Abstract

Purpose

Patient-matched instrumentation is advocated as the latest development in arthroplasty surgery. Custom-made cutting blocks created from preoperative MRI scans have been proposed to achieve perfect alignment of the lower limb in total knee arthroplasty (TKA). The aim of this study was to determine the efficacy of patient-specific cutting blocks by comparing them to navigation, the current gold standard.

High tibial osteotomy is a well established joint preserving procedure for the treatment of unicompartmental knee osteoarthritis. Of particular interest are the alterations in knee loading compartments during dynamic activities such as locomotion. Computer modelling can indirectly assess contact and muscle forces in the patient. This study aimed to develop a valid model representative of high tibial osteotomy to assess the medial joint contact force at the knee during gait.

Software for Interactive Musculoskeletal Modelling (version 2, SIMM Inc, USA) was used to develop a model to replicate the effects of high tibial osteotomy surgery on tibial alignment. The program was then used to perform a detailed analysis on gait data collected from two high tibial osteotomy patients preoperatively and 6 months post operatively. Inverse dynamics simulations were conducted to investigate knee joint contact force on the medial compartment of the two patients during the stance phase of their operated limbs.

Significant decreases (p<0.05) in the medial joint contact force were observed during both early and late stance for both patients. Force generated in muscles crossing the knee was found to be the major contributor to the joint contact force. Total muscle force was found to increase significantly (p<0.05) following surgery, however decreased loads were calculated for the medial compartment. The pattern and magnitude of joint reaction force was found to be consistent before and after surgery and replicated the results of previous studies. The HTO-specific model was valid and sensitive to changes in joint reaction force, medial joint contact force and muscle forces crossing the knee.

High tibial osteotomy reduced the medial joint contact force at the knee as a result of the coronal realignment of the limb. Osteoarthritis symptoms were relieved in terms of knee pain and function. Finally, a difference in compensatory strategies was observed between patients. This novel technique allows non-invasive assessment of the mechanical effect of procedures such as HTO. This should allow more accurate planning and assessment of such surgical procedures.

Patient-matched instrumentation is advocated as the latest development in arthroplasty surgery. Custom-made cutting blocks created from preoperative MRI scans have been proposed to achieve perfect alignment of the lower limb in total knee arthroplasty (TKA). The aim of this study was to determine the efficacy of patient-specific cutting blocks by comparing them to navigation, the current gold standard.

25 TKA patients were recruited to undergo their surgery guided by Smith & Nephew Visionaire Patient-Matched cutting blocks. Continuous computer navigation was used during the surgery to evaluate the accuracy of the cutting blocks. The blocks were assessed for the fit to the articular surface, as well as alignment in the coronal and sagittal planes, sizing, and resection depth. Actual postoperative alignment was then assessed by detailed CT scans following the Perth protocol, comparing the results with intraoperative measurements.

All patient-matched cutting blocks were a good fit intra-operatively. Significant differences (p<0.05) in the resection depths of the distal femur and tibial plateau were observed between the cutting blocks and computer navigation for the medial compartment. Cutting block alignment of the femur and tibia in the coronal and sagittal planes also differed significantly (p<0.05) to navigation measurements. In addition, intraoperative assessment of sagittal femoral alignment differed to planned alignment by an average of 4.0 degrees (+/−2.3).

This study suggests the use of patient-matched cutting blocks is not accurate, particularly in the guidance of the sagittal alignment in total knee arthroplasty. Despite this technique creating well fitting cutting blocks, intraoperative monitoring, validated by postoperative CT scans, revealed an unacceptable degree of potential limb mal-alignment, resulting in increased outliers particularly when compared with standard computer navigation.

Computer assisted surgical navigation has played an increasingly central role in total knee arthroplasty (TKA). Given the recognized importance of subtle component position changes in knee function, navigation has emerged as a promising tool for reducing the occurrence of significant malalignment. The ability of this technology to reliably measure multiple parameters intraoperatively allows analysis to possibly identify a correlation between intraoperative computer assisted surgical navigation data and functional outcomes of patients undergoing elective total knee arthroplasty.

Intraoperative navigation data was collected for 121 patients undergoing cemented, posterior stabilized TKA. Three forward stepwise regression analyses were performed to associate intraoperative coronal alignment correction, tibiofemoral external rotation, and alignment under varus and valgus stress with one year outcomes, including range of motion, Oxford and SF-36 scores.

