Breeding Strategies to Prevent Canine Hip Dysplasia with Dr. Gail K. Smith, VMD, PhD

Dr. Gail Smith, VMD, PhD is Professor Emeritus of Veterinary Orthopedic Surgery at the University of Pennsylvania and the Director of PennHIP. His research focused on the early diagnosis of hip dysplasia and resulted in the development of the PennHIP method for measuring the looseness of hip joints, which is a main cause of canine degenerative joint disease, also known as osteoarthritis.

Dr. Smith has no shortage of amazing accomplishments! He has contributed more than 100 publications on dog health, including research on the effects of diet restriction on aging and osteoarthritis, estimates of the prevalence of hip dysplasia, and risk factors for joint disease. He conducted a longitudinal study that followed a cohort of Labrador Retrievers for their entire lives to understand the trajectory of hip dysplasia. He has received research awards and accolades from the American College of Veterinary Surgeons (AVMA), the World Small Animal Veterinary Association, and Smith Kline Beacham among others, and he was a 2017 recipient of the Albert Nelson Marquis Lifetime Achievement Award.

If you liked Dr. Smith's talk, you can find all of the Health Symposium videos here!

Transcript

Gail K. Smith [0:15] I’m going to talk today a little bit about orthopedic screening. My past is one of orthopedic surgeon, bioengineer. I spent pretty much my whole professional life working on hip dysplasia. There are other orthopedic diseases that we should be screening for (and I might say something about them at the end very briefly): patella instability and elbow dysplasia. I have less to say about them; that’s for sure. There’s less research. You also have to know that the dog is a wonderful model for most human orthopedic diseases. There are rarer diseases but the dog serves as a really good model for a lot of diseases. Obviously I can’t talk about all those diseases, but today I’m going to focus on hip dysplasia. I’m going to try to get through so we have some time for questions. Hopefully you’ll have some questions!

[1:03] Why am I most interested in hip dysplasia? The reason I’m most interested in hip dysplasia is it’s really common—really, really common; I’m going to press that upon you—and it costs dog owners a whole lot of money. We did an analysis a few years ago. We haven’t published this, but of the $103.6 billion per year that Americans pay for pet expenses (this was by the AVMA just this year), hip dysplasia (by our estimates with insurance company data that are not published yet) costs Americans $800 million to $2 billion dollars a year. That’s 2% of the overall money spent on all pet expenses. It’s a huge problem. And an expensive problem. Keep that in mind, because often people say, “It’s just too expensive to do tests. It’s too expensive to go that route.” The disease is more expensive than not looking and eliminating the disease. PennHIP is a global effort. (I am the inventor of PennHIP, so that’s full disclosure. You can accuse me of being biased, but I’ve done a lot of research, and I’m going to share with you as much of that research as I can.)

[2:24] I should say I am now retired and an emeritus professor at Penn. I now work for AIS, which is a division of Mars Incorp.  Let’s review hip dysplasia. I think probably most people in the audience know this. It’s a developmental disease. It’s not a one-gene disease. I wish it were. It would be a lot simpler to deal with. It has complex inheritance. It’s a polygenic disease. That makes it very difficult. It’s also multifactorial, so you can put certain dogs in different environments, and even though they have the same susceptibility, some will manifest disease and some won’t. That makes it really hard to tease out the disease. But let me just make this whole lecture really simple, easy, straightforward: the key factor to hip dysplasia—not the only factor, but the key factor that we have identified—is hip laxity. That’s not even new. What’s new in PennHIP is the way we measure that hip laxity. We have always (as a veterinary profession) been focused on hip laxity. Nothing new there. This is an example of a normal dog, normal hips, in a standard OFA projection. This is not a normal dog. It’s pretty easy to see. These hips are nicely seated, and these hips are not. There’s loose hips. If you follow the 6-month Golden Retriever, just 15 months, it has all the tell-tale signs of hip dysplasia, signs of OA, a lot of mushrooming of the femoral head. That is the basis of hip dysplasia. Why that bothers is this is a really painful disease. This transitions to OA from just laxity is painful. 

[4:25] Look at a normal hip up in the upper left here. This is an 11-year-old dog that had hips all its life, and you see the glistening cartilage in the joint capsule. It’s really thin. As compared to this schematic, this is a diagram of having cartilage erosion of hip dysplasia. That’s because this part of the femoral head was rubbing on this part of the acetabulum. The membranes thicken, and you have all these osteophytes that form cartilage bone. This is just a terrible example of a dog with hip dysplasia that shows how non-round the head is, and that it’s bone rubbing on bone. It’s at times like this when people have this much erosion of cartilage, you get total hip replacement. 

