Abstract
Musculoskeletal side effects are a widely reported consequence of administration of particular matrix metalloproteinase inhibitors (MMPi) in clinical trials. We describe here histopathological findings during dog studies with a fairly selective MMPi AZM551248, that are consistent with these human clinical changes. They were characterized by a dose-and time-dependent formative connective tissue alteration we have termed “fibrodysplasia.” The most sensitive site was the subcuticular connective tissue, although musculoskeletal tissues were also extensively involved. In the subcutis, changes occurred initially around pre-existing blood vessels, but then more diffusely. There was proliferation of cells showing myofibroblast differentiation identified by elevated levels of alpha-smooth muscle actin, fibronectin, and transforming growth factor β, and the deposition of collagen type III with a lesser quantity of collagen type I. On longer-term administration at lower doses, there was evidence of active fibrodysplasia arising and resolving during the dosing period, resulting in the multifocal deposition of mature collagen. Although there was organ specificity, essentially identical changes occurred at multiple connective tissue sites. We conclude that MMPi-induced fibrodysplasia in animals and, by inference, musculoskeletal side effects in humans are potentially diffuse connective tissue disorders.
Keywords
Introduction
The matrix metalloproteinases (MMPs) are a family of approximately 30 zinc-and calcium-dependent enzymes that enable turnover of the extracellular matrix in normal physiology and pathology (Chang and Werb 2001; Clark and Parker 2003; Coussens, Fingleton, and Martisian 2002; Lohi et al. 2001; Murphy et al. 2002). They are the only mammalian proteinases that can specifically degrade triple helical collagens at neutral pH, although they have different specificities for the fibrillar collagens types I, II, and III (Davidson et al. 2006; Knauper et al. 1996). Matrix metalloproteinases also act on nonmatrix substrates, including cell-surface and matrix-bound growth regulators, releasing them from stores, and coordinated regulation of MMPs and TIMPs (tissue inhibitors of metalloproteinases) govern the cleavage and release of many important growth factors, including TGF-β (Chang and Werb 2001). Matrix metalloproteinases are secreted by numerous cell types and play sophisticated roles in modulating normal cellular behavior and cell–cell communication (McCawley and Matrisian 2001), although substrate specificity varies. On the basis of this specificity and other molecular structural factors, vertebrate MMPs can be divided into at least six different groups (McCawley and Matrisian 2001; Visse and Nagase 2003).
Matrix metalloproteinase activity is implicated in a number of disease processes, including rheumatoid arthritis, osteoarthritis, osteoporosis, peridontal disease, metastatic tumorigenesis and neoplastic growth, aneurysm, atherosclerosis, and heart failure (Peterson 2004, 2006), and small-molecule MMP inhibitors (MMPi) have been investigated by pharmaceutical companies for the treatment of such diseases (Hodgson 1995). However, there have been clinical trials where a number of these agents failed as a result of musculoskeletal side effects (MSS). These trials used drugs that included Marimastat, RS-13-0830, CGS-27023A, PG-116800, and BB-2983. Musculoskeletal side effects resulting from these agents were characterized by a tendonitis-like fibromyalgia with joint pain, stiffness, edema, skin discoloration, and reduced mobility. The symptoms often started in the small joints of the hand and in the shoulder, and if dosing continued, symptoms spread to involve other joints (Krzeski et al. 2007; Peterson 2004, 2006; Rosemurgy et al. 1999; Shalinsky et al. 2000). In addition to arthralgia and joint stiffness, subcutaneous skin thickening of the palmar surface of the hands associated with contracture of the digits has been observed (Tierney et al. 1999). These changes were compared to Dupuytren’s contracture, a palmar fibromatosis characterized by the formation of dense subcutaneous fibrous bands leading to flexion deformity at the metacarpophalangeal joints.
Clinical observations from animals during preclinical toxicology studies appear consistent with the MSS reported for humans. Marimastat-treated rats exhibit signs of compromised ability to rest on their hind feet, gait changes including reluctance to move, and hind paw swelling (Renkiewicz et al. 2003). Histological examination revealed physeal and subphyseal trabecular thickening, synovial hyperplasia and hypertrophy, pannus formation with lymphocyte infiltrates and fibrin within the joint space, fibrosis of perisynovial adipose tissue, and other soft tissue fibroplasias. Except for growth plate changes, almost all other joint-related histopathological changes observed in rats have also been seen in monkeys (Drummond et al. 1999; Renkiewicz et al. 2003).
