Abstract
Post-traumatic joint contracture (PTJC) is a debilitating condition, particularly in the elbow. Previously, we established an animal model of elbow PTJC quantifying passive postmortem joint mechanics and histological changes temporally. These results showed persistent motion loss similar to what is experienced in humans. Functional assessment of PTJC in our model was not previously considered; however, these measures would provide a clinically relevant measure and would further validate our model by demonstrating persistently altered joint function. To this end, a custom bilateral grip strength device was developed, and a recently established open-source gait analysis system was used to quantify forelimb function in our unilateral injury model. In vivo joint function was shown to be altered long-term and never fully recover. Specifically, forelimb strength in the injured limbs showed persistent deficits at all time points; additionally, gait patterns remained imbalanced and asymmetric throughout the study (although a few gait parameters did return to near normal levels). A quantitative understanding of these longitudinal, functional disabilities further strengthens the clinical relevance of our rat PTJC model enabling assessment of the effectiveness of future interventions aimed at reducing or preventing PTJC.
Introduction
Post-traumatic joint contracture (PTJC) affects up to 50% of patients following joint dislocation or fracture, often leading to permanent stiffness and debilitating motion loss [1,2]. The elbow is particularly susceptible to contracture following traumatic injury due, in part, to its high degree of natural congruency [3,4]. In order to help improve range of motion (ROM), nearly 12–15% of patients with PTJC in the elbow require surgery to release contracted tissue [5]. While the clinical relevance and severity of contracture is compelling, the elbow remains one of the most understudied joints in the human body.
Previously, we established an animal model of elbow PTJC quantifying passive postmortem joint mechanics and histological changes temporally [6–9]. These results showed persistent motion loss similar to what is experienced in humans. Although previous results demonstrated altered joint mechanics, data were obtained postmortem and did not consider functional use of the injured/healing limb. Evaluation of the in vivo functional consequences of PTJC would provide a clinically relevant measure of joint function. Such measurements could further validate our animal model if joint function were shown to be persistently altered similar to human patients. Thus motivated, this study developed and utilized two experimental techniques to quantify forelimb function in our rat model of PTJC: grip strength assessment and spatiotemporal gait analysis.
Grip strength has been used previously to assess functional limb differences in rodents subjected to various injuries and conditions, including ankle joint inflammation [10], upper extremity muscle and tendon overuse [11], and systemic collagen-V deletion in tendons/ligaments [12], as well as to determine forearm muscle strength in mice with motor neuron disease [13], and healthy mice aging naturally [14]. For this study, a novel dual limb grip strength device was developed, permitting longitudinal evaluations to specifically assess how PTJC affects side-specific forelimb use and strength.
Similar to grip strength, gait analysis has been used previously to quantify functional changes in rodent models of a variety of injury conditions including knee instability and osteoarthritis [15–17], patellar tendon degeneration [18], Achilles tendon injury [19], rotator cuff damage [20], and spinal cord injury [21]. Additionally, a recent study evaluated gait changes due to elbow contracture in a murine model [22]; however, in this study animals were only evaluated at a single time point for a single gait parameter. In this study, spatiotemporal gait compensations were investigated longitudinally in our rat PTJC model using an open source, fully automated technique validated to detect rodent gait abnormalities with high sensitivity [23,24]. We hypothesized (1) the unilateral, persistent range of motion loss observed previously [7] would be manifested as an asymmetric gait with imbalanced duty factor and reduced grip strength when compared to uninjured controls, and (2) metrics of gait and grip strength would partially recover over time but not reach levels of control animals.
Materials and Methods
Animal and Injury Model.
Long-Evans rats (Charles River Laboratories International, Wilmington, MA) were selected based on previously described criteria [6]. These animals exhibit similarities to human elbow functional ROM, bony architecture, and periarticular soft tissues providing the ability to flex-extend and pronate–supinate at the elbow [25–28].
In this Institutional Animal Care and Use Committee approved study, male Long-Evans rats (n = 20; 330–380 g; 9–10 weeks old) were randomized into a surgically induced injury group and uninjured control group (Fig. 1). To simulate damage that occurs clinically in elbow dislocation, animals were anesthetized (n = 10), and a randomized elbow (n = 5 each for left and right injury) was subjected to a unilateral anterior capsulotomy and lateral collateral ligament transection [6]. Due to surgery complications, one animal (injury in left limb) was excluded, leaving nine injured animals. Single doses of antibiotic (7.5 mg/kg enrofloxacin; Bayer Health LLC, Shawnee Mission, KS) and nonsteroidal anti-inflammatory drug (5 mg/kg carprofen; Pfizer Animal Health, New York, NY) were administered pre-operatively by subcutaneous injection, and a single dose of analgesic (0.5 cc of 5 mg/mL bupivacaine; Hospira, Lake Forest, IL) was administered postoperatively under the closed incision via subcutaneous injection.
Fig. 1.
Analysis time points for control and injured groups in this study. Following an acclimation period to the grip strength device and gait arena, data were collected weekly for both groups (time is in days).
Following surgery, injured limbs were immobilized for 42 days using tubular elastic netting (Nich Marketers, Inc., Gulf Breeze, FL) and self-adhering Vetrap bandaging (3 M, St. Paul, MN). Contralateral limbs were not constrained and allowed unrestricted movement. Animals were checked five times per week to ensure injured limbs were immobilized and to identify any signs of discomfort or distress. Bandages were replaced weekly and any sores or cuts were treated topically with antibiotic powder/cream (nitrofurazone (Neogen Corporation, Lexington, KY); silver sulfadiazine (Dr. Reddy's Laboratories, Shreveport, LA)) and/or chafing cream (Prestige Brands, Tarrytown, NY). After 42 days of immobilization, bandages were removed and animals were allowed unrestricted cage activity for an additional 42 days, as previously described [6]. Age-matched control animals (n = 10) were neither injured nor immobilized and were allowed unrestricted cage activity for the entire 84-day period.
