BSE is characterised by a combination of changes in behaviour, sensation and movement, although the contributory signs that define these categories can be variable [3, 30]. It has been shown that BSE should be considered in cattle that display a combination of apprehension, over-reactivity (to touch or sound) and ataxia, since at least one of these signs featured in 97% of 17,154 cases . Because of the experimental designs of the two source studies in the present investigation these criteria are not applicable to the entire population of animals studied. For example, using the approach for at least two of the signs apprehension or nervousness, over-reactivity (as evaluated by tests of over-reactivity including repeated spontaneous startle responses) and ataxia or hypermetria, not all experimentally inoculated, BSE-confirmed cattle were clinical suspects under these criteria. This was to be expected in study 1 where animals were culled at pre-defined time points post challenge and therefore some cattle were in the early stages of disease, or not infected, and thus would not necessarily display the full spectrum or any of the signs associated with BSE. In contrast, the cattle in study 2 were culled at clinical end-point when the signs were convincing enough to suspect BSE, unless, in some cases, concerns for staff and animal health and welfare due to the animal's behaviour precluded monitoring further progression of signs, such as ataxia. The selection of criteria to define a clinical suspect as part of this investigation may explain why some BSE-positive cases were not identified as clinical suspects: for example, one BSE case inoculated with distal ileum from orally dosed cattle culled at 10 mpi developed only over-reactivity (marked aversion to touch in combination with frequent startle reaction), which was convincing enough to suspect BSE, although neither nervousness nor ataxia were displayed.
Conversely, 17 cattle of study 2 that were inoculated with tissue from BSE-infected cattle were not confirmed to have BSE based on postmortem tests and no alternative diagnosis was established but they did display the sign combination for a clinical BSE suspect. While it is possible that other conditions affecting the CNS or other organ systems might cause behavioural, sensory or locomotor abnormalities, which in combination are suggestive of BSE, no such circumstances occurred in controls. Similar observations have been made in a separate study in casualty slaughtered cattle .
The design, particularly group sizes, of the source studies presented difficulty in analysing the clinical signs with regards to the choice of statistical tests. Two different methods were chosen to compare the frequency of expressed clinical signs, either taking into account repeated observations or using only the last assessment prior to cull. BSE is a slowly progressive disease and signs are displayed more frequently, more consistently and more severely as the disease progresses, which makes Cox regression survival analysis an appropriate method. However, in study 1 cattle were culled at a predetermined time point post challenge or, in both studies, at a predetermined end point when the development of BSE was not expected in surviving animals. Therefore, some of the animals may have been at an early stage of the disease when signs associated with BSE were observed only shortly before culling and not yet repeatedly. In these cases, the Fisher's exact test was more applicable, which only considered the occurrence of a sign prior to cull. Regardless of the statistical method, the results confirmed, as expected, that the selected signs associated with BSE occurred more frequently in groups with diagnostic changes of vacuolation and/or PrPd accumulation.
The histopathological diagnosis based on vacuolar changes is less sensitive than detection of PrPd by IHC, which has been demonstrated in experimental BSE [13, 18, 32], since, in general, histopathological changes are apparent later in the course of the disease than detection of PrPd. The more frequent occurrence of locomotor and sensory signs in cattle with vacuolar changes compared to those with no or equivocal vacuolation (Table 2) suggests that vacuolation coincides with a more advanced clinical stage. This is in agreement with the findings in a Swiss study, where clinical suspect BSE cases had more severe spongiform changes in the brain than BSE cases slaughtered as apparently healthy cattle .
The great variation in the group sizes meant that for some signs statistical analysis was compromised because the number of animals per group was too small. It was considered appropriate to combine BSE-positive animals dosed with 1 g and 100 g to increase the sample size for comparison of signs prior to cull although this was less appropriate for the analysis of signs over time, which may be dependent on the dose. It has been demonstrated in cattle orally challenged with BSE that the incubation period followed a lognormal distribution, with decreasing mean as dose increased , and thus a similar effect of dose on onset of the first clinical signs and possibly the disease duration is expected. In study 2 in particular, the number of animals in some groups was considerably greater than in others, ranging from 10 to 240. As a result of the disproportionate group sizes, the differences in the signs displayed by BSE-negative cattle inoculated with tissues from cattle orally dosed with BSE-affected brain and controls were usually not significant, or signs were even less frequently observed over a time period than in controls. In these instances, comparison of signs on a case-to-case basis may be more appropriate, like the selection of clinical markers to identify clinical suspects as demonstrated in Table 1.