The amount of alignment correction and the maximum flexion achieved intraoperatively were significantly correlated (p <0.05, R-sq = 13%) with clinically measured maximum flexion at one year. Maximum flexion achieved intraoperatively, external tibiofemoral rotation and maximum varus under stress were also significantly associated (p < 0.05, R-sq = 31%) with the physical component of the SF-36 outcome score.

Analyses of computer navigation in TKA to date have primarily focused on precision of sagittal plane correction. Alternatively we have identified four intraoperative parameters that correlate with functional outcome at one year. Correct intraoperative interpretation of navigation data may allow surgeons to make subtle changes in real time to produce superior short-term outcomes for patients.

The question of whether to reconstruct an ACL-deficient knee as early as possible following injury or to delay surgery remains unanswered. Early reconstruction potentially reduces the risk of secondary damage. However, there is also concern regarding the risk of arthrofibrosis if surgery is undertaken too soon. The aim of this study was to investigate whether injury-to-surgery delay determines ACL-reconstruction outcomes at up to 2years post-operatively.

A retrospective analysis of prospectively collected data from 211 knees with isolated primary ACL ruptures was performed. Patients were examined preoperatively, at 6months, 1 year, and 2 years post-operatively using International Knee Documentation Committee (IKDC) and Lysholm scores. Side to side differences in knee laxity were also measured with a KT1000 arthrometer. Spearman's rho correlations were used to associate injury-to-surgery delay with outcome scores.

Outcomes scores significantly increased for both IKDC (p<0.05) and Lysholm (p<0.05) questionnaires. Significant positive correlations (p<0.05) were also found between injury-to-surgery delay and IKDC and Lysholm subjective scores. Strongest correlation coefficients were noted at the 2yr follow-up for both IKDC and Lysholm scores (r = 0.79 and 0.8 respectively). Side-to-side laxity measures also showed significant positive correlations with injury-to-surgery delay at 1 year (r = 0.17) and 2 year (r = 0.41) follow ups. The positive correlation suggests that delayed surgery is positively related to subjective outcomes, as well as objective measures of knee laxity.

However, this relationship also suggests that other factors such as the patient's functional status at time of surgery may play a role in their post-operative function. For example, those who can compensate for the ruptured ligament may function well following delayed surgery. These findings highlight the need for more detailed investigation of the interaction between functional status, injury-to-surgery delay and post-operative recovery.

An ACL reconstruction is designed to restore the normal knee function and prevent the onset and progression of degenerative changes such as osteoarthritis. However, contemporary literature provides limited consensus on whether knee degeneration can be attenuated by the reconstruction procedure. The aim of this pilot study was to identify the presence of early osteoarthritis after ACL reconstruction using MRI analysis.

19 patients who had undergone an ACL reconstruction (9 isolated ACL rupture, 8 ACL rupture and meniscectomy, 2 ACL rupture and meniscal repair) volunteered for this study. MRI's were collected preoperatively and postoperatively for analysis with a mean follow up of 23 months. The Boston-Leeds Osteoarthritis Knee Score (BLOKS) was used for the analysis of the articular cartilage by a consultant orthopaedic surgeon. Scores ranged from 0–3, with 0 being total coverage and thickness of the cartilage and 3 being no coverage. Qualitative analysis was then conducted on each patient to determine if the articular cartilage improved, degenerated, or did not change between preoperative and follow-up scans.

All patients with isolated ACL rupture were found to either have no change or improved articular cartilage scores in their follow up scans compared preoperatively. In contrast, patients with a meniscal repair displayed worse cartilage scores postoperatively. Lastly, of the patients who had an associated meniscectomy, 6 had worse follow-up results, with the remaining patients showing no change or improved cartilage scores.

The present results indicate that patients with an isolated ACL rupture have a reduced risk of developing OA compared to those with associated meniscal injuries. This has implications for analysing the outcome of current ACL reconstruction techniques and in predicting the likelihood of patients developing OA after ACL reconstruction. Future work will involve confirming this pattern in a larger patient sample, as well as exploring additional factors such as time to surgery delay and rehabilitation strategy.

Femoral nerve block is a reliable and effective method of providing anaesthesia and analgesia in the peri-operative period but there remains a small but serious risk of neurological complication. We aimed to determine incidence and outcome of neurological complications following femoral nerve block in patients who had major knee surgery.

