Epiphyseal cartilage vascular architecture at the distal humeral osteochondritis dissecans predilection site in juvenile pigs

Failure of endochondral ossification due to interruption of the vascular supply to the epiphyseal cartilage is a critical step in the development of osteochondritis dissecans (OCD). Herein we describe the vascular architecture of the distal humeral epiphyseal cartilage in pigs and identify characteristic features that have been associated with sites predisposed to OCD development across species. Distal humeral specimens were harvested from pigs (n = 5, ages = 1, 10, 18, 30, and, 42 days old) and imaged at 9.4T magnetic resonance imaging (MRI) using a 3D gradient recalled echo sequence. The MRI data were processed using a quantitative susceptibility mapping (QSM) pipeline to visualize the vascular architecture. Specimens were also evaluated histologically to identify the presence of ischemic epiphyseal cartilage necrosis (osteochondrosis [OC]‐latens) and associated failure of endochondral ossification (OC‐manifesta). The QSM data enabled visualization of two distinct vascular beds arising from the perichondrium at the lateral and medial aspects of the distal humeral epiphysis. Elongated vessels originating from these beds coursed axially to supply the lateral and medial thirds of epiphyseal cartilage. At 18 days of age and older, a shift from perichondrial to transosseous blood supply was noted axially, which appeared more pronounced on the lateral side. This shift coincided with histologic identification of OC‐latens (30‐ and 42‐day‐old specimens) and OC‐manifesta (18‐ and 42‐day‐old specimens) lesions in the corresponding regions. The vascular anatomy and its evolution at the distal humeral epiphysis closely resembles that previously reported at predilection sites of knee OCD, suggesting a shared pathophysiology between the knee and elbow joints.

Osteochondritis dissecans (OCD) is a developmental orthopaedic disease affecting children and young animals.It is characterized by formation of osteochondral flaps or fragments in developing joints that cause pain, locking of the joints, and disability. 1 Despite having been recognized for over 130 years, the etiology and pathogenesis of OCD in children remain incompletely understood, with ischemia, genetic predisposition, and repetitive trauma all being implicated. 1 Accordingly, identification of factors that allow accurate prediction of the course of the disease, along with evidence-based guidelines to inform treatment of clinical patients, are absent, as concluded by the American Academy of Orthopaedic Surgeons. 2 address these knowledge gaps, research has been focused on elucidating the pathogenesis of OCD.Because of the inability to conduct invasive experiments in children, most studies utilize large animal models.Indeed, histologic studies performed in asymptomatic juvenile pigs [3][4][5] and horses 6,7 at OCD predilection sites demonstrated that discrete areas of epiphyseal cartilage necrosis (termed osteochondrosis [OC]), caused by focal failure of the epiphyseal vascular supply, are the clinically silent precursor lesions of OCD.Findings from a subsequent study using distal femoral specimens obtained from cadavers of children identified similar lesions, suggesting a shared etiology that similarly involves failure of the vascular supply to the epiphyseal growth cartilage. 8These results were corroborated by the findings of a comparative study demonstrating a nearly identical vascular architecture at the distal femoral OCD predilection site in children and juvenile pigs. 9Vessels supplying the lateral and medial femoral condylar epiphyseal cartilage were found in both species to be elongated and without anastomoses, providing a compelling explanation for how failure of individual vessels may result in extensive ischemic necrosis of the epiphyseal cartilage.
Most comparative studies to date have focused on the knee joint, specifically the femoral condyles, which is the primary OCD predilection site in the pelvic limb in children and pigs. 10Conversely, the elbow joint, the second and third most commonly affected site in male and female children, respectively, has received limited study. 11Although elbow joint OCD is often present in both children and pigs, the within-joint location of lesions differs between these species.While most OCD lesions affect the humeral capitulum in children 12 (i.e., the lateral aspect of the distal humerus), in pigs the majority of lesions are identified medially, involving the humeral trochlea. 13This apparent disparity in predilection site within the elbow suggests, that in addition to the quadruped nature of pigs, there may also be characteristic differences in the distal humeral vascular architecture between species that contribute to elbow OCD development.
