Abstract
Scrapie is a fatal, neurological disease of sheep and goats and belongs to the transmissible spongiform encephalopathies. In 1998, a new type of scrapie, designated scrapie Nor98, was detected in Norway. Scrapie Nor98 differs from classical scrapie in the distribution of pathological changes and of the scrapie prion protein, the Western blot profile of the prion protein, and with isolated cases usually being observed in the case flocks. In 2004, a case–control study was conducted on scrapie Nor98 with 28 cases and 102 randomly selected controls. The questionnaire included questions on demographic data, animal contact between sheep flocks, indirect contact with equipment, use of concentrate feed and supplemental feeds, and use of medicines and vaccines. The data were analysed by using logistic regression with the sheep flock as the statistical unit. In the final model, the detection of scrapie Nor98 was related to the practice of not removing all afterbirths, the use of vitamin and mineral feed supplements, the absence of concentrate feed of swine or poultry on the farm and the presence of dogs on the farm. The results show that the epidemiology of scrapie Nor98 differs from that of classical scrapie in that no risk factors that indicate transmission of scrapie Nor98 between flocks by movement or direct contact between animals were found. Furthermore, the association between scrapie Nor98 and mineral intake shown herein should be explored further. Although the possibility that scrapie Nor98 has a low transmissibility between animals under natural conditions cannot be ruled out, the results would also be in accordance with a spontaneous aetiology.
-
↵†Present address: Norwegian Meat Research Centre, PO Box 396 Økern, NO-0513 Oslo, Norway.
INTRODUCTION
Scrapie is a fatal, chronic neurological disease of sheep and goats. Scrapie belongs to the transmissible spongiform encephalopathies (TSEs), which also include bovine spongiform encephalopathy (BSE) in cattle and Creutzfeldt–Jakob disease in humans. Scrapie has been a notifiable disease in Norway since 1965, and the first verified scrapie case in indigenous Norwegian sheep was detected in 1981 (Hopp et al., 2000). In 1998, a new TSE strain, designated scrapie Nor98, was found in Norwegian sheep (Benestad et al., 2003). During the period from 1998 to 2004, scrapie Nor98 was the most commonly detected scrapie strain diagnosed in 46 of the 53 case flocks (Valheim et al., 2005). Similar strains, designated atypical scrapie to distinguish them from classical scrapie (European Food Safety Authority, 2005), have been detected in several other European countries (Buschmann et al., 2004; Gavier-Widén et al., 2004; Onnasch et al., 2004; Orge et al., 2004), indicating that atypical scrapie is more common than believed previously.
TSEs are associated with the conversion of the host-encoded prion protein PrPC to a partly protease-resistant form designated PrPSc (Prusiner, 1991), which accumulates in the central nervous system. The molecular mechanism for the conformational change from PrPC to PrPSc is not known. Nevertheless, by using in vitro trials, metal ions such as copper and manganese have been shown to influence the conformational change from PrPC to PrPSc (Cerpa et al., 2005; Kim et al., 2005). Furthermore, there are at least 20 different PrP alleles (Heaton et al., 2003), including five alleles, designated VRQ, ARQ, ARH, AHQ and ARR, that are shown to be especially important for susceptibility to classical scrapie (reviewed by Hunter, 1997). Scrapie Nor98 differs from classical scrapie in the distribution of pathological changes and of PrPSc in the central nervous system, in the PrP genotypes affected most frequently and in the age of the case animals at the time of detection (Benestad et al., 2003; Moum et al., 2005).
