Rationale
The current study was undertaken to evaluate the effects of abnormal tibial torsion and corrective rotational osteotomies in a group of subjects without notable additional neurological, muscular, or osseous problems. Specifically, this research tested three hypotheses. Hypothesis 1 was that subjects with abnormal tibial torsion exhibit abnormal knee and ankle kinetics, Hypothesis 2 was that distal tibial rotational osteotomies would significantly improve ankle and knee kinetics in these subjects, and Hypothesis 3 was that these improvements in joint kinetics would be reflected in overall gait kinematics. Furthermore, functional correlations between the foot progression angle and two key kinetic parameters, ankle power and frontal knee moment, were investigated in the preoperative condition.
Participants
The present multicenter cohort study was approved by local institutional review boards and began as a prospective design to determine the changes in ankle and knee kinetics secondary to isolated distal tibial rotational osteotomy in children and young adults. In order to be considered for inclusion, an individual had to have a diagnosis of idiopathic tibial torsion causing pain or activity limitations that required surgical rotational correction of at least 20° to normalize the tibial torsion and had to be six to twenty years of age. A subject was excluded if he or she had a diagnosis other than idiopathic tibial torsion or exhibited excessive femoral retroversion, genu varum or valgum, or any other notable lower extremity deformities. Two movement-analysis laboratories participated in subject recruitment and data collection. All subjects were initially assessed by an experienced orthopaedic surgeon and were determined to have excessive tibial torsion with functional difficulties such as pain and activity limitation.
A sample of convenience was used. Over a three-year period, fifteen subjects were enrolled after providing informed consent; however, only eleven completed the study. Due to this small number, seven additional subjects meeting inclusion criteria were added retrospectively with institutional review board approval. The earliest included data were from March 2001. Of the eighteen final participants, eleven had unilateral rotational osteotomy and seven had bilateral osteotomy, yielding twenty-five limbs in the study group. Participants were divided into two groups, with eight subjects (eleven limbs) having excessive inward (often termed internal) tibial torsion and ten subjects (fourteen limbs) having excessive outward (often termed external) tibial torsion (Table I). Of note, fourteen (78%) of the eighteen participants were obese (>85th percentile) according to their age and body-mass index, and one participant was underdeveloped (<5th percentile). All children walked without assistive devices (Gross Motor Functional Classification System Level I)14.
Surgery
To correct the excessive tibial torsion, all subjects underwent distal tibial rotational osteotomy without fibular osteotomy. The osteotomies were performed at the supramalleolar level, with either a compression plate or crossed percutaneous smooth Steinmann pins and a cast being used for fixation (for four weeks). Parallel smooth Steinmann pins were inserted above and below the osteotomy level. Once the ankle and distal fragment were rotated, a goniometer was used to measure the amount of correction. Subjects were restricted to toe-touch weight-bearing with crutches (unilateral) or standing for wheelchair transfers (bilateral) for four weeks postoperatively. The mean reported operative correction (and standard deviation) was 25° ± 5° (range, 15° to 30°) for the inward tibial torsion group and 30° ± 7.5° (range, 20° to 40°, estimated to the nearest 2.5°) for the outward tibial torsion group.
Data Collection
Standard gait-analysis techniques were employed both preoperatively and postoperatively (Table I) with use of a modified Newington Children's marker set15. The research personnel at the two laboratories participating in the present study underwent similar standardized training in physical examination measurement and model application techniques, had similar hardware (MX and 612; Vicon, Centennial, Colorado) and software (Workstation, Nexus, and Plug-in-Gait; Vicon, Centennial, Colorado), and used the same data-processing techniques. On the basis of the biomechanics of the torsional deformity, data analysis focused on sagittal plane ankle and frontal plane knee kinetics. These data were correlated with the foot progression angle.
Variables
Torsional correction was assessed by evaluating the mean foot progression angle in stance as well as by means of physical examination and marker-based measures of tibial torsion. Tibial torsion during physical examination was evaluated by means of goniometric measurement of the transmalleolar axis with respect to the knee axis. Marker-based torsion was determined with use of markers that were placed on the medial and lateral malleoli and a knee alignment device. Tibial torsion was determined as the difference between the knee flexion axis and the transmalleolar axis projected onto the transverse plane. To test Hypotheses 1 and 2, peak ankle power generation and the values of frontal knee moment were evaluated at the times of both contralateral foot-off and contralateral foot contact. The values at the times of contralateral foot events closely correspond to the two normal peaks of the gait cycle occurring during loading and push-off. Peak knee extension in stance, maximum dorsiflexion in stance, and the Gait Deviation Index16 were employed to measure gait kinematics as specified in Hypothesis 3.
