Vascular Disease is Associated with Facet Joint Osteoarthritis

Abstract

Objective

Epidemologic studies have demonstrated associations between vascular disease and spinal degeneration. We sought to examine whether vascular disease was associated with lumbar spine facet joint osteoarthritis (OA) in a community-based population.

Design

441 participants from the Framingham Heart Study multi-detector computed tomography (MDCT) Study were included in this ancillary study. We used a quantitative summary measure of abdominal aortic calcification (AAC) from the parent study as a marker for vascular disease. AAC was categorized into tertiles of ‘no’ (reference), ‘low’, and ‘high’ calcification. Facet joint (FJ) OA was evaluated on CT scans using a 4-grade scale. For analytic purposes, FJ OA was dichotomized as moderate FJ OA of at least one joint from L2-S1 vs. no moderate FJ OA. We examined the association of AAC and FJ OA using logistic regression before and after adjusting for age, sex and BMI. Furthermore, we examined the independent effect of AAC on FJ OA after including the known cardiovascular risk factors diabetes, hypertension, hypercholesterolemia, and smoking.

Results

Low AAC (OR 3.84 [2.33-6.34]; p=<0.0001) and high AAC (9.84 [5.29-18.3]; =<0.0001) were strongly associated with FJ OA, compared with the reference group. After adjusting for age, sex, and BMI, the association with FJ OA was attenuated for both low AAC (1.81 [1.01-3.27]; p=0.05) and high AAC (2.63 [0.99-5.23]; p=0.05). BMI and age were independently and significantly associated with FJ OA. The addition of cardiovascular risk factors to the model did not substantially change parameter estimates for either AAC tertile.

Conclusions

Abdominal aortic calcifications were associated with FJ OA in this community-based population, when adjusting for epidemiologic factors associated with spinal degeneration, and cardiovascular risk factors. Potentially modifiable risk factors for facet degeneration unrelated to conventional biomechanical paradigms may exist. This study is limited by cross-sectional design; longitudinal studies are needed.

Introduction

Lumbar facet joint osteoarthritis (OA) is prevalent in 60-67% of adults in the general population¹. Facet joint (FJ) OA has been proposed as a common source of low back pain (LBP), and is seen frequently in patients with facet-mediated pain^2-3. The prevalence of lumbar facet joint pain based on controlled diagnostic blocks ranges from 15% in a population of injured US workers⁴, to 40-45% in specialty rheumatology and pain management practices^5-6. Nevertheless, epidemiologic studies of the association between radiographic FJ OA and LBP have had conflicting results^{1, 3, 7}. FJ OA may be related to other features of spinal degeneration and the production of pain through complex biomechanical processes².

Risk factors for extremity osteoarthritis (OA) include increased body mass index (BMI), female sex, genetic predisposition, joint injuries, and malalignment⁸. Although there have been few studies of risk factors for the development of FJ OA specifically, biomechanical factors are often implicated². The theory of the ‘spinal degenerative cascade’, popularized by Kirkaldy-Wallis, describes a sequence of degeneration, whereby changes in the lumbar disks of the anterior vertebral elements may lead to changes in the posterior vertebral elements of the spine, including the lumbar facet joints, and vice versa^9-11. Other biomechanical explanations for FJ OA include abnormalities in facet joint angulation, and facet joint asymmetry, though prior investigations of these risk factors have yielded inconsistent findings ^12-14. As with extremity OA, female sex has also been considered as a risk factor for FJ OA, although prior studies examining this relationship have either found no association between FJ OA and female sex, or a significant association with FJ OA only at certain spinal levels^{1, 15}.

Vascular disease has recently drawn attention as a potentially under-recognized risk factor for OA^16-17. Atherosclerosis has been proposed as a risk factor for OA progression^17-18. Prior studies have found associations between abdominal aortic calcifications (AAC) and the anterior vertebral changes of vertebral osteophytosis^19-20. Although vertebral osteophytes per se are not commonly considered to be direct causes of spine-related pain, the prior work raises the question of whether similar associations may exist with the posterior vertebral structures, including the lumbar facet joints.

The aims of the present study were: 1) to determine whether abdominal aortic calcification, as a marker of vascular disease, is associated with FJ OA in a community based population, adjusting for other risk factors for spinal degeneration, and 2) to examine the effect of controlling for the known cardiovascular risk factors on the association, if any, between AAC and facet joint OA.

