To compare the knee-alignment angle from a full-limb radiograph (mechanical axis) with the anatomic-axis angle as measured by physical examination using a goniometer and by 2 other radiographic methods.
To compare the knee-alignment angle from a full-limb radiograph (mechanical axis) with the anatomic-axis angle as measured by physical examination using a goniometer and by 2 other radiographic methods.
The knee-alignment angle was measured in 114 knees of 57 subjects who had radiographic osteoarthritis (OA), with a Kellgren/Lawrence grade of ≥1 in at least one knee. The mechanical axis was defined as the angle formed by the intersection of 2 lines, one from the center of the head of the femur to the center of the tibial spines, and a second from the center of the talus to the center of the tibial spines. The anatomic axis was defined as the angle formed by 2 lines, each originating from a point bisecting the femur and tibia and converging at the center of the tibial spine tips. The anatomic-axis angle was measured by 3 methods: 1) physical examination using a goniometer, 2) a posteroanterior (PA) fixed-flexion knee radiograph (anatomicPA axis), and 3) an anteroposterior (AP) full-limb radiograph (anatomicAP axis).
Significant correlations were found between the mechanical-axis angle and the anatomic-axis angle measured by each of the 3 methods: by goniometer (r = 0.70, P < 0.0001), by anatomicPA axis (r = 0.75, P < 0.0001), and by anatomicAP axis (r = 0.65, P < 0.0001). The anatomic axis was offset a mean 4.21° valgus from the mechanical axis (3.5° in women, 6.4° in men), which was consistent across all methods.
Knee alignment assessed clinically by goniometer or measured on a knee radiograph is correlated with the angle measured on the more cumbersome and costly full-limb radiograph. These alternative measures have the potential to provide useful information regarding the risk of progression of knee OA when a full-limb radiograph is not available.
The impact of osteoarthritis (OA) is profound. Symptomatic knee OA occurs in ∼6% of Americans ages 30 years and older (1). It is a major cause of long-term disability in people older than age 50 years (2). Identifying the factors influencing progression of OA has implications for disease prognosis and therapeutic interventions. One of the factors shown to play a role in progression of knee OA is knee alignment. A recollection of knee malalignment in childhood (previous bow legs or knock knees) is associated with a 5-fold increased risk of OA progression in adulthood (3). Varus or valgus malalignment of the knee increases the odds of radiographic progression of tibiofemoral OA over an 18-month period (4), and may influence the risk of patellofemoral OA (5). Malalignment also has been shown by magnetic resonance imaging to increase the loss of cartilage volume (6).
In addition to being a primary factor in progression of knee OA, knee alignment acts in a synergistic manner with other mediators, such as obesity and baseline OA severity, to increase the risk of OA progression (7–9). The recognition of the importance of knee alignment has led to strategies, such as knee bracing and wedged insoles, to normalize the malaligned knee in an attempt to reduce pain and OA progression (10, 11). Tibial osteotomy has also long been an option for restoring normal alignment, to postpone the need for total knee arthroplasty, particularly in younger patients (12).
Currently, the gold standard measure of knee alignment is the angle formed by the intersection of the lines drawn from the hip to the knee and from the knee to the ankle on a radiograph of the entire lower extremity. This full-limb radiograph entails administration of radiation to the pelvis and requires a grid and large graduated cassette, as well as the expertise of a radiology technologist to overcome the inherent technical difficulties of the study (the substantial difference in the amount of soft tissue at the hip and the ankle). The goal of the present study was to investigate alternative methods of measuring knee alignment. We compared the knee-alignment angle from a full-limb radiograph (mechanical axis) with the anatomic-axis angle measured by physical examination with a goniometer or measured on a posteroanterior (PA) fixed-flexion knee radiograph (anatomicPA axis) or on an anteroposterior (AP) full-limb radiograph (anatomicAP axis). To our knowledge, this is the first report to describe the relationship between these 3 alternative methods of determining knee alignment.
A subset of 57 participants in the National Institutes of Health–funded Prediction of Osteoarthritis Progression Study were evaluated. Participants met the American College of Rheumatology criteria for symptomatic OA of at least one knee (13). In addition, all participants met the radiographic criteria for OA by having a Kellgren/Lawrence (K/L) grade of ≥1 in at least one knee (14). Participants were excluded on the basis of knee replacement or bilateral knee OA of K/L grade 4. Knee radiographs were read by 2 graders (VBK and GM). The sample included 4 participants who had undergone unilateral hip replacement for OA and 1 patient who had undergone unilateral internal hip fixation for traumatic fracture. A total of 114 knees were included in the analysis.
