Biomechanical, Histological, and Molecular Characterization of a New Post-traumatic Model of Arthrofibrosis in Rats

Abstract

Experimental analyses of post-traumatic knee arthrofibrosis utilize a rabbit model as a gold standard. However, a rodent model of arthrofibrosis offers many advantages including reduced cost and comparison with other models of organ fibrosis. This study aimed to characterize the biomechanical, histological, and molecular features of a novel post-traumatic model of arthrofibrosis in rats. 48 rats were divided into two equal groups. An immobilization procedure was performed on the right hind limbs of experimental rats. One group was immobilized for 4 weeks and the other for 8 weeks. Both groups were remobilized for 4 weeks. Limbs were studied biomechanically via assessment of torque versus degree of extension, histologically via whole knee specimen, and molecularly via gene expression of posterior capsular tissues. Significant differences were observed between experimental and control limbs at 4 N-cm of torque in the 4-week (knee extension: 115° ± 8° vs. 169° ± 17°, respectively; p=0.007) and 8-week immobilization groups (knee extension: 99° ± 12° vs. 174° ± 9°, respectively; p=0.008). Histologically, in each group experimental limbs demonstrated increased posterior capsular thickness and total area of tissue when compared to control limbs (p<0.05). Gene expression values evaluated in each group were comparable. This study presents a novel rat model of arthrofibrosis with severe and persistent knee contractures demonstrated biomechanically and histologically. Statement of clinical significance: Arthrofibrosis is a common complication following contemporary TKAs. The proposed model is reproducible, cost-effective, and can be employed for translational investigations studying the pathogenesis of arthrofibrosis and efficacy of neoadjuvant pharmacologic agents.

INTRODUCTION

Arthrofibrosis continues to be a major failure mode following primary total knee arthroplasty (TKA), affecting 4% of all patients¹. Currently, clinical interventions are limited to early physical therapy, manipulation under anesthesia, or in severe cases, open synovectomy and revision TKA^2–4. There are limited data demonstrating the efficacy of pharmacologic agents in the prevention or mitigation of arthrofibrosis in humans, particularly after a joint replacement. However, there is emerging evidence identifying new molecular mechanisms that are functionally implicated in the disease process that may provide novel therapeutic targets to investigate⁵.

As new pharmacologic and genetic targets are recognized, there is an increasing desire to test these both in vitro and in vivo. The vast majority of in vivo studies investigating post-traumatic arthrofibrosis utilize a rabbit model as it has been demonstrated to be reproducible and informative in regards to the biomechanical, histological, and molecular determinants of the disease^6–14. Despite the rabbit model’s proven efficacy, there remains interest in the development of smaller animal models (e.g. rat) to increase cost-effectiveness and experimental versatility. Additionally, there is a broader understanding of systemic pharmacokinetics in rats as a result of studies of other non-musculoskeletal (MSK) fibrotic disorders. These disorders may serve as a guide for further translational arthrofibrotic studies in rats^15–17.

In the study of atraumatic forms of arthrofibrosis (e.g., immobilization secondary to debility or paralysis), rats frequently have been utilized to study the pathogenesis^{18; 19}. In these studies, there have been extensive investigation into the timeline of contracture formation, histopathologic features, and molecular signature of the models^{20; 21}. Importantly, evaluation of atraumatic forms of contracture development at clinically relevant time points have demonstrated that knee stiffness is the result of myogenic contractures (i.e. shortening of the muscular attachments of the knee) which are distinct from the arthrogenic contractures encountered with clinical arthrofibrosis following musculoskeletal joint surgery^{12; 22}. The critical difference is that myogenic contractures improve with stretching of the muscles whereas arthrogenic contractures are the result of capsular thickening that persists despite therapy.

In previous post-traumatic models of arthrofibrosis in rats, joint contracture was established through an intra-articular injury followed by immobilization of the animals^{20; 21; 23} analogous to previously described rabbit animal models^{12; 24}. Intra-articular injury was either via non-cartilaginous cortical defects^{21; 23} or via direct injury to the trochlear cartilage of the femur²⁰. Similarly, immobilization techniques differed in each model with either suture^{20; 23} or wire²¹ utilized to hold the knee in a flexed position. To date, rat models of post-traumatic have focused on the biomechanical implications of joint contracture, importantly distinguishing the contracture to be arthrogenic in nature^{20; 21; 23}. However, these models have not addressed important questions regarding the optimal period of immobilization required to maximize joint contracture formation or quantified capsular tissue properties. Likewise, though the biomechanical properties of these models have been investigated, important questions regarding the histopathologic and molecular changes that characterize the rat model of post-traumatic arthrofibrosis persist.

