Osteoarthritis (OA) is a common musculoskeletal disease characterized by progressive structural damage to the joint leading to symptoms such as joint stiffness and pain [1]. OA affects the whole joint and involves both cartilage and subchondral bone degeneration, and synovial inflammation. While the disease is characterized by cartilage degradation, the role of subchondral bone in the development and progression of OA has received increasing recognition in recent years [2–4]. A high degree of subchondral bone turnover was shown to predict subsequent joint space narrowing in OA patients using scintigraphy [5] and increased subchondral bone turnover in OA patients was demonstrated using biomarkers [6]. Bone attrition, which reflects subchondral bone loss and flattening, probably due to inadequate bone quality, is associated with bone marrow lesions [7], cartilage loss [8] and structural abnormalities such as malalignment [9]. Alterations in subchondral bone turnover seem to occur early in OA and may precede cartilage lesions [10]. However, it is also well recognized that there is extensive cross-talk between subchondral bone and articular cartilage and early biomechanical changes in one compartment are likely to affect the other [11]. Progress has been made in understanding the molecular signaling mechanisms between these two compartments, but they are not yet completely understood. Importantly, it has been shown in clinical trials that agents that inhibit bone resorption such as strontium ranelate [12], risedronate [13] and calcitonin [14] also show positive effects on OA-related endpoints such as joint space narrowing and on patient reported outcomes such as WOMAC scores in Phase II studies. However, either inconsistent efficacy or safety concerns have precluded approval of these agents for OA.

Cathepsin K is predominately expressed in osteoclasts and is a key enzyme involved in bone resorption through cleavage of type I collagen [15]. Cathepsin K is also expressed in chondrocytes in cartilage, where it can cleave type II collagen and aggrecan, the main components of the cartilage matrix [16–18]. Patients with pycnodysostosis, a disorder caused by inactivating mutations in the cathepsin K gene, exhibit abnormally dense bone (osteopetrosis), and this effect is reproduced in transgenic mice that are deficient in cathepsin K [19, 20]. Selective cathepsin K inhibitors increase bone mineral density in ovariectomized monkeys [21, 22] and clinically in post-menopausal women with osteoporosis [23, 24]. Furthermore, the most advanced cathepsin K inhibitor, odanacatib, reduced the incidence of vertebral and hip fractures in postmenopausal women in a Phase III study [25].

While numerous studies have demonstrated that cathepsin K inhibitors can provide beneficial effects on osteoporotic bone, less is known about the role of cathepsin K on subchondral bone and articular cartilage. Available data in preclinical models of joint degeneration support an active role for cathepsin K in the disease process, as demonstrated in studies using experimental cathepsin K inhibitors. Connor et al. [26] showed that the cathepsin K inhibitor SB553484, albeit with poor selectivity versus other cathepsins, exerted cartilage protection in dogs subjected to partial medial meniscectomy. Hayami et al. [27] demonstrated that the selective cathepsin K inhibitor L-006235 reversed subchondral bone loss in rabbits subjected to anterior cruciate ligament transection (ACLT) and that the subchondral bone protection was associated with some protection of cartilage. However, neither of these two compounds has progressed into clinical development. Thus, it is difficult to interpret if the doses used, the exposures reached and the effects seen in the preclinical models bear any relevance for a potential therapeutic effect in clinical OA.

MIV-711 is a highly potent and selective cathepsin K inhibitor [28] currently in Phase II for OA. The results from the initial Phase IIa study were recently reported [29]. Knee OA patients receiving once daily treatment with MIV-711 for 6 months demonstrated benefit on joint structure, with significantly lower increases in bone area and cartilage thinning in the diseased knee, as assessed by magnetic resonance imaging (MRI), compared to patients who received placebo. The current article summarizes the effects of MIV-711 on joint pathology in rabbits subjected to ACLT and dogs subjected to partial medial meniscectomy, two surgical models of OA. The doses of MIV-711 used in the animal models were aimed to produce clinically relevant exposures that are known to be safe, have been shown to engage cathepsin K in healthy volunteers and post-menopausal women and are intended to be reached in clinical studies in OA patients.

