SUMMARY OF THE RESULTS OF SCRAPIE SURVEILLANCE IN SHEEP IN GREAT BRITAIN JANUARY MARCH 2003

SUMMARY OF THE RESULTS OF SCRAPIE SURVEILLANCE IN SHEEP IN GREAT BRITAIN JANUARY 2002 - MARCH 2003 A document prepared by: With contributions from: John Wilesmith + Danny Matthews + Judi Ryan + Heather Elliott ± Victor Del Rio Vilas + Mark Arnold + Simon Gubbins + Kumar Sivam + + Centre for Epidemiology and Risk Analysis, VLA Weybridge ± Neuropathology Department, VLA Weybridge The Veterinary Laboratories Agency is an Executive Agency of the Department for Environment, Food and Rural Affairs

CONTENTS Section Page Executive summary. 3 1 Abattoir survey. 5 2 Survey of fallen stock.. 8 3 Goats. 11 4 Genotyping... 12 5 Scrapie notifications of clinical (passive surveillance)... 15 6 Estimating the prevalence of infection in the gb sheep flock... 17 7 References... 19 8 Annex 1. Sampling methods and test sensitivity 20 Figure 1. Cutting plan for the brainstem in the head survey.. 25 Figure 2. Cutting plan for the brainstem in the brainstem survey 26

SCRAPIE SURVEILLANCE IN GREAT BRITAIN, JANUARY 2002 MARCH 2003 EXECUTIVE SUMMARY I. Four sources of surveillance are currently available for analysis: the results of routine notifications of clinically suspect of scrapie; an abattoir survey of sheep >18 months of age conducted between January 2002 and March 2003; a survey of fallen stock also conducted between January 2002 and March 2003; and an anonymous postal questionnaire survey of breeding flocks in GB conducted in 2002. II. In the abattoir survey, 29,201 sheep were tested using the Biorad Platelia ELISA on brainstem with confirmation by immunohistochemistry (IHC) (the brainstem survey), and 21,429 were tested using the Prionics Check western blot with additional IHC on obex, submandibular and medial retropharyngeal lymph nodes and tonsil (the head survey). III. The abattoir survey indicates a crude prevalence of infection of 0.13% (67/50,630) in the slaughter population of sheep > 18 months of age. The prevalence of infection found in the head survey (0.20%) was significantly higher (p<0.0001) than that in the brainstem survey (0.08%). Modelling studies using these results provide an estimate of the prevalence of infection in the whole GB sheep flock of 0.33% (95% CI: 0.24 0.44). IV. In addition to the 24 confirmed positives identified by the Biorad test, there were 28 positive Biorad results which could not be confirmed. These are the subject of further investigation. V. The results of the fallen stock survey provide a minimum estimate of the prevalence of scrapie affected flocks of 0.39% (95% CI: 0.00 0.78). The anonymous postal survey recorded that 1% of farmers thought that they had experienced at least one case of scrapie in the previous 12 months.

VI. The results of genotyping scrapie positive sheep in the abattoir survey indicate a disproportionately greater number of animals with the ARR/VRQ genotype, and a disproportionately lesser number with the VRQ/VRQ and ARQ/ARQ genotypes, compared to the results from notified clinical scrapie. This is most likely to be due to the longer incubation period of scrapie associated with the first genotype, and the shorter incubation period associated with the last two genotypes. VII. An analysis of the full genotyping data is in progress and the results will be published separately. There is general agreement between the abattoir survey data and the original VLA genotyping model, but for two breeds, Swaledale and Suffolk, the slight changes in genotype frequencies are consistent with the programme of breeding for resistance in the two breeds. VIII. The results of routine notification of clinically suspect of scrapie indicated that flock owners were reverting to reporting in 2002 following the disruption caused by the FMD epidemic, but there were early indications of a 10% reduction in the first half of 2003. IX. The abattoir survey and the survey of fallen stock indicate that the Prionics Check western blot and ELISA screening tests have a lower sensitivity than immuno-histochemistry (IHC) examination of the brainstem. This is likely to be due to the intrinsic sensitivity of the IHC at the obex, and the effect of specific brain sampling sites on PrP concentration in pre-clinically infected sheep. Also, estimates of the prevalence of scrapie are unlikely to be improved by the examination of LRS tissues by IHC in addition to that of brainstem samples. The results do indicate, however, that if the objective of the surveillance is case finding, then testing both brain and lymphoid tissue will improve the ascertainment of infected sheep.