This study aimed to determine biomechanical properties of the posterior capsule in terms of contracture severity and stiffness based on immobilization period, to define a method for assessing posterior capsule histological changes, and to characterize the molecular changes associated with contracture formation through quantitative polymerase chain reaction experimentation (qPCR).

ANIMALS AND METHODS

Ethical treatment of Animals

All experiments were conducted with the guidance of the Guide for the Care and Use of Laboratory Animals (2011). Experiments were approved by the Institutional Animal Care and Use Committee (Mayo Clinic IACUC # A00004211–19) prior to study initiation.

Husbandry

Animals were socially housed (4/cage) in 1800 cm² polysulfone cages (Allentown Inc., Allentown, New Jersey). Animals were monitored twice daily and were provided cage enrichment resources. The animals’ diet consisted of a commercially-supplied pelleted cow, PicoLab^® Rodent Diet 20 (St. Louis, Missouri). Filtered tap water was available ad libitum. Animals were housed at a constant room temperature (23° C) with 12 hours of light exposure daily.

Study Design

A total of 48 female, skeletally mature 14-week old Sprague Dawley rats²⁵ – Rattus norvegicus domestica – with mean weight of 223 grams (range, 211 – 239) were utilized in the study (Envigo, North America). A single sex was utilized in an effort to reduce size and genetic variation among animals. The 48 rats were divided randomly into 2 groups (i.e. groups 1 and 2), each with 16 experimental rats and 8 control rats. In the 16 experimental rats, the right limb received the experimental intervention (referred to as experimental limb) while the left limb served as an internal control (referred to as contralateral limb). In the 8 control rats with no surgery, both the right and left limbs served as controls for a total of 16 limbs.

The difference between group 1 and group 2 was the experimental right limb was immobilized for 4 weeks or 8 weeks, respectively. Otherwise, the groups were identical (Figure 1). Because only the period of immobilization varied, the groups will be referred to as 4-week and 8-week groups according to their period of immobilization.

Figure 1.

Study design and experimental timeline. Groups were immobilized for a period of 4 and 8 weeks, respectively, to promote contracture formation. After this period, the immobilizing wire was removed and animals were allowed cage free activity for 4 weeks. Following the remobilization period, animals were sacrificed. 50% of animals were allocated to biomechanical measurements and posterior tissue harvesting whereas the other 50% were utilized for histological analysis.

Surgical Procedure

All surgical procedures were conducted under general anesthesia via inhalation of isoflurane (1–2% in oxygen) administered via a nosecone. Cefazolin (30 mg/kg) and buprenorphine SR (0.6 mg/kg) were injected subcutaneously prior to surgical incision for antibiotic prophylaxis and postoperative analgesia, respectively. The animals were prepped with povidone iodine solution and draped in sterile fashion for the procedure. A midline skin incision was made over the stifle (knee) joint and a lateral parapatellar arthrotomy was utilized to gain intra-articular access. The cruciate ligaments were transected and 1 mm drill bit was used to create cortical defects in the medial and lateral femoral condyles with care taken to preserve the collateral ligaments (Figure 2A). The leg was then hyperextended 45° to disrupt the posterior capsule of the stifle joint (Figure 2B).

Figure 2.

Surgical procedure for contracture formation. The surgical procedure was conducted in three primary steps. First, the cruciates were sectioned and the extra-cartilaginous regions of the femoral condyles drilled to create an intra-articular injury (A). Next, the posterior capsule was disrupted through hyperextension of the knee joint (B). The knee was then fixed in > 135° of flexion with stainless steel wire to immobilize the knee (C).

Through a separate incision centered over tibia, a 2.0 316L stainless steel wire (Ethicon, Johnson & Johnson; New Brunswick, NJ) was passed from lateral to medial under the tibia. A lateral thigh incision was used to pass a blunt hemostat submuscularly along the femur and tibia to retrieve the wire. Once the wire was shuttled proximally, it was passed from lateral to medial under the femur. The knee was hyper–flexed to 150° and the wire twisted to tighten the construct (Figure 2C).