Although there is not yet full agreement on the aetiological agent of scrapie, the predominant theory is that PrPSc is the infectious agent (Enserink, 2005). Classical scrapie is transmitted between animals under natural conditions, although the mechanism is not fully known. However, PrPSc has been found in the placenta and amniotic fluids, and oral uptake of placenta and amniotic fluids or materials contaminated by these is thought to play a major role in transmission between animals (Andréoletti et al., 2002). For introduction into flocks, a study on risk factors for scrapie in Norwegian sheep flocks found that the purchase of female sheep from scrapie flocks, sharing pastures with scrapie flocks, and sharing breeding rams were the main risk factors for scrapie (Hopp et al., 2001). This agrees with the view that the main routes for transmission of scrapie between flocks are movement of animals or animal-to-animal contact (reviewed by Detwiler & Baylis, 2003). In France, purchase of proprietary concentrates was found to be a risk factor for scrapie, and the spread of the agent by contaminated meat-and-bone meal, analogous to the situation with BSE, was suspected (Philippe et al., 2005). On two occasions, the spread of scrapie has been related to contaminated vaccines (Agrimi et al., 1999; Gordon, 1946).
However, it is not known whether the source of the infectious agent of scrapie Nor98 is the same as that for classical scrapie. The observation that only a single animal with scrapie Nor98 has been detected in 20 of 21 sheep flocks, in which all animals older than 18 months were examined during stamping out (B. Bratberg, personal communication), has led to the speculation that Nor98 scrapie might be a spontaneous prion disease (Benestad et al., 2003) analogous to sporadic Creutzfeldt–Jakob disease in humans.
The aim of this study was to analyse the relationship and the strength of association between potential risk factors and the detection of scrapie Nor98 in Norwegian sheep flocks.
METHODS
Design.
The study was designed as a case–control study, using the sheep flock as the statistical unit.
Case flocks.
All Norwegian sheep flocks in which at least one sheep had been detected with scrapie Nor98 from the start of 2002 until 1 June 2004 were considered for inclusion in the study as cases (30 flocks in total). These flocks had been detected through surveillance of slaughtered animals (n=11), surveillance of found-dead animals (n=13) and examination of animals showing clinical signs that indicated the possible presence of scrapie (n=6). Scrapie Nor98 diagnosis was based upon detection of PrPSc by either ELISA (BSE Platelia or TeSeE; Bio-Rad) or immunohistochemistry. All of these were confirmed by Western blot (TeSeE Sheep/Goat; Bio-Rad), with a profile characterized by a fast-migrating band at about 12 kDa (Benestad et al., 2003). Two farmers refused to participate in the study, leaving 28 case flocks for the study (Fig. 1⇓). The detection year of the cases varied from 2002 to 2004, with nine, 14 and five cases detected in 2002, 2003 and 2004, respectively.
Geographical distribution of the case and control flocks in a case–control study of scrapie Nor98 in 136 sheep flocks in Norway. The country is divided into the four different regions used in the analysis. Sheep flocks within the striped area were excluded from the study.
Control flocks.
Approximately three control flocks per case flock were selected from the ‘Register of Production Subsidies' of 31 July 2003. Sheep flocks with ten or more breeding sheep, located throughout Norway except for Finnmark County and the northern part of Troms County, and with no known history of scrapie were accepted as controls. A total of 102 control flocks (Fig. 1⇑) were selected randomly from the 16 207 sheep flocks that were eligible. When farmers refused to participate in the study, they were replaced by randomly selected reserves within the same municipality as the original control. A total of 47 control flocks were replaced due to (i) the presence of fewer than 10 breeding animals in the flock (n=6), (ii) unavailability of the farmer by telephone (n=7), (iii) inconvenience of the scheduled time for interview (n=10), (iv) declination by the farmer to participate without specifying the reason (n=22) and (v) other reasons (n=2).
Data collection and variable description.
The data were collected during summer 2004 by personal interviews on each farm using a standardized questionnaire prepared for the study (available from the corresponding author). Four different interviewers were engaged. The questionnaire covered information at both flock and animal levels. At the animal level, information was collected on the case animal or the oldest animal at 1 January 2004 in the case or control flocks, respectively. When considered relevant, information was collected for the last 10 years preceding the detection of a case in the case flocks or from 1994 to 1 January 2004 in the control flocks. In order to improve readability, separate questionnaires were used in the case and control flocks, the difference being the wording describing the relevant time period.
