1. Introduction
Chronic progressive lymphedema (CPL) is a common disease in draught horses [
1,
2,
3,
4,
5]. The first reports of CPL in draught horses date from the early 1900s and indicate a considerable number of German draught horses were affected [
6,
7]. Horse breeds mentioned in this first report also include Danish draught, Belgian draught, Shire, and Clydesdale [
6]. In a review on the causes of premature leave among 979 coldblooded stallions from the stud in Kreuz, Germany, CPL was found to be the primary cause in 6.6% of the cases [
8]. Stallions belonged to different breeds, including Shires (n = 28), Saxon-Thuringians (n = 20), Belgian Draught horses (n = 10), Clydesdales (n = 3), Rhenish Germans (n = 2), and Percherons (n = 1). A survey in the French departments Cluny and Annecy showed a prevalence of CPL in 46–47% of the Ardenn, 10–12% of the Percheron, and 2–3% of the Breton breed [
9]. Wallraf et al. investigated the prevalence of CPL in all German draught horse breeds and found the highest prevalence in Rhenish German, Saxon-Thuringian, Schleswig, and Mecklenburg Draught horse breeds, with 96.0, 84.1, 81.4 and 74.5%, respectively [
1]. Black Forest and South German showed a lower prevalence, with estimates at 24.0 and 39.0% [
1,
10]. Studies from Belgium and USA reported CPL in the Belgian draught horse, Shire, Clydesdale, gypsy vanners, as well as the Friesian [
2,
3,
5].
The characteristic clinical signs of CPL include progressive swelling, scaling, hyperkeratosis, crusts, skin folds, nodules, often with exudative wounds and ulcerations, fibrosis of the distal limbs, and, in some cases, verrucous lesions with clearly demarcated calluses and wart-like proliferations [
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12]. Secondary bacterial and recurrent parasitic infections may aggravate lymphedema and clinical signs [
3,
11,
12,
13]. In contrast to these observations, infestation with
chorioptes mites was not correlated with the prevalence of the different stages of CPL-lesions in German draught horses [
1]. CPL-lesions start to appear at a younger age and prevalence increases with age, regularly leading to limb disfigurement and disability [
11]. Pathological examinations of the distal limbs reveal epidermal hyperplasia, dermal oedema, dilated and tortuous lymphatic vessels, development of thick-walled lymphatics, especially in the palmar/plantar region of the fetlock joints, fibrosis of dermis and subcutaneous tissues, and fragmentation and disorganization of elastic fibres [
11,
14]. The pathological changes in the lymphatic system correspond with the severity of the clinical lesions [
11]. Lymphoscintigraphy indicated a delay in lymphatic clearance in the limbs of affected horses [
14]. In addition, lymph flow stagnates completely in severely affected limbs [
14]. These findings strengthen the hypothesis that the clinical signs of CPL may be due to stagnant lymph flow [
14]. Defects in the network of elastic fibres surrounding dermal lymphatics are assumed to be associated with chronic lymphedema [
15]. Elastin content of the skin can be quantified by measuring cutaneous desmosine levels. CPL-affected Clydesdale and Shire horses show higher cutaneous desmosine levels than clinically normal horses of the same breeds, but controls of the breed Percheron have significantly higher cutaneous desmosine levels than CPL-unaffected Clydesdale and Shire horses [
15]. Mildly CPL-affected horses have the highest cutaneous desmosine levels; however, levels decrease in more severely CPL-affected horses. Accumulation of elastin and amounts of desmosine are highest in superficial dermis in the distal limb and neck [
16,
17]. However, a study in Belgian draught horse stallions could not confirm the use of cutaneous desmosine levels as a diagnostic aid for CPL [
18]. In summary, progressive lymphedema and tissue fibrosis, along with disorganized elastic fibres supporting lymphatic vessels, impair lymphatic clearance and correlate with disease progression; however, the primary causative factors for the development of CPL still remain unknown [
13,
14,
15].
CPL is an incurable disease that requires intensive management to slow down the disease progression [
13,
19]. Treatment of secondary infections is difficult due to severe feathering, treatment failures, and recurrences [
19,
20]. Clipping of feathering, manual lymph drainage, bandaging with short stretch bandages, and exercise are recommended [
19,
21]. Removing multiple multifocal verrucous masses using dissection and electrocautery from the distal hindlimb of a draught horse did not result in regrowth after a 24-month period [
22].