During the period January 2003 to August 2008, medical records of all patients undergoing knee surgery by Dr Myles Coolican and Dr David Parker, who had been administered femoral block for peri-operative analgesia, were evaluated.

Patients with a neurological complication were invited take part in the study. A detailed physical examination including sensory responses, motor response and reflexes in both limbs was performed by an independent orthopaedic surgeon. Subjective outcome and pain specific questionnaires as well as clinical measurements were also collected.

Out of 1393 patients administered with femoral nerve block anaesthesia during this period, 28 subjects (M:F= 5:23) were identified on the basis of persistent symptoms (more than three months) of femoral nerve dysfunction. All the patients had sensory dysfunction in the autonomous zone of femoral nerve sensory distribution. The incidence of neurological complications was 2.01%. One patient was deceased of unrelated causes and five patients declined to participate in the study. 14 patients out of the 22 have been examined so far. Nine cases had a one shot nerve block and five had continuous peripheral nerve block catheter. Areas of hypoesthesia/anaesthesia involving femoral nerve distribution occurred in 7 subjects and hyperaesthesia/paresthesia occurred in four. One subject had a combination of hypoesthesia and hyperesthesia in different areas of the femoral nerve distribution. Three subjects had bilateral symptoms following bilateral simultaneous nerve blocks. Dysesthesias in the affected dermatomes were found in seven cases and paresthesias were found in eight cases. Douleur Neuropathique en 4 questions (DN4) score of ï³ 4 was found in all the patients (average value: 5.55). The average scores for tingling, pins and needles and burning sensation (in a scale from 0 to 10) are 3.8, 3.1 and 2.9 respectively.

The incidence of persistent neurological complication after femoral nerve block in our series is much higher compared to the reported incidence in the contemporary literature (Auroy Y. et al. Major complications of regional anesthesia in France: Anesthesiology 2002; 97:1274 80). The symptoms significantly influence the quality of life in the affected cases and question the value of the femoral nerve block in knee surgery.

Osteotomies are performed in patients with lower limb malalignment, usually associated with osteoarthritis of the knee or instability. The surgery realigns the mechanical axis of the leg by either an opening or closing wedge procedure with the goal of decreasing symptoms, improving function, and delaying the progression of osteoarthritis.

The 103 patients that had undergone osteotomy surgery were studied prospectively, and data was analysed one year post surgery. We examined subjective outcomes, patient history and surgical variables using backwards stepwise multiple regression models to determine whether there were any associations between these.

Subjective outcomes from a total of 103 osteotomy patients at one year post surgery were compared to patient history and surgical variables. All categories of KOOS and WOMAC scores were improved after surgery.

The multivariate models showed that variables significantly influencing the outcomes were pre-operative flexion, pre-operative weight, the size of the HTO plate used and tourniquet time.

Greater pre-operative flexion; lower weight; larger plate used, indicating larger corrections; and lower tourniquet times were shown to result in improved scores. Not all variables influenced all categories of the scores. While flexion and pre-operative weight influenced across the categories of both scores, plate size influenced KOOS pain and symptoms and tourniquet time influenced KOOS sport and quality of life.

Knee flexion and body weight were the most influential variables when considering KOOS and WOMAC outcome scores as a measure of success. The size of the correction may have influenced the pain and symptom scores because patients with greater malalignment may have initially had worse symptoms and their perception of their current function and pain is affected by their previous levels of pain and function. Osteotomy results in improved function and pain scores and our results indicate that there are several variables which significantly influence patient outcomes and may be of greater importance than other variables.

Anterior Cruciate Ligament (ACL) reconstruction is a well established procedure for restoration of stability following ACL rupture. Several methods exist for fixation of soft tissue grafts on the tibia, without general agreement about the optimal method. This study compared two different methods of tibial fixation using hamstring grafts in ACL reconstruction.

113 consecutive patients were randomized into two groups at the time of surgery. In group one, fixation was with a metal interference screw (RCI) and staples and in group two, with a polyethylene screw and sheath (Intrafix). Evaluation of outcomes was conducted using KT-1000 arthrometer, Lysholm, IKDC subjective and Mohtadi scores.