Goniometer readings (to within 1°) were obtained with the participant in a standing position and barefoot, with toes placed forward and feet shoulder-width apart. Measurements were made with a goniometer (Lafayatte Instrument, Lafayatte, IN), the arms of which could be extended 50 cm in both directions, permitting the examiner to visually bisect the thigh and lower leg along their lengths. The centers of both the patella and ankle were located and marked with a pen. The center of the goniometer was placed on the center of the patella, and the arms of this goniometer were extended along the center of the thigh and along the axis of the lower leg to the center of the ankle. The intraobserver reliability of goniometer readings was evaluated on a subset of 20 participants, whose measurements were repeated once during the day of the clinical examination. The radiographic determinations of knee alignment were made by an examiner who was blinded to the results of the clinical assessment of knee alignment.
The mechanical axis (to within 0.5°) was measured according to the method described by Sharma et al (4) on a weight-bearing AP radiograph that included both lower extremities, which were imaged from the pelvis to the ankle (a full-limb radiograph). The mechanical axis was defined as the angle formed by the intersection of a line from the center of the head of the femur to the center of the tibial spines, and a second line from the center of the ankle talus to the center of the tibial spines (4). Concentric circles, known as Mose circles and imprinted on a transparent template, were superimposed on the femoral head to precisely locate its center (15). The gold standard criterion, the mechanical axis, was used as the reference standard, with angles <180° defined as varus alignment, and angles >180° defined as valgus alignment.
The anatomic-axis angle (anatomicPA axis) (to within 0.5°) was measured on a fixed-flexion PA knee radiograph obtained with the SynaFlex x-ray positioning frame (Synarc, San Francisco, CA) (16). The SynaFlex frame allows convenient, reproducible positioning of the knee without the need for foot maps. The feet are externally rotated 10°, the knees and thighs touch the vertical platform anteriorly, and the x-ray beam is angulated 10° caudally. The anatomic axis was defined as the angle formed by the intersection of 2 lines originating from points bisecting the femur and tibia and converging at the center of the tibial spine tips. The origin of these lines was 10 cm from the knee-joint surfaces when included in the field of view on the radiograph; otherwise, the furthest distance from the knee-joint surfaces (up to 10 cm) was used. The lines could be extended to the full 10 cm in both directions on the knee radiographs of all but 3 subjects (4 knees total, equal to 3.5% of all knees). The angle of the anatomic axis was also determined on an AP full-limb radiograph (anatomicAP axis) using the same landmarks, originating 10 cm from the knee-joint surfaces. The offset of the anatomic axis from the mechanical axis was determined for all 3 methods, based on linear regression.
Statistical analyses included linear regression to evaluate the correlation between angles measured by the different methods, and two-way analysis of variance to calculate the intraclass correlation coefficient (ICC) of repeated measurements by goniometer. Multivariate regression modeling of the dominant knee was used to evaluate the association of the degree of malalignment (the absolute difference between the mechanical-axis angle and 180°) and OA, body mass index (BMI), age, and sex, and to evaluate sex-related differences between the mechanical and anatomic axes. All statistical analyses were performed with JMP software (SAS, Cary, NC).
The sample consisted of 57 participants with symptomatic knee OA. The participants (40 women, 17 men) ranged in age from 35 years to 85 years, with a mean ± SD age of 67 ± 11 years. The majority (80%) of this cohort had medial compartment dominant disease. The descriptive characteristics of this cohort are summarized in Table 1. Based on the standard K/L radiographic grading system, the proportions of the 114 knees within each category of OA were as follows: 2.6% with K/L grade 0, 21.9% with K/L grade 1, 26.3% with K/L grade 2, 36.9% with K/L grade 3, and 12.3% with K/L grade 4. The mean ± SD alignment of the 3 normal (K/L grade 0) knees was 2 ± 1.3° varus, based on the mechanical-axis angles.
