An Ongoing Assessment of Osteoarthritis in African Americans and Caucasians in North Carolina: The Johnston County Osteoarthritis Project

Abstract

Osteoarthritis (OA) is the most common type of arthritis and is frequently associated with significant disability. Its public health impact is increasing due to the aging of the population and the obesity epidemic. The Johnston County Osteoarthritis Project is an ongoing, population-based prospective cohort begun in 1990 to fill knowledge gaps about prevalence, incidence, and progression of OA, and its risk factors, in African American and Caucasian men and women in North Carolina. Critically important phenotypic differences were observed in patterns of multi-joint OA burden, with African Americans much less likely than Caucasians to have hand OA and much more likely to have multiple large joint involvement. Racial differences also exist in systemic bone and joint tissue biomarkers. Novel potentially modifiable risk factors identified in this cohort include selenium and blood lead levels. Selected key findings of this ongoing study will be discussed.

INTRODUCTION

Osteoarthritis (OA) is the most common joint condition worldwide, affecting 27 million people in the United States in 2005 (1). Joints affected include the knee, hip, hand, great toe, and spine; OA frequently occurs in multiple joints. Given its predilection to knees and hips, OA impacts mobility and is associated with significant disability, demand for total joint replacement, and health care utilization (2,3). Knee OA is strongly associated with aging and obesity, each increasing in our society; as such, the demand for total knee replacement is estimated to reach almost 3.5 million procedures annually in the United States by 2030 (4).

We have long known that OA is uncommon in those younger than 40 years, unless accompanied by an underlying metabolic disorder or a prior joint injury (2). Early insights into OA etiology emphasized it as a non-inflammatory “wear and tear” condition primarily affecting the articular cartilage, which occurred almost inevitably with aging (2). We now appreciate that OA is a condition affecting all joint tissues, including synovium, bone, meniscus, tendon, and joint capsule. Additionally, chronic inflammation does indeed occur in OA, with activation of inflammatory cytokines and chemokines, leukotrienes and prostaglandins, adipokines, complement, and proteolytic enzymes (5,6).

Knowledge about the role of aging in OA has also evolved, and OA is no longer considered an inevitable consequence of aging. Instead, aging renders the joint increasingly susceptible to oxidative and catabolic damage and less responsive to anabolic factors, effects that can be compounded by altered biomechanics from age-related sarcopenia and diminishing strength and function (7,8).

What did we know about the epidemiology of OA in the late 1980s, when the Johnston County OA Project (JoCo OA) was being designed? Indirect descriptions of blacks in Africa and the Caribbean, compared to European Caucasians, suggested that blacks had lower rates of hip OA (9,10); however, the first National Health and Nutrition Examination Survey, conducted in the United States in the 1970s, did not find differences in prevalence of radiographic hip OA in African Americans and Caucasians (11). The same survey suggested that black women in the United States were twice as likely as their Caucasian counterparts to have knee OA, whereas the prevalence of knee OA in men was not statistically significantly elevated (12). Methodological issues limited the utility of these cross-sectional surveys, and little was known about potential explanatory factors for these observed differences or the impact of OA in these demographic sub-groups.

The JoCo OA was designed as a population-based prospective cohort of knee and hip OA to respond to these knowledge gaps in African American and Caucasian men and women (13). The study has been continually funded by the Centers for Disease Control and Prevention since 1990, with additional funding from multiple entities of the National Institutes of Health and private sources since 1993. More than 40 ancillary studies have been appended to the parent study, examining various risk factors associated with radiographic OA, joint symptoms, multiple chronic conditions, and disability. This article describes selected results from this study, particularly emphasizing racial differences in various aspects of OA.

METHODS

A critically important facet of the study design was its sampling and recruitment strategies that would allow population estimates of prevalence and incidence that could be generalized or applied to other populations in the United States. Such estimates would also be useful to determine burden of disease and need for health services.

A detailed description of the sampling design can be found in the appendix of the article by Jordan et al (13). In short, at baseline, household enumeration of six townships, characterized by their having a small town surrounded by a largely rural area and adequate numbers of African Americans for the study aims, was conducted. Households were mapped and visited by study personnel, and approximately 3,200 African American and Caucasian men and women aged 45 years and older were recruited, with over-sampling of African Americans. Eligible individuals completed two interviewer-administered home interviews and a clinic examination at which radiographs of the knees and hips, joint examination, functional assessment, anthropometrics, and venipuncture occurred.