The occurrence and severity of CPL-lesions showed interactions between sex and age in previous studies [
1,
10,
12,
23,
24,
25]. A study with 37 cases, including mainly German draught horses, three Percherons, one Belgian and one Polish draught horse, and two controls showed a significant correlation between the severity of CPL-lesions and age of the horses [
12]. In 912 draught horses, including all different German draught horse breeds, sex and linear regression on age were significant for the prevalence of CPL and number of limbs affected with CPL [
1]. Both studies with 431 and 980 records of Belgian draught horses confirmed the significant effects of age and sex by age on CPL severity [
2,
24]. Stallions tended to develop CPL-lesions quicker than mares [
24]. Along with more severe CPL-lesions, cannon bone circumference [
12] and skinfold thickness increased [
1,
10,
12,
24,
25]. Contrasting results were obtained for all German draught horse breeds, where residual correlations among skinfold thickness and CPL prevalence were close to zero in multivariate animal models [
1,
10,
25].
Horse farm-related factors affect the severity of CPL-lesions. Horses kept in outside pens on rubber meadows were less severely affected than those kept on sand or soil [
12]. Stable hygiene and stable quality were correlated with the prevalence [
10,
25] and severity of CPL-lesions [
12,
25]. A restrictive feeding management with refrain of concentrates and hay silage reduced the risk of developing CPL-lesions [
10,
26,
27].
The large differences in prevalence of CPL among cold blooded breeds and the disposition of draught horses to CPL suggests the involvement of genetics [
3,
7,
8]. Heritability has been estimated for CPL-scores in Belgium draught horses [
23] and in South German, Black Forest, Schleswig, Rhenish German and East German (Mecklenburg, Saxon-Thuringian) draught horses for the prevalence of CPL, the different stages of CPL, and the number of affected limbs [
1,
10,
25]. Analyses in Belgian draught horses restricted to horses older than 3 years revealed higher heritability estimates with 0.26, compared to data with horses of all ages with 0.11 [
23]. Animal model analyses across German draught horse breeds resulted in heritability estimates of 0.21 for the prevalence of CPL-lesions and 0.24 for the number of limbs affected [
10]. Heritability estimates for the South German and Black Forest draught horses were lower than for Schleswig, Rhenish, and East German breeds [
25]. Genetic correlations with skinfold thickness (0.28–0.43) and pronounced fetlock tufts of hairs (0.32–0.34) were moderate in German draught horses [
10], but inconclusive in Belgian draught horses for both measures, skinfold thickness, and hair diameter due to their large standard errors [
23]. A genome-wide scan with 318 microsatellites for 378 German draught horses using multipoint linkage analyses revealed four chromosome-wide significant quantitative trait loci on ECA 1, 9, 16 and 17 [
28]. A genome-wide association study using 307,474 single nucleotide polymorphisms (SNPs) for Friesian horses with 26 cases (CPL-affected) and 19 controls did not identify significantly associated loci for CPL [
4].
The clinical relevance of CPL and its associated impact on animal welfare, health, and reduced life expectancy are factors that contribute to the importance of determining the prevalence of CPL and possible risk factors [
5,
11]. Previous reports indicated that CPL may have a significant impact on the health of Rhenish German draught horses [
7,
25] and therefore breeders may be discouraged from keeping this breed. In this way, CPL could be a reason why the population size of the Rhenish German draught horse population continues to decrease below 1000 breeding animals. Therefore, our objective of this study was to examine the current prevalence and severity of CPL as well as the age at onset in Rhenish German draught horses in Germany. In addition, we evaluated risk factors associated with CPL. We employed a CPL scoring system based on Wallraf [
25], Affolter [
3] and de Keyser et al. [
2] to be applied for each limb in order to monitor CPL-scores based on a standardised system.
At each horse farm visit, we collected data on animal variables and housing, feeding, management, exercise of horses, and hoof care. These data should allow evaluation of associations between CPL and animal as well as horse farm-related variables. Due to the large number of studs across Germany, stud and animal variables were not confounded with each other. Thus, we employed multivariable models to evaluate the effects of risk factors on CPL.
