Use of Pooled Fecal Culture for Sensitive and Economic Detection of Mycobacterium avium subsp. paratuberculosis Infection in Flocks of Sheep

Abstract

Ovine Johne's disease, or paratuberculosis, occurs in many countries. In Australia, surveillance using serology is used as part of a control program, but the testing regime is costly relative to its sensitivity. For this reason, culturing of Mycobacterium avium subsp. paratuberculosis in fecal samples pooled from a number of sheep was evaluated. Initially, the effect of pooling on the sensitivity of fecal culture was evaluated using samples from 20 sheep with multibacillary paratuberculosis and 20 sheep with paucibacillary paratuberculosis, each confirmed histologically. All multibacillary cases and 50% of paucibacillary cases were detected by culturing of feces at a pooling rate of 1 infected plus 49 uninfected sheep. In a pilot-scale study in 1997, M. avium subsp. paratuberculosis was detected by pooled fecal culture on 93% of 27 infected farms which were identified originally based on history, clinical signs, and one or more rounds of testing using serologic and histopathologic examinations. Pooled fecal culture was compared with serologic examination for submissions from 335 farms where both tests had been conducted on the same sheep and was significantly more sensitive (P < 0.001). Computer simulation of random sampling indicated that the testing of 6 pools of 50 sheep would provide 95% confidence in detecting ≥2% prevalence of infection. The estimated laboratory cost of pooled fecal culture when applied as a flock test is approximately 30% that of serologic examination, and sample collection costs are lower. It is recommended that pooled fecal culture replace serologic examination for detection of M. avium subsp. paratuberculosis infection at the flock level.

Johne's disease, or paratuberculosis, is a chronic enteropathy caused by Mycobacterium avium subsp. paratuberculosis. Most herbivores are susceptible to the infection. Treatment is not considered to be economically justifiable and, once developed, the infection may eventually be fatal. The usual route of infection is fecal-oral, with young animals becoming infected by contact with infected adults or an environment contaminated by infected feces. Paratuberculosis is recognized globally in farm livestock, and disease control programs are being developed in many countries to deal with the problem. These programs are aimed at either control or eradication. For control, laboratory tests are often used to identify infected individuals, which are then culled to reduce transmission to young, susceptible animals. For eradication in Australia, whole flock or herd depopulation, management of the pasture to reduce contamination, and restocking with uninfected stock are used. Thus, it is necessary to diagnose infection at the flock or herd level in order to define the geographic extent of infection and to evaluate the regional prevalence of infection. Conversely, it is also necessary to identify uninfected flocks or herds as a source of replacement stock for farms that have eradicated infected livestock. Paratuberculosis usually occurs in a population at a relatively low prevalence, but there are two features of small-ruminant extensive-grazing systems in many countries that make herd diagnosis of paratuberculosis particularly difficult: large flock or herd sizes and low individual animal value. Disease testing must be conducted using epidemiologically valid sampling so as to detect a prescribed prevalence of infection (for example, 2%) with defined confidence (usually 95%). If a serologic test with an assumed sensitivity of 30% is used, then it is necessary to test more than 450 individuals, compared to about 300 for a test with a sensitivity of 50% (1, 11). Thus, there is a strong incentive for the development of sensitive, low-cost diagnostic tests, particularly tests that can be applied efficiently to define the status of whole herds or flocks.

Diagnostic tests for paratuberculosis include indirect tests for the host's immune response or direct microbiological tests for the organism. Tests for cell-mediated immunity are not yet sufficiently developed for routine use, and tests for anti-M. avium subsp. paratuberculosis antibody are insensitive. Estimates of the sensitivity of enzyme-linked immunosorbent assays (ELISA) and agar gel immunodiffusion (AGID) tests for sheep vary with the stage of the disease but have not exceeded 30% for unselected populations (7, 12). Antibody responses develop quite late in the disease process, by which time environmental contamination due to fecal shedding of M. avium subsp. paratuberculosis has commenced (2, 10, 14, 15). Although culture media suitable for ovine strains of M. avium subsp. paratuberculosis were developed only recently, it is already clear that culturing of the organism from feces of individual sheep is a more sensitive test for the infection than serologic testing and is able to detect infection at an earlier stage of the disease (2, 24, 25).

