Materials and Methods

Animals—Sheep included in the study originated from sheep farms across the United States. Although typically not detected in sheep < 14 months of age,¹² results of another study¹³ suggested the possibility of detecting PrP^Sc in RAMALT of younger sheep. Therefore, sheep were enrolled in the study when they were r 4 months of age and were considered at high risk of having scrapie (ie, had clinical signs consistent with scrapie, were offspring of scrapie-positive ewes, were exposed at birth to the lambing of a scrapie-positive sheep or to a premises on which a scrapie-positive sheep was born, or were genetically susceptible and had resided in flocks that had contained scrapie-positive sheep). Genetic susceptibility to scrapie was confirmed via detection of the following PNRP genotypes: homozygous for glutamine at codon 171 (QQ₁₇₁) or heterozygous for alanine and valine at codon 136 and heterozygous for glutamine and arginine at codon 171 (AV₁₃₆QR₁₇₁). The latter requirement pertained only to sheep epidemiologically linked to infected sheep with a PNRP genotype of AV₁₃₆ or VV₁₃₆.

Tissue and data collection for each sheep was conducted by the veterinarians and veterinary technicians responsible for implementation of the NSEP in the geographic region of the sheep. Sampling methods were opportunistic; that is, sheep meeting inclusion criteria were enrolled in the study when they were selected for depopulation by the NSEP and when the owner and veterinary official agreed to participate in the study. The study was performed in compliance with institutional guidelines of the USDA, APHIS, Veterinary Services. Animal procedures were approved by the National Veterinary Services Laboratories Institutional Animal Care and Use Committee.

Data collection—Veterinarians completed standard regulatory USDA laboratory submission forms to collect information on each sheep, including sheep and flock identification number, state of residence, sex, breed, and age. To complete these forms, veterinarians used clinical and historical information obtained from flock records or the owner. Veterinarians examined sheep for clinical signs of scrapie including pruritus, wool loss, ataxia, anxious behavior, hyperesthesia, bruxism, tremor, and emaciation. The following information was also recorded for each sheep: PNRP genotype at codons 136 and 171, detection and characteristics of clinical signs consistent with scrapie, and any complications that developed following the RAMALT biopsy procedure. Genotype was determined via examination of records of genotyping performed as part of flock investigations conducted by the NSEP.

Tissue sampling—Five types of tissues were collected from each sheep for IHC evaluation. Biopsy specimens of RAMALT were obtained from the RV (ie, 3 to 5 o'clock), LV (ie, 7 to 9 o'clock), and LD (ie, 10 to 11 o'clock) positions of the rectum. One biopsy specimen of third-eyelid lymphoid tissue was collected from each eye. Postmortem tissue samples of medulla oblongata at the level of the obex, retropharyngeal lymph node, and palatine tonsil were also obtained. Specimens of RV and LV RAMALT and third-eyelid tissues were obtained from live sheep; however, specimens of LD RAMALT were collected postmortem.

Rectoanal mucosa–associated lymphoid tissue comprises lymphoid tissue that circumferentially occupies the mucosa of the anal mucocutaneous junction and spreads rostrally, occupying the most aboral 1 to 2 cm of rectal mucosa.¹⁴ Prior to collecting biopsy specimens of RAMALT, ophthalmic 0.5% proparacaine hydrochloride or 2% lidocaine gel was administered topically to the rectum of each sheep. Biopsy specimens were collected from the LV position and then the RV position. To make the rectal mucosa visible, abaxial or lateral pressure was placed on the perianal region or a rectal speculum was used. To aid visualization, headlamps or chemical restraint were sometimes used. To perform biopsies, the rectal mucosa at the mucocutaneous junction was retracted caudally with rat-tooth forceps. The initial cut, performed with scissors, began at the anal mucocutaneous junction (junction of nonhaired skin of the anus and rectal mucosa) and extended 1 cm cranially by superficially undermining mucosa, producing a 1.5 × 2-cm oblong biopsy specimen. The specimen was rolled out flat, mucosal side down, within a sponge-lined histology cassette.

Biopsy specimens of third-eyelid tissue were collected as described elsewhere.⁸ Briefly, ophthalmic 0.5% proparacaine hydrochloride was applied to conjunctiva topically. The third eyelid was retracted with rat-tooth forceps, and lymphoid tissue was collected from the bulbar surface with Metzenbaum scissors. Specimens were laid flat, with bulbar surface facing down within a sponge-lined cassette.