2. Materials and Methods
2.1. Ethical Approval
The study was performed according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of the University of Veterinary Medicine Hannover (Foundation) and the state veterinary offices from the different German Federal States for North Rhine-Westphalia (registration number 81-02.05.40.19.083), Lower Saxony (registration number 33.8-42502-05-19A465), Thuringia (registration number 22-2684-04-TIH-20-101), Brandenburg (registration number 2347-A-19-1-2020) and Saxony (registration number 25-5131/521/20). The sampling and handling of the horses followed European Union guidelines for animal care and handling and the Guidelines of Good Veterinary Practices.
2.2. Sample Collection
The study was performed in cooperation with the Westphalian Breeding Association, the Rhenish Breeding Association, and the North Rhine-Westphalian State Stud. We presented the project to the breeders in several meetings and published articles on the aims and study design in horse breeders’ journals. We invited all Rhenish German Draught horse breeders in North Rhine-Westphalia to participate in the study. In addition, we offered breeders from other German states to be involved in this project. All breeders who agreed to participate in this project received an information letter and were contacted to arrange a visiting appointment for examination of the horses and recording management, housing and feeding data of the horses. All breeders recruited for this study participated voluntarily and gave written informed consent for participating in this research on CPL and possible risk factors. The horse farms were located in Brandenburg, Lower Saxony, Rhineland, Westphalia, and Thuringia.
In total, we examined 493 horses from 96 horse farms (
Table 1). Data were collected from February 2019 until October 2021. The mean number of sampled horses per farm was 5.1 ± 8.9 with a minimum of one horse and a maximum of 61 horses per farm. The work of the breeding organizations is limited to a certain region and this means that all horse-farms from a certain region are members of the respective breeding organization.
The average age of the examined horses was 7.77 ± 6.37 years, with a maximum age of 33.3 years and a minimum of 1–2 months.
Figure 1 represents the distribution of the 493 sampled horses by age. The numbers of horses older than 1, 2, 5, 8, 11, and 19 years were 425, 354, 256, 186, 124, and 42, respectively. Most of the horses > 17 years old were females (70.2%) and only a few geldings (15.8%) and males (14.0%). The horses from the oldest group were sampled in Westphalia (n = 23), Thuringia (n = 8), Brandenburg (n = 3), and Rhineland (n = 8). Overall, we sampled 111 males (22.5%), 57 geldings (11.6%), and 325 (65.9%) females. The distribution by sex and age in years is given in
Supplementary Figure S1. Average ages of males, geldings and females were 5.2 ± 5.4 years, 10.2 ± 6.1 years, and 8.2 ± 6.5 years, respectively. The oldest male, gelding, and female reached 21.7, 27.8, and 33.3 years, respectively.
2.3. Data Collection
During each horse farm visit, an interview with the horse owner was conducted. The stud was inspected to record data on the type of stable, the type of use or work applications for the horses, the bedding type, manure management, type of outdoor access and average hours of outdoor exercise per day in autumn and winter or in spring and summer, average hours of pasture grazing, frequency of shoeing, and type of roughage and concentrates fed in the summer and winter months. The answers were documented on a written questionnaire (
Supplementary Table S1). The questionnaire contained yes-no and closed questions.
2.4. Examination of the Horses
Sampling of horses was carried out by a veterinarian. The handheld RFID reader APR600 (Agrident, Barsinghausen, Germany), which can scan transponder types HDX and FDX-B, was used to scan the transponders and record the data of the sampled horse, with a pre-programmed task mode stored on the reader. The scanned transponder number was matched with the equine passport. The Universal Equine Life Number (UELN), name, date of birth, and coat colour of the horse were electronically recorded at the farm. In addition, the skinfold thickness [
24] on the neck (Cutimeter, Hauptner, Solingen, Germany) and the hoof conformation were recorded by measuring the length of the dorsal wall, heel length, angle of the dorsal border, and hardness of the hoof horn at the dorsal wall (Shore D, Zwick-Roell, Ulm, Germany) from the left front and right hind limb. Since the differences between left and right hoof are meaningless [
25], we measured one front hoof and one rear hoof. To take into account the sides of the body, we chose one hoof from the left and one hoof from the right side. The pigmentation of the hoof colour was documented for the left front hoof and the right hind hoof.
