Treatment of Monolateral Legg-Calvé-Perthes Disease with Autologous Bone Marrow Mononuclear Cells in 32 Dogs

Legg-Calvé-Perthes disease (LCPD) was first identified in children by Legg, Calvé and Perthes in 19101–3; in dogs, the disease was first described by Tutt in 1935.4 In canines, the average time of disease onset is at 7 months of age, with no gender differences. In 10 to 15% of cases, LCPD is bilateral; the incidence is 2.8 cases per 1000.5 Legg-Calvé-Perthes disease has well-defined clinical and radiologic characteristics, although the aetiology and pathogenesis have not yet been completely revealed. The pathological features of LCPD are typical for avascular necrosis of bone tissue, followed by attempts to revascularization and bony remodelling of the femoral head. The most recent epidemiologic hypothesis in LCPD involves the expression

Introduction of a defect in endochondral ossification and considers the disease to be a type of osteochondritis. 6 Several treatment methods have been proposed in LCPD aimed at promoting blood supply, such as core decompression, or conservative approach such as physiotherapy to reduce pain, and femoral head and neck resection as treatment of choice. 7,8 Recently, many authors researching treatment options for avascular necrosis of the femoral head (ANFH) in humans have proposed the use of bone marrow (BM) [9][10][11] and BM-derived cells. [12][13][14][15][16][17][18] Bone marrow-derived cells have been employed in dogs, to verify their efficacy as a translational model for ANFH in humans. 19,20 Several favourable biological characteristics make bone marrow mesenchymal stem cells (BMSCs), an ideal therapeutic agent for bone regeneration. [21][22][23][24] The aim of this study was a retrospective evaluation of dogs with LCPD we treated with autologous bone marrow mononuclear cell (BMMC) implantation during the past 12 years.

Materials and Methods
Population Clinical records of 32 dogs affected by monolateral LCPD and treated with BMMC implantation between 2007 and 2019 were included.
All treatments were performed according to the Italian guidelines on the use of stem cells for clinical applications in veterinary medicine (Gazzetta Ufficiale (Italy) serie generale n. 277 del 26/11/2013) and each dog owner provided informed consent.

Clinical Examination
In preoperative and postoperative evaluations, we considered hip pain during passive movement of the coxofemoral joint, the degree of lameness, gait and findings of radiologic and computed tomography (CT) examinations. The diagnosis of LCPD was based on the following clinical findings: monolateral hip pain, lameness and radiographic lesions graded 1 to 3 based on the Ljunggren scale. 8 Cases of Ljunggren grade 4 and 5 were not included because of the severe modifications of the femoral head and neck.

Bone Marrow Harvesting
After sedation with acepromazine (0.02 mg/kg) and methadone (0.3 mg/kg), induction and maintenance of general anaesthesia were obtained with propofol (5 mg/kg). The hip region was prepared for aseptic surgery. Bone marrow was harvested from the iliac crest using an 18-gauge Jamshidi needle or, a 24-gauge epidural needle. Aspiration of bone marrow was performed using a 20 mL syringe containing heparin (2500 IU/20 mL) (►Fig. 1). Ten mL of BM was considered as the minimum volume for the procedure. If this amount was not reached with a single site, then another harvesting was performed in the opposite iliac crest.

Bone Marrow Processing and BMMC Isolation
Bone marrow was diluted 1:1 in phosphate buffered saline (PBS), then layered 1:1 on Biocoll separation medium (Ficoll density gradient 1.077 g/mL; Biochrom, Berlin, Germany) and centrifuged at 2000 rpm Â 30 minutes. The separated cells were counted using nuclear staining (0.1% methyl violet in 0.1 M citric acid). The cells were rinsed twice with PBS and then diluted in an adequate amount of fibrin glue (Tissucol; Baxter Spa, Rome, Italy).
To evaluate the capacity of the cells to form colonyforming unit fibroblasts (CFU-f), 100 μL of cells were seeded before and after density gradient centrifugation in two Petri dishes (100 mm diameter) in complete medium (Coon's F-12 medium, 10% foetal bovine serum, 100 IU/mL penicillin and streptomycin, 5% of a 200 μM solution of L-glutamine; Biochrom), and incubated at 37°C in a humid 5% CO 2 /air (carbogen) atmosphere for 2 weeks, with the medium replaced twice a week. The cells were then washed with PBS, pH 7.2, fixed with 4% buffered formalin and stained with 1% methylene blue in borate buffer (10 mM, pH 8. 8).