Paratuberculosis in sheep occurs in two forms, multibacillary and paucibacillary, distinguished by the number of acid-fast bacilli present in granulomatous lesions in intestinal tissues (4, 16). Recently, M. avium subsp. paratuberculosis was enumerated in the feces of sheep with multibacillary disease (26). The numbers were such that dilution of feces, even by several orders of magnitude, would be unlikely to affect the sensitivity of fecal culture for multibacillary disease, raising the possibility that pooled samples could be used for flock diagnosis (23). The aim of this study was to develop and evaluate a method for culturing of fecal samples pooled from a number of sheep in order to provide an economical test for M. avium subsp. paratuberculosis infection in flocks. Specific aims were to determine an acceptable rate of pooling of fecal samples, to compare the sensitivities of pooled fecal culture and an AGID test, to evaluate the practicality of sample collection, and to develop recommendations for sampling rates for confirmation of M. avium subsp. paratuberculosis infection in flocks.

MATERIALS AND METHODS

Experiment 1: determination of an acceptable rate of pooling for feces.

Feces from 20 sheep each with multibacillary and paucibacillary paratuberculosis were removed from storage at −80°C and thawed. Feces from the multibacillary cases had been stored for 12 to 15 months and were derived from sheep with multifocal coalescing to severe diffuse intestinal lesions containing large numbers of acid-fast bacilli. Feces from the paucibacillary cases had been stored for 12 to 20 months and were derived from sheep with small to multifocal coalescing intestinal lesions containing few (n = 13) or no (n = 7) acid-fast bacilli. The number and distribution of acid-fast bacilli in lesions were graded as follows: 0, no acid-fast bacilli; 1, individual or small numbers, limited foci; 2, small numbers, multiple foci; 3, moderate numbers, diffuse; and 4, large numbers, diffuse. There were likely to have been approximately 10⁸ viable M. avium subsp. paratuberculosis cells per g of feces in the sheep with multibacillary paratuberculosis, based on enumeration of samples from sheep with similar lesions (26). Each fecal sample was diluted pellet-wise at rates of 1 plus 9, 1 plus 49, and 1 plus 99 with feces from control sheep. The controls were housed sheep known to be free of paratuberculosis. Feces were mixed in an electric blender as described below. The original fecal samples and the dilutions prepared from them were cultured as described below.

Experiment 2: pilot-scale field evaluation of pooled fecal culture.

Pooled fecal samples (>10 to ≤50 sheep per pool) were solicited from veterinarians during routine surveillance for ovine paratuberculosis in 1997. In some instances, blood samples were collected for serology. An estimate of the prevalence of ovine paratuberculosis in a flock was made when possible, based on the number of mortalities attributable to ovine paratuberculosis per annum. A low-prevalence flock was defined as one where this mortality rate was ≤2%.

Experiment 3: large-scale field evaluation of pooled fecal culture.

Pooled fecal culture and serologic examination were compared in a large-scale regional survey for ovine paratuberculosis in 1998. Flocks were tested when there was a suspicion of paratuberculosis based on the presence of clinically affected sheep, when surveillance indicated movement of sheep from an infected farm to a farm of unknown status (trace-forward), when trace-backward investigation identified putatively infected farms, or when testing was undertaken to confirm freedom from disease. Random sampling rates were prescribed for serologic examination to enable the detection of 2% prevalence with 95% confidence, assuming a test sensitivity of 30%. Alternatively, biased sampling of sheep selected on the basis of age (old) or body condition (poor) was recommended, these conditions being considered likely to increase the number of infected animals in the sample (12, 13).

Experiment 4: modeling of sample size required to detect flock infection.

In experiment 2, pooled fecal samples were collected from all sheep ≥2 years old on 14 farms where ovine paratuberculosis was known to be present and flock size was >500 sheep. Sampling was simulated to determine the probability of detection of infection, depending on the number of pools sampled. For each submission, the number of pools that were tested and the number of pools that were culture positive were entered into a spreadsheet (Excel; Microsoft). The HYPGEOMDIST function was then used to simulate random sampling of 1 to 10 pools from the total number of pools for each submission and to calculate the probability of including at least one culture-positive pool at each level of sampling.

Collection of fecal samples.

For all experiments, veterinarians were requested to take one fecal pellet from the rectum of each sheep ≥2 years old and pool these in lots of up to 50 in a sterile screw-cap jar with pellets from sheep from the same mob, to change gloves between pools, to collect a blood sample only from sheep from which a pellet had been obtained, to refrigerate samples, and to forward them directly to the laboratory. Fecal samples were submitted via regional laboratories to the central laboratory at Elizabeth Macarthur Agricultural Institute (EMAI). The referring laboratories were asked to refrigerate samples if shipment could not be effected immediately, and samples were transported on wet ice. Upon receipt at EMAI, samples were placed at 4°C or at −80°C if they could not be processed within 4 days.