Grip Strength Testing.
A custom grip strength device was designed and built to allow for the simultaneous measurement of left and right limb grip strength, which is necessary to track side-specific recovery over time in our unilateral injury model (Fig. 2). Unlike typical devices for measuring grip strength in rodents [13,22,29], our device consists of two sets of custom-made grip bars/ladders, each of which is secured via linear slides and connected to separate load cells (Loadstar Sensors, Fremont, CA). During the first week after bandage removal, animals were familiarized to the grip strength device by performing six practice trials. After gently placing their front paws on matching grip bars of each ladder, animals were horizontally pulled by the base of their tails in a direction away from the grip strength device until they released their paws. Upon paw release, animals were immediately cradled underneath, never allowing a fall of more than a few centimeters. Grip strength measurements were collected weekly beginning one week after bandage removal on injured animals (n = 9) and equivalent time points on controls animals (n = 10; Fig. 1). In an attempt to reduce environmental stressors that could impact repeatability of results, data collection was completed by the same handlers in the same environments at similar times of the days. Maximum left- and right-limb grip strength measurements were recorded simultaneously for six trials per animal at each time point. Any grip strength values of the six trials that fell outside of two standard deviations of the mean for that animal and time point were excluded. In total, one trial was excluded from an injured animal at day 49 and one trial from a control animal at day 56. One control animal refused to grip the bars during testing at day 56, 63, 70, and 84; all grip strength data from this animal were excluded, thus reducing the control group to nine animals.
Fig. 2.
The custom-built grip strength device allowed for simultaneous bilateral grip measurements. The device consists of two sets of grip bars/ladders, each held in place by a linear slide and connected to separate load cells. The front paws of the animals were placed on the same grip bar of each ladder and the animals were pulled by the base of their tails horizontally (parallel to the ladders and linear slides). Scale bar = 1 cm.
Spatiotemporal Gait Testing.
A custom-built gait arena with associated hardware and open-source analysis software were used to acquire spatiotemporal gait data [17,23,24] (additional information available online1). Briefly, the arena is an enclosed, transparent acrylic box (60 in. long × 5 in. wide × 10 in. tall) with colored vinyl covering the lid and back wall Fig. 3. The arena sits on a transparent base above a 45 deg-angled mirror so that images of the sagittal and ventral plane can be recorded simultaneously. The width of the arena is sufficiently narrow to force a walking direction perpendicular to the camera. The arena is lit with two 20,000 lumen light bars covered with white fabric to diffuse the light. During the first week after bandage removal, animals were acclimated to the gait arena by placing them in the arena for ∼20 min on three sequential days. Gait data were collected weekly beginning one week after bandage removal on injured animals (n = 9) and equivalent time points on controls animals (n = 10; Fig. 1). During gait collection, animals voluntarily explored (i.e., were not forced or prompted) the arena at self-selected velocities while being recorded with a high-speed video camera (FASTCAM Mini UX50, Photron, Tokyo, Japan) at 500 frames per second (fps). Utilizing similar exclusion criteria as previous studies using this techniques [15,23], trials with constant walking speed and containing at least three complete gait cycles were collected on injured and control animals. Based on a previous work, a minimum of five trials (collected on at least eight animals at 500 fps) were deemed necessary to ensure sufficient sensitivity to detect gait changes of 1% in symmetry and duty factor and 0.25 cm in stride length [23]. Because of postprocessing exclusion criteria (color thresholding errors or varying velocity), attempts were made to collect in excess of 10 trials per animal per time point. In total, after all exclusion criteria were met, gait data included n = 5–11 trials per animal per time point. Because of the voluntary nature of the gait collection and limits of time available for acquisition, only three trials were collected on one control animal at day 84 (n = 3; data were included). Gait data could not be collected on one injured animal at day 49 because walking occurred without placing the injured paw on the ground; data for the remaining time points of this animal were included.
Fig. 3.
(a) The gait collection setup included a custom-built gait arena above an enclosed mirror. A high-speed video camera recorded simultaneous views of the sagittal and ventral plane. (b) During gait collection, animals voluntarily explored the arena lit by two light bars. Trials with constant walking speed and containing at least three complete gait cycles were collected on controls and injured groups.
Spatiotemporal gait parameters were calculated using an open source, fully automated analysis software that has been validated against manual digitization at frames rates at or above 125 fps [23,24]. Briefly, a setup script was run to define several parameters including video frame rate, video region of interest, and display options. A standard matlab (MathWorks, Natick, MA) function (colorThresholder, MathWorks, Natick, MA) was used to create color filters to isolate the animal body in both planes and the paws in the ventral plane. Videos from a single day of collection (∼50 videos) were then batch processed, with spatial and temporal gait parameters automatically calculated. The computed parameters included forelimb duty factor (i.e., percent stance time), duty factor imbalance, temporal symmetry, spatial symmetry, and stride length for each video trial; these spatiotemporal gait parameters have been previously defined and extensively reviewed [30,31].
Mechanical Testing.
All animals were euthanized by CO2 inhalation and immediately stored in a −20 °C freezer. For each animal in the injured (injured limbs: n = 9; contralateral limbs: n = 9) and control (n = 20 limbs) groups, mechanical testing was performed on both forelimbs. Forelimbs were prepared postmortem and mechanically tested, as previously described [6,7,9]. Briefly, a custom mechanical testing system was used to evaluate limbs in flexion–extension joint testing for five cycles to ±0.75 N (±11.25 N·mm of torque) at 0.3 mm/s. A rack and pinion gear converted the linear displacement to angular rotation about the elbow joint center, and overhead images were taken at maximum flexion and extension. Force and displacement data were converted to torque and angular position, and elbow motion was quantified using a custom-written matlab program. Measurements included maximum extension, maximum flexion, total ROM, and neutral zone (NZ) length. The NZ length represents an estimate of the functional ROM and is defined by the flatter region between maximum extension and flexion on the torque–angular position curve [6]. Angles were computed in terms of degrees of flexion with a horizontal line from the humerus representing 0 deg flexion (or full extension).
