Nonspherical Femoral Head Shape (Pistol Grip Deformity), Neck Shaft Angle, and Risk of Hip Osteoarthritis

Transcription

1 ARTHRITIS & RHEUMATISM Vol. 58, No. 10, October 2008, pp DOI /art , American College of Rheumatology Nonspherical Femoral Head Shape (Pistol Grip Deformity), Neck Shaft Angle, and Risk of Hip Osteoarthritis A Case Control Study Michael Doherty, 1 Philip Courtney, 2 Sally Doherty, 1 Wendy Jenkins, 1 Rose A. Maciewicz, 3 Kenneth Muir, 1 and Weiya Zhang 1 Supported by the John Monk Hip Foundation. Infrastructure support for Academic Rheumatology, University of Nottingham, was provided by the Arthritis Research Campaign, UK (ICAC grant 14851). Support for the initial assembly of the Genetics of Osteoarthritis and Lifestyle (GOAL) Study was provided by AstraZeneca, Macclesfield, UK. 1 Michael Doherty, MA, MD, FRCP, Sally Doherty, SRN, Wendy Jenkins, RGN, RM ONC, Kenneth Muir, BSc, PhD, FRSS, Weiya Zhang, PhD: University of Nottingham, Nottingham, UK; 2 Philip Courtney, MD: Nottingham University Hospitals Trust, Nottingham, UK; 3 Rose A. Maciewicz, BA, MA, PhD: AstraZeneca, Macclesfield, UK. Dr. Courtney has received an honorarium from Merck, Sharp, and Dohme (less than $10,000). Dr. Maciewicz owns stock or stock options in AstraZeneca and has submitted patent applications for nonpharmalogical biologic products. Dr. Zhang has received consulting fees, speaking fees, and/or honoraria from Rottapharm and the LMS Group (less than $10,000 each). Address correspondence and reprint requests to Michael Doherty, MA, MD, FRCP, Academic Rheumatology, Clinical Sciences Building, City Hospital, Nottingham NG5 1PB, UK. Michael.Doherty@nottingham.ac.uk. Submitted for publication February 19, 2008; accepted in revised form June 27, Objective. To determine whether 2-dimensional measures of femoral head shape and angle are associated with hip osteoarthritis (OA). Methods. We compared cases with symptomatic radiographic hip OA with asymptomatic controls with no radiographic hip OA. On anteroposterior pelvis radiographs, we measured pistol grip deformity for each hip (visually categorized as nonspherical, indeterminate, or spherical), the femoral head to femoral neck ratio as an interval measure of femoral head shape, and the femoral neck shaft angle. The relative risk of hip OA associated with each feature was estimated using odds ratios (ORs) and 95% confidence intervals (95% CIs), adjusted for possible confounders using a logistic regression model. Results. Of 1,007 cases, 965 had definite radiographic hip OA; of 1,123 controls, 1,111 had no radiographic OA. The prevalence of pistol grip deformity in at least 1 hip was 3.61% in controls and 17.71% in cases (OR 6.95 [95% CI ]), and the prevalence of abnormal femoral head to femoral neck ratio in at least 1 hip was 3.70% in controls and 24.27% in cases (OR [95% CI ]). The risk of hip OA increased as the femoral head to femoral neck ratio decreased (P for trend < 0.001) and with each extreme of neck shaft angle (P < 0.05). In cases with unilateral hip OA, the prevalence of abnormal femoral head to femoral neck ratio in the unaffected hip was 2 times greater than that in controls (OR 1.82 [95% CI ]); in contrast, an abnormally low, but not abnormally high, neck shaft angle was more common in unaffected hips than in controls (OR 1.79 [95% CI ]). Conclusion. Our findings indicate that pistol grip deformity is associated with hip OA. The increased prevalence of pistol grip deformity and an abnormally low neck shaft angle in unaffected hips of cases with unilateral OA suggests that they are risk factors for development of hip OA. However, both a nonspherical head shape and an increase in neck shaft angle may occur as a consequence of OA. Osteoarthritis (OA) of the hip is an important cause of pain and disability in the community (1,2). As with OA at other sites, hip OA is considered a common complex disorder with multiple genetic, constitutional, and environmental risk factors that result in heterogeneity of phenotype and variability in clinical outcome (3,4). Hip OA shows marked familial predisposition, with heritability estimates of 0.6 in women (5) and up to a 5-fold increased risk in siblings of patients undergoing 3172

2 NONSPHERICAL FEMORAL HEAD SHAPE AND RISK OF HIP OA 3173 Figure 1. Anteroposterior radiographs of the pelvis, showing a spherical femoral head shape (left) and a nonspherical pistol grip deformity (right). The measured femoral head to femoral neck ratios for these 2 examples were 1.47 (spherical head) and 1.25 (nonspherical head). surgery for hip OA (6). The mechanism of this heritable risk is unknown. Other recognized risk factors include severe developmental hip dysplasia (7 9), childhood disorders such as Legg-Calvé-Perthes disease and slipped femoral epiphysis (7,10), occupational activity such as farming (11 13), high-impact recreational activity (13,14), high bone density (4,13), obesity (4,13), and aging (4,13). Apart from severe acetabular dysplasia and overt femoral head disease presenting in childhood, it is possible that more minor acetabular dysplasia or subtle variation in proximal femur morphology might compromise the joint biomechanically and lead to OA (7 9,15). The term pistol grip (16) or tilt (17) deformity refers to a characteristic shape of the proximal femur, with loss of normal concavity of the anterosuperior region of the head neck junction, resulting in a nonspherical head (Figure 1). In a large study of human skeletal remains, Goodman et al (10) found an 8% prevalence of pistol grip deformity and a higher prevalence of severe OA in the pistol grip deformity group compared with the non pistol grip deformity group (38% versus 26%). They attributed this shape to mild subclinical slipped femoral epiphysis and proposed it as a risk factor for OA. Resnick (18), however, noted this deformity in 48 of 100 patients undergoing surgery for hip OA and concluded that it was mainly secondary to bone remodeling and osteophyte formation rather than a sign of preceding epiphysiolysis. In a recent magnetic resonance imaging (MRI) study, Ito et al (19) demonstrated that a nonspherical femoral head could result in damage to the acetabular labrum by either the head pressing against the labrum and adjacent cartilage ( cam effect) or the femoral neck abutting against the labrum ( pincer effect). Such femoroacetabular impingement is associated with histologic changes of OA in femoral head cartilage (20). Therefore, a