[5:16] So I’m going to go through diagnosis really quickly. I’m going to just scan through what doesn’t work: clinical signs doesn’t work. It’s not a definitive tell for hip dysplasia because some dogs have terrible hip dysplasia but they don’t have clinical signs. Palpation is something that’s done routinely in the human nursery, and it’s often done in puppies in veterinary medicine, but it hasn’t been standardized, so it’s not received wholesale endorsement. In fact, radiologists back in the ’60s ridiculed people who based their diagnosis on joint laxity. Keep this in mind! They were ridiculed because the people who were trying to measure joint laxity with their hands (not quantifying)—they were saying that just about every dog had hip dysplasia. The radiologists said that can’t be true. I’m going to show you data to show that’s pretty much the truth. We stopped doing palpation in the ’60s. Had we kept doing that, I think we’d be further along than we are now. Radiography became the convention. Even though radiographs are used all the time, there is no gold standard. Radiographs do not accurately tell you whether a dog is absolutely diseased or absolutely normal. 

[6:38] I talked to you about clinical signs. I just threw this in. I’ll go, quickly, over it. Here’s a dog. A 1-year-old Golden Retriever. Really painful. Shows no OA yet, just shows a lot of laxity. This dog hurts. This dog puts all its weight to the front legs. Can’t do stairs. Can’t walk half a block. This dog—a 6-year-old Golden Retriever—the owner thinks is normal. Well, he has all this osteophyte formation on the femoral head; it looks absolutely horrible. The owner thinks this dog just doesn’t have a problem: walks with him, jogs with him, and he doesn’t get lame. 

[7:25] He did when he was a puppy, but that was the only time. We also have palpation methods in veterinary medicine called the Ortolani test. You can palpate laxity in young dogs. The problem is there are too many false negative diagnosis; that is, when hips begin to remodel, this thump that you feel on the femoral head pops back into the joint and is diminished. You don’t feel the thump, and therefore you’ll say this dog has good hips when it doesn’t. So it’s not as reliable, not as well researched.

[8:01] What is the convention globally? The hip extended radiograph. This radiograph here, with the radiographic plate behind the dog’s pelvis including in these, and in the United States, we evaluate dogs at 2 years of age. In the rest of Europe and most of the world, 1 year of age is the standard age. I’ll have something to say about that. That is a problem. But in the United States, the OFA rates dogs as normal (excellent, good, and fair) or dysplastic (mild, moderate, severe). We have a borderline category for dogs they can’t decide. Those dogs are, in 6 months, re-evaluated. That’s the OFA method. We call this the hip extended radiograph. 

[8:48] When you look at dogs, you can see the range of changes. Here’s a normal dog, nicely positioned with the femoral heads in the acetabulum, deep-seated. And then you have a dog that would have mild hip dysplasia, and that means the dog doesn’t have OA yet. The dog, however, on this left side does have laxity, some laxation; the femoral head looks like it’s too far out. And this dog has severe hip dysplasia. This dog has loads of osteophyte formation and OA around the femoral head and the acetabulum on both sides. This would be severe hip dysplasia. Everything in between can be diagnosed.

[9:31] So, what the radiologists key on when they make a diagnosis of hip dysplasia is pretty simple. It’s two things. One is subluxation, and I’ll call that laxity on the hip extended view, and this would be the hip extended view. The other is osteoarthritis, shown here—all the osteophyte formation in the remodeling of the joint. If you see any subluxation or if you see any osteoarthritis, the dog is given a diagnosis of one grade or another of hip dysplasia. 

[10:04] It’s that simple. Subluxation comes in a full range, from really tight tips to really loose hips. In the US, we don’t measure laxity. It’s eyeballed by the radiologists. In Europe, in the BBAKC (the British System), they use the Norberg angle to measure the laxity. Somewhere between this hip and this hip, it probably triggers the mild diagnosis of hip dysplasia. It’s not moderate hip dysplasia yet because there’s no OA. That’s one key factor. Subluxation. 