Statistical Analysis
Grip Strength Testing.
Linear mixed models were developed using random effects to account for the temporal and intersubject correlation in the data to appropriately evaluate the fixed effects associated with the hypotheses relevant to our research. Model fitting was done in JMP (SAS Institute, Cary, NC). For grip strength data, random effects included individual animals and repeated measures, while fixed effects included treatment group, time point, weight, and subsequent interactions. Weight was controlled for because injured animals were lighter than control animals at similar time points and because the analysis needed to avoid confounding effects of weight gain across the study (control: 545.2±25.3 g (day 49) to 605.9±24.8 g (day 84); injured: 443.6±16.3 g (day 49) to 560.1±26.1 g (day 84)). Rather than using weight as a covariate directly, a linear model was fit to adjust weight for treatment group and time point. The studentized residuals from that model were used in the final linear mixed model. These residuals are interpreted as the excess weight of a subject given a treatment group and time point. This procedure allowed for the assessment of the impact of weight independent of time and treatment group effects. The interaction of weight, treatment group, and time was hypothesized to be significant because the impact of weight across time might depend on treatment group. This third-order interaction was built into the model requiring all lower order interactions to be included. For consistency, the same model was used for all the three grip strength comparisons.
In total, three separate linear mixed models were developed to compare (1) injured limb grip strength to the left and right control limb intratrial average grip strength, (2) contralateral limb grip strength to the left and right control limb intratrial average grip strength, and (3) injured-to-contralateral limb grip strength difference for injured animals to left-to-right difference for control animals. Model adjusted outcomes for an animal of average weight (excess weight = 0) were calculated and are presented as 95% confidence intervals (95% C.I.). Pairwise comparisons between groups were performed for model outcomes at each time point to identify statistically significant differences (p < 0.05).
Spatiotemporal Gait Testing.
Similar to grip strength data, linear mixed models were developed to account for random and fixed effects in spatiotemporal gait data in JMP. For the gait data, random effects included individual animals and repeated measures, while fixed effects included treatment group, time point, weight, velocity, and subsequent interactions. The majority of gait parameters are known to depend on velocity and must be controlled for when performing gait analysis [30,32]. Similar to how weight was considered in grip strength analysis, velocity was adjusted in the gait data based on treatment group and time point. Residual velocity is interpreted as excess velocity of a subject for a given treatment group and time point. This allowed for the evaluation of the impact of velocity independent of treatment group and time point. The interaction of velocity, treatment, and time was hypothesized to be significant. Injured rats were expected to walk slower than uninjured control rats and velocity was assumed to change over time because of recovery (in the injured group) and familiarity with the testing setup (in both groups). For this reason, this interaction (i.e., velocity, treatment, and time) and all lower-order interactions were built into the model. The interaction of weight, treatment, and time was hypothesized to be significant for reasons similar to grip strength data. For consistency, the same model was used for all the six gait comparisons. Nonsignificant effects remained in the model to preserve its comprehensiveness since computational demand was negligible. Similar to the grip strength analysis, studentized residual weights and velocities were used in the final linear mixed models rather than actual animal weights and velocities.
In total, six separate linear mixed models were developed: two to compare injured and contralateral forelimb duty factor to the average forelimb duty factor in control animals, and four to compare forelimb duty factor imbalance, temporal symmetry, spatial symmetry, and stride length in injured animals to control animals. Model adjusted outcomes for an animal of average weight and velocity were calculated and are presented using 95% C.I. Pairwise comparisons were performed between groups for model outcomes at each time point to identify significant differences (p < 0.05).
Mechanical Testing.
Mechanical testing parameters (e.g., max extension, max flexion, total ROM, and NZ length) for right control limbs were subtracted from paired left limbs, while contralateral limbs were subtracted from paired injured limbs. One sample t-tests compared left-to-right or injured-to-contralateral differences to a theoretical mean of zero (μ0 = 0). Unpaired Welch's unequal variances t-tests compared the control animal differences to injured animal differences. Significance was defined as p < 0.05 for all statistical tests.
Results
Grip Strength Testing.
Fit results of the linear mixed models for grip strength data are represented by the summary of fit, significance of fixed effects, and restricted maximum likelihood (REML) variance (Table 1). In general, grip strength models fit the data well with all fixed effects being significant in at least one of the models, except for the interaction of studentized residual weight crossed with treatment group. Random effects in the injured and contralateral limb grip strength models explained an approximately equal percentage of the REML variance. In the grip strength difference model, repeated trials (subsample) accounted for approximately twice the REML variance as compared to individual animals (rat ID).
Table 1.
Grip strength linear mixed model summary of fit, significance of fixed effects, and REML variance
| INJ limb grip strength | CL limb grip strength | Grip strength difference | ||
| R2 | 0.812 | 0.561 | 0.655 | |
| Summary of fit | RMSE | 82.7 | 86.7 | 111.9 |
| Source | p-value | |||
| Fixed effects | Time | <0.0001 | <0.0001 | <0.0001 |
| Treatment group | <0.0001 | 0.1612 | <0.0001 | |
| Time × Treatment group | <0.0001 | <0.0001 | <0.0001 | |
| Stud Res Wt | 0.0104 | 0.0292 | 0.9979 | |
| Stud Res Wt × Time | 0.0003 | 0.0006 | 0.1069 | |
| Stud Res Wt × Treatment group | 0.1643 | 0.0750 | 0.7390 | |
| Stud Res Wt × Treatment group × Time | 0.1644 | 0.3392 | 0.0435 | |
| Random effect | Variance component (percent of total) | |||
| REML variance | Rat ID | 6642.4 (49.3) | 5812.7 (43.6) | 6882.3 (35.5) |
| Subsample | 6836.2 (50.7) | 7525.0 (56.4) | 12,527.7 (64.5) | |
Note: INJ/CL limb grip strength = model details for injured/contralateral limb compared to control limbs average; grip strength difference = model details for limb differences in injured animals compared to limb differences in control animals; time = time point (day 49, 56, … 84); treatment group = control or injured; Stud Res Wt = studentized residual weight; Rat ID = individual animal identification; subsample = repeated measurements on an animal at given time point (bold = significant effect; p < 0.05).