1. ABATTOIR SURVEY 1.1 In January 2002 Defra began a programme of active surveillance for sheep scrapie. This began as an abattoir survey commissioned by the Spongiform Encephalopathy Advisory Committee (SEAC) to estimate the prevalence of sheep scrapie in the British sheep flock, but by April 2002 was subsumed within a statutory survey as required by Commission Regulation 999/2001. 1.2 This EU requirement involved the examination of a random sample of 60,000 sheep over 18 months of age between April 2002 and March 2003. Within this survey, 20,000 of the sheep were to have additional tissues tested to satisfy the SEAC requirements. 1.3 The design of the abattoir survey involved a systematic sample of sheep (>18 months of age) slaughtered in Great Britain. On administrative grounds, it was decided that those abattoirs currently recorded as slaughtering over 10,000 sheep over 18 months of age formed the sampling frame. These abattoirs slaughter 93% of sheep in GB. The number of samples required from each of the abattoirs was proportional to their throughputs. 1.4 With respect to the testing of the anticipated 60,000 samples, a sampling protocol was designed such that 40,000 sheep were identified for testing of brainstem samples (brainstem survey) and 20,000 sheep were identified for testing of samples of brainstem (obex), submandibular and retropharyngeal lymph nodes and tonsils (head survey). The sampling of the 20,000 sheep started in January 2002 to satisfy the SEAC requirements, and the additional 40,000 required for EU started in April 2002. They both continued through to March 2003. The brainstem sampling was carried out in the selected abattoirs, with samples being sent to VLA s Regional Laboratory at Shrewsbury. The 20,000 head survey sheep were decapitated in the abattoir and the heads sent to the nearest VLA Regional Laboratory, or one of the Scottish Agricultural Colleges, for sampling. 1.5 For the 40,000 sample of sheep, the brainstem was screened using the BioRad Platelia ELISA and samples positive to this screening test were confirmed by immuno-histochemistry (IHC). 1.6 The 20,000 sheep heads collected for additional testing, as recommended by the SEAC, were subjected to the following tests. The brainstem was tested by the Prionics Check western blot (WB); immunohistochemistry (IHC) was undertaken on fixed brainstem (obex),

submandibular and medial retropharyngeal lymph nodes, and the tonsil. This protocol also satisfied the requirements of Commission Regulation 999/2001, but went further without compromising the statutory objectives. 1.7 All positive IHC, arising from the ultimate sample of 60,000 sheep, were re-read and confirmed by specialist staff at the VLA, Weybridge. 1.8 In reporting the outcome of the survey it has been decided in the interest of simplicity to assume that the approaches in each part of the survey are of equivalent sensitivity. In other words, Prionics Check WB plus IHC on brain, plus IHC on lymphoid tissue is equivalent to Bio-rad Platelia testing of the brainstem followed by IHC on ELISA positive samples. This is because from discussion of interim reports on the survey the separation of the two components led to difficulties in interpretation for the lay reader. Despite some reservations about the inappropriateness of data pooling in this way, the overall message from the surveillance programme is not significantly influenced. Issues of test sensitivity and appropriate sampling methods will be discussed later in this report (Annex 1).The specific issues will be thoroughly aired in the subsequent scientific paper. 1.9 A total of 54,145 sheep were sampled. Brainstem samples were taken from 32,133 of these sheep for testing by ELISA only, with confirmatory test as appropriate (brainstem survey). Additional tissues were taken from the remaining 22,012 sheep according to the extended testing protocol defined as the head survey. However, 3,515 (6.5%) of the total samples taken were deemed to be unsuitable for testing. 1.10 Of the 50,630 suitable samples, sixty-seven (0.13%) were positive (Table 1). Of these 24 were identified by the Biorad Platelia ELISA, 28 by Prionics-check WB, 12 by IHC on the obex (and were negative to the Prionics Western Blot), and three by IHC on lymph nodes and tonsil (these were negative to Prionics Western Blot and IHC on the obex). 1.11 The percentage of positive samples in the head survey (0.20%) was significantly higher (p < 0.001) than that for the brainstem survey (0.08%). This indicates the greater sensitivity of the IHC method compared with the ELISA which is likely to be a result of a combination of the intrinsic sensitivity of the IHC and the brain sampling sites used for each diagnostic method.