We have evaluated a number of MMPi during preclinical safety studies in which we have more fully characterized the histopathological changes. A consistent pattern of abnormal connective tissue fibrosis, which we have termed “fibrodysplasia,” has been observed during these investigations, although site specificity differed between the different species. In this report, we characterize the fibrodysplasia seen in the dog with one of these compounds: AZM551248, an aryl piperazine reverse hydroxamate. It has excellent pharmacodynamic and pharmacokinetic properties and is a potent inhibitor against a number of matrix metalloproteases including MMP2, MMP8, MMP9, MMP12, and MMP13 (unpublished data).
Materials and Methods
Test Materials
AZM551248 was supplied by the formulation and analytical support group, AstraZeneca, Alderley Park. It was formulated in water containing 0.5% w/v hydroxypropyl methylcellulose and 0.1% w/v polysorbate 80. Test items were tested for identity, and the purity was determined in all studies to be above 97%. The MMPi selectivity profile was assessed by in vitro recombinant human MMP assays using a FRET-based synthetic substrate. IC50 values were found to be: MMP2-0.1nM; MMP8-0.5nM; MMP9-0.9nM; MMP12-0.8nM; MMP13-0.7nM (unpublished data). Activity against other tested MMPs including MMP14 were of an order of magnitude lower (at least twenty-fold).
Animals and Treatments
The studies were planned in accordance with the standards of animal care and ethics described in “Guidance on the Operations of the Animals (Scientific Procedures) Act 1986” issued by the U.K. Home Office. They were conducted so that any clinical expression of toxicity remained within a moderate severity limit as described in guidelines agreed with the U.K. Home Office inspector. Beagle dogs, at least eleven months of age, were obtained from the Dog Breeding Unit, AstraZeneca, Alderley Park. They were group housed with access to an exercise area with others of the same sex and group for the major part of each day, and acclimatized for at least four weeks before the first dose.
A set weight of Teklad Dog Diet 2021 supplied by Harlan UK Ltd. was offered within one hour of the end of dosing each day (available for two hours), and water from the site drinking water supply was freely available. Each animal was uniquely identified within studies by an animal reference number and a microchip implanted in the ear.
Three separate studies were performed: a one-month pilot study to provide a preliminary assessment of toxicity to set doses for the three-month dog study; a three-month general toxicity study that defined the general nature, severity, and location of target organ pathology; and a time-course study to fully characterize the development of the fibrodysplastic changes. Details of the dosing and necropsy schedules for the three studies are included in Table 1.
The formulated test or control materials were dosed by oral gavage once daily up to and including the day of necropsy between 8:00 a.m. and 12:00 p.m. Animals were generally inspected at least twice daily and clinical signs recorded, and they were weighed once weekly, including during the prestudy period. In the time-course study, skin thickness in the dorsal cervical region was measured using calipers on days −7, −3, −1, 2, 4, 6, 8, 12, 14, and 16. For the one-and three-month studies, standard plasma chemistry, urinary chemistry, and hematological assessments were made at periods throughout the studies, including (for the three-month study) at the end of the three-month recovery period. For the time-course study, blood and urine samples were collected throughout the study for specific fibrodysplasia biomarker assessments (to be reported separately).
Necropsy and Histology
For the one-month study, major organs were sampled at necropsy. Also sampled were synovial membrane from the femorotibial joint, femur head, full-depth skin/subcutaneous tissue samples (sections including the epidermis, dermis, and deeper connective tissues generally down to the deeper musculature) from the dorsal cervical and ventral abdominal regions, and skeletal muscle (longitudinal sections) from the left and right gastrocnemius with calcaneal tendon, left and right cranial tibial with associated tendon, and left and right quadriceps with patella tendon. In the three-month study, a comprehensive list of tissues was sampled from all animals at necropsy consistent with international guidelines for repeat dose toxicity studies. Additional skin/subcutaneous tissue (as listed for the one-month study but including left and right hind limb adjacent to calcaneal tendon) and skeletal muscle samples (as for the one-month study) were taken. For the time-course study, kidneys, skeletal muscle (as for the one-and three-month studies), skin/subcutaneous tissue (left and right dorsal cervical, left and right ventral abdominal, left and right dorsal lumbar), and synovial membrane (left and right from the femorotibial joint) were the only tissues sampled.
