Opportunities for enhanced surveillance of foot‐and‐mouth disease in endemic settings using milk samples

Summary Under‐reporting of foot‐and‐mouth disease (FMD) masks the true prevalence in parts of the world where the disease is endemic. Laboratory testing for the detection of FMD virus (FMDV) is usually reliant upon the collection of vesicular epithelium and fluid samples that can only be collected from acutely infected animals, and therefore animals with sub‐clinical infection may not be identified. Milk is a non‐invasive sample type routinely collected from dairy farms that has been utilized for surveillance of a number of other diseases. The aim of this study was to examine the application of milk as an alternative sample type for FMDV detection and typing, and to evaluate milk as a novel approach for targeted surveillance of FMD in East Africa. FMDV RNA was detected in 73/190 (38%) individual milk samples collected from naturally infected cattle in northern Tanzania. Furthermore, typing information by lineage‐specific rRT‐PCR assays was obtained for 58% of positive samples, and corresponded with the virus types identified during outbreak investigations in the study area. The VP1‐coding sequence data obtained from milk samples corresponded with the sequence data generated from paired epithelial samples collected from the same animal. This study demonstrates that milk represents a potentially valuable sample type for FMDV surveillance and might be used to overcome some of the existing biases of traditional surveillance methods. However, it is recommended that care is taken during sample collection and testing to minimize the likelihood of cross‐contamination. Such approaches could strengthen FMDV surveillance capabilities in East Africa, both at the individual animal and herd level.


Summary
Under-reporting of foot-and-mouth disease (FMD) masks the true prevalence in parts of the world where the disease is endemic. Laboratory testing for the detection of FMD virus (FMDV) is usually reliant upon the collection of vesicular epithelium and fluid samples that can only be collected from acutely infected animals, and therefore animals with sub-clinical infection may not be identified. Milk is a non-invasive sample type routinely collected from dairy farms that has been utilized for surveillance of a number of other diseases. The aim of this study was to examine the application of milk as an alternative sample type for FMDV detection and typing, and to evaluate milk as a novel approach for targeted surveillance of FMD in East Africa. FMDV RNA was detected in 73/190 (38%) individual milk samples collected from naturally infected cattle in northern Tanzania. Furthermore, typing information by lineagespecific rRT-PCR assays was obtained for 58% of positive samples, and corresponded with the virus types identified during outbreak investigations in the study area. The VP1-coding sequence data obtained from milk samples corresponded with the sequence data generated from paired epithelial samples collected from the same animal. This study demonstrates that milk represents a potentially valuable sample type for FMDV surveillance and might be used to overcome some of the existing biases of traditional surveillance methods. However, it is recommended that care is taken during sample collection and testing to minimize the likelihood of cross-contamination.
Such approaches could strengthen FMDV surveillance capabilities in East Africa, both at the individual animal and herd level.

K E Y W O R D S
East-Africa, foot-and-mouth disease virus, milk, real-time RT-PCR, surveillance, Tanzania This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2019 The Authors. Transboundary and Emerging Diseases published by Blackwell Verlag GmbH.
Foot-and-mouth disease (FMD) is a highly contagious disease of cloven-hooved mammals and is of great global economic importance (Knight-Jones & Rushton, 2013). There are seven serotypes of FMD virus (FMDV), O, A, C, Asia 1 and Southern African Territories (SAT) 1, SAT 2 and SAT 3 (Robson, Harris, & Brown, 1977), four of which (O, A, SAT 1 and SAT 2) currently circulate in domestic livestock in East Africa, where the disease is endemic. Vaccination is the most effective control measure for FMD prevention. In order to identify the appropriate vaccines and the time of their application, a thorough understanding of contemporary serotypes/strains is necessary (Casey-Bryars et al., 2018;Kasanga et al., 2012;Paton, Sumption, & Charleston, 2009). However, rapid viral detection and characterization can often be problematic in endemic areas due to limited resources and capacity to undertake surveillance.
Pan-serotypic real-time reverse-transcription polymerase chain reaction (rRT-PCR) assays have been described for the rapid detection of FMDV in typical clinical samples (Callahan et al., 2002;King et al., 2006;Reid et al., 2002;Shaw et al., 2007). These assays target highly conserved genomic regions that are shared among all serotypes and topotypes, but do not differentiate between them. To enable rapid typing of FMDV, rRT-PCR assays have also been developed for the detection and differentiation between FMDV lineages specific to particular geographical regions, including East Africa (Bachanek-Bankowska et al., 2016). These region-tailored typing assays target lineage-specific conserved regions within the variable VP1-coding sequence (Bachanek-Bankowska et al., 2016;Reid et al., 2014;Saduakassova et al., 2017).
Surveillance for FMD in an endemic setting such as East Africa often relies on passive surveillance, depending on farmers or veterinarians to observe and report infected herds, and is only rarely supplemented by targeted case finding (Kasanga et al., 2012;Namatovu et al., 2013). Furthermore, sample types for laboratory FMDV diagnosis, typically vesicular epithelium and fluid, are invasive and labour intensive to obtain, and therefore, they are collected infrequently, resulting in under-reporting (Knight-Jones, McLaws, & Rushton, 2016). As a consequence, FMD reporting is inherently biased towards clinically affected animals, failing to capture viruses circulating sub-clinically that may play a role in disease transmission.
Milk is routinely collected from dairy farms, and has been exploited as a surveillance tool for the detection of other diseases of veterinary importance, for example bovine viral diarrhoea, border disease and bluetongue (Beaudeau et al., 2001;Berriatua et al., 2006;Kramps, van Maanen, Mars, Popma, & van Rijn, 2008). It has been demonstrated that the mammary gland is a highly susceptible organ for FMDV replication, and that during infection, FMDV RNA can be detected in milk by rRT-PCR before, during and after the appearance of clinical signs (Armson et al., 2018;Blackwell & McKercher, 1982;Burrows, Mann, Greig, Chapman, & , 1971;Reid et al., 2006). However, only a small number of studies have described the detection of FMDV RNA in milk from naturally-infected animals. These include FMDV detection in milk during the 2007 FMD outbreak in the United Kingdom (Armson et al., 2018), in cattle and buffaloes in Pakistan (Ahmed et al., 2017;Saeed et al., 2011)

