Familial Clustering and Genetic Analysis of Severe Thumb Carpometacarpal Joint Osteoarthritis in a Large Statewide Cohort

Abstract

Purpose:

Our goals were to 1) identify individuals who required surgery for thumb carpometacarpal osteoarthritis (CMCJ OA), 2) determine if CMCJ OA clusters in families, 3) define the magnitude of familial risk of CMCJ OA, 4) identify risk factors associated with CMCJ OA, and 5) identify rare genetic variants that segregate with familial CMCJ OA.

Methods:

We searched the Utah Population Database to identify a cohort of CMCJ OA patients who required surgery. Affected individuals were mapped to pedigrees to identify high-risk families with excess clustering of CMCJ OA. Cox regression models were used to calculate familial risk of CMCJ OA in related individuals. Risk factors were evaluated using logistic regression models. Whole exome sequencing was used to identify rare coding variants associated with familial CMCJ OA.

Results:

We identified 550 pedigrees with excess clustering of severe CMCJ OA. The relative risk of developing CMCJ OA requiring surgical treatment was significantly elevated in first- and third-degree relatives of affected individuals, and significant associations with advanced age, female sex, obesity, and tobacco use were observed. We discovered candidate genes that dominantly segregate with severe CMCJ OA in four independent families, including a rare variant in Chondroitin Sulfate Synthase 3 (CHSY3).

Conclusions:

Familial clustering of severe CMCJ OA was observed in a statewide population. Our data indicate that genetic and environmental factors contribute to the disease process, further highlighting the multifactorial nature of the disease. Genomic analyses suggest distinct biological processes are involved in CMCJ OA pathogenesis.

Clinical Relevance:

Awareness of associated comorbidities may guide the diagnosis of CMCJ OA in at-risk populations and help identify individuals who may not do well with non-operative treatment. Further pursuit of the genes associated with severe CMCJ OA may lead to assays for detection of early stages of disease and have therapeutic potential.

Introduction

Thumb carpometacarpal joint osteoarthritis (CMCJ OA) is a common form of hand OA and little is known about the biological mechanisms that are responsible for disease onset and progression.^{1, 2, 3, 4} The risk factors for developing CMCJ OA in cohorts defined by diagnostic codes or radiographic evidence include age,³ sex,^5–7 and hypermobility^{8, 9} Additionally, a recent study found that being overweight (men) and obesity (both sexes) are risk factors associated with CMCJ OA despite the joint not being a weight-bearing unit.¹⁰

Most CMCJ OA cohort studies are based on inclusion of individuals using diagnostic codes or radiographic evidence.^{3, 5–10} Radiographic severity does not correlate with clinical severity,⁵ and not all patients with a radiographic diagnosis have a clinical phenotype with pain or hand dysfunction.¹¹ These cohorts are likely heterogeneous and do not necessarily represent individuals with symptomatic CMCJ OA. Identification of individuals requiring surgical treatment for CMCJ OA captures those with potentially severe symptoms/disease, reduces cohort heterogeneity, and allows higher resolution to identify risk factors and genes that contribute to pathologic forms of CMCJ OA.

There are no drugs that specifically target the progression of CMCJ OA since few of the pathways that drive the disease have been identified.^{12, 13} Studies have indicated that CMCJ OA is heritable in closely related individuals^14–16 and others have identified risk loci and candidate genes.^17–22 Given that few genes have been associated with CMCJ OA, there is clearly a need to identify more pathways that have a strong contribution to disease susceptibility. One way to identify genes with a determinant effect on disease is to study families that have severe forms of disease.^23–27

We employed a unique resource, the Utah Population Database (UPDB), to 1) identify individuals with severe CMCJ OA (defined as requiring surgery), 2) determine if severe CMCJ OA clusters in large, multigenerational families, 3) define the magnitude of familial risk of CMCJ OA, 4) identify risk factors associated with CMCJ OA and quantify the magnitude of risk, and 5) discover rare genetic variants that segregate with severe familial CMCJ OA.

Materials and Methods

The Institutional Review Boards of the University of Utah (IRB # 79442) and Intermountain Healthcare (IRB # 1050554), and the Resource for Genetic and Epidemiologic Research approved this study.