[10:38] The other key factor is osteoarthritis. This is a 13-year-old dog. I show this every time I give a hip dysplasia lecture because this dog has profound osteoarthritis—more than I’ve ever seen. I have to tell you, at 13 years of age, this mixed-breed dog came to me for a cruciate rupture, not for hip dysplasia. You can see how far I can pull the hips into extension. It’s unbelievable. Hip OA can be profoundly advanced or it can be like this. This dog received an OFA as good. But what we see here are some new osteophytes. In these three papers, one’s called the CCO. That’s this comma-shaped piece of bone here. The other is called the CFHO that surrounds the cartilage bone. We have followed dogs longitudinally. We feel pretty strongly that the evidence supports that these are either osteophytes or predictors of later osteophyte formation. These studies have created a lot of interest and now there are 5-6 studies that radiologists have done; it’s interesting literature to review. I would advise everyone to pick up the literature and review it because it’s still being decided what meaning these new osteophytes have. 

[12:07] Maybe this sounds obvious, but why are we involved in hip screening? When I was a student, it was for one reason. It was to help dogs in future generations. We want to apply the principles of quantitative genetics, and I’ll talk about some of those. If we select normal parents and breed them to normal parents, we should have normal offspring. That’s the view. That’s what we’ve held as a view for decades, generations! But now we know there are things we can do to help dogs in the current generation. We can identify dogs at risk and highlight PennHIP as a way to do this. And then we can impose preventative measures. One of the most profound, important preventative measures you should take home with you was shown in a lifespan study (I was fortunate to be invited into that study—phenomenal study following 48 Labrador Retrievers for life). By keeping the counterpart littermate of this group of 48 fit, by feeding them less, they had huge health benefits. Even though they had the same susceptibility to getting OA as a pairmate in the litter, the ones that were leaner had it less severe, got it later, and suffered less from it. This is the most important thing you can do to help any dog that you suspect is susceptible to having hip dysplasia. There are medications like Adequan that are supposed to later the formation of OA, slow it down. That’s one that’s FDA-approved. There are others kind of in the pipeline, not any that are really ready for general consumption. And there are a few surgeries that are said to be preventative. They also require a little more research. Some look pretty promising, but we really would like to have something we can do. If we know a dog is going to go on and get OA and the pain of hip dysplasia, we’d like to offset that as much as possible. But treating the dogs in the current generation and in future generations requires accurate screening tests.

[14:28] I have concerns, and so do a lot of people in the literature. It’s pretty full of those concerns. The studies have been done. Of the OFA convention, this isn’t just the OFA. It’s the hip extended, subjectively scored radiograph. I know people are probably throwing things at the screen as I say this. I’m not saying it has no value. I’m going to give you studies to show you the value of hip extended radiology. But just hear me out before you get too biased against me. The subjective rating system is highly variable. That’s been shown by many, many studies globally. Different radiologists read the same field films differently. Two years of age in the USA (and definitely 1 year of age elsewhere) is too late for breeders, and it’s too early for OA. I’m going to prove that to you in a second with the lifespan study. Voluntary radiographic submission is terrible. It causes selection bias. Only the best hips get sent in for evaluation, and this is a flaw in all the systems that are available globally. It’s not just the OFA. We don’t know the true estimate of how much disease is out there. It underestimates the true incidence. We can’t monitor progress. How do you know you’re doing well if you have a biased data set? I’m going to prove to you that the OFA excellent phenotype is not a good target phenotype. It’s not bad, but it’s flawed. I’ll have data from a brand-new study in 2019 to prove that. We’re making slow progress because this thing I’ll share with you in a second called EBVs are not routinely calculated. Most dogs are bred based on their individual phenotype, on their own hip status, not on relatives. EBVs incorporate the entire family, the kindred, of the dog of which you have interest.

[16:33] Here’s how it works. I’m not trying to blow you away, but if you’re looking for the estimated breeding value (EBV) of your dog, that male, you don’t just look at that male’s hip. He looks at the hips of its progeny, its sibs, its parents. You put it into this formula and you get a value which is actually the genetic value of that dog. This dog could have the best possible hips in the world. All the relatives could be horrible, and therefore it’ll be reflected in this EBV. If you don’t do this evaluation, you think you’re breeding your dog with the best hips in the world. It’s that simple. 

