1. Introduction
The domestic dog exhibits a profound degree of phenotypic diversity in size. Two particular conditions affecting size, referred to as chondrodystrophy and chondrodysplasia, are characterized by shortened limbs and are common across many dog breeds [
1,
2]. The causes for both chondrodystrophy and chondrodysplasia were identified as two separate fibroblast growth factor 4 (
FGF4) retrogenes on chromosome 12 (12-
FGF4RG) and chromosome 18 (18-
FGF4RG), respectively [
1,
2]. The most severe form of disproportionate dwarfism is seen in breeds that carry both
FGF4 retrogenes, such as Dachshunds, Basset Hounds, and Corgis [
1,
3]. FGF signaling is involved in early embryonal development [
4,
5], and appropriate levels of
FGF4 are necessary for normal limb formation [
6]. Higher levels of
FGF4 transcripts were seen in dogs with either of the 12-
FGF4RG or 18-
FGF4RG insertions, leading to the conclusions that the
FGF4 retrogenes are expressed and that the short-limb phenotype is associated with overexpression during development [
1,
2]. Similarly in humans, achondroplasia, the most common form of dwarfism, is caused by the gain of function variants in the fibroblast growth factor receptor 3 that results in increased signaling [
7,
8].
In addition to shortened limbs, chondrodystrophic breeds are also characterized by chondroid metaplasia of the nucleus pulposus leading to premature degeneration and calcification of the intervertebral discs [
9,
10,
11]. This degeneration predisposes chondrodystrophic dogs to intervertebral disc disease (IVDD), a debilitating disorder associated with protrusion or extrusion of intervertebral disc components into the vertebral canal, resulting in pain and/or neurological dysfunction [
12]. Disc calcification may be visualized radiographically, and chondrodystrophic dogs with increased numbers of radiographically visible calcified discs have been shown to be at higher risk for clinical IVDD [
13,
14,
15]. Calcification of the nucleus pulposus has also been described in older, non-chondrodystrophic dogs in the late stages of degeneration; however, this process generally occurs at an earlier age in the chondrodystrophic dog breeds [
10,
16,
17].
While all breeds can be affected by IVDD, chondrodystrophic breeds are at particularly high risk [
9,
10,
12]. Hansen classified the IVDD that occurs in the chondrodystrophic dog breeds as Type I, typified by acute extrusion of degenerate, often calcified nucleus pulposus through degenerate annulus fibrosis into the vertebral canal [
10,
18]. Hansen Type II IVDD generally occurs at a later age in larger breed (non-chondrodystrophic) dogs and more commonly involves chronic protrusion of degenerative disc material. Historically, type II IVDD has been reported to involve fibrous rather than chondroid disc degeneration. However, despite some specific differences in macro and microscopic pathology and in disease progression, Hansen’s original work and more recent studies have shown that chondroid metaplasia is a common underlying pathological process in both chondrodystrophic and non-chondrodystrophic breeds [
9,
19].
Current consensus supports the use of decompressive surgery to remove the disc material impinging on the spinal cord in dogs severely affected by IVDD [
12,
17], although the cost of surgery can be prohibitive for many owners. Dogs susceptible to IVDD may also suffer multiple disc herniations at different locations throughout their lifetimes [
20]. While chondrodystrophic dog breeds with 12-
FGF4RG alone, such as the French Bulldog and Beagle, are at high risk for IVDD, the chondrodysplastic breeds with 18-
FGF4RG alone, such as the Scottish Terrier and the West Highland White Terrier, are not considered at high risk [
1,
10]. However, in dogs with both
FGF4 retrogenes, the contribution of 18-
FGF4RG to disc degeneration and thus IVDD is unknown. It is also unclear whether or not the
FGF4 retrogenes act in a completely dominant manner or whether any additive effect exists. Since many breeds are homozygous for the
FGF4 retrogenes, the determination of the relative risk for intervertebral disc herniation is challenging. Segregating breeds and mixed breed dogs provide an opportunity to evaluate the risk of herniation in the presence of the retrogenes.