Classification of CPL signs is shown in
Supplementary Table S2. The examination of all four limbs included inspection and palpation, starting at the hoofs and ending at the knee or elbow joints. The findings were documented for each limb. Each horse received a final score for each limb and an overall score for all four limbs.
2.5. Statistical Analysis
Statistical analysis of data was performed using SAS, version 9.4 (Statistical Analysis System, Cary, NC, USA, 2022). Descriptive statistics were calculated with the SAS procedures MEANS and distributions and exact binomial 95% confidence intervals (CI) with the SAS procedure FREQ. We used generalised linear mixed models to evaluate associations of CPL-scores with breeding association, sex, month-year-classes at time of examination, coat colour, age, and limb. There was no influence of the investigator to be considered, as all examinations were performed by an experienced veterinarian. Analyses were performed using the SAS procedures GLIMMIX and MIXED. Dependent variates used were the overall CPL-score across all four limbs (CPL-score), the highest CPL-score per horse (CPL-max) on a scale from 0 to 5, the sum of all CPL-scores over all four limbs (CPL-sum), as well as CPL-score > 0, CPL-score > 1, and CPL-score > 2, and CPL-score > 3 as 0/1-variates. Overall CPL-scores were calculated using the CPL-scores per limb. An overall CPL-score of 0 was determined when the sum of CPL-scores per limb was not greater than 1. Overall, a CPL-score of 1 was set when the sum of CPL-scores per limb was >1 and <4 and the maximum CPL-score per limb was 2. Overall, a CPL-score of 2 was deduced when the sum of CPL-scores per limb was >3, at least one limb had a CPL-score of 2, and CPL-scores per limb did not exceed a value of 2. Overall, a CPL-score of 3 was applied when the sum of CPL-scores per limb was >4, at least one limb showed a CPL-score of 3, and no limb had a CPL-score >3. Overall, a CPL-score of 4 was given when the sum of CPL-scores per limb was >7, at least on one limb a CPL-score of 4 was seen, and CPL-scores per limb were <5, and, overall, the CPL-score was 5 when the sum of CPL-scores per limb was >8 and at least one limb showed a CPL-score of 5.
In the generalised mixed linear models, we employed a binomial distribution function and logit as link function for binary variates, and for categorical variates a multinomial distribution function with a cumulative logit link function or a normal distribution and identity link function to parameterize the probability of an overall or specific CPL-score for individual horses.
The final model 1 included the fixed effects of breeding association, coat colour, sex, and the linear and quadratic regressions on age at examination within sex. The final generalised mixed linear model 1 using cumulative logits for probabilities of CPL-scores (
θijklmn) or as quasi-quantititive trait (
yijklmn) with an identity link function was as follows:
where
μ is an unknown constant common to all horses in the linear model, and in the multinomial model, unknown constants for the thresholds to which the observed CPL-score belongs, breeding association
i = fixed effect, with i = Brandenburg-Anhalt, Rhenish, Saxon-Thuringian and Westphalian including Lower Saxon; coat colour
j = fixed effect, with j = chestnut, black and bay; sex
k = fixed effect, with k = female, gelding and male; b
1(age * sex)
l = linear regression on age in years within the three different sexes; b
2(age
2 * sex)
m = quadratic regression on age in years within the three different sexes; e
ijklmn = unknown random residual effect.
CPL-scores and prevalence were not significant between limbs, thus CPL-scores for the single limbs were not used in all following analyses. In addition, we did not find a significant effect for month-year-classes for time at examination. All two-way interactions other than age by sex tested with a stepwise forward and backward selection strategy were not statistically significant and, thus, were omitted in the final model. The stud was tested as a random effect, but did not reach statistical significance. Coat colours distinguished were bay, chestnut, and black. Roan coat colour was not significant as separate fixed effect nor in a two-way interaction with basic coat colours. This final model 1 was employed for all horses sampled and subsets by age groups. Herein, we restricted the analyses for a minimum age of 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 years in order to test for the importance of the age by sex effects on the different CPL-scores. Analyses of variance were also performed for CPL-max, CPL-sum, CPL-score > 0, CPL-score > 1, and CPL-score > 2, and CPL-score > 3.
