A convenience sample of tissues from horses euthanized at the institution for reasons other than stifle disease was used. The main PL and IFP vascular pattern was described based on computed tomography (CT) scanning of barium perfused specimens, whereas the histological architecture was described from tissue processed with H&E and toluidine blue staining.
Upon obtaining informed consent from animal owners, an opportunistic cohort of one 10-day old foal and seven adult horses (age 3 – 18 years) euthanased at the institution for reasons unrelated to stifle pathology were recruited to the study; case details and reasons for euthanasia are detailed in Supplementary table 1. The study was approved by the ethical review board at the institution and was in accordance with national legislation regarding use of animals in research (FOR-2015–06-18–761).
Prior to euthanasia, adult horses were subjected to a standardized lameness examination including walking and trotting in a straight line on a hard surface and were included in the study when no baseline hind limb lameness was detected (defined as AAEP grade 0). The stifle area was palpated for soft tissue swelling such as joint effusion and periligamentous thickening. A standardized B-mode ultrasonographic examination (Esaote MyLab Gold ultrasonography machine, using a multifrequency linear probe set at 10–15 MHz) of the stifle was performed in a weightbearing position. In brief; following a routine skin preparation including hair removal (size #40 clipper blades) and application of alcohol and coupling gel, the medial, cranial and lateral aspect of each stifle was examined in longitudinal and transverse planes. Structures evaluated at the medial aspect included the tibial plateau; medial meniscus and medial collateral ligament. Presence or absence of medial femorotibial joint effusion; osteophyte formation; and ligamentous or meniscal injuries were recorded. At the lateral aspect, the lateral collateral ligament; lateral meniscus and lateral femorotibial and femoropatellar synovial pouches were assessed for the same abnormalities. The cranial aspect of the stifle was then evaluated focusing on the intermediate, medial and lateral PLs as well as the medial and lateral femoral trochlea and the trochlear groove. The three patellar ligaments were scanned in transverse and longitudinal planes from their origin to the tibial insertions. Ligament shape, size, echogenicity and fiber pattern was evaluated and considered normal when findings were in line with previously reported literature [1,2,3,4,5,6]. Abnormal findings such as deviations in size, shape and/or echogenicity pattern were noted and considered normal if findings were bilaterally symmetric.
Included animals were administered i.v. heparin (500 IU/kg) as an adjunct to the routine euthanasia protocol which consisted of premedication with detomidine / butorphanol, anesthesia induction with ketamine/ midazolam, followed by an overdose of pentobarbital. Immediately after euthanasia, one randomized hind limb was subjected to a barium perfusion procedure and subsequent CT scanning, while the contralateral hind limb was used for histology only. Histology was also obtained from two barium perfused limbs after CT scanning, to validate that the barium identified on the CT scans was located intravascularly, thus representing blood vessels.
Barium perfusion and CT scanning
Limbs were harvested at the coxofemoral joint and suspended from the distal limb for 3 h prior to being subjected to barium perfusion using modifications of a previously published protocol . In brief, this protocol includes stepwise perfusion of the limb via the femoral artery with 1) isotonic saline until the effluent runs clear; 2) 20% v/v micronized barium suspended in saline; and finally, 3) 20% v/v micronized barium suspended in 10% neutral buffered formalin (NBF). Perfused limbs were refrigerated for 48 h to allow the barium to set in the arterial vasculature prior to en block harvesting of the patella, the PLs and the IFP for further tissue fixation (10% NBF for a minimum of 48 h followed by ethanol) and subsequent CT scanning. Tissue harvesting was achieved by sharp dissection of periarticular soft tissues from the patella and PLs, taking care to separate the caudodistal portion of the medial PL from the common aponeurosis of the gracilis and sartorius muscles, and to separate the lateral aspect of the lateral PL from the aponeuroses of the biceps femoris and the fascia lata muscles . The tibial tuberosity was cut from parent bone using a bone saw: a transverse cut was placed immediately cranial to the menisci, and a horizontal cut was placed distal to the tibial tuberosity, taking care to leave all PL insertions intact. In the foal and one adult horse, the IFP was left intact with the en block specimens for tissue fixation and scanning, whereas in 6/7 adult horses, the IFP was removed by sharp/blunt dissection from the intermediate PL for separate fixation and scanning.