Homogenization of fecal samples.

Samples of pooled pellets were transferred to a sterile stainless steel vessel that had rotating metal blades (Waring MC-3, 50 to 250 ml; catalog no. 222452; Selby Sci, Mulgrave North, Victoria, Australia) and that was attached to an electric motor (Waring two-speed motor; catalog no. 222420; Selby). A piece of sterile heavy-gauge aluminium foil was used to cover the vessel. Pellets were thoroughly mixed to form a fecal homogenate. This involved several cycles of blending for approximately 5 s and scraping material from the sides of the blender with a sterile spatula. Other methods of mixing feces, including stomaching and grinding, were tried initially but required the addition of liquid and were less effective than blending. The homogenate was returned to the original specimen jar and placed at 4°C or at −80°C if further processing could not be undertaken within 2 days. Samples were thawed at 4°C overnight or at room temperature on the day they were used. To minimize sample-to-sample cross contamination, the homogenization step was done with a class II biosafety cabinet by an operator wearing latex gloves. The base of the cabinet was lined with an absorbent mat. After each group of samples from a given farm was processed, the absorbent mat was discarded, the operator changed gloves, and the surfaces of the cabinet were wiped with a phenolic disinfectant (Phensol; Whitely Industries, Rosebery, New South Wales, Australia). At the end of each blending run, usually each morning and each afternoon, a negative control fecal sample was subjected to the homogenization protocol and then included in the subsequent culture steps. The negative control fecal sample consisted of a pool of 50 fecal pellets that were collected from paratuberculosis-free sheep housed at EMAI and that were stored at −20°C. The class II biosafety cabinet was treated with UV radiation each morning and each afternoon, prior to each blending run.

Culturing and identification of M. avium subsp. paratuberculosis.

The fecal homogenate was cultured using both modified BACTEC 12B radiometric medium and modified Middlebrook 7H10 (7H10) medium with mycobactin J (MJ) as the primary media (24). The growth index was measured weekly for 12 weeks. A 0.2-ml sample was removed from each BACTEC vial for PCR when its growth index was first found to be ≥200, again when the growth index reached 999, and then again 1 week later. The sample was prepared for PCR using ethanol as described previously (24), but if the PCR result was negative, DNA was purified using a commercial kit (Wizard PCR Preps; Promega) and the sample was retested (25). In experiment 3, an additional sample of 0.5 ml was removed for subculture when the growth index was first found to be ≥200. In most instances, a sample of 0.7 ml was removed from the vial when the growth index was first found to be 999 because the growth index rapidly increased to this level; therefore, PCR and subculture were conducted with the same sample. For subculture, five 0.1-ml aliquots of BACTEC medium were spread on slopes of the following media: Lowenstein-Jensen medium, Herrold's egg yolk medium without pyruvate but with and without MJ, and 7H10 medium with MJ (7H10+MJ) and without MJ (24). These cultures were examined biweekly for up to 20 weeks. All incubations were done at 37°C. The purpose of subculture to Lowenstein-Jensen medium and Herrold's egg yolk medium without pyruvate was to identify the growth of organisms other than ovine M. avium subsp. paratuberculosis that might have contributed to the growth index, while the purpose of subculture to 7H10 medium was to evaluate MJ dependency as an aid to the identification of M. avium subsp. paratuberculosis. Colonies with morphology typical of ovine strains of M. avium subsp. paratuberculosis were removed from 7H10+MJ medium slopes and prepared for PCR as described previously (24).

A positive control fecal sample was included with each batch of cultures. This sample consisted either of a pool of approximately 5 g of feces from 7 sheep with paratuberculosis or of a pool of feces from 13 sheep with paratuberculosis that was mixed at a rate of 1 in 10 with negative control feces. Positive control feces were dispensed in aliquots of 5 ml and stored at −80°C until used.

PCR to detect the IS900 sequence and restriction endonuclease analysis (REA) of the PCR product with AlwI were conducted as described previously (24, 25). The amount of DNA was assessed from agarose gels stained with ethidium bromide.

A positive fecal pool was defined as one from which M. avium subsp. paratuberculosis was isolated in primary culture in BACTEC medium and/or in primary or secondary culture on 7H10+MJ medium and identified by both PCR and REA. An inconclusive pool was one in which PCR was positive for a product of the expected size (413 bp) but REA could not be done due to insufficient PCR product. A negative pool was defined as one from which M. avium subsp. paratuberculosis was not isolated.

Serologic and histopathologic tests.