[17:16] Again, I got invited into the Nestle Purina. It was Ralston Purina at the time. A study back in the ’80s. It taught me more about hip dysplasia than was known at that point. It’s unbelievable, the study. I’m just going to show you one dog. This is something we never thought could happen. This is a dog who had normal hips at 2 years of age (a Labrador), normal hips at 5 years of age, normal hips at 8 years of age. If the study ended at 8 years of age, you’d say keep breeding that dog! That’s the best dog we ever had! However, at 11 years of age, this dog started to show OA and, amazingly, by 14 years of age, this dog had severe OA. 

[18:03] Right now it’s 2 years of age. 

[18:06] Here it’s 8 years of age. Hips are normal.

[18:09] Here it’s 11 years of age, and the hips are just starting to show early OA here in the neck region on the right hip. 

[18:19] By 14 years of age, the hips are like this. Severe OA. We never knew this could develop. You don’t know, until you follow this along. There’s a saying that goes, “If nothing kills success clinically, then follow up.” You really have to follow dogs longitudinally before you can assume they’re normal for the end of their life. That’s one of the points I always try to make when I talk to veterinarians and breeders. We should get away from this practice of just doing one evaluation. We want to check that one evaluation later. We’re going to make sure our diagnosis is correct. Nothing helps better than just the passage of time. Let them get older. Take the test again. 

[18:59] I’m not going to show you all the studies. It’s just not enough time. I want to have some questions. But this is the most recent paper, 2020. I see it’s cut off down here. It’s in the Journal of Frontiers in Veterinary Medicine. This is from the BBAKC. It’s about elbow dysplasia: the effectiveness of canine hip dysplasia and elbow dysplasia accrued in the program in 6 UK pedigree breeds. It just shows some improvement in the decline of hip cases. But marginal improvement in elbow dysplasia. We know so much less about elbows. But I’ll show you some of the data for 2 of the dogs through the breeds. This says Newfoundland up here. This is 20 years following. This is in 1990 and this is in 2010 or so. You can see the British system scores from 0 to 106. A lot of numbers. But you can see the dogs up here with the highest scores—you don’t see so many of them. By selective breeding, you’re getting rid of some of the seriously diseased dogs. The problem is you’re not getting this niche bump here from Newfoundlands to shift too much. We would like this to shift to the left. You’d like it to be closer to perfect, as you do these selections. The same goes for Labs. I hope I’m reading this correctly. The letters are so small on my screen. But the same goes for Labradors where there’s this kind of distribution, disease, abnormality in Labradors in 1990. Twenty years later, it looks like you get a spike here in dogs between 0 and a perfect score. You get a little bit higher so-called leptokurtic action to make more dogs. But the mean of the breed is about the same. There’s not a big change after 20 years of selection. Now this is mass selection. This isn’t using EBVs. The paper says that we need to switch to EBVs if you want to go faster in the direction of better hips. I’ll have something to say about that, from the paper we just published in 2019.

[21:15] The interim conclusions are that selective breeding has questionable efficacy using mass selection. Mass selection means you just breed using the individual dog’s record or phenotype. It’s probably because the inheritability of the trait is too low. The inheritability is a number that runs from 0 to 1, and 0 means the dog’s trait has no genetic underpinnings whatsoever; 1 means that its trait is entirely predicted by the genes. 100%. The environment has nothing to do with it. Very few traits are at the extremes. Most are in between. Those traits in between are called quantitative genetic traits. The reason that selective breeding is low or questionable is that I think you can’t create enough selection pressure. There are two things you need to make genetic change. One is inheritability has to be considerable and you need to apply selection pressure. I’m going to demonstrate that in a formula. If we use EBVs, everyone knows and everyone suggests at the end of the paper that, yeah, the progress is pretty slow. But if we just use EBVs, it would be much faster. We’ve got to do that. Military working groups are doing that and are making unbelievable progress. I’m going to share some of that progress with you. Hang in there.

[22:45] This is the formula. I know a lot of people hate math. This is really simple. This is about quantitative genetic traits. This is something everybody can understand. If you’re looking at the genetic change to expect in one generation. What’s that litter going to be if I mate these two parents? It’s very simple. It’s based on the inheritability of the trait, so you want inheritability that’s as high as possible. It’s based on the selection pressure you apply. The mean of the parents minus the mean of the population. It’s that simple. Let’s say you have average parents. When you take the mean of the parents minus the mean of the population, what’s the selection pressure? Zero. No matter how high the inheritability is, you cannot make genetic change in a generation. It just won’t happen. If inheritability is very low, below 0.23, which we suspect it is for the phenotype, then even if you have some selection pressure, you’re not going to get much over that. Keep that in mind. Genetic change is the inheritability (genetic underpinning of the phenotype) times the selection pressure that’s applied.