The sample has been followed approximately every 5 years with repeat questionnaires; radiographic, clinical, and functional examinations; and venipuncture. Additional radiographic assessment of the hand, lumbar spine, foot, and ankle have been performed over the years; other data have also been collected including bone density and body composition, foot biomechanics, and whole blood and toenail assessment of various metal exposures, to name a few. Cohort refreshment in 2003−2004 enrolled an additional 1,100 people.

Statistical analyses have varied according to the analyses described. Analyses of radiographic differences between African American and Caucasians were predominantly evaluated by multiple logistic regressions controlling for age and body mass index (BMI). Differences in severity or other outcomes with more than two levels were examined by proportional odds modeling. Assessment of serum biomarkers as outcomes were conducted with linear regression with transformation of biomarker levels if assumptions of normality were violated. Specific analytic issues will be discussed with each selected result if required for clarification or interpretation.

All radiographs were scored for Kellgren-Lawrence (K-L) grade and individual radiographic features by a single bone and joint radiologist with high inter- and intra-rater reliability. Radiographic OA as described herein refers to OA defined by K-L grade of 2 or greater (14); severe radiographic OA to K-L grades 3 and 4; and symptomatic OA refers to a joint with both radiographic OA and symptoms usually defined by pain or pain, aching, or stiffness.

RESULTS

African American men and women displayed discrete differences, not previously described, in both radiographic knee and hip OA (13,15–17). In adjusted analyses, African American men were approximately 36% more likely than Caucasian men to have radiographic tibiofemoral knee OA (adjusted odds ratio, (95% confidence interval = 1.36 [1.00–1.86]), and twice as likely to have more severe disease (adjusted proportional odds ratio [aPOR] = 2.08 [1.19–3.65]). There was no significant racial difference in the prevalence of radiographic tibiofemoral knee OA in the women, but African American women were approximately 50% more likely than Caucasian women to have more severe involvement (aPOR = 1.56 [1.06–2.29]) (16). African American men were three times as likely as Caucasian men to have involvement of all three compartments of the knee, the medial, lateral and patellofemoral, whereas African American women were twice as likely as their Caucasian counterparts to have tri-compartmental involvement. African Americans also had greater prevalence and severity of osteophytes and joint space narrowing (JSN) and higher likelihood of sclerosis; some features were more likely in the lateral compartment (16). Differences were not limited to radiographic OA, but African Americans were more likely to have more severe knee symptoms that were not explained by radiographic severity, but were related to obesity and depression (18).

Racial differences were also observed in the radiographic features of hip OA. African American women were more likely than Caucasians to have superior and medial JSN, moderate or severe axial JSN, medial or lateral osteophytes, and subchondral cysts. African American men were more likely than Caucasian men to have superior or medial JSN and lateral osteophytes, but less likely to have axial JSN (17). Some of these features are associated with risk of joint replacement, potentially pointing toward unmet need in this demographic group.

Significant racial differences were also noted in the joint distribution of OA, with African Americans much less likely to have hand OA or multi-joint involvement that included hand OA; instead, multiple large joints were more likely to be affected in this group. The Caucasians, on the other hand, were more likely to have hand OA, and various phenotypes that co-existed with hand OA, such as hand with hip OA, hand with knee OA, or hand with knee and hip OA (19).

Examination of systemic biomarkers of OA in the JoCo OA revealed racial and gender differences in cartilage oligomeric matrix protein, a non-collagenous matrix protein found in all joint tissues and elevated in OA. There were no racial differences in this marker in men, but African American women had higher levels than Caucasian women (20). Other biomarkers have been examined in this cohort but were beyond the scope of the presentation and this summary.

Multiple potential risk factors for OA have also been examined in this cohort. One particular focus has been on novel potentially modifiable environmental factors. Examination of selenium and metal exposures was prompted by recognition that Kashin-Beck disease, an endemic, early-onset chondro-osteoarthropathy, is found in areas of China and other areas in Asia with some of the lowest levels of selenium in the soil anywhere on the planet. Kashin-Beck disease is strongly associated with selenium deficiency and is diminishing in frequency with selenium supplementation (21). We wondered if OA might also be related to selenium and conducted the first study of toenail selenium levels and knee OA using data from the JoCo OA. We observed that compared to those in the third quartile, those with toenail selenium levels in the lowest quartile were more likely to have radiographic knee OA. Low selenium levels were also associated with worse pain and function.