Serum samples were tested using an AGID test with antigen from M. avium subsp. paratuberculosis strain 316V (20). Reactors in this test were subjected to postmortem examination, and their intestinal tissues were examined histologically (20). The tissues examined were the ileocecal valve and adjacent ileum, three additional sections of small intestine taken at 1-m intervals proximal to the ileocecal valve, the ileocecal lymph node, the caudal jejunal lymph node, cecum, proximal colon, and any tissues with gross lesions suggestive of ovine paratuberculosis. The results of the histologic test were defined as positive if there were granulomatous lesions together with acid-fast bacilli, inconclusive if there were granulomatous lesions without acid-fast bacilli, and negative if there were no serologic reactions or if there were no granulomatous lesions at these sites in a serologic reactor. The results for blood samples, histopathologic examinations, and fecal pools submitted from the same farm on separate occasions were combined to give a single set of results for each farm. A farm was considered infected if one or more histologically positive animals was identified.

RESULTS

Negative and positive control feces.

M. avium subsp. paratuberculosis was not recovered from any of the negative control fecal samples in any experiment but was isolated from every positive control fecal sample in each experiment.

Experiment 1: determination of an acceptable rate of pooling for feces.

When BACTEC 12B radiometric medium was used, all 20 multibacillary cases were detected at a dilution of 1 in 50, while 18 of 20 were detected at a dilution of 1 in 100. The rates of detection of the paucibacillary cases were lower, with only 10 being detected at the 1-in-50 dilution (Table 1). All the sheep with multibacillary disease had large numbers of acid-fast bacilli (grade 4) in their intestinal lesions. The six sheep with culture-negative paucibacillary disease had very small numbers of acid-fast bacilli in their lesions (grade 0 or 1), but some cases with similarly low intensities were detected at a dilution of 1 in 100. Primary culture on 7H10+MJ medium was less sensitive than that in BACTEC medium, particularly for paucibacillary cases (Table 1).

TABLE 1.

Effect of rate of dilution of feces from sheep with multibacillary (n = 20) and paucibacillary (n = 20) paratuberculosis on rate of isolation of M. avium subsp. paratuberculosis

Dilution rate	No. of sheep with a positive culture at each dilution tested with the following medium:
	BACTEC 12B radiometric		7H10+MJ
	Multibacillary	Paucibacillary	Multibacillary	Paucibacillary
Undiluted	20	14	16	5
1 in 10	20	12	10	3
1 in 50	20	10	6	2
1 in 100	18	9	5	2

A growth index of >200 was obtained in 3 to 5 weeks for the highest positive dilution for all the multibacillary cultures, compared to 4 to 6 weeks for the paucibacillary cultures; however, growth on the solid medium was delayed by many weeks.

Mixing of 50 fecal pellets was achieved readily using the blending device, and a pooling rate of up to 50 was chosen for all further experiments. At this dilution, the sensitivity of detection of multibacillary cases was 100% (95% confidence interval, 83 to 100%), while that for paucibacillary cases was 50% (27 to 73%).

Experiment 2: pilot-scale field evaluation of pooled fecal culture. (i) Comparison of pooled fecal culture with flock infection status.

A total of 57 submissions comprising 378 individual pools of feces from 51 independent farms were included in the evaluation. Farms were located in the Central Tablelands, Condobolin, Forbes, Goulburn, Gundagai, Molong, Moss Vale, Wagga, and Young regions of New South Wales (illustrated in reference 9). The number of pools submitted per farm ranged from 1 to 24, and the number of sheep per pool ranged from 11 to 50. There were 27 submissions (312 pools) from flocks in which paratuberculosis had been confirmed by means other than culture and in which the sheep sampled were considered likely to have been infected (flock status: infected), 19 submissions (33 pools) from flocks identified as suspect but in which paratuberculosis had not been confirmed (status: suspect), and 11 submissions (33 pools) from flocks of unknown status. Using flock status as the “gold standard,” infection was confirmed by pooled fecal culture for 93% of submissions (75% of pools) from infected flocks, 37% of submissions (21% of pools) from suspect flocks, and 9% of submissions (3% of pools) from flocks of unknown status.

(ii) Comparison of pooled fecal culture with serologic examination.

A comparison of pooled fecal culture with the AGID test was possible for 39 submissions where feces and serum were submitted in parallel from the same sheep. Infection was detected by pooled fecal culture on 14 farms, while serologic examination resulted in the detection of infection on only 4 farms. For 11 submissions, only pooled fecal culture was positive; for 3 submissions, both pooled fecal culture and serologic examination were positive; and for 1 submission, only serologic examination was positive. The difference between the rates of detection of infection for the two tests was highly significant (McNemar's test: chi-square value, 6.75; P < 0.01).