Tissues were fixed in buffered 10% formalin (twenty-four to forty-eight hours) and processed to wax blocks. For the one-and three-month studies, single sections from each block were prepared and stained with hematoxylin and eosin (H&E) for examination by light microscopy. For the time-course study, similarly stained single sections of kidneys, muscle, and synovial membrane were prepared. For the skin/subcutaneous tissue samples, four sections were cut from each sample at approximately 500 μm steps. Additionally, isolated (at necropsy) subcutaneous connective tissue samples from the six skin sites listed above were similarly prepared as single sections for light microscopy.
The severity of any changes was graded subjectively: minimal, mild, moderate, or severe.
Electron Microscopy
Subcutaneous connective tissue from the left and right dorsal cervical skin regions of each time-course study dog and selected animals from the three-month study were immersion-fixed in 2.5% glutaraldehyde fixative. Glutaraldehyde-fixed samples were postfixed in 1% osmium tetroxide and processed to Araldite resin blocks. From selected samples (detailed in results section), thin 70-to 90-nm resin sections were cut and stained using uranyl acetate and lead citrate. Ultrastructural morphology was examined on a Hitachi H7100 transmission electron microscope using a 75 kV accelerating voltage.
Immunohistochemistry
To assess the utility of specific immunohistochemical markers for characterizing/detecting fibrodysplastic tissue, sections of selected formalin-fixed, paraffin-embedded skin/subcutaneous connective tissue from the time-course study (generally dorsal cervical, but also other selected sites from four control dogs and from dogs showing frank fibrodysplasia: three at day 11, two at day 14, and one at day 17) were immunohistologically stained for the presence of specific proteins: vimentin, desmin, alpha smooth muscle actin, Ki-67, cleaved caspase 3, MAC387, c-lamin A, TGF beta 1, ED-A fibronectin, collagen type I, and collagen type III. Sections were cut at 4 μm onto “Colorfrost Plus” glass slides (ThermoShandon, 9991002). They were dried at 37°C overnight, then dewaxed in xylene and rehydrated through graded alcohols. Sections were then washed in tap water and manually pretreated as indicated in Table 2 to enable antigen retrieval. Immunohistochemistry (IHC) was carried out on a Labvision LV-1 Autostainer using a standard polymer detection method. Briefly, sections were washed with Tris-buffered saline with 0.05% Tween (TBST), then incubated with a 3% solution of hydrogen peroxide (diluted in TBST) to quench endogenous peroxidase activity. Sections were washed in TBST, then a 5% solution of normal goat serum (Dako, X0907) diluted in TBST was applied for twenty minutes. Normal serum was blown off each section, then primary antibody was applied for between 30 and 120 minutes at room temperature, or overnight at 4°C, at the appropriate concentration (see Table 2). Reagents were diluted in TBST where required. Following a further two washes in TBST, antibody was detected using either a mouse or rabbit Envision Polymer-HRP Kit (Dako, K4007and K4011, respectively) according to kit instructions for thirty minutes. Immunostaining was visualized with DAB (supplied with Dako Polymer Kit). At the end of the immunostaining procedure, sections were removed from the Labvision Autostainer and counterstained manually in Carazzi’s hematoxylin (Clintech, 64230). Sections were dehydrated through graded alcohols, then two changes of xylene, and coverslipped with Hystomount.
To further characterize the temporal development of fibrodysplasia using markers of potential utility (determined from the initial immunohistochemical assessment), subcutaneous connective tissue samples (generally dorsal cervical, but also selected tissues from other sites) from all five control dogs, all five dogs from day 8, three dogs from day 11, two dogs from day 14, and four dogs from day 17 were immunohistochemically assessed, or reassessed, for collagens type I and III; the thin contractile filament alpha smooth muscle actin; the cell proliferation marker Ki-67 (present during all active phases of mitosis); and EDA fibronectin (associated with myofibroblast differentiation). The tissue samples assessed were selected for their presence of normal subcuticular loose connective tissue, fibroplasia in an undosed dog, and a range of severity and temporal development of fibrodysplasia.