and in cattle in
India (Ranjan et al., 2016). The limited milk samples used in these studies were collected either as an additional sample type to validate molecular assays, or to investigate the possible role of milk in FMDV transmission; nonetheless these studies provide useful evidence that FMDV RNA can be detected and typed by rRT-PCR, in milk from naturally infected animals. Consequently, further investigation into the potential of milk as an alternative non-invasive sample type for routine FMDV detection and surveillance is warranted, particularly in areas where surveillance infrastructure is limited. Therefore, the aim of this study was to examine the use of milk for surveillance in endemic settings of East Africa where this approach had not been investigated to date.  (Table S1).
An initial screen of all the milk samples and the lesion material from four of the cows was performed. For this, RNA was extracted using the MagMAX ™ -96 Viral RNA Isolation Kit (Applied Biosystems ® ). The OIE recommended pan-serotypic rRT-PCR assay was carried out on an Applied Biosystems ® 7500 Real-time PCR System, using the Superscript III Platinum ® One-Step qRT-PCR Kit (Invitrogen ™ ), with primers and probes targeting the conserved three-dimensional region of the FMDV genome (Callahan et al., 2002;OIE Terrestrial Manual, 2017), and thermal cycling conditions as previously reported (Shaw et al., 2007). Positive samples were then tested using East Africa (EA) typing rRT-PCR assays, as previ-  Table S1). Milk samples that were observed to have a C T value of above 38 using the pan-serotypic rRT-PCR assay were unable to be typed. In addition, a positive signal from more than one typing assay was identified in 18 milk samples, including three samples each positive for three serotypes (O, A and SAT 1). In samples with a positive signal for two FMDV types, O and SAT 1 were the most common types detected, while types A and SAT 1 were identified in one sample only. It is possible that these animals were co-infected with multiple FMDV serotypes, as has been previously described in endemic areas (Ferris, Oxtoby, & Hughes, 1995;Woodbury, Samuel, Knowles, Hafez, & Kitching, 1994). However, alternative explanations should also be considered, including the possibility that these results represent (a) contamination due to contact with materials infected with other FMDV types during sample collection in the field, transport or testing in the laboratory; or (b) cross-reaction between the individual typing rRT-PCR assays, although no evidence of this has been To determine if milk is a suitable alternative sample type to vesicular lesion material (epithelium/fluid) for FMDV detection and typing, both sample types collected from the same animal were tested and the results compared (Table 1). In the pan-serotypic assays, the C T values of the lesion material samples were stronger (lower C T values) than of the milk samples, confirming previous observations of higher virus concentrations in vesicular lesions .
Typing results were comparable for all pairs, with the exception of Animal A, where no signal was observed in any of the typing assays, possibly due to the reduced viral load observed in this animal. In three animals (B, C and D), SAT 1 was detected in both milk and lesion material samples (Table 1). In animals C and D, the C T values of the pan-serotypic and the SAT 1-specific assays were comparable, while in animals A and B the differences in the values were greater.
In animal A, SAT 1 was detected in the vesicular lesion sample only. cells. Upon comparison of the SAT 1-specific primers/probe with the VP1-coding sequence data obtained from milk and vesicular samples, it was evident that the difference in C T values between the pan-serotypic and the SAT 1-type specific assay may occur due to nucleotide differences at the 3' end of the primer binding region of the typing assay. At least one nucleotide difference was identified within the SAT 1-specific typing assay binding region in sequences obtained from animals A and B, while no such differences were observed in sequence data obtained from animals C and D. As the VP1coding sequence is the most variable genome region, mismatches between the primers and probes of the typing assays and the template are expected. Therefore, it is recommended to use typing assays alongside the more sensitive pan-serotypic assay as a screening tool (Bachanek-Bankowska et al., 2016).
This study demonstrates that milk could represent a valuable sample type as an alternative to the traditional diagnostic samples collected for FMD surveillance: vesicular epithelium or fluid. Milk from individual animals can be routinely collected and FMDV RNA can be detected and typed by rRT-PCR in milk samples in a region where FMD is endemic, albeit with weaker C T values than from vesicular samples. The study also demonstrates that VP1 sequence data may be obtained from milk samples, enhancing the possibility of further, in-depth virus characterization. Milk sampling as a targeted surveillance approach shows promise given the concordance between typing data from milk samples and confirmed reports from outbreak investigations. Follow-on studies are required to assess the application of pooled milk in combination with herd clinical status for FMDV surveillance. In conclusion, milk is a simple-to-collect, non-invasive sample type which might be utilized in targeted surveillance campaigns in FMD endemic regions.
However, due to the high analytical sensitivity of molecular tests used to detect FMDV, appropriate care needs to be taken to minimize the possibility for cross-contamination during sample collection, transport and testing in the laboratory. The use of milk as a diagnostic sample might help to address some of the potential biases of traditional surveillance methods and improve surveillance capabilities for reporting of disease at the individual and herd level in East Africa.

ACK N OWLED G EM ENTS
The authors thank colleagues in the WRLFMD, The Pirbright

CO N FLI C T O F I NTE R E S T
The authors have no competing interests.