The one submission that was found positive by serologic examination with the AGID test but negative by pooled fecal culture was from a group of 300 sheep that were examined in more detail as part of an unrelated experiment. There was only one AGID test reactor, but this sheep was culture negative (sample from the terminal ileum), suggesting that it was a false-positive reactor in the AGID test. The flock was classified as infected only by culturing terminal ileum samples from three sheep that were selected after reaction in an experimental ELISA. All 300 sheep lacked histological lesions in the terminal ileum. The results confirm that the prevalence of infection in this flock was extremely low and that the infection was subclinical.

(iii) Sensitivity of pooled fecal culture in relation to the prevalence of ovine paratuberculosis.

Of the 27 submissions from known infected flocks, 13 were from farms which were estimated to have an annual mortality rate due to paratuberculosis of <2%; the estimated mortality rates were between 2 and 10% for 9 flocks and >10% for 5 flocks. Of the 13 low-prevalence flocks, all but 2 were pooled fecal culture positive, including 3 for which only one pool was submitted and 4 for which between four and eight pools were submitted. All 14 submissions from flocks with estimated mortality rates of >2% were pooled fecal culture positive.

Experiment 3: large-scale field evaluation of pooled fecal culture. (i) Samples.

Pooled fecal samples and blood samples were received from 296 farms in 26 Rural Lands Protection Board districts in New South Wales (illustrated in reference 9); the sample distribution (number of fecal pools cultured, number of farms represented) was as follows: Australian Capital Territory (21, 3), Armidale (34, 23), Bega (9, 1), Bombala (19, 3), Braidwood (9, 1), Cooma (73, 11), Central Tablelands (545, 106), Dubbo (14, 3), Forbes (15, 7), Glen Innes (2, 2), Goulburn (133, 26), Gundagai (64, 6), Hume (26, 5), Molong (33, 15), Moss Vale (26, 2), Mudgee-Merriwa (9, 1), Murray (8, 2), Narrabri (6, 5), Narrandera (4, 1), Tamworth (14, 4), Wagga (88, 14), Wentworth (3, 2), Yass (200, 27), and Young (180, 26). Overall, there were 1,535 fecal pools containing fecal pellets from approximately 70,000 sheep. The practicality of collection of fecal pellets was recorded for 246 of 343 submissions and was high in 61%, moderate in 34%, and low in 5%. The main reason given for low practicality of collection was yarding of sheep for prolonged periods, which resulted in many sheep having defecated, resulting in an empty rectum. There were relatively few examples where pools submitted did not contain pellets from 50 sheep.

By far, the largest number of samples came from the districts of Central Tablelands and Goulburn and the adjacent districts of Yass and Young, where ovine paratuberculosis was endemic, but a wide range of geographic areas was represented. Most of the culture-positive results were obtained from farms in areas where ovine paratuberculosis was endemic.

The most common numbers of pools submitted per farm were 1 (29% of farms) and 9 (27% of farms); however, the number of pools submitted per farm ranged from 1 to 11. Culture-positive outcomes were obtained with submissions of 1 to 11 pools, with 54% of culture-positive outcomes from farms submitting either 1 or 9 pools, corresponding to the most frequent sample sizes.

(ii) Culture results in relation to the method of sampling.

The method of selection of sheep was recorded for 205 of the 343 submissions and was biased for 99 and random for 106. The number of pools submitted ranged from 1 to 11 regardless of the method of sampling. There was no significant difference between random and biased sampling in the detection of infection using either serologic with histopathologic testing or pooled fecal culture for the most frequent submissions, namely, those with one pool or nine pools, or for all submissions regardless of the number of pools (Table 2). These data may be confounded by the reasons for submission, but these were not always recorded, and further analysis could not be done.

TABLE 2.

Detection of infection according to the method of sampling in experiment 3

Category	Resulta obtained with the following method of sampling:
	Biased	Random
1 pool
No. of submissions	41	18
Serologic with histopathologic test result	29.3	22.2
Pooled fecal culture result	36.6	27.8
9 pools
No. of submissions	15	37
Serologic with histopathologic test result	13.3	24.3
Pooled fecal culture result	33.3	43.2
1–11 pools
No. of submissions	99	106
Serologic with histopathologic test result	38.4	48.1
Pooled fecal culture result	35.4	47.2

^aResults for serologic with histopathologic tests and pooled fecal culture are shown as percentages of submissions with a positive result. P values, determined by the chi-square test, for biased versus random sampling were >0.05 for all comparisons. 

