Validation of a Dynamic Joint Contracture Measuring Device in a Live Rabbit Model of Arthrofibrosis

Abstract

The current method of measuring arthrofibrosis in live rabbits is critically limited. Specifically, this method involves radioactive fluoroscopy, error-prone goniometric measurements, and static joint angle outcomes that fail to approximate the compliance of tissues surrounding the joint. This study aims to validate a novel method of capturing the compliance of contracted tissues surrounding the joint without the use of fluoroscopy or animal sacrifice. Surgically induced contractures of one-hundred and eight rabbits were measured using the current standard of contracture measurement (a pulley system) as well as a newly designed dynamic load cell (DLC) device. The DLC device was highly reliable when compared to the pulley system (r = 0.907, p < 0.001). Finally, the DLC device produced joint stiffness hysteresis curves capable of approximating the compliance of stiff joint tissues, ultimately calculating a mean joint stiffness of 1.57 ± 1.31 N · m · rad⁻¹ (range, 0.33–6.37 N · m · rad⁻¹). In conclusion, the DLC device represents a valid method for measuring joint contractures. Further, the DLC device notably improves current techniques by introducing the capacity to approximate the compliance of contracted tissues in living rabbits.

The development of a robust animal model is essential for research into the arthrofibrosis disease process.¹ Not only must this animal model approximate human joint contractures, but the outcomes measured must adequately approximate the established characteristics of arthrofibrosis. More specifically, the biomechanical outcomes measured in an animal model of arthrofibrosis should be capable of documenting the compliance of contracted tissues surrounding the joint.

The current standard for measuring joint contractures in experimental rabbits prior to sacrifice employs a pulley system to capture a joint angle at a given torque. However, this method has several limitations. First, the pulley system requires the use of fluoroscopy which necessitates bulky machinery, time-consuming rabbit positioning that is prone to error, and potentially harmful radiation exposure to laboratory staff.² Second, the use of manual goniometric measurements of the knee is prone to reproducibility error.³ Finally, and most importantly, the current method is incapable of approximating the compliance of contracted tissues surrounding the joint.⁴ The pathophysiology of arthrofibrosis involves several tissue types with nuanced contractile features. The current standard for measuring contractures, however, captures a static contracture angle—a measurement that fails to capture critical contractile features of the stiff joint. For these reasons, it is important to improve upon the current standard of measuring joint contractures in living rabbits.

This study introduces a novel dynamic load-cell (DLC) device with the potential to improve the measurement of joint contractures in living rabbits. The primary aim of this study was to validate the DLC device’s ability to obtain reliable static measurements. The second aim was to eliminate the use of fluoroscopy for joint contracture measurements in living rabbits by using the DLC device to approximate the joint contracture angle at a given torque. The third and final aim was to optimize a method characterizing the compliance of contracted tissues using the DLC device.

MATERIALS AND METHODS

Study Design

One hundred and eight skeletally mature New Zealand White female rabbits (2.5–3.5 kg) were used to induce a knee contracture as previously described.⁴ In short, after an index surgical procedure to induce a joint trauma, the operated knee was immobilized in full flexion for 8 weeks. At 8 weeks, the rabbits’ limbs were re-mobilized and the rabbits were allowed free cage activity for eight more weeks. The severity of the contracture in the operative limb was measured at sixteen weeks using a pulley device as previously described⁴ (Fig. 1A and B). Secondarily, each contracture was measured using the new DLC device (Fig. 2). The DLC device was aligned perpendicularly to the rabbit’s tibia using a goniometer. Each non-operated contralateral limb was measured as an internal control. All animal protocols were validated and approved by the Institutional Animal Care and Use Committee (IACUC), and all animals were socially housed in facilities accredited by the AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care).

Figure 1.

(A) Static pulley device. The string is attached to the tibia at 10 cm below the patella to apply a strain perpendicular to the bone (red arrow). In the other end of the string, sequentially, weights of 2–4 Newtons were applied. (B) Superior view illustrating the lateral radiograph beam (blue trapezoid) to measure flexion contracture.

Figure 2.

Dynamic load-cell device. The rabbit leg is attached 10 cm below the patella to the inferior hook of the device (1 = Load-cell sensor attached the animal limb; 2 = Sliding arm; 3 = X-ray tube, 4 = Image intensifier; 5 = Animal support).