Methods

Study Sample

This cross-sectional study was an ancillary project to the Framingham Heart Study. The Framingham Heart Study began in 1948 as a longitudinal population-based cohort study of the causes of heart disease. Initially, 5209 men and women between the ages of 30 and 60 years living in Framingham, Massachusetts were enrolled in the Original cohort. In 1971, 5,124 offspring of the Original cohort and their spouses were entered in the Offspring cohort²¹. In 2002, 4095 men and women who were children of the Offspring cohort were enrolled in the Third Generation cohort²². 3529 participants from the Offspring and Third Generation cohorts aged 40-80 years underwent abdominal multi-detector computed tomography (MDCT) to assess aortic calcification. The recruitment and conduct of CT scanning have been reported previously ^23-24. Subjects for this ancillary study were selected randomly from the MDCT cohort, with oversampling of the Offspring cohort to enrich the sample for older individuals. Individuals whose CT scans were of insufficient quality to allow assessment of FJ OA were excluded.

Imaging parameters

Study participants were imaged with an eight-slice MDCT scanner (Lightspeed Ultra, GE, Milwaukee, WI, USA). Each subject underwent unenhanced abdominal CT that was performed using a sequential scan protocol with a slice collimation of 8 mm × 2.5 mm (120 KVp, 320/400 mA for 220 lbs body weight, respectively) during a single end-inspiratory breath hold (typical duration 18 seconds). For the abdominal scan, thirty contiguous 2.5 mm thick slices of the abdomen were acquired covering 150 mm above the level of S1.

Quantitative AAC evaluation

The MDCT scans were quantified for the presence and quantity of AAC by an experienced reader using a dedicated off-line workstation. A calcified lesion was defined as an area of at least 3 connected pixels with CT attenuation > 130 Hounsfield units applying 3D connectivity criteria (6 points). A modified Agatston score was computed by multiplying each lesion area by a weighted MDCT attenuation score in Hounsfield units within the lesion. This method of scoring has been described elsewhere and has demonstrated reliability (ICC r>0.96)^23-24.

Facet Joint OA evaluation

FJ OA evaluation was performed using eFilm Workstation software (Version 2.0.0). All CT studies were read blinded to clinical information and to the results of the quantitative AAC evaluation. Lumbar facet joints were graded on both the left and right side at levels L2-L3, L3-L4, L4-L5, and L5-S1. Four grades of facet joint OA were defined using criteria designed for research purposes that have been used in multiple studies^{14, 25}, and are based on earlier criteria by Pathria et al.²⁶ and Weishaupt et al.²⁷ This system grades facet joint OA according to the grading of the individual subcategories of joint space narrowing (JSN), osteophytes, articular process hypertrophy, sclerosis, subarticular erosions, subchondral cysts, and vacuum phenomenon.

The following criteria were used for facet joint OA evaluation:

• Grade I (Normal): No JSN (joint space > 2mm); no osteophytes or possible small osteophytes; no articular process hypertrophy; no sclerosis or doubtful sclerosis; no subchondral erosions; no subchondral cysts; no joint space vacuum phenomenon.

• Grade II (Mild): Joint space 1-2 mm; and/or definite small osteophytes; and/or mild articular process hypertrophy; and/or definite sclerosis; no subchondral erosions; no subchondral cysts; no joint space vacuum phenomenon.

• Grade III (Moderate): Joint space <1 mm; and/or moderate osteophytes; and/or moderate articular process hypertrophy; and/or mild subchondral erosions; and/or mild subchondral cysts; and/or joint space vacuum phenomenon.

• Grade IV (Severe): Severe JSN (bone to bone); and/or large osteophytes; and/or severe articular process hypertrophy; and/or severe articular erosions; and/or severe subchondral cysts.

Examples of grading for FJ OA on CT are shown in Figure 1.

Figure 1. Examples of Facet Joint Osteoarthritis Grading.

Reliability of CT readings for facet joint osteoarthritis

CT assessment of facet joint osteoarthritis was performed by a board-certified, fellowship-trained physiatrist (PS) researcher specializing in spine care, who was trained by an experienced research musculoskeletal radiologist (AG). A standard atlas of FJ OA grading was created and used throughout the reading process. Calibration of the primary reader to the musculoskeletal radiologist was performed using a training set prior to the start of the formal reads, and intra- and inter-rater reliability were calculated for two readers at the start of the reading process. All CT scans were then analyzed in a blinded fashion. Recalibration of the reader to scans read previously by the radiologist was performed at additional time points during the reading process. To evaluate for reader-drift, intra-rater and inter-rater reliability was reassessed by inserting one repeated scan for every 10 new scans, and repeating reliability calculations at the middle and end of the reading process. Intra-observer reliability assessed with the κ statistic varied between 0.68 and 0.87, and inter-observer reliability varied between 0.68 and 0.84. This range of kappa statistics represents moderate to excellent reproducibility.