7 reinjuries occurred within the time frame of the study, mostly related to sporting injuries, with 5 in group 2. 81% of remaining participants were successfully followed at 2 years post surgery. No significant difference in mean KT-1000 side-to-side measurements was found between groups at an average follow-up of 30 months (1.5 ± 1.9mm and 1.8 ± 1.9mm, respectively; p > 0.05). The mean Lysholm score for group one was 65.2 ± 15.5 preoperatively and 90.8 ± 9.5 postoperatively; for group two these scores were 62.0 ± 20.7 preoperatively and 88.8 ± 14.3 postoperatively. This improvement in scores after surgery was similar for both groups and was not significantly different between groups (p > 0.05). Both the IKDC subjective and Mohtadi scores showed significant (p < 0.05) improvements postoperatively compared with pre-operatively but no significant difference between fixation groups.

There were no significant differences between the two groups for any outcome value, with both methods of graft fixation producing good results. The newer Intra-fix device had a higher reinjury rate but this was not significantly different from the screw and staple fixation, and on all other outcome measures the Intrafix device was equivalent to an interference screw and staples for tibial-sided graft fixation in ACL reconstruction.

Multi-ligament knee injuries require complex surgery. Hinged external fixators propose to control the tibio-femoral relationship, protect reconstructions and allow early mobilisation. However, a uniaxial hinge may be too simplistic for such a complex joint. We investigated the influence of an external fixation device on ligament strains and joint contact forces.

Six fresh frozen cadaveric lower limbs (41–56 years old) were obtained. Displacement transducers (Microstrain, USA) were attached to mid-substance lateral (LCL) and medial collateral (MCL) ligaments, and the anterior and posterior cruciate (PCL) ligaments through minimal soft tissue incisions. Joint pressures were measured by transducers (Tekscan) introduced in the medial and lateral compartments through small sub-meniscal arthrotomies. Flouroscopic imaging was used to construct the hinged fixator centred over the epicondylar axis. Ligament tensile strains and joint contact forces were determined through a passive arc of 20 to 110 degrees of flexion and extension, with and without the external fixator (ExFix^™, EBI Biomet Australia).

The application of the external fixation device resulted in minimal change in the mean peak percentage strain of the PCL, MCL and ACL ligaments, while the LCL peak percentage strain decreased. Generally the peak percentage strain for each ligament occurred at or near the same flexion angle in both the un-instrumented and instrumented case within each limb, but the peak percentage strain flexion angles varied significantly across limbs. Peak joint contact forces increased significantly (p < 0.05) in the lateral compartment after attachment of the external fixation device. There was no difference seen in the medial compartment joint contact forces.

This study shows that a uniaxial hinged external fixator can be used in a multi-ligament reconstructed knee to maintain joint congruence and allow early postoperative rang of motion without compromising the results of reconstructions or repairs.

Understanding of mechanical factors influencing knee joint loading is crucial for insight into OA progression and development of prevention and treatment strategies. High tibial osteotomy (HTO) changes knee alignment. Forces and moments should also be altered and reduce loading on one compartment.

15 subjects undergoing high tibial osteotomy were enrolled in the study. Markers were placed on prominent anatomical landmarks to indicate 12 body segments. Three dimensional positions of each marker were calculated using fourteen cameras (Eagle 8 mm, Motion Analysis Corp.) recording at 100Hz and a motion analysis system (EvaRT4.6, Motion Analysis Corp.). Three-dimensional external moments and inter-segmental joint forces were calculated using inverse dynamics in the Kintrak^™ software. Kinematic and kinetic data from

The AMMFL is an anatomical variant of the attachment of the anterior horn of the medial meniscus to the posterolateral wall of the intercondylar notch. It is distinct from the meniscofemoral ligaments of Wrisberg and Humphrey. This large series prospectively documented its incidence and any associated meniscal or chondral pathology. The study period was from September 2006 until December 2007.

All patients that underwent arthroscopy of the knee for meniscal, chondral or ligamentous pathology including arthroscopic anterior cruciate ligament reconstruction were included. The procedures were performed by the two senior authors, according to their standard protocols at one of two hospitals.

All the findings from the arthroscopies were prospectively recorded in a standardized datasheet. This recorded all meniscal, chondral and miscellaneous pathology including the presence or absence of an AMMFL. This data was entered into a database including all patients.

