation depends mostly on size of the dataset, distribution of data and on value of GAM. Eq. 1 proved to be slightly more stable but Eq. 2 made cleaner cuts at the BP, i.e. fewer impossible doses (the doses in between the estimated dose levels). Due to technical difficulties and limited time, estimation with SAEM could not be performed.
Conclusions: Using FOCE and Eq. 2 seems to be the best of the tested options both because it made cleaner cuts and because it does not have the variable (BW) as base. The small difference in stability seems to be possible to overcome by finding an optimal GAM.
Place, publisher, year, edition, pages
2012. , 28 p.
Medical and Health Sciences
IdentifiersURN: urn:nbn:se:uu:diva-185664OAI: oai:DiVA.org:uu-185664DiVA: diva2:572319
Subject / course
Master of Science Programme in Pharmacy
Kjellsson, MariaKarlsson, Mats
Hammarlund-Udenaes, Margareta, Professor