As a representation of overall results, model adjusted outcomes for an animal of average weight were calculated (Fig. 4). Grip strength was significantly decreased for injured limbs compared to control limb at all time points (p < 0.002; Fig. 4(a)). However, there was consistent recovery over time for injured limbs, improving from a ∼54% reduction at day 49 to ∼27% reduction at day 84. Grip strength for contralateral limbs closely followed control limbs, with no significant differences except at day 70 (p = 0.0105; Fig. 4(b)). Side-to-side grip strength difference (injured minus contralateral limb grip strength for injured animals and left minus right limb grip strength for control animals) followed a similar trend to grip strength values for injured limbs, remaining significantly decreased compared to controls at all time points (p < 0.006; Fig. 4(c)). Limb differences for injured animals steadily recovered from day 49 (−380.8 ± 31.6 g) to day 84 (−114.5 ± 31.6 g), but failed to reach control values by the end of the study.
Fig. 4.
(a) Grip strength values were decreased for injured limbs long-term compared to controls while (b) contralateral limb grip strength was not different from control (except at day 70). (c) Grip strength difference values remained decreased for injured animals compared to control at all time points. (Data shown are model adjusted outcomes for an animal of average weight; 95% C.I.; INJ = injured; CL = contralateral; L − R = left minus right control limb; INJ − CL = injured minus contralateral limb; * = different from control at specified time point; p < 0.05).
Gait Testing.
Fit results of the linear mixed models for gait data are represented by the summary of fit, significance of fixed effects, and REML variance (Table 2). In general, gait models fit the data well with 10 out of 12 fixed effects being significant in at least one of the models. While many gait parameters were dependent on velocity (Table 2), there were no differences between walking speeds in control and injured groups (control: 42.5 ± 9.1 cm/s; injured: 39.4 ± 8.5 cm/s). Random effects in the contralateral limb duty factor and stride length models explained an approximately equal percentage of the REML variance. In the remaining four models, repeated trials (subsample) accounted for approximately twice the REML variance as compared to individual animals (rat ID).
Table 2.
Gait linear mixed model summary of fit, significance of fixed effects, and REML variance
| INJ limb duty factor | CL limb duty factor | Duty factor imbalance | Temporal symmetry | Spatial symmetry | Stride length | ||
| R2 | 0.664 | 0.735 | 0.665 | 0.626 | 0.695 | 0.848 | |
| Summary of fit | RMSE | 0.027 | 0.024 | 0.034 | 0.036 | 0.026 | 0.765 |
| Source | p-value | ||||||
| Fixed effects | Time | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 |
| Treatment group | <0.0001 | 0.0363 | <0.0001 | <0.0001 | <0.0001 | 0.0003 | |
| Time × Treatment group | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | <0.0001 | |
| Stud Res Wt | 0.0256 | 0.8900 | 0.0145 | 0.0002 | 0.0074 | 0.6920 | |
| Stud Res Wt × Time | 0.0347 | 0.9031 | 0.0043 | 0.0040 | 0.0002 | 0.2167 | |
| Stud Res Wt × Treatment group | 0.0033 | 0.3396 | 0.3692 | 0.5475 | 0.5172 | 0.5446 | |
| Sud Res Wt × Treatment group × Time | <0.0001 | 0.3112 | 0.0001 | 0.1306 | 0.1533 | 0.2192 | |
| Stud Res Vel | <0.0001 | <0.0001 | 0.1007 | 0.0320 | 0.0110 | <0.0001 | |
| Stud Res Vel × Time | 0.0147 | 0.5557 | 0.0083 | 0.0007 | 0.0771 | 0.0017 | |
| Stud Res Vel × Treatment of group | 0.4699 | 0.0056 | 0.3034 | 0.5830 | 0.0172 | 0.8031 | |
| Stud Res Vel × Treatment group × Time | 0.1265 | 0.4306 | 0.8640 | 0.0806 | 0.5728 | 0.5904 | |
| Stud Res Wt × Stud Res Vel | 0.0910 | 0.1828 | 0.4724 | 0.2924 | 0.6061 | 0.1394 | |
| Random effect | Variance component (percent of total) | ||||||
| REML variance | Rat ID | 0.00034 (31.9) | 0.00062 (52.4) | 0.00052 (32.0) | 0.00075 (36.2) | 0.00040 (37.4) | 0.66628 (53.3) |
| Subsample | 0.00072 (68.1) | 0.00057 (47.6) | 0.00112 (68.0) | 0.00132 (63.8) | 0.00067 (62.6) | 0.58463 (46.7) | |
Note: INJ/CL limb duty factor = model details for injured/contralateral limb duty factor compared to average forelimb duty factor in controls; duty factor imbalance/temporal symmetry/spatial symmetry/stride length = model details for specified injured animal parameter compared to control parameter; time = time point (day 49, 56, … 84); treatment group = control or injured; Stud Res Wt = studentized residual weight; Stud Res Vel = studentized residual velocity; Rat ID = individual animal identification; subsample = repeated measurements on an animal at given time point (bold = significant effect; p < 0.05).