nonspherical head might be a local risk factor for hip OA acting via repetitive mechanical impingement (19 21). Such studies of nonspherical femoral heads and impingement, however, have included relatively few subjects, have focused mainly on young men, often involve 3-dimensional (3-D) assessments (using cadavers or MRI), and have not adjusted for other risk factors for hip OA. The angle between the femoral shaft and the femoral neck might also compound impingement, influence the orientation between the femoral head and the weight-bearing surface of the acetabulum (acetabular sourcil), and mechanically predispose to OA (22 25). However, to our knowledge, this morphologic variant has not been formally investigated in a large case control study. We wished to examine morphologic variables of the proximal femur that can be assessed using standard hip radiographs. We therefore undertook this case control study in an existing large cohort of wellcharacterized individuals with available hip radiographs

3 3174 DOHERTY ET AL to determine 1) the prevalence and symmetry of pistol grip deformity and abnormal (increased or decreased) femoral neck shaft angle and whether these differ between men and women, and 2) whether pistol grip deformity or neck shaft angle is associated with hip OA. Rather than just dichotomize as the presence or absence of pistol grip deformity, we developed a simple measure (the femoral head to femoral shaft ratio) to capture the change from spherical to nonspherical shape. Because these morphologic features could be secondary to the OA process itself, we determined risk of OA by focusing on the unaffected (normal, non-oa) hip of subjects with unilateral hip OA. SUBJECTS AND METHODS Cases and controls. Subjects were men and women originally recruited in Nottingham between 2002 and 2006 to participate in a case control study (the Genetics of Osteoarthritis and Lifestyle [GOAL] study) designed primarily to investigate gene environment interaction in knee and hip OA. Approval for assembly of this cohort for the purposes of research into OA was obtained from the Nottingham Research Ethics Committee. Hip OA cases were recruited from hospital orthopedic surgery lists (current and from the previous 5 years) and from a rheumatology OA clinic. All cases had been referred to the hospital with symptomatic, clinically severe hip OA, and the majority had undergone unilateral or bilateral total hip replacement (THR) within the previous 5 years. Subjects were excluded from the study if they had other major arthropathy (e.g., rheumatoid arthritis or ankylosing spondylitis), Paget s disease of bone affecting the pelvis or femur, overt childhood hip disease (e.g., Legg-Calvé-Perthes, slipped femoral epiphysis, or severe acetabular dysplasia), THR due to trauma or avascular necrosis of the femoral head, or a terminal illness. Control subjects were recruited from lists of people who had been in the hospital for intravenous urography within the last 5 years. Individuals who had no radiographic evidence of hip OA on their screening intravenous urography radiograph and who had no hip or knee symptoms and none of the exclusions listed above for cases were invited to take part in the study. Cases and controls were further characterized by interview and examination. Height and weight were measured to calculate body mass index (BMI). The presence of interphalangeal nodes was determined clinically by established methods (26,27), and a nodal phenotype was defined as Heberden s and/or Bouchard s nodes that affected at least 2 rays of each hand. Calcaneal bone density (single heel determined by hand dominance) was measured using the Norland Apollo 501A00Z, and the age-adjusted Z score was used for bone mineral density (BMD) in the analysis. The index finger to ring finger type was visually classified as type 1 (index finger longer than ring finger), type 2 (index finger equal to ring finger in length), and type 3 (ring finger longer than index finger), by examination of hand radiographs (28). History of hip injury was defined as any self-reported significant hip injury (fracture, severe trauma requiring medical attention, immobilization, or crutches for 2 weeks) due to occupation, sport, or leisure activities during the subject s lifetime. Regular exercise was defined as energetic activity performed for 20 minutes at a time and 3 times per week that caused sweating, breathlessness, or increased pulse rate in subjects of any age who participated in such activity between the ages of 20 and 49 years (28). Radiographic assessment of hips. For cases who had undergone THR, the presurgical radiograph was copied using Hipax Dicom digitizing software (Hipax, Vorsetten, Germany) to enable scoring. In cases who had not undergone THR and in all controls, new radiographs were obtained unless the subject had undergone radiography of the pelvis (using the same standardized method) not more than 2 years previously. Anteroposterior views of the pelvis were obtained using a standardized protocol with the participant supine and feet internally rotated 10 (70 kv exposure, broad focus, 100-cm focus-to-film distance). Each film was scanned and written to CD using the Hipax Dicom 4.2 x-ray image processor. CD images were read using the Hipax Private Health disc image viewer, which enables straight-line measurements between visually determined points to an accuracy of 0.01 mm (automatically taking into account any magnification utilized in viewing) and measurement of angles to an accuracy of 1 (manufacturer precision). Prior to the main study, we devised a simple, reproducible measure of pistol grip deformity on plain radiographs of the pelvis. We determined that pistol grip deformity is associated with a high femoral head to femoral neck ratio, and that the ratio of maximal femoral head diameter divided by the minimum parallel femoral neck diameter successfully differentiates hips with the deformity from those without. In practice, this measure requires identification of the equatorial center of the femoral head, using concentric circles on the Perspex template of the Lequesne arthrometer (29), and a point in the center of the femoral neck, measured to be equidistant from the superior and inferior borders (Figure 2). A line is then drawn between these 2 points. The maximum femoral head diameter at right angles to this line is identified, using the arthrometer, and then measured electronically; similarly, the minimum