Relationship Between Frontal Knee Alignment and Reference Axes in the Distal Femur

Abstract

The two transepicondylar axes (the clinical and surgical epicondylar axes), the posterior condylar axis, and the anteroposterior axis were constructed using computed tomography scans in 111 (66 patients) knees with symptomatic arthritis. The relationships between angles made by these reference axes and two angles indicating frontal knee alignment (the tibiofemoral valgus angle and the femoral valgus angle) were investigated. In ¼ of the knees, the surgical epicondylar axis could not be constructed because the sulcus of the medial epicondyle was not recognizable. The condylar twist angle was almost constant and averaged 6° when the femoral valgus angle was 9° or less, but increased gradually when the angle was greater than 9°. The difference between the condylar twist angle and the posterior condylar angle was constantly 3°. The anteroposterior axis was almost at right angles to the clinical epicondylar axis, and the relationship between these axes was constant, independent of the femoral valgus angle. With 3° to 6° external rotation relative to the posterior condylar axis, the femoral component could be set parallel to the transepicondylar axis in common varus or neutral knees. In cases with a larger femoral valgus angle, the anteroposterior axis would be a more reliable reference axis. Preoperative computed tomography scans are recommended for patients with knees with severe valgus deformity or severe hypertrophic osteoarthritis.

Recently, a concept of knee kinematics has been proposed in which flexion and extension motion of the knee occurs about one fixed axis in the posterior femoral condyles. ^6,7,10,21,24 This functional axis passes the origins of the medial and lateral collateral ligament, coinciding with the transepicondylar axis. This fixed flexion and extension axis theory, which is based on three-dimensional observation of the knee, is replacing the classic concept of the variable flexion and extension axis theory, which is based on observation in the sagittal plane. ^5,8,19,20,22 If this new concept is assumed to be correct, it is logical to set the rotational alignment of the femoral component parallel to the transepicondylar axis so that more physiologic ligamentous balance, patellar tracking, and kinematics can be achieved.

Many orthopaedic surgeons usually determine the rotational alignment of the femoral component by using a line connecting the posterior aspects of the femoral condyles (the posterior condylar axis) and set the component in a slightly externally rotated position relative to the axis. ^1,11,13,15 The external rotation is thought to be necessary to compensate for the angular discrepancy that results from the proximal tibial cut perpendicular to the tibial shaft axis because the tibial articular surface is in approximately 3° varus in normal knees. ¹⁶ Thus, many instrumentation systems used for total knee replacement adopt external rotation cutting of 3° relative to the posterior condylar axis to attain a parallel flexion gap. However, it is not clear that the 3° external rotation could set the femoral component parallel to the transepicondylar axis in arthritic knees.

The reference axes of the distal femur and angles made by these axes have been studied in detail using anatomic specimens. Some authors reported that the condylar twist angle (an angle between the transepicondylar axis and the posterior condylar axis) was approximately 5°, although the angle varied among individuals to some extent. ^3,14,24 Poilvache et al ¹⁷ measured the condylar twist angle at surgery and reported that the transepicondylar axis was at 3.6° ± 2.0° external rotation relative to the posterior condylar axis in arthritic knees, which is close to the results of previous anatomic studies. According to these data, 4° to 5° external rotations seem to be appropriate for setting the femoral component parallel to the transepicondylar axis.

Berger et al ⁴ proposed two transepicondylar axes: the clinical epicondylar axis and the surgical epicondylar axis. The former is a line connecting the lateral epicondylar prominence and the most prominent point of the medial epicondyle, which is on the crescent-shaped attachment of the superficial fibers of the medial collateral ligament. The latter is a line connecting the lateral epicondylar prominence and the sulcus of the medial epicondyle, which is the attachment of the deep fibers of the medial collateral ligament (Fig 1). Berger et al ⁴ showed that the condylar twist angle (an angle between the clinical epicondylar axis and the posterior condylar axis) is 4.7° in males and 5.2° in females, and that the posterior condylar angle (an angle between the surgical epicondylar axis and the posterior condylar axis), which is made by the surgical epicondylar axis and the posterior condylar axis, is 3.5° in males and 0.3° in females. In addition, they showed that the surgical epicondylar axis is more reliable and reproducible than the clinical epicondylar axis. According to their observations, 0° to 3° external rotation may be appropriate to set the component parallel to the surgical epicondylar axis. However, in arthritic knees with a varus or valgus deformity, in which distortion of the medial and lateral femoral condyles may vary among individuals, it is unclear how much external rotation should be provided relative to the posterior condylar axis to set the femoral component parallel to the transepicondylar axis.