Table 1. Number of samples collected and tested by survey type and time period i) January March 2002 Total Total Suitable Percent Survey type Collected for testing Negative Positive positive Head survey 1,045 939 938 1 0.11% Brainstem survey 0 ii) April 2002 March 2003 Head survey 20,967 20,490 20,448 42 0.20% Brainstem survey 32,133 29,201 29,177 24 0.08% iii) January 2002 March 2003 Head survey 22,012 *21,429 21,386 43 0.20% Brainstem survey 32,133 29,201 29,177 24 0.08% Total 54,145 50,630 50,563 67 0.13% * Of the 21,429 sheep tested, 167 were not subjected to the full range of tests. Head survey: Prionics Check western blot on brainstem plus immunohistochemistry (IHC) on obex, lymphoid tissues and tonsil. Brainstem survey: Biorad Platelia ELISA on brainstem with confirmation of positives by IHC. 1.12 In addition to the 24 confirmed positives identified by the Biorad test, there were 30 positive Biorad Platelia results which could not be confirmed by IHC, the current gold standard. For 28 of these the confirmatory test was negative, and these have been reported as negative in this report. In the absence of confirmation by an OIE approved test, this approach is consistent with Commission Regulation 999/2001. There was insufficient tissue for the remaining two for confirmatory tests to be carried out. These results are the subject of further investigation, and are not considered further in this report. However, it should be pointed out that the Biorad Platelia test has not been formally validated for use in sheep tissues. Therefore the diagnostic significance of these results cannot be determined without further detailed scientific investigation.

2. SURVEY OF FALLEN STOCK 2.1 At the same time as the abattoir survey a representative sample of 6,000 fallen sheep were required by the EU to be sampled. As these could not be sampled at the collection sites, the heads were taken and delivered to VLA Regional Laboratories and Scottish Agricultural Colleges. It was therefore logistically possible to sample additional tissues as for the head survey. 2.2 The epidemiological design of the fallen stock survey was initially based upon a random selection of registered owners of sheep. Of the 120,000 sheep owners in Great Britain, a sample of 30,000 owners were contacted and requested to submit two sheep over 18 months of age that had died. A sampling quota was allocated to each Animal Health Office catchment area based on the geographical distribution and seasonality obtained from the throughput of sheep presented for post mortem at the VLA Regional laboratories. The response declined very rapidly after the first few months. An additional 30,000 flock owners were contacted by post in August 2002. Then, in December 2002, a large publicity campaign was launched to stimulate and maintain submissions. 2.3 In an attempt to increase the numbers of fallen stock sampled, sheep which died in transit to the abattoir were included from November 2002, and the restriction on the number of sheep per farm was eased. 2.4 Samples of the brainstem, tonsil and medial retropharyngeal and submandibular lymph nodes were taken. The brainstem was tested using the Prionics Check WB and, where possible, the additional tissues were examined by immunohistochemistry. Those samples not suitable for IHC (due to autolysis) were examined for Scrapie Associated Fibrils (SAF) by electron microscopy. 2.5 The ascertainment of fallen stock has proved difficult. The number of submitted in each month is shown in Table 2. There is some evidence of seasonality in the ascertainment of, with submissions being greatest during the lambing period, but this is confounded by the efforts made to secure animals for sampling.

Table 2 Number of fallen stock and sheep dying in transit to the abattoir, submitted per month during 2002-03. Year Month Number Submitted Number Suitable Number Positive 2002 Jan 17 17 0 2002 Feb 255 223 5 2002 Mar 198 164 2 2002 Apr 204 155 0 2002 May 89 70 0 2002 Jun 39 32 0 2002 Jul 25 21 0 2002 Aug 11 9 0 2002 Sep 14 11 0 2002 Oct 77 74 0 2002 Nov 62 59 0 2002 Dec 85 78 1 2003 Jan 312 295 1 2003 Feb 491 456 2 2003 Mar 628 549 1 Total 2507 2213 12 2.6 Of the 2,507 sheep submitted, 2,408 were fallen stock on farms and 99 were from sheep dying in transit to the abattoir. Of the 2,507 sheep submitted for the survey, 294 (11.7%) were not suitable for testing. Of the remaining 2,213 sheep, twelve (0.54%) were positive to at least one of the tests (Table 2). None of the 99 sheep which died in transit to the abattoir were positive. Eleven of the twelve were positive under the EU criteria as they were positive to the Prionics Check test and confirmed by IHC on the obex. One animal was negative to the Prionics test, but was positive by the IHC. There was no evidence of seasonality in the numbers of positive samples, however there was a significantly higher proportion of positives in January March 2002 compared to that in January March 2003 (p < 0.01). 2.7 The flock of origin was known for 2,395 of the 2,408 sheep, and these were obtained from 1,319 flocks. Their distribution is shown in Table 3. Eighty-five percent of the flocks (1,122) submitted one or two animals. Thirteen owners (1%) submitted 10 or more sheep, with one submitting 38. All of the positive sheep were submitted from different flocks. The 1,319 flocks submitting fallen stock represent 1.9% of the 70,000 breeding flocks in GB.