The results of 401 arthroscopic procedures were recorded during the study period. Of these patients, 14 were found to have AMMFLs, resulting in an incidence of 3.49%, higher than previously reported. (Anderson et al describe an incidence of 0.44% from a combined retrospective and prospective review.)

The associated pathology was most commonly a radial tear of the medial meniscus, found in six patients. One had a bucket handle tear of the medial meniscus. There were three lateral meniscus tears, two of which were associated with an ACL rupture. Two patients had an ACL rupture as the only other pathology at arthroscopy. One patient was found to have a ruptured AMMFL as her only pathology.

This anatomical variant is more common in this Australian sample than has been described in the literature, and there seems to be a relationship between the presence of the AMMFL and a particular pattern of medial meniscus tear, suggesting an influence of this anatomical variant on meniscal pathology.

Introduction and Aims: Anterior cruciate ligament (ACL) injuries have historically been classified as non-contact or contact based on the mechanism of injury. The purpose of this study was to establish a detailed correlation between mechanism and the associated osteochondral, meniscal and other injuries to improve understanding of this common injury and its outcome.

Method: A descriptive analysis of prospectively collected data on ACL injuries requiring reconstruction between 2000 and 2004 was completed. Mechanism of injury was clearly elicited and correlated with clinical, radiologic and operative findings. Magnetic resonance imaging (MRI) was performed on all patients to analyse patterns of ACL rupture and associated osteochondral, meniscal and ligament injuries. Osteochondral injuries were analysed by a musculoskeletal radiologist according to location, intensity and depth. Intra-operative documentation of intra-articular injury pattern was also performed and correlated with MRI findings. Classification into ‘active’ (non-contact) and ‘passive’ (contact) mechanisms was completed and correlated with injury pattern.

Results: Seventy patients were identified with appropriate clinical, radiologic and operative data. A thorough review of the events surrounding the injury was documented. Forty-six patients described an active mechanism and 24 patients a passive mechanism of injury. Clinical examination demonstrated a similar proportion of medial collateral ligament injuries in each group. MRI within three months of injury demonstrated occult osteochondral lesions or ‘bone bruises’ in the majority of patients. Clear distinguishing patterns of femoral and tibial osteochondral injury were identified in the active and passive groups. Depth of osteochondral injury was most commonly classified as at least two-thirds the distance to the physeal scar in both groups. Intensity of the abnormal edema-like signal in the marrow of the distal femur and proximal tibia was most commonly classified as severe in both groups. Lateral meniscus injury was more common than medial, and was found in the majority of patients, more commonly in the passive group.

Conclusion: Although surgical techniques continue to improve, the ACL injury mechanism and its relation to intra-articular pathology is less well defined. This study defines either ‘active’ or ‘passive’ mechanisms, with implications for likely associated osteochondral and meniscal injury. This gives valuable insight into the ACL injured knee, its management, and eventual prognosis.

Introduction and Aims: Knee dislocation is an uncommon but serious injury. This study assessed the initial mechanism of injury, pattern of ligament injury, osteochondral and peri-articular soft-tissue trauma, and associated neurovascular injuries in the multi-ligament knee injury. Outcomes following operative and non-operative management were reviewed.

Method: Retrospective review of patients with multi-ligament knee injuries was performed. Inclusion criteria were either a confirmed knee dislocation, or complete rupture of two or more ligaments requiring reconstruction. Systematic review of hospital records and imaging was performed for all patients, and clinical assessment, including validated outcome scores, were performed in the majority of patients. All reconstructive surgery was performed by the two senior authors.

Results: Forty-five patients with 47 knee injuries were identified over a 13-year period (1990–2003). The most common mechanisms of injury were motorcycle and motor vehicle accidents. Other mechanisms included pedestrians hit by cars, sporting injuries and falls. Approximately half had a documented knee dislocation, while the remainder were located at the time of presentation. Vascular injury occurred in around 25% of patients, all having positive clinical findings. Routine angiography was not performed in the absence of positive clinical findings. Neurological injury also occurred in approximately 20% of patients. Transient neuropraxia was more common than permanent nerve palsy, and there was an association between neurological and vascular injury. Associated injuries were varied, the most frequent being long-bone fracture. A significant number of patients had no associated injuries. Injury patterns were varied, though correlations were found between the reported mechanism and the pattern of ligament rupture and osteochondral injury. The majority of cases were managed with operative repair, and assessment of outcomes revealed that most returned to a good level of function, with some minor objective residual laxity and/or stiffness.