femoral neck diameter, again at right angles to the line, is determined with the aid of the arthrometer and then measured electronically. The ratio is the head diameter divided by the neck diameter. This measure was validated in a test series of 22 pelvis radiographs (44 hips) comprising 24 hips with definite nonspherical heads and 20 with definite spherical heads as judged by visual assessment alone. A single observer (PC), who was blinded with regard to the visual assessment, scored both hips in all radiographs twice in random order. This measure showed good intrarater reliability. The median ratio for normal spherical hips (1.4) was significantly higher than that for hips with pistol grip deformity (1.1) (P 0.01). For the main study, a single trained observer (WJ) measured the femoral head to femoral neck ratio in both hips in all pelvis radiographs. The same observer also measured the neck shaft angle (Figure 2) and the center edge angle (30) in all hips. The neck shaft angle was determined by drawing a line between at least 2 points, measured to be in the middle of the femoral shaft, below the lesser trochanter (the lowest point being as low as the radiograph permitted). The angle between this line and the initial

4 NONSPHERICAL FEMORAL HEAD SHAPE AND RISK OF HIP OA 3175 Figure 2. Measurement of the femoral head to femoral neck ratio and the neck shaft angle (NSA). A line (L) was drawn between the equatorial center of the head (1) and the midpoint of the neck (2). The maximum width of the head (H) and the minimum width of the neck (N) were both measured at a 90 angle to the line. The neck shaft angle is the angle between the original line and a second line (M) drawn between 2 measured midshaft points below the lesser trochanter (3 and 4). line drawn for the femoral head to femoral neck ratio was then measured electronically. A second observer (SD) scored all pelvis radiographs using the Kellgren/Lawrence (K/L) scale for hip OA (descriptionbased scale of 0 4) (31) and the Croft grade of hip OA (scale of 0 5) (32). The same observer determined the minimum joint space width (JSW; in mm) and the site of maximum narrowing (superolateral, superointermediate, superomedial, superior indeterminate, axial, medial, concentric, or unable to determine). The degree of narrowing was scored on a scale of 0 3, and osteophyte formation in the acetabulum, femoral head, and femoral neck was scored on a scale of 0 3 for each assessment, using a radiographic atlas (33). The presence and location (in the acetabulum or femoral head) of subchondral cysts and sclerosis were determined. The same observer also visually classified each hip qualitatively as normal/spherical (code 1), indeterminate (code 2), or having a definite pistol grip deformity (code 3). Hip OA was ascertained radiographically when JSW was 2.5 mm (32). Subjects in the case group with JSW 2.5 mm in either hip were classified as having hip OA. Subjects in the control group with JSW 2.5 mm in both hips were retained as controls. Cases were also divided into 2 groups, those with bilateral hip OA (JSW 2.5 mm in both hips) and those with unilateral hip OA (JSW 2.5 mm in only 1 hip). Measurement of reproducibility. Reproducibility of all radiograph scores and measurements was determined using a sample of 20 films specifically chosen to include normal hips as well as a broad spectrum of severity of radiographic features. This was assessed approximately halfway through the scoring period. This time point was selected to capture the average reproducibility during the whole period, avoiding the early phase (when observers were possibly the least well trained and had the least experience) and late phase (when observers were possibly the best trained and had the most experience). Each of the 2 observers read the selected films twice, with a 2-week interval between readings. The observers were blinded with regard to subjects identity and to initial scores. Statistical analysis. Reproducibility was assessed using the kappa statistic (for dichotomous data), the weighted kappa statistic (for categorical readings) and the one-way random intraclass correlation coefficient (ICC) (for continuous data) as appropriate. Symmetry of the morphologic markers between right and left hips was examined in the control population using McNemar s test (for discontinuous data) and the paired t-test (for continuous data). Abnormal femoral head to femoral neck ratio and neck shaft angle were defined based on the distribution of these makers in the control population. Under the assumption of the normal distribution, the mean 1.96 SD was used as a threshold for the femoral head to femoral neck ratio. Similarly, the mean 1.96 SD and the mean 1.96 SD were used as thresholds for determining an abnormal neck shaft angle (either very high or very low). The prevalence of visual pistol grip deformity and the abnormality of the femoral head to femoral neck ratio and the neck shaft angle were calculated and compared between subjects on the basis of at least 1 hip affected. We also calculated the prevalence of these abnormalities in both the affected and unaffected hips of cases with unilateral hip OA and compared them with those of controls with no radiographic hip OA. The chi-square test was used to compare prevalence between subjects of different ages and sex. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were used to estimate the relative risk in cases with hip OA versus controls. We also categorized the femoral head to femoral neck ratio and the neck shaft angle into tertiles and calculated graded risks associated with hip OA; for this analysis, the worst affected hip (i.e. the hip with the smaller minimum JSW) was used. The independence of each measure was examined using a logistic regression model, adjusted for other morphologic markers, age, sex, BMI (kg/m 2 ), BMD (Z score), type 3 index finger to ring finger ratio (i.e., index finger shorter than ring finger), physical exercise, history of hip injury, hand nodes, and center edge angle. The relationship between the visual classification and the measured femoral head to femoral neck ratio was examined using the analysis of variance test for the trend of linearity. The agreement between visual and measured pistol grip deformity was examined using the kappa statistic. P values less than 0.05 (2-tailed) were considered significant. All statistical analyses were performed using SPSS software, version 14 (SPSS, Chicago, IL).