A broad analysis of allele frequency across dog breeds was performed for both FGF4 retrogenes, identifying breeds that segregate or are fixed for one or both retrogenes. A referral hospital DNA database was utilized to obtain information from dogs that had received decompressive surgery for IVDD, and prospective samples were collected for two years from surgical cases to obtain a large, across breed sample of 569 dogs that had IVDD defined by surgery. These samples were genotyped for both 12-FGF4RG and 18-FGF4RG. Breed, weight, sex, age at time of first surgery, and the presence of calcified discs at the time of surgery were used to determine the contribution of 12-FGF4RG and 18-FGF4RG to disease phenotype using linear and logistic regression. A separate logistic regression was performed in mixed breed dogs to determine characteristics contributing to IVDD surgery itself, and a relative risk for 12-FGF4RG was calculated in segregating breeds.
4. Discussion
Both 12-FGF4RG and 18-FGF4RG are common across many dog breeds, as they were found in at least one dog in 53% and 43%, respectively, of the 75 breeds tested. Among the dogs treated surgically for IVDD, dogs that carried at least one copy of 12-FGF4RG were significantly younger, smaller and more likely to have radiographically calcified discs than the dogs without 12-FGF4RG. 12-FGF4RG was the only retrogene with a statistically significant effect on disc calcification in an additive manner. Age and breed also had modest effects on disc calcification. Under multivariable logistic regression, 12-FGF4RG was the only factor contributing to IVDD surgery in mixed breed dogs. The relative risk for 12-FGF4RG varied among segregating breeds, with mixed breed dogs carrying 12-FGF4RG the most at risk for IVDD.
The 12-
FGF4RG has been described in association with the chondrodystrophic phenotype [
1], as previously defined clinically and pathologically by Hansen, Braund, and others [
10,
11]. Consistent with these historical data, presence of the 12-
FGF4RG in this large data set was found to be significantly associated with four major clinico-pathological phenotypes associated with chondrodystrophy-associated IVDD; specifically, small size, anatomical location of IVDD, early age of onset, and presence of calcification. IVDD in 12-
FGF4RG carrying dogs was mostly located in the thoracolumbar region, while only 14.2% of the extruded discs localized to the cervical region. Hansen observed that only 15% of the extruded discs in the chondrodystrophic breeds were cervical, and other studies have also shown that thoracolumbar herniation is most common in the chondrodystrophic breeds [
18,
22,
23]. In contrast, among the dogs with zero copies of 12-
FGF4RG, 42.2% of the extruded discs localized to the cervical region. The caudal cervical region, in particular, was the most affected, although thoracolumbar and lumbosacral IVDD was also present among the dogs with zero copies of 12-
FGF4RG, results which are consistent with previous studies in non-chondrodystrophic breeds [
18,
23,
24].
While age at time of surgery across all breeds was significantly lower for 12-
FGF4RG dogs, significant differences were present within specific chondrodystrophic breeds suggestive of additional factors, either genetic or environmental, that may contribute to overall disease presentation. As has been described previously, French Bulldogs had a mean age at time of surgery significantly lower than other breeds, at 4.1 years [
25]. Alterations in WNT pathway signaling have been consistently implicated in aging and degeneration of the intervertebral disc [
26,
27,
28]. Down regulation of WNT signaling has been described specifically in chondrodystrophic dog degenerate nucleus pulposus [
27], and a downregulating frameshift variant in the WNT pathway gene
Dishevelled 2 (
DVL2) was also recently identified and associated with screw tail and brachycephaly in Bulldogs, French Bulldogs, and Boston terriers [
29]. Bulldogs and Boston Terriers rarely carry the 12-
FGF4RG and are rarely reported with IVDD; however, it is interesting to speculate that the significantly earlier onset of IVDD in French Bulldogs carrying the 12-
FGF4RG may be related to additional perturbation of WNT signaling exacerbating
FGF4 retrogene-related pathology.
Dachshunds, the breed with the highest prevalence of disc disease in this study and elsewhere, surprisingly had a significantly older age of onset than mixed breeds. While fewer individuals were identified from other breeds, there is a trend that breeds with a high allele frequency of 12-FGF4RG have a later median age of onset of disc herniation. It is possible that breeds with high allele frequencies of 12-FGF4RG, such as Beagles and Dachshunds, have undergone additional selection with younger onset affected animals being more likely to be excluded from the breeding pool. It is also possible that owners treat their dogs differently since they are aware of the risk of disc herniation within these breeds. Within breed selection for protective effects could also explain why mixed breed dogs suffer the greatest relative risk for IVDD associated with 12-FGF4RG, as they would not benefit from any protective alleles.