We also examined whole blood lead (Pb) levels and OA. Before the 1970s, when limitations on its use in gasoline and soldering were instituted, Pb was ubiquitous. It is stored in bone with an exceedingly long residence and is not inert. We found that although Pb was not associated with the presence of knee OA, those with higher blood Pb levels were more likely to have more severe and bilateral knee OA than those with lower levels (22). Blood Pb was also associated not only with biomarkers of bone breakdown (urinary NTX-1), but also with type II collagen breakdown (urinary CTX-II) in women (23).

Finally, the JoCo OA derived the first estimates of lifetime risk of symptomatic knee and hip OA (24,25). The lifetime risk of symptomatic knee OA was approximately 45% in both African Americans and Caucasians, increasing to 57% in those with a history of knee injury and 60% in those who were obese (24). The lifetime risk of symptomatic hip OA was 25% with no differences by race, gender, educational level, hip injury, or BMI (25).

DISCUSSION

The JoCo OA is the only population-based, prospective study of OA in African Americans and Caucasians in the United States. It was the first prospective OA study to include African Americans and has provided heretofore unavailable data on the epidemiology of knee, hip, and multi-joint OA, joint symptoms and function, and the consequences of OA. Several long-held notions about OA have been challenged by results of this study. This was also the first study to derive population-based lifetime risks of symptomatic knee and hip OA.

First, African American men and women were both found to have more severe knee OA, were more likely to have tri-compartmental OA, and more likely to have lateral compartment involvement than Caucasians (16). It should be noted that the NHANES-I did not include the patellofemoral joint in the assessment of knee OA (12), making the data about tri-compartmental involvement from the JoCo OA cohort original. African Americans in the Osteoarthritis Initiative were more likely to have valgus thrust consistent with potential increased involvement of the lateral compartment (26). These findings have clinical relevance regarding potential preventive and therapeutic measures for knee OA. They have also prompted us and others to further investigation of joint shape and joint inter-relationships along the entire lower extremity kinetic chain from hip to knee to ankle and feet.

Data suggesting that hip OA prevalence was equivalent, if not increased, in African Americans (15) certainly were unexpected based on prior studies of African and Caribbean blacks (9,10). Some of the radiographic features more common in African Americans were associated with risk of joint replacement in other studies, potentially pointing toward unmet need in this demographic group (17). Additional studies by our group not discussed in the presentation have confirmed that African Americans may be less likely to develop hip OA, but equally or more likely for it to progress, thereby explaining the difference between cross-sectional prevalence figures and longitudinal incidence and progression data (27). Hip OA should not be overlooked in African Americans.

Caucasians, on the other hand, were more likely to have hand OA, and various phenotypes that co-existed with hand OA, such as hand with hip OA, hand with knee OA, or hand with hip and knee OA (19). This is highly clinically relevant since the concept of “generalized OA” typically includes hand OA with its common correlate of OA in multiple joints. The menopausal Caucasian woman is the individual in whom this classic phenotype is expected. By such a definition, African Americans would not be included, while having clear evidence of multiple joint involvement, just not including the hand.

Demographic differences in systemic biomarkers of joint metabolism are critical to the development of standards for the anticipated use of biomarkers to define OA at early, pre-radiographic stages and to select people at high risk of progression to participate in clinical trials of disease-modifying interventions. The JoCo OA, with its population-based selection that includes those with and without OA, and its detailed phenotypic characterization of multiple joints, is being used to help define biomarker levels for just such purposes.

Age and BMI are well-known risk factors for knee OA. The JoCo OA's calculation of lifetime risk of symptomatic knee OA emphasizes the importance of BMI in increasing the risk substantially in the setting of obesity. That lifetime risk of hip OA is similar in men and women and in African Americans and Caucasians and does not vary by obesity status, educational level, or hip injury, indicates the conundrum we have clinically in identifying modifiable risk factors for this difficult-to-treat disease.

Finally, the observations regarding novel risk factors of selenium and Pb levels prompt further investigation into the biology of cartilage and other joint tissues upon exposure to these agents. A tight therapeutic window for selenium makes plans to use this as a potential treatment premature and worthy of further study. Longitudinal analyses, now 25 years and counting, will continue to elucidate the natural history of this condition in ways that other studies of shorter duration cannot.