CT scanning was performed using a 4-slice GE CT scanner (GE Medical Systems Bright Speed S); slice thickness was 0.625 mm. The digital CT images were subsequently processed using ImageJ and Osirix Dicom 3D imaging software. The technical quality of the barium perfusion was assessed and deemed of adequate quality for analysis when perfused vessels were clearly visible within each of the PLs; the patella; and within the tibial tuberosity. Potential perfusion artefacts (overfilling and/or barium leakage) was recorded. Quantitative and qualitative assessments of perfused vessels commenced. For all PLs, the main intraligamentous vessel morphology including trajectory and branching pattern was described from 3D reconstructions; for the specimens in which the IFP was left in place for the scan, the vascular association between the IFP and intermediate PL was described. Within each ligament, the number of distinct, barium-filled longitudinal vessels were counted on 2D transverse images at 3 different locations (proximally, defined as immediately distal to the patellar origin; at the mid-portion, halfway between the patella and the tibia; and distally, immediately proximal to the tibial insertion), and the vessel cross-sectional location (central vs. peripheral) was recorded and reported as the proportion of centrally located vessels. In the IFP, the number of distinct, barium-filled vessels were counted on 2D transverse images at 3 different locations; 1/3 from the proximal pole; at the mid portion; and 1/3 from the distal pole.
En block harvesting of the patella, PLs, IFP and tibial tuberosity of the contralateral limb was performed as described above and placed in 10% NBF for 48 h. Five 5 mm tissue blocks were sharply cut perpendicular to the longitudinal fiber orientation of each ligament as illustrated in Fig. 3, from the origin at the patella/medial parapatellar fibrocartilage, to the insertion at the tibial tuberosity, taking care not to include bone and/or cartilage at the ligament origin/ insertion. Five 5 mm tissue blocks were sharply cut from the IFP, perpendicular to its longitudinal axis. Tissue blocks were embedded in paraffin and 4—6 µm sections were routinely processed with H&E staining. Tissue blocks in which metaplastic tenocytes were identified in H&E sections were also processed with toluidine blue staining. Sections were evaluated at 2.5x, 10 × and 40 × magnification by light microscopy and photographed using a Zeiss™ digital photomicroscope (AxioCam ERc5s) and then copied and stored onto a computer using designated software (Zen blue, Carl Zeiss).
All sections were assessed for diagnostic quality, and sections in which ligament structure and cellular morphology could not be assessed due to processing artefacts were omitted from the analyses. Normal ligament structure was defined as polygonal fascicles consisting of collagen bundles with scattered tenocytes, separated by endotenon recognized as interstitial connective tissue septa carrying blood vessels, lymph vessels and nerves, as illustrated in Fig. 3. Tenocytes were classified as normal when nuclei were spindle shaped (type 1) or when nuclei were plump and cigar shaped (type 2). Tenocytes with rounded nuclei situated in semi-lacunae were defined as abnormal and referred to as metaplastic tenocytes. Tissue vascularity was defined as normal when blood vessels were confined to the endotenon and abnormal when vessels were seen within fascicles.
The interfascicular endotenon was subjectively classified as thin when it was 1–2 cell layers thick, and as distinct when it consisted of multiple cell layers; the distinct interfascicular endotenon were further characterized as having a predominantly discrete or prominent vascularity as illustrated in Fig. 2. Interfascicular endotenon thickness was objectively measured on ruler-calibrated images captured at 2.5 × magnification using imaging software (ImageJ); cross-junctional areas of the endotenon were excluded from the analysis as they formed irregular structures.
Fascicle cellularity was defined as normal when scattered tenocytes were seen within the fascicles, whereas hypercellularity was defined as areas of tenocyte clustering (Fig. 2). Contrary, hypocellularity was defined as areas almost devoid of cells, which coincided with areas of metaplastic tenocytes (Fig. 2).
Ligamentous fatty infiltration was defined as normal when adipocytes were confined to the endotenon, and as abnormal when intra-fascicular adipocytes were seen. When confined to the endotenon, the amount of fatty infiltration was subjectively graded as none; slight; moderate or marked as illustrated in Fig. 2.
In the toluidine-blue sections, the extracellular matrix glycosaminoglycan content was subjectively assessed as increased in areas of increased basophilia.
Normal histological architecture of the IFP was defined as lobes of adipose tissue separated by connective tissue septa, with a synovial membrane lining along its caudal aspect and with no presence of inflammatory cell infiltration. Interlobular connective tissue septa were subjectively assessed as thin or distinct; septal thickness as well as lobular diameter were objectively measured in images acquired at 2.5 × magnification using imaging software (Image J).
Descriptive statistics were applied to all measured parameters. Due to non-normal distribution, the interfascicular endotenon thickness measurements were compared between ligaments using the Wilcoxon method. For each ligament, the association between age and endotenon characteristics and fatty infiltration was tested using univariable logistic regression analyses; model fit was assessed with the effects likelihood ratio test. Significance was defined as P < 0.05 for all analyses.