Is Osteoarthritis a Systemic Disorder of Bone?

Transcription

1 ARTHRITIS & RHEUMATISM Vol. 50, No. 2, February 2004, pp DOI /art , American College of Rheumatology Is Osteoarthritis a Systemic Disorder of Bone? Juliet Rogers, Lee Shepstone, 1 and Paul Dieppe 2 Objective. To describe the osteologic findings associated with osteoarthritis (OA) of a variety of joints. Methods. We performed visual examination of 563 skeletons of which >80% of the skeleton was available, from an archaeologic site in England. The surfaces and margins of several different joints (shoulders, elbows, wrists, hips, hands, knees, and ankles) were studied for evidence of eburnation and osteophytes, respectively, and the entire skeleton was examined for evidence of generalized enthesophyte formation. Associations between changes in different joint sites and between enthesophyte formation and evidence of OA were sought. Results. Eburnation and osteophyte formation at the hand, hip, and knee were strongly associated with eburnation and osteophytes at other joint sites not commonly thought to be prone to OA, including the elbow and wrist. Only the ankle was rarely involved. There was also a strong relationship between both bone eburnation and osteophytes and generalized enthesophyte formation. These findings remained statistically significant after adjustment for the age, sex, and historical period of the skeletons. Conclusion. Our findings indicate that skeletal OA is more widespread in the body than is apparent from clinical studies and are consistent with other data suggesting that OA is a disease that is primarily dependent on systemic predisposition to a particular type of bone response to mechanical stress. Dr. Rogers work was supported by the Arthritis Research Campaign. Dr. Dieppe s work was supported by the Medical Research Council. The Barton-on-Humber excavation was funded by English Heritage. 1 Lee Shepstone, PhD: University of East Anglia, Norwich, UK; 2 Paul Dieppe, BSc, MD, FRCP, FFPHM: Medical Research Council Health Services Research Collaboration, University of Bristol, Bristol, UK. Dr. Rogers is deceased. Address correspondence and reprint requests to Lee Shepstone, PhD, School of Medicine, Health Policy and Practice, University of East Anglia, Norwich NR4 7TJ, UK. L.Shepstone@ uea.ac.uk. Submitted for publication August 9, 2002; accepted in revised form November 14, Osteoarthritis (OA) is a difficult condition to investigate. Although very common and a frequent cause of chronic pain and physical disability in a limited number of joint sites (principally the hand, knee, hip, and spine), its pathophysiology remains poorly described. OA is characterized by areas of focal loss of articular cartilage associated with changes in the subchondral and marginal bone of synovial joints, and variable degrees of synovitis and capsular thickening (1,2). When OA is severe, the bone involvement can be detected on plain radiographs, but there are no other simple tests for OA: no blood test, for example. Other imaging techniques, such as arthroscopy (chondroscopy), scintigraphy, or magnetic resonance imaging, can provide information on both bone and soft tissue changes (3), but they are so expensive or invasive as to be inappropriate for routine use. Therefore, radiography remains the standard means of diagnosing OA and of assessing its severity (3,4), despite the fact that radiographs can provide only a unidimensional shadowgram of the complex 3-dimensional structure of a joint, and little information about the non-bone aspects of OA pathophysiology. The fact that radiographic findings in the joints relate very poorly to pain and disability further limits the value of the main technique used to assess OA (5,6). OA was first differentiated from other forms of joint disease at the beginning of the 20th century, on the basis of the hypertrophic changes seen in bone (7) hence, the term osteo arthritis. It is entirely appropriate, then, that we should look to osteology (the study of bones) to provide further insights into this disease. A unique advantage offered by the study of skeletons is the potential to directly examine the surfaces of all joints from any angle, as well as all other parts of the skeleton. We report herein the findings from a study of OA based on a large collection of human skeletons excavated from a single site in northern England. We were particularly interested in exploring the hypothesis that generalized osteoarthritis (GOA) represents a specific subset of the disease with a limited distribution (8,9), 452

