Design and Analysis of End Points in Clinical Trials for Visceral Leishmaniasis

Abstract

Visceral Leishmaniasis (VL) is a vector-borne parasitic disease characterized by fever, substantial weight loss, swelling of the spleen and liver, and anaemia. Clinical trials in VL generally take long to conclude due to challenges with patient recruitment and the fact that evidence for definitive cure (DC) can only be seen at least 6 months after treatment completion. Analysis of efficacy at extended follow-up using the triangular test (IT) design is not straight forward in the case of VL trials. In eastern Africa, the combination of Sodium Stibogluconate (SSG) and Paramomycin (PM) is used as a standard treatment for VL while in India, SSG is no longer used due to a poor safety profile particularly cardiotoxicity. The objectives of this study were to develop alternative analyses approaches for DC at extended followups in TT designs, estimate an optimum time point besides 6 months in the analysis of DC and establish the safety of SSG treatment for VL in eastern Africa. Comparisons between the maximum likelihood estimator (MLE), shrinkage estimator (SHE) and probability tree estimator (PTE) in the analysis of DC following the IT design was carried out using bias, root mean square error (RMSE),variances and coverage probabilities of the confidence intervals (CI). An estimate of optimal timing for DC using multi-state models and a review of existing safety data on SSG treatment has been undertaken to assess its association with cardiotoxicity. The results after analysis indicated that Bias, RMSE and variances were low with high coverage probabilities for both the SHE and PTE. The 95% CI of no change in cure status between month 3 and month 6 has a probability of between 98% and 99.9%. Cardiotoxicity was reported in < 1%of the patients treated with SSG in combination with PM. Both the SHE and PTE are viable alternative approaches in the estimation of efficacy at extended follow-up following IT. There were very few cases in which changes in treatment outcomes occurred between month 3 (M3) and month 6 (M6)follow-up time points. There was not enough evidence to suggest association of SSG use with cardiotoxicity in eastern Africa thus its continued use in combination with PM as a first line treatment for VL is acceptable. Although both the PTE and SHE are good alternatives, analysis approaches for efficacy at extended follow-up to MLE, other analysis approaches of the Bayesian nature need to be explored. Assessment of DC at month 3 gives comparable results to month 6 end point. SSG is still a safe treatment for VL when used in combination with PM but continued monitoring through post-market surveillance is required. This thesis contributes to the improvement in existing knowledge and understanding of the design and analysis methods regarding the cond uct of clinical trials in VL.