Surgical Procedure

Each rabbit underwent two surgical procedures. The index procedure produced a severe and sustained joint contracture.⁵ The second procedure removed the Kirschner-wire (K-wire; Stryker; Minneapolis, MN) and allowed for remobilization.

The index procedure began with a lateral parapatellar arthrotomy of the knee. Next, two 3-mm cortical bone defects were created on the non-cartilaginous portions of the medial and lateral femoral condyles to replicate an intra-articular fracture with hematoma formation. The anterior cruciate ligament (ACL) and posterior cruciate ligament (PCL) were transected, and the joint was hyperextended to 45° in order to disrupt the posterior capsule. Finally, a 1.6-mm K-wire was placed through the tibia and hooked around the femur to immobilize the knee in 160° of flexion. We have previously published the ability of this technique to induce a severe joint contracture for 24 weeks.^1,4–13

The remobilization procedure was performed eight weeks after the index procedure. In this procedure, the K-wire was removed. In addition, all heterotopic ossification (HO) and excessive callus formations were removed from the extra-articular spaces. The rabbits returned to their cages for 8 weeks of free cage activity.

Timing of Joint Contracture Measurements

All joint contracture measurements were taken 16 weeks from the index surgery. Rabbits were anesthetized and both static pulley and DLC device measurements were performed for each rabbit at that time.

Static Pulley Device

A previously validated joint measuring device was used to obtain a baseline contracture measurement.⁴ This device employs a pulley system and associated fluoroscopy to measure the knee range of motion at a given torque (Fig. 1B). In order to obtain this measurement, a landmark was placed on the rabbit leg 10 cm below the patella. This landmark was used to fasten a string which was placed perpendicular to the tibia. The string was installed into the pulley device, and sequential weights of 2–4 Newtons (N) were applied to create a torque of 20, 30, and 40 Newton centimeters (N · cm). At each known torque, a fluoroscopic image was taken to measure the angle of extension at the knee (Fig. 1B). This process was repeated for the contralateral limb.

These torque values (20, 30, and 40 N · cm) were chosen based on results obtained from experimentation prior to the initiation of this study. Specifically, we utilized recently euthanized rabbits in the same joint contracture model (data not published) to determine a viscoelastic yield point which would serve as a threshold for approximating the contracture stiffness. From this information, we were able to decide how we could accurately measure joint angles without disrupting the capsular adhesion fibers of the contracture. This was necessary to ensure the natural in vivo simulation of arthrofibrosis was maintained for the duration of the study. We determined this threshold to be at approximately 80 N · cm, beyond which we began to observe lesions of the fibers and a disruption of the contracture. Thus, when accounting for inter-individual variability, the decision to utilize 20, 30, and 40 N · cm torque loads in the live rabbit would then safely provide accurate joint measurements without concern for breaching the yield point of the contracture, which would undermine the larger aims of this study if this occurred.

DLC Device

The DLC device is a manually operated and computerized device capable of recording force and displacement (Fig. 2). The sliding arm is connected to a load sensor to determine the traction torque applied to the rabbit limb. A potentiometer measures the linear displacement of the device as it manipulates the knee. To operate the device, the rabbit is laid in the supine position with the tibia attached perpendicular (10 cm below the patella) to the manually operated sliding arm. Two sets of measurements were taken. The first involved the sequential application of 20, 30, and 40 N · cm of extension torque on the knee, with lateral fluoroscopic images taken at each interval for radiographic measurement. The second set of measurements involved stretch-relaxation cycles that were applied by extending the knee with 50 N · cm of torque and relaxing to 10 N · cm. Five stretch-relaxation cycles replicated in immediate were succession. Prior to each measurement the device was re-zeroed by applying known forces to the load cell.

All fluoroscopic images were analyzed by two observers (WHT and NR) using the software ImageJ 1.50i (National Institute of Health). Flexion contracture angles were all defined as number of degrees of contracture between the axis of the femur and the tibia with 180° operating as proxy for full extension. The anatomic femoral axis was defined as the line bisecting the medullary canal (Fig. 3).

Figure 3.

The displacement of the tibia (d) between 20 and 40 N · cm is a reflection of the variation of flexion angle.