Following final tightening of the wire, the parapatellar arthrotomy and skin were closed with absorbable sutures. At the conclusion of the procedure, a fluoroscopic image was obtained to confirm wire placement and rule out iatrogenic fracture. Animals were injected with 5 ml of subcutaneous saline and transferred to a warmed recovery cage.

Wire Removal Procedure

Following the prescribed period of immobilization, a second procedure was performed on the experimental animals to remove the wire. Procedural preparation and anesthesia were identical to the index procedure. Prior to skin incision, a fluoroscopic image was obtained to confirm the wire was intact. If the wire was not intact, the second procedure was not performed and the rat was humanely euthanized via carbon dioxide inhalation. If the wire was intact, removal proceeded.

To remove the wire, a lateral thigh incision was made over the femur and a needle driver was used to untwist the wire. The wire was then removed through a separate anterior tibial incision. Following conclusion of the procedure, a fluoroscopic image was obtained to confirm no iatrogenic fracture had occurred. Animals were recovered as described above.

Wire removal was followed by a 4 week period of remobilization in each group after which experimental and control animals were humanely euthanized via carbon dioxide inhalation and limbs were harvested for biomechanical, histological, and molecular analyses. Half of experimental and control animals were utilized for histological analyses and the other half were used for both biomechanical testing, followed by harvesting of posterior capsular tissue for molecular analyses.

Complications

A single experimental animal (group 1) died during anesthesia induction for the wire removal procedure and was not replaced. Four animals (2 in group 1and 2 in group 2) had wire breakage at time of wire removal. Wire failure occurred at sites of metal contact caused by forming a figure of eight versus the desired loop (Figure 2C) when passing the wire around the tibia and femur. These animals were euthanized and not replaced.

Biomechanical Testing

Hind limbs utilized for biomechanical testing were prepared by disarticulating the limbs at the hip and ankle joints. Care was taken to preserve the posterior capsule and periarticular soft tissue within 1 cm proximal and distal to the knee joint; all other soft tissues were denuded. The femoral end of the hind limb was potted in a 1 cm polyvinyl chloride tube with polymethylmethacrylate (PMMA) cement (Stryker; Mahwah, NJ) (Figure 3A). The potted hind limb was mounted to a metal bracket and a fluoroscopic image was obtained to ensure that the knee was in the center of the mounting device (Figures 3B and 3C). The mounting device was then attached to a dynamic load cell to measure the torque applied to the limb at a constant rate of 1°/s until failure (Figure 3D). Data obtained from the measurement device were transformed into graphs of passive extension versus torque using Matlab 2016a (Mathworks, Inc.; Natick, MA).

Figure 3.

Limb preparation and biomechanical measurement device setup. Disarticulated hind limbs from were potted in a cement vessel (A) and mounted on the measurement device. Fluoroscopic x-ray images were performed to confirm the knee was centered in the device (B). The tibia of the specimen was placed in a lever that passively extends the knee (C). An electronically controlled motor drives a belt that extends the knee at a constant rate of 1°/s (D).

The angle of displacement at varying torques (i.e. 2, 4, and 8 N-cm) was determined for all limbs tested and defined as the passive extension angle (PEA) (Figure 4). The torque values studied were chosen as they represented a range of values that stress the control limbs from physiologic range of motion (i.e. 0–170°) to supraphysiologic range of motion (i.e. >170°). The PEA was determined for both limbs of the control animals as well, 71- as the experimental (operated) and contralateral (non-operated) limbs.

Figure 4.

Measurement schema for normal and pathologic rat hind limbs. Rats inherently have a physiologic extension deficit (red line). As such, the passive extension angle (PEA) of a normal limb fails to reach 180° of extension under basal conditions (black line). The PEA decreases following contracture formation (blue line).