BMMC Injection
The dogs were re-anaesthetized using propofol (5 mg/kg). For injection of the BMMCs into the femoral head and neck, the BMMCs were suspended in 1 mL of fibrinogen. After injection of BMMCs þ fibrinogen, 0.3 mL of thrombin was immediately injected using the same needle, to clot the suspension and avoid the backflow. This procedure was conducted under radiographic imaging or CT guidance using an 18-gauge Jamshidi needle for insertion into the femoral head via an intertrochanteric injection or directly into the femoral head (►Fig. 2).

Follow-Up
In the postoperative period, in each dog antibiotic medications were administered for 7 days (Amoxicillin/Clavulanic acid 25 mg/kg q 12h). Nonsteroidal anti-inflammatory drugs were avoided post-surgically to exclude any possible interference with the BMMC effect. Orthopaedic, radiographic and CT examinations were performed at 3 weeks and 3 months after the procedure in all cases. Long-term radiographic follow-up at 1, 2, 4 and 6 years was available in selected cases.

Histology, Histochemistry and Immunohistochemistry
Among the total number of dogs treated, four had persistent pain and lameness at 3 months after treatment, indicating the need for femoral head and neck ostectomy. In these cases, the femoral head and neck were harvested and fixed in 10% buffered formalin, and then decalcified in a solution of 10% ethylenediamine-tetra-acetic acid in 0.1 M phosphate buffer, pH 7-8, for $8 weeks, changing the solution once a week. Ethylene-diamine-tetra-acetic acid decalcification is optimal for proteoglycan preservation and is highly recommended for sections destined for immunohistochemistry (IHC) 25 After paraffin inclusion, the same tissues were cut into 4-μm thick sections, placed on pre-treated slides (Bio-Optica, Milan, Italy) to promote adhesion, and dried overnight at 37°C To obtain controls, we retrieved paraffinized tissue samples from untreated dogs with LCPD who previously underwent femoral head and neck resections; for this analysis, femoral head-neck samples from three male dogs (a 7-month-old Papillon and two Yorkshire Terriers aged 9 and 10.5 months) with a history of severe clinical signs of pain and lameness were selected from the Veterinary Pathology Unit archives, University of Camerino. These samples had been obtained immediately following a surgical procedure in dogs that presented for elective surgery (femoral neck resection controls, FNRC).
Eight non-consecutive sections, obtained at 100-μm intervals, were measured per each femoral head (in both the histological study group [HSG] and FNRC groups). The following parameters were evaluated at the central level of the femoral head: cartilage thickness, proteoglycan content ratio based on safranin O staining intensities (PC), chondrocyte density (CD) and subchondral bone staining intensities (BI). The degree of user interaction varied from manually tracing objects to nearly complete computer automation. For cartilage parameters, particularly evaluation of the new cartilage regeneration process inside the femoral head, four sections were stained with safranin O 26 and counter-stained with light green. For bone parameters, four additional non-consecutive sections were stained using Herovici polychrome dye, 27 to evaluate the different tissues and for pre-collagen and collagen differentiation. All slides were stained with haematoxylin and eosin (H&E) for general morphological evaluation. Finally, the presence of probably derived BMMCs was immunohistochemically evaluated using mouse monoclonal antibody anti-CD34 (Clone BL-35C; Zymed Inc., San Francisco, California, United States). Specific primary antibody substituted with Tris buffered saline or non-immune serum was used as the negative control in immunohistochemical techniques. For IHC tests, sections were treated as described elsewhere. 28 Immunohistochemical evaluation was made for the presence of CD34þ cell expression in cartilaginous areas and interspersed or localized in perivascular fashion throughout the damaged area of the femoral head, assessed at 100Â magnification. A score of 0 (absence of antigen expression), 1 (weak and spotted antigen expression), 2 (weak but diffuse antigen expression throughout the entire specimen) and 3 (diffuse and strong antigen expression) was assigned in a semiquantitative evaluation of the immunohistochemical reaction. Computed tomography, expressed in micrometres, was measured at 20Â magnification, using an image-analysis system consisting of a light microscope (Carl Zeiss, Jena, Germany) attached to a Javelin JE3462 high-resolution camera and a personal computer equipped with a Coreco-Oculus OC-TCX frame grabber and high-resolution monitor. Computerized colour-image analysis was performed using Image-Pro Plus software (Media Cybernetics; Rockville, Maryland, United States). After manual delimitation of the cartilage between two reference marks (1 and 2) and the subchondral bone, the resulting area of cartilage was automatically segmented (using a threshold for black and white extraction with a green image), and the CT was calculated as the mean length of all the segments generated from each pixel situated on the border of the corresponding area of the cartilage.
The PC was approximated by quantification of the safranin O staining intensity (SOI) of the histological section. The area of measurement was centrally positioned between reference marks 1 and 2, and segmentation of the cartilage was automatically generated for measurement of the CT. The intensity was evaluated in the superficial (S) and deep (D) zones of the cartilage with manual delimitation at the upper margin of the D zone. Intensity was automatically measured as the optical density calculated on a greyscale obtained using the red component of the light passing through the section. To minimize the differences in safranin O staining between the sections and because PC varies differentially during experimental optics acquisition in the S and D zones, the following ratio was calculated: SOI in the S zone divided by SOI in the D zone; this justifies expression of the PC as a ratio.
The CD was evaluated at 200Â magnification with manual counting of the cells divided by automatic calculation of the cartilage surface. Chondrocyte density is expressed in number of cells/mm 2 . Cartilage parameters (CT and CD) were successively measured in each of the four non-consecutive sections stained with safranin O, and the bone parameter (BI) was measured in the four non-consecutive sections stained with Herovici polychrome dye and evaluated in terms of O.D. of different red colour variations. Finally, for CD34þ cells, the count was performed in four randomly selected cartilaginous fields, as follows: the number of IHC-positive cells was expressed as the number of stained cells in 100 randomly counted cells, as a percentage. For each dog, the data of each parameter were the mean of the four measurements obtained from the four sections. All sections were read blindly.