5 3176 DOHERTY ET AL Table 1. Characteristics of the cases with hip OA and controls* Cases (n 965) Controls (n 1,111) P Age, mean SD years Women JSW 2.5 mm NA JSW 1.5 mm Croft grade K/L grade BMI, mean SD kg/m Calcaneal BMD, mean SD Z score Hand nodes Type 3 hand History of hip injury Regular exercise Visual pistol grip deformity Center edge angle * Except where indicated otherwise, values are the percentage of subjects. Cases were defined as having hip osteoarthritis (OA) if joint space width (JSW) was 2.5 mm. NA not applicable; K/L Kellgren/Lawrence; BMI body mass index; BMD bone mineral density. Heberden s and/or Bouchard s nodes affecting at least 2 rays of each hand. Index finger shorter than ring finger. Self-reported significant injury (fracture, severe trauma requiring medical attention, immobilization, or use of crutches for 2 weeks). Activity performed for 20 minutes at a time and 3 times per week between the ages of 20 and 49 that caused sweating, breathlessness, or increased pulse rate. RESULTS Characteristics of the study subjects. Of the 2,130 participants studied, 1,007 had clinically severe hip OA, and 1,123 were controls who had undergone intravenous urography and had apparently normal screening hip radiographs. After formal radiographic scoring, 965 cases were ascertained to have radiographic evidence of hip OA (defined as JSW 2.5 mm) and 1,111 controls had no such evidence of OA. The classification of cases was supported by other radiographic criteria, such as JSW 1.5 mm, overall Croft score 3 (i.e., 2 or more of the following were present: osteophytes, narrowing, subchondral sclerosis, or cysts) and K/L score 2 (Table 1). The hip OA group was significantly older, had significantly higher mean BMI and calcaneal BMD, a smaller mean center edge angle, and a higher prevalence of hand nodes and type 3 hands, and included more subjects with a history of hip injury and more subjects taking part in regular exercise (Table 1). Reproducibility. ICC scores were 0.88 for visual pistol grip deformity, 0.84 for femoral head to femoral neck ratio, and 0.76 for neck shaft angle. ICC (or kappa) scores were 0.98 for JSW, 0.94 for Croft grade, 0.89 for K/L scale, 0.93 for JSN, 0.84 for the presence of acetabular osteophytes, 0.89 for the presence of osteophytes in the femoral head, and 0.60 for the presence of osteophytes in the femoral neck. All agreements were statistically significant (P 0.05). Symmetry of pistol grip deformity, femoral head to femoral neck ratio, and neck shaft angle in controls. Visually determined pistol grip deformity was a bilateral feature in 63% of affected subjects, and was modestly more frequent in the right side, with a rightto-left ratio of 1.2:1. The test for asymmetry was not statistically significant (P 0.30) (Table 2). The femoral head to femoral neck ratio was strikingly symmetrical between right and left hips within individuals (Table 2). The mean SD paired difference was , which was not significantly different from 0 (P 0.08). Similarly, the neck shaft angle showed marked symmetry between right and left hips within individuals. The mean SD paired difference was , which was not significantly different from 0 (P 0.993) (Table 2). Based on the symmetry of these 2 measures, the mean value between right and left was used to calculate the threshold for abnormality. This yielded a cutoff value of 1.27 (mean 1.96 SD) for abnormality of the femoral head to femoral neck ratio (measured pistol grip deformity) and a cutoff value of (mean 1.96 SD) or (mean 1.96 SD) for abnormality of the neck shaft angle. Prevalence and distribution of pistol grip deformity, femoral head to femoral neck ratio, and abnormal neck shaft angle in controls. The prevalence of visual pistol grip deformity in at least 1 hip was 3.61% in the control population (Table 3). It was more prevalent in men (6.37%) and uncommon in women (0.39%) (P 0.001). Although pistol grip deformity was reduced with Table 2. Symmetry of morphologic makers in the right and left hips in control subjects* Right hip Left hip Paired difference P Visual pistol grip deformity, no. (%) (n 1,109) 35 (3.2) 30 (2.7) 15 (1.4) Femoral head to femoral neck ratio (n 1,109) Neck shaft angle (n 1,103) * Except where indicated otherwise, values are the mean SD. Asymmetry was examined using McNemar s test for pistol grip deformity and the paired t-test for femoral head to femoral neck ratio and neck shaft angle.

6 NONSPHERICAL FEMORAL HEAD SHAPE AND RISK OF HIP OA 3177 Table 3. Prevalence of femoral head abnormalities and estimated relative risk of hip OA* Prevalence, no. positive/no. analyzed (%) OR (95% CI) Hip OA Control Crude Adjusted Visual pistol grip deformity Men 128/423 (30.26) 38/597 (6.37) 6.38 ( ) 6.62 ( ) Women 19/407 (4.67) 2/512 (0.39) ( ) 7.83 ( ) All 147/830 (17.71) 40/1109 (3.61) 5.75 ( ) 6.95 ( ) Femoral head to femoral neck ratio 1.27 Men 176/461 (38.18) 32/597 (5.36) ( ) ( ) Women 41/458 (8.95) 1/512 (0.20) ( ) ( ) All 223/919 (24.27) 33/1109 (3.70) ( ) ( ) Neck shaft angle Men 30/467 (6.42) 23/592 (3.89) 1.70 ( ) 1.30 ( ) Women 73/464 (15.71) 31/511 (6.07) 2.89 ( ) 3.25 ( ) All 103/931 (11.06) 54/1103 (4.90) 2.42 ( ) 2.24 ( ) Neck shaft angle Men 29/464 (6.25) 19/592 (3.21) 2.01 ( ) 2.31 ( ) Women 18/465 (3.87) 13/511 (2.54) 1.54 ( ) 1.04 ( ) All 47/929 (5.06) 32/1103 (2.90) 1.78 ( ) 1.75 ( ) * The threshold for normal femoral head to femoral neck ratio was the mean 1.96 SD ratio in the control population. The range for normal neck shaft angle was the mean 1.96 SD value in the control population to the mean 1.96 SD value in the control population. The prevalence of each feature was calculated based on the presence of that feature in at least 1 hip. See Table 1 for other definitions. A logistic regression model was used to calculate the odds ratio (OR) (95% confidence interval [95% CI]), adjusted for age, sex (when both men and women were included in the analysis), BMI, BMD, physical activity, history of hip injury, type 3 hand (index finger shorter than ring finger), hand nodes, and center edge angle. Visual pistol grip deformity (or femoral head to femoral neck ratio) and neck shaft angle were included in the same model for independent effect. A stepwise forward calculation was used to select significant variables. age in men, the trend was not statistically significant (P 0.116) (Figure 3). Similarly, the prevalence of abnormal femoral head to femoral neck ratio ( 1.27) was 3.70% in the control population. It was much more common in men than in women (5.36% versus 0.20%; P 0.001) and was not age dependent (P 0.40 in men and in women) (data not shown). The prevalence of an abnormally high neck shaft angle ( ) was 4.90%, and the prevalence of an abnormally low neck shaft angle ( ) was 2.90%. The neck shaft angle showed no clear relationship to sex or age (data not shown). Pistol grip deformity, femoral head to femoral neck ratio, and neck shaft angle and relative risk of hip OA. The prevalence of visual pistol grip deformity in at least 1 hip was significantly greater in subjects with hip OA (17.71%) compared with controls (3.61%), giving an Figure 3. Prevalence of visual pistol grip deformity (PGD) in men (diamonds) and women (squares) in the control population. P for trend in men and in women. Figure 4. Femoral head to femoral neck (FH:FN) ratio and relative risk of hip osteoarthritis (OA). Values are the adjusted odds ratio (OR) (95% confidence interval [95% CI]) for development of hip OA in each femoral head to femoral neck ratio tertile. Tertile 3 was used as the reference, so the greater the femoral head to femoral neck ratio the greater the risk of hip OA.