Histologic Processing

Following disarticulation of the hind limbs, knees that included a 1 cm length of the distal femur and proximal tibia were immediately immersed in 40 mL of 10% neutral buffered formalin, stored at room temperature (22°C) for 72 hours, and sent overnight to Premier Laboratory, LLC (Longmont, CO) for processing, sectioning, staining, and whole slide scanning. The knees were decalcified in 10% formic acid for 7 days and then processed to paraffin. The knees were embedded and sectioned at 5 μm intervals in the sagittal plane at multiple levels to capture the medial, central, and lateral regions of the joint. Medial and lateral sections were considered adequate if the posterior horn of the meniscus, tibial plateau, and femoral condyle were visualized. Central sections were considered adequate when both cruciate ligaments were observed in the section (Figure 5). For each knee medial, central and lateral sections were stained with hematoxylin and eosin (H&E), Masson’s trichrome, and toluidine blue utilizing standard methods, resulting in a total of 9 stained slides per knee. Slides were digitized at 20x magnification using an Aperio ScanScope AT2 (Leica Biosystems, Inc.; Buffalo Grove, Illinois).

Figure 5.

Representative medial, central, and lateral hematoxylin and eosin stained sections for control and experimental limbs.

Histopathologic Evaluation

All sections were assessed using an E600 brightfield light microscope (Nikon Inc.; Melville, New York) by an American College of Veterinary Pathologist board-certified veterinary pathologist (BB). Features of the periarticular connective tissue were scored using a tiered, semi-quantitative scale (Table 1). For all three stains these criteria were initially established by an informed (non-blinded) histopathologic evaluation, after which the definitive data set was produced by a coded (blinded) histopathologic assessment.

Table 1.

Histology Grading Criteria for Fibrosis

Score	Degree of Fibrosis	Description of Periarticular Connective Tissue
0	Normal	Intra-articular and peri-articular tissue (excluding ligaments) consists entirely of white fat
1	Minimal	Intra-articular and peri-articular tissue (excluding ligaments) consists mainly of white fat, with small linear foci of dense connective tissue along the margins of fat pads
2	Mild	Intra-articular and peri-articular tissue (excluding ligaments) replaced by dense connective tissue in about 15% of 30% of section area
3	Moderate	Intra-articular and peri-articular tissue (excluding ligaments) replaced by dense connective tissue in about 35% of 50% of section area
4	Marked	Intra-articular and peri-articular tissue (excluding ligaments) replaced by dense connective tissue in about 55% of 80% of section area
5	Severe	Intra-articular and peri-articular tissue (excluding ligaments) replaced by dense connective tissue in more than 85% of section area

Additional histological assessment was performed by two blinded observers (ARO and LD) utilizing the digital images obtained from Aperio ScanScope. In order to quantify the characteristics of the posterior capsular tissue, three measurement techniques were performed using Imagescope software (Leica Biosystems, Inc.). Posterior capsule morphology was assessed utilizing a previously reported method²⁶ that assesses the posterior capsule thickness along a line perpendicular to the patellar tendon (Figure 6A). Additionally, an original protocol was created that measured the posterior capsule thickness at three equidistant points along the axis of the femoral and tibial posterior capsule insertion sites (Figure 6B). Finally, the area of the posterior capsular tissue was determined for each section by manually contouring of the limits of the posterior capsule (Figure 6C). The total area representative of connective tissue – defined here as fibrosis area – was determined by applying Aperio Positive Pixel Count Algorithm (Leica Biosystems, Inc.). The algorithm quantifies the amount of stain present on the slide which eliminates areas of fat and/or areas of processing artifact.

Figure 6.

Quantitative histologic measurements. The posterior capsule tissue thickness was determined by measuring the width of tissue along an axis perpendicular to the patellar tendon (A). Posterior capsule thickness was further investigated at three equidistant points (1: proximal, 2: middle, 3: distal) along an axis defined by the femoral and tibial insertions of the posterior capsule (B). The area of the posterior capsule tissue was measured by manually contouring its limits (C).

Gene Expression

Posterior capsular tissue was harvested from each limb utilized for biomechanical analysis. At time of harvest, tissues were placed in RNAlater (Ambion, Inc.; Austin, Texas) and stored at −20°C. Total RNA was extracted utilizing a standard kit (QIAGEN; Germantown, Maryland) and converted to complimentary DNA. Quantitative PCR was performed using a CFX384 machine (BIO-RAD, Hercules, California) with 10 ng cDNA per 10 μl using the QuantiTect SYBR Green PCR Kit (QIAGEN). Samples were analyzed as biological triplicates using the glyceraldehyde 3-phosphate dehydrogenase (GAPDH) housekeeping gene as a reference. Collagenous, non-collagenous, matrix metalloproteinases and myofibroblastic markers of fibrosis were studied (Supplementary Table 1).