Table 3. Distribution of sheep submitted per flock. No. of sheep per flock No. of flocks submitting sheep Percentage of flocks sampled in survey No. of flocks with a positive sheep Number of sheep submitted No. of positive sheep 1 794 60.20 6 794 6 2 328 24.87 3 565 3 3 83 6.29 249 4 56 4.25 2 224 2 5 15 1.14 75 6 14 1.06 1 84 1 7 10 0.76 70 8 5 0.38 40 9 1 0.08 9 10 1 0.08 10 11 3 0.23 33 12 3 0.23 36 13 1 0.08 13 14 1 0.08 14 15 1 0.08 15 17 1 0.08 17 18 1 0.08 18 38 1 0.08 38 Total 1,319 100.00 12 2395 12 2.8 Of the 1,319 flocks from which fallen stock were submitted, 30 (2.2%) have reported at least one confirmed clinical case of scrapie in the past 5 years. During this period scrapie was confirmed in 486 flocks. (See Section 5 for more details of scrapie notifications.) Of the 1,289 flocks in which scrapie had not been confirmed previously, seven have now been identified as scrapie affected as a result of this survey. Over the period of 15 months in which the survey was carried out, this provides a minimum prevalence of flocks affected with clinical scrapie of 0.54% (95% CI: 0.14 0.94), on the assumption that scrapie was responsible for the death of these fallen stock.

2.9 Restricting the data to the 12 month period of April 2002 to March 2003, 1,032 flock owners submitted sheep, of which 26 (2.5%) had previously reported a clinical case. Of the 1,006 flocks in which scrapie had not been confirmed previously, four have now been identified as a result of this survey. This gives a minimum estimate of the annual prevalence of flocks affected with clinical scrapie of 0.39% (95% CI: 0.00 0.78), on the assumption that scrapie was responsible for the death of these fallen stock. 2.10 The anonymous postal survey of breeding flocks in GB (Sivam et al, 2003) conducted in 2002 estimated that 1.0% of farmers thought that they had experienced at least one case of scrapie in the previous 12 months. 3. GOATS 3.1 During the same period, the UK was required to test 240 healthy and 24 fallen goats. 3.2 Only 75 healthy goats were sampled, 66 of these being suitable for testing. All were negative. 3.3 Eleven fallen goats were sampled, one of which was unsuitable for testing. The remaining ten were all negative.

4. GENOTYPING 4.1 All sheep and goats in the above surveys were subject to the determination of the PrP allelic variants at codons 136, 154 and 171. In all the genotyping was conducted on brain tissue, following validation of the process, rather than through the usual medium of blood. 4.2 The distributions of genotypes found in each survey group are shown in Table 4. The genotype distribution for clinical of scrapie from 1998 to 2002 is shown for comparison (Table 5). There was no difference in the genotype distributions between the brainstem and head abattoir surveys. 4.3 The frequency distribution of the positive animals from the abattoir survey indicates a disproportionately greater number of ARR/VRQ genotypes and disproportionately lesser numbers of VRQ/VRQ and ARQ/ARQ genotypes compared with the distribution in the notified (Table 4). The most likely explanation for this is the longer incubation period in sheep with the ARR/VRQ genotype and the shorter incubation period of the VRQ/VRQ and ARQ/ARQ genotypes (Table 7) and the fact that the majority of sampled animals would have been >36 months of age. This is simply because in Great Britain there are very few sheep slaughtered between 18 and 36 months of age. 4.4 An analysis of the full genotyping data is in progress and the results will be published separately. There is general agreement between the abattoir survey data and the original VLA genotyping model (Arnold et al, 2002), but for two breeds, Swaledale and Suffolk, the slight changes in genotype frequencies are consistent with the programme of breeding for resistance in the two breeds.