As a representation of overall results, model adjusted outcomes for an animal of average weight and velocity were calculated (Fig. 5). Injured limb duty factor remained significantly decreased at all time points and showed only slight recovery from day 49 to day 56 (p < 0.001; Fig. 5(a)). Contralateral limb duty factor was initially increased compared to control (i.e., day 49 and day 56; p < 0.001), but fully recovered by day 63 and remained similar to control limb averages through day 84 (Fig. 5(b)). Duty factor imbalance (or side-to-side limb difference) was dramatically reduced for injured animals at day 49 and partially recovered over time, but remained significantly decreased compared to control, with an apparent plateau from day 63 onward (p < 0.0005; Fig. 5(c)). Temporal (p < 0.004; Fig. 5(d)) and spatial (p < 0.006; Fig. 5(e)) symmetry for injured animals showed a similar trend to duty factor imbalance; both parameters were significantly decreased at day 49 and only partially recovered, plateauing by day 70. Stride length for injured animals steadily recovered over time yet was significantly decreased compared to control at all time points (p < 0.04) except day 84 (Fig. 5(f)).
Fig. 5.
(a) Duty factor values were decreased for injured limbs long-term and (b) contralateral limb duty factor was initially increased compared to controls. (c) Duty factor imbalance, (d) temporal symmetry, and (e) spatial symmetry remained decreased long-term for injured animals compared to age-matched control animals. (f) Stride length was initially decreased but fully recovered by day 84. (Data shown are model adjusted outcomes for an animal of average weight and velocity; 95% C.I.; INJ = injured; CL = contralateral; * = different from control at specified time point; p < 0.05).
Mechanical Testing.
Raw mechanical testing parameters of injured limbs showed differences compared to their uninjured contralateral limb and uninjured control limbs for maximum extension (Fig. 6(a)), total ROM (Fig. 6(c)), and NZ length (Fig. 6(d)). For maximum flexion, injured limbs were similar to their uninjured contralateral limb and to uninjured control limbs (Fig. 6(b)). Furthermore, uninjured contralateral limb and uninjured control limbs were similar for all biomechanical parameters (Fig. 6). Due to the anatomical structure of the elbow joint in Long-Evans rats, animal limbs cannot reach a max extension of 0 deg.
Fig. 6.
Mechanical test parameters of: (a) maximum extension, (b) maximum flexion, (c) total ROM, and (d) NZ length are shown to highlight raw differences between limbs in each group. Injured and contralateral limb differences were statistically different from zero for (e) maximum extension, (g) total ROM, and (h) NZ length; however, (f) maximum flexion was not different from zero. Differences between left and right control limbs were not statistically different from zero in any mechanical testing parameters (e)–(g). When comparing between groups, limbs differences in injured animals were different from limb differences in control animals for (e) maximum extension, (g) total ROM, and (h) NZ length, but not (f) maximum flexion. (Average ± standard deviation; L = left control limb; R = right control limb; INJ = injured limb; CL = contralateral limb; L − R = left minus right control limb; INJ − CL = injured minus contralateral limb; # = different from theoretical mean of zero (μ0 = 0); * = injured animal difference different from control difference; p < 0.05).
Differences between injured and contralateral limbs (i.e., injured limb minus contralateral limb) were statistically different from zero (p < 0.001) for maximum extension (Fig. 6(e)), total ROM (Fig. 6(g)), and NZ length (Fig. 6(h)); however, maximum flexion was not different from zero (Fig. 6(f)). Differences between left and right control limbs (i.e., left limb minus right limb) were not statistically different from zero for any of the mechanical test parameters. When comparing between groups, side-to-side differences in injured animals were significantly different compared to control animals (p < 0.0002) for maximum extension (Fig. 6(e)), total ROM (Fig. 6(g)), and NZ length (Fig. 6(h)), but not maximum flexion (Fig. 6(f)).
Discussion
This study showed significant deficits remained long-term in joint mechanics, grip strength, and gait in our animal model of unilateral post-traumatic elbow contracture. This study used our previously established animal model of PTJC [6,7,9], and joint mechanics were comparable to our previous results [7] with significant differences in injured limb maximum extension, total ROM, and NZ length compared to control limbs. Importantly, this study marks the first quantification of in vivo active, longitudinal, and functional measurements in our animal model via grip strength and spatiotemporal gait analysis.
Linear mixed statistical models were necessary to account for temporal and intersubject correlations and interactions in our longitudinally collected grip strength data. A comprehensive predictive model consisting of 658 total trials collected on 19 animals over 6 weeks was developed. Model results of grip strength are indicative of a unilateral injury. Specifically, the injured limb remained significantly different from control limbs while the contralateral limb was not different (except at day 70). Side-to-side grip strength difference values in control animals remained ∼0 as expected, while similar values in injured animals were significantly decreased at all time points. Interestingly, injured limb grip strength and injured–contralateral differences never appeared to plateau, suggesting these parameters might return to control levels at a later time point. A recently developed mouse model of elbow contracture evaluated grip strength at 28-days postinjury [22] and observed significant differences between injured and control animals, similar to our results. To the best of our knowledge, this is the only other small animal model of elbow joint contracture.
Our novel bilateral grip strength device allowed for the simultaneous measurement of left and right forelimbs, which was important for investigating the impact of a unilateral injury in this study. Conventional devices typically consist of a grid, ring, or T-bar grasping device connected to a single load cell or strain gauge [29]. Unilateral differences can only be assessed by restraining one limb during testing on these devices. Our device allows for a more natural bilateral positioning of the limbs. While not tested here, our device was also designed to enable measurement of grip strength at varying pronation–supination positions. Specifically, each ladder can be rotated inward or outward to measure strength in a pronated, supinated, or neutral position. This study was completed in the neutral position; however, future work could investigate other positions as our previous work has shown differences in pronation–supination joint mechanics between injured and contralateral limbs in this animal model [8].