Statistical Analysis
Using MedCalc 14.8.1, parametric data were evaluated for normal distribution, and a comparison between the HSG and FNRC groups was performed with a one-way analysis of variance test. Nonparametric data were compared with a Kruskal-Wallis test. Significance was set at p < 0.05.
Dogs presented with different degrees of lameness, myoatrophy and shortening of the affected legs. Physical examination showed hip pain during passive movement of the coxofemoral joint, particularly during leg abduction. Radiographic and CT examinations confirmed the diagnosis and allowed staging in LCPD grade 1 (n ¼ 3), grade 2 (n ¼ 11) and grade 3 (n ¼ 18) according to the Ljunggren scale (►Figs. 3 and 4).
No major complications or infections were seen neither after bone marrow harvesting nor after intra-osseous injection of BMMCs. For the latter, laterolateral X-rays or CT guidance were helpful to allow accurate injection into the necrotic and perinecrotic area and to avoid the potential risk of damage to the retinacular vascular blood supply of the femoral head; in fact, the dorsal and ventral retinacular artery remained throughout unaffected by trajectory of the needle. We harvested a mean of 14.3 AE 5.2 mL bone marrow. The number of BMMCs obtained after Ficoll separation was 104.7 AE 46.5 Â 10 6 /mL. The total CFUs from bone marrow were 71.6 AE 51.9 Â 10 2 /mL.