Sheep were humanely euthanized within 4 weeks after collection of antemortem tissue specimens. Postmortem tissue collection began within 24 hours after euthanasia. The caudal rectum was removed, opened longitudinally, and examined for any gross abnormalities including, but not limited to, rectal prolapse, cellulitis, rectal stricture, hemorrhage, perirectal abscess, and peritonitis. Samples of rectal mucosa resembling those collected from live sheep were collected from the LD position of the rectum.

Immunohistochemical testing for PrP^Sc—All tissues collected for IHC evaluation were immersed in neutral-buffered 10% formalin and fixed for at least 48 hours. Formalin-fixed tissues were processed conventionally. Care was taken to embed RAMALT and thirdeyelid specimens flat within paraffin wax, such that the surface composed of rectal mucosa or bulbar conjunctiva was positioned in the same plane as the surface of the block. Sections of RAMALT specimens were obtained at a depth of 200 to 400 Mm inward from the surface of the paraffin-embedded tissue block. Tissue samples were sectioned at 5 Mm, mounted on positively charged glass slides, and air-dried overnight. Slides were rehydrated with xylene and graded alcohols. For antigen retrieval, sections were treated with 95% to 98% formic acid for 5 minutes and then washed in Tris buffer. Next, slides were placed in modified citrate buffer^a and were autoclaved at 120°C for 20 minutes in a medical pressure cooker.^b,^c

Immunohistochemical testing was performed by use of an automated immunostainer.^d Slides were incubated with the primary antibody,^e a mouse monoclonal 99/97 anti–prion protein antibody, for 32 minutes. The remaining procedure was performed by means of a commercially available alkaline phosphatase immunostaining technique,^f according to manufacturer's instructions. Slides were counterstained with hematoxylin. Positive control tissues, consisting of obex and retropharyngeal lymph node from a scrapie-positive sheep, were prepared with each set of 20 slides.

Immunohistochemically stained tissues were evaluated by 1 of 4 pathologists experienced in TSE diagnosis. For the purposes of the study, a lymphoid follicle was defined as a discrete or expansile aggregate of mononuclear cells that ranged in size from approximately 250 Mm to > 1,500 Mm. The result for any tissue was considered as positive for scrapie when globular red reaction product within the cytoplasm of tingible body macrophages or fine granular red reaction product within a germinal center of a lymphoid follicle or within gray matter of the obex was identified within a tissue section. If no signal was identified within a section, the test result for that tissue was considered negative for scrapie. Results from sections of lymphoid tissue lacking scrapie-specific immunostaining and containing < 6 lymphoid follicles were recorded as insufficient lymphoid follicles for determination and deemed inconclusive. Tissues that were positive for scrapie were reevaluated by and results were confirmed by a second pathologist.

Diagnostic grade—Approximately 10 months after the study was initiated, it became apparent that specimens of RAMALT that contained insufficient lymphoid follicles for determination were common among specimens of RAMALT that contained a substantial proportion of mucosa composed of nonkeratinizing stratified squamous epithelium. Because squamous mucosa exists caudal to the rectoanal junction, it was decided that such specimens had been obtained from an inappropriate area. A grading scheme was devised to help determine the association of caudal biopsy location (as evidenced by the presence of squamous mucosa) with the presence of insufficient follicles. Grade 1 was assigned to RAMALT specimens that contained no squamous mucosa. Grade 2 was assigned to RAMALT specimens in which < 50% of the mucosa comprised squamous epithelium, and grade 3 was assigned to those in which r 50% of the mucosa comprised squamous epithelium.

Classification of test results and scrapie status—Test results for all tissues were dichotomized into positive or negative results. When an inappropriate location of obex had been sampled (ie, area of medulla oblongata not containing the dorsal motor nucleus of the vagus nerve), the result was excluded from analysis (ie, the result was treated as a missing value). When a lymphoid tissue was deemed to contain insufficient follicles, these results were considered inconclusive and were excluded from test sensitivity analysis (ie, these data were treated as missing values).