Statistical Analysis

GraphPad Prism version 8.0.0 for Windows (GraphPad Software; San Diego, California) was used to perform all statistical analyses. Group sample size was based upon biomechanical quantification of contracture and determined by assuming a 5% type 1 error and 80% power to detect an effect difference of 20° with a standard deviation (SD) of 10°, thus 8 limbs were required for the experimental and control groups. These numbers were doubled for the present study as limbs utilized for biomechanical testing were taken to posterior capsular failure; therefore, these posterior capsules were not able to be utilized for whole knee histology. For all experimental data, Kolmogorov-Smirnov normality tests were conducted and either parametric or non-parametric comparisons between each treatment group and its respective control group were carried out using either Student’s t-tests (for parametric variance) or Wilcoxon-rank sum tests (for nonparametric variance). Data were reported as group means and SDs. Statistical significance was considered p < 0.05. When applicable, significance is noted in the figures with a standard asterisk convention (* = p ≤ 0.05, ** = p ≤ 0.01, *** = p ≤ 0.001, **** = p ≤ 0.0001).

A reliability assessment was conducted for the three quantitative histological measurements using interclass and intraclass correlation coefficients (ICCs)²⁷. To compute the ICCs, histologic measurements were performed by two independent observers (ARO and LD) for all sections and repeated by a single observer (ARO) on a random selection of 10% of sections of experimental, contralateral, and control limbs.

Receiver operating characteristic (ROC) curve analysis was performed to evaluate the quantitative histological methods utilized. Area under the curve (AUC) and threshold values were computed.

RESULTS

Biomechanical Results

Rats in the 4-week immobilization group demonstrated significant differences in PEAs at all torques evaluated when comparing experimental (operated) limbs to both contralateral (non-operated) and control limbs. Likewise, rats in the 8-week immobilization group demonstrated significant differences in PEAs at all torques evaluated when comparing experimental limbs to contralateral and control limbs. Notably, there was no difference in PEAs between control limbs of non-operated animals and the contralateral (non-operated) limbs of experimental animals in either group (Figure 7).

Figure 7.

Biomechanical data for 4- and 8-week immobilization groups. The passive extension angle at torques ranging from 2 to 8 N-cm were compared between experimental, contralateral and control limbs for 4-week (A) and 8-week groups (B). Significance was observed at all observed torque values for 4- and 8-week groups between experimental and control as well as experimental and contralateral limbs. Contralateral limbs did not demonstrate a difference in passive extension angle compared to controls.

To compare the biomechanics of contractures generated by the 4-week and 8-week immobilization groups, the PEAs were compared between groups. At 4 N-cm of torque, the 4-week immobilization group showed a PEA of 115° ± 8° vs. 99° ± 12° in the 8-week immobilization group. Despite the large differences in PEAs of operated limbs between the 4-week and 8-week groups, no differences were observed when comparing the PEAs directly at 2, 4, or 8 N-cm (Supplemental Figure 1).

However, direct comparisons failed to account for the physiologic deficit that is inherent to individual rats as they assume a quadrupedal gait (Figure 4)²¹. As there were no differences between the contralateral and control limbs (i.e., the contralateral limb served as an internal control in the experimental animals), a novel method of evaluating the degree of contracture – the extension deficit – was calculated by taking the difference of the experimental and contralateral limb from the same animal. Using this method, there were significant differences in the degree of contracture between experimental groups with the 8-week group producing a greater contracture at 4 and 8 N-cm (p<0.05) (Figure 8).

Figure 8.

Extension deficit biomechanical data for 4- and 8-week immobilization groups. The extension deficit is defined as the difference between the contralateral limbs passive extension angle (PEA) and the experimental limbs PEA (A). Using this method of analyzing the biomechanical data, the 8-week immobilization group was observed to have significantly greater extension deficits when compared to the 4-week immobilization group at 4 and 8 N-cm of torque (B). Secondary to the strength of the contracture, greater torque must be applied to observe the differences between 4- and 8-week immobilization groups. As greater torque is applied the deficit, the difference between contralateral and experimental limb, increases in the 8-week group but remains constant in the 4-week group (C).