Our study utilizes data drawn from the Utah Population Database (UPDB) (https://uofuhealth.utah.edu/huntsman/utah-population-database/). The UPDB provides person-based interlinked records documenting genealogy, medical records, and vital statistics for over 11 million individuals from the late 18th century to the present. Medical records derive from statewide facility data from the two largest healthcare providers in Utah (Intermountain Healthcare (IHC) and University of Utah Health (UUH)), Medicare claims, and the Utah and Idaho Cancer Registries. Privacy of individuals whose data is available through UPDB is strictly protected through the Utah Resource for Genetic and Epidemiological Research.

We identified individuals with severe CMCJ OA between 1996 and 2020 in the UPDB from statewide facility data (inpatient and ambulatory surgery) and IHC and UUH Enterprise Data Warehouse using the following diagnosis and related procedure codes: diagnostic codes for CMCJ OA - ICD-9 715·04 or 715·14; ICD-10 M18·0 or M18·1x or M19·04; procedure codes for CMC arthroplasty and fusion - ICD-9 8269, 8174, or 8175; CPT 25447 or 2544. Inclusion as a case required presence of one diagnostic code and one procedure code. To identify individuals with idiopathic CMCJ OA, we excluded individuals if they had a traumatic event to the thumb CMC joint, inflammatory arthritis, and ligamentous hyperlaxity (see Supplementary Methods). Controls were age- and sex-matched and were selected if they had no history of a CMCJ OA diagnosis or related procedure or were excluded based on criteria mentioned above (Supplementary Methods). Affected individuals were required to have relatives in the UPDB to be included in our study cohort so we could link them to pedigrees.

Results

We applied a strict definition of CMCJ OA to identify individuals with potentially severe forms of CMCJ OA. We defined “severe” CMCJ OA as having been diagnosed and treated using a surgical procedure (CMC fusion or arthroplasty). We searched 7,311,712 UPDB records and identified 4707 individuals with severe CMCJ OA. Mean birth year of the CMCJ OA cohort was 1950 (± 10.3), 76.1% were female, and 86.9% of individuals were white (Supplementary Table 2). We performed manual chart review on a random subset of 25 individuals to verify the CMCJ OA diagnosis (Supplementary Methods and Supplementary Fig. 1) and found that all 25 individuals had a verifiable CMCJ OA diagnosis confirmed by radiographic evidence.

To test if there is significant familial clustering of severe CMCJ OA in our cohort, we analyzed cases with severe CMCJ OA that linked to a pedigree using the Familial Standardized Incidence Ratio (FSIR) calculation.^{25, 27, 28} We identified 550 unrelated, multigenerational, high-risk pedigrees that had at least 4 living members in the UPDB and an increased clustering of severe CMCJ OA, defined by a FSIR ≥ 2.0 (p-value < 0.05). We set the threshold FSIR value at 2.0 to select for families in which the incidence of disease is two times higher than the expected incidence in control pedigrees. Of the 550 high-risk pedigrees, the FSIR ranged from 2.03 to 21.08 (mean 3.4 ± SD 1.8). Founder birth year, number of descendants, number of affected individuals, and FSIR values are indicated for 10 representative high-risk pedigrees in Table 1. Figure 1 illustrates two multigenerational high-risk pedigrees. One pedigree has at least 20 known affected individuals and a FSIR of 2.1 (Figure 1A) and another has at least 4 known affected individuals and a FSIR of 4.6 (Figure 1B), indicating a 2.1- and 4.6-fold increase respectively, in CMCJ OA incidence in these families compared to control pedigrees. The identification of high-risk pedigrees indicates significant familial clustering of severe CMCJ OA in our cohort.

Table 1 -.

High-Risk Pedigrees with Excess Familial Clustering of Thumb Carpometacarpal Joint Osteoarthritis

Founder Birth Year	Number of Descendants	Number of Affected Individuals	FSIR#
1766*	22,460	20	2.1
1740	22,450	24	2.1
1788	16,344	19	2.
1761	11,909	15	3.7
1754	5,105	12	4.1
1788	4,628	8	6.3
1815	1,127	7	7.8
1783	2,228	6	9.1
1809	979	5	12.5
1848	525	4	21.1

^*indicates pedigree represented in Fig. 1A.

^#indicates that all FSIR P-values are P < 0.05.