Statistical Analysis
To test Hypothesis 1, both the inward and outward torsion groups were compared with a similarly aged group of seventeen typically developing subjects (six girls and eleven boys ranging from nine to seventeen years of age) (total, thirty-four sides) with use of unpaired Student t tests. These individuals were previously recruited from relatives and friends of hospital employees at one site and were evaluated during an institutional review board-approved protocol to collect normative gait data. To test Hypothesis 2, groups were compared longitudinally with use of paired Student t tests to assess the effects of surgery on the key variables. Additionally, Hypothesis 2 was tested by considering whether improvement resulted in normalization of variables by means of comparison with the typically developing group with use of unpaired Student t tests. To test Hypothesis 3, both paired Student t tests (to analyze surgical changes) and unpaired Student t tests (to compare both preoperative and postoperative data with those for typically developing children) were performed. Regression analyses were undertaken to investigate the relationships between key variables; groups were pooled and plotted relative to differences from the mean control values to determine quality of fit and significance by means of analysis of variance. Additionally, for key variables, the effect of pooling prospective and retrospective data was analyzed with use of repeated-measures analysis of variance and evaluating between-subject effects for prospective versus retrospective data.
Source of Funding
No external funding was used for this study.
Comprehensive results for all variables examined are listed in Table II (with corresponding p values provided in the Appendix), with selected key variables (foot progression angle, ankle power, frontal knee moment at the time of contralateral foot contact, and gait deviation index) plotted in Figure 1. Statistical evaluation indicated that the pooling of retrospective and prospective data in the present study was acceptable.
Tibial Torsion and Foot Progression Angle
The mean intraoperative correction (and standard deviation) as measured with goniometry was 25° ± 5° for the inward tibial torsion group and 30° ± 7.5° for the outward tibial torsion group. Both inward and outward torsion groups had mean tibial torsion measures (based on both physical examination and marker measures) that improved significantly following rotational osteotomy. The outward torsion group included four subjects, all with unilateral involvement, in whom the initial foot progression angle was within one standard deviation of normal (—10.8° ± 6.4°), indicating transverse plane compensations at the hip and/or pelvis. For both the inward and outward torsion groups, the mean foot progression angle in stance was significantly different from normal (p < 0.001). Postoperatively, the angle for the outward torsion group improved by 13.3° ± 8.6° and was no longer significantly different from normal (p = 0.35) whereas that for the inward torsion group improved by 15.8° ± 8.8° but remained significantly different from normal (p < 0.001). Neither physical examination-based nor marker-based torsion was significantly different from normal according to unpaired Student t tests.
Ankle Power
Preoperative maximum ankle power was significantly reduced in the inward torsion group (p = 0.003) but was not different from normal in the outward torsion group. With surgical correction, the maximum ankle power for the inward torsion group significantly improved (p = 0.016) and returned to within a normal range. Further statistical evaluation with adjustment for the effects of body-mass index did not change the findings attained from paired t tests. Preoperative ankle power was explored as a function of foot progression angle to investigate the concept of lever arm dysfunction17,18. It was anticipated that power would decrease with either increased inward or outward torsion from the norm. Thus, to enable grouping of subjects from the inward and outward torsion groups, deviation from the mean foot progression angle in stance for the typically developing group (10.2°) was used (Fig. 2). A linear fit (R2 = 0.25) demonstrated a significant value of the slope (p < 0.001) on analysis of variance.
Frontal Knee Moment
Internal frontal knee moments or loading peaks at contralateral foot-off were not different from those in typically developing subjects and were not significantly altered with surgery in either torsion group. Internal frontal knee moments at the time of contralateral foot contact, or push-off peaks, however, were significantly different preoperatively from those for typically developing subjects (p < 0.001 for both groups) and were significantly altered by surgery (p < 0.001 for both groups) such that postoperative values were not different from those for typically developing subjects. Further statistical evaluation with adjustment for the effects of body-mass index did not change the findings attained from paired t tests. The correlation between foot progression angle and internal frontal knee moment at the time of contralateral foot contact was explored by combining all groups (Fig. 3). A linear fit (R2 = 0.53) demonstrated a significant value of the slope (p < 0.001) on analysis of variance.