Conclusion: Multi-ligament injuries of the knee are uncommon but serious injuries with potentially catastrophic consequences. This detailed analysis provides correlation between mechanism and resulting injury to the knee and periarticular structures. The analysis of this large series provides valuable information to better understand natural history, and improve future management.

Introduction Reflex sympathetic dystrophy (RSD) as a complication of total knee arthroplasty (TKR) is rarely mentioned. The literature has been limited to 58 cases of RSD in TKR, a prevalence of 0.8% of all TKR done. No previous reports give a clear understanding as what to expect in the long term after the diagnosis of RSD post TKR has been made nor do they report the struggle that patients undergo to achieve their result.

Methods We report on 11 cases of RSD diagnosed post TKR, operated on by one of us from 1991 to 2001. All patients met diagnostic criteria for Complex Regional Pain Syndrome, Type 1. Specifically they exhibited slow post-operative recovery and delayed return of normal function. Flexion was limited and cutaneous hypersensitivity was present along with temperature changes in the limb. These patients were evaluated using general and disease specific outcome tools previously validated in the literature, the SF-36 and WOMAC scores. In addition they were evaluated clinically at minimum two years following resolution of symptoms.

Results We found that once appropriate treatment had been instituted, which in our case was manipulation under anaesthetic in the painfree phase of CRPS-I, the majority of our patients reported higher scores on the bodily pain section of SF-36 however these were still lower than age matched controls of pre-operative osteoarthritic patients as determined by WOMAC scores. In general, though patients had poorer SF-36 and WOMAC scores than primary uncomplicated TKR, they did significantly better than primary osteoarthritics without surgery.

Conclusions This would suggest that when appropriately managed, RSD after TKR does not hold the dire prognostic consequences as previously thought.

Introduction The aim of this study was to assess trends in the circumstances of spinal cord injury in all codes of football played in Australia in 1997 to 2002, and to combine and contrast these findings with those of identical studies done covering earlier years (1960 to 1996).

Methods A retrospective review of all spinal cord injuries occurring in all codes of football 1997 to 2002, combining and contrasting the results with identical studies done covering the years 1960 to 1985 and 1986 to 1996. Every football player with a documented spinal cord injury admitted to one of the spinal cord injury units across Australia was included. Data was recorded by way of record and radiograph review, and patient interview.

Results Fifty-four footballers were admitted to the spinal injury units over the period. The average yearly frequency of injuries over the study period was higher than the period 1986 to 1996, and similar to the period 1977 to 1985. The annual incidence of injury was lower in every sport except soccer, although data still remains to be collected from Victoria which may affect the incidence pertaining to Australian Rules. Rugby League had the biggest decrease in incidence. Most notable was the absence of any scrum injuries in league, down from nine (24% of all league injuries) in the prior study. Scrums sustained at engagement remained a prevalent cause of injury in Union. They by far predominated over those in collapsed scrums, reversing the trend towards the latter noted in the prior study. One-third of scrum injuries were in adult front-rowers who had played between one and four games in the front-row in their careers. The incidence of schoolboy injuries overall decreased substantially. The tackle accounted for all League and 40% of Union injuries. Over 75% of known tackle injuries on the ball carrier involved two or more tacklers at once. A much smaller percentage of patients remain wheelchair dependent (30%) than in the last study, and nearly 15% returned to near normality.

Conclusions Spinal cord injuries remain a significant concern in football, particularly the rugby codes. While the incidence overall may have slightly decreased, attention is needed to enforcing scrummaging laws, particularly in adult rugby, and focusing on the gang tackle as a cause of increased injuries in League and Union. An adequate compensation scheme and a national registry also need realisation.

We report the results of arthroscopic removal of loose bodies and abnormal synovium from 18 knees with primary synovial chondromatosis. After a mean of three years, six months (range one to 10 years), 14 knees were either symptom-free or had only minor symptoms. Three of these had required two arthroscopic operations. Three patients were improved but not cured and there was one failure. The results were better than the published results of open operation for this condition. Three patterns of macroscopic appearances were noted: four knees had large lesions covered by normal synovium, 10 had small fragments of cartilage lying in or on the synovium and four had only free fragments of cartilage in the joint cavity but none in, on, or under, the synovium. These three appearances may represent three different disease processes.