Figure 1 –

High-risk thumb carpometacarpal joint osteoarthritis (CMCJ OA) pedigrees. (A, B) Examples of two high-risk pedigrees segregating severe CMCJ OA identified from the Utah Population Database. Circles = females, squares = males, slash = deceased. Filled circles/squares = affected individuals; open circles/squares = individuals with an unknown CMCJ OA diagnosis. C, D E, and F indicate subjects for whom radiographs are shown. FSIR of pedigree A is 2.1 and pedigree B is 4.6. (C-F) Hand radiographs of individuals in the pedigrees shown in A and B.

To determine whether there is an increased risk of CMCJ OA among closely related individuals, we examined the relative risk of developing CMCJ OA in first-, second-, and third-degree relatives (Supplementary Table 3). The risk of developing CMCJ OA was approximately 3.62-fold greater in first-degree relatives of both sexes of patients compared to controls (RR, 3.62 [95% CI, 2.7-4.86], P < 0.001), while males and females had a different magnitude of risk; males: RR= 6.26 [95% CI 3.7-10.6], P < 0.001; females: RR=3.14 [95% CI 2.28-4.32], P < 0.001) (Table 2). We were unable to detect a significantly elevated risk of CMCJ OA in second-degree relatives. For third-degree relatives, we observed a slight increase in relative risk for CMCJ OA development in both sexes (RR 1.18 [95% CI, 1.00-1.39], P = 0.048), but higher in males (RR=1.38[CI 1.01-1.88], P = 0.042) (Table 2). Our data indicate that within CMCJ OA families there is elevated risk of disease among family members and the risk for males is greater than the risk for females. Together with the familial clustering of severe CMCJ OA, these data indicate a genetic component to the disease.

Table 2 -.

Increased Familial Risk of Thumb Carpometacarpal Joint Osteoarthritis

Both Sexes
	N of relatives of CMCJ OA cases	N of relatives of matching controls	P-value	Relative Risk	95% CI Lower Limit	95% CI Upper Limit
Had a FDR with CMC	145	195	<0.001	3.62	2.70	4.86
Had a SDR with CMC	44	167	0.208	1.24	0.89	1.75
Had a TDR with CMC	207	775	0.048	1.18	1.00	1.39
Male
	N of relatives of CMCJ OA cases	N of relatives of matching controls	P-value	Relative Risk	95% CI Lower Limit	95% CI Upper Limit
Had a FDR with CMC	39	30	<0.001	6.26	3.70	10.60
Had a SDR with CMC	7	30	0.835	1.09	0.48	2.50
Had a TDR with CMC	60	190	0.042	1.38	1.01	1.88
Female
	N of relatives of CMCJ OA cases	N of relatives of matching controls	P-value	Relative Risk	95% CI Lower Limit	95% CI Upper Limit
Had a FDR with CMC	106	165	<0.001	3.14	2.28	4.32
Had a SDR with CMC	37	137	0.197	1.28	0.88	1.85
Had a TDR with CMC	147	585	0.267	1.11	0.92	1.35

FDR = first-degree relatives; SDR = second-degree relatives; TDR = third-degree relatives

FDR include mothers, fathers, sons, daughters, and full siblings. SDR include grandparents, grandchildren, half-siblings, aunts, uncles, nieces, and nephews. TDR include cousins, grand-nieces and grand-nephews, great-aunts and great-uncles, great-grandchildren, great-grandparents, half-aunts and half-uncles, and half-nieces and half-nephews.

OA in general is more prevalent in females, and the same has been reported for CMCJ OA, with almost three quarters of affected individuals identified as female.^{1, 2, 29} To determine if there is an age and sex bias associated with severe CMCJ OA, we examined age-standardized sex-specific incidence rates in our cohort from 1996-2020. We found a significant association of sex and age with severe CMCJ OA in our cohort. Of 6,083,729 unaffected individuals and 4,540 CMCJ OA cases, a higher proportion of cases were female (76.3% of CMCJ OA cases versus 48.7% of controls) and were older (mean ± SD birth year 1950.3 ± 10.3 for CMCJ OA cases and 1980.2 ± 24.6 for controls) (P <0.001) (Supplementary Table 4). We also determined that females had a significantly higher rate of severe CMCJ OA from the ages of 30–89 years compared to males, with the highest female-to-male incidence ratio being 4.68 (95% CI 4.07-5.37) in the age group 50–59 years (Table 3). Our results indicate that being female of advanced age is a significant risk factor for severe CMCJ OA in the general population.

Table 3 -.