[23:59] Let me move on to PennHIP and just make some definitions. Then we’re going to compare the two, and I think you’re going to see that. PennHIP is a radiographic diagnostic procedure that’s based on evidence-based medicine. We’ve got almost 40 publications in peer-reviewed journals relate to hip dysplasia and the use of PennHIP. What’s different than any other testing methodology is it trains veterinarians to do it. This is a global effort. We certify them to perform the procedure. What’s most important is that we put all this data in the database and register it. What I should say is in addition to training veterinarians, we require every veterinarian who takes a PennHIP radiograph to submit that PennHIP radiograph. If someone doesn’t want to submit it, then they don’t get a PennHIP radiograph. If a veterinarian doesn’t send in the radiographs, that’s grounds for dismissal from the program. We are serious about hip dysplasia. It’s a painful disease we can be rid of. I mentioned this already, but in 2013, AIS required PennHIP from the University of Pennsylvania. Mars then acquired AIS in 2018. We now do online training. I used to go all over the world training people. Thankfully, AIS helped. We made an online training seminar. It’s very simple. We certify vets in their practices by looking at the repeatability of the radiographs they send us. Once they pass that, they’re certified members. The whole thing is free and provides CE credit. If there are any veterinarians out there, I invite you to get certified. You can also get certified as a technician for veterinary practice. 

[25:54] PennHIP requires heavy sedation or anesthesia. We would prefer anesthesia but heavy sedation works. It’s a radiographic procedure using 3 views, not 1, but 3. There’s the OFA view, the hip extended radiographic (as used globally), but then we do 2 other views in a neutral hip position. There’s a compression view, with the hips pushed in the joint. And one is a distraction view, with the hips pulled laterally. Again, we require anyone who does PennHIPS to send it in. Send the films. Because we do not want to have that selection bias that we talked about being so harmful to our understanding of the disease. Hip scoring is centralized. We read all the radiographs at our analysis center. Hip laxity is quantified with a distraction index. We now do this via computerized applications. It has been done for decades. We subjectively score for OA. We do it just as it's always been done: mild, moderate, severe OA. 

[27:10] I’ll go through how it looks. I think this was a Rhodesian Ridgeback. The hips extended, and then you get the standard extended radiograph—beautiful picture of the hips. That’s radiograph number one. Radiograph number two is a compression view. That’s shown up here in this upper line. The dog’s hips are put in the neutral position. The femurs now poke up at you, looking down at the dog in this kind of position. The veterinarian rotates the tibias externally. That creates the force that puts the hips in the joints. You see this beautiful seating of the round on round. You get a sense for what it really should be like. Then we do the third radiograph, which is the distraction position. We have especially made this distractor that we put between the legs. An assistant holds that down, or we strap it down. When you push the knees together, you fulcrum the hips out of the joint. The dogs have to be sedated. Otherwise this would be uncomfortable. This amount (in a nutshell) of laxity that we measure—the magnitude—is directly related to the susceptibility that this dog will develop OA later in life. It’s that simple. It’s really logical. It’s straight-forward.

[28:28] Just to show you how we calculate the distraction, we have special computerized circle gauges that we put on the head here and we put on the acetabular landmark, and we measure the distance between the centers of the femoral head, and we divide by the radius. That normalizes, for all dog sizes (young dogs, old dogs, big dogs, small dogs), a 0.61 is shown here. It would look the same in any dog that you evaluate. The second view I talked about (the compression view) is this view. We measure that laxity just to see that it's close to 0. We did some research to show if it’s not close to 0, the dog is likely heading for disease. We won’t go into that further at this point.

[29:20] What was really helpful for me in working out this method is I was an engineer first. I could do this mechanical testing on cadaver dogs. Let me say first: no dog was sacrificed at any point for PennHIP. These were cadaver dogs who died for other reasons. In my laboratory at the medical school, we paid for painstaking research on all the different hip positions that would show us maximum laxity. We found the positions where putting the hip will give you 2.5-11x more laxity than you find on the hip extended radiograph. This is the hip in extension. What we found and published on is when you pull the hip into extension, you wind up the fibers in the joint capsule. That actually pushes the hip in the joint. For young dogs especially, before the joint capsule changes, you’re actually masking the very laxity that you want to see. That was published in 1990. 