7 3178 DOHERTY ET AL Table 4. Prevalence of femoral head abnormalities in controls and in the unaffected and affected hips of patients with unilateral hip OA* Controls Unaffected hips Affected hips Visual pistol grip deformity Men 38/597 (6.37) 46/296 (15.54) 59/295 (20.00) Women 2/512 (0.39) 1/271 (0.37) 5/271 (1.85) All 40/1109 (3.61) 47/567 (8.29) 64/566 (11.31) Adjusted OR (95% CI) ( ) 4.00 ( ) Femoral head to femoral neck ratio 1.27 Men 32/597 (5.36) 28/296 (9.46) 81/288 (28.13) Women 1/512 (0.20) 3/271 (1.11) 19/263 (7.22) All 33/1109 (2.98) 31/567 (5.47) 100/551 (18.15) Adjusted OR (95% CI) ( ) 7.98 ( ) Neck shaft angle Men 23/592 (3.89) 6/296 (2.03) 13/291 (4.47) Women 31/511 (6.07) 18/270 (6.67) 29/268 (10.82) All 54/1103 (4.90) 24/566 (4.24) 42/559 (7.51) Adjusted OR (95% CI) ( ) 1.71 ( ) Neck shaft angle Men 19/593 (3.20) 20/296 (6.76) 6/291 (2.06) Women 13/511 (2.54) 9/270 (3.33) 2/268 (0.75) All 32/1104 (2.90) 29/566 (5.12) 8/559 (1.43) Adjusted OR (95% CI) ( ) 0.45 ( ) * Except where indicated otherwise, values are the number of positive subjects/number of subjects analyzed (%). The threshold for normal femoral head to femoral neck ratio was the mean 1.96 SD ratio in the control population. The range for normal neck shaft angle was the mean 1.96 SD value in the control population to the mean 1.96 SD value in the control population. The prevalence of each feature was calculated based on the presence of that feature in at least 1 hip. Compared with controls, the adjusted odds ratio (OR) (95% confidence interval [95% CI]) was calculated using a logistic regression model, adjusted for age, sex (when both men and women were included in the analysis), BMI, BMD, physical activity, history of hip injury, type 3 hand (index finger shorter than ring finger), hand nodes, and center edge angle. Visual pistol grip deformity (or femoral head to femoral neck ratio) and neck shaft angle were included in the same model for independent effect. A stepwise forward calculation was used to select significant variables. See Table 1 for other definitions. estimated relative risk of 7 (OR 6.95 [95% CI ]) for the association of pistol grip deformity with hip OA (Table 3). This result was supported by the measured pistol grip deformity (i.e., an abnormal femoral head to femoral neck ratio of 1.27) (Table 3). Furthermore, the risk increased as the femoral head to femoral neck ratio decreased (P for trend 0.001) (Figure 4). These associations were independent of other risk factors for hip OA, such as age, sex, BMI, BMD, physical activity, history of hip injury, type 3 hands, hand nodes, and center edge angle. In addition, we found that an abnormal neck shaft angle ( or ) was associated with greater risk of hip OA (Table 3). To determine whether these individual morphologic features were likely to be constitutional risk factors for developing hip OA or were secondary to OA bone remodeling, we examined individuals with unilateral hip OA (Table 4). The prevalence of visual pistol grip deformity increased from 3.61% in the control population to 8.29% in the unaffected hips of the subjects with unilateral hip OA; it further increased to 11.31% in the affected hips of these subjects. The relative risk was 3 times greater in the unaffected hips (OR 2.72 [95% CI ]) and 4 times greater in the affected hips (OR 4.00 [95% CI ]) compared with controls (Table 4). A similar increased risk in unaffected hips and even greater risk in affected hips was seen for the femoral head to femoral neck ratio (Table 4). For the presence of an abnormally high neck shaft angle, there was no difference between unaffected hips and controls (OR 0.92 [95% CI ]), although in affected hips there was a modest but statistically significant association (OR 1.71 [95% CI ]) (Table 4). An abnormally low neck shaft angle was more frequent in the unaffected hips of OA patients than in controls (OR 1.79 [95% CI ]), whereas in affected hips there was no significant difference from controls (Table 4). We divided patients with unilateral hip OA into 2 groups according to the presence (n 39) or absence (n 528) of pistol grip deformity in the unaffected hip, and examined the pattern of femoral head migration in

8 NONSPHERICAL FEMORAL HEAD SHAPE AND RISK OF HIP OA 3179 their affected hips. In both subgroups, the site of maximum narrowing was predominantly superior (87.7% of patients with pistol grip deformity and 88.5% of patients without pistol grip deformity had maximum narrowing in one of the superior sites). The superolateral site was the site of maximum narrowing in 33% of pistol grip deformity positive cases, compared with 29% of pistol grip deformity negative cases. The difference was not significant. Relationship between visual pistol grip deformity and femoral head to femoral neck ratio. The mean of the femoral head to femoral neck ratio was greatest in hips categorized as having a spherical femoral head, followed by hips classified in the indeterminate category, and then by hips with pistol grip deformity. The linear trend was statistically significant for both right and left hips (P for trend 0.001). The agreement between visual pistol grip deformity and the abnormal measured femoral head to femoral neck ratio was fair to moderate, with kappa values of 0.46 (P 0.001) and 0.35 (P 0.001) for right and left hips, respectively. DISCUSSION This is the first large radiographic study to present prevalence data for pistol grip deformity in normal adults and to show an association between pistol grip deformity and hip OA, taking into account other known risk factors. It is also the first study to use a simple interval measure (the femoral head to femoral neck ratio) in an attempt to quantify the spectrum of femoral head shape, ranging from spherical to nonspherical. As the femoral head to femoral neck ratio decreases, there is a progressive association with hip