Gait Kinematics
The overall gait kinematics as measured with the Gait Deviation Index were significantly below normal in both torsion groups preoperatively (p = 0.002 for the inward torsion group and p < 0.001 for the outward torsion group). With rotational osteotomy, both groups had significant improvement (p < 0.001). The mean value for the inward torsion group (87) remained less than one standard deviation below normal, whereas the mean value for the outward torsion group (91) improved to within normal bounds. However, both groups remained significantly different from typically developing subjects (p = 0.006 for the inward torsion group and p = 0.004 for the outward torsion group). With regard to both peak knee extension and maximum dorsiflexion in stance, both groups had preoperative values within the normal range, which did not change significantly following surgery.
All three hypotheses were supported by the data. With regard to Hypothesis 1, distal rotational osteotomies significantly improved, but did not fully correct, deviations in the foot progression angle. With regard to Hypothesis 2, while lever arm dysfunction was not apparent in the outward tibial torsion group as a whole, it was evident in the inward torsion group and was corrected with rotational osteotomy. Significant deviations in frontal knee moment were evident in both groups and were corrected with surgery. Significant, moderately correlated relationships between foot progression angle and both ankle power and frontal knee moment were determined. With regard to Hypothesis 3, deviations in the overall gait kinematics were evident in both groups and were improved with surgery but remained different from normal.
Body-Mass Index
When compared with an age-matched group of typically developing subjects, the idiopathic tibial torsion group had significantly higher body-mass index. Most of the effects of increased weight are accounted for in the results as all force measures (moments and powers) are normalized by body mass. Some compensatory mechanisms such as wide-based gait, however, may confound the results19.
Foot Progression Angle
Rotational osteotomies significantly improved both static measures of tibial torsion and the foot progression angle. Correcting stance phase foot progression angle is the principal aim of tibial derotation surgery, so this finding of significant improvement is not surprising. Although the foot progression angle for the outward torsion group corrected such that it was not significantly different from that for typically developing subjects, the inward torsion group, despite a 16° improvement, continued to demonstrate significant inward rotation in comparison with typically developing subjects (p < 0.001). One interpretation may be that surgeons should be more aggressive with the degree of surgical correction and should aim for overcorrection. An argument against this interpretation is that frontal knee moment corrected, and indeed exhibited mean overcorrection, with the surgical corrections performed in both of these groups.
Copers and Noncopers
Several subjects in the present study exhibited normal kinematic and/or kinetic parameters preoperatively, indicating that they were able to compensate for their deformity. Four subjects (four sides; 16%), all with unilateral outward tibial torsion, exhibited a normal foot progression angle prior to surgery. Three of these subjects exhibited normal kinematic patterns according to the Gait Deviation Index but complained of knee pain that limited their activity. Tibial torsion in these subjects was confirmed with CT. All subjects reported decreased pain or resolution of pain after surgery. Two of these subjects exhibited an abnormally elevated valgus frontal knee moment when the opposite effect would be expected. Thus, compensation strategies for maintaining a normal foot progression angle deleteriously and unexpectedly changed the forces at the knee. Five subjects with inward torsion and one with outward torsion (six sides; 24%) exhibited normal frontal knee moments. Eleven subjects (fourteen sides; 56%) exhibited normal ankle power. This finding suggests that frontal knee moments are more sensitive to torsional changes than ankle power is.
Lever Arm Dysfunction
Decreased ankle power of one standard deviation or more from typically developing individuals was observed in just five sides (36%) in the outward torsion group and six sides (55%) in the inward torsion group. All subjects but one exhibited improved ankle power with surgery, with four of the five sides in the outward torsion group and four of the six sides in the inward torsion group returning to within normal limits. Thus, in the majority of subjects, excessive torsion did not significantly reduce peak ankle power. In the subjects in whom it did, surgery returned these values to normal in 73% (eight) of eleven limbs. A regression analysis was undertaken to further study the relationships between ankle power and foot progression angle (Fig. 2). A significant relationship was determined, with the foot progression angle accounting for 25% of the variance. According to the regression, 30° of abnormal torsion is required to reduce ankle power below one standard deviation and >45° of abnormal torsion is required to reduce ankle power below two standard deviations from the norm.