Information on the following topics was collected: flock characteristics (no. breeding animals, breeds represented in the flock, age of eldest ewe and ram, preferred age of culling, no. ewes 6 years and older, other production animal species kept on the farm); contact between sheep flocks (purchase of ewes and rams, import of breeding material, sharing of rams, pastures and equipment with other sheep flocks, contact with scrapie-infected farms); management (housing of rams and ewes, inspections during lambing, handling of the afterbirth, use of manure for fertilizing); feed (use of concentrate feed, milk replacements, vitamin and mineral feed supplement and lick stones) with brand name and suppliers; feed for non-ruminants (brand name, storage and use); use of vaccines. Information on diseases and treatment of the sheep (clinical signs, reason for culling, use of medicines and surgical operations) was collected at the individual level.
All variables collected on an ordinal or continuous scale were dichotomized to derive two evenly sized groups. When information was collected as yes/no answers, on an ordinal scale and as exact numbers, the appropriate level of accuracy was chosen after dichotomizing the variables, considering the numbers of missing values. The location of the sheep flocks was categorized into four geographical regions: south-eastern, western, middle and northern Norway (Fig. 1⇑).
Information on feeding stuff with brand name and suppliers was collected as open-ended questions. When information on brand name was missing, the suppliers were contacted to acquire data on the products that they distributed. For each feedstuff category, the copper content was obtained from information on whether the feed was produced for sheep (without copper in certain regions) or for other ruminants (with copper).
Statistical analysis.
The data were entered by using Microsoft Office Sharepoint Portal Server 2003. All further variable processing and statistical analyses were performed in the SAS-PC system, version 9.1.3 for Windows (SAS Institute Inc.). The descriptive analyses were conducted by using frequency tables (proc freq).
Unconditional logistic regression, with the Nor98 status as the outcome variable and the sheep flock as the statistical unit, was performed by using proc logistic. The independent categorical variables were expressed as dummy variables. Flock-level factors were considered as candidates for the multivariate analysis, whereas the animal-level factors were used for description of the data. All of the variables were initially run in a univariate logistic regression analysis, and those with Wald χ2 P values of <0.20 were selected for further analyses. Every possible pair of non-categorical variables was checked for correlation before dichotomization by using Spearman's test for ranked variables (proc corr). If a pair of variables had a correlation coefficient of >0.60 and both variables had a P value of <0.20, the variable judged as most biologically plausible was included in the multivariate analysis.
The multivariate logistic regression was performed by using a combined manual forward and backward selection process, in which the Wald χ2 P value was used to test the significance (P value <0.10) of adding or subtracting one variable at a time from the model. The same approach was used to test the significance of the two-way interaction terms between the independent variables in the final multivariate model. The odds ratios were calculated from the estimated coefficients in the final model, as a measure of strength of association.
RESULTS
Description of the material
The case and the control flocks were scattered geographically (Fig. 1⇑). In the case flocks, the mean number of breeding ewes was 78 (range, 17–187), compared with a mean of 64 (range, 8–330) breeding ewes in the control flocks. Norwegian white sheep (including Dala sheep, Rygja sheep and Steigar sheep) and Spæl sheep were the breeds reported most frequently, although breeding material from 12 of the 15 Norwegian breeds was present within the flocks participating in the study. The case animals, all ewes, were born between 1994 and 2000 and their ages varied from 3 to 9 years, with a mean age of 6 years.
A total of three case and ten control flocks reported contact with other scrapie flocks during the last 10 years. Of these, two case and three control flocks had contact with flocks with scrapie Nor98. During the last 10 years, ewes were introduced through purchase into 13 case flocks and 37 control flocks. No live animals or embryos were imported into any of the flocks. Participation in ram circles during the last 10 years was reported by 11 case flocks and 35 control flocks. All flocks except two (one case and one control flock) reported the use of rams from other flocks during the last 10 years. A total of three case flocks and 23 control flocks had not shared pastures in the last 10 years.