OA. Because of the usual symmetry of morphologic measures between right and left hips, the data obtained in subjects with unilateral hip OA support the hypothesis that pistol grip deformity and a decreasing femoral head to femoral neck ratio are both risk factors for the development of OA (17) and features that develop as a consequence of OA (18). Pistol grip deformity and a low femoral head to femoral neck ratio are predominantly found in men, and their very low prevalence in women (both found in 0.4% of controls) suggests that they have a low attributable risk for hip OA in women. With respect to neck shaft angle, a low angle appears to be a risk factor for OA, whereas a high angle is associated with OA only as a secondary phenomenon and not as a risk factor for development. In controls, we found pistol grip deformity to be a predominantly bilateral feature that was present in 6% of men but only 0.4% of women. This prevalence in men is very similar to the 8% prevalence observed in male skeletons by Goodman et al (10). However, that study found a much higher prevalence in female skeletons (6%), and in both sexes, the deformity was unilateral in 31% of the cases and occurred more often in the left hip (3:1), which was in contrast to the slightly more frequent occurrence in the right side (1.2:1) observed in our study. Neither study found an association of pistol grip deformity with age. Several factors may explain the differences in findings. For example, we undertook 2-D assessment of femoral head shape using radiographs of controls selected for the absence of OA, whereas the skeletal survey used 3-D assessment of cleaned femora in a collection that included femora with OA. Also, the subjective dichotomized categorization of pistol grip deformity used in the skeletal survey is likely to bias toward discordance. To date, there are no published studies showing radiographic prevalence of pistol grip deformity in normal hips with which to compare our findings. However, a recent study used a statistical active shape model on standard pelvis radiographs to investigate the change in femoral head shape that accompanies OA (34). Two groups were studied, those with K/L grade 0 both on baseline radiographs and on repeat radiographs 6 years later (55 non-oa controls), and those with grade 0 or 1 at baseline who showed worsening of at least 3 grades at 6 years (55 OA cases). Shape analysis was undertaken on 1 hip only, and, apart from age and sex, there was no adjustment for other OA risk factors. Unfortunately, the inferior femoral neck contour could not be included in the model, so neck width did not contribute to the 10 modes that were analyzed. In the group of patients with hip OA, there was an increase in modes 1 and 3 (both nonspherical head shapes) during the 6-year interval, reflecting the expected flattening of the femoral head with OA, but also a negative association with mode 6 (a more spherical shape) at baseline, with even further reduction in frequency of this mode at 6 years. The nonspherical, flattened head modes were particularly prevalent in men. Therefore, although pistol grip deformity was not specifically examined and consideration of neck width was omitted, the findings of that study indirectly support our conclusions, in that they showed a predominance of nonspherical head shape in men, an association between nonspherical head shape and predisposition to hip OA, and an increase in the frequency of this shape as a consequence of OA. In our study, there appeared to be a decrease in prevalence of pistol grip deformity with age in male

9 3180 DOHERTY ET AL controls, although this was not statistically significant (Figure 3). This might support the notion that pistol grip deformity is a constitutional risk factor for OA in men, in that left censorship of pistol grip deformity might occur as controls increasingly convert to cases with age. This hypothesis is supported by the results of comparison between the unaffected hips of patients with unilateral hip OA and the hips of controls, which showed a significant difference in men but not in women (Table 4). Although cross-sectional case control studies only demonstrate associations, we attempted to explore possible cause and effect by using the strategy of two hips, one person. Although there was some asymmetry of visually assessed pistol grip deformity, the striking symmetry between right and left hips of the measured features in normal controls and the apparent absence of an effect of age lend strong support to the hypothesis that these morphologic features are fixed constitutional characteristics of individuals. Therefore, in patients with unilateral hip OA, it seems reasonable to assume that the appearance of the unaffected hip is likely to represent the predisease appearance of the contralateral hip. Important caveats to this strategy are the possibility that bone remodeling and alteration of shape may be a generalized bone abnormality that predisposes to OA (34,35) and that bone remodeling and shape alteration may precede evidence of other radiographic changes in OA (34). Clearly, only a prospective study of incident cases can give definitive evidence of causation. If pistol grip deformity predisposes to hip OA via femoroacetabular impingement, it would be expected to cause an excess of superolateral OA. However, although we found a numerical increase in superolateral maximal narrowing (femoral head migration) in the unaffected hips of patients with pistol grip deformity, this was not statistically significantly different from that found in OA not associated with pistol grip deformity. The overall predominance of superolateral and indeterminate superior OA may have masked any such difference, and again, a prospective study that includes patients with less severe OA is required to examine this issue. Use of the femoral head to femoral neck ratio has