Molecular xenomonitoring and xenosurveillance: exploring the potential of entomological screening for the integrated surveillance of neglected tropical diseases

Substantial progress has been made in recent years towards the elimination of neglected tropical diseases (NTDs) including Lymphatic Filariasis (LF) and onchocerciasis, but traditional surveillance methods have limited usefulness for assessments of transmission interruption and post-interruption monitoring. To avoid the risk of recrudescence, there is a need for new methods that are both sensitive and sustainable. Molecular xenomonitoring (MX), the screening of vector insects for the presence of pathogen genetic material, may provide a non-invasive solution. However, its use has been limited by a poor understanding of the sensitivity of MX, the relationship between MX results and human prevalence, and the influence of sampling strategy on the likelihood of detecting exposed vectors. Mosquitoes can be screened for multiple pathogens at once, including those not vectored by mosquitoes, providing optimism that MX could one day provide a mechanism for the integrated surveillance of NTDs. The feasibility of this approach, and a knowledge of which pathogens will be amenable to mosquito-based surveillance, also remains to be fully explored. This research project aimed to investigate these areas of uncertainty, firstly by employing meta-analytical methods to draw novel conclusions from previously-conducted studies, and secondly by using laboratory and field-based methods to collect new data. The results of this research suggest that, in comparison to traditional methods based on the detection of microfilaria (mf) in human biological samples, MX is highly sensitive to communities with ongoing transmission of LF and onchocerciasis. Furthermore, evidence of a significant linear relationship between MX rates and mf prevalence was found. For both diseases, approximately half of the variation in MX rate was explained by variation in the human mf prevalence. The remaining variation may be explained by differences in study site and sampling methodology. For LF, studies that used consistent mosquito collection methods in the same location over time provided evidence of a strong correlation between MX rates and mf prevalence (R2 = 0.78, p