Presumed cholesterinic granulomas detected on CT in horses are associated with increased lateral ventricle height and age

Abstract Cholesterinic granulomas are mass‐like lesions that form at the choroid plexus of the ventricular system. Large cholesterinic granulomas within the lateral ventricles have been reported to cause severe neurological signs. However, little data are available about their prevalence or appearance in the overall population. The objective was to report the prevalence of presumed cholesterinic granulomas on CT in a population of horses, and investigate associations between presumed cholesterinic granuloma presence, lateral ventricle size, age, and neurological signs. The study was cross sectional, CT scans of the head were assessed for presumed cholesterinic granuloma presence and size, and lateral ventricle height. Computed tomography findings and clinical information were compared using nonparametric testing. Computed tomography scans of 139 horses were included. Presumed cholesterinic granulomas were found in 22 horses (15.8%), nine were unilateral and 13 bilateral. A significant increase in prevalence was observed with age (P < .0001), with 38% of horses over 15 years old affected. The median volume of presumed cholesterinic granulomas was 242 mm3 with a range from 51 to 2420 mm3. The mean lateral ventricle height was significantly increased in horses with presumed cholesterinic granulomas present (P = .004), with a median of 7.3 mm compared to 4.9 mm without. Neurological signs were not associated with presumed cholesterinic granuloma presence or lateral ventricle height. Fourth ventricle mineralizations were found in seven horses, which may represent cholesterinic granulomas. In conclusion, presumed cholesterinic granulomas occurred in a large proportion of the examined population and are associated with increased lateral ventricle dilation and advanced age.

is unknown; inflammation, compression of surrounding tissues, and obstructive hydrocephalus have all been hypothesized. 3,[8][9][10] As a result, the characteristics that influence the relevance of cholesterinic granulomas are unknown. The ventricular system of the brain involves complex flow of cerebrospinal fluid from choroid plexus through the lateral ventricles, the third and fourth ventricle to the central canal of the spinal cord and subarachnoid space. 11 Ventricular enlargement can occur secondarily to inflammation and obstruction both of which have been suggested in cases of cholesterinic granulomas. 3,9 The CT and MRI characteristics of cholesterinic granulomas in horses are described in a number of case reports and case series, 1,3,8,12 predominantly in older horses. Cholesterinic granulomas are described as usually bilateral masses of the choroid plexus in the base of the lateral ventricles, which on CT are heterogeneous, irregularly mineralized, and unreliably contrast enhance. 1,2,9 Cholesterinic granulomas of the choroid plexus of the fourth ventricles are reportedly common, 4,6 although no confirmed cases on CT were found. 13 Computed tomography is a valid choice for investigating intracranial structures of the horse, which allows assessment of the ventricular system and the presence of cholesterinic granulomas, and unlike MRI, is available in the standing horse. 1,3,[14][15][16][17] Based on our review of the literature, cholesterinic granulomas are only described in clinical cases, 1,2,9 and descriptions of smaller or incidental cholesterinic granulomas and relevance in a larger population are lacking in the current literature.
The objective of the study is to report the appearance and prevalence of presumed cholesterinic granulomas in a population of horses and their relevance by examining potential associations with age and gender, lateral ventricle height and symmetry, and presence of neurological signs. As such we present the following alternative hypotheses 1 : Prevalence of presumed cholesterinic granulomas increases with age, 2 mean lateral ventricle height increases with presumed cholesterinic granuloma presence, and 3 presence of presumed cholesterinic granulomas increases prevalence of neurological signs.

Data recording and analysis
The  The lateral ventricle height was measured in a multiplanar reconstruction of a parasagittal plane. The height was defined as the greatest perpendicular distance between the ventral (hippocampus/caudate nuclei) and dorsal (corpus callosum/corona radiata) border of each lateral ventricle caudal to the interventricular foramen ( Figure 2). The lateral ventricle height was recorded for each side individually, as a F I G U R E 2 Parasagittal CT image at a modified soft tissue window width 150 and window length 40. Measurement was made as demonstrated by the line at the maximum distance from the base to floor (hippocampus or caudate nuclei) to top (corpus callosum or corona radiata) of the lateral ventricles mean of the two sides (mean lateral ventricle height), and as a ratio of left:right lateral ventricle height. The mean lateral ventricle height was used to represent overall size of the lateral ventricles, mean lateral ventricle height to brain height ratio was used to give a measure relative to horse size, and the ratio of left to right ventricle height to allow statistical analysis of symmetry. Brain height was measured on midline from the rostrodorsal aspect of the basisphenoid bone (immediately caudal to the sphenopalatine sinus) to the dorsal aspect of the cranium. Lateral ventricle to brain height ratio was calculated by dividing the former by the latter. In addition to the first reviewer, a second blinded European College of Veterinary Diagnostic Imaging resident (D.S.W.) repeated measurements on a sample of 40 randomly chosen horses to enable reliability testing of the lateral ventricle height and presumed cholesterinic granuloma measurements. The presence of presumed cholesterinic granulomas (unilateral or bilateral cases were combined unless stated otherwise) was tested for association with gender using a two-sided Fisher's exact test.

