Advances in cancer therapy have prolonged patient survival even in the presence of disseminated disease and an increasing number of cancer patients are living with metastatic bone disease (MBD). The proximal femur is the most common long bone involved in MBD and pathologic fractures of the femur are associated with significant morbidity, mortality and loss of quality of life (QoL). Successful prophylactic surgery for an impending fracture of the proximal femur has been shown in multiple cohort studies to result in longer survival, preserved mobility, lower transfusion rates and shorter post-operative hospital stays. However, there is currently no optimal method to predict a pathologic fracture. The most well-known tool is Mirel's criteria, established in 1989 and is limited from guiding clinical practice due to poor specificity and sensitivity. The ideal clinical decision support tool will be of the highest sensitivity and specificity, non-invasive, generalizable to all patients, and not a burden on hospital resources or the patient's time. Our research uses novel machine learning techniques to develop a model to fill this considerable gap in the treatment pathway of MBD of the femur. The goal of our study is to train a convolutional neural network (CNN) to predict fracture risk when metastatic bone disease is present in the proximal femur. Our fracture risk prediction tool was developed by analysis of prospectively collected data of consecutive MBD patients presenting from 2009–2016. Patients with primary bone tumors, pathologic fractures at initial presentation, and hematologic malignancies were excluded. A total of 546 patients comprising 114 pathologic fractures were included. Every patient had at least one Anterior-Posterior X-ray and clinical data including patient demographics, Mirel's criteria, tumor biology, all previous radiation and chemotherapy received, multiple pain and function scores, medications and time to fracture or time to death. We have trained a convolutional neural network (CNN) with AP X-ray images of 546 patients with metastatic bone disease of the proximal femur. The digital X-ray data is converted into a matrix representing the color information at each pixel. Our CNN contains five convolutional layers, a fully connected layers of 512 units and a final output layer. As the information passes through successive levels of the network, higher level features are abstracted from the data. The model converges on two fully connected deep neural network layers that output the risk of fracture. This prediction is compared to the true outcome, and any errors are back-propagated through the network to accordingly adjust the weights between connections, until overall prediction accuracy is optimized. Methods to improve learning included using stochastic gradient descent with a learning rate of 0.01 and a momentum rate of 0.9. We used average classification accuracy and the average F1 score across five test sets to measure model performance. We compute F1 = 2 x (precision x recall)/(precision + recall). F1 is a measure of a model's accuracy in binary classification, in our case, whether a lesion would result in pathologic fracture or not. Our model achieved 88.2% accuracy in predicting fracture risk across five-fold cross validation testing. The F1 statistic is 0.87. This is the first reported application of convolutional neural networks, a machine learning algorithm, to this important Orthopaedic problem. Our neural network model was able to achieve reasonable accuracy in classifying fracture risk of metastatic proximal femur lesions from analysis of X-rays and clinical information. Our future work will aim to externally validate this algorithm on an international cohort.
Alarmins- also referred to as damage associated molecular patterns (DAMPS)- are endogenous molecules mobilized in response to tissue damage known to activate the innate immune system and regulate tissue repair and remodelling. The molecular mechanisms that regulate inflammatory and remodelling pathways in tendinopathy are largely unknown therefore identifying early immune effectors is essential to understanding the pathology. S100A8 and S100A9 are low molecular weight calcium binding proteins primarily released by activated phagocytes in an inflammatory setting and also secreted as a heterodimeric complex that exhibits cytokine like functions. Based on our previous investigations we sought evidence of S100A8/A9 expression in human tendinopathy and thereafter, to explore mechanisms whereby S100 proteins may regulate inflammatory mediators and matrix regulation in human tenocytes. Torn supraspinatus tendon (established pathology) and matched intact subscapularis tendon (representing ‘early pathology’) biopsies were collected from patients undergoing arthroscopic shoulder surgery. Control samples of subscapularis tendon were collected from patients undergoing arthroscopic stabilisation surgery. S100A8/A9 expression was analysed at transcript and protein level using quantitative RT-PCR and immunohistochemistry, respectively. Primary human tenocytes were cultured from hamstring tendon tissue obtained during hamstring tendon ACL reconstruction. The Immunohistochemistry of tendinopathic tissues demonstrated the presence of S100 A8/A9 in diseased tissues compared to control tissue. In addition, early pathological diseased tissue indicated greater S100A9 expression compared with established diseased pathology. These findings were reflected by data obtained at transcript level from diseased tissues. Recombinant human S100A8, A9 and A8/A9 complex led to significant increase in expression of inflammatory mediators, including IL-6 The presence of S100A8 and S100A9 in early tendinopathic lesions suggests expression is upregulated in response to cellular damage. S100A8 and S100A9 are endogenous ligands of Toll-like receptors (TLRs) and receptor for advanced glycation end products (RAGE). These receptors have known regulatory effects on immune mediated cytokine production. We propose S100A8 and S100A9 as active alarmins in the early stages of tendinopathy and thus targeting of its downstream signalling may offer novel therapeutic approaches in the management of human tendon disorders.