The capacity of angle prediction related to our third aim was assessed by the comparison of the measured angle of contracture, between 20 and 40 N · cm of torque, and the angle expected regarding the displacement of the device. The isosceles triangle formed, on a lateral view, by the three landmarks: condyles, ankle at 20 N · cm of traction, ankle at 40 N · cm, could be associated with the trigonometric function $\theta =2\times {\text{tan}}^{-1}\left(\frac{d}{2\times 100}\right)$, where θ is the angle calculated, d is the displacement in millimeters of the load cell, and 100 represents the distance from the knee of the landmark as described above.

To evaluate the compliance of the contracted tissues, we considered the rotational stiffness, k, given by $k=\frac{M}{\theta }$, where M is the moment applied and θ is the rotation (N · m per radian according to international system of units, that is, N · m · rad⁻¹). This parameter was measured on the contracted limbs of 24 animals. The slope of the joint stiffness curve observed during the cycles was assumed to be linear to validate the viscoelastic behavior. Then, the data were extracted for the cycles between 20 and 40 N · cm of torque using Matlab 2016a (Mathworks Inc., Natick, MA).

Statistical Analysis

Angles and displacement were measured on the operative and non-operative limbs. The data were compared in both devices. Reliability and consistency were assessed by a Student’s t-test to evaluate equality of means between the two methods. The correlation between the two techniques for each measurement was analyzed using Pearson coefficients (r). Bland and Altman plotting¹⁴ was then performed to evaluate the agreement of both techniques. This plot compared the score differences against the score means. Differences occurring with <5% likelihood of being due to chance were considered statistically significant (p < 0.05). Based on measurement of 20% of the sample, we calculated an intraclass correlation (ICC) which provided the intra-observer and inter-observer reliability (ICC was considered excellent when higher than 0.9).

RESULTS

Validity of DLC Device

The contracture angles measured fluoroscopically had an inter-observer ICC grade of “excellent” (r = 0.989, p < 0.001). Intra-observer reliability was also graded as “excellent” with r = 0.971 (p < 0.001).

The mean angle measured was 84.3 ± 17° (range, 29–149°) on the operated limb and 138.2 ± 9° (range, 110–162°) on the non-operated limb. The overall reliability of the DLC device (using the pulley system as reference) was 0.829 (p < 0.001) at 20 N · cm, 0.881 (p < 0.001) at 30 N · cm, and 0.907 (p < 0.001) at 40 N · cm (Fig. 4). This correlation was higher in the operative limb (r = 0.911 and p < 0.001; r = 0.939 and p < 0.001; r = 0.952 and p < 0.001, respectively) compared to the contralateral limb (r = 0.643 and p < 0.001; r = 0.771 and p < 0.001; and r = 0.748 and p < 0.001, respectively).

Figure 4.

Reliability of both methods of measurements per moment applied to the rabbit knees in the operated side.

The regression for the operated limb was linear with a regression coefficient of 0.940 (p < 0.001). The regression for the contralateral limb was linear as well, but with lower reliability (regression coefficient of 0.753 with p < 0.001). These results are summarized in Figure 5. The coefficient for regression was 0.607. Bland–Altman plotting was satisfactory on visual analysis for both techniques with no proportional bias (p > 0.05 in each). Yet, the contralateral limb had a higher number of measurements that appeared to be outliers, as seen in Figure 6.

Figure 5.

Measurement of both methods of measurement according to the limbs analyzed. The angle measured in operated limb (blue dots) is more consistent than the contralateral limb (green triangles) despite a similar regression line.

Figure 6.

Bland–Altman plot. The agreement in both limb is high but the contralateral limb have more outliers. Y-axis shows the difference between the two paired measurements and the X-axis represents the average of these measures.

Approximation of Joint Angle

When measured radiographically, joint angles correlated to a mean displacement of 10.2 ± 5 mm (range, 2–23 mm) when utilizing the trigonometric function $\theta =2\times {\text{tan}}^{-1}\left(\frac{d}{2\times 100}\right)$. Displacement calculated by the DLC device was higher at 21 ± 8 mm (range, 3.8–35.2 mm). The correlation for accuracy of angle prediction from the DLC device alone was r = 0.517 (p = 0.01).