2 IS OA A DISORDER OF BONE? 453 Table 1. Demographic and boneformer characteristics of the 563 skeletons studied* Sex Period Boneformer Total Male Female Pre-1500 Post-1500 Negative Positive Age, years (63) 139 (56) 130 (73) 116 (62) 152 (63) 239 (71) 35 (35) (37) 109 (44) 49 (27) 71 (38) 88 (37) 96 (29) 64 (65) Unknown 129 Sex Male 286 (52) 131 (49) 152 (55) 194 (45) 92 (78) Female 264 (48) 138 (51) 122 (45) 238 (55) 26 (22) Unknown 13 Period Pre (50) 211 (48) 65 (55) Post (50) 226 (52) 54 (45) Unknown 7 Boneformer Negative 443 (79) Positive 120 (21) * Values are the number (%). and suggestions that OA of major joint sites such as the hand, hip, or knee may be associated with more generalized skeletal changes (10). MATERIALS AND METHODS Source of skeletons studied. The study material consisted of a collection of skeletons excavated from St. Peter s Church in Barton-on-Humber in northeastern England, described elsewhere (11). This collection contained some 3,000 skeletons buried between 900 and 1850 AD. From archaeologic evidence it was possible to separate skeletons likely to have been buried prior to 1500 AD from those with later burial dates. The work described here is based on a subset of 563 of the adult skeletons from the site, in which there was preservation of at least 80% of the skeleton, including at least 1 joint surface from the hips, knees, and hands. Assessment of demographic and joint characteristics. We have limited this report to include only the main findings in peripheral joints, ignoring spinal changes. The following joint sites were considered: the hip (comprising the surfaces of the femoral head and acetabulum), the knee (the surfaces of the patella, distal femur, and proximal tibia), the elbow (the surfaces of the distal humerus, proximal radius, and ulna), the hand (the surfaces of the phalanges, distal metacarpals, and first carpometacarpal joint), the shoulder (the surfaces of the glenoid fossa and humerus), the wrist (the distal surfaces of the radius and ulnar and articulating surfaces of the scaphoid and lunate), and the ankle (the surfaces of the trochlear of the talus and articulating surfaces of the tibia and fibula). Each joint site was reviewed separately for evidence of osteophytes and subchondral eburnation. Evidence on any joint surface (either left or right) was sufficient to consider the joint site affected. If there were no apparent signs of these features the joint was considered unaffected, provided at least half of the joint surfaces comprising the joint were available for inspection. If, due to postmortem damage, more than half of the joint surfaces were missing, the entire joint site was considered missing. The skeletons were divided into male and female groups based on standard methods (12). Assessment of age was also based on standard techniques using pelvic and dental examination (12), but because of the uncertainty of older age, we divided each of the collections (male and female) into 2 main age categories: 45 years or 45 years, as described elsewhere (13). A third, nonstandard demographic variable we considered was boneformer status. This has been discussed in detail elsewhere (13). Briefly, it is a categorization based on the presence and degree of enthesophytes. At a number of ligament and tendon insertion sites a score (0 3) was assigned based on the degree of enthesophytes present. An average grade of 0.3 defined an individual as a boneformer. The label is used to indicate those individuals who appear to have had a propensity to producing new bone in response to stresses on the skeleton. Statistical methods. Prevalence data are presented with exact 95% confidence intervals (95% CIs). The association between variables was quantified using the Mantel- Haenszel odds ratio (OR) with 95% CIs. Odds ratios were deemed statistically significantly different from unity if the 95% CI excluded 1. Stratification was used to adjust ORs by other factors. All analyses were carried out using SAS version 8.2. RESULTS Demographic details of the study sample (n 563) are shown in Table 1. Age at death could not be determined in 129 subjects; sex could not be determined in 13. A higher proportion of males than females were in the older age group. Subjects classified as boneformers were more frequently older and male.