Femoroacetabular impingement (FAI) is a significant cause of osteoarthritis in young active individuals but the pathophysiology remains unclear. Increasing mechanistic studies point toward an inflammatory component in OA. This study aimed to characterise inflammatory cell subtypes in FAI by exploring the phenotype and quantification of inflammatory cells in FAI versus OA samples. Ten samples of labrum were obtained from patients with FAI (confirmed pathology) during open osteochondroplasty or hip arthroscopy. Control samples of labrum were collected from five patients with osteoarthritis undergoing total hip arthroplasty. Labral biopsies were evaluated immunohistochemically by quantifying the presence of macrophages (CD68 and CD202), T cells (CD3), mast cells (mast cell tryptase) and vascular endothelium (CD34). Labral biopsies obtained from patients with FAI exhibited significantly greater macrophage, mast cell and vascular endothelium expression compared to control samples. The most significant difference was noted in macrophage expression (p<0.01). Further sub typing of macrophages in FAI using CD202 tissue marker revealed and M2 phenotype suggesting that these cells are involved in a regenerate versus a degenerate process. There was a modest but significant correlation between mast cells and CD34 expression (r=0.4, p<0.05) in FAI samples. We provide evidence for an inflammatory cell infiltrate in femoroacetabular impingement. In particular, we demonstrate significant infiltration of mast cells and macrophages suggesting a role for innate immune pathways in the events that mediate hip impingement. Further mechanistic studies to evaluate the net contribution and hence therapeutic utility of these cellular lineages and their downstream processes may reveal novel therapeutic approaches to the management of early hip impingement.
Femoroacetabular impingement (FAI) is a significant cause of osteoarthritis in young active individuals but the pathophysiology remains unclear. Increasing mechanistic studies point toward an inflammatory component in OA. This study aimed to characterise inflammatory cell subtypes in FAI by exploring the phenotype and quantification of inflammatory cells in FAI versus OA samples. Ten samples of labrum were obtained from patients with FAI (confirmed pathology) during open osteochondroplasty or hip arthroscopy. Control samples of labrum were collected from five patients with osteoarthritis undergoing total hip arthroplasty. Labral biopsies were evaluated immunohistochemically by quantifying the presence of macrophages (CD68 and CD202), T cells (CD3), mast cells (mast cell tryptase) and vascular endothelium (CD34). Labral biopsies obtained from patients with FAI exhibited significantly greater macrophage, mast cell and vascular endothelium expression compared to control samples. The most significant difference was noted in macrophage expression (p<0.01). Further sub typing of macrophages in FAI using CD202 tissue marker revealed and M2 phenotype suggesting that these cells are involved in a regenerate versus a degenerate process. There was a modest but significant correlation between mast cells and CD34 expression (r=0.4, p<0.05) in FAI samples. We provide evidence for an inflammatory cell infiltrate in femoroacetabular impingement. In particular, we demonstrate significant infiltration of mast cells and macrophages suggesting a role for innate immune pathways in the events that mediate hip impingement. Further mechanistic studies to evaluate the net contribution and hence therapeutic utility of these cellular lineages and their downstream processes may reveal novel therapeutic approaches to the management of early hip impingement.
The objective was to seek evidence of hypoxia in early human tendinopathy and thereafter, to explore mechanisms whereby tissue hypoxia may regulate apoptosis, inflammatory mediators and matrix regulation in human tenocytes. Fifteen torn supraspinatus tendon (established pathology) and matched intact subscapularis tendon (representing ‘early pathology’) biopsies were collected from patients undergoing arthroscopic shoulder surgery. Control samples of subscapularis tendon were collected from 10 patients undergoing arthroscopic stabilisation surgery. Markers of hypoxia were quantified by immunohistochemical methods. Human tendon-derived primary cells were derived from hamstring tendon tissue obtained during hamstring tendon ACL reconstruction. The impact of hypoxia upon tenocyte biology Increased expression of HIF 1a, Bcl-2 and clusterin (hypoxic and apoptotic markers) was detected in subscapularis tendon samples compared to both matched torn samples and non matched control samples (p<0.01). Hypoxic tenocytes exhibited increased production of proinflammatory cytokines (p<0.001), altered matrix regulation (p<0.01) with increased production of Collagen type III operating through a MAPK dependent pathway. Finally, hypoxia increased expression of several mediators of apoptosis and thereby promoted tenocyte apoptosis. Hypoxia promotes expression of proinflammatory cytokines, key apoptotic mediators and drives matrix component synthesis towards a collagen type III profile by human tenocytes. We propose hypoxic cell injury as a critical pathophysiological mechanism in early tendinopathy offering novel therapeutic opportunities in the management of tendon disorders.
