When can field trials contribute practically to understanding of - PowerPoint PPT Presentation

When can field trials contribute practically to understanding of efficacy (and when do they not?) Professor James Wood Disease Dynamics Unit, Dept Veterinary Medicine and Interdisciplinary Research Centre in Infectious Diseases University of Cambridge

Alternatives to field trials • Modelling transmission studies • Experimental transmission studies What are the advantages and limitations of these approaches? Should there be stipulations? Are there times that field efficacy data are essential?

Impact of vaccination • Vaccines should obviously prevent disease in vaccinated individuals • One of the largest impacts of vaccination is through indirect protection that comes about through reduced transmission. – Many experimental / dossier studies fail to consider transmission at the population level – Unstructured pharmacovigilance data cannot be used to assess transmission • Measuring transmission (and reductions) is highly complex

Case Study 1: Equine influenza vaccination • Equine influenza vaccines need to be adequately potent against homologous virus – But continuous antigen drift also interferes • even if not linear • When do vaccines need to be updated with new strains? – Experimental investigation in ponies – Comparison of in date with out of date vaccination – Carefully parameterised models demonstrate that impact is far larger than expected from experimental data

Difference 1: Seroconversion depends on antibody level and homology of viruses 1 Homologous vaccinates Heterologous vaccinates 0.9 0.8 Probability of seroconverting 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 50 100 150 200 250 Pre-challenge antibody level (srh, mm 2 ) Park et al (2004) Proc Roy Soc B 271 , 1547-1555

Difference 1: Seroconversion depends on antibody level and homology of viruses 1 Homologous vaccinates Heterologous vaccinates 0.9 0.8 Probability of seroconverting 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 50 100 150 200 250 Pre-challenge antibody level (srh, mm 2 ) Park et al (2004) Proc Roy Soc B 271 , 1547-1555

Difference 2: Not all seroconverters excrete virus 1 0.9 0.8 0.7 pr(excreting|sero+) 0.6 0.5 0.4 0.3 0.2 0.1 0 homologous heterologous Park et al (2004) Proc Roy Soc B 271 , 1547-1555

Difference 3: Latent/infectious periods Homologous latent periods Homologous infectious periods 20 10 8 15 Frequency Frequency 6 10 4 5 2 0 0 1 2 3 4 5 6 1 2 3 4 5 6 Days Days Heterologous latent periods Heterologous infectious periods 20 10 8 15 Frequency Frequency 6 10 4 5 2 0 0 1 2 3 4 5 6 1 2 3 4 5 6 Days Days Park et al (2004) Proc Roy Soc B 271 , 1547-1555

Impact: Heterology significantly increases modelled risk in populations of racehorses Probability of epidemic >= 5% 1 0.8 0.6 0.4 0.2 0 5 10 15 20 25 30 35 40 45 50 Week (starting Jan 1) Probability of epidemic >= 20% 1 0.8 0.6 0.4 0.2 0 5 10 15 20 25 30 35 40 45 50 Week (starting Jan 1) Park et al (2004) Proc Roy Soc B 271 , 1547-1555

Bovine TB vaccination • An aspiration for GB government policy • Requires a change in EU law – Invited ESPA opinion states need for field trials that show reduced transmission • Field trials would require vaccination to be a supplement to current measures – Positive animals are removed on detection which reduces low transmission to very slow • We used carefully parameterised within herd transmission models to consider the necessary scale of trials Conlan, Vordermeier, de Jong & Wood submitted

Necessary field trial scale • Removal of test-positive animals from herds obscures benefits of vaccination – severely limits potential to discern impact of vaccination on transmission • 100 herds required to demonstrate impact of vaccination at animal level – But farmers and policy makers need information on farm scale impact as controls operate at level of farm • >1000 herds required to demonstrate impact at herd level Conlan, Vordermeier, de Jong & Wood submitted

Alternative experimental transmission studies • Experimental transmission studies – Use 50:50 mix of infected seeder animals and sentinel animals – Consider joint final size probability distributions • Very sensitive to assumed / estimated R0 in cattle and herd size effects • For effect size: 75% and power: 80%, R0 1.5 – required in-contact time 1-6 years depending on the transmission scenario, with a group size of 52 animals – For 50% effect size, group size 128 & duration 1-5yrs

Conclusions • Modelling can provide greater understanding of meaning of experimental data than simple statistical analysis of data from vaccine trials – Massive non-linearities in transmission make impacts of interventions very unpredictable • Pharmacovigilance data will fail to demonstrate full impact of interventions in many cases • Field trials can be massively inefficient and costly – Can fail to estimate transmission changes – Can fail to answer question of interest because of scale – Should in many cases be replaced by more efficient experimental design