Vaccines had been used in the last 10 years in 22 case and 71 control flocks. Of the four different vaccine brands, three were combined vaccines against clostridium toxins and one was against listeriosis. Medical treatment had been received by four of the case animals for four different conditions before they showed clinical signs that could be related to scrapie. None of the case animals had undergone surgery.
All flocks had used concentrate feed for sheep, which was obtained from more than 30 suppliers, representing six different companies. Concentrate feed for swine or poultry had been stored at the holding of 48 flocks, all of them keeping either swine or poultry. The use of concentrate swine feed or poultry feed for sheep was reported from one case flock and six control flocks. In all instances, the feed was used for less than 1 week during the last 10 years, and in five instances the feed was given by accident. Vitamin and mineral feed supplements had been purchased from three producers, and lick stones for use on pasture from three producers. Dogs and cats had been fed pet food in the sheep barn on 11 farms, three of which were cases. None of the farmers reported the use of meat-and-bone meal or imported concentrate feed.
Univariate analysis
There were 17 potential risk factors with a P value of <0.20 in the univariate analysis (Table 1⇓). Of the four variables related to purchase of sheep, one had a P value of <0.20 in the univariate analysis. None of the four variables that measured use of concentrate feed for sheep had a P value of <0.20 in the univariate analysis. There was no significant association between scrapie Nor98 and feedstuffs with or without added copper for concentrate feed, lick stones, or vitamin and mineral supplemental feed.
Potential risk factors with a P value of <0.20 in a univariate logistic-regression analysis in a case–control study of scrapie Nor98 in 130 Norwegian sheep flocks
Multivariate analysis
Table 2⇓ shows the risk factors that were found to be related significantly to the scrapie Nor98 status in the final multivariate assessment. The practice of not removing all afterbirths and the use of vitamin and mineral feed supplements increased the odds for the flock being detected as positive for scrapie Nor98. If feed for swine or poultry was kept on the farm, risk for scrapie Nor98 in the flock was significantly lower than if such feed was not kept on the farm, whilst the presence of dog(s) at the farm increased the risk of Nor98 significantly. The final model included none of the factors that indicated movement of animals between flocks or direct contact with animals from other flocks as risk factors for scrapie Nor98. No two-level interaction terms were found to be related significantly to the scrapie Nor98 flock status.
Results from the multiple logistic-regression model in a case–control study of scrapie Nor98 in sheep flocks in Norway [n=129; information from one flock was not used due to missing information on ‘Dog(s) present on the farm’
Abbreviations: LRT, likelihood-ratio test; CI, confidence interval; OR, odds ratio.
DISCUSSION
In this study, we did not find any significant risk factors indicating that scrapie Nor98 had been transmitted between sheep flocks by animal movement or animal-to-animal contact. This suggests that the transmissibility of scrapie Nor98 in natural conditions is low, if present at all. This was in contrast to the previous study on risk factors for scrapie-positive flocks based on material collected in 1995–1997 in Norway (Hopp et al., 2001), where factors such as purchase of female sheep from scrapie flocks, sharing pastures with scrapie flocks, and sharing breeding rams increased the odds for scrapie in the flock. Based upon the pathological changes and the PrP genotypes affected, it is now assumed that most of the case flocks included in the previous study had classical scrapie (B. Bratberg, personal communication). Therefore, the conflicting results between these two studies are probably due to two different types of scrapie being examined, with different abilities to transmit between animals.