Statistics
Age of groups with and without presumed cholesterinic granulomas present was compared with an Independent-Samples-Mann-Whitney U test, and a Chi-squared test was used to test distribution within different age categories. The relationship between the volume of the presumed cholesterinic granuloma present (or largest volume if bilateral) and the age was assessed with a related-samples Spearman's rank correlation test; for this and all following uses, correlation coefficients were ranked from "very weak" (<0.19) to "very strong" (0.8-1.0) and an independent samples Kruskal-Wallis test was applied to age groups.
A related-samples Wilcoxon signed rank test significance was used to test for difference between the left and right lateral ventricle height in all horses without presumed cholesterinic granulomas present.
Mean lateral ventricle height and a ratio of mean lateral ventricle height to brain height ratio were tested for association with sex using an Independent-Samples-Mann-Whitney-U test and age with a related-samples Spearman's rank correlation test. Independent-Samples-Mann-Whitney-U tests were used to compare the mean lateral ventricle height and a ratio of mean lateral ventricle height to brain height ratio between cases with a bilateral presumed cholesterinic granulomas present, unilateral presumed cholesterinic granuloma present, or no presumed cholesterinic granuloma present. Asymmetrical dilation of the lateral ventricle with presumed cholesterinic granuloma presence was tested by comparing a ratio of the left:right lateral ventricle height with the presence of left and then right unilateral presumed cholesterinic granulomas, compared to presumed cholesterinic granuloma absent. Spearman's rank correlation tests were then used to examine correlation between mean lateral ventricle height and a ratio of mean lateral ventricle height to brain height ratio, to the volume of presumed cholesterinic granuloma (both total volume and volume of the largest presumed cholesterinic granuloma present were tested).
A two-sided Fisher's exact test was used to test for association between the presence of presumed cholesterinic granulomas and neurological signs. The mean lateral ventricle height, a ratio of mean lateral ventricle height to brain height ratio, and left:right lateral ventricle height ratio was also tested for associated with to neurological signs using Independent-Samples-Mann-Whitney U tests.
Fourth ventricular mineralizations were tested for association with age and mean lateral ventricle height (using Independent Samples-Mann-Whitney-U tests) and a ratio of mean lateral ventricle height to brain height ratio, presumed cholesterinic granuloma presence, and neurological signs (using two-sided Fishers exact tests).

Imaging technical parameters
The CT scans were acquired in standing horses with a 64-slice sliding gantry CT scanner (SOMATOM Definition AS. Siemens AG, München, Germany). The protocol utilized was 140 KVp, 328 mAs, pitch 0.8, rotation time 0.5 seconds, and 0.6 mm slice thickness. Bone (H60f) and soft tissue (H31f) reconstruction algorithms with a slice thickness of 1 and 2 mm, respectively, were applied and a standard matrix of 512 × 512 pixels used. Contrast was not utilized.

Interobserver reliability
Interobserver reliability was analyzed for the measured presumed  Table 1; volume is used for further comparisons.

Correlation of volume and age
No significant associations between presumed cholesterinic granuloma volume and age groups could be found for horses with presumed cholesterinic granulomas present. A correlation coefficient of −0.065 and P = 0.8 when correlating the largest (if bilateral) presumed cholesterinic granuloma present. Largest presumed cholesterinic granuloma volume was further assessed within ordinal age groups with no significant differences found (P = 0.9) (Figure 4).

Lateral ventricle height in horses without cholesterinic granulomas
The median "mean lateral ventricle height" was 4.9 mm and "mean lateral ventricle height to brain height ratio" was 0.053 in horses without presumed cholesterinic granulomas ( Table 2). In horses without presumed cholesterinic granulomas present, the median left lateral ventricle height (5.0 mm) was significantly greater than the right (4.6 mm; P = .02). The median "mean lateral ventricle height to brain height ratio" in horses without presumed cholesterinic granulomas present

Association of lateral ventricle height with presumed cholesterinic granulomas
The median "mean lateral ventricle height" was 7.3 mm and median "mean lateral ventricle height to brain height ratio" was 0.077 in horses with presumed cholesterinic granulomas, both of which were significantly greater than those with no presumed cholesterinic granulomas (P = .004 and P = .001, respectively). The significant difference was maintained after exclusion of all horses 0-4 years old (P = .01 and P = .006, respectively). No significant difference between horses with unilateral or bilateral presumed cholesterinic granulomas was found (P = .5 and P = .4 for the absolute and ratio values, respectively;    (Table 2).