Surgical site infection (SSI) is an important outcome indicator. It is estimated that 70% of post-operative infections present after discharge. A reliable post-discharge surveillance (PDS) method is yet to be described. The aim of this prospective cohort study was to assess the reliability of patient self-diagnosis. Telephone questionnaires were used following hip and knee prosthetic surgery. A trained validation nurse checked the wounds of all patients reporting problems and a sample of those who did not. 376 elective hip and knee arthroplasty procedures from 363 patients were included. In-patient infection rate was 3.1% (13 of 422 procedures) and post-discharge infection rate was 5.2% (22 of 422 procedures). Results suggest that patients can reliably self diagnose SSI. The sensitivity of the procedure (the probability that the telephone surveillance will detect an infection given that the patients has an infection) was 90.9%. The specificity (the probability that the telephone surveillance will report no infection given that no infection is present) was 76.6%. Hence telephone PDS of SSI is a valuable means of identifying accurate rates of hospital acquired infection following surgery. In this study population, 41% of infections were diagnosed post discharge, which is lower than has previously been estimated. PDS of SSI is necessary if accurate rates of hospital acquired infection following surgery are to be available.
The incidence of infection remains 1–2% after primary total joint arthroplasty and even higher after revision procedures in spite of advances in prophylactic antibiotics and clean air operating theatre environment. Detection of low-grade infection in a prosthetic joint can often be very difficult. None of the investigations available so far have 100% sensitivity and specificity. This has huge implications on the subsequent treatment, cost and patient morbidity. Revision of an unrecognized infected arthroplasty may lead to less satisfactory results in a high proportion of cases. We utilized Polymerase Chain Reaction, a molecular biology technique to detect bacterial DNA from the synovial fluid of patients undergoing revision surgery. We prospectively assessed 70 patients undergoing revision arthroplasty (57 hips and 13 knees). Each patient was pre operatively assessed clinically and radiologically. ESR and CRP results were noted. During revision, synovial fluid and tissue cultures from capsule, bone and bone-cement interface were obtained. None of the patients received pre or intra operative antibiotics till the specimens were taken. Standard microbiology and histology study were done on tissue samples. In addition Polymerase Chain Reaction study was done on the synovial fluid. In this method, DNA is extracted from the bacterial cell, it is polymerized and finally visualized by gel electrophoresis. Post operatively patients were followed up at regular intervals. Diagnosis of infection included correlation between clinical, radiological and laboratory investigations along with intraoperative findings, tissue culture and histology results and a period of post operative follow up (12 months to 36 months). Six (8%) of the 70 cases that had revision arthroplasty were clinically infected. Polymerase chain reaction was positive in 25 cases, tissue cultures were positive in 5 cases and histology was positive in 5 cases for infection. PCR showed sensitivity and specificity of 83% and 69% respectively. Tissue culture showed sensitivity and specificity of 83% and 100% respectively. Histology showed sensitivity and specificity of 83% and 100% respectively. 20 out of 25 PCR positive cases did not show any clinical evidence of infection. It is unclear whether this represents contamination during surgery or in the PCR lab. Alternatively this may represent true positive PCR results in cases with low bacterial count that can be detected by ultrasonication of implant and immunofluorescence methods. PCR is more sensitive in detection of bacterial DNA. However it has low specificity and combination of tissue cultures and histology can still provide a reliable diagnosis of infection.
Aseptic loosening of orthopaedic implants has a major financial impact on the Health Service. The process is thought to be caused by wear particles that are phagocytosed by macrophages and hence stimulate bone resorption via a cytokine response. Previous work suggests that factors inhibiting or enhancing bone resorption act through regulation of the OPG and RANK-L mechanism. The objective of this study was to identify the role of RANK-L and OPG within the cytokine response leading to orthopaedic implant loosening. Ten samples of cellular membrane obtained during revision arthroplasty surgery were analysed with basic histological staining, immunohistology and polymerase chain reaction (PCR). In vitro studies were also carried out using explanted cancellous bone, to which PMMA particles were added and bone resorbing osteoclastic cells were identified by their Tartrate-Resistant Acid Phosphatase (TRAP) activity. PCR identified the presence of OPG in all of the periprosthetic samples, with RANK-L shown in 40% of the specimens. Immunoreactivity was shown for CD3, CD68 and RANK-L. In vitro studies confirm that there is an initial burst of inflammatory cytokine activity that then subsequently plateaus. A balance of RANK-L and OPG regulates bone resorption at the bone/implant interface of implants by stimulating a significant initial inflammatory response which leads to loosening.