The objective of our study was to describe the three-dimensional vascular architecture of the distal humeral epiphyseal cartilage in developing pigs to generate critical data for future comparisons to children and to further our understanding of the pathogenesis of elbow OCD.Histologic evaluation of each porcine specimen was also performed to identify areas of necrotic epiphyseal cartilage characteristic of early OCD and to determine their association with MRI findings at the individual level.We hypothesized that, similar to the distal femoral condyle, the epiphyseal vascular architecture at the distal humeral OCD predilection sites would be comprised of elongated vessels traveling parallel with the articular surface, and that OC lesions would be identified in these regions.

| Animals
Right distal humeral specimens (n = 5) were harvested from discarded carcasses of five domestic pigs aged 1, 10, 18, 30, and 42 days that were presented as fresh carcasses at the University of Minnesota Veterinary Diagnostic Laboratory for diagnostic necropsy for nonorthopedic reasons.Briefly, elbow joints were disarticulated by severing the joint capsule and all ligamentous and tendinous attachments.Soft tissue covering the distal humerus was then sharply excised and the humeri were transected approximately 1 cm proximal to the distal humeral growth plate.Harvested specimens were wrapped in saline-soaked towels and stored in a freezer at −20°C until further processing.No approval from the Institutional Animal Care and Use Committee was necessary because only cadaveric specimens were used in this study.

| Magnetic resonance imaging (MRI) acquisition
Specimens were imaged using a preclinical 9.4T MRI scanner (Varian NMR Systems) interfaced to VnmrJ software (VnmrJ version 4.2 revision A).A quadrature volume transceiver coil (Millipede; Varian NMR Systems) was used in the acquisitions.For imaging, specimens were immersed in perfluoropolyether (Fomblin; Solvay Specialty Polymers) for a clean background and to avoid artifacts at tissue-air interfaces and were oriented with the humeral shaft approximately along the main magnetic field B 0 of the scanner.Data were acquired as previously described 14,15 using a 3D gradient recalled echo (GRE) sequence with an isotropic spatial resolution of 78 µm (specimen ages 1 and 10 days, field-of-view = 30 × 30 × 30 mm 3 ) or 100 µm (specimen ages of 18-42 days, field-of-view = 38.4× 38.4 × 38.4 mm 3 ), using an acquisition matrix size of 384 × 384 × 384.Bandwidth was set to 14 kHz, repetition time (TR) was 40 ms, and echo time (TE) was 15.3 or 15.6 ms for the smaller or larger fields of view, respectively.Scan time with one average and full k-space acquisition was 98 min.

| MRI data processing
The 3D GRE data were processed using a quantitative susceptibility mapping (QSM) pipeline in MATLAB (Version 2022b; The Mathworks), similar to that previously described by Nykänen et al. 16 Briefly, masks containing only the articular-epiphyseal cartilage complex were created using ITK-SNAP software (http://www.itksnap.org/) 17with a region growing segmentation algorithm.The raw phase data was unwrapped using Laplacian unwrapping 18 then background field contributions were removed using V-SHARP (sophisticated harmonic artifact reduction for phase with variable kernel size) 19 filtering.The radius of the V-SHARP filter was set to vary from 1 to 8 voxels, with a threshold of 0.85 to allow for as little erosion of the tissue boundary as possible. 16The susceptibility maps were calculated from the unwrapped and background field corrected phase (i.e., field map) data using the morphology enabled dipole inversion (MEDI-L1) method 20 available in the MEDI-Toolbox (http://pre.weill.cornell.edu/mri/pages/qsm.html)with the lambda value (regularization strength) set at 80,000.The susceptibility maps were used to enhance the GRE magnitude images as QSM-weighted images, mapping susceptibility values from [<0, 0 to 0.1, >0.1] ppm to [1, 1 to 0, 0], to create a "positive" weighting mask. 15Using the QSM-weighted images, minimum intensity projections of 2-4 mm thickness were created in the coronal and axial planes as well as through the growth plate to visualize the vasculature.For 3D visualization of the distal humerus, the secondary ossification center was segmented like the susceptibility mask using ITK-SNAP.
Segmentation of the vasculature within the epiphyseal cartilage was accomplished using automatic detection of an appropriate threshold to binarize the QSM data; the optimal threshold was determined by finding the maximum correlation between the original susceptibility map and a binary image generated with variable threshold levels.These segmentation images were then saved as 3D stereolithography (STL) mesh files from MATLAB and subsequently combined in a 3D design software (Rhinoceros version 6 SR35, Robert McNeel & Associates) for 3D visualization.