[30:20] We found early on that things were different in the distraction. Here’s another OFA “fair.” A Golden Retriever. Hips look pretty good. This was early on, in the ’80s. When we did the distraction, there was quite a bit of laxity. No idea what this meant! We had to do the studies to follow up to determine what this laxity means later on. 

[30:45] The studies have been done. By 2001, we had thousands of dogs within each common breed. I’ll take you through this because I know data is foreign to a lot of people. If this is the distraction index, then the tightest possible hip to the loosest possible hip—this is the probability of those dogs that have those indices getting OA. You’ll notice that the tight hip dogs (below 0.2) almost have a 0 probability of showing OA. I think the mean age here was 27-29 months. As the laxity goes up (0.5, 0.6), the probability of showing OA goes up to 100% — almost all dogs with that laxity will get OA, radiographically, by 29 months of age. We did note that German Shepherds separated themselves from the other breeds (these were Goldens, Labs, and Rottweilers and included German Shepherds). German Shepherds are more laxity intolerant. I think it’s because of the way they stand. I wasn’t very popular by mentioning this at the German Shepherd Dog Club. They stand in a way where the hips are flexed even in normal position. When you walk around with your hips flexed, you load your hips many times your body weight. Look at this way. The distraction index is at 0.5. The average Lab will have a 5% chance of showing OA by the time he’s 29 months of age. A German Shepherd will have almost a 45% chance. Same laxity. 45% chance of showing OA! My appeal to the German Shepherd breeders and to those who want performance dogs especially is to try to get them to where they were back in the early 1900s, when German Shepherds were squared off. Just my opinion. I can’t prove that.

[32:50] This is a really important slide. It puts it all together. Hundreds of thousands of dogs. This is 10-year-old data. I don’t know how many dogs are in it at this point. Most popular breeds. These are box and whisker plots. This is like a bell curve looking from the top down. The mean is the dot and the median is the line. The shaded area is 50% of the data and between points, it’s 90% of the data. It’s how the data is distributed. What you can see is that Borzois and Greyhounds—almost all of them have tight hips. Not surprisingly. It’s not a coincidence. Almost all of them are free of having hip dysplasia, especially the performance components. Look at all the other breeds. Goldens, Labs, Rottweilers—the distribution is all greater than 0.3. Remember 0.3 is that point where dogs below it are not susceptible. Dogs above it are. When that guy, John Bardens, in the 1960s was palpating hips and saying all these dogs have laxity—he was right! Look at the laxity in these dogs. Vast amounts of laxity: 0.6, 0.7 in this distribution. Some dogs have no members that are below 0.3. Golden Retrievers—1.8% of them are tighter than 0.3. That means that 98.2% are looser than 0.3. That’s a telling statistic.

[34:38] Let’s talk about the Seeing Eye experience. I’m going to try to wrap up in 5 minutes. Seeing Eye in Marstown, NJ, had 40 years of selection and did everything right. My friend and colleague Eldin Leighton wrote this paper. We pulled together 40 years of data on dogs bred based on the EBV since 1979. I think we ended it in 2015. Shepherds, Labrador Retrievers, Golden Retrievers. All the hips were scored the same by the same radiologist, over that 40-year-period, Dr. Barry. He was a collaborator. He used a 7-point subjective scoring, similar to the OFA type scoring. Different than the OFA; the dogs were younger: 12-14 months. They weren’t 2 years. All along, from the very beginning, only normal dogs entered the breeding pool. Those were hips that were graded 6 and 7. That would be good and excellent. Those were the dogs that were bred. In 1990, when PennHIP came along, we started films at the Seeing Eye as well. 

[35:50] Both tests were done. This is a very complicated-looking graph, but it’s not hard to understand so just bear with me. This is looking at 10 generations of Labrador Retrievers. The data is the same for Shepherds and Goldens—10 generations, that’s 40 years of breeding dogs. This is the proportion of those dogs that were within these qualitative hip-scoring OFA categories. In the beginning, they picked all dogs except for one. They had one dog they bred that wasn’t normal. But these were all excellent or good dogs that were bred in the first generation. Notice in the next generation, within all excellent and good, you get a bunch of dogs that are diseased: severe, mild, or moderate hip dysplasia. They didn’t breed based on the individual phenotype. They bred based on EBV. Just look at what they’ve done. Over those 10 generations, they’ve gotten rid of disease. There are no dogs with disease in the 10th generation. In fact, they’re all excellent. Why do we need PennHIP? This works! You just have to use EBV and everything will work out? No. And I’ll show you why.