Histological Analysis

Coded semi-quantitative scoring of sagittal knee sections stained with H&E and Masson’s trichrome showed significant posterior capsular fibrosis present in both experimental cohorts (p<0.05) (Figure 9). Interestingly, the degree of fibrosis was variable according to sagittal sections evaluated with predominantly mild fibrosis (Grade 2) observed in medial sections versus marked to severe fibrosis (grades 4 and 5, respectively) in central and lateral sections (Figure 9).

Figure 9.

Histologic stains performed for 4- and 8-week immobilization groups. Hematoxylin and eosin (A) and Masson’s trichrome (B) stains showed significant difference between experimental and control limb at both 4 and 8 weeks in all sections (i.e. medial, central, and lateral).

For quantitative methods, ICCs demonstrated good interobserver reliability and excellent intraobserver reliability for all methods (Table 2). Quantitative measurements utilizing the method previously described by Watanabe et al²⁶, demonstrated no difference in posterior capsular thickness in medial sections of operated knees. In contrast, central and lateral knee sections demonstrated significantly increased posterior capsular thickness in experimental knees when compared to either control or contralateral limbs in both groups (p<0.05) (Figure 10A).

Table 2.

Interclass and intraclass correlation coeffients for histologic measurements

Measurement	Interclass Correlation	Intraclass Correlation
Posterior capsule thickness	0.73	0.99
Area	0.67	0.99
Fibrosis Area	0.75	0.97
Ratio	0.83	0.94

Figure 10.

Quantification of posterior capsule tissue morphology between 4- and 8-week immobilization groups. Posterior capsule thickness was greater in central and lateral sections at 4 and 8 weeks; however, no difference was observed in the medial sections of either group (A). The area of the posterior capsule was greater in experimental limbs in all sections for the 4-week group, but only for the lateral section of the 8-week immobilization group (B). By removing the fat and staining artifact from the area of the posterior capsule more consistent increases in area were observed in both 4- and 8-week immobilization groups across medial, central and lateral sections (C). The ratio of fibrosis area to overall area were compared and demonstrated that experimental limbs consisted of greater fibrosis area than control or contralateral limbs from the same histologic sectioning region (D).

Evaluation of the total area of the posterior capsular tissue demonstrated increased area in experimental limbs relative to control and/or contralateral limbs in the lateral sections of both 4 and 8 week cohorts and also in the central section of the 4 week cohort (p<0.05) (Figure 10B).

Utilizing a positive pixel counter, the fibrosis area was quantified for both 4-week and 8-week immobilization cohorts. In both groups, experimental limbs consistently demonstrated increased fibrotic area of posterior capsular tissue in comparison to control and contralateral limbs (p<0.05) (Figure 10C).

The ratio of the total area measurement to the fibrosis area measurement was performed to characterize the proportion of non-connective tissue present in the studied limbs. Utilizing this method, both the 4-week and 8-week immobilization groups showed an increased ratio of fibrotic tissue in experimental limbs as compared to control or contralateral limbs in observed sections (Figure 10D).

The predictive ability of the measurement strategies to determine contracted versus non-contracted limbs was investigated by ROC curve analysis. ROC curve analysis demonstrated that posterior capsular thickness along the axis perpendicular to the patellar tendon was an acceptable predictor of contracture formation for 4-week and 8-week immobilization groups (4-week: AUC = 0.84, 95%CI = 0.71 – 0.98; 8 week: AUC = 0.75, 95%CI = 0.59 – 0.90). For the 4-week immobilization group a threshold value of 1 mm demonstrated sensitivity of 81% and specificity of 91% whereas a threshold of 1 mm for the 8-week group showed sensitivity of 70% and specificity of 70% (Figures 11A and B).

Figure 11.

ROC curves for histologic measurements performed. ROC curve for contracture formation with threshold value of 1 mm posterior capsule thickness at both 4 weeks (AUC = 0.84) (A) and 8 weeks (AUC = 0.75) (B). ROC curve for contracture formation with threshold ratio of 0.67 for area ratio at both 4 weeks (AUC = 0.73) (C) and 8 weeks (AUC = 0.96) (D).

Similarly, the ratio of the fibrosis area to total area was studied via ROC curve analysis. For the 4-week immobilization group, the ratio was determined to be an acceptable predictor for the 4-week group with AUC = 0.73 (95%CI = 0.58 – 0.89); a threshold ratio of 0.67 had a sensitivity of 71% and specificity of 73% (Figure 11C). The 8-week immobilization group demonstrated excellent predictive ability with AUC = 0.96 (95%CI = 0.89 – 1.00) and a threshold ratio of 0.67 had a sensitivity of 94% and a specificity of 91% (Figure 11D).