In classical scrapie, the prion protein PrPSc has been found in the placenta and amniotic fluids and the afterbirth is considered as an infectious source for other animals and for contamination of the environment (Andréoletti et al., 2002). In a previous case–control study on scrapie in Irish sheep flocks, it was found that failure to retrieve the placenta from the lambing pen or disposal of placenta in the compost increased the odds of (classical) scrapie (Healy et al., 2004). Our finding that the practice of not removing all afterbirths increased the risk of scrapie Nor98 might suggest that the agent of scrapie Nor98 is also found in the placenta of infected dams. In classical scrapie, the detection of PrPSc in the placenta has been reported in dams with the PrP genotypes ARQ/ARQ, ARQ/VRQ and VRQ/VRQ. For these genotypes, PrPSc has also been found in the lymphoid tissue (Andréoletti et al., 2002; Tuo et al., 2002). To our knowledge, the scrapie Nor98 agent has so far only been detected in the central nervous system and not in lymphoid tissues (Benestad et al., 2003), which might indicate that PrPSc is not found in the placenta in scrapie Nor98. The examination of the placenta of dams infected with scrapie for PrPSc might give insight into the possibility of the placenta being an infectious source of scrapie Nor98.
The increased risk for scrapie Nor98 in flocks in which vitamin and mineral feed supplements had been used might be explained by the feed supplements that either included factors predisposing for scrapie Nor98 or had been contaminated with the scrapie Nor98 agent. Both minerals and vitamins might be potential predisposing factors from the feed and, on the molecular level, it is shown that PrPC has metal-binding properties. It has been suggested that manganese may induce the conformational change from PrPC to PrPSc (Kim et al., 2005), and copper may induce, as well as reduce, this conformational change (reviewed by Cerpa et al., 2005). Furthermore, an increased level of manganese in brain material from human cases with Creutzfeldt–Jakob disease and hamsters with scrapie has been reported (Kim et al., 2005; Wong et al., 2001). Therefore, the association found between scrapie Nor98 and feed supplements might be explained biologically by the intake of essential elements that influence the development of the disease.
A few epidemiological studies have investigated the associations between mineral intake and TSEs. In a study on risk factors for BSE, no association between BSE and mineral intake was found (Wilesmith et al., 1988). Chihota et al. (2004) investigated potential associations between mineral content in pasture and scrapie in the UK and found that farms with an excess of molybdenum had higher odds of having scrapie. However, the authors suggested that this might be a false-positive result (type I error). Although previous epidemiological studies have not been able to confirm any association between mineral intake and TSEs, we cannot rule out the possibility that this might be biologically plausible. We suggest that this relationship should be explored further in studies with a design suitable for the purpose.
The feeding regime of sheep in Norway varies considerably and the sources for essential elements are concentrate feed, lick stones, forage and grazing, in addition to the vitamin and mineral feed supplement. We would expect sheep given vitamin and mineral feed supplements to have a higher mineral intake than sheep not fed supplemental feed. Both the vitamin and mineral feed supplements and the salt lick stones produced for sheep generally included the essential elements: calcium, magnesium, manganese, selenium, iodine, cobalt, zinc and iron. When the farmer had used feed supplements produced for other ruminants, copper was also included. It would have been preferable to perform the analysis by using information on the quantitative intake of the different essential elements, but in our study design, the data only allowed differentiation between use of feedstuffs with and without copper. The fact that there was no significant association between scrapie Nor98 and feedstuffs with or without added copper indicates that the increased risk for scrapie Nor98 was not related specifically to the copper content of the feed.
The result that the use of feed supplements gave a larger risk for scrapie Nor98 raises the question as to whether these might have been contaminated with the scrapie Nor98 agent through raw materials of animal origin. In Europe, BSE (Wilesmith et al., 1988) and perhaps also (classical) scrapie (Philippe et al., 2005) have spread through concentrate with meat-and-bone meal contaminated with the respective agent. In Norway, meat-and-bone meal was prohibited in commercial feed for ruminants in November 1990, and products of ruminant origin were prohibited as feed for ruminants in June 1994 (Høgåsen & Hopp, 2002). Although meat-and-bone meal has been excluded from ruminant concentrate feed since 1994, meat-and-bone meal was allowed in concentrate feeds for poultry and swine until 2001 (Høgåsen & Hopp, 2002) and some cross-contamination cannot be ruled out. However, the fact that the practice of keeping concentrate feeds for poultry or swine on the farm reduced the risk for scrapie Nor98 suggests that such feed has not been a source of the scrapie Nor98 agent.