Fig 1.: The locations of the landmarks for the clinical and surgical epicondylar axes are shown. a = the lateral epicondylar prominence; b = the most prominent point of the medial epicondyle; c = the sulcus of the medial epicondyle.

Recently, Griffin et al ⁹ reported, based on intraoperative measurements in osteoarthritic knees, that the posterior condylar angle was variable and the posterior condyles were potentially unreliable references for femoral component rotation in some knees. The authors considered that this observation needed to be verified by a more precise measuring method because in some cases it may be difficult to determine the location of the sulcus of the medial epicondyle at surgery.

The purpose of the current study was to investigate the relationship between frontal knee alignment on plain radiographs and angles made by these reference axes (the clinical and surgical epicondylar axes, posterior condylar axis, and the anteroposterior [AP] axis ^3,23) in osteoarthritic and rheumatoid arthritic knees with various deformities. The reference axes in the arthritic knees were constructed spatially using multiple computed tomography (CT) scans. The authors provide information for determining the rotational alignment of the femoral component in total knee replacement.

MATERIALS AND METHODS

Plain AP radiographs and CT scans were used to evaluate 111 knees (66 patients) with symptomatic arthritis. The diagnoses were osteoarthritis in 46 knees in 27 patients (mean age, 67.2 years) and rheumatoid arthritis in 65 knees in 39 patients (mean age, 60.2 years). Patients with traumatic osteoarthritis and osteonecrosis of the knee were not included. Women accounted for 88% of all patients. Three observers (MA, TN, EY) made radiographic measurements. One observer (MA) constructed the reference axes of the distal femur on CT scans, and one observer (TN) measured the angles made by these axes. The third observer (EY) drew lines on plain radiographs and measured the angles made by these lines. The nomenclature of the reference axes and angles made by these axes followed that reported previously by Berger et al ⁴ and Yoshioka et al. ²⁴

The tibiofemoral valgus angle and the distal femoral valgus angle ¹² on plain AP radiographs obtained with the patients in the supine position were chosen to represent frontal knee alignment (Fig 2). Alignments of the knees were classified according to the tibiofemoral valgus angle as varus (angle < 0°), neutral (angle between 0° and 7°), or valgus (angle > 7°). Alignment of the femurs also was classified according to the femoral valgus angle as varus (angle < 8°), neutral (angle between 8° and 10°), or valgus (angle > 10°).

Fig 2.: Angular measurements on an AP radiograph of the knee. α = tibiofemoral valgus angle; β = distal femoral valgus angle. These angles were chosen to represent frontal knee alignment.

Transverse CT scans were made from the level of the proximal pole of the patella to the distal end of the femur at 3-mm intervals so that the scan was parallel to a line connecting the distal ends of the condyles. By using four continual scans on which the medial or lateral epicondyles could be recognized, the posterior condylar, clinical epicondylar, surgical epicondylar, and AP axes were drawn (Fig 3). The medial and lateral prominences of the epicondyles, the sulcus of the medial epicondyle, and the most posterior points of the femoral condyles were located spatially on a series of CT scans by the following method. The most prominent points of the medial and lateral epicondyles on the first image were marked on a transparent sheet. This sheet was overlaid on the second, third, and fourth images, matching the spatial position, and the most prominent points of the epicondyles in these images also were marked on the sheet. From these four points for each epicondyle, the most medial and lateral points were chosen as the most prominent points of the medial and lateral epicondyles for that case. A line connecting these most prominent points on a series of CT scans was defined as the clinical epicondylar axis. The deepest point of the sulcus on the medial epicondyle was determined on the sheet, and a line connecting this point and the most prominent point of the lateral epicondyle on a series of CT scans was defined as the surgical epicondylar axis. In many cases it was difficult to recognize the sulcus, because of flattening of the medial epicondyle or because of bone formation on it, so the condition of the medial epicondyle was classified as Type 1 (the sulcus is well recognizable on CT scans), Type 2 (the sulcus barely is recognizable on CT scans), or Type 3 (the sulcus is not recognizable on CT scans) (Fig 4). The most posterior points of the medial and lateral femoral condyles were determined by the same method, and a line connecting these two points on a series of CT scans was defined as the posterior condylar axis.