Age-Specific Incidence Rates of Thumb Carpometacarpal Joint Osteoarthritis by Sex and Female-to-Male Incidence Ratios

Age, years	Male		Female		Female-to-male ratio (95% CI)
	N cases	Rate Per 10000	N cases	Rate Per 10000
< 20	<11	0.002	0	0	-
20-29	<11	0.003	<11	0.003	1.04 (0.15, 7.41)
30-39	<11	0.016	28	0.058	3.70 (1.69, 8.12)
40-49	57	0.136	235	0.581	4.28 (3.21, 5.72)
50-59	241	0.721	1120	3.370	4.68 (4.07, 5.37)
60-69	466	2.030	1324	5.550	2.73 (2.46, 3.04)
70-79	251	1.860	638	4.170	2.25 (1.94, 2.60)
80-89	46	0.770	108	1.330	1.72 (1.22, 2.43)
90+	<11	0.321	<11	0.057	0.18 (0.02, 1.71)

We analyzed several risk factors to determine if they are associated with severe CMCJ OA in our cohort (Supplementary Table 1).^29–31 We examined the association of severe CMCJ OA with obesity, diabetes, tobacco use, alcohol use, and being related to an individual diagnosed with and surgically treated for CMCJ OA (to the third degree) in the same cohort used for the age- and sex-specific analysis. We examined the relative risk of these risk factors independently in males and females while adjusting for demographic features (race and ethnicity), and diabetes (Methods and Supplementary Table 3). When adjusting for diabetes, obesity was identified as a risk factor for males (RR of 1.59 [95% CI 1.36-1.86]) and females (RR 1.24 [95% CI 1.14-1.36]). Tobacco use was a risk factor for both sexes (males: RR 1.33 [95% CI 1.15-1.53]; females: RR 1.45 [95% CI 1.33-1.58]) (Table 4). Independent of diabetes, having a first-degree relative (males: RR 3.78 [95% CI 2.43-5.89]; females: RR 3.21 [95% CI 2.46-4.19]) or a third-degree relative (males: RR 1.58 [95% CI 1.15-2.18]) with severe CMCJ OA was a significant risk factor (Table 4). We found the same risk factors when not adjusting for diabetes (Table 4). Our analyses indicate that Hispanics are slightly less likely to develop CMCJ OA (non-white vs white and Hispanic vs non-Hispanic for both sexes, irrespective of adjustment for diabetes, e.g., both sexes adjusted for diabetes - RR 0.39 [95% CI 0.33-0.46]) (Supplementary Table 5). These data indicate that obesity, tobacco use, and having a first- and third-degree relative with severe CMCJ OA are all significant risk factors for severe CMCJ OA in the general population.

Table 4 -.

Risk Factors Associated with of Thumb Carpometacarpal Joint Osteoarthritis

Males – adjusted for diabetes
	CMCJ OA cases	Controls	P-value	Relative Risk	95% CI Lower Limit	95% CI Upper Limit
Ever diagnosed with alcoholism	55	340	0.173	0.81	0.59	1.10
Ever diagnosed with diabetes	348	1571	0.402	1.07	0.92	1.25
Ever diagnosed with obesity	327	1169	<0.001	1.59	1.36	1.86
Ever smoked/used tobacco	417	1862	<0.001	1.33	1.15	1.53
Had at least one FDR with CMCJ OA	36	49	<0.001	3.78	2.42	5.89
Had at least one TDR with CMCJ OA	55	180	0.005	1.58	1.15	2.18
Females – adjusted for diabetes
	CMCJ OA cases	Controls	P-value	Relative Risk	95% CI Lower Limit	95% CI Upper Limit
Ever diagnosed with alcoholism	113	471	0.454	1.09	0.87	1.35
Ever diagnosed with diabetes	843	4469	0.175	0.94	0.85	1.03
Ever diagnosed with obesity	1104	4900	<0.001	1.24	1.14	1.36
Ever smoked/used tobacco	952	3778	<0.001	1.45	1.33	1.58
Had at least one FDR with CMCJ OA	96	151	<0.001	3.21	2.46	4.19
Had at least one TDR with CMCJ OA	132	616	0.722	1.04	0.85	1.26
Males only – not adjusted for diabetes
	CMCJ OA cases	Controls	P-value	Relative Risk	95% CI Lower Limit	95% CI Upper Limit
Ever diagnosed with alcoholism	55	340	0.172	0.81	0.59	1.10
Ever diagnosed with obesity	327	1169	<0.001	1.62	1.39	1.88
Ever smoked/used tobacco	417	1862	<0.001	1.33	1.16	1.54
Had at least one FDR with CMCJ OA	36	49	<0.001	3.78	2.43	5.89
Had at least one TDR with CMCJ OA	55	180	0.005	1.58	1.15	2.18
Females – not adjusted for diabetes
	CMCJ OA cases	Controls	P-value	Relative Risk	95% CI Lower Limit	95% CI Upper Limit
Ever diagnosed with alcoholism	113	471	0.448	1.09	0.87	1.35
Ever diagnosed with obesity	1104	4900	<0.001	1.22	1.13	1.32
Ever smoked/used tobacco	952	3778	<0.001	1.44	1.32	1.57
Had at least one FDR with CMCJ OA	96	151	<0.001	3.21	2.46	4.19
Had at least one TDR with CMCJ OA	132	616	0.717	1.04	0.85	1.26

