Hello, I'm Kevin Shea and we're going to talk about non-operative management of OCD lesions. I have no relevant financial conflicts about this talk. All my conflicts are up-to-date in both the Academy and AOSSM's brochures and online information. Roadmap, we're going to define OCD, talk about its common locations, and particularly talk about healing potential based on age and distinguishing stable versus unstable lesion, and talk about some of the clinical imaging and other things we consider, and talk about non-operative treatment protocols as well. Loss of chondritis, the rock group defined this and this definition has undergone a little bit of evolution, but it's a focal idiopathic alteration of subchondral bone with risk for instability and disruption of adjacent articular cartilage that may result in premature osteoarthritis. There's a clear distinction between the parent or normal and the progeny or abnormal OCD involved bone. The clinical problem is that some of these go on to heal very nicely, but some don't, and this can lead to significant osteoarthritis in young patients in their third or fourth decade of life. We think of this in somewhat as two forms, as a juvenile form, basically people who have open FICEs, about 5 to 15 year olds approximately, maybe a little younger in females. We know that overall there seems to be a better prognosis for healing in many, but not all of these cases, and Eric Wall has looked at this very nicely in his classic publication in JBGS about 15 years ago. Anderson et al. has also outlined the better prognosis for healing in younger patients. In kids and young adults in which the FICEs is closed or closing, the prognosis is worse and you have to follow these more closely and may require surgical intervention. Most children who present to clinic have stable OCD lesions, and so keep this in mind, the most common thing you'll see are those that are more stable, but there are some complaints that they'll show up with, including very nonspecific complaints of pain, maybe activity-related pain. It's usually on the anterior aspect of the knee and might mimic patellofemoral pain to some degree. Physical examination for unstable lesions, they may have more significant pronounced mechanical symptoms. An occasional limp is commonly reported by our parents and patients. Frequently is the presence of a knee fusion and occasionally some mechanical symptoms or crepitants. I think it's important to emphasize that there are probably at least 30% of cases in which OCD will be bilateral, and so you always want to think about the other knee, both with imaging or examination, perhaps even if they're not reporting significant pain on the other side. Plane radiographs are important. Both AP and lateral views are pretty standard, but you may not always see these on the typical AP view, so looking closely at the lateral view and also the tunnel notch views can be very helpful because these will help you visualize the mid to posterior lesions a little bit better, and if you're looking at trochlear or patellar OCD, the merchant or skyline views can be very important, so keep these in mind that three or four views in each series is probably beneficial. Other imaging studies, obviously we might think about doing a better job of characterizing lesion. Sometimes MRI and other modalities can give us a little better idea about the prognosis or severity of the lesion and the chance of healing with operative versus non-operative care. Unfortunately, there's no perfect imaging protocol that will predict the success of non-operative treatment in all cases, and I think that we like to emphasize the importance of what we think is a lesion stability and its potential for healing. Unstable lesions are probably not going to heal, whereas stable lesions have a better prognosis. Radiographs, if you see any loose fragments or an incongruent fragment or there's a clear boundary or separation between the parent and progeny bone, those are concerning findings for suggestions that it won't heal. MRI, there's many things we look at for stability features and we'll talk about that. Arthroscopy though probably remains the gold standard for stability, at least the surface stability, and it helps us distinguish both the stable or immobile lesions and contrast these or distinguish these from unstable or the immobile lesions. Rock group, as we've talked about previously, has broken down into the stable group of the cue ball, the shadow, and the wrinkle in the rug. Lesions are pretty descriptive. The unstable lesions include the locked door, the trap door, and the crater in which the lesion has been lost. I tend to think of OCD lesions in terms of stable lesions kind of like the skin of an apple. If you've got a minor bruise on an apple but the skin is intact, that's a lesion that's probably got pretty good healing potential. But think of a major injury to the apple skin in which the apple skin is disrupted and there may be some separation of the cartilage or the meat of the apple. And here's an example of a trochlear OCD that is quite unstable and starting to detach and produce mechanical symptoms. Other things we look at, DeSmet and other people have looked at four different signs and perhaps a few more. If you have a high signal intensity that's more than five millimeters between the OCD lesion and the underlying bone, especially a linear mark, if there's a lot of homogeneous signal within the lesion as well, especially on T2 images. If you have a focal defect in the articular cartilage surface, that also is a predictor of instability. And if there's a high signal traversing the subchondral plate into the lesion and actually sending the joint, these are signs for instability. DeSmet found that if you have high signal behind the lesion or high signal intensity on a T2, this was predictive of instability on about 70% of lesions, especially after arthroscopic evaluation. This doesn't seem to be quite as predictive in younger patients. Sometimes the lesions look bad in the MRI and yet they're quite stable, but it may be more predictive in older patients. Well, just to summarize some of the things we've learned, it's about who may heal with non-operative