The finding that the presence of dog(s) on the farm represented a higher risk for scrapie indicates that feed offered to dogs might have been contaminated with the scrapie Nor98 agent, analogous to feline spongiform encephalopathy in cat, which is suggested to originate from cat feed contaminated with the BSE agent (Bradley, 1997). However, when asked about the feeding practices for dogs, only three case and eight control farmers admitted having fed their dog(s) in the sheep barns. To our knowledge, there is no report describing any relationship between the risk of scrapie and contact between sheep and dogs. Therefore, the association found is not explained easily and we think that the result must be considered in the context of a study with a limited number of cases and controls and with the possibility of a false-positive result.
The positive association between scrapie Nor98 and PrP genotypes with the alleles AHQ and AF141RQ has been documented previously (Moum et al., 2005). We were not able to find any relationship between the breed represented in the flock and the risk of scrapie Nor98 that could support the findings of the previous Norwegian study. However, in our study, the flock level was the unit of concern, which might not have been an optimal design for measuring this relationship. In future, such a relationship might be studied on the animal level and by using the PrP genotype as a measure of genetic susceptibility.
The latent period for scrapie is long and, for some factors, we considered it important to collect historical information and recall bias cannot be excluded. The questionnaire included questions for a period of up to 10 years, in accordance with the time period for which the farmers were obliged to keep the economic records, to allow the use of information collected from written sources and thereby reduce recall bias. Farmers with scrapie-affected flocks have been obliged to collect information on all trade for the authorities, and knowing that (classical) scrapie is transmitted through animal-to-animal contact, they probably would have tried to recollect any contact with other flocks. The control farmers would not necessarily have done so, which might have generated a differentiated recall bias. This would potentially exaggerate the effect from these factors. Despite this, no factors indicating transmission between animals were found to be significant in the final model.
The case flocks were asked for the last year of activity at the farm and then 10 years back, whereas the control flocks were asked for the time period 1994–2004. The different time periods of cases and controls might have introduced a problem if the exposure of the risk factors has changed during the time period. However, the cases differ from the controls by a maximum of 3 years, which, with regard to Norwegian sheep farming, we would consider a short time period for the risk factors investigated. Nevertheless, when considering the different time periods of the cases, a design where the cases and controls were matched for time period would have been preferable.
This study was designed as a case–control study in which the control flocks were selected randomly. However, for practical reasons, some adjustments were made to a true random sample of the total Norwegian sheep population. In order to reduce the cost of information collection, potential control flocks in the northern-most part of Norway, a total of 623 flocks (4 %), were excluded from the study. Northern Norway has a harsh climate and different management routines would be expected. Therefore, extrapolation of the results to the northern-most part of Norway is not necessarily valid. Furthermore, if a control refused to participate in the study, it was replaced by a reserve from the same municipality as the control that it replaced. This selection procedure was chosen to avoid (expensive) changes in the travel itinerary during the collection of the data. We do not know of any factor that might be affected by this as a result.
Despite the low number of cases used in this study, we were able to reveal several factors that were associated significantly with scrapie Nor98. The study has shown to be a cost-effective way of screening a large number of potential risk factors representing several hypotheses on the origin of the disease. The results of the study support the hypothesis that the epidemiology of scrapie Nor98 differs from that of classical scrapie. Scrapie Nor98 has a low transmissibility or might not be transmitted between animals under natural conditions. Furthermore, an association between mineral intake and scrapie Nor98 is biologically plausible and should be explored further in studies designed for the purpose. The association with mineral intake would be in accordance with a spontaneous aetiology of scrapie Nor98.
Acknowledgments
We would like to thank the farmers for their active participation in the study and veterinary students Margrethe Handeland, Silje R. Normann, Tonje Seim and Andreas Svendsen for conducting the interviews. This study was financed by a grant from the Research Levy on Agricultural Products, project number 1617487/110.