DISCUSSION

Schuster, New York: The data in African-Americans could be viewed in two different ways. One would be a genetic approach, although the more I see of that literature, race as a genetic disease is vanishing and race as a social construct is clearly on the emergence. If you looked at the osteoarthritis in the Chinese and you thought race was a genetic disease, you might think that that's a genetic disease in the Chinese. Yet you have an environmental culprit at hand there. In the African American population that you are looking at — take out genetics — what environmental or social etiologies can you imagine or hypothesize that might explain the differences?

Jordan, Chapel Hill: I agree with you that these observed differences must be understood in the context of differences in socio-economic status (SES). Our group has been very interested in the role of not only the SES of the person, but the SES of the community in which that person lives, and the effect upon OA outcomes. There clearly are associations between knee and hip OA and SES, especially symptomatic OA. There are also environmental factors that may be modifiable, and BMI is one critical one. We have looked at some dietary factors as well to see if some of those might be impacting OA. Our studies looking at lead and selenium exposure also suggest environmental factors that may be modifiable and may vary according to SES. Few studies have data on genetics in African Americans; we have been collaborating with the Osteoarthritis Initiative (OAI), a study of Caucasian and African American people at high risk of developing knee OA from four different sites in the US. There do seem to be differences in associations between single nucleotide polymorphisms and OA in Caucasians or reported in Asians, and the African-American community. We have also been particularly intrigued by anatomic differences in joint shape that may be relevant in these observed racial/ethnic differences. We have published data that show racial/ethnic differences in hip anatomy may influence not only the development of hip OA, but potentially also the development of knee osteoarthritis, showing how an issue in one joint of the lower extremity can affect the risk of OA in another joint.

Stevenson, Palo Alto: My question relates to obesity. Most people think about the relationship between obesity and this kind of problem as maybe weight related. But there are different kinds of fat, and they do different things in terms of signaling. Have you begun to look at the relationship of abdominal fat distributions and this problem as compared to other kinds of fat?

Jordan, Chapel Hill: We have not directly looked at that exactly. We have DEXA scan data on body composition and when we looked at this — and this was just in women at the time when we analyzed this — it looked as though adding fat mass to a model that was already taken BMI into account did not really make that much of a difference. We have data on waist and hip circumference, and we have not — but should — look at body fat distribution further to examine this. Thank you for this question.

Stevenson, Palo Alto: Given the magnitude of the obesity epidemic in children, I think this is actually something that should be looked at in much greater detail, because we know things that we do in childhood have a huge impact on health in general and this might be something that is very important.

Jordan, Chapel Hill: Actually looking at events from earlier in the life course was part of our thought process when we examined the effects of lead exposure. Before lead was limited in the environment in 1972, everybody was exposed to lead; all of us were as children. And lead deposits in the bones and just sits there forever basically. So we were curious to see that early life exposure was related, and as we discussed, it was related to OA severity. But it does get to that point of needing to understand the effects of exposures in earlier and earlier stages and trying to look backwards. One of the things that Dr. Leigh Callahan in our group is also doing in relation to SES is going back and collecting data about the SES of childhood and our participants' parents to see if that is a proxy for some of the exposures and other things happening during childhood that might affect OA. One of the things we would like to do in our next cohort recruitment is to recruit younger people into the study mainly because of the obesity issue as well as joint injuries happening to people at younger and younger ages. Because of both of those factors, we anticipate that people may be developing joint symptoms and OA at younger ages.

Alexander, Atlanta: There is a lot of evidence about the sameness of the multiple chronic diseases that we get at the late stage, clinical phenotypes which may be quite different from the earlier pathways which appear to follow anti-senescence aging pathways. They basically go together. Did you look at your population and assess co-morbidities and how other diseases of aging were progressing or not?

Jordan, Chapel Hill: We collect a lot of data on co-morbidities. In fact, the current grant that we have funded from CDC is to look specifically at how diabetes, heart disease, and obesity impact OA cumulatively. We will be looking not only at the severity of the OA in the setting of co-morbidities, and severity of symptoms, but also how these multiple chronic conditions can impact disability. We are even looking at mortality as an outcome, something not usually thought about as being affected by arthritis, but data are starting to emerge showing that the combination of some of these conditions with OA actually contribute to higher mortality.