Fig 3.: Construct of the reference axes on transverse CT scans of the distal femoral condyle is shown. CEA = clinical epicondylar axis; PAA = line perpendicular to the anteroposterior axis; PCA = posterior condylar axis; SEA = surgical epicondylar axis. These axes were determined from a series of multiple scans (see text).

Fig 4.: The types of medial epicondyles are shown on CT scans. Type 1 = the sulcus is well recognizable; Type 2 = the sulcus barely is recognizable; Type 3 = the sulcus is not recognizable. The prominence of the medial epicondyle is recognizable in all types (arrowhead).

The AP axis was defined as a line connecting the deepest point of the patellar groove anteriorly and the center of the intercondylar notch posteriorly. In cases with osteophyte formation in the patellar groove or in the intercondylar notch, the deepest point of the groove and the center of the notch were determined after elimination of these osteophytes. The deepest point of the patellar groove and the center of the notch usually were determined by using the most distal one of a series of CT scans because the deepest point of the groove could be located more clearly in that scan. The condylar twist angle, the posterior condylar angle, and the angle made by the clinical epicondylar axis and the line perpendicular to the AP axis were measured in each case.

All variables measured were entered into a computer data sheet. Statview 4.02 (Abacus Concepts, Berkley, CA) was used for statistical analysis by analysis of variance (ANOVA) and Scheffe’s method as a post hoc test.

RESULTS

In all cases, the medial and lateral epicondylar prominences could be determined by the methods mentioned. Thus, the posterior condylar axis, the clinical epicondylar axis, and the AP axis could be constructed in all cases so that the condylar twist angle, the angle between the clinical epicondylar axis, and the line perpendicular to the AP axis were measured. However, in 29 cases it was impossible to recognize the sulcus of the medial epicondyle because of flattening or because of bony formation on it (Type 3 medial epicondyle). Consequently, in only 82 cases (Type 1 medial epicondyle in 27 cases and Type 2 medial epicondyle in 55 cases), the surgical epicondylar axis could be constructed so that the posterior condylar angle and the angle between the clinical and surgical epicondylar axes were measured.

Table 1 shows results of the angular measurements (the tibiofemoral valgus angle, femoral valgus angle, the condylar twist angle, and the angle made by the clinical epicondylar axis and the line perpendicular to the AP axis) according to the subdivisions of the frontal knee alignments. There was a significant difference (p < 0.001) between the mean condylar twist angles of valgus (8.8°; standard deviation, 3.2°) and varus knees (6.4°; standard deviation, 1.4°), but not between those of varus and neutral knees (7.2°; standard deviation, 2.0°) or between those of neutral and valgus knees. The mean condylar twist angle of the valgus femurs (9.1°; standard deviation, 2.4°) was significantly larger than those of the neutral (6.4°; standard deviation, 1.6°) and varus (6.1°; standard deviation, 1.4°) femurs (p < 0.0001), which were not significantly different from each other.

TABLE 1: The Angular Measurements According to the Frontal Knee Alignments

The relationship between the condylar twist angle and the tibiofemoral valgus angle was examined by linear regression analysis. However, there was a poor correlation between these parameters. The relationship between the condylar twist angle and the femoral valgus angle was examined (Fig 5), and a much better correlation was found compared with the relationship between the condylar twist angle and the tibiofemoral valgus angle. Based on the authors’ observation of the scatter diagram, three knees in two patients were not included in the statistical analysis because large osteophyte formations were seen on the posterior aspect of the medial femoral condyle (Cases 1 and 2) and on the posterior aspect of the lateral femoral condyle (Case 3). A lateral plain radiograph and CT scans of Case 2 are shown in Figure 6. The relationship between the femoral valgus angle and the condylar twist angle in 108 cases (excluding the three problematic cases) were examined by linear regression analysis, which was done separately in two ranges (the femoral valgus angle of 9° or less, or the angle greater than 9°). The condylar twist angle was almost constant at approximately 6° when the femoral valgus angle was 9° or less, but increased gradually when the femoral valgus angle was greater than 9° (Fig 5). When the femoral valgus angle was greater than 9°, the correlation between the condylar twist angle and the femoral valgus angle was relatively good and significant (r = 0.65; p < 0.0001).

Fig 5.: The relationship between condylar twist angle and femoral valgus angle is shown. When the femoral valgus angle was 9° or less, the condylar twist angle was almost constant at approximately 6°. When the femoral valgus angle was greater than 9°, the condylar twist angle increased gradually. Cases 1, 2, and 3 are shown as black circles. CEA = clinical epicondylar axis; PCA = posterior condylar axis.