| Histology
At the conclusion of the MRI sessions, the distal humeri were fixed in 10% neutral buffered formalin for a minimum of 72 h, decalcified in 10% ethylenediaminetetracetic acid, and serially sectioned in the coronal (transverse) plane from anterior (cranial) to posterior (caudal) into 2.0-mm-thick slabs that spanned the total width of the epiphysis.Individual slabs (4-6/humeri) were processed into paraffin blocks for histological evaluation.At least two 5-μm-thick sections were collected from the surface of each slab and stained with hematoxylin & eosin (H&E).Additional serial sections at 50-µm intervals were obtained and stained with H&E to confirm identification of all suspected OC lesions present in the evaluated specimens (3-4 sections per suspected lesion).Histological sections were assessed by a board-certified veterinary pathologist with experience in musculoskeletal pathology (ARA).OC lesions were defined as areas of chondronecrosis associated with necrotic vascular profiles that were confined to the epiphyseal cartilage (OC-latens) or associated with a delay in endochondral ossification (OC-manifesta). 21

| RESULTS
3D visualizations of the vasculature supplying the distal humeral epiphyseal cartilage were consistent with the presence of two distinct vascular networks originating from the perichondrium at the lateral and medial aspects of the joint (Figure 1 and Video S1).Elongated vessels arising from perichondrium traveled parallel to the articular surface to supply the medial and lateral thirds of the epiphyseal cartilage.An avascular area was apparent in the central third of the epiphyseal cartilage, immediately adjacent to the sagittal ridge.By the age of 18 days and older, an apparent shift from perichondrial to transosseous blood supply was apparent axially in the vascularized epiphyseal cartilage, most prominently on the lateral side (Figure 1).These observations were further confirmed by analysis of minimum intensity projections generated in the coronal and axial planes (Figure 2).The epiphyseal growth cartilage of the medial epicondyle remained extensively vascularized until the appearance of its ossification center in the 18-, 30-, and 42-day-old specimens (Figure 1).A high density of transphyseal vessels traversing the distal humeral growth plate was apparent in the 1 and 10-day-old specimens (Figures 2 and 3), but they nearly completely disappeared by the age of 18 days and older.A few thick-appearing vessels were noted in the 1-and 18-day-old specimens (Figures 2 and 3).This appearance is likely a manifestation of "blooming artifact," an accentuated and enlarged signal change (signal void) in both raw magnitude MR images and susceptibility maps.
Histological evaluation of the harvested specimens identified the presence of ischemic epiphyseal cartilage necrosis in 3 of the 5 specimens.In the 18-day-old specimen, an OC-manifesta lesion was identified at the trochlea characterized by a region of chondronecrosis with loss of staining of the epiphyseal cartilage and a corresponding focal delay of ossification (Figure 4).At 30 days, an OC-latens lesion was present in the same location (Figure S1).In the 42-day-old piglet, there was an OC-latens lesion at the trochlea and an OC-manifesta lesion at the capitulum.
All lesions were located at weight-bearing points in the joint, approximately midway between the cranial-and caudal-most points of the distal humerus.Using the histologic findings as a guide, individual lesions, along with the vessels supplying them, could also be identified in the QSM-weighted images after they were manually resliced and oriented to match the orientation of histologic sections (Figure 4 and Figure S2).

| DISCUSSION
The vascular supply of the distal humeral epiphyseal cartilage and growth plate was successfully visualized using QSM of 3D GRE data.Vessels arising from the perichondrium covering the medial and lateral aspects of the distal humeral epiphysis gave origin to elongated branches coursing parallel with the articular surface in the axial direction.These elongated vessels terminated before reaching the mid-portion of the joint, creating an area devoid of vasculature resembling a "watershed" region.This vascular F I G U R E 1 3D reconstructions of the segmented secondary ossification center, epiphyseal cartilage, and the epiphyseal vasculature, depicting the vascular architecture to the distal humeral epiphyseal cartilage in pigs aged 1-42 days.Elongated vessels, arising abaxially from the perichondrium, supply the lateral and medial third of the joint, while the central portion of the epiphyseal cartilage is avascular.At 18 days of age and later, at the lateral portion of the distal humerus, blood supply to the axial segment has shifted from vessels originating from the perichondrium to vessels traversing the ossification front (white ellipses).The secondary ossification center of the medial epicondyle (black circles) is apparent at 18, 30, and 42 days of age.Image orientation: Lateral is to the left and anterior (cranial) is to the top.