Follow-Up
A progressive and clear reduction of pain upon passive abduction and weight bearing and an improvement in gait was observed in 28 (87.5%) of the treated dogs within 3 weeks.
In all dogs, at 3 weeks, radiography did not show any radiographic changes, as compared with preoperative examinations. In four of the treated dogs (12.5%), we observed persistent hip pain and lameness.
After 3 months, the 28 improved dogs showed no signs of pain and normal gait. Radiographic examination at this follow-up showed an increase in sclerosis of the femoral head and a reduction in radiolucent zones with only mild periarticular changes. However, radiography and CT scan also revealed persistent collapse of the femoral head and incongruence of the coxofemoral joint (►Fig. 5). The longterm follow-ups on selected cases revealed normal gait despite the radiographic persistence of the articular collapse and joint incongruence (►Fig. 6) In 4 of the 32 dogs, pain and lameness persisted; therefore, femoral head and neck resection was performed.

Histology, Histochemistry and IHC
Morphological examination of the femoral head and neck of control dogs not subjected to stem cell therapy, revealed the presence of large erosive-necrotic processes, particularly at the level of the neck and in some areas of the head (►Fig. 7A), associated with a complete lack of safranin O staining, indicating a loss of proteoglycan content in the articular cartilage and    growth plate (►Fig. 7B). In the head and neck of all controls, Herovici polychrome stain evidenced diffuse secondary ossification (►Fig. 7C, arrowheads) and a notable lack of newly formed collagen, lack of neo-angiogenesis and of signs of regeneration. Morphological and histochemical staining was performed on the HSG group that had not clinically recovered by 3 months post-treatment; evidence was found of an inflammatory process and new fibrocartilage deposition in the BMMC-injected area (►Fig. 7D and 7G). In particular, nuclei of cartilage formation (►Fig. 7G, arrows) and proteoglycan synthesis and deposition (arrow heads, ►Fig. 7E and small arrow heads, ►Fig. 7H) were stained orange with safranin O. In HSG dogs, Herovici polychrome staining revealed large areas of extracellular matrix deposition around the point of BMMC injection (►Fig. 7F and 7D, arrow heads), associated with neoangiogenesis. Quantification of CD34þ cells in histological sections of the femoral head and neck from HSG group and control FNRC animals revealed a high number of CD34þ cells, mainly aggregated around capillaries in areas of new fibrocartilage deposition (►Fig. 8A insert). Computed tomography PC, CD and BI were compared between the group HSG and control FNRC group. All analysed parameters are summarized in ►Table 1 and ►Fig. 9. In the HSG, the CT, PC, CD and BI were significantly higher than the FNRC group. CD34þ cells were only found in stem cell-injected femoral heads and were completely absent in controls (►Fig. 8A).