Neurological signs
Neurological signs were seen in 25 horses of the total of 139, of which nine displayed headshaking, six vestibular/cerebellar ataxia, five cranial nerve deficits, four abnormal mentation, and two epilepsy.
Neurological signs showed no significant association with sex or age (P = .6 and P = .3, respectively). Neurological signs were seen in two of

Fourth ventricle and other intracranial mineralizations
Fourth ventricle mineralizations were found bilaterally in seven horses (5%, 95% confidence interval 2.3-9.6), no unilateral cases were observed. All incidences were seen as mineralizations, without a defined mass lesion presumably due to border effacement with the surrounding brain parenchyma (in contrast to presumed cholesterinic granulomas in the lateral ventricles that were predominantly surrounded by cerebrospinal fluid). The median age of horses with fourth ventricular mineralizations present (17 years, range 9-33) was significantly greater than those without (10 years, range 1-24; P = .02).

Association of fourth ventricle mineralizations with presumed cholesterinic granulomas
No association between fourth ventricle mineralization presence and neurological signs was found (P = 1.0). Fourth ventricle mineralizations were significantly more likely to occur in horses with presumed cholesterinic granulomas present (P = .001); five of the seven horses with fourth ventricular mineralizations also had presumed cholesterinic granulomas present (two unilateral and three bilateral). The median "mean lateral ventricle height" was 7.15 mm in horses with fourth ventricle mineralizations compared to 5.13 mm without, the difference did not reach a level of significance (P = .1).
Similarly, the difference between the median lateral ventricle height to brain height ratios (0.055 in horses without and 0.78 for horses with fourth ventricular mineralizations) did not reach significance (P = .7).

Other mineralizations
Additionally, several intracranial mineralizations were found that did not fit the criteria for presumed cholesterinic granulomas or fourth ventricular mineralisations. These included: bilateral mineralizations within the ventral piriform lobes of three horses, a small mineralization in the white matter of the frontal lobe in a single horse, and a central 1 mm diameter mineralization of the brain stem in one horse.

DISCUSSION
This study has further investigated the presence and possible rele- The prevalence of cholesterinic granulomas has previously been a subject of disagreement, with prevalence reported as 15-20% in older horses 4 or 15% of a population of epileptic horses. 6 Our study found the overall prevalence of presumed cholesterinic granulomas in the sample population of 15.8% and a much higher prevalence in older horses (15 years or older) of 40%. This difference could be due to altered rates of postmortem detection rather than a representative live population 4,5 or differences in region and demographic. Breed was not examined in the current study as an uneven distribution, prevented accurate comparisons.
A significant difference in prevalence of presumed cholesterinic granulomas between younger and older horses was confirmed leading to acceptance of our first hypothesis. No significant association between age and the size of presumed cholesterinic granuloma was found, which would seem to concur with a previous paper that examined a case of bilateral cholesterinic granulomas over time and found an overall reduction in size. 2 Although further longitudinal CT data collection is required to make meaningful conclusions about change in cholesterinic granuloma size over time.
Cholesterinic granulomas have been suggested to cause neurological signs due to primary pressure changes, and consequent to lateral ventricle dilation and hydrocephalus. 9 For this reason, lateral ventricle height was also investigated. This is the first study to describe lateral ventricle height in horses. Lateral ventricle height is a simple and repeatable radiological measure of lateral ventricle size, it avoids the confounding presence of presumed cholesterinic granuloma within the measurement region and has been demonstrated as effective as area and volume in detecting hydrocephalus in dogs. 20 The method of measurement of lateral ventricle height and brain height used were subtly altered from the method used in dogs, as MRI enabled accurate assessment of both values in a single plane, however, relatively limited soft tissue detail necessitated choosing the greatest height of the lateral ventricle for measurement rather than a simple midline height. Additionally, bony rather than soft tissue landmarks were chosen for brain height to increase reliability of measurement on CT. 20,21 The repeatability of the measure was confirmed by excellent interobserver reliability testing. The lateral ventricle median height was 4.9 mm, and significantly greater on the left than right side. This finding coincides with previous studies in dogs finding that the left lateral ventricle is larger than the right in the majority of cases. 22 Overall results between absolute lateral ventricle height and a ratio of lateral ventricle height to brain height varied little. However, the ratio did reveal a significant increase in relative lateral ventricle height with increased age. This could be due to reduction in brain volume with age a phenomenon previously reported in dogs. 21 Mean lateral ventricle height was significantly increased in cases with either a bilateral or unilateral presumed cholesterinic granulomas.
This would appear to support the suggestion that lateral ventricle dilation occurs secondarily to cholesterinic granuloma presence. 3  However, at present, CT and pathological correlation are limited in the literature unlike those found in the lateral ventricles. A mineralized fourth ventricle cholesterinic granuloma has been described but not confirmed in a horse. 13 One case was confirmed in a dog, however, that case presented with a large central mass like lesion. 28 As such other differentials such as other choroid plexuses tumours (papilloma etc) or siderocalcinosis are possible. 29 Siderocalcinosis is considered as a differential, despite its occurrence in the white matter of the cerebellum not choroid plexus. This is due to the limited soft tissue contrast on CT images, due to the minimal amount of observed fluid in the lateral recesses of the fourth ventricles and the location adjacent to the petrous temporal bone. 14,29 The piriform lobe mineralization is of unknown relevance.
Although this study increases understanding of prevalence and