Busulfan is a bifunctional alkylating agent used in combination with other chemotherapeutics for the ablation of dysfunctional bone marrow prior to haematopoietic stem cell transplantation. Busulfan use is complicated by a large inter-individual and inter-occasion pharmacokinetic variability. Furthermore, the exposure of busulfan (estimated as a cumulative area under the curve) has been associated with dose-limiting toxicities such as sinusoidal obstruction syndrome, which can lead to multi-organ failure and death if left untreated.
The research presented in this thesis retrospectively explores SOS incidence in 337 HSCT patients over an eleven-year study period (2006 – 2017) across seven institutions in Australia. Out of 344 busulfan-based conditioning occasions, there were 64 cases of SOS reported. A population pharmacokinetic analysis was developed to explore the pharmacokinetic variability present. Association studies were conducted post hoc and non-parametric and Cox proportional hazards models were developed to better understand the development of SOS post-busulfan use. Lastly, an exploration of the influence of patient genotypes on clinical outcome was conducted.
A total of 3241 observations informed the selection of a one-compartment pharmacokinetic model. Inter-occasion and inter-individual variability were characterized for both clearance and volume. Adjusted-ideal bodyweight (kg) and a sigmoidal Emax, maturation function were incorporated as covariates. A post hoc analysis into concomitant medications found a significant decrease in busulfan CL for patients co-administered metronidazole (difference in median CLNORM = 0.05 L/h/kg, n = 17, P < 0.0001), and a marginal increase with dexamethasone (difference in median CLNORM = 0.01 L/h/kg, n = 49, P < 0.01).
Overall, cAUC was not significantly associated with SOS, although median Cmax observed on Day 1 of busulfan therapy was higher in patients with SOS (2 μg/mL vs. 2.61 μg/mL, P < 3.7 x 10-5). A multivariate Cox proportional hazard model characterized the hazard of developing SOS as a combination of risk factors: low AIBW, low pre-transplant albumin, younger age and high Cmax.
A linear regression analysis on an ADME panel of 67 SNPs in a sub-cohort (223 patients) found found no significant effect of SNPs on busulfan clearance. A logistic regression analysis of the same number of patients over the panel of enzyme-related SNPs failed to report any significant genetic associations with SOS. A separate analysis of 189 patients for polymorphisms of GSTA1 (an enzyme previously associated with busulfan metabolism) showed a 14% – 18% lower clearance (difference in mean CLNORM = 0.03 L/h/kg, P = 0.004 and 0.04 L/h/kg, P = 0.0003) in patients heterozygous and homozygous for the *B allele, respectively. A further categorisation of GSTA1 polymorphisms into activity-based groups was of no additional benefit.
In conclusion, the high degree of inter-occasion and inter-individual variability in busulfan pharmacokinetics was reconfirmed through this thesis. Patient specific factors such as GSTA1 polymorphisms affected pharmacokinetic variability, but cAUC was not associated with risk of SOS.