The original classification of type I and type II IVDD made by Hansen was done based on histopathological examinations of intervertebral discs and signalment. For this study, in the absence of histopathology, when descriptions of calcified or mineralized disc material were available from the surgery or radiographic reports, the cases were classified as Group A or Group B. Under this classification system, the majority (87%) of dogs in Group A had at least one copy of 12-
FGF4RG, consistent with a dominant mode of inheritance. Interestingly, 46 out of the 378 cases classified as Group A had no copies of 12-
FGF4RG. Previous studies have shown that Hansen type I IVDD can occur in non-chondrodystrophic breeds, and, more recently, it has also been shown that the histopathological progression of disc degeneration is similar between the chondrodystrophic and non-chondrodystrophic breeds [
17,
19,
22]. Therefore, it is reasonable to expect that some older non-chondrodystrophic dogs could present with clinical IVDD resembling that seen in the chondrodystrophic breeds. In this study, the Group A dogs with zero copies of 12-
FGF4RG were on average 22 months older than the Group A dogs with one or two copies of 12-
FGF4RG. It is unclear whether non-chondrodystrophic dogs presenting with either calcified intervertebral discs or chronic disc protrusions are reflections of a spectrum of presentations within a common phenotype, similar to heterogeneity that may be seen even within chondrodystrophic breeds, or whether additional genetic factors may be present in non-chondrodystrophic dogs, resulting in histopathologically similar end points of disc degeneration with a later onset of disease.
Although the current and previous data support 12-
FGF4RG as the chondrodystrophy locus, many chondrodystrophic breeds expressing the 12-
FGF4RG also carry the 18-
FGF4RG, possibly reflecting prolonged breeding selection for a short limbed phenotype through two different loci (
Supplementary Table S1;
Table 1). The effect of 18-
FGF4RG was modest and only significant in one copy and not two. The number of dogs in this category (heterozygous for 18-
FGF4RG) was low (N = 73) compared to the number of homozygous animals (zero copies 18-
FGF4RG N = 243; two copies 18-
FGF4RG N = 253). These differences in allele frequency are reflective of the allele frequency results within breeds reported here and elsewhere [
2]. The fact that two copies of 18-
FGF4RG does not significantly reduce the age of onset of IVDD argues that the significant effect seen in this dataset for one copy is less likely to be biologically significant. Logistic regression using calcification as an outcome also found that the addition of 18-
FGF4RG did not improve the regression equation compared to 12-
FGF4RG alone. Multivariable logistic regression also showed that 18-
FGF4RG did not significantly contribute to IVDD surgery in mixed breed dogs, where 12-
FGF4RG alone explained the outcome. While this study does not rule out that 18-
FGF4RG is contributing to the IVDD disease phenotype in minor ways, such as a younger age of onset (13 months) in dogs with one copy of 18-
FGF4RG, the effect of 12-
FGF4RG was found to be far greater on all aspects of the disease.
Progression of the nucleus pulposus from normal to a radiographically visible degenerate and mineralized pathology appears to be under the influence of copy number of 12-
FGF4RG since there is an additive effect. This is in contrast to the effect on age of onset of herniation, which is the same between dogs with one or two copies of 12-
FGF4RG. In previous work evaluating height in the Nova Scotia Duck Tolling Retriever, the effect of 12-
FGF4RG was also additive. Beagle neonatal puppies with two copies of 12-
FGF4RG were previously shown to have 20-fold higher expression of
FGF4 in the intervertebral disc compared to Cane Corso neonatal puppies with no
FGF4 retrogenes [
1]. It is possible that continued high expression in the disc in adults is contributing to the rate of mineralization in an additive fashion but that the presence of degeneration is enough to predispose dogs to herniation.