Similar to grip strength models, linear mixed models were necessary to account for inter-subject correlation and interactions in our longitudinally collected gait data. In total, 950 gait trials collected on 19 animals over 6 weeks were used to develop the gait parameter predictive model. Considering the overall set of model predicted gait parameters, the differences and trends seen in the injured rats were indicative of a persistent unilateral forelimb injury. Limb duty factor values for injured limbs were expected to be decreased compared to controls as the rat shifted weight from the injured limb to the contralateral limb [30]. Interestingly, limb duty factor values for contralateral limbs returned to control levels by day 63 suggesting that hindlimbs may have started compensating for the gait abnormality; we were not able to confirm this since hindlimb gait was not analyzed in this study. As expected, duty factor imbalance values, defined as the difference in duty factor between opposing limbs within an animal, were approximately zero in control animals, corresponding to a balanced gait. The significant decrease in duty factor imbalance in injured animals compared to control animals was expected and shows an imbalance of stance times between the limbs. Temporal and spatial symmetry were expected to be ∼50% in control animals. As anticipated, symmetry values (both temporal and spatial) for unilateral injuries were less than 50%, which quantitatively demonstrates an asynchronous foot-strike pattern where animals hesitated to apply, then rapidly removed, weight from their injured limb [30]. Unexpectedly, stride length fully recovered by the end of the study, suggesting the contracted elbow and corresponding reduced ROM were not factors in this measurement. Here, stride length and spatial symmetry are reported rather than step distances. However, step distance is merely an algebraic conversion of these two measures [30] and can be calculated by multiplying stride length by spatial symmetry. Considering results for these two parameters in injured animals, injured limb step distances would be decreased while contralateral limb step distances would be increased compared to controls. Thus, spatial symmetry and injured limb step distance are likely better indicators of reduced ROM. Only one previous study has evaluated gait in an animal model of elbow contracture [22]. Using a murine model of PTJC, significantly decreased mean step distances were observed at 28-days postinjury, similar to this study.
This study is not without limitations. First, grip strength analysis measures more than strictly elbow joint function as these measurements are impacted by wrist, forearm, upper arm, and shoulder mechanics/strength. However, grip strength is still a meaningful way to measure overall limb strength and function. The results from this study indicate that elbow injury involving insult to joint stabilizing tissues and immobilization impacted limb use and strength. Second, while differences exist between quadruped gait compensations associated with a load bearing limb and clinically relevant outcomes for elbow PTJC patients, the goal of this work was to develop a technique capable of longitudinal assessment of elbow joint function in our animal model to understand the impact of PTJC and assess future therapy interventions. Specifically, the progression of recovery in functional measures in combination with postmortem mechanical and histological assessment will provide a comprehensive assessment of therapy intervention success.
Conclusion
Previous work with our animal elbow model of PTJC was limited to postmortem, passive assessment of joint function. In this study, active functional consequences associated with the unilateral injury were shown to persist long-term and never fully recover, thereby further validating the clinical relevance and utility of our animal model. This study described a custom bilateral grip strength device and demonstrated its ability to measure functional changes in rodent forelimbs; results showed deficits in forelimb grip strength in the injured limb persisted over time. Furthermore, gait patterns remained imbalanced and asymmetric throughout the study (although stride length and contralateral limb duty factor recovered). A quantitative understanding of the longitudinal, functional disabilities associated with this injury model will help assess the effectiveness of future treatment interventions aimed at reducing or preventing PTJC.
Footnotes
Contributor Information
Alex J. Reiter, Department of Mechanical Engineering , and Materials Science, , Washington University in St. Louis, , St. Louis, MO 63130
Griffin J. Kivitz, Department of Mechanical Engineering , and Materials Science, , Washington University in St. Louis, , St. Louis, MO 63130
Ryan M. Castile, Department of Mechanical Engineering , and Materials Science, , Washington University in St. Louis, , St. Louis, MO 63130
Paul C. Cannon, Seed Production Innovation, , Bayer Crop Science, , St. Louis, MO 63146
Emily H. Lakes, J. Crayton Pruitt Family Department , of Biomedical Engineering, , University of Florida, , Gainesville, FL 32610
Brittany Y. Jacobs, J. Crayton Pruitt Family Department , of Biomedical Engineering, , University of Florida, , Gainesville, FL 32610
Kyle D. Allen, J. Crayton Pruitt Family Department , of Biomedical Engineering, , University of Florida, , Gainesville, FL 32610
Aaron M. Chamberlain, Department of Orthopaedic Surgery, , Washington University in St. Louis, , St. Louis, MO 63130
Spencer P. Lake, Department of Mechanical Engineering, and Materials Science, , Department of Orthopaedic Surgery, Department of Biomedical Engineering, Washington University in St. Louis, , St. Louis, MO 63130 , e-mail: lake.s@wustl.edu
Funding Data
National Institute of Health (Grant Nos. R01 AR071444, R01 AR068424, R01 AR071431, and K99/R00 AR057426; Funder ID: 10.13039/100000002).
National Science Foundation (Grant No. DGE-1745038 Funder ID: 10.13039/501100008982).
Washington University Office of Undergraduate Research (Funder ID: 10.13039/100007268).