Table 4. Distribution of genotypes for all sheep tested in the active and passive surveillance programmes. NSP Passive surveillance 1998-2002 Sheep abattoir Sheep fallen stock Goat abattoir Goat fallen Group Pos % Neg % Pos % Neg % Pos % Neg % Neg % I ARR/ARR 9524 19.57% 383 21.63% Total 9524 19.57% 383 21.63% ii ARR/AHQ 1 0.06% 4563 9.38% 166 9.37% ARR/ARH 694 1.43% 53 2.99% ARR/ARQ 6 0.34% 15153 31.14% 559 31.56% 1 10.00% Total 7 0.40% 20410 41.95% 778 43.93% 1 10.00% iii AHQ/AHQ 6 0.34% 849 1.74% 1 1.49% 28 1.58% AHQ/ARH 147 0.30% 12 0.68% AHQ/ARQ 38 2.16% 4259 8.75% 1 1.49% 127 7.17% ARH/ARH 3 0.17% 177 0.36% 15 0.85% ARQ/ARH 5 0.28% 529 1.09% 30 1.69% 1 9.09% ARQ/ARQ 340 19.29% 6955 14.29% 3 4.48% 212 11.97% 65 100.00% 9 90.00% Total 392 22.23% 12916 26.54% 5 7.46% 424 23.94% 1 9.09% 65 100.00% 9 90.00% iv ARR/VRQ 44 2.50% 2660 5.47% 15 22.39% 86 4.86% Total 44 2.50% 2660 5.47% 15 22.39% 86 4.86% v AHQ/VRQ 2 0.11% 833 1.71% 1 1.49% 31 1.75% ARH/VRQ 79 4.48% 88 0.18% 3 4.48% 4 0.23% ARQ/VRQ 906 51.39% 2044 4.20% 38 56.72% 60 3.39% 9 81.82% VRQ/VRQ 333 18.89% 183 0.38% 5 7.46% 5 0.28% 1 9.09% Total 1320 74.87% 3148 6.47% 47 70.15% 100 5.65% 10 90.91% Totals 1763 100.00% 48658 100.00% 67 100.00% 1771 100.00% 11 100.00% 65 100.00% 10 100.00%

Table 5. Distribution of genotypes for clinical scrapie reported to Defra (passive surveillance) 1998-2002. NSP Genotypes Number %total Number %total Number %total Number %total Number %total total number %total Group confirmed confirmed confirmed confirmed confirmed confirmed confirmed confirmed confirmed confirmed confirmed confirmed 1998 reported 1999 reported 2000 reported 2001 reported 2002 reported 1998-2002 reported I ARR/ARR 0 0.00% 0 0.00% 0 0.00% 0 0.00% 0 0.00% 0 0.00% II ARR/AHQ 0 0.00% 1 0.20% 0 0.00% 0 0.00% 0 0.00% 1 0.06% II ARR/ARH 0 0.00% 0 0.00% 0 0.00% 0 0.00% 0 0.00% 0 0.00% II ARR/ARQ 0 0.00% 1 0.20% 2 0.41% 2 0.85% 1 0.31% 6 0.34% II Total 0 0.00% 2 0.39% 2 0.41% 2 0.85% 1 0.31% 7 0.40% III ARQ/ARH 1 0.48% 0 0.00% 0 0.00% 1 0.43% 3 0.92% 5 0.28% III ARQ/AHQ 6 2.86% 12 2.35% 9 1.86% 4 1.70% 7 2.15% 38 2.16% III AHQ/AHQ 2 0.95% 3 0.59% 0 0.00% 1 0.43% 0 0.00% 6 0.34% III ARH/ARH 0 0.00% 0 0.00% 1 0.21% 0 0.00% 2 0.62% 3 0.17% III AHQ/ARH 0 0.00% 0 0.00% 0 0.00% 0 0.00% 0 0.00% 0 0.00% III ARQ/ARQ 29 13.81% 99 19.41% 84 17.39% 44 18.72% 84 25.85% 340 19.29% III Total 38 18.10% 114 22.35% 94 19.46% 50 21.28% 96 29.54% 392 22.23% IV ARR/VRQ 6 2.86% 9 1.76% 8 1.66% 9 3.83% 12 3.69% 44 2.50% V AHQ/VRQ 0 0.00% 0 0.00% 0 0.00% 1 0.43% 1 0.31% 2 0.11% V ARH/VRQ 6 2.86% 23 4.51% 25 5.18% 7 2.98% 18 5.54% 79 4.48% V ARQ/VRQ 120 57.14% 248 48.63% 268 55.49% 129 54.89% 141 43.38% 906 51.39% V VRQ/VRQ 40 19.05% 114 22.35% 86 17.81% 37 15.74% 56 17.23% 333 18.89% V Total 166 79.05% 385 75.49% 379 78.47% 174 74.04% 216 66.46% 1320 74.87% Totals 210 100.00% 510 100.00% 483 100.00% 235 100.00% 325 100.00% 1763 100.00% Genotypeunknown 289 88 85 60 79 601 TotalCases 499 598 568 295 404 2364