[37:12] These are called violin plots, and they really are goofy-looking. I know that. These are Labrador Retrievers right here. If you ask the question: Of those Labrador excellent hip dogs, how many were susceptible to getting OA? These are excellents here. About 50-70% of them are above 0.3 in terms of their distraction index. They are susceptible to getting OA. So while you think you’ve hit that target phenotype that you’ve always wanted—all the dogs are excellent—you haven’t! Those dogs are still susceptible. But a smaller proportion. If you were breeding mild hip dysplasia, these are all susceptible. If you’re breeding fair, more of them are susceptible, above 0.3. Good, the same way. The thing is: the excellent hip score is not a perfect target phenotype. You’ve got to keep understanding that. Seventy percent or more of those dogs are still susceptible to getting OA later in life. Yes, less susceptible because they’re a tighter bunch of dogs. It helps to make them tighter as you’re breeding for better hips. Same thing for the Goldens. Goldens are even worse. Very few Goldens are below 0.3. German Shepherds aren’t bad. There are plenty below 0.3 and they have a narrower range of hip laxity. Keep that in mind. That’s from 2019. If you see these slides, you can copy the DOI and find that paper. 

[38:53] Another paper that just came out in 2020 (JAVMA) is from a colleague, Paul Waggoner at the Vapor Dog Center in Auburn. I wasn’t aware they were doing this study, but this is a very telling study because they were not using EBVs. In their study, they had been breeding dogs for 17 years based on the individual dogs’ hip scores. In the last 3 years, they used PennHIP. They judged both. They did both PennHIP scores and OFA scores. They started doing PennHIP. The normal dogs are the open bars. The black bars are the dogs with dysplasia. The higher the bar, the more dogs have dysplasia. Notice after they started using PennHIP, no dogs had a hip extended score dysplastic. That was a very telling paper because they didn’t use it. I think if they’d used EBV, they would even have better progress. In other words, we don’t have the DIs—the PennHIP’s DIs—from these dogs, so we don’t know the distribution. 

[40:02] As a breeder, how should you use all this information? It’s really simple and straight-forward. If Borzois lived here, in this laxity range almost entirely below 0.3 and these are Golden Retrievers or Labrador Retrievers or Rottweilers—if you just breed dogs where the mean laxity is better than the mean for the overall breed, you will create positive selection pressure. The more the tightness, the DI is better than the mean, the greater the selection pressure. You’d like to breed from over here, if possible. If you start doing that—only breed from this 50th percentile—you will begin to shift this distribution to the left. We don’t know how many generations it’s going to take, but with EBVs, it’s going to take fewer. That’s for sure. Eventually, it’s in the cards through quantitative genetic logic, that we can make Golden Retrievers look like this. It’s simple.

[41:13] This was from 1980. The first pair of dogs we bred. German Shepherds at the time had a wide range. The average German Shepherd was 0.42. Several breeders wanted to get involved. We found one female as 0.19 and one male as 0.21. We bred them. If the heritability was 0, puppies would be all the way over the whole range here. They wouldn’t be under the parents. If the heritability is 1, the puppies will fall entirely under the parents. What happens is the puppies fall somewhat between the mean and the parents. If you look at the distance between the mean of the litter versus the mean of the population versus the mean of the parents, it’s 60% of the way. That means the heritability—just in this one baby—is 0.6. Now that there are many people globally that have done heritability testing on PennHIP DI, it’s somewhere between 0.65 and 0.85. That’s also in the new paper that we published. 