FDR = first-degree relatives; SDR = Second-degree relatives; TDR = third-degree relatives.

All conditions are yes vs. no.

To identify genes that are major risk factors for development of severe CMCJ OA, we analyzed the exomes of four high-risk pedigrees in which CMCJ OA segregated as an apparent autosomal dominant trait (Fig. 2A and Table 5).^{24, 32} In the pedigree designated CMC4 we analyzed the exomes of one unaffected and five affected individuals from a family (Fig. 2A and B). We identified a rare coding variant in Chondroitin Sulfate Synthase 3 (CHSY3). (PVAAST: p-value = 4.7x10⁻⁵; LOD 1.5; PHEVOR score: 4.1, final rank = 5) (CHSY3; NM_175856: exon3:c.G1885A:p.G629R, rs145272862, MAF 0.0001312). The rs145272862 SNP results in the non-synonymous substitution of an invariant glycine (in the vertebrate lineage) to an arginine in the chondroitin N-acetylgalactosaminyltransferase domain of CHSY3 (Fig. 2C). CHSY3 is a glycosyltransferases involved in initiation and elongation of the of the chondroitin sulfate glycosaminoglycan side chains on a core protein,³³ including the major chondroitin sulfate proteoglycan of cartilage, Aggrecan (ACAN).

Figure 2 –

A dominant CHSY3 mutation segregates with thumb carpometacarpal joint osteoarthritis (CMCJ OA). (A) CMC4 pedigree. Severe CMCJ OA segregates as an apparent autosomal dominant trait. Arrow marks the proband (II-2). The founder (deceased, I-0) was not genotyped, but had a medical history of CMCJ OA. Exomes were sequenced from two generations of the family (individuals II-1-4 and III-5 and 6). All affected individuals genotyped (II-1-4 and III-6) were heterozygous for the rare variant (rs145272862) and the unaffected individual (III-5) genotyped was homozygous for the reference allele. (B) Right hand radiograph of the proband demonstrating CMCJ OA. (C) Schematic diagram of the CHSY3 protein indicating the transmembrane domain (TM) and the chondroitin N-acetylgalactosaminyltransferase domain and the location of the p.Gly629Arg mutation in the catalytic domain.

Table 5.