References
- [1]. Gaston, P. , Will, E. M. , and Keating, J. F. , 2005, “ Recovery of Knee Function Following Fracture of the Tibial Plateau,” J. Bone Jt. Surg., Br., 87(9), pp. 1233–1236.https://www.ncbi.nlm.nih.gov/pubmed/16129749 [DOI] [PubMed] [Google Scholar]
- [2]. Anakwe, R. E. , Middleton, S. D. , Jenkins, P. J. , McQueen, M. M. , and Court-Brown, C. M. , 2011, “ Patient-Reported Outcomes After Simple Dislocation of the Elbow,” J. Bone Jt. Surg., Ser. A, 93(13), pp. 1220–1226. 10.2106/JBJS.J.00860 [DOI] [PubMed] [Google Scholar]
- [3]. Charalambous, C. P. , and Morrey, B. F. , 2012, “ Posttraumatic Elbow Stiffness,” J. Bone Jt. Surg., Ser. A, 94(15), pp. 1428–1437. 10.2106/JBJS.K.00711 [DOI] [PubMed] [Google Scholar]
- [4]. Monument, M. J. , Hart, D. A. , Salo, P. T. , Befus, A. D. , and Hildebrand, K. A. , 2013, “ Posttraumatic Elbow Contractures: Targeting Neuroinflammatory Fibrogenic Mechanisms,” J. Orthop. Sci., 18(6), pp. 869–877. 10.1007/s00776-013-0447-5 [DOI] [PubMed] [Google Scholar]
- [5]. Myden, C. , and Hildebrand, K. , 2011, “ Elbow Joint Contracture After Traumatic Injury,” J. Shoulder Elbow Surg., 20(1), pp. 39–44. 10.1016/j.jse.2010.07.013 [DOI] [PubMed] [Google Scholar]
- [6]. Lake, S. P. , Castile, R. M. , Borinsky, S. , Dunham, C. L. , Havlioglu, N. , and Galatz, L. M. , 2016, “ Development and Use of an Animal Model to Study Post-Traumatic Stiffness and Contracture of the Elbow,” J. Orthop. Res., 34(2), pp. 354–364. 10.1002/jor.22981 [DOI] [PubMed] [Google Scholar]
- [7]. Dunham, C. L. , Castile, R. M. , Havlioglu, N. , Chamberlain, A. M. , Galatz, L. M. , and Lake, S. P. , 2017, “ Persistent Motion Loss After Free Joint Mobilization in a Rat Model of Post-Traumatic Elbow Contracture,” J. Shoulder Elbow Surg., 26(4), pp. 611–618. 10.1016/j.jse.2016.09.059 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8]. Dunham, C. L. , Castile, R. M. , Chamberlain, A. M. , Galatz, L. M. , and Lake, S. P. , 2017, “ Pronation-Supination Motion Is Altered in a Rat Model of Post-Traumatic Elbow Contracture,” ASME J. Biomech. Eng., 139(7), p. 071011. 10.1115/1.4036472 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9]. Dunham, C. L. , Castile, R. M. , Havlioglu, N. , Chamberlain, A. M. , and Lake, S. P. , 2018, “ Temporal Patterns of Motion in Flexion-Extension and Pronation-Supination in a Rat Model of Posttraumatic Elbow Contracture,” Clin. Orthop. Relat. Res., 476(9), pp. 1878–1889. 10.1097/CORR.0000000000000388 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10]. Montilla-García, Á. , Tejada, M. , Perazzoli, G. , Entrena, J. M. , Portillo-Salido, E. , Fernández-Segura, E. , Cañizares, F. J. , and Cobos, E. J. , 2017, “ Grip Strength in Mice With Joint Inflammation: A Rheumatology Function Test Sensitive to Pain and Analgesia,” Neuropharmacology, 125, pp. 231–242. 10.1016/j.neuropharm.2017.07.029 [DOI] [PubMed] [Google Scholar]
- [11]. Gao, H. G. L. , Fisher, P. W. , Lambi, A. G. , Wade, C. K. , Barr-Gillespie, A. E. , Popoff, S. N. , and Barbe, M. F. , 2013, “ Increased Serum and Musculotendinous Fibrogenic Proteins Following Persistent Low-Grade Inflammation in a Rat Model of Long-Term Upper Extremity Overuse,” PLoS One, 8(8), p. e71875. 10.1371/journal.pone.0071875 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12]. Sun, M. , Connizzo, B. K. , Adams, S. M. , Freedman, B. R. , Wenstrup, R. J. , Soslowsky, L. J. , and Birk, D. E. , 2015, “ Targeted Deletion of Collagen V in Tendons and Ligaments Results in a Classic Ehlers-Danlos Syndrome Joint Phenotype,” Am. J. Pathol., 185(5), pp. 1436–1447. 10.1016/j.ajpath.2015.01.031 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13]. Smith, J. P. , Hicks, P. S. , Ortiz, L. R. , Martinez, M. J. , and Mandler, R. N. , 1995, “ Quantitative Measurement of Muscle Strength in the Mouse,” J. Neurosci. Methods, 62(1–2), pp. 15–19. 10.1016/0165-0270(95)00049-6 [DOI] [PubMed] [Google Scholar]
- [14]. Ge, X. , Cho, A. , Ciol, M. A. , Pettan-Brewer, C. , Snyder, J. , Rabinovitch, P. , and Ladiges, W. , 2016, “ Grip Strength Is Potentially an Early Indicator of Age-Related Decline in Mice,” Pathobiol. Aging Age-Relat. Dis., 6(1), p. 32981. 10.3402/pba.v6.32981 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15]. Allen, K. D. , Mata, B. A. , Gabr, M. A. , Huebner, J. L. , Adams, S. B. , Kraus, V. B. , Schmitt, D. O. , and Setton, L. A. , 2012, “ Kinematic and Dynamic Gait Compensations Resulting From Knee Instability in a Rat Model of Osteoarthritis,” Arthritis Res. Ther., 14, p. R78. 10.1186/ar3801 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16]. Kloefkorn, H. E. , Jacobs, B. Y. , Loye, A. M. , and Allen, K. D. , 2015, “ Spatiotemporal Gait Compensations Following Medial Collateral Ligament and Medial Meniscus Injury in the Rat: Correlating Gait Patterns to Joint Damage,” Arthritis Res. Ther., 17(1), pp. 1–15. 10.1186/s13075-015-0791-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17]. Jacobs, B. Y. , Dunnigan, K. , Pires-Fernandes, M. , and Allen, K. D. , 2017, “ Unique Spatiotemporal and Dynamic Gait Compensations in the Rat Monoiodoacetate Injection and Medial Meniscus Transection Models of Knee Osteoarthritis,” Osteoarthritis Cartilage, 25(5), pp. 750–758. 10.1016/j.joca.2016.12.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18]. Fu, S. C. , Chan, K. M. , Chan, L. S. , Fong, D. T. P. , and Lui, P. Y. P. , 2009, “ The Use of Motion Analysis to Measure Pain-Related Behaviour in a Rat Model of Degenerative Tendon Injuries,” J. Neurosci. Methods, 179(2), pp. 309–318. 10.1016/j.jneumeth.2009.02.011 [DOI] [PubMed] [Google Scholar]