(iii) Agreement between the results of pooled fecal culture and those of serologic and histopathologic testing.

For the purposes of analysis and to provide a conservative classification, inconclusive pooled fecal culture test results were regarded as negative, while inconclusive serologic or histopathologic test results were regarded as positive. Using these criteria, 32.1% of the 296 farms were found positive by pooled fecal culture, and 23.3% were so found by serologic with histopathologic testing (Table 3). Given that culture is a specific test, pooled fecal culture was significantly more sensitive than serologic with histopathologic testing (McNemar's test with continuity correction: chi-square value, 14.9; P < 0.001).

TABLE 3.

Comparison of pooled fecal culture and serologic with histopathologic test results in experiment 3

Pooled fecal culture result	No. of farms with the following serologic with histopathologic test result:
	Positive	Inconclusive	Negative	Total
Positive	58	3	34	95
Inconclusive			9	9
Negative	5	3	184	192
Total	63	6	227	296

Of the five farms where sheep had positive serologic with histopathologic test results but negative pooled fecal culture test results, the lesions in the affected sheep on three farms contained small numbers of acid-fast bacilli. However, on one farm, there were large numbers of acid-fast bacilli in lesions.

Of the 34 farms with positive pooled fecal culture test results and negative serologic with histopathologic test results, two had serologic reactors which were not made available for histologic confirmation. Four of the farms were found to be infected in independent investigations prior to or within 6 months of the completion of experiment 3, three because of confirmed clinical cases of ovine paratuberculosis and one because of detection using serologic and histopathologic testing. Twenty-two farms were suspect, 12 because of tracing sheep movements forward or backward from a farm with confirmed infection and 10 because of a shared boundary with an infected farm. Further evidence suggesting that infection was present on these farms was that there were multiple BACTEC medium culture-positive pools from 12 farms: four culture-positive pools from 2 farms, three culture-positive pools from 2 farms, and two culture-positive pools from 8 farms. In addition to positive BACTEC medium cultures, cultures on solid medium were positive for 21 of the 34 infected farms.

Experiment 4: modelling of sample size required to detect flock infection.

The effectiveness of various levels of random sampling of sheep for the detection of infection was evaluated for 14 submissions from infected farms where all sheep ≥2 years old were sampled (Table 4). For 13 of these, random sampling of four pools provided 95% confidence of detecting at least one infected pool, while for the remaining submission (farm 1), random sampling of eight pools provided this level of confidence. The prevalence of infection on farm 1 was very low, and probably fewer than 1% of sheep were infected (19).

TABLE 4.

Simulated random sampling for the detection of one or more infected pools in experiment 4

Submission	No. of pools	No. (proportion) of pools positive	No. of pools needed for 95% confidence of detection
1	20	5 (0.25)	8
2	20	11 (0.55)	4
3	20	12 (0.60)	3
4	20	13 (0.65)	3
5	24	17 (0.71)	3
6	20	15 (0.75)	2
7	20	16 (0.80)	2
8	17	15 (0.88)	2
9	10	9 (0.90)	2
10	24	22 (0.92)	2
11	18	18 (1.00)	1
12	20	20 (1.00)	1
13	20	20 (1.00)	1
14	23	23 (1.00)	1

DISCUSSION

Compared to serology, the benefits of pooled fecal culture for flock diagnosis of ovine paratuberculosis are increased diagnostic sensitivity and a large reduction in cost. Pooling of samples is possible due to the massive numbers of M. avium subsp. paratuberculosis present in the feces of sheep with multibacillary disease. The average rate of excretion of M. avium subsp. paratuberculosis in such sheep was 1.09 × 10⁸ organisms per g of feces (26). As the analytical sensitivity of similar culture methods has been estimated to be 100 CFU per g of feces (3), the pooling rate could be much greater than that used in this study without a loss of detection of multibacillary cases, and it was not surprising that all 20 multibacillary cases were detected when feces were diluted 1 in 50. However, paucibacillary cases may comprise a significant proportion of the cases of paratuberculosis in a flock, and fecal culture for such cases was susceptible to the dilution effect caused by pooling of samples. In an earlier study of undiluted feces with the same culture methods, a lower isolation rate (48.4%) was reported for paucibacillary cases (24), confirming that the proportion of paucibacillary cases in a flock is important in determining the sensitivity of pooled fecal culture. Another factor limiting the pooling rate is the efficacy of mixing of pooled samples. It is essential to ensure that pooled samples are completely mixed, and this goal can be difficult to achieve. In this study, mixing was achieved most easily and completely using a high-speed blending apparatus.