CMCJ OA Family and Candidate Gene Details

Family	OA Phenotypes and Individuals Analyzed	Candidate Gene(s)	SkeletalVis Expression				PolyPhen/SIFT Prediction
			log2 fold change (adj. p-value)	Accession Number	Tissue	Description and Comparison
CMC4	Four affected siblings (two female and two male) and two daughters (one affected and one unaffected) of one affected individual. Pedigree in Figure 2.	CHSY3 - rs145272862, MAF - 0.0001312, NM_175856: exon3:c.G1885A:p.G629R	1.4 (0.00000451)	22659600	Human cartilage	human hip OA; OA vs nonOA	Polyphen - probably_damaging SIFT - deleterious
CMC1	Two affected sisters.	B4GALNT2 - rs565993034, MAF - 0.00005657, NM_001159387:exon3:c.G323T:p.R108l	1.24 (6.17e-15)	GSE74220	Human cartilage	4h IL-1β stimulation of primary human chondrocytes; IL1B vs Control	Polyphen - probably_damaging SIFT - deleterious
		LTF - rs199678659, MAF -0.0003113, NM_001199149:exon16:c.C1898T:p.T633I	−2.9 (3.2e-7)	GSE51588	Human bone	Subchondral Bone in Osteoarthritis; OA-vs Normal-Subchondral bone from medial tibial
CMC3	Four affected siblings (three female and one male).	STAT4 - rs761161672, MAF - 0.00002830, NM_001243835:exon2:c.A125G:p.D42G	1.99 (0.0307)	GSE10575	Human cartilage	Osteoarthritis chondroprogenitor cells; Osteocytes/-blasts from human bone chips of patients with OA vs healthy chondrocytes	Polyphen - probably_damaging SIFT - deleterious
		TRPM3 - rs200939544, MAF - 0.0002024, NM_001007471:exon10:c.C1358T:p.S453L	1.24 (0.00883)	E-MTAB-6266_B	Human cartilage	Human intact knee osteoarthritic subgroups and nonOA cartilage; OA vs nonOA
CMC743	Two affected (female - seperated by 12 meioses) and two unaffected (female) individuals.	MERTK - rs199779970, MAF - 0.0003748, NM_006343:exon5:c.C791G:p.A264G	−1.01 (0.00000988)	E-MTAB-6266_B	Human cartilage	Human intact knee osteoarthritic subgroups and nonOA cartilage; OA vs nonOA	Polyphen - probably_damaging SIFT - deleterious
		DOK3 - rs146755459, MAF - 0.0004378, NM_001144875:exon3:c.C193T:p.R65W	−0.987(6.81e-7)	GSE51588	Human bone	Subchondral Bone in Osteoarthritis; OA-vs Normal-Subchondral bone from medial tibia

To define biological pathways that confer susceptibility to CMCJ OA, we analyzed three additional CMCJ OA pedigrees using the above methods. In the CMC1 pedigree, we identified candidate variants in Beta-1,4-N-acetyl-galactosaminyltransferase 2 (B4GALNT2) and Lactotransferrin (LTF). We identified variants in Signal transducer and activator of transcription 4 (STAT4) and Transient receptor potential cation channel subfamily M member 3 (TRPM3) in the CMC3 pedigree and variants in MER proto-oncogene, tyrosine kinase (MERTK) and Docking protein 3 (DOK3) in the CMC743 pedigree (Table 5). These genes were shown to be expressed in joint tissues associated with homeostasis and disease (using SkeletalVis³²) and the variants were predicted to be damaging/deleterious using SIFT and Polyphen (Table 5). These data indicate that multiple biological processes are involved in CMCJ OA pathogenesis.

Discussion

In this study, we used a unique statewide medical genetics resource, the UPDB, to identify a cohort of individuals diagnosed with severe CMCJ OA. Previous studies primarily focused on CMCJ OA cohorts defined by radiographic evidence of OA, which does not correlate with symptomatic severity.^5,11 Our study focuses on a unique population of CMCJ OA individuals; those with symptoms severe enough to require surgical management of symptoms (CMC fusion or arthroplasty). From this cohort of individuals with severe CMCJ OA we have i) demonstrated familial enrichment of severe CMCJ OA, ii) determined that first- and third-degree relatives of an individual with severe CMCJ OA are at approximately 3.66-fold and 1.19-fold increased risk of developing the disease, respectively, iii) determined that age, sex, obesity, tobacco-use, and having a first- or third-degree relative with severe CMCJ OA are significant risk factors associated with the disease, and iv) and identified rare, dominantly segregating coding variants in four CMCJ OA pedigrees. These data suggest that both genetic and physiological factors contribute to the development of severe CMCJ OA in a large population-based cohort.

The risk factors in our surgical CMCJ OA cohort are consistent with other studies that have examined risk factors for CMCJ OA in other populations.^{2, 5–7, 9, 10, 16} Although prior studies have associated tobacco use with less hand OA,³⁰ this is the first time tobacco use has been associated with increased risk of CMCJ OA. Awareness of these comorbidities may help guide the clinical diagnosis of this condition in at-risk populations, particularly those with affected family members, and assist in the identification of individuals who are unlikely to do well with non-operative treatment. The mechanisms by which environmental and physiological risk factors interact with an individual’s genetic background to contribute to CMCJ OA remain to be elucidated.