Dichotomous test results were created for parallel interpretations of results for RV and LV RAMALT specimens, right and left third-eyelid specimens, and RV and LV RAMALT specimens and right and left third-eyelid specimens. For parallel interpretation of test results, when 1 or more biopsy specimens were deemed positive for PrP^Sc, the result was considered positive. When none of the results for a given specimen were positive, the result was considered negative. Results from parallel interpretations were treated as missing values when 1 or more of the required tests were not completed or when all required specimens contained insufficient follicles.

Scrapie status of sheep was determined by use of dichotomized results of the reference test (a gold standard). The reference test consisted of parallel interpretation of results of IHC testing for PrP^Sc in obex and tonsil, retropharyngeal lymph node, or both. Scrapie was considered confirmed in sheep (scrapie-confirmed sheep) when 1 or more of these tissues had positive test results. Scrapie was considered as not being confirmed when all evaluated tissues lacked positive test results.

Lymphoid follicles—For each sheep in which scrapie was confirmed, the total number of lymphoid follicles and the number of lymphoid follicles containing scrapie-specific immunostain were counted in each specimen of RAMALT and third-eyelid tissue. The number of follicles in specimens from the right and left eyelids were added together, producing a value for total eyelid lymphoid follicles evaluated.

Statistical analysis—Descriptive and analytic statistics were obtained by use of commercially available software.^g For all analyses, values of P < 0.05 were regarded as significant. Data for sheep were included in statistical analyses when scrapie status was verified via the reference test. Verification was considered complete when test results were obtained for obex tissue in addition to retropharyngeal lymph node or palatine tonsil tissue. Verification was considered incomplete when obex tissue or both retropharyngeal lymph node and palatine tonsil tissues were unavailable.

Sheep were classified into breed groups according to face color. Crossbred sheep were classified according to the breed that was considered most representative. When features of a breed did not predominate or were not recognized, crossbred sheep were classified by face color and treated as crossbred sheep.

For continuous variables (ie, age and follicle number), normality of data was evaluated by means of an Anderson-Darling normality test and normal probability plot. A Wilcoxon rank sum test was used to evaluate the difference in median age between sheep according to confirmed scrapie status and test results. For each test evaluated, sensitivity was calculated as the proportion of scrapie-confirmed sheep that had positive test results. Exact 95% confidence limits for the sample proportions were estimated by use of a binomial distribution. Scrapie-confirmed sheep were cross-classified according to the dichotomous results of 2 tests: the McNemar χ² test, which accounts for paired samples, was used to test for differences in sensitivity between 2 biopsy sites; the Pearson χ² test was used to test for differences in the proportion of positive RAMALT results between breed groups (black-face, white-face, and other), age groups (< 2 years, 2 to 5 years, and > 5 years), and PNRP genotype at codon 136 groups (AA₁₃₆ and not AA₁₃₆).

The Pearson χ² test was used to test for associations of diagnostic grade and age group (< 2 years or r 2 years) with dichotomized test results (for specimens with insufficient lymphoid follicles vs specimens with sufficient follicles). The McNemar χ² test was used to test for the difference in proportions of specimens with insufficient follicles between biopsy sites. The difference in median total number of lymphoid follicles per specimen between biopsy sites was evaluated by use of a Wilcoxon signed rank test. The difference in median total number of follicles per biopsy between age groups was tested by use of a Wilcoxon rank sum test.

Results

Animals—From November 2006 through October 2007, 774 sheep were enrolled in the study, of which 762 were included in the statistical analysis. Of the 12 sheep not included, 5 were missing a test result for obex or both tonsil and retropharyngeal lymph node, and 7 were sheep that could not be confirmed as meeting criteria for enrollment. For the purpose of this report, sheep included in the analysis are referred to as the study population.

The study population comprised 717 females and 45 males and represented 13 breeds and 7 different PNRP genotypes (Table 1). Of sheep enrolled in the study with a known PNRP genotype, 90.3% had polymorphisms associated with increased susceptibility to scrapie (QQ₁₇₁ or AV₁₃₆QR₁₇₁). Tissue specimens for IHC testing were collected by approximately 42 veterinarians or veterinary technicians and were obtained from 71 flocks of 22 states.