Pooled fecal culture is a sensitive method for flock diagnosis of ovine paratuberculosis. In experiment 2, about 93% of 27 infected flocks were detected by pooled fecal culture, a remarkable outcome given that flock infection status was derived from lengthy historical information, often based on the results of sequential field and laboratory investigations. In many instances, a positive culture result was obtained from a single pool of 50 fecal pellets. Overall, pooled fecal culture appeared to be considerably more sensitive than serologic examination for detection of flock infection. In experiment 2, more than three times as many flocks were detected by pooled fecal culture than by serologic examination of the same sheep, while in experiment 3, with a much larger sample size, about 1.5 times as many farms were detected by pooled fecal culture than by serologic examination.

The absolute value for the diagnostic sensitivity of pooled fecal culture is not known but lies somewhere between the sensitivities of detection of individual multibacillary cases (83 to 100%) and paucibacillary cases (27 to 73%) when feces are pooled at a rate of 1 in 50. Thus, the probability of detection of flock infection is very high if multibacillary cases are present in a pool. In a separate study, excretion of M. avium subsp. paratuberculosis by sheep with multibacillary disease was shown to be continuous, so that one would expect a single fecal sample from such sheep to be satisfactory (26). The proportion of multibacillary cases would be expected to vary from flock to flock and would be affected by factors such as time since introduction of infection to the flock, contamination levels in the pasture, and age of sheep but may reach 70% of clinical cases (4).

The data obtained in experiments 1 and 4 permit analysis of the sample sizes required to detect infection using pooled fecal culture. Using conservative values of sensitivity of 95 and 40% for multibacillary and paucibacillary cases, respectively, and a 20:80 distribution of multibacillary and paucibacillary cases in a flock, the average sensitivity value is (0.95 × 0.2) + (0.4 × 0.8) = 0.51. Using an estimate of 50% for the average sensitivity of pooled fecal culture at the individual animal level, a sampling rate of 300 animals (six pools of 50) would provide a level of assurance exceeding 95% confidence of detecting a prevalence of infection of ≥2%. In experiment 4, the sample sizes required to detect flock infection were evaluated by statistical modelling of randomly sampled pools for 14 known infected flocks. Culturing of four pools was needed to be 95% certain of detecting infection in 13 flocks, and culturing of eight pools was needed for the other flock, which had a low prevalence of infection.

Despite pooled fecal culture having higher sensitivity than serologic examination, the results of this study suggest that it may be difficult to diagnose infection with pooled fecal culture in flocks with a low prevalence of sheep with ovine paratuberculosis. Several low-prevalence flocks were included in experiment 2, and pooled fecal culture failed to detect infection in one. Infection was confirmed in this flock only by intensive investigation conducted during another research trial. Three of 20 sheep in this flock were selected based on weak reactivity in an ELISA for anti-M. avium subsp. paratuberculosis antibodies, and 1 was found culture positive (intestinal samples) in the absence of histologic lesions of paratuberculosis. Despite these observations, in experiment 3, more than 30 flocks were found to be infected by pooled fecal culture, despite an apparent lack of clinical and serologic evidence of M. avium subsp. paratuberculosis infection. There were objective reasons to believe that these flocks were infected, and it is likely that many had either a very low prevalence of infection or sheep only in the early stages of the disease. Several have since been confirmed infected. Conversely, several flocks that were known to be infected based on serologic and histopathologic test results were not detected by pooled fecal culture (1 of 14 infected farms in experiment 2 and 5 of 63 infected farms in experiment 3). Thus, no single test at a single point in time can be used to rule out the presence of M. avium subsp. paratuberculosis infection in a flock of sheep. A high level of assurance that a flock is not infected with M. avium subsp. paratuberculosis can be gained only after successive negative tests.

M. avium subsp. paratuberculosis is an obligate pathogen and parasite of animals that is identified by slow growth, mycobactin dependency (22), and the presence of IS900 (6). The specificity of pooled fecal culture is assumed to be equivalent to the specificity of culture in confirming the taxonomic characteristics of M. avium subsp. paratuberculosis, that is, 100%. Although we have no evidence that any of the following occurred, factors that could affect apparent specificity include misidentification of individual sheep or flocks, sample cross contamination in the field, sample mislabeling in the field, sample mislabeling and transcription errors in the laboratory, sample cross contamination in the laboratory at any stage of the testing protocol, false-positive reactions in PCR, and errors in reporting from the laboratory. These potential problems can be prevented by quality control protocols, care, and common sense.

