Development of preclinical OA models is crucial to the study of OA pathophysiology and evaluation of DMOAD efficacy. However, models must be extensively characterized to ensure that appropriate conclusions are drawn from the studies that use them. In this study we characterized a surgical rodent model of OA, in which ACL-T and PM lead to joint destabilization, and thus OA pathology. Notably, this model more closely reflects secondary forms of OA, which arise from trauma or other disorders [60
]. Nonetheless, it may also have application in primary OA studies. We evaluated OA activity through histomorphometric analysis using the quantitative OARSI scoring method [42
], quantitative analysis of bone mineral density [61
], and biochemical analysis of cartilage breakdown [40
]. Furthermore, we are the first to evaluate the effects of FM on pathogenesis in a rat model of OA, and we assessed chondrocyte hypertrophy in OA pathogenesis. To date, a comprehensive, longitudinal evaluation of a preclinical surgical rodent model of OA, as shown here, has not been reported.
Our histologic results indicate that in this model, articular cartilage degradation consistently begins as early as 2 weeks after surgery and is worse with FM. Early in pathogenesis, the profile of cartilage degradation initially reflects the edema and delamination of the superficial layer, and development of fissures into the mid-zone that are commonly observed during early stages of human OA [60
]. At later time points the model also exhibits features characteristics of late-stage human OA including denudation, and osteophytes and fibrocartilage-like tissue are present at the denuded surface when FM is applied [60
]. Interestingly, although proteoglycan loss occurred at earlier stages in regions with more severe lesions, proteoglycan loss was not progressive over the time course. In fact, proteoglycan staining was more intense near the end of the time course, particularly in repair tissues, which is probably due to a compensatory anabolic repair response. Accordingly, quantitative analyses that include proteoglycan loss in addition to other features of degradation are necessary to achieve a comprehensive understanding of disease progression.
In addition, early loss of subchondral bone density and trabecular architecture were also present and are reminiscent of human OA [10
]. Ultimately, these properties are likely to persist to end-stage OA, where joint failure occurs and invasive arthroplastic intervention is required. Longitudinal analysis allows evaluation of both the early and late stages of OA development. This is highly effective in rodent models in particular, because the time course to overt pathology is relatively short, and a larger number of animals can be managed. As previously shown, longitudinal three-dimensional vBMD analysis in rabbits [52
] and OARSI scores in rodents [42
] are precise tools for assessing OA development in animal models. Overall, our findings correspond with current assessments of OA in humans, and the model produces significant, predictable, and reproducible results. Therefore, we conclude that the ACL-T/PM model of OA with FM is appropriate for use in preclinical studies of OA, providing a means to study the onset, development, and characteristics of lesions that appear to be similar to those in human OA. However, although this model mimics certain aspects of human OA, extrapolation of joint lesions to human OA should be considered with caution, as with any animal model.
There is often disagreement as to whether sham surgery is the most appropriate control in surgical models of OA [28
]. Accordingly, in addition to the ipsilateral knee, we investigated OA activity in the contralateral joint. Histology and OARSI scores revealed minor OA activity in contralateral articular cartilage, including significantly higher scores in contralateral joints at 2 and 12 weeks, compared with shams. We also saw induction of type X collagen and MMP-13 expression in contralateral cartilage late in the 5-month time course. Evidence of contralateral OA activity, however, did not worsen over the time course. Nonetheless, these findings indicate that the surgically unaltered contralateral joint is affected to a minor extent by OA induction in the model. The effects may be due to alterations in weight-bearing during rest or activity [65
] or to systemic factors (for example, circulating inflammatory factors [66
]) yet to be identified in this model [68
]. Interestingly, subchondral bone and vBMD profiles were not altered in contralateral joints (compared with sham joints), perhaps protecting them from OA advancement. Furthermore, we recently demonstrated that contralateral chondrocyte gene expression profiles are altered, relative to shams, emphasizing the importance of sham controls in gene expression studies [69
]. By extension, the contralateral joint may be susceptible to developing OA caused by changes in gait or systemic effects. Accordingly, we conclude that a sham operation in independent animals is the most appropriate control in genetic and biochemical studies [70
]. However, in studies focused on subchondral bone (for example, micro-CT studies) the contralateral joint is a sufficient control, because we did not observe any contralateral changes in subchondral architecture or vBMD over 5 months (compared with shams), and is advantageous for controlling inter-animal variables such as age and weight.