Scrapie was confirmed by the reference test in 139 (18.2%) sheep. These scrapie-confirmed sheep originated from 22 of 71 flocks (31.0% of flocks). Of 124 scrapie-confirmed sheep with known PNRP genotype, 111 (89.5%) had AA₁₃₆, and 9 (7.3%) had AV₁₃₆. Of 124 scrapie-confirmed sheep with known PNRP genotype, all but one (99.3%) had classical scrapie and had a genotype of QQ₁₇₁. The remaining sheep had a nonclassical strain of scrapie (Nor98-like)¹⁵ and had a genotype of AA₁₃₆RR₁₇₁.

Of 10 sheep with clinical signs consistent with scrapie, 3 were scrapie confirmed. These sheep had ataxia, weakness, and pruritus. Most of the 7 sheep with clinical signs that were not scrapie confirmed were suspected to have scrapie on the basis of thin body condition, wool loss, or both, which are general signs of unthriftiness and not specific to scrapie or any particular disease.

The median age of sheep was 3.0 years (range, 0.3 to 11.0 years) for the study population and 3.0 years (range, 1.0 to 10.0 years) for scrapie-confirmed sheep. The proportion of scrapie-confirmed sheep varied significantly (P < 0.01) among age groups, and the highest proportion of scrapie-confirmed sheep was among those aged 2 to 5 years (Table 1).

Complications associated with the RAMALT biopsy procedure—Biopsy specimens of RV and LV RAMALT were collected from 582 (76.5%) sheep on the same day that postmortem LD RAMALT specimens were obtained, precluding detection of potential complications associated with the biopsy procedure. The median interval between antemortem and postmortem specimen collection was 0 days (range, 0 to 28 days).

For the 180 sheep that were not euthanized on the same day as RAMALT was biopsied, complications potentially associated with the RAMALT biopsy procedure were reported by veterinarians for 3 (1.7%) sheep. One sheep developed a rectal prolapse following the biopsy. Substantial rectal hemorrhage was observed in 1 sheep, and the biopsy procedure perforated the rectum of another sheep.

Diagnostic test sensitivity—Sensitivity estimates for IHC testing of RAMALT and third-eyelid biopsy specimens for PrP^Sc and contingency tables summarizing dichotomized test results were compared with scrapie status as determined by the reference test (Table 2). Sensitivity estimates for tests of RV and LV RAMALT specimens were stratified by breed class, age group, and PNRP genotype at codon 136 (Table 3). There was no significant difference between sensitivities for tests of RV RAMALT and LV RAMALT (P = 1.0) or between sensitivities for tests of RV RAMALT and postmortem LD RAMALT (P = 0.26).

Biopsy specimens of RV and LV RAMALT and right and left third-eyelid tissue were tested for PrP^Sc in 46 scrapie-confirmed sheep, allowing comparisons between the sensitivity of these 2 tests. Of these sheep, 5 (10.9%) were deemed positive for scrapie via parallel interpretation of results for RV and LV RAMALT and negative via parallel interpretation of results for right and left third-eyelid specimens. In contrast, none of the sheep were deemed positive via parallel interpretation of results for right and left third-eyelid specimens and negative via parallel interpretation of results for RV and LV RAMALT. For the same 46 sheep, when the sensitivity of the IHC test for PrP^Sc IHC was compared between each tissue type, the sensitivity was significantly (P < 0.05) higher for results for RV and LV RAMALT considered in parallel, compared with results for right and left third-eyelid specimens considered in parallel.

False-negative results (relative to the reference test) from IHC tests of RV and LV RAMALT that were considered in parallel were evident for 16 of 133 (12.0%) scrapie-confirmed sheep that were tested by use of RV and LV RAMALT. Eight of the 16 sheep were deemed positive for PrP^Sc when obex was tested but were negative for scrapie when other tissues were tested (3 specimens of RAMALT, 2 specimens of third-eyelid biopsies, and tissue from retropharyngeal lymph node and tonsil). One of the 8 sheep that had a positive test result for PrP^Sc in obex tissue but a negative result for PrP^Sc in lymphoid tissue was an 8-year-old Suffolk with a genotype of AA₁₃₆RR₁₇₁. The lymphoid tissues evaluated in this sheep included 3 specimens of RAMALT and 1 each of retropharyngeal lymph node and tonsil; no third-eyelid biopsies were collected. The microanatomic distribution of immunostaining for PrP^Sc (in a section of obex, immunostaining was limited to trigeminal ganglia)¹⁵ and the distinct glycotype profile of PrP^Sc identified with western blot¹⁶ (data not shown) were consistent with infection with a Nor98-like strain of scrapie agent. The PNRP genotype of the remaining 7 obex-positive, lymphoid tissue–negative sheep was QQ₁₇₁; information regarding the genotype of the 136 PNRP codon was available for 3 of these sheep and was AA₁₃₆.