Approximation of Joint Stiffness

The stretch-relaxation measurements obtained from the DLC device were plotted using a hysteresis curve (Fig. 7). This “joint stiffness” curve approximates the compliance of contracted tissues surrounding the knee. Analysis of the joint stiffness showed highly variable values among the data extracted for 108 rabbits. The mean was 1.57 ± 1.31 N · m · rad⁻¹ (range, 0.33–6.37 N · m · rad⁻¹).

Figure 7.

Hysteresis curves after stretch-shortening cycles. Red arrow is the slope between 20 and 40 N · cm, that is, a proxy for the rotational stiffness of the knee studied.

DISCUSSION

Current methods of measuring joint contractures in experimental rabbits prior to sacrifice provide only a static contracture angle that fails to approximate the compliance of stiff joint tissues. This study introduces a novel joint measuring device, the DLC device, which attempts to improve earlier methods of measuring contractures in living rabbits. Specifically, this study (i) validated the DLC device; (ii) established its ability to approximate joint contracture angles with the aid of fluoroscopy; and (iii) optimized the measurement of joint tissue compliance in live rabbits.

The DLC device showed high levels of reliability and precision when compared to a previously validated method for measuring static joint contracture angles.⁶ The regression coefficient was extremely accurate with a value of 1.005. Of note, the DLC device’s reliability improved at higher torque (40 N · cm, 0.907 compared to 20 N · cm, 0.829). This suggests that torque greater than 20 N · cm, but lower than the threshold of contracture disruption, allows for a more reliable static indicator of contracture severity. Further, this finding may relate to the deformation response of viscoelastic materials.

The DLC device more reliably measured contracture angles in the operative limb compared to the non-operative limb (regression coefficient of 0.940 compared to 0.753, respectively). This is, however, to be expected because torque on the contralateral limb introduces greater variability through increased motion at the hip. For example, low torque (20 N · cm) is often capable of fully extending the non-operated limb, causing higher levels of hip extension when compared to the contracted limb. This hip extension likely introduces variability in the joint contracture angle measured fluoroscopically with both the pulley and the DLC device. This is important because previous studies relied on the non-operated limb as internal controls. Yet, the findings in this study suggest that the non-operated limb may be an unreliable internal control when the DLC device is utilized. Future studies using this model might consider eliminating the contralateral limb as an internal control.

Most importantly, these results validate the use of the DLC device as a reliable and precise instrument for measuring joint contractures. While this does not usurp the established standard pulley model,⁶ this is, an essential first step in establishing the device’s ability to eventually eliminate the use of fluoroscopy and characterize the compliance of stiff joint tissues in living rabbits.

The secondary aim of this study was to eliminate the use of fluoroscopy for the measurement of contractures in living rabbits. The DLC device was designed with the intention of collecting computerized force and displacement data that could be approximated to a static joint contracture angle without the use of fluoroscopy.

A comparison of the limb displacement measured on fluoroscopic images taken from the pulley device and DLC device at known torque showed a correlated angle prediction of 0.517 (p = 0.01). These approximations, however, have higher levels of variability than contracture angles obtained exclusively from the pulley device and might suggest that the DLC device warrants further research to determine if the use of fluoroscopy can be completely eliminated. This may be due to the fact that the DLC device’s manually operated sliding arm must be firmly held by the technician at a given force read by the load-cell. The dynamic nature of the sliding arm creates a challenge for the technician to maintain a steady torque while simultaneously capturing a fluoroscopic image. This technique is inherently less consistent than the pulley, as the weight of the pulley is not subject to the natural tremor of a technician manually operating the sliding arm. This is certainly a limitation of the DLC device, and could be improved in future DLC models by implementing automated steps that could allow precise measurement once a given torque is reached.

The final aim of this study was to establish whether the DLC device could approximate joint stiffness in the contracted limb of a living rabbit. Previous methods were incapable of capturing the compliance of stiff tissues surrounding the joint. This indicated a critical limitation in the ability of arthrofibrosis animal models to adequately characterize joint contractures in vivo. This study found the DLC device is capable of capturing the compliance of stiff joint tissues through the application of cyclic stress on the contracture—a critical biomechanical feature of arthrofibrotic joints. This data was plotted along a hysteresis curve (Fig. 7).