Frontal Knee Moments
The normal pattern of frontal knee moment in stance results in two distinct peaks during the gait cycle, one with weight acceptance and one with push-off. Net internal moments are reported; thus, an inwardly rotated foot producing a medially offset ground-reaction force would be expected to push the knee into varus, requiring an increased internal valgus moment to balance. The pattern of frontal knee moments in our subjects was notably different from typically developing subjects as the inwardly rotated group exhibited a normal weight acceptance peak and increased valgus moment during the push-off peak and the outwardly rotated group exhibited a normal weight acceptance peak but a diminished valgus or relative varus push-off peak. Clinically, these occurrences are often noted visually with associated frontal plane motions of the knee as either valgus or varus thrust. Frontal knee moments were corrected to within normal limits with surgery. The relationship between foot progression angle and frontal knee moment was more sensitive in the outward torsion group. According to the linear relationship (Fig. 3), 10° of increased outward torsion reduced valgus moments to below one standard deviation from normal and 25° reduced them to below two standard deviations. The inward torsion group required 25° of excessive torsion to fall one standard deviation from normal and 40° to reach two standard deviations. In addition, more subjects in the inward torsion group demonstrated normal values, indicating better ability to cope with the deformity.
Kinematics
Isolated tibial torsion can lead to a variety of secondary compensatory or induced kinematic abnormalities during gait. These may be subtle and multilevel in nature, may be variable across subjects, and may include coupled pelvic rotation and hip abduction/adduction; hip, knee, and ankle transverse rotations; and contralateral limb compensation. In the present study, the primary kinematic parameter (foot progression angle) and the primary kinetic parameters (ankle power and frontal knee moment) were studied on the basis of the biomechanics of the deformity and goals of the surgery. To assess overall kinematic deficiencies and changes subsequent to surgery, a gait index that evaluates profiles of gait kinematics was used. According to the Gait Deviation Index, both groups were initially more than two standard deviations from normal, with the index for the inward torsion group (70) being slightly worse than that for the outward torsion group (75). With rotational osteotomy, both groups improved 17 points on the average, placing the index for the outward torsion group (91) within one standard deviation of normal and the index for the inward torsion group (87) within one and one half standard deviations of normal.
All subjects but one demonstrated improved gait kinematics postoperatively. The one subject who was the exception had a 4.4-point decrease in gait function. This subject initially presented with inward foot progression of 19.5°, which only improved to 15° subsequent to surgery. The primary changes in the subject's gait pattern were increased pelvic tilt and coupled hip flexion. This subject's body-mass index increased from 31.4 kg/m2 preoperatively to 36.4 kg/m2 postoperatively, which were both the largest values and the largest increase in the study, and the data may reflect the difficulty inherent with marker placement on extremely obese subjects. Increased body-mass index in the surgical group compared with the typically developing group may also account for the persistent significant differences that were found when the postoperative gait indices were compared with the values in typically developing subjects.
Comparisons with the Literature
In the present study, according to regression analysis, differences from typically developing individuals of one standard deviation required large amounts of torsion (at least 10° for frontal knee moments and 30° for ankle power). In contrast, Turner12 determined correlations of tibial torsion values of just 6° with knee disorders in adults. Over time, even small abnormal moments on the torsioned knee may have deleterious effects, highlighting the need for early intervention. In the myelomeningocele population, both Lim et al.6 and Ounpuu et al.7 found abnormal knee moments in groups of patients with tibial torsion. Lim et al.6 indicated that excessive outward tibial torsion of 20° would result in knee moment values with adverse effects, but ankle power was not examined. Ounpuu et al.7 correlated adverse knee moment with trunk motion, which is excessive in children with spina bifida. This emphasizes the advantage of examining torsional relationships in the idiopathic population as torsional effects can be isolated from muscle weakness, contracture, spasticity, and other problems.
Limitations
The current study had several inherent limitations. First, it included a mix of retrospective and prospective data. Second, bilateral data from the same subjects were treated independently, although changes in one side may affect the other. Third, because of recognized measurement uncertainty20, no attempt was made to correlate actual torsion measures with kinematic or kinetic variables. Instead, correlations were made to the foot progression angle, which can be compensated for by subjects. Finally, a small number of subjects were available for each group; thus, it was not possible to control for sex, unilateral versus bilateral involvement, or other factors.
Overview
Distal tibial rotational osteotomy effectively restored ankle power in patients with increased inward tibial torsion and effectively restored frontal knee moments in patients with both inward and outward tibial torsion. While gait kinematics significantly improved with surgery, complete normalization was not achieved. The concept that lever arm dysfunction reduces ankle power, while biomechanically sound, may only practically be appreciated in patients with extreme levels of excess torsion. Conversely, frontal knee moments are more sensitive to tibial rotational deformities and were more highly correlated with the foot progression angle.