Results
Clinical Findings
During the one-month study, dogs at all dose levels exhibited reddening of the abdomen from week 2. Tightening of the skin characterized by a generalized loss of mobility of the skin relative to underlying tissue, particularly notable in the region of the scruff, was observed in both dogs dosed 50 mg/kg/day during week 3 and in both dogs dosed at 10 mg/kg from the start of week 4. In the three-month study, the only clinical sign considered to be related to treatment was an alteration in skin tone, of a similar nature to that seen during the one-month study, in all animals dosed 2 and 5 mg/kg/day. This finding was noted from days 68 or 78 (5 mg/kg/day and 2 mg/kg/day, respectively) until termination and persisted for four or ten weeks following cessation of dosing (males and females, respectively; recovery animals). During the time-course study, a red discoloration of the skin, particularly on the ventral surface, was seen from day 7 with an increasing incidence subsequently. An adherence of the skin to deeper tissues was observed in a small number of animals from day 11 (four dogs affected in total), and the skin of two dogs appeared “hot” from day 12. Following calliper measurement, two dogs showed increases in skin thickness from days 12 and 16, respectively (up to 1.9-fold increase compared with the prestudy average).
By scheduled termination of the one-month study, animals dosed at 50 mg/kg/day showed some hematological changes, including decreases in hematocrit (up to 18%), and in hemoglobin and red cell count (up to 30%) compared to prestudy. There were also slight decreases in creatinine and total calcium, moderate decreases of total protein, albumin, ALT, AST, and ALP, and slight increases of glucose, potassium, and inorganic phosphate. Animals dosed 10 mg/kg/day showed an increase in platelets by day 27 (up to 84%); slight decreases of albumin, creatinine, and ALT; and some decrease of total protein. Animals dosed 2.5 mg/kg/day showed no changes. During the three-month study (up to 5 mg/kg/day), no important changes in hematology and plasma chemistry were seen. For the time-course study, there were no dose-related changes in the standard parameters measured in urine chemistry.
Necropsy/histopathology
One-month Study:
At necropsy of dogs dosed 10 and 50 mg/kg/day, there was adhesion of the skin to the underlying muscle associated with an extensive deposition of white/cream-colored dense fibrous tissue in the subcutaneous regions. This was particularly notable in the dorsal cervical region, but also in other areas, including adjacent to superficial tendons such as the calcaneal tendon.
On histological examination, fibrodysplasia was the only change attributable to administration of AZM551248. It was present in the skin, subcutis, synovial membrane, articular cartilage, ligaments, tendons, and associated connective tissues, as summarized in Table 3. Fibrodysplasia was characterized by the proliferation and accumulation of medium-sized to large fibroblasts forming abundant, dense, and irregular collagen sheets and bands resulting in the apparent loss and replacement of adipose tissue. The cellular component was generally relatively monomorphic and contained few other cellular infiltrates (Figure 1). Focally, particularly in deeper subcutaneous tissues, there was a variable infiltrate of mixed inflammatory cells, mainly monocytic/histiocytic, but with a variable polymorphonuclear cell component and some edema. The fibrodysplastic tissue showed an extensive distribution involving the more superficial and deep subcutaneous tissues, including connective tissue adjacent to deep subcuticular muscle, resulting in thickening of the affected regions. The dorsal cervical areas appeared more affected than tissues from the ventral abdomen (Table 3). The dermis was also affected, but to a more limited degree. Changes here were characterized by a diffuse increase in cellularity as a result of the proliferation of fibroblasts around the bundles of dense dermal collagen. Accumulations of mononuclear inflammatory cells in the superficial dermis were common.
The musculoskeletal tissues were also affected at 10 and 50 mg/kg/day (Table 3). Tendons were involved at multiple sites, most severely at the aponeuroses, but also more distally along their length. There was replacement of the normal dense, relatively acellular regular linear arrays of collagen by abundant large, proliferating fibroblasts depositing irregularly oriented collagen bundles (Figure 2a). The local loose connective tissue associated with the muscle, tendons, and other tissues in these areas were also involved, and where tendons were located in proximity to other tendons, fusion occurred. When close to subcutaneous tissues, fibrodysplastic tissue extended from the tendon to the dermis. Intra-articular ligaments (from the femoral head) were similarly affected.
The subsynovial connective tissues sampled from the femorotibial joint showed a variable severity of fibrodysplasia (Table 3), and occasional hemorrhage was observed. In some cases, the fibrodysplasia was associated with a moderate synovial hyperplasia, synovial degeneration, and pannus formation characterized by a fibrinous exudate overlying the hyperplastic or degenerate membrane with cellular debris in the joint space (Figure 2b).