Discussion
The aim of this retrospective study was to evaluate the use of BMMCs as a minimally invasive treatment for LCPD in dogs.
The results showed that this technique gives good results in terms of swift recovery from pain and also leads to gradually increasing weight bearing on the affected limb, up to a complete remission of clinical signs at 3 months post-stem cell injection. Moreover, our data show that in case of failure of the BMMCs treatment (12.5%), dogs can still be safely treated with the classical femoral head and neck excision.
The clinical effectiveness of the method described in this study has been shown previously in human medicine, but the role of transplanted cells remains unclear. Our histological and IHC findings showed an increased articular cartilage and subchondral bone thickening, with a marked neoangiogenesis in treated samples, but complete absence thereof in the control group. Moreover, the presence of CD34þ cells, a marker found in BMMCs, could have been responsible for the regenerative effect, directly or indirectly through cell recruitment and chemotaxis, as found in other mesenchymal tissues. It has been documented that BMMCs secrete a wide spectrum of factors with immunomodulatory, anti-inflammatory, antiapoptotic, proangiogenic, proliferative or chemoattractive capacities. Pain improvement is probably owing to the anti-inflammatory capacity of stromal cells. Indeed, mesenchymal stromal cells have therapeutic potential for regenerative and anti-inflammatory therapy, [29][30][31] and bone marrow-derived MSCs could suppress T cell proliferation. 32 Currently, several soluble factors that are either produced constitutively by BMSCs or as the result of cross-talk with target immune cells, have been attributed to the immunomodulatory properties of BMSCs. [33][34][35]   The amount of the injected BMMCs was very variable among our patients (104.7 AE 46.5 Â 10 6 /mL), since the technique used to obtain the cells (for density gradient centrifugation) is depending on several factors including the original quality of harvested bone marrow, its volume and patient-related individual factors. Although some authors suggest that there is a direct correlation between the number of implanted cells and the treatment efficacy, at the moment, a clear reference target number is not available and the current literature suggests using the largest number of cells available. 36 Our good clinical results also seem related to increased precocious bone sclerosis, which is normally the last outcome of the disease, proven by radiographic and CT findings of persistent collapse of the femoral head and incongruence of the coxofemoral joint. Considering the minimally invasive and conservative approach of the proposed treatment, we suppose that a re-injection of the BMMCs could be an option in dogs not responding to the first treatment and/or with a recurrence of the symptoms. Further studies are required to confirm this hypothesis.
It has been considered that the reason for insufficient substitution in bone remodelling after osteonecrosis in the femoral head may be the small number of progenitor cells present in the femoral head. 37,38 Others suggested that the effectiveness of bone marrow cells may be related to the availability of stem cells endowed with osteogenic properties in the femoral head, which increase after bone marrow implantation. [39][40][41][42] Another possible explanation for the therapeutic effect of bone marrow implantation is that injected marrow stromal cells secrete angiogenic cytokines, resulting in increased angiogenesis and subsequent improvement in osteogenesis. Hernigou and colleagues stated that BMMCs can elicit the formation of new blood vessels via the presence of endothelial cell progenitors or haemangioblasts. This may be owing to both the supply of progenitor cells and to angiogenic cytokines produced by bone marrow cells. Endothelial progenitors can actively engage in angiogenesis in tissue devoid of vessels and in neo-angiogenesis from pre-existing capillaries via the angiopoietin-1/Tie2 pathway. In addition, local ischaemia that activates HIF1A signalling and mobilization of circulating progenitors via the SDF1-dependent pathway may provide permanent stimuli for blood vessel repair, as well as a supply of new cells for bone regeneration with chemiotaxis or differentiated or undifferentiated resident cells for functional tissue restoration. 43 In a study on cartilage repair using freshly isolated BMMCs as controls, Jiang and colleagues stated that cells obtained from in vitro culture and expansion are not recommended for cell-based therapy and that fresh BMMCs are the best choice. 44 The absence of a clinical control group can be considered a limitation of this study.
The present results confirmed clinically what has already shown in experimental models on dogs. Indeed, Yan and colleagues investigated the survival and differentiation of MSCs transplanted in skeletally mature mongrels in which osteonecrosis of the femoral head was surgically induced; patients were treated with either saline (control) or autologous BMSC transplantation. These previous results demonstrate that transplanted autologous BMSCs can survive, proliferate and differentiate into osteoblasts directly, which contributes to acceleration of the repair process. 45 Hang and colleagues and Liu and colleagues evaluated the efficacy of vascular endothelial growth factor 165 in transgenic BMSCs for the repair of earlystage osteonecrosis of the femoral head in mature mongrel dogs. After core decompression, transgenic and non-transgenic autologous BMSCs were implanted and therapeutic efficacy was obtained, with new bone formation and neovascularization in the femoral head confirmed in collateral examinations. 46,47 Jin and colleagues investigated the effects of autologous BMSCs administered via arterial perfusion on vascular repair and angiogenesis in an experimental model of osteonecrosis of the femoral head in dogs. Three weeks after establishment of ischaemia in the femoral head, a good therapeutic effect was obtained, with BMSCs promoting vascular repair and angiogenesis. 19 In conclusion, on the basis of our results, we consider the technique described as minimally invasive, with no disadvantages, and that implantation of autologous bone marrow cells can be a valid alternative approach to the classical surgical treatment of LCPD with femoral head and neck excision; in case of non-response to cell treatment, surgery may still be performed successfully.