3 454 ROGERS ET AL Table 2. Prevalence of eburnation and osteophytes* Joint Eburnation Osteophytes group n % positive 95% CI n % positive 95% CI Any Hip Knee Elbow Hand Shoulder Wrist Ankle * 95% CI 95% confidence interval. Table 2 presents data on the prevalence of osteophytes and eburnation at different joint sites. More than one-fifth of the sample had eburnation somewhere on the joints of interest. Osteophytes were found in more than two-fifths. Table 2 also shows the prevalence of these changes at different joint sites. Eburnation was most common in the small joints of the hand and least common in the ankle (just 1 individual had evidence of eburnation at this site). The elbow had a relatively high prevalence of eburnation (slightly higher than the prevalence in the knee), and the wrist was also frequently affected. Osteophytes were very common in the hand, but more so in the shoulder. Again, a high prevalence was seen at the elbow, and the only joint to be rarely affected was the ankle. Due to the low prevalence of changes in the ankle, this joint was not included in any further analysis. Mantel-Haenszel ORs were calculated to quantify the association of eburnation and osteophytes with each of the demographic variables in turn (age, sex, historical period, and boneformer status) while adjusting for the remaining 3 demographic variables. Age showed a clear relationship with both eburnation and osteophytes, both features being more prevalent in the age 45 group at all joint sites. For eburnation, adjusted ORs ranged from 2.95 (95% CI ) for the shoulder to 6.94 (95% CI ) for the wrist. For osteophytes, adjusted ORs ranged from 2.33 (95% CI ) for the hip to 5.17 (95% CI ) for the shoulder. After adjustment for age and boneformer status, there were no significant differences in prevalence of osteophytes between males and females, but females had a higher prevalence of eburnation at the knee than did males (5.8% versus 2.5%; adjusted OR 5.63 [95% CI ]). Historical period had no apparent effect on prevalence of either osteophytes or eburnation at any joint site other than the elbow, where both features were more common in the post-1500 group than the pre-1500 group (2.6% versus 6.0%; OR 2.70 [95% CI ]). Table 3 shows the relationships between boneformer status and both eburnation and osteophytes. There was a strong and consistent association between boneformer status and osteophytes at all joints; there was also a relationship between enthesophyte formation and eburnation at all joint sites, with those that were labeled boneformer positive having a consistently higher prevalence of eburnation than those labeled negative. Table 3. Prevalence of eburnation and osteophytes by boneformer status* Eburnation Osteophytes Joint group % negative % positive OR (95% CI) Adjusted OR (95% CI) % negative % positive OR (95% CI) Adjusted OR (95% CI) Any ( ) 2.65 ( ) ( ) 9.03 ( ) Hip ( ) 5.36 ( ) ( ) 6.99 ( ) Knee ( ) 4.12 ( ) ( ) 8.57 ( ) Elbow ( ) 0.92 ( ) ( ) 6.94 ( ) Hand ( ) 2.15 ( ) ( ) 4.04 ( ) Shoulder ( ) 2.32 ( ) ( ) 7.46 ( ) Wrist ( ) 4.17 ( ) ( ) 7.08 ( ) * Boneformer defined as an average enthesophyte grade of % CI 95% confidence interval. Odds ratios (ORs) were calculated by Mantel-Haenszel test. Adjusted ORs were stratified by age, sex, and period.

4 IS OA A DISORDER OF BONE? 455 Table 4. Associations between joint groups* Hip Knee Elbow Hand Shoulder Wrist Hip 2.50 ( ) 0.84 ( ) Knee 4.24 ( ) ( ) 4.21 ( ) 7.04 ( ) Elbow 3.68 ( ) 5.40 ( ) 2.88 ( ) 5.60 ( ) Hand 2.16 ( ) 1.77 ( ) 2.51 ( ) 1.82 ( ) 5.72 ( ) Shoulder 3.08 ( ) 3.82 ( ) 2.93 ( ) 2.21 ( ) 2.73 ( ) Wrist 2.92 ( ) 1.94 ( ) 2.38 ( ) 4.61 ( ) 3.66 ( ) * Values are the odds ratio (95% confidence interval). Values above the diagonal are for eburnation; values below the diagonal are for osteophytes. Odds ratio could not be calculated due to zero cell frequencies. For a number of joint sites (hip, knee, and wrist), there was a statistically significant association even after adjustment for other demographic variables. Table 4 presents associations between joint sites (Mantel-Haenszel ORs with stratification by age, sex, historical period, and boneformer status). Strong, statistically significant associations for both the presence of osteophytes and the presence of eburnation were found between various joint sites. Knee eburnation was strongly associated with eburnation at the elbow as well as the hand, hip, and wrist; hand eburnation was associated with eburnation at the wrist; hip eburnation was associated with eburnation at the hand. Similar, but less strong, associations were found for osteophyte formation at different joint sites. DISCUSSION Through detailed study of whole human skeletons from an