Covariates

Covariates were measured at the seventh Offspring and first Third Generation examinations, including information on age, sex, body mass index, and cardiovascular risk factors. Body mass index (BMI) was calculated as the ratio of weight (in kg) divided by height (meters²), and categorized based on the Classification of Overweight and Obesity by the National Heart Lung and Blood institute: Underweight/Normal (BMI <25.0 kg/m2), Overweight (BMI 25.0-29.9 kg/m2), Obesity I (BMI 30.0-34.9 kg/m2), Obesity II (BMI 35.0-39.9 kg/m2), and Obesity III (BMI 40.0+)²⁸. Due to small numbers of individuals in the Obesity II and Obesity III categories (likely due to a weight limit of < 350 lbs for inclusion in the MDCT study), these categories were combined. Fasting samples were used to measure plasma glucose and total cholesterol. Diabetes was defined as plasma glucose of 126 mg/dl or greater, current treatment with either a hypoglycemic agent or insulin, or a prior diagnosis of diabetes. Participants who reported smoking regularly within the past year were defined as current smokers. Hypertension was defined as systolic blood pressure of ≥140 mm Hg, diastolic blood pressure of ≥90 mm Hg, or the use of antihypertensive therapy. Hypercholesterolemia was defined as total cholesterol of ≥240 mg/dL.

Statistical analysis

In the absence of detailed prior data on FJ OA, this study was powered to detect a relative risk of 1.7, assuming β= 0.8, α=0.05, and a 40% prevalence of aortic calcification in controls, for a target recruitment of 452 CTs. 441 participants had CT scans that permitted adequate visualization of FJ OA. We initially characterized the sample using means and standard deviations for continuous variables, and frequencies and proportions for categorical variables. AAC measurements were highly right skewed, and roughly one third of the study sample demonstrated no AAC. We therefore categorized the quantitative AAC measurements into the tertiles of ‘no AAC’, ‘low AAC’ (Agatston score ≤959.2), and ‘high AAC’(Agatston score >959.2); this method has been used previously²⁹. In our clinical experience, and experience with prior studies of FJ OA, we have found that mild osteoarthritic changes in the facet joints as detected by CT scan are nearly ubiquitous in adults. We therefore planned a priori to use normal/mild FJ OA (grade I/II) as the reference group. The primary outcome was defined as the presence of moderate (grade III) FJ OA in any joint at the L2-S1 spinal levels. We first compared the subgroups of individuals with and without moderate FJ OA using the chi-square test for dichotomous variables, and the Student's T-test for continuous variables. We used bivariate logistic regression to produce odds ratios (ORs) for the association between AAC tertile and FJ OA, using no AAC as the reference group. We then used multivariate logistic regression to examine the relationship between AAC tertile and moderate FJ OA while adjusting for age, sex, and BMI category. Last, we added the cardiovascular risk factors diabetes, hypertension, hypercholesterolemia, and current smoking in the multivariate model.

In a series of secondary analyses, we further characterized the relationship between AAC and FJ OA. First, we examined the effect of treating AAC as a dichotomous variable (AAC absent vs. present). Next, we constructed a multivariate model including AAC tertile, age, sex, and BMI category, but instead using the outcome of severe (grade 3) FJ OA in any joint at the L2-S1 spinal levels. Last, we explored whether a dose response was present by treating the total number of facet joints with moderate OA as a count outcome. Negative binomial regression was used to model the association between the predictor variables and the total number of facet joints with moderate OA, to account for overdispersion with Poisson regression. All statistical analyses were performed using SAS software, (SAS Institute Inc, Cary, North Carolina, release 9.1).

Results

The study sample (n=441) had a mean age of 54.5±11.5 years, was 46.0% female, and had a mean BMI of 28.1± 5.2 kg/m². The study sample was slightly older than the general MDCT cohort (mean age 50.9), but comparable with respect to sex and BMI (48.1% female and 27.8 kg/mg², respectively). 40.8% of individuals were overweight, and 28.6% were obese. 34.9% of individuals had no AAC, 32.6% had low AAC, and 32.6% had high AAC. 37.1% of individuals had hypertension, 24.9% had hypercholesterolemia, 12.1% were current smokers, and 6.4% had diabetes. 70.3% of the sample had moderate FJ OA in at least one joint L2-S1, and 34.7% had severe FJ OA in at least one joint L2-S1.

Table 1 displays a comparison of individuals with and without moderate FJ OA. Increasing AAC tertile, age (years), female sex, and BMI category each were significantly associated with the presence of moderate FJ OA. The cardiovascular risk factors of hypertension and hypercholesterolemia were also significantly associated with moderate FJ OA, though diabetes and smoking in the past year were not.