The current article demonstrates that the potent and selective cathepsin K inhibitor MIV-711 reduces biomarkers of bone resorption and cartilage degradation in the ACLT model in rabbit and the partial medial meniscectomy model in dog. The doses of MIV-711 used resulted in near complete protection of subchondral bone in rabbits. MIV-711 also attenuated cartilage lesions in the dog model and decreased cartilage swelling in the rabbit model. The potency-corrected exposures of MIV-711 reached in these two preclinical models of OA were in the same range as exposures reached in human and that have been shown to be safe and well tolerated. The reduction of biomarkers in dog and rabbit were of similar magnitude as in human, thus offering a translational potential for MIV-711.

The ACLT model in the rabbit is characterized by subchondral bone loss together with cartilage swelling and focal degradation [31–33]. In the current study, ACLT surgery significantly elevated HP-1 levels, which may be indicative of an early increase in bone resorption, either due to the induced joint instability or altered loading patterns on the knee joint after surgery [33]. MIV-711 reduced both HP-1 and CTX-II biomarkers in a parallel and dose-dependent manner, demonstrating successful target engagement of cathepsin K at the doses used in the rabbit. MIV-711 exposure was consistent throughout the 7-week dosing period. The effect of MIV-711 on the biomarker levels, measured as % of the vehicle, decreased numerically over time, but this was more related to decreases in biomarker levels in the vehicle group rather than loss of target engagement.

Analysis using µCT showed that the therapeutic administration of MIV-711 reversed the effects of ACLT on most subchondral bone parameters in the rabbit, with administration starting 7 days after surgery, at a time point where the HP-1 levels were close to maximal. Interestingly, in most cases, the effects of MIV-711 on the subchondral bone plate were not dose-dependent, with complete reversal of bone loss observed already at the low dose. Similar effects on the subchondral bone plate were seen with the low and high doses of MIV-711, even though the high dose resulted in tenfold higher exposures and twofold larger reductions in the HP-1 levels. In contrast, it appeared that there were some dose-dependent effects on trabecular bone, with the higher dose of MIV-711 producing more consistent protective effects. This suggests that relatively low doses of MIV-711 are sufficient for protective effects on the subchondral bone plate while higher doses are required for protection of underlying trabecular bone. In dog OA models it has been suggested that bone loss in the subchondral bone plate correlates well with cartilage changes and is thus more intrinsically linked to the OA process than the trabecular bone loss which may reflect less loading of the destabilized knee joint [34, 35].

Although areas of focally eroded cartilage were present in response to ACLT in rabbits, the current study shows that ACLT also evoked increases in cartilage thickness and volume, in all four areas examined, when quantified by µCT. Our results are in line with recent data demonstrating increases in the femur and tibia cartilage thickness at 8 weeks following ACLT in rabbits using µCT [36]. Similar changes in cartilage volume have been demonstrated in partial medial meniscectomy [32] and ACLT [33] rabbit models of OA using MRI. The increase in cartilage volume has been attributed to cartilage swelling and has been suggested to precede the loss of subchondral bone [32]. Interestingly, cartilage swelling was numerically reduced by both doses of MIV-711 in all areas examined although the effects only reached statistical significance in the medial cartilage of the tibia plateau with the low dose. Since there was a lack of dose-dependency and cartilage swelling was not completely reversed, it may suggest that the reduced cartilage swelling is indirectly due to improved quality of the underlying subchondral bone plate. If cartilage swelling precedes subchondral bone loss as in the study by Calvo et al. [32] and if the subchondral bone is the main target for MIV-711, then it is not surprising that the subchondral bone plate was wholly protected while cartilage swelling was merely attenuated. Unlike µCT which quantifies all articular cartilage, histological assessment depends on the specific section analyzed. Since the cartilage parameters analyzed using Mankin scores were very focal, we observed large degrees of variation precluding any detection of efficacy with MIV-711. An increased osteophyte volume in ACLT rabbits was quantified using µCT and histomorphometry but was not affected by MIV-711 treatment. The effects of anti-resorptive drugs on osteophyte formation in experimental OA models have been variable (see [37] for a review). For instance, in the dog ACLT model, the bisphosphonate NE-10035 was shown to inhibit trabecular bone resorption but not affect osteophyte formation when treatment was started the day after surgery [38], while alendronate inhibited osteophyte formation in the rat ACLT model when treatment was started before surgery [39]. This could be due to that osteophyte formation can be detected quickly after ACLT (within days) and that earlier interventions or longer treatment times may be required to observe an effect [37]. However, Hayami et al. [27] showed that cathepsin K inhibition could reduce osteophyte formation using a similar treatment protocol as in the current study.