archaeologic site in England, we have found that evidence of classic OA (eburnation on the articulating surfaces of synovial joints commonly affected by OA) is associated with widespread skeletal changes elsewhere, including widespread eburnation of bone and osteophyte formation, as well as excess formation of enthesophytes. Many of these skeletal changes would not be apparent from standard clinical or radiologic examinations. Osteology provides us with the unique opportunity to study the bones of an individual directly and examine each part of the bony surface of all joints unimpeded by the limitations of conventional radiography. Clinical studies have focused on a narrow range of joints, such as the knees, hips, and hands, dictated by the frequency of patient-reported symptoms. Freed from such restrictive investigation, osteology allows a new perspective on diseases that affect bone, including OA, providing data on bones and joints hitherto neglected (14,15). We investigated 2 types of bony change that are both easily identified in skeletons and associated with OA: osteophytes and eburnation (15). However, osteophytes alone may not indicate OA (16), and eburnation is generally regarded by osteologists as the gold standard (14,15). Most of our findings and associations are similar for both osteophytes and eburnation, but one striking discrepancy is apparent in the shoulder, which had a high prevalence of osteophytes but a relatively lower prevalence of eburnation. We believe the likely explanation for this is that there was a high prevalence of rotator cuff disease resulting in apparent osteophytosis. The advantages of using skeletal material to study the pathophysiology of OA do come at a cost. Typically, very little is known about the individual being studied, and such mundane details as age and sex can be difficult to ascertain. Indeed, in this study age at death could not be estimated for nearly one-quarter of the study sample. Further missing data arises with postmortem damage to bones or missing bones. While there is little reason to believe that there is any systematic pattern to the missing values, these can present difficulties in interpretation. For example, when any analysis is adjusted for age, this analysis pertains only to the subset of individuals for whom age was available. Any difference in the results between this subset and the full study sample may be due to genuine age effects or simply to the pattern of missing data. This problem is not unique to osteology; it bedevils the clinical epidemiologist as well, but the latter has a better chance of guarding against this in the design and execution stages of any study. A further difficulty, with respect to age at death, is the imprecision with which it can be estimated. In this study, age at death was dichotomized into 45 years and 45 years. This is clearly crude since the variation within each group will be great. Thus, although we have attempted to adjust for age in our analyses (as an obvious potential confounder), we cannot eliminate the possibility of associations be-

5 456 ROGERS ET AL tween joint sites being due to the confounding effect of older age. A further weakness of this study, but again one that can occur in many situations, is the possibility of observer bias. While the evaluation of each joint site was carried out by a single, experienced paleopathologist who was an expert in the field (JR), each joint was not evaluated with blindedness to the others within the same individual. Thus, for example, the presence of osteophytes at one joint site may well have led to a more thorough search for osteophytes elsewhere. Similarly, knowledge of an individual s age at death may have influenced the degree to which evidence of OA was sought. Conversely, evidence of OA may well have influenced the estimation of age at death. Nonetheless, this study has yielded rare and valuable data that provide new insights into the nature of OA, gathered from a large sample, and presents fresh questions about the condition. Two main issues arise: first regarding the high prevalence of changes characteristic of OA in joints, such as the elbow, that are not considered to be an important site for the disease today, and second regarding the association with widespread changes throughout the skeleton. Each issue is discussed below. Idiopathic OA is sometimes divided into 2 main categories: monarticular (large joint) OA or generalized OA (17), with the implicit belief that single-joint OA is more dependent on local injury and generalized OA on a systemic predisposition to the disease (17,18). The data presented here challenge these notions and question the description of GOA as mainly affecting hips, knees, and