Toluidine blue staining was performed to evaluate the cartilage matrix integrity of medial, central and lateral sections according to an original protocol (Table 3). Utilizing this methodology, control and contralateral limbs had normal cartilage surfaces and matrix composition in all sections evaluated. In contrast, experimental limbs exhibited mild to moderate degenerative cartilage changes, especially in central and lateral sections (Figure 12).

Table 3.

Histology Grading Criteria for Cartilage

Score	Loss of Cartilage Matrix	Description of Cartilage Surface
0	Normal	Diffuse staining of superficial articular cartilage on distal femur and proximal tibia
1	Minimal	Focal loss of articular cartilage staining (usually caudal margin of femoral condyle); ≤ 10%
2	Mild	Multifocal loss of articular cartilage staining (femur and tibia); 15% to 30%
3	Moderate	Multifocal extensive loss of articular cartilage staining (femur and tibia); 35% to 55%
4	Marked	Multifocal extensive loss of articular cartilage staining (femur and tibia); 60% to 85%
5	Severe	Diffuse loss of articular cartilage staining (femur and tibia); ≥85%

Figure 12.

Toluidine blue stains performed for 4- and 8-week immobilization groups to evaluate cartilage integrity. Mild to moderate degenerative changes were observed in experimental animals in central and lateral sections for each group.

Gene Expression

Posterior capsular gene expression levels of collagenous, non-collagenous, and myofibroblastic markers in 4-week and 8-week immobilization groups were comparable (Figure 13A – C). The matrix metalloproteinase (MMP) family of genes demonstrated variable results in the 8-week experimental cohort with decreased gene expression of MMP9 in experimental limbs whereas MMP28 demonstrated increased gene expression (p <0.05) (Figure 13D). Differences in the expression of MMPs may reflect distinctions in extracellular matrix remodeling in contracted versus non-contracted limbs.

Figure 13.

Quantitative PCR analyses for fibrosis related genes for 4- and 8-week immobilization groups. No difference in gene expression levels was observed in collagenous (A), myofibroblastic (B), or non-collagenous (C) markers of fibrosis in either group. Matrix metalloproteinase (MMPs) 9 and 28 demonstrated variable results in the 8 week experimental tissues as MMP9 showed reduced expression in whereas MMP28 showed increased expression relative to other tissues (P<0.05) (D).

DISCUSSION

Arthrofibrosis continues to affect a clinically important subset of patients undergoing TKAs. To facilitate large-scale investigation of potential therapeutic targets, there is interest in utilizing a validated small animal model of arthrofibrosis that can provide reliable data in regard to the biomechanical, histological, and molecular changes that may occur during the pathogenesis of arthrofibrosis.

The current study demonstrated persistent contracture formation in both 4-week and 8-week immobilization groups. Importantly, this study further demonstrates significant differences in the degree of contracture formation between immobilization groups with the 8-week group producing a more severe contracture. This information is useful as detection of therapeutic efficacy relies on both the ability to adequately create a contracture in the model and having the sensitivity to measure differences. Similar to our findings, previous investigations of atraumatic knee contracture formation in rats have identified that maximal knee contracture formation occurs at 8 weeks with minimal further contracture afterwards²². Likewise, these prior studies demonstrated that contractures persisted, though to a reduced degree, after the immobilization period, even after 32 weeks of remobilization²⁸. Analogous, longitudinal studies have not been performed in post-traumatic rat models with reported models immobilizing animals for only two²⁰ or four²¹ weeks followed by inconsistent periods of remobilization after contracture formation.

The current model evaluated knee contractures at various torque to observe both the minimum and maximum load required to observe biomechanical changes associated with contracture. Although differences were noted at all torques evaluated (i.e. 2, 4, and 8 N-cm), it is recommended that future studies not stress the knee beyond 4 N-cm as loads beyond this limit extended the knee into supraphysiologic ranges. In our model, torque was evaluated on a continuous scale, which allows for biomechanical measurements such as stiffness that have been evaluated in larger animal models of joint contracture^{9; 11; 20; 21; 29}. In contrast, previous investigations evaluated the degree of contracture by hanging a single weight from the animal’s disarticulated limb^{20; 21; 29}.