- [19]. Freedman, B. R. , Salka, N. S. , Morris, T. R. , Bhatt, P. R. , Pardes, A. M. , Gordon, J. A. , Nuss, C. A. , Riggin, C. N. , Fryhofer, G. W. , Farber, D. C. , and Soslowsky, L. , 2017, “ Temporal Healing of Achilles Tendons After Injury in Rodents Depends on Surgical Treatment and Activity,” J. Am. Acad. Orthop. Surg., 25(9), pp. 635–647. 10.5435/JAAOS-D-16-00620 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20]. Sarver, J. J. , Dishowitz, M. I. , Kim, S. Y. , and Soslowsky, L. J. , 2010, “ Transient Decreases in Forelimb Gait and Ground Reaction Forces Following Rotator Cuff Injury and Repair in a Rat Model,” J. Biomech., 43(4), pp. 778–782. 10.1016/j.jbiomech.2009.10.031 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21]. Giszter, S. F. , Davies, M. R. , and Graziani, V. , 2008, “ Coordination Strategies for Limb Forces During Weight-Bearing Locomotion in Normal Rats, and in Rats Spinalized as Neonates,” Exp. Brain Res., 190(1), pp. 53–69. 10.1007/s00221-008-1451-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22]. Moore-Lotridge, S. N. , Oelsner, W. K. , Ihejirika, Y. , Desai, M. J. , Gebhart, S. S. , and Schoenecker, J. G. , 2018, “ Novel Preclinical Murine Model of Trauma-Induced Elbow Stiffness,” J. Exp. Orthop., 5(1), p. 36. 10.1186/s40634-018-0155-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23]. Kloefkorn, H. E. , Pettengill, T. R. , Turner, S. M. F. , Streeter, K. A. , Gonzalez-Rothi, E. J. , Fuller, D. D. , and Allen, K. D. , 2017, “ Automated Gait Analysis Through Hues and Areas (AGATHA): A Method to Characterize the Spatiotemporal Pattern of Rat Gait,” Ann. Biomed. Eng., 45(3), pp. 711–725. 10.1007/s10439-016-1717-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24]. Jacobs, B. Y. , Lakes, E. H. , Reiter, A. J. , Lake, S. P. , Ham, T. R. , Leipzig, N. D. , Porvasnik, S. L. , Schmidt, C. E. , Wachs, R. A. , and Allen, K. D. , 2018, “ The Open Source GAITOR Suite for Rodent Gait Analysis,” Sci. Rep., 8(1), p. 9797. 10.1038/s41598-018-28134-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25]. Whishaw, I. Q. , and Pellis, S. M. , 1990, “ The Structure of Skilled Forelimb Reaching in the Rat: A Proximally Driven Movement With a Single Distal Rotatory Component,” Behav. Brain Res., 41(1), pp. 49–59. 10.1016/0166-4328(90)90053-H [DOI] [PubMed] [Google Scholar]
- [26]. Whishaw, I. Q. , Gorny, B. , Foroud, A. , and Kleim, J. A. , 2003, “ Long-Evans and Sprague-Dawley Rats Have Similar Skilled Reaching Success and Limb Representations in Motor Cortex but Different Movements: Some Cautionary Insights Into the Selection of Rat Strains for Neurobiological Motor Research,” Behav. Brain Res., 145(1–2), pp. 221–232. 10.1016/S0166-4328(03)00143-8 [DOI] [PubMed] [Google Scholar]
- [27]. Sacrey, L. A. R. , Alaverdashvili, M. , and Whishaw, I. Q. , 2009, “ Similar Hand Shaping in Reaching-for-Food (Skilled Reaching) in Rats and Humans Provides Evidence of Homology in Release, Collection, and Manipulation Movements,” Behav. Brain Res., 204(1), pp. 153–161. 10.1016/j.bbr.2009.05.035 [DOI] [PubMed] [Google Scholar]
- [28]. Aoki, S. , Sato, Y. , and Yanagihara, D. , 2012, “ Characteristics of Leading Forelimb Movements for Obstacle Avoidance During Locomotion in Rats,” Neurosci. Res., 74(2), pp. 129–137. 10.1016/j.neures.2012.07.007 [DOI] [PubMed] [Google Scholar]
- [29]. Maurissen, J. P. J. , Marable, B. R. , Andrus, A. K. , and Stebbins, K. E. , 2003, “ Factors Affecting Grip Strength Testing,” Neurotoxicol. Teratol., 25(5), pp. 543–553. 10.1016/S0892-0362(03)00073-4 [DOI] [PubMed] [Google Scholar]
- [30]. Jacobs, B. Y. , Kloefkorn, H. E. , and Allen, K. D. , 2014, “ Gait Analysis Methods for Rodent Models of Osteoarthritis,” Curr. Pain Headache Rep., 18, p. 456. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31]. Lakes, E. H. , and Allen, K. D. , 2016, “ Gait Analysis Methods for Rodent Models of Arthritic Disorders: Reviews and Recommendations,” Osteoarthritis Cartilage, 24(11), pp. 1837–1849. 10.1016/j.joca.2016.03.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32]. Hildebrand, M. , 1989, “The Quadrupedal Gaits of Vertebrates,” Bioscience, 39(11), pp. 766–775. [Google Scholar]