hands (8,9). Statistically significant associations were found for eburnation between pairs of joints (including the elbow and wrist) not considered within the traditional GOA subset. Of the joints studied, only the ankle appeared to be spared. Our findings are compatible with other pathologic evidence, notably that of Heine, who described widespread OA from a large pathologic study carried out in the first part of the last century (19); this suggests that our results are not due to changes in the distribution of OA over time. However, this pathologic distribution contrasts sharply with clinical descriptions of OA (20,21), except in relation to sparing of the ankle (22). Thus, pathologic GOA would appear to involve a broader set of joints than symptomatic GOA. We believe the most likely explanation for this difference is that some joints affected by the structural changes of OA remain asymptomatic. This concept is supported by other observations in OA, including the poor correlation between the degree of radiologic joint damage and the severity of symptoms in joints such as the knee (5,23) and the frequent observation that in people with OA, joints such as the elbow may have some loss of function (generally a mild flexion deformity) even though they remain free of pain (a finding that was commonly recorded in our OA500 series [21]). However, this still begs the question of why a joint like the elbow should remain asymptomatic in the setting of severe structural OA changes. We can only speculate that this might relate to aspects of the biomechanics or innervation of different joints. Investigation of this issue might enhance global understanding of the nature of OA. Our study revealed strong and statistically significant associations between boneformer status (widespread enthesophyte formation in the skeleton) and the presence of both eburnation and osteophytes. The latter relationship is of no surprise; the original report coining the term boneformer (13) noted the strong relationship between the degree of enthesophyte formation and osteophyte formation in a separate skeletal collection. The relationship with eburnation, however, was unexpected. Three possible hypotheses suggest themselves to explain the link between increased enthesophyte formation and the presence of eburnation, which is indicative of full-thickness cartilage loss. First, both may be the consequence of physical activity inducing stresses upon the body that cause local bone responses. This would imply that all OA is related to physical activity, and there is some epidemiologic evidence to support this (24,25). However, one aspect of our findings would provide evidence against this hypothesis: the large distance between joints affected by OA (eburnation) and the enthesophytes for example, skeletons with eburnation confined to joints of the upper limb often had florid enthesophyte formation in the spine and lower limbs. An alternative explanation lies in our interpretation of eburnation. We have considered this to be evidence of full-thickness cartilage loss and therefore defining of OA. However, this phenomenon could be an indicator not of OA per se but of a particular response to OA, in much the same way as osteophytes can be considered as such, i.e., eburnation could be indicative of a boneformer s increased bone reaction to cartilage loss, leading to the apparent association. A third possible explanation is that both enthesophyte development and cartilage loss/eburnation are the consequence of a generalized predisposition to a particular type of skeletal remodeling in response to mechanical stresses. This hypothesis would suggest that OA is part of a systemic disorder of bone. We believe

6 IS OA A DISORDER OF BONE? 457 this is consistent with other evidence that has indicated the importance of bone turnover in OA. A negative association between OA and osteoporosis has been found by many investigators and appears to affect the whole skeleton (26,27), and data suggest that there may be abnormal bone turnover both in the subchondral bone of affected joints (28 30) and throughout the skeleton (10,31). Hulth has suggested that OA may be a generalized disease of the mineralized layer of cartilage (30), and Burger and colleagues have noted the links between OA, obesity, age, and bone density, suggesting that there could be a common connection if OA is dependent on bone responsiveness to growth factors (31). In conclusion, we have shown that skeletal OA is more widespread than is suggested by clinical studies, and that it is associated with generalized enthesophyte formation. We believe these findings are consistent with the hypothesis that OA is a condition that is due primarily to a systemic predisposition to a certain type of bone response to mechanical stresses. ACKNOWLEDGMENT This article is dedicated to the memory of Dr Juliet Rogers, who was the inspiration behind this work and who carried out all of the skeletal examinations herself prior to her death at the end of REFERENCES 1. Bullough PG. Osteoarthritis and related disorders: pathology. In: Klippel JH, Dieppe PA, editors. Rheumatology. 