|Women (n = 80 knees)||Men (n = 34 knees)|
|Age, mean ± SD years||65.7 ± 11.9||68.8 ± 8.1|
|BMI, mean ± SD kg/m2||31.7 ± 6.7||29.6 ± 18.3|
|Mechanical-axis angle, mean ± SD degrees||178.3 ± 4.8||177.4 ± 3.9|
|AnatomicAP angle, mean ± SD degrees||180.9 ± 4.5||182.5 ± 3.8|
|AnatomicPA angle, mean ± SD degrees||181.0 ± 4.9||181.3 ± 5.3|
|Goniometer angle, mean ± SD degrees||182.1 ± 3.9||181.7 ± 3.9|
|Knee alignment, no. (%)|
|Genu varus||40 (50.0)||19 (55.9)|
|Genu valgus||31 (38.7)||5 (14.7)|
|Normal alignment||9 (11.3)||10 (29.4)|
The mechanical axis measured on a full-limb radiograph is the gold standard method for assessing knee alignment. The mechanical-axis angles in all 114 knees ranged from a minimum of 164° to a maximum of 193° (16° varus to 13° valgus alignment). In bivariate analyses, the degree of malalignment was not associated with age or sex, but was associated with BMI (r = 0.36, P = 0.005) and OA severity (r = 0.41, P = 0.001). In a multivariate model adjusted for age and sex, the degree of malalignment was associated with OA severity (parameter estimate = 1.50, P = 0.0005) but not with BMI (P = 0.15). As OA severity increased, the variation in knee alignment increased, as demonstrated by the difference between the maximum and minimum knee-alignment angles within each K/L grade: 2.5° for K/L grade 0, 11° for K/L grade 1, 13.5° for K/L grade 2, 22° for K/L grade 3, and 21° for K/L grade 4. This variation was also demonstrated by the mean ± SD angles of the knees in each K/L grade, as follows: 178.0 ± 1.3° for K/L grade 0, 179.6 ± 2.8° for K/L grade 1, 177.8 ± 4.0° for K/L grade 2, 177.9 ± 5.0° for K/L grade 3, and 176.0 ± 6.4° for K/L grade 4.
The knee-alignment angles recorded by goniometer ranged from 170° to 191°. Although the absolute values of knee alignment differed due to the different frames of reference, a significant correlation was found between the angle measured by the goniometer and the mechanical-axis angle (r = 0.70, P < 0.0001) (Figure 1A). The relationship between the mechanical-axis angle and the goniometer angle was as follows: mechanical-axis angle = [0.82(goniometer angle)] + 28.85. For 20 participants, knee alignment was measured by goniometer twice on the day of the clinical evaluation, at least 1.5 hours apart. The ICC of these 2 measurements was high, with a value of 0.94.
Knee alignment was also measured from a fixed-flexion knee radiograph (anatomicPA axis) using landmarks bisecting the femur and tibia and 10 cm from the surfaces of the knee joint. The angle of the anatomicPA axis ranged from 167.5° to 195.5°. The angle of the anatomicPA axis was strongly associated with the angle of the mechanical axis (r = 0.75, P < 0.0001) (Figure 1B). Using regression analysis, the relationship between the mechanical-axis angle and the anatomicPA-axis angle was as follows: mechanical-axis angle = [0.69(anatomicPA axis)] + 53.69. The correlation of these 2 measures in women (r = 0.76, P = 0.0001) was similar to that in men (r = 0.77, P = 0.0001). Correlations of similar magnitude were obtained from the right knees and the left knees, and the results were also unchanged when the 5 participants with unilateral hip prostheses were excluded. A significant association was also found between the angle measured by goniometer and that of the anatomicPA axis (r = 0.56, P < 0.0001), with the relationship calculated as follows: anatomicPA-axis angle = [0.72(goniometer angle)] + 49.28.
The anatomic axis was also measured on the same full-limb AP radiograph from which the mechanical-axis angle was measured, using the same landmarks (10 cm above and below the joint surfaces) as for the PA fixed-flexion radiograph. Under these conditions, the correlation of the anatomicAP-axis and mechanical-axis angles was high (r = 0.65, P < 0.0001) (Figure 1C), calculated as follows: mechanical-axis angle = [0.67(anatomicAP axis] + 55.86. The correlation between the angles of the anatomic axis measured by the 2 methods, AP and PA fixed-flexion radiographs, was also high (r = 0.73, P < 0.0001), calculated as follows: anatomicPA-axis angle = [0.83(anatomicAP axis)] + 30.62.
The value of 180° on the mechanical axis defined the threshold between varus and valgus alignment. By interpolation, we could calculate the angle offset for the alternative frames of reference using the relationships determined by regression analysis. The anatomic axis was offset a mean 4.21° valgus from the mechanical axis (3.5° in women, 6.4° in men). The neutral angles determined from the 3 methods of measuring the anatomic axis are listed in Table 2. The offset for each method was consistently greater for men than for women, and the difference was statistically significant. For instance, the absolute difference between the anatomicPA-axis angle and mechanical-axis angle was greater for men (P = 0.01, parameter estimate = 1.23°, in a multivariate model adjusted for age, BMI, and OA severity). There was no association of height with the absolute difference between the mechanical- and anatomic-axis angles. When the neutral angles in Table 2 were used as cutoff points to classify knees as having either varus or valgus alignment, the 3 methods of measuring the anatomic-axis angle classified 80% of the knees correctly, using the mechanical axis as the gold standard.