Three-month Study:
No macroscopic changes attributable to treatment were noted at necropsy following three months of administration at 1, 2, or 5 mg/kg/day.
Fibrodysplasia was the only notable histopathological change seen, and the distribution and incidence is summarized in Table 4. It was present with a dose-related incidence and severity in subcutaneous tissues. As in the one-month study, the dorsal cervical region showed a greater sensitivity than the ventral abdominal area. The active proliferative lesions were more focal than the changes seen at higher doses for shorter periods. They showed discrete bands or nodules of dense collagen deposits containing irregularly orientated activated fibroblasts (Figure 3), within a background matrix of loose “normal” connective tissue. Fibroblasts were quite variable in number, although generally more sparse than those seen at the earlier time points. Indeed, some foci of increased collagen density appeared relatively acellular. The dermis appeared unaffected. Fibrodysplasia was also occasionally seen in the tendons of animals dosed at 5 mg/kg/day. It was commonly present in the adjacent loose connective tissue and capsular region and was consistent with that seen in the subcutis. In some cases, it also involved a length of tendon and was characterized by a generalized increase in cellularity, with enlarged fibroblasts oriented along its length associated with somewhat irregular, more lightly stained collagen (Figure 4). No fibrodysplastic changes were detected in dogs dosed for three months at 5 mg/kg/day and left undosed for a further three months.
The kidneys of one dog dosed 5 mg/kg/day showed a moderate bilateral nephropathy with numerous groups of immature glomeruli, undifferentiated mesenchyme in the cortex and medulla, along with interstitial inflammatory cell infiltration, primitive ducts, and atypical tubular epithelium. These changes were consistent with renal dysplasia (Maxie and Newman 2007). However, a very notable feature was a marked and extensive deposition of dense irregularly oriented bundles of interstitial collagen in association with numerous active fibroblasts (Figure 5a). On immunohistochemical staining for collagen types I and III, the predominant collagen in subcuticular fibrodysplastic tissue and the renal interstitium was found to be type III, with a smaller proportion of type I. This dog had shown an elevation in plasma creatinine and urea prestudy (day −14 and day −4) and throughout (days 37 and 87) the study period (urea: 16.0 to 20.1 mmol/L; normal approx 5.0 mmol/L) (creatinine: 108 to 117 μmol/L; normal approx. 70 μmol/L).
Time-course Study:
At necropsy following eight or more days of treatment, the skin from a proportion of dogs (one at day 8, two at day 11, five at day 14, three at day 17) showed a gelatinous appearance of the subcutis and/or diffuse thickening and/or red discoloration.
Hematoxylin and Eosin Histology
The distribution, incidence, and severity of proliferative connective tissue changes following histopathological assessment is summarized in Table 5.
The only proliferative “fibroplastic” change observed in the skin/subcutis of control group dogs was in the dorsal lumbar region of one animal. Multiple small compact aggregates of large collagen-forming, fibroblast-like cells were present, generally focused on blood vessels, replacing subcutaneous adipose tissue. The pathogenesis of this change was uncertain, but it may have been a result of local trauma.
No fibrodysplastic changes were observed in the skin/subcutaneous tissue of animals dosed at 20 mg/kg/day for four days. By day 8, frank fibrodysplasia of minimal to moderate severity was present in the dorsal cervical or dorsal lumbar subcuticular connective tissue of two of five dogs. Focal and more widespread areas of loose connective/adipose tissue in the superficial and deep subcutis showed proliferations of medium-sized to large spindle-, strap-, or polygonal-shaped fibroblast-like cells, with numbers of mitotic figures and collagen deposition. Proliferations of minimal severity were often focused around pre-existing blood vessels (Figure 6). There was evidence of degeneration of adipocytes within these fibrodysplastic areas characterized by mature fat cells showing shrinkage, and infiltration by enlarged vacuolated histiocytic cells or occasionally more mixed inflammatory cell infiltrates and edema. The assessment of step sections from multiple sites illustrated that the changes did not affect the subcutaneous tissues diffusely at day 8; rather, they were multifocal and may not have been detected in some animals without thorough sampling.