several advantages over visual classification of pistol grip deformity. It provides an interval measure rather than a dichotomous variable, it characterizes the whole spectrum of shapes from spherical to nonspherical head, it is very simple to determine, it is more objective than pattern recognition used to identify pistol grip deformity, and it is more sensitive to change associated with remodeling (Table 4). The use of a cutoff value of 1.27 (mean 1.96 SD) to define abnormal femoral head to femoral neck ratio resulted in findings similar to those obtained by visual assessment of pistol grip deformity, but the greater symmetry of the ratio in controls suggests that the femoral head to femoral neck ratio is a more robust measure with less noise. The difference in performance between femoral head to femoral neck ratio and visual assessment of pistol grip deformity suggests that components of the ratio may capture more than just the nonspherical shape of the femoral head. Although an active shape model applied to the whole contour of the head and neck would seem to be an excellent way to assess bone shape from radiographs, the femoral head to femoral neck ratio offers a simple and cheap alternative, especially for large epidemiologic studies. The mean value for neck shaft angle in our controls (128 ) was similar to the value (126 ) reported in a previous study of 225 normal women (36) and was within the normal range ( ) proposed by Bombelli et al (25,37); the cutoff values used to define abnormally high valgus neck shaft angle (140.3 ) or low varus neck shaft angle (116.4 ) were clearly outside this range. One previous case control study, which compared radiographs of 28 patients with hip OA with those of 16 femora removed at autopsy, demonstrated reduced neck shaft angle in hip OA (group reduction of 5.1 ) (22), but a similar study comparing radiographs of 44 patients with hip OA with those of 33 cadaver specimens found no difference in neck shaft angle (23). Consistent with the results of the study by Moore et al (22), we found a higher prevalence of low neck shaft angle ( ) in cases than in controls (OR 1.78 [95% CI ]). Importantly, we also found an increased prevalence of a low angle in the unaffected hips of patients with unilateral OA (OR 1.79 [95% CI ]), supporting the notion that this feature is a risk factor for development of OA. In contrast, we found an increase in neck shaft angle in OA hips, but values similar to those found in controls in unaffected hips of cases with unilateral OA, suggesting that a high neck shaft angle is not a risk factor for development but may occur as a secondary consequence of OA. Taken together, these data suggest that the femoral bone remodeling that accompanies OA tends to increase but not decrease the neck shaft angle, as one might expect from the marked new bone formation ( buttressing ) that often involves the inferior femoral neck. We found no significant interaction between abnormal neck shaft angle and femoral head to femoral neck ratio, suggest-

10 NONSPHERICAL FEMORAL HEAD SHAPE AND RISK OF HIP OA 3181 ing that these exert independent effects on development of OA, presumably via different biomechanical means. In addition to the cross-sectional case control design, there are a number of limitations to our study. We examined 2-D images on plain radiographs, and, although we used a standardized protocol, variability in positioning may have influenced our assessments. The use of 3-D imaging (e.g., MRI) could provide more information, but such techniques are expensive and impractical for such a large study. Furthermore, although we accounted for currently known risk factors for OA, unrecognized factors may have confounded our results. The GOAL cohort was assembled primarily to study gene environment interaction in a Caucasian population. Cases were patients referred to the hospital with clinically severe hip OA, and controls were referred for an intravenous urography, so the generalizability of our findings to the whole population requires further study. There are no universally agreed upon criteria for case definition of hip OA, and we opted to use the minimum JSW for this purpose. Although this definition has been suggested to be appropriate for the purpose of epidemiologic study (32) and is the one that we used previously in a genetic study of hip OA (6), some researchers may prefer other definitions. Finally, a nonspherical femoral head shape may be an early feature of OA, present prior to other radiographic features, and thus be an early signal of developing OA rather than a predisposing constitutional risk factor. In conclusion, this large, appropriately adjusted case control study provides the first robust evidence that a nonspherical femoral head shape not only occurs as a consequence of OA, but itself may be a morphologic risk factor for development of hip OA, primarily in men. The femoral head to femoral neck ratio provides a simple and convenient way of measuring this shape on plain radiographs. A low neck shaft angle appears to be a risk factor for OA but does not result from the remodeling that accompanies OA, whereas an increased neck shaft angle may develop as a secondary feature of hip OA but does not appear to predispose to its development. Future prospective studies are required to confirm these findings. ACKNOWLEDGMENT We are indebted to Professor Michel Lequesne for the gift of an arthrometer. AUTHOR CONTRIBUTIONS Dr. Michael Doherty had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study design. Michael Doherty, Maciewicz, Muir, Zhang. Acquisition of data. Courtney, Sally Doherty, Jenkins. Analysis and interpretation of data. Michael Doherty, Zhang. Manuscript preparation. Michael Doherty, Maciewicz, Zhang. Statistical analysis. Zhang. REFERENCES 1. Van Saase JL, van Romunde LK, Cats A, Vandenbrouke JP, Valkenburg HA. Epidemiology of osteoarthritis: Zoetermeer survey. Comparison of radiological osteoarthritis in a Dutch population with that in 10 other populations. Ann Rheum Dis 1989;48: Gabriel SE, Crowson CS, O Fallon WM. Costs of osteoarthritis: estimates from a geographically defined population. J Rheumatol Suppl 1995;43: Brandt K, Doherty M, Lohmander S. The concept of osteoarthritis as failure of the diarthrodial joint. In: Brandt K, Doherty M, Lohmander S, editors. Osteoarthritis. 