The mechanisms underlying the differential phenotypes associated with the
FGF4 retrogenes remain to be elucidated. The relatively minimal effects of 18-
FGF4RG on disc disease compared to that of 12-
FGF4RG could be due to differences in expression patterns between the two
FGF4 retrogenes. Although retrogenes are often regarded as non-expressing pseudogenes due to the frequent lack of defined regulatory elements [
30], the 5′ end of the
FGF4 retrogenes contains a highly conserved CpG island which is predicted to function as a promotor [
31,
32]. Both
FGF4 retrogenes have been shown to be transcriptionally active, although associated clinical phenotypes appear to be different [
1,
2]. It was previously theorized that expression of 12-
FGF4RG in the intervertebral disc is based on the chromosomal environment in which it was inserted, since all nearby genes were shown to be expressed in the intervertebral disc [
1]. Temporal and tissue-specific expression profiles of the two
FGF4 retrogenes (with associated different clinical phenotypes) may, therefore, be more dependent on the genomic context at the different insertion sites.
Dachshunds are the most commonly affected breed with IVDD, and most of the demographic and breed selection related studies have been conducted relating to the various Dachshund breed varieties [
13,
24,
33,
34]. Dachshunds, similarly, made up the largest portion of this study population, accounting for 31.6% of all retrospectively collected surgical cases, while only making up 4.7% of the DNA repository. While Dachshund variety information was not defined in this study, weights indicated that only two of the 148 surgical cases were Standard Dachshunds, while 146 were miniatures. The allele frequency for 12-
FGF4RG was high within the breed; however, there was variation depending on where the samples were collected (0.98 in USA/UK samples and 0.94 in Swiss samples), indicating that some populations may be less homozygous than others. It is possible that some varieties of Dachshund segregate 12-
FGF4RG more than others, and studies on IVDD in Dachshunds outside the USA suggest that this may be true: A previous genetic analysis of disc calcification in Wirehaired Standard Dachshunds registered to the Danish Kennel Club identified an associated region on chromosome 12 near the 12-
FGF4RG, indicating that the population studied likely segregated 12-
FGF4RG [
35], and wire-haired varieties also appear to be less often clinically affected by IVDD [
18,
21,
33,
36].
Several studies have defined incidence of calcification in Dachshunds, the relationship between calcification and risk of clinical disc disease, and the heritability of disc calcification, providing a body of data to try and inform selective breeding to reduce IVDD incidence in the breed [
15,
21,
37]. Interestingly, recent pilot data (Proschowsky and Fredholm, Gravhunden 1-2018 pp 12-13, Magazine for members of the Danish Dachshund Club) evaluating genotype of 12-
FGF4RG and calcification scores in Dachshunds from Denmark showed an OR of 6.1 for high calcification scores (K6–K15) associated with either one or two copies of 12-
FGF4RG allele within the wire haired variety. It is possible that data from historical calcification and heritability studies, particularly when segregating varieties were included, may have been a reflection of 12-
FGF4RG allele frequency within the heterogeneous Dachshund populations studied [
15,
21,
37].
Determining whether disease incidence as opposed to age of onset increases with two versus one copy of the retrogene is important information for the purpose of IVDD breed eradication strategies. Circumstantial evidence may support an increase in disease incidence with two copies given the association between historical calcification scoring (at a defined age) and disease incidence in one breed of dog [
13,
14], and the correlation of radiographically calcified discs and 12-
FGF4RG allele frequency in this study. However, age at surgery in this study was highly variable, and data relating to radiographic presence of calcification should be interpreted in this context, since it has been shown that the number of radiographically calcified discs declines with age [
21,
38]. Selection against higher numbers of disc calcifications through radiographic screening programs in Dachshunds has been implemented in some countries as a way of reducing the incidence of IVDD [
14,
38], although progress has been limited to date [
15]. This may reflect the inherent sensitivity and specificity issues associated with using disc calcification scoring, and its application over potentially heterogeneous 12-
FGF4RG populations of Dachshund varieties. Among the Dachshund surgical cases in this study, 9.4% had no radiographic evidence of disc calcification. These results are similar to previous retrospective studies in Dachshunds and Pekingese that found that 13% and 17% of cases with disc extrusions had no radiographic calcification [
39,
40], likely reflecting low sensitivity (0.3–0.6) compared to histopathological assessment, as well as limitations of calcification as the sole marker for “clinically” relevant pathology [
41,
42].