The capacity of the DLC device to represent force and displacement data along a hysteresis curve is a major improvement over the previous standard which only captured a single static contracture measurement. A limited number of studies have examined the compliance of contracted periarticular tissue in live animal joint contracture models.^15,16 Limitations of these previous studies included difficulties in standardizing measurement techniques, such that a reproducible torque could not be accurately defined in live anaesthetized animals. Woo et al.¹⁷ developed an “arthrograph” device to measure the biochemical changes of periarticular connective tissue correlated to joint stiffness in live anaesthetized rabbits following prolonged immobilization. This device, however, was unable to measure maximum joint extension due to its design, and permitted only a single cycle of biomechanical testing in vivo before sacrifice. Hildebrand et al.¹⁸ designed a similar device to measure maximum knee extension in a rabbit model of knee joint contractures, yet the sequential cycling measurements produced by this device were shown to progressively decrease, indicating a stretching of the soft tissues around the joint with each repeated loading. In the present study, the ability to control dynamic cycling below a pre-determined threshold of a viscoelastic yield point (data not shown) avoids disrupting capsular adhesion fibers so that accurate repeated measurements of passive knee extension could be made. Further, earlier methods capable of capturing the compliance of stiff joints required sacrificing the rabbit and dis-articulating the diseased limb. Rabbit sacrifice is costly, and does not allow researchers to characterize the compliance of stiff tissues along a continuum of disease progression. The DLC device, however, is capable of approximating joint stiffness in living rabbits—potentially saving considerable resources. Further, the DLC device allows for the mapping of the compliance of joint tissues along a continuum for a single animal as it develops arthrofibrosis. This feature will offer previously inaccessible insight into the development of the arthrofibrosis disease process, and may provide insight into the effect of potential treatments over time.

Of note, the DLC device’s rotational stiffness measurements varied among the cohort. This can likely be attributed to the diversity of contracture severity seen in this rabbit cohort as measured with the validated pulley device. This is intuitive, as each rabbit will respond to the index surgery in a different and individualized manner. To our knowledge, the rotational stiffness has not been previously described for the biomechanical features of arthrofibrosis. This new parameter defines inter-individual variability. It emphasizes and characterizes the individual inflammatory response involved in trauma healing. Yet, the device is not without functional limitations, specifically the manual nature of the sliding arm. The use of a motorized sliding arm could improve this measurement, and future directions might focus on the development of a motorized sliding arm. Further, it is possible that motion of the hip during DLC device’s stress cycling introduced variability in the measurements. Free motion was given to the hip during these measurements in an effort to minimize contracture disruption by applying potentially unknown torque (from an un-validated device). It is important to note that motion of the hip is an inherent limitation of both the pulley and DLC device. Further, the hip was given free motion for the measurements of both devices, functionally standardizing the potential bias of hip motion across measurements. Finally, as the results from this study validate the use of the DLC device, hip stabilization is justified in future studies.

This study has several limitations. The use of a limited measurement technique (the pulley system) to validate a new device risks a circular limitation. Yet, the pulley system was the only previously validated device for measuring joint contractures in living rabbits, and as such was the only available means of validating an improved device. Additionally, this study does not incorporate previously out-lined data widely used to characterize joint contractures in this animal model—specifically the use of histology, molecular analysis, and dynamic stiffness of sacrificed rabbit limbs. The use of these additional parameters would bolster the validation of the DLC device, and is a limitation of the present study. However, the decision not to incorporate these data arose from the necessity to keep the rabbits alive in order to more deeply analyze disease over an extended time-course beyond the purview of this experiment.

The current method for measuring joint contractures in living rabbits is unable to approximate compliance of contracted tissues surrounding the joint, and is therefore critically limited. Further, the current method necessitates the use of bulky, inefficient, and potentially dangerous fluoroscopy. This study attempted to validate a new device capable of better characterizing joint stiffness in live animals, while also eliminating the use of fluoroscopy when collecting this biomechanical data.

This study demonstrated that this novel device, the DLC, is a valid and reliable instrument for measuring joint contractures. The DLC device alone does not completely eliminate the need for fluoroscopy in the collection of this biomechanical outcome, yet ongoing device engineering aims to improve the efficiency of this animal model. Finally, the DLC device is capable of approximating the compliance of contracted tissues surrounding the living rabbit knee. For these reasons, the DLC device represents a substantial improvement on the ability to accurately and efficiently characterize joint contractures in vivo.