Table 1. Comparison of individuals with and without facet joint osteoarthritis^* (FJ OA).

	No FJ OA (n=131)	FJ OA (n=310)	p-value‡
Abdominal Aortic Calcification (AAC)†
None	82 (53.6%)	71 (46.4%)
Low AAC	33 (23.1%)	110 (76.9%)	<0.0001
High AAC	15 (10.5%)	128 (89.5%)
Demographics
Age	46.7 ± 9.7	57.8 ± 10.6	<0.0001
Female Sex	51 (38.9%)	152 (49.0%)	0.05
BMI (kg/m2)
Normal/Underweight† (BMI <25.0 kg/m2)	59 (45.4%)	71 (54.6%)	<0.0001
Overweight (BMI 25.0-29.9 kg/m2)	47 (26.1%)	133 (73.9%)
Obesity I (BMI 30.0-34.9 kg/m2)	16 (18.4%)	71 (81.6%)
Obesity II/III (BMI 35.0+ kg/m2)	9 (23.1%)	30 (76.9%)
Cardiovascular Risk Factors
Diabetes	5 (3.8%)	23 (7.4%)	0.16
Current smoking	15 (11.5%)	38 (12.3%)	0.80
Hypertension	33 (25.2%)	130 (42.1%)	0.001
Hypercholesterolemia	22 (16.8%)	88 (28.4%)	0.01

^*Moderate facet joint OA at the L2-S1 level

^†Categorized by tertile of aortic abdominal calcification

^‡Chi-square test

Bivariate logistic regression using the predictor variable of AAC tertile and the response variable of moderate FJ OA yielded ORs and 95% confidence intervals (CIs) of 3.84 (2.32-6.34) for low AAC, and 9.84 (5.29-18.31) for high AAC, respectively. Adjusted ORs for all variables included in the multivariate logistic regression model are presented in Table 2; due to small numbers of individuals in obesity II and obesity III, these groups were combined. Increasing tertile of AAC was independently associated with moderate FJ OA when adjusting for age, female sex, and BMI category, though the difference in adjusted OR between low AAC (1.81[1.01-3.27]) and high AAC (2.34[0.99-5.23]) was not large. Increasing age in years and female sex were significantly associated with moderate FJ OA. Higher categories of BMI were associated with greater odds of moderate FJ OA for overweight and obesity I, but not for obesity II/III. However, when BMI was treated instead as a continuous variable, the independent effect of a unit (kg/m²) increase in BMI was 1.08 (1.02-1.14). The addition of cardiovascular risk factors to the model did not produce a substantial change in parameter estimates for AAC tertile (>15%), and no cardiovascular risk factor was independently associated with FJ OA (data not shown).

Table 2. Multivariate associations between predictors and facet joint osteoarthritis^* (FJ OA).

	Adjusted Odds Ratio (95% confidence interval)
Abdominal Aortic Calcification (AAC)†
None	reference
Low AAC	1.81 (1.01-3.27)
High AAC	2.34 (0.99-5.23)
Demographics
Age (year)	1.09 (1.06-1.13)
Female Sex	1.86 (1.11-3.12)
BMI (kg/m2)
Normal/Underweight† (BMI <25.0 kg/m2)	reference
Overweight (BMI 25.0-29.9 kg/m2)	2.83 (1.61-4.98)
Obesity I (BMI 30.0-34.9 kg/m2)	4.86 (2.27-10.4)
Obesity II/III (BMI 35.0+ kg/m2)	2.07 (0.80-5.37)

^*Moderate facet joint OA at the L2-S1 level

^†Categorized by tertile of aortic abdominal calcification

In secondary analyses, we modeled the association between the predictor variables from Table 2 (AAC, age, sex, and BMI category) and the response variable of moderate FJ OA, while treating AAC as a dichotomous variable. AAC remained independently and significantly associated with moderate FJ OA (OR 1.90 [1.07-3.37]. In order to evaluate relationships with more severe FJ OA, we then examined the relationship between AAC tertile and the outcome of severe FJ OA at the L2-S1 levels. In multivariate analyses including the predictor variables of AAC tertile, age, sex, and BMI category, and the response variable of severe FJ OA, low AAC was not significantly associated with severe FJ OA (OR 1.60 [0.85-3.01]), but high AAC was (OR 2.71 [1.26-5.82]). When the outcome of moderate FJ OA was treated instead as a count outcome for the total number of facet joints affected (0-8), the presence of low AAC (1.42; p=0.005) and high AAC (1.52; p=0.005) were independently and significantly associated with the number of arthritic joints. This means that individuals with low AAC or high AAC had a significantly greater average number of arthritic joints, as compared to those without AAC. For all secondary analyses, increasing age in years and female sex were significantly associated with the outcome of FJ OA (data not shown). Individual BMI categories were generally significantly associated with FJ OA, but this association was strongest when moderate FJ OA was treated as a count outcome. The addition of cardiovascular risk factors to these models did not produce substantial changes in parameter estimates for any categorization of AAC (data not shown).