It is thought that OA may arise from abnormal articulations and repetitive joint loading. Accordingly, we explored the effects of FM (inducing repetitive joint loading and maximal flexion and extension of the knee joint) on OA development in the ACL-T/PM model. Because NM animals in this study could voluntarily remain stationary, it was hypothesized that such a paradigm of FM would accelerate pathogenesis. Indeed, this was the case. FM animals had increased cartilage degradation at 2 weeks after surgery, higher OARSI scores, articular surface deformation and subchondral plate failure, and earlier subchondral bone sclerosis than did NM animals. Similar to human OA, FM also induced abnormal repair processes, including fibrocartilage-like tissue and osteophytes, which were not present in NM animals. This was probably due to subchondral plate failure caused by FM, allowing infiltration of bone marrow stromal cells and activation of repair processes such as that seen following osteochondral fracture [72
]. Furthermore, biochemical analysis of type II collagen breakdown indicated that cartilage degradation is elevated in FM animals at 4 weeks. Together, these data indicate that FM in this model effectively accelerates both the onset and progression of OA pathogenesis, resulting in a disease state that is reminiscent of human OA.
The clinical implications of FM are unclear. FM exercise, as applied in this study, involves walking slowly on a rotating drum for 30 min, three times per week, and thus is not vigorous exercise. Rather, FM forces maximal joint flexion and extension and causes repetitive increased load bearing that is not necessarily exhaustive but is deleterious when applied to the destabilized joint. Importantly, however, FM exercise had no deleterious effects on nondestabilized joints. Although we know from this and many other previous studies that joint destabilization in animal models leads to OA pathogenesis, we conclude here that FM accelerates OA pathogenesis, and that repetitive load bearing exercise is deleterious to the destabilized joint, at least in this OA model. A question faced by many patients who suffer OA is whether they should exercise. Exercise is often recommended for patients with hip or knee OA [73
], and the American College of Rheumatology recommends aerobic exercise for patients with knee and hip OA [74
]. However, fear that exercise is damaging their joints causes some patients with OA not to exercise [75
]. The therapeutic importance of exercise is therefore a complex issue. One point of departure may be in the form of exercise. Indeed, many studies have demonstrated the beneficial effects of light to moderate exercise in animal models of OA and patients [35
], whereas more strenuous exercise and repetitive load bearing has a deleterious effect, at least in experimental animals [39
]. We must stress that our study does not suggest that nonrepetitive load bearing forms of exercise accelerate pathogenesis, nor that they are detrimental. However, our study does indicate that both the type and timing of exercise after a joint injury should be considered with care, in addition to the stability of the joint in question.
Several studies have proposed that recapitulation of hypertrophic chondrocyte differentiation is involved in cartilage degradation in OA [79
]. We investigated this hypothesis in our OA model by analyzing the expression of hypertrophic chondrocyte markers, including MMP-13, alkaline phosphatase, and type X collagen. We demonstrated that all three markers of chondrocyte hypertrophy are increased in FM ipsilateral cartilage, supporting this hypothesis. MMP-13, when activated, will digest the territorial matrix, and deposition of type X collagen and secretion of factors that facilitate calcification of the cartilage matrix (such as alkaline phosphatase) are deleterious to articular cartilage. Furthermore, degradation of the matrix by catabolic factors such as MMP-13 make the cartilage susceptible to further mechanical erosion. Accordingly, hypertrophic differentiation of articular chondrocytes will facilitate degradation in OA. Moreover, we conclude that hypertrophic-like differentiation of articular chondrocytes is one facet of, and probably contributes to, OA pathology in this model.
Preclinical studies are often limited by time or financial constraints. Consequently, accelerating pathology with FM (as presented here), while maintaining similarity to human OA, will maximize productivity in preclinical studies. Of course, no OA model can reproduce human OA pathology identically [81
]. For example, the NM animals used in this study did not develop osteophytes, which are known to develop in human knee OA [82
]. Interestingly, the use of FM induced osteophyte formation. Such limitations must be considered when designing and drawing conclusions from a preclinical OA model. Instead, OA models are highly useful for investigating specific properties of OA. For example, this model accurately mimics the articular cartilage degradation and subchondral changes that are observed in many types of human OA, such as post-traumatic arthritis [84
]. Although we were unable to monitor disease progression in individual animals over time (because of sacrifice for histology), we now have a predictable index of pathogenesis in this model. We propose that future studies on disease progression in the joint should use the 2-week to 8-week time points for early, and 12-week to 20-week time points for late stages of OA, and are most relevant to human OA when combined with FM in this model. Thus, future studies will require only end-point histology, and disease progression may be monitored solely with in vivo
techniques such as high resolution magnetic resonance imaging [85
] and micro-CT.