Unlike the ACLT model in rabbits and dogs, the partial medial meniscectomy model has limited effects on subchondral bone and is more reflective of a model of cartilage degeneration [32, 40]. In the current dog study, neither urinary CTX-I or CTX-II levels increased in response to the surgery, similar to previous findings [26]. In contrast, urinary CTX-II levels were decreased in vehicle-treated animals after 28 days. This may be due to the age of the animals (6–8 months) in this study. Although growth plates were closed at this age, there could still be ongoing growth plate activity giving rise to relatively high CTX-II levels which would subsequently decrease with age and during the study [41]. MIV-711 reduced the CTX-I and CTX-II biomarkers measured in urine by 80–86% at the sacrifice which is consistent with the effects observed by the cathepsin K inhibitor SB553484 [26]. Concentrations of MIV-711 were also detected in synovial fluid, and CTX-II levels in this matrix were also reduced in response to MIV-711 treatment. The concentrations of MIV-711 in synovial fluid were low, but this is to be expected since the exposure was measured at trough, i.e. 24 h after the dose. Interestingly, significantly higher MIV-711 levels were detected in the injured joint compared to the contralateral control joint. This might be due to increased vascularization of the injured joint enabling more MIV-711 to reach this site, or to a higher degree of partitioning of non-basic MIV-711 into the joint where a pH as low as 5.5 has been measured in OA cartilage surface [17]. However, the relevance of increased MIV-711 exposure in the injured joint is difficult to appreciate concerning efficacy since the CTX-II reduction was similar in both joints.

Macroscopic and microscopic lesion scores in femur condyles and tibial plateaus in the dog were consistently reduced in response to MIV-711 treatment although the reductions were of borderline statistical significance when compared individually. Since many of the 101 variables analyzed were correlated, principal components analysis was used to derive scores that are combinations of the correlated individual variables. The first two scores explained 26 and 15% respectively of the total variance in the dataset, and both were significantly different between treatment groups. The first score was derived primarily from gross and histological morphological variables, and the second from PK and biomarker data, indicating that in this analysis of the aggregated data, treatment had a significant effect on both morphologic and biomarker endpoints. The magnitude of reduction at the macroscopic level was similar to the data observed with the cathepsin K inhibitor SB553484 [26]. However, although SB553484 is almost 30-fold more potent than MIV-711 against canine cathepsin K (0.055 nmol/L vs. 1.5 nmol/L), the daily doses of SB553484 were sixfold higher, and plasma exposures were not reported. Also, SB553484 has only marginal selectivity towards cathepsin K versus other cathepsins (15-fold, vs. > 1300-fold for MIV-711) making the contribution of cathepsin K in these two studies difficult to compare. In the study by Settle et al. [42] the protective effects on cartilage were more significant when using the MMP-13 inhibitor PF152. To our knowledge neither SB553484 nor PF152 has progressed into clinical development, thus making a translation of the preclinical data to the clinical relevance of the doses used and the effects achieved in the preclinical models difficult. In the current study, however, the potency-corrected exposures of MIV-711 in the dog are clinically relevant, since they are known to be safe, have been shown to engage cathepsin K in healthy volunteers and post-menopausal women and are intended to be reached in ongoing human studies in OA patients. Since the dog partial medial meniscectomy model has limited bone loss [32, 40], the attenuated cartilage degradation with MIV-711 may suggest direct effects of cathepsin K on cartilage per se. Indeed, cathepsin K can cleave collagen type II in vitro [18, 43]. However, it cannot be excluded that both an indirect effect on bone and a direct effect on cartilage could be involved.