architecture is nearly identical to that of the femoral condylar epiphyseal cartilage, the primary OCD predilection site in the pelvic limb in children and pigs, where the vascular supply also originates from the axial and abaxial portions of each condyle, courses axially and terminates before reaching the mid-portion of the femoral condyles. 9,22ongated vessels arising from the perichondrium and running parallel with the articular surface have long been implicated in the development of OCD at multiple predilection sites, including the distal tibia 23 and femur. 24,25Results of studies using histology 23,24 and/or microcomputed tomography 26 lend strong support to the theory that failure of these elongated vessels occurs when the advancing ossification front reaches their mid-portion, interrupting the perichondrial blood flow to the distal vascular segment.If this event is not accompanied by formation of new anastomoses between vessels originating from the bone marrow and the distal portions of the elongated vessels, the latter lose their blood supply, leading to ischemic necrosis of the surrounding epiphyseal cartilage. 5,23is hypothesis is further supported by the observation that animal species that do not rely on these elongated vessels to supply the axial portions of their epiphyseal cartilage are protected from the development of OCD. 9 Interestingly, despite of the high resolution of the MRI studies (78 µm [specimens 1 and 10 days old] and 100 µm [specimens 18-42 days old] isotropic resolution), we were unable to visualize any vessels in the midsagittal watershed region.Findings from a previous study using a similar approach showed that the smallest vessels detectable with a 9.4T and 7T magnets were on average (range) 104 µm (55-154) and 200 µm (134-301) in diameter, 14 suggesting that all but the smallest capillaries were identified in the current study.This finding, along with the absence of lesions in the central, avascular portion of the distal humeral epiphyseal cartilage, suggests that this region receives sufficient amounts of nutrients and oxygen via diffusion from the joint cavity and from the axialmost portion of the vessels comprising the lateral and medial vascular beds.
Multiple studies conducted in pigs demonstrated that the majority of OC lesions involve the medial aspect (trochlea) of the distal humerus. 3,13One potential explanation our study offers for this phenomenon is that, in the axial segment of the lateral portion of the joint, the shift in the blood supply from perichondrial vessels to transosseous (originating from the bone marrow) appears more pronounced than that of the medial portion of the joint.This shift to a transosseous blood supply was first identified at the age of 18 days and older, a time that coincided with the histologic appearance of OC lesions.These observations suggest that future studies investigating why most elbow joint OCD lesions affect the lateral portion of the distal humerus (capitulum) in children should carefully evaluate the vascular architecture, along with its evolution over time, in the distal humeral epiphyseal cartilage.
Three out of four OC lesions identified in the evaluated specimens were found in the medial aspect of the distal humerus, the primary predilection site of OC in the porcine elbow joint. 3,13While the earliest histologically-confirmed OC-latens and OC-manifesta lesions previously reported were found in pigs aged 4 and 12 weeks, respectively, 27 our findings demonstrate that these lesions can occur much earlier; we identified an OC-manifesta lesion in a piglet as young as 18 days (<3 weeks) of age.Our observed shift in the vascular supply in the axial portion of the distal humeral epiphyses from perichondrial to transosseous, first noted at 18 days of age provides, provides a plausible explanation for the early appearance of OC lesions.
F I G U R E 2 Coronal and axial plane minimum intensity projections generated from the 3D QSM images depicting the distal humeral epiphyseal vasculature in pigs aged 1-42 days.Elongated vessels arising from the perichondrium coursing axially (black arrows) are apparent both in the medial and lateral thirds of the distal humeral epiphyseal cartilage, while the central third of the epiphyseal cartilage is avascular (black ellipse).Vessels traversing the growth plate (white arrowheads) are present in the 1-and 10-day-old specimens.There is progressive thinning of the epiphyseal cartilage with increasing age.Blooming MRI artifacts (black arrowheads) in the 1-and 18-day-old specimens are apparent.Black insets in the right lower corner identify the region the minimum intensity projections depict.Coronal image orientation: lateral is left and proximal is to the top.Axial image orientation: lateral is left and anterior (cranial) is to the top.MRI, magnetic resonance imaging.
