Genetics of Cruciate Ligament Rupture 61significant locus on chromosome 24. The major-
ity of the carbohydrate-binding genes also had a
role in immune function as pattern-recognition
receptors. Of note, this cluster also included
aggrecan (ACAN), hyaluronan and proteogly-
can link protein 3 (HAPLN3), also known as
cartilage link protein. Aggrecan and link pro-
tein are important for maintaining hydration in
collagenous tissues. Tissue hydration is impor-
tant for the efficient distribution of mechanical
load and cellular repair. A link between aggre-
can and canine CR is also supported by earlier
work (Wilkeet al. 2009; Wilke 2010), which iden-
tified an association with a microsatellite within
aggrecan core protein precursor. The upreg-
ulation of aggrecan has also been associated
with human ACL rupture (Mannionet al. 2014;
Johnsonet al. 2015), as well as equine degen-
erative suspensory ligament desmitis (DSLD),
a debilitating disorder of horses that leads to
collagen disruption and eventual rupture of the
suspensory ligament (Plaaset al. 2011).
While GWAS has most often been used for
the identification of candidate genes and bio-
logical pathways that may be contributing to
disease pathogenesis, this approach may also
be used to develop genomic prediction algo-
rithms for genetic screening. Here, research is
less concerned with the biological effect of a
mutation tagged by a SNP, but instead focuses
on the statistical effect of a particular geno-
type and its ability to predict a disease out-
come, often in combination with other SNP
genotypes. In a recent study, GWA was per-
formed separately in a population of 46 New-
foundlands (22 cases and 24 controls) and 333
Labrador Retrievers (190 cases and 143 controls)
using a mixed linear model approach. Associ-
ated SNPs from GWAS were used to develop a
classification tree, which is a statistical method
that evaluates each SNP for its classifying abil-
ity and selects the best SNPs to create a clas-
sification model. In Newfoundlands, 19 SNPs
were used for diagnostic model assessment. The
model selected three SNPs for best classifica-
tion. Cross-validation of the model yielded an
area under the receiver operator characteristic
(ROC) curve of 95.5%, indicating that the model
had a good ability to classify cases and controls
in the sample population. In Labrador Retriev-
ers, 13 SNPs were used in the same procedure.
The diagnostic model selected all 13 SNPs for
best classification, with an area under the ROC
curve of 88.4%, indicating that the model was
also able to classify cases and controls in the
sample population, but slightly reduced com-
pared to Newfoundland dogs (Wilkeet al. 2015).
The SNPs identified in this GWAS did not over-
lap with regions identified in other Newfound-
land or Labrador GWAS (Wilkeet al. 2009; Baird
et al. 2014b; Bakeret al. 2017). While the results
of these studies are promising, they should
not be over-interpreted as the population used
for GWAS was also used for diagnostic model
development and testing. A truly predictive
model must be tested in new populations that
were not used to train the model. Nevertheless,
these results speak to the concept that develop-
ment of a genetic test for CR is an achievable
research goal, as accurate stratification of cases
and controls can be achieved (Wilkeet al. 2015;
Bakeret al. 2017).Canine and human cruciate
ligament ruptureThe young (human) female athlete is another
cohort that shares characteristics of the CR phe-
notype found in dogs. This cohort experiences
anterior cruciate ligament (ACL) injuries two to
eight fold more frequently than their counter-
part young athletic males (Arendt & Dick 1995;
Gwinnet al. 2000; Lohmanderet al. 2004). ACL
tears in people often involve contact trauma;
however, in this select population ACL rup-
ture usually occurs via a non-contact mecha-
nism (Arendt & Dick 1995).
Several studies have evaluated the potential
genetic contribution to ACL injuries in humans.
The first was a case-control study that found
those with an ACL injury to be twice as likely to
have had a family member with an ACL injury
as those study participants that did not have an
ACL injury (Flynnet al.2005). Additionally, a
more recent study noted that individuals with
a first-degree relative with an ACL injury were
at 2.2 times greater risk of both graft rupture
and contralateral injury (Websteret al. 2014).
These findings support a familial predisposition
to ACL rupture.
Further research has targeted mutations in
collagen genes as risk factors for ACL injuries.
Two separate studies have evaluated a muta-
tion (G1023T; rs1800012) in intron 1 ofColla-
gen Type 1 alpha 1(COL1A1), the binding site