Subsynovial connective tissue sampled from the femorotibial joint of one dog at day 8 showed a mild fibrodysplasia. One further dog showed a unilateral acute hemorrhagic synovitis, and another a bilateral chronic synovitis characterized by hyperplasia of the synovial membrane and lymphocytic infiltrates in the underlying tissues. Dermal connective tissues were generally unaffected at day 8, and tendons appeared unaffected.
From day 11 to 17, the subcutaneous tissues of all dosed dogs were affected by fibrodysplasia (Table 5), with the changes becoming more extensive and severe. All skin regions were affected, although the dorsal cervical region commonly showed the most severe changes. There was a progressive involvement of dermal connective tissues by expansion of the subcuticular lesion into the deeper dermis, and by the appearance of enlarged active fibroblasts sparsely throughout the superficial to mid-dermis, or more densely around hair follicles. For the dermis, the ventral abdominal region appeared most severely affected. Synovial connective and adipose tissues, and connective tissue adjacent to tendons, were also progressively involved.
Immunohistochemistry
A number of the immunohistochemical assessments showed no clear utility in specifically characterizing fibrodysplastic cells, or differentiating fibrodysplastic tissue from normal fibrous tissue or the focal fibroplasia seen in an undosed animal. These assessments included the most widely distributed intermediate filament vimentin, which showed a mild to moderate cytoplasmic activity in normal fibroblasts from all three tissues, although it was slightly less prominent in fibrodysplastic fibroblasts; the type III intermediate filament desmin (generally present in muscles cells from early in development) showed no clear positivity in any fibroblasts from all three tissues, although it was positive in arterial medial cells; the intermediate filament c-Lamin A showed no clear activity in any fibroblasts from the three tissues; and the macrophage marker MAC387 showed no clear positivity in any fibroblasts, although it was positive in monocytes and macrophages.
A few markers aided the characterization of fibrodysplastic fibroblasts and matrix but were not of potential utility for the detection of very early changes. For control subcuticular connective tissue, collagen stains showed a moderate activity for collagen type I and marked activity for collagen type III. For frank fibrodysplastic tissue at any time point, collagen staining was more variable, with collagen I showing a mild to moderate presence, and type III moderate to marked. Alpha SMA showed a minimal to moderate activity in fibroblasts from frank fibrodysplastic tissue, minimal activity in the control fibroplasia, and negative activity in the control inactive fibroblasts (Figure 6). Fibronectin showed no activity in fibroblasts from inactive control connective tissue, minimal activity in the control fibroplastic lesion, and generally mild to moderate activity in frank fibrodysplastic tissue (Figure 6). TGFβ 1 showed a generally mild presence in inactive control fibroblasts, but moderate presence in fibroblasts within fibrodysplastic tissue and the control fibroplasia (Figure 6).
Only one marker showed potential utility for the detection of “subhistological” fibrodysplastic changes. The cell proliferation marker Ki-67 showed nuclear staining of occasional fibroblasts (minimal) in the control subcuticular connective tissue, mild activity for fibroblasts in the fibroplastic lesion from the control dog, and mild to moderate activity in all frank fibrodysplastic lesions. However, dorsal cervical subcutis samples from two dogs dosed for eight days, showing no frank fibrodysplasia by H&E histopathological assessment, exhibited mild to moderate activity in groups of fibroblast-like cells focally, notably around blood vessels. The dorsal cervical subcutaneous connective tissue samples (left and right) from all five dogs dosed for four days, when assessed for Ki-67 activity, showed no areas of increased cellular proliferation.
Ultrastructure
Fibroblasts (fibrocytes) from the subcutaneous connective tissue of control dogs had attenuated cytoplasm with a diminished fine structure showing no evidence of activation, although they commonly showed a dilated endoplasmic reticulum containing amorphous material of a medium density. They were surrounded by regular arrays of uniformly banded collagen.
Activated fibroblasts from fibrodysplastic tissue showed an abundant cytoplasm with extensive rough endoplasmic reticulum (RER) and numerous vesicles often seen to contain apparent filamentous inclusions (Figure 6). Associated extracellular collagen showed apparently normal morphology.