Discussion

AAC was associated with the presence and the extent of moderate lumbar FJ OA in this community based population, after adjusting for the covariates of age, sex, and BMI category. A greater amount of AAC was associated with greater odds of having moderate FJ OA. Increased age, female sex, and increased BMI category were also significantly and independently associated with FJ OA. The association of AAC with FJ OA appeared to be independent of cardiovascular risk factors.

Although clinical studies have suggested a relationship between aortic calcifications and composite measures of disk degeneration, no prior study has examined the relationship between AAC and FJ OA. The relationship between AAC and OA in the posterior vertebral elements (such as in FJ OA) may be analogous to the relationship between AAC and OA in the anterior vertebral elements. Karasik et al conducted a study of the Original cohort of Framingham using plain radiographs, and found that AAC was associated with the presence of anterior lumbar osteophytes, but not with hand osteophytes¹⁹. These authors suggested two local mechanisms to explain the association of AAC with lumbar osteophytes, which are likely applicable also to the association of AAC and FJ OA. First, they proposed that decreased nutrient supply due to lumbar artery stenosis may cause disc degeneration, which may then lead to osteophytosis secondary to biomechanical changes and altered local blood flow¹⁹. This notion applied to FJ OA would suggest a direct, causal link between the vascular changes seen in AAC and FJ OA. Second, Karasik et al proposed that local inflammatory factors such as cyclooxygenase-2 (COX-2) may play a role in both arterial calcification and osteophytes. COX-2 is expressed both in osteophytes and atherosclerotic plaques, and may suggest an inflammatory process common to these two seemingly disparate problems. Indeed, systemic inflammation has been proposed elsewhere as a common risk factor for vascular disease and osteoarthritis^30-33. This notion applied to FJ OA suggests an indirect link-that AAC and FJ OA are each associated with a common important risk factor- and points away from a causal relationship. Other confounding factors associated with both AAC and FJ OA may also play a role, including different inflammatory mediators^33-35, the metabolic syndrome^36-37, dietary factors such as vitamin D^38-40, and calcium metabolism^41-42.

The increased prevalence of FJ OA with chronologic age is well known^{15, 43}. Our finding of an association between female sex and obesity and facet joint OA, however, is worthy of further mention. Despite a documented higher prevalence of OA in most other anatomic locations⁴⁴, prior epidemiologic studies did not find an association between female sex and FJ OA prevalence^{1-2, 45}. In fact, one prior study of cadaver specimens from the early 1900s concluded that male sex was associated with a greater prevalence of FJ OA at all lumbar levels⁴⁶. Our finding of a strong association between female sex and the prevalence of moderate and severe FJ OA is supported by immunohistochemical studies demonstrating estrogen receptors in facet cartilage, and a correlation between amount of estrogen receptor expression and FJ OA⁴⁷. To our knowledge, no prior study has reported an increased prevalence of FJ OA by BMI category, though this association is well-documented in epidemiologic studies of OA in other weight-bearing joints³⁵. Our study may have had advantages over prior studies in detecting these associations due to its large sample size and use of a well-developed measure of FJ OA.

The primary limitation of this study is its cross-sectional design, which precludes firm causal inferences. Nevertheless, this study demonstrates a strong and novel relationship between AAC and FJ OA, which can be further examined in prospective studies. Future studies should include factors which may mediate the relationship between AAC and FJ OA- including inflammation, sex hormones, and other factors mentioned above- while adjusting for the independent risk factors of age, sex, and BMI category, identified in the current study.

The relationship of vascular disease to facet joint osteoarthritis is potentially important, as biomechanical theories of degeneration may not account for all patterns of degeneration^{10, 48}. Our understanding of the mechanisms by which AAC is related to spinal degeneration may be complemented by the knowledge gained from future studies of the association between AAC and incident cardiovascular disease. These parallel advances hold the hope of discoveries which permit preventive strategies for musculoskeletal disease that overlap with the treatment of other conditions (such as cardiovascular disease) for which preventive care is already an accepted standard. Future studies of vascular disease and FJ OA should investigate mechanisms of degeneration, the production of pain, and should include a longitudinal design.