Fibrosis associated with contracture formation was observed both quantitatively and via blinded histopathologic scoring, and was found to be statistically significantly different between experimental and control limbs. In our study, medial posterior capsule sections demonstrated reduced fibrosis in comparison to central and lateral histologic sections. The specific reason for this finding is unclear. However, the combination of a lateral parapatellar arthrotomy used for the index surgery combined with the increased inflammatory response incited by the controlled intra-articular trauma may have resulted in a more robust inflammatory reaction in neighboring lateral and central capsular tissues. Additionally, when evaluating specific sagittal histologic sections, the central sagittal section most consistently demonstrated histological difference between experimental and control animals. As such, this sagittal section – readily identified by the presence of the cruciate ligaments – should be considered the principal histologic section for evaluation of posterior capsular histopathology^{20; 21}.

In regards to specific measurement techniques, linear histological measurements (i.e. thickness of posterior capsule) demonstrated differences between control and experimental animals. However, the predictive natures of these one-dimensional measurements were inferior to those of two-dimensional measurements (i.e. area) when compared with ROC curves (Figure 10). Measurements that accounted for the thickness in two planes (i.e. area) offered greater sensitivity and specificity in detecting contracture formation and may be considered the standard for future experiments assessing the difference in size and tissue characteristics in quantitative analysis of fibrotic tissues. In prior reports, histological data regarding post-traumatic knee arthrofibrosis in rats was limited to one-dimensional measurements of a single sagittal histological section^{20; 21}.

To fully evaluate the histological changes in the joint, we evaluated the cartilage of the tibiofemoral joint and found there were degenerative cartilaginous changes in the experimental (i.e. contracted) limbs that were absent in contralateral and control limbs. This is in opposition to reports of rabbit models of arthrofibrosis where the articular cartilage is unaffected by the periarticular changes of arthrofibrosis^{30; 31}. Interestingly, our report is supported by previous studies of atraumatic models of rat arthrofibrosis which demonstrated cartilaginous changes at apposed and unopposed regions of the tibiofemoral joint following contracture development^{32; 33}. These cartilaginous changes must be considered when evaluating the results of proposed experiments and planning for human trials as they represent a potential confounding factor and complicate interpretation of results.

The present study was unable to demonstrate molecular differences between control and experimental groups. However, gene expression is highly time-dependent and the terminal time points evaluated in our study may not reflect the optimal time points to determine expression differences between groups. Previous studies in rabbits demonstrated significant differences at both early³⁴ and late³⁵ experimental time points. Future investigations evaluating these time points in rats are required to better characterize the molecular basis for contracture in the rat model.

There are limitations to the current study. Foremost, the presence of degenerative cartilage changes in experimental limbs which confound the outcomes and interpretations of data aimed towards investigation of arthrofibrosis. Interestingly, this appears to be a phenomenon experienced in rats (traumatic and atraumatic models), but not other known animal models of arthrofibrosis (i.e. rabbits). Though this does not preclude the utilization of rats in future endeavors, it must be accounted for when drawing conclusions from these studies. Another limitation is that the rats utilized in the current study were not genetically identical. There is emerging data that genetics plays a role in arthrofibrosis formation. However, this was not accounted for in the current study. Finally, though the current study evaluated differences in immobilization periods, the effect of remobilization duration was not investigated. This decision was based on previous data showing motion gains plateaued after 4 weeks of remobilization²¹.

In summary, the current study presents a novel model of post-traumatic arthrofibrosis in rats that created a severe and persistent knee contracture demonstrated both biomechanically and histologically. Furthermore, the current model identified the cartilage degeneration associated with contracture formation previously uninvestigated by other post-traumatic models of arthrofibrosis in rats. Although no molecular conclusions can be drawn from the present study, the techniques and resources utilized are ready to be employed at physiologically relevant time points to elucidate molecular events at play in arthrofibrosis formation. In conclusion, this animal model of post-traumatic arthrofibrosis provides a useful rodent model to evaluate the efficacy and safety of novel antifibrotic therapeutics.

Supplementary Material

fS1

Supplemental Figure 1. Biomechanical data for 4- and 8-week immobilization groups compared. No difference was observed between 4 and 8 week groups at 2, 4, or 8 N-cm when comparing the control, contralateral, and experimental limbs.