5. SCRAPIE NOTIFICATIONS OF CLINICAL CASES (PASSIVE SURVEILLANCE) 5.1 Scrapie in sheep and goats was made a notifiable disease within the EU from January 1993. The Scrapie Notifications Database (SND) was established at the VLA in July 1998 for recording details of reported in GB, when the diagnosis of scrapie was supplemented by the genotyping of confirmed. Available back data from 1993 to June 1998 was included in the database. Between January 1993 and December 2002, scrapie has been confirmed in 4142 sheep from 1571 locations and 8 goats from 8 goat herds. From 1 January to 12 th July 2003, 197 have been confirmed in sheep from 73 flocks. There have been no in goats during 2003 to date. 5.2 In terms of the observed incidence, the most notable finding is the effect of the FMD epidemic in 2001 which significantly suppressed the reporting of scrapie (Table 6). The number of notifications of suspect which were confirmed as scrapie during 2002 indicated that flock owners were reverting to notifying suspect of scrapie to the State Veterinary Service of Defra. However, during 2003, there has been a reduction in the number of and flocks reported and confirmed compared to 2002. Confirmation figures can be misleading because of the delay in the diagnosis for any particular period. Reported figures are therefore used for this comparison and these indicate that there is a 10% reduction in the number of reported to Defra in 2003 compared with the same period in 2002 (1 January to 12 July Table 6). Table 6. Number of confirmed of scrapie reported to Defra 1998 2002 and number reported between January and 12 July 2002 and 2003. Year No. confirmed Confirmed Reported 1998 499 Jan-12 July 2002 240 334 1999 598 Jan-12 July 2003 197 298 2000 568 Reduction 18% 10% 2001 295 2002 404

5.3 In 2003, the 298 reported clinically suspect originated from 116 flocks (2.6 reported per flock), and the 197 subsequently confirmed of scrapie were from 73 flocks (2.7 confirmed per flock). 5.4 The age at report for confirmed appears to be influenced by genotype. The average age at reporting varies significantly (analysis of variance, p<0.0001) among the most frequent genotypes (2002 data). This is shown in Table 7. This is consistent with findings in previous years. Table 7. Age at report for different genotype groups NSP Group Genotype N Average age (years) V VRQ/VRQ 54 38.5 months (3.2) V ARH/VRQ 15 38.9 months (3.2) III ARQ/ARQ 75 42.8 months (3.6) V ARQ/VRQ 128 47.6 months (4.0) III ARQ/AHQ 7 59.7 months (5.0) IV ARR/VRQ 11 67.7 months (5.6)

6. ESTIMATING THE PREVALENCE OF INFECTION IN THE GB SHEEP FLOCK 6.1 On their own, the results of the abattoir survey provide an estimate of the prevalence of detectable infection in the cull ewe population (Table 1). By allowing for the variation in the prevalence of infection with age and test sensitivity and specificity with stage of incubation it is possible, however, to use the results of the abattoir survey to estimate the prevalence of scrapie infection in the GB sheep population (Gubbins and others, 2003). This involves using a modelling approach that accounts for these factors and, moreover, which provides a method to relate the prevalence of infection in the abattoir population to that in the GB population as a whole. 6.2 To analyse the results, we group the animals sampled according to which combination of diagnostic tests was used on them (Table 8). Sample group 1 comprises those animals tested as part of the brainstem survey while sample groups 2 & 3 comprise those animals tested as part of the head survey. For the purposes of the model we can only make use of three of the tests used in the surveys, the Biorad Platelia ELISA, the Prionics Check western blot and immunohistochemistry (IHC) on the tonsil. Table 8. Results from the 2002/3 abattoir survey used in the analysis. Sample group Diagnostic tests used Number of animals tested Number of positive samples 1 *ELISA only 29173 24 2 western blot (WB) only 900 2 WB only 5 3 **WB and tonsil IHC 20524 * Excludes the 28 unconfirmed Biorad positives. ** Excludes 5 samples not tested by WB. tonsil IHC only WB and tonsil IHC 10 21

6.3 The methods used to analyse these data are presented in detail elsewhere (Gubbins and others, 2003). In the analysis we assume that: (i) the age distribution of animals sampled was the same as for animals over 18 months old in the 1997/8 abattoir survey; (ii) all positive samples came from animals over 36 months old; and (iii) all diagnostic tests are 100% specific. 6.4 Analysis of the results of the 2002/3 abattoir survey yields an estimate for the prevalence of infection of 0.33% (95% CI: (0.24, 0.44)). 6.5 By extrapolating from the results of the 2002/3 abattoir survey, we have estimated that the prevalence of infection in the GB population as a whole is 0.33%. 6.6 The differences between the prevalence estimate (0.33%) and the proportion of positive samples (0.20% for the head survey and 0.08% for the brainstem survey; Table 1) arise for a number of reasons. First, the sensitivities of the diagnostic tests used are less than 100% and, hence, not all infected animals will be detected. Second, the prevalence of infection is likely to be lower in older animals (such as cull ewes) and, thus, there is an apparent increase in prevalence when extrapolating to the population as a whole.