Discussion
On evaluation of a number of MMPi in preclinical safety studies with multiple species, we have observed a consistent pattern of abnormal connective tissue fibrosis (unpublished data). During these studies, although preferential site specificity differed between species, the temporal, proliferative, and morphological character of the changes was essentially identical. As these changes are primarily an abnormal fibroplasia we have termed them “fibrodysplasia.” In the dog, the subcutaneous connective tissues appear most sensitive. In addition, our comprehensive assessment of the skin and subcutaneous tissues from a range of sites following administration of AZM551248 has shown that the dorsal cervical subcutaneous connective tissue is particularly sensitive. At 20 mg/kg/day, evidence of proliferative changes in the subcutis (increased expression of Ki-67) first appear around blood vessels after day 4 of treatment, but then more diffusely throughout the subcuticular connective tissues, and generally less markedly in the dermis. Indeed, at lower doses (5 mg/kg/day) after three months, the dermis was spared. Fibrodysplasia was characterized by the proliferation of large fibroblast-like cells in association with expansive arrays of irregular collagen. By day 8 (following 20 mg/kg/day), extensive changes were seen in a proportion of the treated dogs, but not all, illustrating an idiosyncratic response. At this and subsequent time points, the subcutaneous and dermal lesions were often severe, widespread, and occasionally associated with infiltrates of inflammatory cells and edema. By three months, after lower doses (2 and 5 mg/kg/day), the subcutis showed active fibrodysplastic changes of a character similar to those described above. However, step-sectioning of tissues revealed that they were generally multifocal, with some sections showing focal active lesions and others only relatively acellular increases in collagen density, indicating that active fibrodysplasia can occur at focal sites throughout the dosing period, with subsequent resolution leaving a residue of relatively acellular collagen. Following a recovery period of three months, active fibrodysplasia or an increased density of residual collagen was not detected histologically. However, resolved collagenous fibrodysplastic tissue would perhaps have a character indistinguishable from pre-existing collagen, resulting in difficulty in determining full recovery.
Ultrastructurally, active fibrodysplastic tissue was characterized by the presence of cells with an abundant cytoplasm and features indicative of marked activation and secretion of collagen. By immunohistochemistry, the cells expressed increased levels of transforming growth factor (TGF) β1, α-SMA, and EDA-fibronectin. These ultrastructural and immunohistochemical characteristics imply both secretory and contractile differentiation consistent with that seen in myofibroblasts during wound healing (Hinz et al. 2007; Gabbiani 2003); α-SMA has, in particular, become a reliable marker of myofibroblastic differentiation (Gabbiani 2003). During wound healing, at least three local events are required to generate α-SMA–positive differentiated myofibroblasts. These events are (1) accumulation of biologically active TGF β1 (the main myofibroblast inducer); (2) the presence of specialized extracellular matrix (ECM) proteins such as EDA-fibronectin (up-regulated by TGF β1); (3) high extracellular stress (owing to loss of ECM rigidity during ECM remodeling) (Desmouliere et al. 1993; Hinz et al. 2007; Gabbiani 2003; Serini et al. 1998; Tomasek et al. 2002). The active cells in the fibrodysplastic tissue show these required characteristics of myofibroblasts. Indeed, subcuticular site specificity may, in part, be a result of local levels of skin mobility and extracellular stress. A spontaneous subcutaneous fibroblastic lesion assessed in a control dog did not show the same level of increased activity of α-SMA, and no evidence of increased EDA-fibronectin, although activity of TGF β1 appeared similar. This finding further illustrates the degree of myofibroblast differentiation in the fibrodysplastic lesion. The extracellular collagen, although increased in content and appearing more “irregular,” retained a normal banding pattern. It was predominantly collagen type III with a smaller type I component. It is noteworthy that in granulation tissue of normal healing wounds, in fibrocontractive diseases, or in normal tissues subject to mechanical stress, collagen type I is replaced to a great extent by collagen type III (Gabbiani 2003; Gabbiani et al. 1976).