CMCJ OA is highly heritable,^14–16 yet few coding variants have been definitively linked to CMCJ OA.^{17, 18, 20–22, 34} Using the UPDB enabled us to discover rare coding variants in high-risk pedigrees, a powerful way to define pathways with a determinant effect on disease development.^{23, 24, 35, 36} Our study is the first to identify a large number of multigenerational severe CMCJ OA pedigrees, determine relative risk among relatives, and identify a coding variants associated with CMCJ OA.

To define pathways that contribute to CMCJ OA, we performed genomic analyses on four high-risk pedigrees and found multiple genes that may have roles in OA pathogenesis (Fig 2 and Table 5). We identified a rare, dominantly segregating coding variant (rs145272862, c.G1885A:p.G629R) in CHSY3. CHSY3 is a glycosyltransferase that initiates and adds chondroitin sulfate glycosaminoglycan side chains to a core protein to from a functional proteoglycan.³³ The variant is predicted to be damaging and is located in a highly conserved amino acid in the enzymatic domain of CHSY3 (Fig. 2 C and Table 5). One hypothesis is the disease allele reduces chondroitin N-acetylgalactosaminyltransferase activity of CHSY3, thereby limiting initiation or elongation of chondroitin sulfate glycosaminoglycan side chains to the core protein. Aggrecan (ACAN) is the major chondroitin sulfate proteoglycan in cartilage, and loss of chondroitin sulfate concentration and reduction of side chain length is associated with OA severity.³⁷ A reduction in CHSY3 activity may alter the chondroitin sulfate glycosaminoglycan content of ACAN, thereby leading to the initiation of cartilage catabolism and onset of OA. Consistent with this, mice lacking Chsy3 display intervertebral disc degeneration, including loss of Acan expression and increased expression of catabolic factors in nucleus pulposus tissue.³⁸ Modification of the chondroitin core protein is a recurrent risk factor for OA as loci containing other enzymes in the pathway have been identified in OA GWAS,^{20, 39} making this pathway an appealing candidate for therapeutic intervention and early identification of at-risk individuals.

We sequenced exomes from three additional independent pedigrees and found several candidate genes (Table 5). In the CMC1 pedigree, we identified variants in B4GALNT2 and LTF. B4GALNT2 is postulated to have potentially protective effects in OA pathogenesis because it was found to be highly expressed in the MRL/MpJ “super healer” mice⁴⁰ in the context of injury-induced OA. LTF is a glycoprotein that has a role in bone homeostasis⁴¹ and is differentially expressed in subchondral bone of OA patients.⁴² In pedigree CMC3, we discovered variants in STAT4 and TRPM3. STAT4 is involved in inflammatory signaling and is highly expressed in the context of inflammatory arthritis. ⁴³ TRPM3 is associated with rheumatoid arthritis⁴⁴ and is co-expressed with mir-204, a stress induced factor that contributes to OA.⁴⁵ In pedigree CMC743, we found variants in MERTK and DOK3. MERTK is a tyrosine kinase family member that has been proposed to be a viable therapeutic target in rheumatoid arthritis.^{46, 47} DOK3 is a negative regulator of the JNK pathway that has a role in bone remodeling.⁴⁸ While were unable to define a single candidate gene in these three additional pedigrees, functional studies will identify the causal gene and provide insight into the biological mechanism of how modulation of these genes leads to the onset and progression of OA.

Our study has several limitations. Our definition of “severe” CMCJ OA (requiring surgical intervention) may be dependent on several factors. There are no concrete indications for CMCJ OA surgery other than pain and disability that is severe enough (from the patient’s perspective) to merit surgical intervention, together with failure of nonoperative treatment (splinting > 6 weeks, NSAIDs, steroid injection). In addition, there is conflicting literature regarding the relationship between radiographic severity and level of pain and disability. While it is possible that disease/radiographic severity and clinical severity may not be related, this definition allows us to focus on a specific CMCJ OA cohort that has not previously been studied and makes identifying families with the phenotype less ambiguous. The relative risk and FSIR calculations are likely underestimates for CMCJ OA in our cohort due to several factors. First, our cohort was limited to individuals identified through medical coding and our analysis can only identify individuals diagnosed in Utah. We are thus missing individuals diagnosed out of state and affected individuals who have not sought medical care. We have a small number of ethnic minorities in our study, which reflects the overall population of Utah. While we detected significant differences in risk factors in the Hispanic population, given the small number of individuals we are cautious not to overinterpret this data until a larger cohort can be studied.

Data Statement:

Data are available upon request.