Quantitative susceptibility mapping also allowed visualization of vessels traversing the growth plate and their abrupt disappearance by 18 days of age in the examined porcine specimens.Although we were unable to determine the origin of the vessels present in the porcine growth plate, in foals the growth plate of the medial femoral condyle is known to receive its blood supply from a mixture of metaphyseal and epiphyseal arteries, with only the latter persisting beyond 10 days of age. 28Vessels traversing the physis are clinically important because they present a conduit for hematogenous spread of infection across the growth plate, 29,30 thus their identification using a noninvasive MRI method merits further investigation.
Processing of the susceptibility-weighted 3D GRE MRI (phase) data was done utilizing a typical QSM pipeline. 16During generation of the calculation masks (regions of interest), small inconsistencies in the mask (missing tissue near cartilage-bone boundaries, or due to susceptibility artifacts) were considered acceptable.As with articular cartilage, 16 minimal erosion near the boundaries was desired for the cleanest possible depiction of vessel origins, and thus V-SHARP was chosen for the background correction step.During the final step in the QSM processing (dipole inversion), streaking artifacts were minimized by the regularization available in the MEDI method. 20e strength of the regularization determines the level of artifact reduction as well as the overall smoothing of the quantitative susceptibility maps.Different regularization strengths ranging from 10,000 to 125,000 were tested in one specimen, informing selection of 80,000 as the optimal value to highlight the vasculature.The blooming artifact, associated with some of the vessels in the 1-and 18-day-old specimens, is most likely due to greater susceptibility differences between these vessels and the surrounding epiphyseal cartilage than elsewhere in the vasculature.A susceptibility-related source for the enlarged signal void around the thick-appearing vessels was further implicated by the relatively large changes around those vessels in the local field maps, produced as an intermediate step in the QSM processing.Such localized susceptibility differences likely have arisen from an increased accumulation of iron (released from the heme molecules during breakdown of blood), that lead to increased paramagnetic susceptibility in the region. 31r study has several limitations.While our sample size is small, with only one animal imaged at each time point, it is known from previous studies conducted in horses, 32 pigs, 24 and goats 33 that the blood supply to the epiphyseal growth cartilage is highly consistent in a given species at a specific age and location.Also, our findings are qualitative in nature, making their rigorous evaluation difficult.The assessment and quantitative comparison of the vascular architecture could be further supported by morphological analyses, 22,25  In conclusion, our study identified characteristic features of the vascular supply to the epiphyseal cartilage of the developing distal humerus in pigs that may contribute to the pathogenesis of OCD lesions.In particular, we found a temporal shift in the blood supply from perichondrial to transosseous in the axial portion of the vascularized epiphyseal cartilage that coincides with the appearance of OC lesions and corresponds to the predilection site of elbow joint OCD in pigs.These findings support a shared pathogenesis of OCD lesions between the knee and elbow joints with clinical relevance to human juvenile OCD.
particularly in larger datasets.Nevertheless, consistent features of the vascular architecture present across individual pigs (i.e., the presence of elongated vessels arising from the perichondrium) along with progressive changes occurring over time (i.e., a shift to transosseous blood supply in the axial segments) remain important observations with potential clinical significance.Lastly, our studies were conducted ex vivo on relatively small specimens that allowed the use of an ultrahigh F I G U R E 3 Axial plane minimum intensity projections generated from the 3D QSM images depicting the vascular supply to the distal humeral growth plate in pigs aged 1-42 days.Black dots (black arrows) present in the central portion of the growth plate in the 1-and 10-day-old specimens correspond to vessels traversing the growth plate.At 18 days of age and later, these transphyseal vessels are no longer present.Blooming MRI artifacts (white arrowheads) in the 1-day-old specimen are apparent.Black insets in the right lower corner identify the regions the minimum intensity projections depict.Image orientation: lateral is left and anterior (cranial) is to the top.MRI, magnetic resonance imaging; QSM, quantitative susceptibility mapping.field magnet.Translating our methods to in vivo imaging of human subjects would be complicated by the decreased resolution and signal to noise ratio available even at the highest clinically relevant field strength of 7T.