Fig 6A–B.: (A) Lateral radiograph (Case 2) shows abnormal bone formation of the posterior aspect of the medial femoral condyles (arrowheads). (B) Computed tomography scan shows abnormal bone formation on the posterior aspect of the medial femoral condyle. On a plain AP radiograph, the tibiofemoral valgus angle was −2° and the femoral valgus angle was 10°; on CT scans, the condylar twist angle was 14° and the angle between the clinical epicondylar axis (CEA) and the line perpendicular to the anteroposterior axis (PAA) was 1°. PCA = posterior condylar axis.

The mean posterior condylar angle in the 82 knees with Type 1 or Type 2 medial epicondyles was 4.2° (standard deviation, 2.1°; range, 1°–12°). The mean of the angle between the clinical and surgical epicondylar axes was 3.2° (standard deviation, 0.7°; range, 2°–5°). The correlation between these axes was strong and significant (r = 0.95; p < 0.0001).

The relationship between the angle made by the clinical epicondylar axis and the line perpendicular to the AP axis and the femoral valgus angle (including the three cases excluded from the previous statistical analysis) was examined (Fig 7). The mean angle between these axes was 0.4° (standard deviation, 1.8°; range, −3.5°–4°). Table 1 shows the mean angles between the condylar twist angle and the line perpendicular to the AP axis according to the frontal knee alignments. There were no significant differences in the mean angles between the varus, neutral, and valgus knees or between the varus, neutral, and valgus femurs.

Fig 7.: The relationship between the angle made by the clinical epicondylar axis and the line perpendicular to the AP axis and the femoral valgus angle. The angle between these axes was almost constant at 0°, independent of variations of the femoral valgus angle. CEA = clinical epicondylar axis; PAA = anteroposterior axis.

DISCUSSION

In previous studies, angles of reference axes in the distal femur have been measured in detail in normal cadaveric specimens. ^3,4,14,24 In the current study, the authors used CT scans to construct reference axes in arthritic knees of patients with symptomatic osteoarthritis or rheumatoid arthritis to measure the angles made by these axes in the distal femur. Because the study population was limited to Japanese patients, the results of this study should be interpreted cautiously. The data included in this study may be typical for an Asian population, and there may be anatomic differences in the Caucasian population.

One of the problems in constructing reference axes and measuring angles on CT scans was that the angles between the reference axes varied among the four or femoral valgus angle scans on which the medial and lateral condyles could be recognized. A method using multiple scans was adopted so that those anatomic reference axes were determined spatially. Another problem was that the sulci of the medial epicondyles were obscured in many cases because of flattening of the epicondyle or because of abnormal bone formation on it. Using anatomic specimens, Berger et al ⁴ showed that the surgical epicondylar axis is a reproducible, secondary anatomic reference axis and claimed that it is useful for the rotational orientation of the femoral component when the posterior condylar surfaces cannot be used. However, in arthritic knees, for which total knee arthroplasty is done, only 25% of sulci were recognized clearly on the medial epicondyle, and in more than 25% of cases, it was impossible to recognize the sulcus and construct the surgical epicondylar axis. This observation may partly explain why the posterior condylar angle measured intraoperatively had large variability in osteoarthritic knees. ⁹ However, the most prominent point of the medial epicondyle could be recognized in all cases. Thus, the authors used the clinical epicondylar axis (the condylar twist angle) for statistical analysis in the current study (Fig 5) and think the most prominent point of the medial epicondyle is more likely than the sulcus to be available as a rotational landmark at the time of total knee replacement.

The relationship between the condylar twist angle and the tibiofemoral valgus angle was weaker than that between the condylar twist angle and the femoral valgus angle. This is because many factors can affect the tibiofemoral valgus angle, such as the femoral valgus angle, the tibial varus angle (destruction of the tibial plateau), balance between the medial and lateral collateral ligaments, and thickness of the medial and lateral joint spaces. The authors do not recommend the tibiofemoral valgus angle for prediction of the extent of the external rotation of the femoral component and think the femoral valgus angle is more favorable.