Uninterpretable test results—There were significant differences in proportions of tissue sections with insufficient lymphoid follicles for determination of scrapie status between LV RAMALT and LD RAMALT biopsies (P < 0.001) and between RV and LV RAMALT considered in parallel and right and left third-eyelid tissues considered in parallel (P < 0.001; Table 4). There was no significant difference in proportions of tissue sections with insufficient follicles between RV RAMALT and LV RAMALT (P = 0.09) or between LV RAMALT and right and left third-eyelid tissues considered in parallel (P = 0.85). The proportion of tissue sections with insufficient follicles was significantly (P < 0.001) higher in sheep with RAMALT that had a diagnostic grade of 3, compared with that for sheep with specimens that had a diagnostic grade of 1 or 2 (Table 5).

Total number of lymphoid follicles in biopsy specimens from scrapie-confirmed sheep varied among tissues (Table 4). The difference in the median total number of follicles between specimens of RV RAMALT and LD RAMALT was significant (P < 0.05), as was the difference between median total number of follicles in RV and LV RAMALT specimens considered together and right and left third-eyelid specimens considered together (P < 0.001). There was no significant difference in median total number of follicles between RV and LV RAMALT (P = 0.70) or between LV RAMALT (P = 0.95) or RV RAMALT (P = 0.98) and right and left third-eyelid tissues considered together.

For biopsy specimens of RV and LV RAMALT, the proportion of specimens with insufficient lymphoid follicles for determination was significantly (P < 0.01) greater for sheep r 2 years of age, compared with that for sheep < 2 years of age. Furthermore, median age of sheep with insufficient follicles was significantly (P < 0.001) higher than that in sheep with sufficient follicles. The median age for sheep with inconclusive RV or LV RAMALT test results was 4 years, whereas the median age for sheep with conclusive RV or LV RAMALT test results was 3 years. The median age for sheep with inconclusive third-eyelid test results was 3.5 years, whereas the median age for sheep with conclusive third-eyelid test results was 2 years.

The median number of follicles in RV RAMALT biopsy specimens from scrapie-confirmed sheep was 44 in sheep < 2 years of age and 22 in sheep r 2 years of age. The median number of follicles in LV RAMALT biopsy specimens from scrapie-confirmed sheep was 73 in sheep < 2 years of age and 23 in sheep r 2 years of age. The median number of follicles in 2 third-eyelid biopsy specimens from scrapie-confirmed sheep was 29 in sheep < 2 years of age and 15 in sheep r 2 years of age.

Median age of sheep that had right and left thirdeyelid tissues with insufficient follicles was significantly (P < 0.001) higher than that of sheep with right and left third-eyelid tissues with sufficient follicles. Although the median total number of follicles in RAMALT and third-eyelid tissues was higher in sheep < 2 years of age than that in sheep r 2 years of age, this difference was only significant for RV (P < 0.001) and LD RAMALT (P < 0.0001) and not for right and left third-eyelid tissues combined (P = 0.11).

Discussion

When applied to biopsy specimens of RAMALT, IHC testing for PrP^Sc was a sensitive method for diagnosing preclinical scrapie in live sheep. A study¹⁰ in which IHC testing for PrP^Sc in postmortem specimens of RAMALT was evaluated in experimentally and naturally infected sheep revealed that the sensitivity of the test was between 86.0% and 97.1% (relative to parallel IHC testing of tissues from obex, retropharyngeal lymph node, tonsil, and ileum), depending on whether the sheep had clinical signs of scrapie. In the study reported here, sensitivity of IHC testing for PrP^Sc in RAMALT collected from live sheep was between 85.3% and 89.4%, depending on the site from which RAMALT was obtained.