|Full-limb radiograph (mechanical)||180 (0)||180 (0)||180 (0)|
|Physical examination (goniometer)||183.97 (3.97)||185.89 (5.89)||184.37 (4.37)|
|Fixed-flexion radiograph (anatomicPA)||183.27 (3.27)||185.87 (5.87)||183.96 (3.96)|
|Full-limb radiograph (anatomicAP)||183.10 (3.10)||187.40 (7.40)||184.31 (4.31)|
There is evidence that knee alignment is associated with progression of OA and decline in physical function. Although a few studies have relied on anatomic-axis determinations of knee-alignment angle (6, 17, 18), quantification of the relative risk of OA progression based on knee alignment has been primarily based on the mechanical-axis angle as measured on a full-limb radiograph (4, 8). This method is cumbersome, requires specialized equipment and expertise, and is costly, particularly for large clinical studies. In our study, we set out to validate more clinically accessible measures of knee alignment against the gold standard criterion of the mechanical axis measured on a full-limb radiograph. To our knowledge, the relationship between these measures has not been previously explored.
The 3 normal (K/L grade 0) knees were, on average, in 2° varus alignment, which is consistent with the reported 1.1–1.5° varus alignment of normal knees (18). We observed a mean physiologic offset of 4.3° valgus for the anatomicAP axis from the mechanical axis. The anatomicPA axis was offset a similar amount (4.0°), as was the goniometer measure, which was centered on a focal point at the patella (4.4°). All of these results are consistent with reported values of physiologic valgus offset for the anatomic axis (results varying by method of ascertainment), ranging from 4° to 6° (19, 20). Of interest was the fact that the offset was less in women than in men. This may, at least in part, be accounted for by reported sex differences in distal femoral valgus (21). Women have significantly more distal femoral valgus than men, in both non-OA and OA patients (21), with variations due to sex of 0.6–0.8° (20). This refers to the fact that the angle on the medial aspect of the knee, formed by the hip-to-knee vector of the mechanical axis (tangent to the condyles), is greater in women. The end result would be a closer approximation of the mechanical axis to the anatomic axis and thus less of an offset of these 2 measures in women. The origin of this difference in femoral geometry is unknown, but it has been speculated that it may result from sex-related differences in femoral bowing (21).
The PA radiograph was obtained with the SynaFlex frame to precisely position the feet and create an ∼30° angle of knee flexion. This frame controls rotation of the femur relative to the tibia and has been used increasingly in clinical trials. Limb rotation affects the measurement of limb alignment and creates more measurement variability of the anatomic axis than of the mechanical axis (22). Varying the flexion of the knee has been shown to be the main source of effect of rotation on limb alignment. The SynaFlex device minimizes one source of measurement error, namely inconsistent limb positioning, and this may account for the fact that a slightly better correlation was observed between the anatomicPA-axis angle and the mechanical-axis angle than was observed with the anatomicAP-axis angle measured from the same full-limb radiograph as the mechanical axis.
The measurement of knee-alignment angle by goniometer has been shown to be highly reproducible when performed by rheumatologists in the evaluation of patients with OA (23). In fact, alignment measured by goniometer was the most reliable physical examination technique of the 42 techniques evaluated. It was not our intention to extensively test interrater reliability of these measures, but rather to assess the interrelationships of knee-alignment angle ascertained by a variety of methods. We did evaluate the intrarater reliability for the goniometer measure for a single examiner, and it was high. The goniometer used in these studies has extendable arms, which facilitated obtaining accurate measurements. Although we contemplated alternative foci for centering of the goniometer, we found the center of the patella to be the most readily identified physical landmark, and we therefore performed all measurements with this focal point. In theory, patellar subluxation may confound assessments by goniometer, but this did not appear to be a major problem in this sample of 114 knees.
In summary, we have found that alternative methods of measuring knee alignment (by knee radiograph and by goniometer), which entail little or no radiation exposure, correlated surprisingly well with the knee-alignment angle measured on the more cumbersome and costly full-limb radiograph that includes pelvic irradiation. Each alternative method has its strengths. The anatomicAP measurement approximates the type of measurement available from the standard AP standing knee radiograph acquired in the clinical setting. The fixed-flexion radiograph minimizes measurement error due to variable limb positioning and rotation. The goniometer measurement would be available to every clinician at little cost. Thus, these more readily accessible measures may provide useful information regarding the risk of knee OA progression, in the absence of full-limb radiographs.
We thank Dr. Leena Sharma for her technical advice and Dr. Greg Samsa for his advice regarding statistical analyses. We also thank Diana Boney-Sessoms and Marie Stone for expert radiology technical services.