2nd ed. Vol. 2. London: Mosby; p Collins DH. The pathology of articular and spinal diseases. London: Edward Arnold; Dieppe P, Peterfy C, Watt I. Osteoarthritis and related disorders: imaging. In: Klippel JH, Dieppe PA, editors. Rheumatology. 2nd ed. Vol. 2. London: Mosby; p Altman M, Hochberg M, Murphy WA, Wolfe F, Lequesne M. Atlas of individual radiographic features of osteoarthritis. Osteoarthritis Cartilage 1995;3 Suppl: Lawrence JS, Bremner JM, Bier F. Osteoarthrosis: prevalence in the population and relationship between symptoms and x-ray changes. Ann Rheum Dis 1966;25: Salaffi F, Cavalieri F, Nolli M, Ferraccioli G. Analysis of disability in knee osteoarthritis: relationship with age and psychological variables but not with radiographic score. J Rheumatol 1991;18: Goldthwaite JE. The treatment of disabled joints resulting from the so called rheumatoid diseases. Boston Med Surg J 1987;136: Kellgren JH, Moore R. Generalised osteoarthritis and Heberdens nodes. BMJ 1962;1: Cooper C, Egger P, Coggon D, Hart DJ, Masud T, Cicuttini F, et al. Generalised osteoarthritis in women: pattern of joint involvement and approaches to defintion for epidemiological studies. J Rheumatol 1996;23: Sowers M, Lachance L, Jamadar D, Hochberg MC, Hollis B, Crutchfeld M. The associations of bone mineral density and bone turnover markers with osteoarthritis of the hand and knee in preand perimenopausal women. Arthritis Rheum 1999;42: Barton-on-Humber Archaeological Report. Barton-on-Humbar Excavation. London: English Heritage; Workshop of European Anthropologists. Recommendations for age and sex diagnosis of skeletons. J Hum Evol 1980;9: Rogers J, Shepstone L, Dieppe P. Bone formers: osteophyte and enthesophyte formation are positively associated. Ann Rheum Dis 1997;56: Rogers J, Waldron T. A field guide to joint disease in archaeology. New York: Wiley; Rogers J, Waldron T, Dieppe P, Watt I. Arthropathies in paleopathology: the basis of classification according to the most probable cause. J Arch Sci 1987;14: Danielsson L, Hernborg J. Clinical and roentgenologic study of knee joints with osteophytes. Clin Orthop 1970;69: Flores RH, Hochberg MC. Definition and classification of osteoarthritis. In: Brandt K, Doherty M, Lohmander S, editors. Osteoarthritis. Oxford: Oxford University Press; p Felson DT, Lawrence R, Dieppe P, Hirsch R, Helmick C, Jordan J, et al. Osteoarthritis: new insights. 1. The disease and its risk factors. Ann Intern Med 2000;133: Heine J. Uber die arthritis deformans. Virchows Archiv 1926;260: Peterson IF. Occurrence of osteoarthritis in the peripheral joints in European populations. Ann Rheum Dis 1996;55: Cushnaghan J, Dieppe P. Study of 500 patients with limb joint osteoarthritis. 1. Analysis by age, sex and distribution of symptomatic joint sites. Ann Rheum Dis 1991;50: Huch K, Kuettner KE, Dieppe P. Osteoarthritis in ankle and knee joints. Semin Arthritis Rheum 1997;26: Creamer P, Lethbridge-Cejku M, Hochberg M. Determinants of pain severity in knee osteoarthritis. J Rheumatol 1999;26: Vingard E, Alfredsson L, Goldie I, Hogstedt C. Occupation and osteoarthrosis of the hip and knee: a register-based cohort study. Int J Epidemiol 1991;20: Hadler NM, Gillings DB, Imbus HR, Levitin PM, Makuc D, Utsinger PD, et al. Hand structure and function in an industrial setting: influence of three patterns of sterotyped, repetitive usage. Arthritis Rheum 1978;21: Hart DJ, Mootoosamy I, Doyle DV, Spector TD. The relationship between osteoarthritis and osteoporosis in the general population. Ann Rheum Dis 1994;53: Sambrook P, Naganathan V. What is the relationship between osteoarthritis and osteoporosis? Baillieres Clin Rheumatol 1997; 11: Mansell JP, Bailey AJ Abnormal cancellous bone metabolism in osteoarthritis. J Clin Invest 1998;101: Dieppe P, Cushnaghan J, Young P, Kirwan J. Prediction of the progression of joint space narrowing in osteoarthritis of the knee by bone scintigraphy. Ann Rheum Dis 1993;52: Hulth A. Does osteoarthritis depend on growth of the mineralised layer of cartilage? Clin Orthop 1993;287: Burger H, van Daele PLA, Odding E, Valkenberg HA, Hofman A, Grobbee DE, et al. Association of radiographically evident osteoarthritis with higher bone mineral density and increased bone loss with age: the Rotterdam study. Arthritis Rheum 1996;39:81 6.