We extended this final model 1 to study horse farm-related factors on CPL-scores. These stud factors were regarded as single additional fixed effects or covariates in model 1. Stud factors included type of stable, outdoor facilities for horses in winter and summer months, bedding type, time interval for cleaning out the stable, type of roughage and concentrate fed in winter months, type of concentrate fed in summer months, other additional feed types for horses, type of hoof care, length of hoof trimming intervals, type of work applications for horses, daily hours of work with horses, and days per week of work with horses. The stud-related factors as additional single factors in model 1 with p-values < 0.05 were then tested to reach the final multivariable generalized mixed linear model. Forward selection using the SAS procedure HPQUANTSELECT was performed to test the consistency of the results. In the final model 2, a p-value < 0.05 was defined as the significance threshold for inclusion of an effect.
Animal variables were height at withers, body length, chest circumference, thickness of skinfold at the neck region, and, at the left front and right hind limb, cannon bone circumference, hoof measures including dorsal wall length, heel length, and front angle, and Shore D hardness of hoof horn at the dorsal wall. We tested animal variables as single additional effects using model 1, as well as all animal variables of the body and variables either of the front or of the hind limb with the forward selection procedure HPQUANTSELECT, in order to validate significant associations with CPL-scores.
4. Discussion
In this study, we analysed the prevalence and progression of CPL with age and evaluated animal- and horse farm-related factors influencing the prevalence and severity of CPL-lesions in Rhenish German draught horses. We employed multivariable generalised linear models to identify risk factors for CPL-scores. Due to the large sample size and large number of horse studs, factors under analysis were not confounded. Participation in this study was voluntary, and breeders were required to agree for taking part in this study. Thus, our study was not completely randomized. However, confidence intervals are reasonably small for the prevalence of CPL, and therefore support the robustness of our results.
The classification of CPL took into account the various progressing forms and severity of CPL [
1,
2,
3,
12,
13,
25]. All recordings were done on horse farms in order to sample all horses per stud and in particular the older horses that were no longer used for breeding. For this reason, we did not visit horse registration or horse breeding events or horse contests for recording, as was the case in a previous Belgian study [
2,
23]. The proportion of horses older than 3 years was lower in the previous Belgian study, 52.6% [
23], than in the present study, 63.5%. Mares at stable visits were significantly older than mares on horse contests [
2]. In addition, horses sampled on horse contests had significantly lower CPL prevalence (54% vs. 66%) and lower CPL-scores in mares (2.49 vs. 5.34) than horses examined at stable visits [
2].
Comparison of the prevalence of CPL between different studies is difficult due to study design, age distribution of horses sampled, and type of sampling. The age when horses are sampled has a major effect on prevalence, as CPL progresses significantly with age [
1,
2,
10,
12,
23,
25]. Often, it is not possible to sample most horses for a certain breed because recording is not mandatory for a breeding program and breeders can decide whether to participate in a study. In the Belgian study [
23], 56% in the full data set with 762 horses and 88% in the dataset with 401 horses older than 3 years were affected by CPL. The mean age in the full data set of the Belgian study was 3.87 ± 3.02 years [
2]. In the previous German study [
1,
10,
25], horses older than 2.5 years were recorded, and the prevalence estimated in a model correcting for systematic effects such as age and sex ranged widely from 24% to 96%. Estimates were between 75 and 96% for breeds with a genealogical relationship with Belgian draught horses [
29], and lowest for Black Forest at 24%. Overall, the prevalence of CPL in Belgian draught horses [
2,
23] appear to be similar to those in the Rhenish-German draught horses in the present study and slightly lower than in the previous German study [
1,
25]. The horses had a higher mean age (8.50 ± 4.38 and 7.77 ± 6.37 years) in both German studies, the previous [
1,
25] and the present study, which may contribute to the similar prevalence compared with the Belgian study [
2,
23].
All studies agree that males or stallions are more frequently affected by CPL [
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20,
21,
22,
23,
24,
25]. The higher susceptibility of males developing more severe CPL-lesions is believed to be a result of the stabling and management of breeding stallions and the breeding aim of large-framed stallions with strong cannon bones in previous times [
10,
30].