According to the results of the current study, the condylar twist angle was almost constant and averaged 6° when the femoral valgus angle was 9° or less. This result suggests that in common varus or neutral knees, the femoral component could be set parallel to the clinical epicondylar axis when the posterior condyles are resected with external rotation of 6° in varus or neutral knees. In 82 cases in which the condylar twist angle and the posterior condylar angle could be measured, these two angles showed good correlation (r = 0.95), and the difference between the two was consistently 3°. As a result, setting the femoral component with an external rotation of 3° relative to the posterior condylar axis would be appropriate in knees with common varus or neutral deformity to ensure they align parallel to the surgical epicondylar axis. When the femoral valgus angle was 9° or greater, there was a significant correlation between the condylar twist angle and the femoral valgus angle, and the condylar twist angle increased gradually with the femoral valgus angle. The mean condylar twist angle in the valgus femurs was significantly larger than those in the varus and neutral femurs. In valgus knees with a femoral valgus angle greater than 9°, greater than 3° external rotation relative to the posterior condylar axis would be necessary to set the component parallel to the surgical epicondylar axis and greater than 6° would be necessary for the clinical epicondylar axis.

The line perpendicular to the AP axis essentially was parallel to the clinical epicondylar axis, so the AP axis is a reliable rotational reference axis, not only in valgus knees with a large femoral valgus angle, but also in varus or neutral knees. Thus, the femoral component would be set parallel to the clinical epicondylar axis when the anterior and posterior femoral condyles are cut perpendicular to the AP axis. However, the angles showed a relatively large range (−3.5°–4°). The cadaveric study reported by Arima et al ³ also showed a relatively large range of angles measured radiographically (0.5°–7.0°).

It is not clear which anatomic axis the femoral component should be set parallel to, the surgical or clinical epicondylar axis, because it is not known which axis is closer to the functional flexion and extension axis. The surgical epicondylar axis may be closer to the functional axis for the femorotibial articulation because the offset of the flexion and extension axis from the condylar surface averages 3° in the coronal and in the transverse planes according to Hollister et al. ¹⁰ In addition, the surgical epicondylar axis is more reproducible than the clinical epicondylar axis, according to the observation of Berger et al. ⁴ However, the clinical epicondylar axis may be closer to the functional axis for the patellofemoral articulation because the AP axis was perpendicular to the clinical epicondylar axis according to the current study. Anouchi et al ² and Rhoads et al ¹⁸ showed with anatomic specimens that setting the femoral component with external rotations of 5° and 10°, respectively, relative to the posterior condylar axis made patellar tracking more physiologic. Additional kinematic and anatomic studies based on accurate three-dimensional observations are needed to clarify the relationship between the functional axes for the flexion and extension rotation, the patellar tracking, and the anatomic reference axes.

Preoperative CT scans would be useful in cases with severe valgus deformity or severe hypertrophic osteoarthritis. In the current study, three knees had exceptional condylar twist angles compared with the femoral valgus angle. Cases 1 and 2 (Fig 6) had very large condylar twist angles (15° and 14°) and a normal femoral valgus angle (10°). Case 3 had a very small condylar twist angle (0°) compared with the femoral valgus angle (11°). These variations appear to be attributable to a large osteophyte formation in the posterior part of the lateral or medial femoral condyle. Destructive change in the posterior part of the femoral condyle and abnormal bone formation can vary the condylar twist angle. If plain radiographs show abnormal bone formation on the posterior condyles, which would restrict knee flexion, the authors recommend preoperative CT scans to show the abnormal femoral condylar geometry, measure the reference axes, and determine the correct rotational alignment of the femoral component. These findings may support the intraoperative observation by Griffin et al ⁹ of unreliability of the posterior condylar axis. Surgeons need to be aware of many of the current instrumentation systems, which automatically provide 3° external rotation of the femoral component for all knees. There is a wide range of anatomic variations in the femoral condylar geometry of arthritic knees, and the femoral component rotation should be determined based on the each patient’s anatomy.

The current results suggest that setting the femoral component with an external rotation of 3° to 6° relative to the posterior condylar axis is appropriate in a knee with common varus or neutral alignment to set it parallel to the transepicondylar axis. In cases of severe valgus deformity with abnormally small lateral condyles or of severe hypertrophic arthritis with abnormal bone formation in the posterior condyles, a combination of the AP axis and the transepicondylar axis is a more reliable method for determining the rotational alignment of the femoral component. The most prominent point of the medial epicondyle is more likely to be available as a rotational landmark at surgery than is the sulcus of the medial epicondyle. Obtaining CT scans and measuring the condylar twist angle before surgery provides a correct external rotation angle of the component relative to the posterior condylar axis in knees with severe hypertrophic osteoarthritis or valgus deformity.