Sensitivity of the test did not differ among the 3 RAMALT sites; however, statistical power for these comparisons was probably very low given the limited number of scrapie-confirmed sheep in the study population. For the purposes of our study, when a specimen of RAMALT or third-eyelid tissue contained < 6 follicles and lacked scrapie-specific immunostaining, test results were considered inconclusive because there was too little tissue to detect disease. Inconclusive test results were not used when estimating sensitivity or when conducting sensitivity analyses. Therefore, individuals who use RAMALT to test for scrapie need to be aware that if inconclusive test results are interpreted as negative test results, the resultant RAMALT sensitivities would be significantly lower than the estimates provided herein.

The proportion of false-negative test results (1 – sensitivity) for specimens of RV and LV RAMALT tested in parallel was 12% (n = 16). Eight of the 16 sheep that were deemed falsely negative for PrP^Sc had positive test results in brain tissue only and had negative results for all lymphoid tissues tested. One of these sheep that had positive test results for brain tissue only had a Nor98-like strain of scrapie agent, and sheep infected with that strain do not accumulate PrP^Sc in lymphoid tissues.¹⁵ In classic scrapie, accumulation of PrP^Sc in lymphoreticular tissue usually precedes its detection in CNS tissue.^{17, 18} However, as evident in 7 sheep of the present study, accumulation of PrP^Sc in neural tissue, without lymphoreticular tissue involvement, occurs in sheep with preclinical and clinical classic scrapie.^13,19–21 These findings underscore the importance of applying assays for PrP^Sc detection to more than 1 tissue type when determining the scrapie status of sheep. Because all of the remaining 8 sheep with false-negative results also had negative results for PrP^Sc in obex, it is possible that these sheep were at an early stage of disease in which PrP^Sc may not have reached detectable concentrations or may not have become widespread throughout RAMALT.

Host factors that are associated with susceptibility to disease, progression of disease, or expression of biomarkers for disease may provide the basis for variation in test performance between subpopulations of animals.²² Overlapping confidence intervals for all stratumspecific sensitivity estimates suggest that there was no significant difference in test sensitivity regardless of age or breed of sheep, identity of PNRP genotype at codon 136, or presence of clinical signs of scrapie. However, the small numbers of scrapie-confirmed sheep represented within each stratum of these variables limited the precision of stratum-specific sensitivity estimates. Additional research is needed to determine whether differences in sensitivity among strata of these variables exist.

The sensitivity of the IHC test for PrP^Sc in RAMALT appears to meet or exceed the sensitivity of the same test applied to third-eyelid tissue. The sensitivity of the IHC test for PrP^Sc in 1 third-eyelid biopsy specimen collected from a live sheep with subclinical scrapie is reportedly 72.5% (relative to the same test applied to postmortem specimens of obex).⁹ Because lymphoreticular tissue usually accumulates PrP^Sc at an earlier stage of disease than does CNS tissue,²³ the exclusion of retropharyngeal or tonsillar tissue from the reference test by the aforementioned study⁹ may have inflated the reported sensitivity estimate (ie, the sensitivity of the test on third-eyelid tissue may actually be < 72.5% because this estimate was based on comparison to a reference test with substandard sensitivity). Nonetheless, the present study revealed that the sensitivity of IHC test for PrP^Sc in 1 biopsy specimen of RAMALT was similar to the sensitivity of the same test applied to 2 specimens (right and left) of third-eyelid tissue, when results of the third-eyelid tests were considered in parallel.

When NSEP-relevant flock investigations are being conducted, postmortem testing is the preferred method for testing high-risk sheep and goats. In situations in which owners will not permit euthanasia, inconclusive results derived from both third eyelids present a dilemma because the possibility of repeat testing is restricted by the limited lymphoid content of that tissue. If highrisk sheep cannot be verified via antemortem testing to have not accumulated PrP^Sc, indemnification and postmortem testing are usually conducted. If a flock owner does not elect for indemnification, additional interventions are required to prevent the spread of scrapie while awaiting postmortem testing because the scrapie status of the flock cannot be evaluated.