There are several explanations for the presence of M. avium subsp. paratuberculosis in fecal samples from individual animals or groups of animals that lacked other clinicopathologic evidence of infection. First, fecal shedding may occur in the absence of an antibody response. Excretion in feces is known to occur months before seroconversion in sheep (2). Second, fecal shedding may occur in the absence of detectable histologic lesions regardless of serologic status. Culturing of feces or intestinal tissues from individual sheep has been shown to be a more sensitive indicator of the presence of M. avium subsp. paratuberculosis than histopathologic examination for infected flocks (24). One obvious reason for this finding is the focal nature of lesions in some sheep and the relatively small amounts of tissue that can be examined histologically. Third, in infected flocks, there may be passive excretion of ingested M. avium subsp. paratuberculosis. This excretion may continue for up to 1 week after ingestion (8, 21). Another factor that could affect the apparent specificity of culture is the existence of strains of M. avium subsp. paratuberculosis of low virulence, but there is no evidence for such strains, and variations in virulence are not considered in the description of the taxon. The validity of the results obtained during experiment 3 is supported by the matching of positive pooled fecal culture results with districts of New South Wales that are known to contain flocks infected with M. avium subsp. paratuberculosis and the absence of positive results from most other districts.

Results for nine farms (11 pools) in experiment 3 were classified inconclusive by pooled fecal culture. This classification was necessitated by technical limitations in the conduct of REA, namely, insufficient DNA to conduct the test in agarose gels with ethidium bromide staining. REA is necessary to increase the level of confidence that a positive result by PCR is in fact due to the DNA from M. avium subsp. paratuberculosis (5). Alternate methods would enable the conduct of REA even when only small amounts of DNA are available. One of the simplest would be to evaluate REAs in acrylamide gels with silver staining. It is important to note that when REA was performed on the IS900 PCR product in this study, results consistent with M. avium subsp. paratuberculosis were obtained in every instance.

The collection and examination of fecal pellets from large groups of sheep for the purpose of diagnosing paratuberculosis have not been undertaken before, and it was therefore considered important to assess the practicality of the technique. Most respondents to the survey noted that collection of fecal samples was practical unless sheep had been kept in yards too long and had already defecated. Transport of fecal samples from farm to laboratory is an important issue. Efforts are required to ensure that the interval between sampling and receipt at the laboratory is as short as possible, preferably overnight, and samples should not be left in transit over a weekend in order to minimize problems with growth of irrelevant microorganisms (R. J. Whittington, unpublished data). The practicality of laboratory aspects of pooled fecal culture is similar to that of individual sample culture, but there is an additional requirement for facilities and equipment to carry out homogenization of the pooled pellets, a rate-limiting step. Facilities are therefore required to store fecal samples pending homogenization, and at present it is suggested that storage be done at −80°C to minimize the loss of viability of M. avium subsp. paratuberculosis (17, 18).

In considering the requirements for sampling from a flock, the necessity for randomization should not be overlooked unless whole-flock examination is undertaken. Biased samples did not improve the rate of detection in this study (Table 2), in contrast to earlier reports on the benefits for serologic diagnosis of using samples biased toward sheep with poor body condition (12, 13). In the present study, sheep that were positive by the pooled fecal culture test generally had no clinical signs of paratuberculosis and may have been in uniform body condition within farms; when biased selection was used, it may have been based merely on age. In other words, the infected sheep were likely to have been in relatively early stages of the disease and/or the prevalence of infection was low. Under such conditions, the biasing of collection in favor of sheep with poor body condition would not be effective.

The costs of pooled fecal culture are lower than those of serologic testing. Collection costs (in Australian dollars) are approximately $0.50 per sheep for feces compared to $1.00 per sheep for blood, excluding the costs of containers, needles, and syringes. Serologic testing is undertaken for approximately $6 per head, compared to $2 per head for pooled fecal culture at a pooling rate of 50. Thus, the costs of flock diagnosis using pooled fecal culture are approximately 30% of those using serologic testing, assuming that equal numbers of sheep are tested. The apparent higher sensitivity of pooled fecal culture also means that fewer sheep need to be sampled to provide equivalent confidence of detecting infected flocks, further reducing costs. Considering that there are over 30,000 flocks of sheep in New South Wales, Australia, and that 30% of these might require testing, the economic benefits of pooled fecal culture are considerable.