treatment. The younger patients, those who are quite skeletally mature with three, four years or more of growth remaining, they don't have significant mechanical symptoms or the presence of infusion. Radiographs show they're pretty small. There's no obvious instability. Evidence of displacement, the contour of the lesion appears normal. And overall, the MRI findings look pretty benign. The Rot Group has developed a little bit of a treatment algorithm and I'll walk you through a couple steps of that. First thing they try to do is determine stability if you've got a more advanced lesion and you've gone beyond x-rays. So you get an MRI and you'll assess the stability. If overall the lesion appears to be stable, you try to have some prediction of the healing that may occur. That's not, we don't have a perfect tool, but we're getting better tools. Dr. Wall presented his study almost 15 years ago when he looked at non-operative treatment and they found almost two-thirds of patients with wide open growth plates would heal after a six-month protocol of neomobilization or activity restrictions. But they went back and looked at their patients and they did identify some predictors of not healing, including larger lesions and the presence of swelling and or mechanical symptoms. And here's the algorithm that they produced. And once again, you assign points for size. Larger lesions have worse prognosis for healing, including length and width. You also assign points for symptoms and this gives you a total point score, which can then be used to give them a probability of healing. In this patient, the size lesion, you know, the symptoms added up to a low probability of healing over 21%. Clearly though, we do see a lot of these patients heal. And here's an example of an 11-year-old male after three months of activity restrictions. In this case, they used an unloader brace and you can show signs of regressive ossification and healing in that lesion, even within a three-month period. So when do you have to think about going to arthroscopy? If initially you have a stable lesion, but at least what appears to be stable, the nomogram predicts an unacceptable probability of healing. This really is a shared decision-making process with the families. Or let's say after three to six months of non-operative treatment, those are times when you might need to think about going to non-operative or surgical treatment and have a careful discussion with the family. We usually start with arthroscopic evaluation of the first step. One thing I like to emphasize though about lesions, you need to follow these long-term. This was a 10-year-old patient who was followed for a couple years and was unfortunately kind of lost for follow-up and then showed up in my clinic at age 16 and showed up like this with a delaminated free body inside his knee. So on age 10 to age 16. So you have to follow these closely, both medially and lateral lesions. We've recently done some work with some of the students at Stanford who are experts in machine learning. We've been looking at the ROCK OCD database to help us develop an algorithm to better predict healing. And we've done some work and I think this can be very important when you're trying to provide family counseling and patient counseling about what the next step is. Success, we define for this machine learning algorithm, is complete healing with returned activity. And the failure of these people who were initially treated with non-operative treatment crossed over into surgery after three months. We had any one subjects that met these criteria and we had 26 successes and 55 failure. And this allowed us to use a logistic regression model. The previous model showed an accuracy of about 65% with an AUC of 0.645. This current machine learning model has an accuracy of almost 75% and an AUC of 0.762. So our model is going to get better with time. Risk factors we identified that predict the absence of healing. Wider lesions, you know, normalized based on the size of patient, but wider lesion and longer lesions don't heal. Lesions that are in the more posterior sagittal zone don't seem to heal as well. Age is a predictor. Younger patients heal better than older patients. Presence of mechanical symptoms is a bad prognostic sign. And once again, the length matters as well. Clearly kind of summarizing size, more posterior location, and age are strong predictors. The machine learning algorithm will improve as we add more data. And we think the database will basically allow us to eventually predict with much more accurate predictions for the family. And as we add more patients, we plan to actually put a healing algorithm on our website so that you can actually plug in patient factors based on imaging, age, etc. And we can give you a score for those patients to counsel families. To break down a non-operative management, we usually think of it as a couple different ways. One is changing activities so you can reduce or remove sports, especially higher-impact running sports. We may try to counsel these children to go into swimming, cycling, golfing, or make some alterations to their training program for less impact. Non-operative management, casting, long-leg casts are used with the ankle-free, but atrophy and stiffness are certainly concerns. And most of our centers have gone to bracing, including unloaded braces, but there are challenges with compliance and getting the kids to wear these and getting to wear them for adequate amounts of time. So just to summarize, we know there's a poor natural history for unhealed lesions. Skeletal and mature have a good prognosis for healing with non-operative treatment. Most, but not all, will heal with non-operative treatment. Imaging can be critical, both x-ray and MRI, to evaluate the potential for healing. But once again, close follow-up is important. In the future, we hope the ROT group will continue working on the healing prediction algorithms and make these available for general use, both for clinicians, patients, and families. And thank you for your attention. If you have any questions, my email is listed below. Thank you.

non-operative management

OCD

lesions

healing potential

prediction algorithms