Although the signs used for comparison of the different groups of cattle in different source studies consisted of behavioural, sensory and locomotor signs associated with BSE [3, 24, 25], many of these signs are unspecific, or not characteristic of a neurological disease and may also be displayed to some degree in healthy cattle. This may explain why some of the signs were more frequent in BSE-positive cattle of one study but not in the other. For example, "BSE series of events", which were defined as behavioural signs in combination with head tossing or nose licking, all of which were associated with BSE , were usually significantly more frequent in BSE-positive cattle of study 2 but were unexpectedly significantly less frequent in BSE-positive cattle dosed with 1 g compared to controls of study 1. Other signs that may be neurogenic, such as tremor, may occur physiologically in anxious animals or may be displayed as hind limb tremor in animals with spastic syndrome, which occurred in some animals but was not considered to be associated with BSE (data not shown). To simplify the analysis, the absence or presence of tremor was assessed regardless of location and underlying condition. In most cases, in particular when the tremor was not confined to the head, it was impossible to determine the cause based on a clinical examination and to differentiate a potential physiological tremor or a tremor caused by other intercurrent diseases from an identical tremor displayed by BSE-affected cattle. Furthermore, the finding that significant differences were not consistently present between BSE-positive cattle and controls, for example in the display of nervousness on head tests or head restraint, made it difficult to interpret the significance of these signs. This may be partly explained by the different treatment of animals between studies, such as regular blood sampling, which was done only in cattle of study 1. We also cannot rule out that the route of inoculation (oral or intracerebral) has an effect on the display of particular signs since the route of infection may result in minor differences in the severity of vacuolar changes [32, 34, 35] although cattle with BSE have generally presented with changes in behaviour, sensation and locomotion irrespective of the route [32, 36].
A predominant sign of BSE is over-reactivity to external stimuli, which has been observed in 75-98% of confirmed clinical field cases . Its presence was so characteristic in BSE that tests of over-reactivity have been refined successfully to diagnose BSE , to predict cases among BSE suspects prior to postmortem test confirmation  and to screen casualty slaughter cattle for BSE . Startle responses are also observed frequently in patients suffering from sporadic Creutzfeldt-Jakob disease [37, 38]. Indeed, startle responses to tests of over-reactivity and apparently spontaneous startle, occurring during examinations or observed on weekly observations, were consistently more frequent in BSE-positive animals compared to controls and this was often statistically significant. Interestingly, the repeated display of startle responses was also significantly more frequent in cattle orally dosed with 1 g and 100 g of BSE brainstem in study 1 despite a negative postmortem test result, regardless of the statistical method used. Similarly, cattle inoculated with tissue from BSE-challenged cattle (study 2, group 3) startled to the observer's movement significantly more often than control cattle when signs over time were compared. Nervousness or apprehension in the corridor and tremor were also significantly more frequently displayed in BSE-negative cattle orally dosed with 100 g using the Cox regression analysis. In the absence of histopathological changes in the brains of these cattle and the exclusion of metabolic diseases by regular blood biochemical analysis of selected cattle in the study (data not shown), it is possible that the signs were actually caused by BSE. There are some precedents for this in other species, such as mice  and pigs  inoculated with the BSE agent and goats naturally infected with scrapie , which presented with signs suggestive of a TSE despite failure to detect PrPd in the brain. In a previous study, of the animals in the present study 1, which sought to predict the relationship between detection of PrPd and incubation period according to dose, the clinical criteria were based on the combination of clinical signs consistent with the "definite" clinical diagnosis of BSE which was considered equivalent to reporting of the animal as a suspect in the field . This approach includes the use of classifications of possible and probable signs of BSE  where "possible" and "probable" indicate a potential earlier clinical stage based on the relative frequency and severity of signs or sign combinations. Using this approach, it has been shown that some animals in both of the present studies may display possible or probable signs in the absence of a positive postmortem test result [14, 21].
Improvement of postmortem test protocols over time has accounted for increased sensitivity of the diagnosis of BSE cases in an experimental oral exposure cattle study [13, 42]. Also, studies utilising brain tissue from selected orally dosed, postmortem test unconfirmed cattle in study 1, have demonstrated the presence of infectivity by assay in transgenic mice overexpressing bovine PrP  and the presence of PrPres by PMCA . There are therefore clear precedents for the prudent application of additional transmissibility, molecular and neuropathological assessments of clinically healthy animals inoculated with a TSE agent to exclude the possibility of subclinical prion infection, which has been often demonstrated in rodent studies . The possible occurrence of clinical disease in cattle unconfirmed by current postmortem tests equally warrants further investigation.