The distribution and activities of MMPs within the subcuticular connective tissues may be implicated in determining this particular site sensitivity in the dog, although normal cell and matrix turnover rates, and local levels of physical stress and mobility, as noted above, may be factors. Indeed, restraining dogs by the scruff during handling may be implicated in the increased sensitivity of this region to the induction of fibrodysplasia. However, in the dog and other species we have examined, including the rat, mouse, and guinea pig (unpublished data), and in monkeys (Drummond et al. 1999; Renkiewicz et al. 2003), the musculoskeletal tissues are also an important, or the primary target, and tendons, ligaments, and synovial are invariably involved. During the current investigations in the dog, tendons at one month (10 and 50 mg/kg/day) showed fibrodysplasia that particularly affected the aponeuroses. At three months (5 mg/kg/day), more diffuse changes to tendinal cellularity and collagen morphology were apparent, and lesions in the synovial membranes were often widespread particularly within the supporting connective tissues. Tendonitis-like musculoskeletal dose-limiting side effects (MSS) have hampered the efficacy assessment of numerous metalloproteinase inhibitors (MMPi) (Peterson 2006) in the clinic, and its character is entirely consistent with the musculoskeletal fibrodysplasia seen here in the dog. If they occurred in patients, the fibrodysplastic changes seen in the dog and other species, with associated disruption of tissue architecture in tendons, ligaments, and synovial tissues, would result in the clinical observations made during human trials.
The current studies, although illustrating a differential sensitivity of particular connective tissues to the induction of fibrodysplasia, also show widespread multiple-organ involvement. In addition, our preclinical experience with a number of MMPi in multiple species has shown that, at high dose, a range of connective tissues can be targeted, including those associated with the mesentery, heart, kidney, and aorta (unpublished data). We therefore consider fibrodysplasia in animals, and by inference MSS in humans, as potentially a diffuse connective tissue disorder. Also, as illustrated here by changes in the kidney of one dosed dog, it appears that preexisting fibrotic lesions may be exacerbated by fibrodysplastic changes. A preexisting nephropathy was indicated by elevated urinary enzymes, but aspects of the histopathological changes following dosing, particularly the extensive active interstitial fibrosis, were perhaps inconsistent with the diagnosed spontaneous renal dysplasia.
The mechanism of induction of clinical MSS and preclinical fibrodysplasia by particular MMP inhibitors is entirely uncertain. Specific or combination effects on various MMPs and their physiological activities have been implicated and then often excluded. These activities include inhibition of MMP-1, MMP-2, MMP-9, MMP-14, MT1-MMP, and sheddase activity; direct targeting of the catalytic Zn2+; and effects on cell surface and matrix-bound growth factors (Chang and Werb 2001; Drummond et al. 1999; Holmbeck et al. 1999; Johnson et al. 2007; Levi et al. 1996; Martignetti et al. 2001; Peterson 2004, 2006; Shalinsky et al. 2000; Whittaker et al. 1999; Yu and Stamenkovie 2000; Zhou et al. 2000). A full review is beyond the scope of this article (those listed above provide further discussion), but it is notable that AZM551248 and many of the other agents reported to induce connective tissue pathology preclinically or MSS clinically are relatively broad-spectrum/nonselective MMP inhibitors, which may indicate that the issue of fibrodysplasia/MSS can be avoided by improved selectivity.
As for fibrodysplasia in the dog, there appears to be a time dependency for clinical onset of the MSS condition. For example, in a study with Marimastat, MSS events requiring dose modification were not observed during the first twenty-eight days, but they occurred in a substantial number of patients under longer-term treatment (Rosemurgy et al. 1999). The mean time to this side effect was forty-five days, and it was considered largely reversible. Other clinical investigations have verified that MSS is dose and time related; involves joints in the hands, arms, and shoulders; and is reversible following discontinuation of dosing (Nemunaitis et al. 1998; Wojtowicz-Praga et al. 1998), although the exact sequence of events appears to depend on the drug used and the dosing regimen (Peterson 2004, 2006). Efficacy and MSS side effects also appear to be closely linked to drug exposure, as the plasma drug levels required for efficacy with Batimastat, Marimastat, CGS-27023A, and Prinomastat also produced MSS (Drummond et al. 1999; Hutchinson et al. 1998; Peterson 2004). Indeed, in many clinical trials, efficacy of MMPi may have been compromised because of attempts to avoid MSS coupled with an inability to adequately assess the therapeutic index, resulting in a dose selection beneath the minimal effective dose (Peterson 2006). At high doses of AZM551248 in the dog, clinical and histological changes appeared between days 4 and 8. However, at lower doses following three months of administration, although clear histological changes were apparent, no clinical findings were noted. These observations in dogs raise the concern that dysplastic connective tissue changes may occur following therapeutic dose before, or even in the absence of, any clinical sign. Also, if clinical signs such as MSS are seen, undetected changes may also be present in additional organs. It may be concluded that proportionate preclinical safety margins are required for the clinical development of MMPi with fibrodysplastic potential.