2nd ed. Oxford: Oxford University Press; p Felson DT, Lawrence RC, Dieppe PA, Hirsch R, Helmick CG, Jordan JM, et al. Osteoarthritis: new insights. Part 1. The disease and its risk factors. Ann Intern Med 2000;133: MacGregor A, Antoniades L, Matson M, Andrew T, Spector TD. The genetic contribution to radiographic hip osteoarthritis in women: results of a classic twin study. Arthritis Rheum 2000;43: Lanyon P, Muir K, Doherty S, Doherty M. Assessment of a genetic contribution to osteoarthritis of the hip: a sibling study. BMJ 2000;321: Harris WH. Aetiology of osteoarthritis of the hip. Clin Orthop Relat Res 1986;213: Smith RW, Egger P, Coggon D, Cawley MI, Cooper C. Osteoarthritis of the hip joint and acetabular dysplasia. Ann Rheum Dis 1995;54: Lane NE, Lin P, Christiansen L, Gore R, Williams EN, Hochberg MC, et al. Association of mild acetabular dysplasia with an increased risk of incident hip osteoarthritis in elderly white women: the study of osteoporotic fractures. Arthritis Rheum 2000;43: Goodman DA, Feighan JE, Smith AD, Latimer B, Buly RL, Cooperman DR. Subclinical slipped capital femoral epiphysis: relationship to osteoarthrosis of the hip [published erratum appears in J Bone Joint Surg Am 1999;81:592]. J Bone Joint Surg Am 1997;79: Thelin A. Hip joint arthrosis: an occupational disorder among farmers. Am J Ind Med 1990;18: Croft P, Cooper C, Wickham C, Coggon D. Osteoarthritis of the hip and occupational activity. Scand J Work Environ Health 1992;18: Cooper C, Inkskip H, Croft PR, Campbell L, Smith G, McLaren M, et al. Individual risk factors for hip osteoarthritis: obesity, hip injury and physical activity. Am J Epidemiol 1998;147: Vingard E, Alfredsson L, Goldie I, Hogstedt C. Sports and osteoarthritis of the hip: an epidemiologic study. Am J Sports Med 1993;21: Tonnis D, Heinecke A. Acetabular and femoral anteversion: relationship with osteoarthritis of the hip. J Bone Joint Surg Am 1999;12; Stulberg SD, Cordell LD, Harris WH, Ramsey PL, MacEwen GD. Unrecognised childhood hip disease: a major cause of idiopathic osteoarthritis of the hip. Proceedings of the Third Open Scientific Meeting of the Hip Society. St Louis (MO): Mosby; p Murray RO. The aetiology of primary osteoarthritis of the hip. Br J Radiol 1965;38: Resnick D. The tilt deformity of the femoral head in osteoar-

11 3182 DOHERTY ET AL thritis of the hip: a poor indicator of previous epiphysiolysis. Clin Radiol 1976;27: Ito K, Minka MA II, Leunig M, Werlen S, Ganz R. Femoroacetabular impingement and the cam-effect. J Bone Joint Surg Br 2001;83; Wagner S, Hofstetter W, Chiquet M, Mainil-Varlet P, Stauffer E, Ganz R, et al. Early osteoarthritic changes of human femoral head cartilage subsequent to femoro-acetabular impingement. Osteoarthritis Cartilage 2003;11: Leunig M, Beck M, Dora C, Ganz R. Femoroacetabular impingement: etiology and surgical concept. Oper Tech Orthop 2005;15: Moore RJ, Fazzalari NL, Manthey BA, Vernon-Roberts B. The relationship between head-neck shaft angle, calcar width, articular cartilage thickness and bone volume in arthrosis of the hip. Br J Rheumatol 1994;33: Reikeras O, Hoiseth A. Femoral neck angles in osteoarthritis of the hip. Acta Orthop Scand 1982;53: Vossinakis IC, Karnezis IA, Parry K, Learmonth ID. Radiographic associations for primary hip osteoarthrosis: a retrospective cohort study of 47 patients. Acta Orthop Scand 2001;72: Bombelli R, Santore RF, Poss R. Mechanics of the normal and osteoarthritic hip: a new perspective. Clin Orthop Relat Res 1984;182: O Reilly S, Johnson S, Doherty S, Muir K, Doherty M. Screening for hand osteoarthritis (OA) using a postal survey. Osteoarthritis Cartilage 1999;7: Thaper A, Zhang W, Wright G, Doherty M. The relationship between Heberden s nodes and underlying radiographic changes of osteoarthritis. Ann Rheum Dis 2005;64: Zhang W, Robertson J, Doherty S, Liu JJ, Maciewicz RA, Muir KR, et al. Index to ring finger length ratio and the risk of osteoarthritis. Arthritis Rheum 2008;58: Lequesne M, Morvan G. Description of the potential of an arthrometer for standard and reduced radiographs suitable to measurement of angles and segments of hip, knee, foot and joint space widths. Joint Bone Spine 2002;69: Wiberg G. A measuring method for distinguishing between a normal and a maldeveloped acetabulum. Acta Chir Scand 1939;83 Suppl 1: Kellgren JH, Lawrence JS. Radiological assessment of osteoarthrosis. Ann Rheum Dis 1957;16: Croft P, Cooper C, Wickham C, Coggon D. Defining osteoarthritis of the hip for epidemiological studies. Am J Epidemiol 1990;132: Burnett S, Hart DJ, Cooper C, Spector TD. A radiographic atlas of osteoarthritis. London: Springer-Verlag; Gregory JS, Waarsing JH, Day J, Pols HA, Reijman M, Weinans H, et al. Early identification of radiographic osteoarthritis of the hip using an active shape model to quantify changes in bone morphometric features: can hip shape tell us anything about the progression of osteoarthritis? Arthritis Rheum 2007;56: Hunter DJ, Hart D, Sneider H, Bettica P, Swaminathan R, Spector TD. Evidence of altered bone turnover, vitamin D and calcium regulation with knee osteoarthritis in female twins. Rheumatology (Oxford) 2003;42: Faulkner KG, Cummings SR, Black D, Palermo L, Gluer GC, Genant HK. Simple measurement of femoral geometry predicts hip fracture: the study of osteoporotic fractures. J Bone Miner Res 1993;8: Bombelli R. Osteoarthritis of the hip. 2nd ed. New York: Springer 1983.