7. References Arnold, M., Meek, C.,Webb, C., Hoinville, L., (2002). Assessing the efficacy of a ram-genotyping programme to reduce susceptibility to scrapie in Great Britain. Prev. Vet. Med. Vol. 56 (3) pp. 227-249. Benestad, S.L., Sarradin, P., Thu, B., Schonheit, J., Tranulis, M.A. and Bratberg B. (2003) Cases of scrapie with unusual features in Norway and designation of a new type, Nor98. Veterinary Record. 153, 202-208. Gubbins, S., Simmons, M.M., Sivam, K., Webb, C.R. and Hoinville, L.J. (2003) Prevalence of scrapie infection in Great Britain: interpreting the results of the 1997/98 abattoir survey. Proceedings of the Royal Society of London Series B (published online 24/07/03, DOI: 10.1098/rspb.2003.2468). Moynagh, J, Schimmel, H and Kramer, G.N. (1999) The evaluation of tests for the diagnosis of Transmissible Spongiform Encephalopathy in Bovines. (http://europa.eu.int/comm/food/fs/bse/bse12_en.html) Moynagh, J and Schimmel, H. (1999) Nature. 400. 105. OIE (2000) Manual of Standards for Diagnostic Tests and Vaccines. Fourth Edition. Office International des Epizooties. Paris. pp. 873-880 Sivam, K., Baylis, M., Gravenor, M.B., Gubbins, S. and Wilesmith, J.W. (2003) Occurrence of Scrapie in Great Britain: Results of a postal survey in 2002. Veterinary Record (in press).

8. ANNEX 1. Sampling methods and test sensitivity 8.1 The sampling arrangements for the survey attempted to address both the statutory needs as demanded by Commission Regulation 999/2001 and scientific concerns that sampling of the brainstem only might lead to significant under ascertainment. As discussed in the cited references, any survey that is based upon detection of changes in the brain is potentially compromised if infected animals are sampled before changes are detectable in the brain. This applies particularly in younger sheep, but as indicated by the data in Table 7 is likely to be significantly influenced by genotype. 8.2 Genotype clearly influences incubation, presumably in part by determining the speed with which infectivity enters the CNS and the subsequent rate of replication, resulting in pathological effects which trigger the onset of detectable clinical signs. Genotype also determines the extent to which there is peripheral distribution of infectivity (or detectable PrP res ), especially in lymphoid tissue. Whereas PrP res can readily be detected in peripheral lymphoid tissue of infected VRQ/VRQ sheep in the first few months post infection, and before it is detectable in the brain, in other genotypes such changes are less easy to detect. 8.3 The head survey was therefore intended to investigate whether or not a survey based upon sampling and testing of peripheral lymphoid tissue would be more appropriate, and sensitive, than one that targeted brain tissue only. 8.4 Consequently, although Commission Regulation 999/2001 only required the retesting for confirmation of sheep that had tested positive by rapid test, samples of brainstem (obex), tonsil, and submandibular and retropharyngeal lymph nodes were tested for all sheep where tissue quality was suitable. All were tested by immunohistochemistry, the only test considered appropriate and interpretable for use on lymphoid tissue at the time of the survey. The cutting plan for the brainstem for the head survey is provided at Figure 1, the cutting plan for the brainstem survey is provided at figure 2.

Table 9. Comparison of the Prionics check test and IHC on additional tissues in the abattoir survey. Prionics Check IHC IHC Western retropharyngeal submandibular Blot IHC obex lymph nodes lymph nodes IHC tonsil Total Positive + + + + 17 + + - + 3 + - - - 5 + - - u 2 + u + + 1 28 WB Pos Negative + + + + 6 + + + - 1 + - - + 1 + - - - 4 - + + + 1 Modified Western Blot - + - + 2 15 SEAC Pos - - - - 20,128 - - - u 586 - - u - 135 - - u u 8 - u - - 131 - u - u 20 - u u - 14 - u u u 5 u - - - 27 u - - u 2 u - u - 1 u u - - 1 u u u u 109 nd nd nd nd 167 21,356 WB Neg IHC IHC retropharyngeal submandibular IHC obex lymph nodes lymph nodes IHC tonsil Total Negative - - - - 47 Not done - - - - 4 - - u - 1 21,429Total tested nd = Not Done u = Unsuitable for testing (ie sample missing or incorrect target area collected).