The frequency of inconclusive results from 1 specimen of third-eyelid tissue collected from live sheep by an experienced technician has been estimated at between 20% and 38.5%.^{8, 9} In the present study, the frequency of obtaining inconclusive results for both right and left third-eyelid tissues was 23.7%. The frequency of obtaining inconclusive results from 1 specimen of RAMALT ranged from 13.1% to 23.8%, depending on the site from which RAMALT was obtained. The frequency of 2 simultaneously collected RAMALT specimens both yielding inconclusive results was 10.1%; this value was significantly lower than that for testing of 2 third-eyelid specimens. Therefore, RAMALT specimens appear to be less likely to yield inconclusive results than do third-eyelid specimens. In addition, should inconclusive results be obtained, RAMALT can be collected repeatedly. Because inconclusive results were likely among RAMALT specimens that contained substantial squamous mucosa (indicating they had been collected from an area of the rectoanal interface that was caudal to the position of RAMALT), an appropriate biopsy technique would help reduce the occurrence of inconclusive results. Use of a headlamp, proper sheep restraint, topical anesthetic, and rectal speculum may improve visualization of the appropriate anatomic site for biopsy.

With respect to RV and RD RAMALT, inconclusive results occurred more frequently for specimens obtained from sheep r 2 years of age versus sheep < 2 years of age, and median total number of lymphoid follicles was higher in sheep < 2 years of age versus sheep r 2 years of age. Therefore, as reported for the third-eyelid specimens,⁸ RAMALT specimens may be most likely to produce suitable test material for diagnosis of scrapie in young sheep with presumably robust lymphoid tissue.

In the present study, complications associated with RAMALT collection were rare and could likely have been prevented with additional training and experience. Veterinarians and technicians are advised that to prevent rectal perforation or excessive hemorrhage, the rectal mucosa should be carefully elevated away from underlying submucosa by grasping only a small portion of mucosa with forceps before collecting the sample to ensure that RAMALT is collected superficially. The biopsy specimen should consist of mucosa only, and there should be no connective tissue attached to the underside of the tissue. To further reduce the risk of rectal hemorrhage, sheep should be kept confined and quiet during and immediately following the biopsy procedure.

During the study period, approximately 60% of eligible flocks under investigation by the NSEP were enrolled. On the basis of the distributions of age and PNRP genotype and of the prevalence of scrapie in the sheep used, we believe the target (high-risk) population of sheep was suitably represented in our study.²⁴ Caution must be used when attempting to generalize the test performance results presented herein to other populations because the test may perform differently in populations with different prevalences of scrapie or with differential expression of disease-associated host characteristics.²²

The frequentist approach to evaluating test performance, which contends that the probability of an event is defined by the frequency of that event according to previous observations, was used in the present study and is based on the assumption that the reference test has perfect accuracy (100% sensitivity and specificity). However, it is known that accumulation of PrP^Sc in any tissue does not develop until months to years after infection, presenting a substantial period in which infected sheep cannot be detected via IHC testing for PrP^Sc. Therefore, IHC testing applied to tissues from obex and palatine tonsil, retropharyngeal lymph node, or both is not 100% sensitive; true sensitivity is unknown. With the frequentist approach, sensitivity estimates for the evaluated test will not take into account imperfect sensitivity of the reference test. In the present study, if scrapie-infected sheep were undetected by the reference test, then they were also undetected by the test under evaluation (because none of the scrapie-nonconfirmed sheep of the study population had positive results when RAMALT or third-eyelid tissues were tested). Therefore, if the true sensitivity of the reference test was < 100%, then the true sensitivities of the tests evaluated in our study would be expected to be lower than estimated.

The sensitivity of tests evaluated in the present study may also be biased as a result of conditional dependence (ie, detection of PrP^Sc accumulation in RAMALT is associated with detection of PrP^Sc accumulation in specimens of obex, tonsil, and retropharyngeal lymph node).²⁵ Latent-class methods, which account for the lack of a gold standard reference test by modeling the true disease status as a latent (unknown) variable, may be used to estimate test sensitivity and can account for the effects of conditional dependence.^{26, 27}

Immunohistochemical testing for PrP^Sc in biopsy specimens of RAMALT appeared to be a sensitive approach for antemortem diagnosis of scrapie in sheep. Sensitivity at least approximated and appeared to exceed the only currently available antemortem test (IHC testing of PrP^Sc in third-eyelid tissues). In contrast to collecting third-eyelid tissues, obtaining biopsy specimens from RAMALT is relatively easy, and when RAMALT is collected from the appropriate location, few inconclusive test results occur. Use of RAMALT for scrapie testing is expected to improve the success and rapidity of the US NESP because scrapie status of flocks can be more expediently and accurately established, facilitating appropriate placement of interventions to control scrapie.