An increase in CPL prevalence with age was found in Belgian [
2,
23] and German draught horse breeds [
1,
10,
25]. In the present study, similar to the Belgian study [
2,
23], we confirmed a significant interaction of sex and age, whereas in the previous studies [
1,
10,
25], an interaction of sex and age was not evident and only a linear increase with age was found. In addition, the increase in CPL prevalence was significantly steeper in Rhenish German, Mecklenburg and Saxon-Thuringian draught horses than in Black Forest and South German [
1]. Disease progression is significantly faster in males, which is in agreement with the Belgian study [
2,
23], but not with previous German studies [
1,
10,
12,
25]. As the clinical signs of CPL progress in older horses, lesions become more painful and interfere with movement, the intervals between recurrent infections become shorter, and the horses lose appetite and their condition deteriorates, which may result in the need to euthanize the horses. Therefore, severely CPL-affected horses may not reach the same age as only mildly affected horses, so the proportion of severely CPL-affected horses may decrease at older ages. An explanation for these differences in disease progression between males and females is thought to be differences in the type of housing and less access to pastures [
25,
26,
27] as well as more intensive selective breeding in stallions [
7,
25,
30].
Differences between breeding associations (regions) of the Rhenish German draught horse were found in the previous [
1,
25] and the present German study. A possible reason for these differences could be genetic differences between these subpopulations [
29]. There seems to be some exchange of breeding horses between these regions, but far from a complete mixing of breeding populations. Even when the model used accounted for interactions between age and sex and coat colour and was extended to include horse farm-related effects, the significant differences between breeding associations remained. Differences in age structure among these breeding associations were not important because we did not detect interactions between age and breeding associations or between breeding association, sex, and age. We also examined the different types of use and work assignments of the horses, but could not demonstrate a significant contribution to higher CPL-scores. Draught horses in eastern Germany were more frequently used in agriculture and dairy production in the past, which might have led to a stricter selection against CPL.
The effects of coat colour do not seem to be consistent among the different previous studies [
2,
23,
25] and may be partly associated with sires or horse families. However, the association of increased CPL prevalence with chestnut and bay Rhenish German horses in Saxony-Anhalt was also stated in an early German study [
31]. A direct effect of horse colour variants on the occurrence of CPL has not yet been considered. Significant quantitative trait loci for CPL were not located on the horse chromosomes where coat colour genes (
ASIP on ECA 22,
MC1R on ECA 3,
SLC45A2 on ECA 21 and
KIT on ECA 3) have been mapped [
28]. Even larger samples seem to be necessary to reach a clearer conclusion.
In agreement with previous studies, we confirmed the negative effects of some feeding rations [
1,
12,
25,
26,
27]. Feeding of concentrates and hay silage are factors promoting CPL [
1,
25]. Access to pasture and improvement in stable hygiene help to reduce the prevalence and progression of CPL [
1,
12,
25]. Optimization of housing conditions, feeding rations and management may have positive effects on preventing disease occurrence and progression, but these environmental effects could not cancel out the effects of the breeding area, sex, and age by sex interactions. Thus, these horse farm-related effects explained only a part of the total variation. Exercise is believed to enhance lymphatic flow and therefore supports the transportation of lymph fluids [
19,
32]. Horses used for riding, carriage, and work in agriculture were less prone to severe CPL-lesions when the type of work application was added as a single factor to model 1. Therefore, it is important to improve opportunities for exercise in draught horses. Even though this effect was not significant in the final model 2, and only a trend in this direction was observed.
We studied animal variables in order to prove whether body traits might show correlated changes with CPL and thus be influenced by the impaired lymphatic flow [
11,
14,
15,
16]. The significant positive correlation between CPL-scores and cannon bone circumference is consistent with previous studies [
7,
12,
30]. The significant correlations with cannon bone circumference at the front and hind limb were still present even if we removed age and sex effects for cannon bone circumference in the analysis. This result means that larger cannon bone circumferences are correlated with CPL-scores, independent of the age and sex of the horses. This might indicate a genetic disposition to more severe CPL-lesions in horses with larger cannon bone circumference. Height at withers was not significant in our study, therefore we cannot confirm that large-framed horses are significantly more likely affected by CPL-lesions [
26,
30]. Similarly, hoof measures and hoof horn hardness do not seem to be influenced by CPL and its underlying stagnating lymphatic flow.