Some anti-resorptive drugs have demonstrated encouraging signs of efficacy in clinical trials in OA and have also been effective in preclinical models of OA. The preclinical effects on bone and cartilage by these agents are in line with our data. Low doses of alendronate prevented bone loss in the estrogen-deficient ovariectomized (OVX) model in the rat, but did not prevent cartilage degradation in ACLT rats [39], and instead higher doses of alendronate were required to demonstrate protection of cartilage. Studies with the cathepsin K inhibitor L-006235 reveal a similar pattern, with complete protection of bone seen in OVX rabbits and ACLT rabbits using a dose of 10 mg/kg [27, 44], while higher doses of 50 mg/kg were needed to show some degree of cartilage protection [27]. In the current article, protective effects on the subchondral bone plate in the rabbit were achieved at relatively low plasma exposures and relatively low effects on biomarkers as summarized in Fig. 8. These exposures seemed to also offer attenuated cartilage swelling in the rabbit model. Higher exposures and more prominent effects on biomarkers seemed to be required for trabecular bone protection and consistent cartilage protective effects as seen in the dog model. Thus, it is essential to consider not only the agent and its mechanism of action but also the dose used, the exposure reached, and the target engagement achieved. Taken together, our results together with others suggest that with anti-resorptives, like bisphosphonates and cathepsin K inhibitors, relatively low doses offer bone protection while higher doses may be required for cartilage protection. If this reasoning is correct, then it is important to evaluate bone-acting agents like cathepsin K inhibitors at doses that are sufficiently high to have a meaningful effect on biomarkers of cartilage degradation, and that doses that are effective for the treatment of osteoporosis may be too low to achieve optimal effects in OA patients.

The effects of MIV-711 on biomarkers preclinically and clinically provide considerable translational value. The bone resorption biomarker HP-1 has not been used much in clinical studies, but rather as an alternative bone resorption biomarker in rabbits since conventional CTX-I assays cannot be used due to species differences. Nonetheless, urinary HP-I levels have been shown to be highly correlated to urinary CTX-I levels and treatment with alendronate reduced CTX-I and HP-I levels in man to a similar degree [45]. In the study by Hayami et al. [27], a dose of L-006235 providing a 60% reduction of HP-1 was sufficient for the protection of bone loss as assessed by histomorphometry. Our data suggest that even lower reductions of HP-1 translate into structural improvements on subchondral bone in rabbits. Figure 8 shows that these exposures can be reached in human and at reasonable doses.

The rapid degenerative process induced in the surgical models used in the current study is unlike the slow progression of disease that characterizes OA, and these rapid changes may, therefore, limit the efficacy of anti-resorptive compounds like MIV-711. However, it is difficult to speculate as to what degree of joint protection is required for translation into a clinical effect. Higher degrees of bone and cartilage protection could perhaps have been reached by further increasing the dose of MIV-711. However, the fact that the decreases in biomarkers were near maximal suggests that there is likely to be a limited additional benefit on the joint structure by further increasing the doses of MIV-711. Lack of clinically relevant exposures is often a shortcoming in preclinical studies trying to provide translational relevance to humans [46]. As shown in Fig. 8, MIV-711 has the potential to reach exposures in man that demonstrated structural protection in these preclinical models, and its effect on biomarkers enables the doses to be selected from early human studies to be tested in the longer-term clinical studies that will be required to show a disease-modifying effect in OA patients. In fact, the recent results from the initial Phase IIa study with MIV-711 in knee OA patients demonstrated significant effects on bone and cartilage structural endpoints (as assessed by MRI) after 6 months and the beneficial effects on the joint structure were associated with substantial reductions in the CTX-I and CTX-II biomarker levels [29]. The results suggest that the structural protection observed by MIV-711 in the preclinical OA models indeed translates into disease-modifying effects in OA patients.

The potent and selective cathepsin K inhibitor MIV-711 attenuates joint pathology and reduces biomarkers of bone resorption and cartilage degradation in experimental models of OA. Similar exposures and changes in biomarker levels can be achieved in man and thus supports the potential of MIV-711 as a disease-modifying agent for the treatment of OA.