8.5 Of the total of 43 sheep considered to be positive, three were positive on lymphoid tissue while negative by Prionics Check and immunohistochemistry on the brainstem. This does not indicate that sampling of lymphoid tissue provide a truer estimate of the prevalence if tested instead of brain. The results do indicate, however, that if the objective of the surveillance is case finding then testing both brain and lymphoid tissues will improve the ascertainment of infected sheep. This result is most probably due to the genotype distribution of the population sampled, in which there was under-representation of VRQ/VRQ sheep. The higher frequency of heterozygote sheep, and in particular of sheep carrying one ARR allele, would have significantly reduced the chances of detecting peripheral lymphoid involvement. 8.6 Only 28 of the 43 IHC positive sheep were positive by the Prionics Check WB in the abattoir survey. For fallen stock the discrepancy was smaller at 11 WB positives out of a total of 12. At face value the results on abattoir samples might indicate that this test is inappropriate for surveillance for scrapie in small ruminants, being significantly less sensitive than testing by immunohistochemistry. This is in contrast to almost equivalent sensitivity when used on brain from clinically affected scrapie suspects at the VLA. Some consideration must however be given to the extent to which the sampling methodology compromised apparent test sensitivity. 8.7 The preservation of the obex for fixation and immunohistochemistry, the preferred approach to scrapie diagnosis in the UK, and in compliance with OIE guidelines (OIE, 2000) inevitably meant that this key target site was not available for testing by Prionics Check WB. Consequently, any significant variation in PrP res concentration within the brainstem may have compromised comparison of test sensitivity. While in the clinically affected sheep it is probable that PrP res concentrations in rostral and caudal medulla are close enough to those at the obex to ensure compatability, in the pre-clinically infected animal, at an unknown stage of incubation, levels may be below the limits of detectability with this rapid test. This may explain the closer correlation in the fallen stock survey which agrees with previous work done at the VLA on clinical. The fallen stock may represent clinically affected scrapie that were unrecognised while alive. 8.8 Consequently, for a direct comparison of test performance on preclinically infected sheep it is essential to ensure that both tests are applied to the same target area, preferrably the obex.

8.9 Whether or not the sampling arrangements have also influenced the outcome of the brainstem survey remains to be determined, but the outcome is somewhat surprising. The prevalence of 0.08% is significantly lower from that found in the head survey (0.2%). Despite the inappropriateness of pooling of results for the head survey (Prionics WB+IHC on brain+ihc on lymphoid tissue),it was expected that the greater analytical sensitivity of the BioRad ELISA (Moynagh and Schimmel, 1999; Moynagh, Schimmel and Kramer, 1999)) would compensate, and that the outcomes of both parts of the survey would be similar. That proportionately fewer infected animals were detected by the use of the BioRad ELISA was a surprise but may in part be due to the same factors identified with regard to the Prionics Check test. In other words, the ELISA test was not conducted on the preferred target site in a pre-clinically infected animal, and the test s greater analytical sensitivity did not compensate for this. Alternatively it must be recognised that along with the other rapid tests used for this surveillance in small ruminants, the tests have not been specifically evaluated for use on sheep. It may therefore be that the BioRad Platelia test, developed for use in bovines, is less effective when used on sheep. This is a possibility that the company recognises to the extent that it has developed a format specifically for sheep that has still to be evaluated and approved. 8.10 Unfortunately in the absence of routine testing of the obex by immunohistochemistry on ELISA-negative samples, it is not possible to confirm that there was a problem with the sensitivity of the test. This assumption comes only from comparison of the outcome of the two components to the survey, given that the selection process in the abattoir was the same. 8.11 Finally, as has been highlighted by the recent publication by Benestad et al (2003), there may still be occasional strains or isolates that will not be detectable by testing of the obex, simply because that is not the primary target site of pathological changes in the brain. It is difficult to design a sampling regime that is compatible with detecting the unusual, while also ensuring minimal disruption to the abattoir at the point of sampling. In particular it would require the collection of whole brain, and examination of multiple target areas before the presence of a strain with unusual targeting of pathology could be recognised. This may be disproportionate to the perceived dangers that such animals may represent, but may be the only way to fully understand the full range of case definitions that apply to scrapie, or TSEs of small ruminants. Nevertheless, it should be stressed that the phenomena described earlier in this annex do not represent deviations from the case definition of scrapie. They should not be assumed

to represent unusual isolates such as Nor98. They simply reflect the dangers of over-interpreting data from multiple diagnostic tests on limited numbers of samples, especially where tests have been carried out on different target areas.

Figure 1. Cutting plan for the brainstem in the head survey Submitted to Weybridge for Modified Prionics western blot for molecular strain typing and confirmation Standard Prionics Check Western blot used for primary test

Figure 2. Cutting plan for the brainstem in the brainstem survey Hemisection of caudal medulla enabled primary test on right half by ELISA, and a repeat on the left half if first test positive.