[42:22] Here’s how it works. If you have a dog like this—a German Shepherd—with nice hips, we would try to test that dog young. You can test them as young as 16 weeks. Obviously 6 months or a year is better because the metric gets more accurate. By looking at this dog, you could say wow, it’s 0.3. Very mild to no risk of showing OA in its life. It can still show it. It’s just that the probability is low. Also, that dog being tighter than the mean (the shaded area here is 90th percentile), it’d be a great breeding dog. It’s going to create selection pressure. If you bred it to a female way over here, no. It’s just going to be out. But if you breed it to one over on this side of the mean, it creates substantial selection pressure. Compare that to this dog, also a German Shepherd with really loose hips. We’ve rarely seen Shepherds with hips this loose anymore because they’re tightening up. But here’s one with a hip laxity of 1. We had Goldens with hip laxity of 1.5. It’s amazing how loose the hips can get in some dogs. This dog would not be a breeding candidate. You couldn’t even mate it with a tight-hipped dog on this side of the scale because it’s so far to the right that you’d never get an average that’s better than the average for the breed. It’s not a breeder. But it is disease susceptible. It almost certainly will get OA within the first 2 years of its life. Almost certain to require some medication for pain. May require some surgery if the pain of the condition gets bad enough or the dog begins to lose function and begins to lose muscle mass. This scale tells us a whole lot about how to manage our patients.

[44:13] I’m going to read these conclusions. I don’t like to have read conclusions, but I want you to take this home with you: Selection for better hips based on the conventional hip-extended radiograph will yield little further improvement. Slow improvement, but not much. Even by utilizing EBVs. Excellent hip-extended radiographic phenotype is a flawed phenotype. Scientific evidence supports the PennHIP method of hip radiography as the method of choice to improve hip quality. High heritability and considerable selection pressure are the reasons. Remember that goes into the delta G. A DNA test for hip dysplasia will require a highly validated phenotype like PennHIP to which genomic association can be made. More research, better molecular methods, and an unknown amount of time are needed. PennHIP is a phenotypic test for all dogs that is available now. You don’t have to wait. You can use it now. My last recommendation is to test early and to test often to confirm our earlier findings. That’s not a financially-motivated mandate. That is so we know more about the disease. 

[45:27] I think my time is up. I only had brief things to say about patella luxation and about elbow dysplasia, in that we don’t know much about them genetically. You’ve got to get tested for elbow dysplasia. Patella luxation is congenital. Usually you can get that diagnosed as a puppy. With that, I think I’m done. I’d be happy to answer any questions. 

Dr. Judi Stella [46:09] That was amazing, Dr. Smith! Thank you so much. We are running short on time today, but we would love to have you back to talk about patellas and elbows and continue this conversation with hips at some point in time. One question I do think would be pertinent for this particular forum and what we’re trying to do today: a breeder asked what is the easiest way to describe the difference between PennHIP and OFA in terms of measurements and process? Because they have a lot of puppy owners that want to know about that. What’s an easy way to explain it?

GS [46:47] That’s hard. I guess both methods, especially in young dogs, are key on laxity. It’s just that PennHIP measures the laxity from an engineering perspective, where it’s maximized. So you’ll see more laxity. If you show that that laxity is related to OA later, that’s key. I guess PennHIP is closer to the key factor, which is hip laxity. OFA view looks at laxity too but it’s a blinded view of the total amount of hip laxity. 

JS [47:25] That makes sense. Because this has been validated for puppies as young as 16 weeks of age, is there a breed difference? Is there a time that is best to test dogs if you’re going to use them in your breeding pool? You did mention at some point that it gets more valid or better as dogs get older, but is there an optimal time you’d recommend?

GS [47:52] These are good questions. We did a lot of research, but we didn’t do all the research. There’s so much to do. There’s hundreds of breeds, and we looked at a few. We looked at a couple of ages. We can do most dogs at 16 weeks of age, but there are probably some breeds that we could do younger, and it would be valid. But we didn’t do the test. We know, for instance, that in German Shepherds, 15% of them will get more laxity after 16 weeks of age. Not a whole lot more, but they’ll get more. We’re finding these things out as we keep doing research. You can’t be hard and fast on anything, but we can say that from the paper we published in 1993, the data is there. You can answer those questions and just see what we’ve done. The score gets more reliable as the dog gets older. By 1 year of age, the repeatability of the hip score from 1 year to 2 years is almost perfect. It’s based on a number. It’s 0.91. 1 is perfect. It’s 0.91. That’s really amazing. You do that at 16 weeks and compare to 2 years of age, it’s 0.81. So 0.91 is better than 0.81. If you’re going to select the time, if you can hang onto the dog for a year, keep them a year at least. 

JS [49:22] That’s great. I think that will help out a lot of our breeders because we do get that question quite frequently. Again, we thank you so much. We’d love to have you back to continue this conversation and have more questions. I’m sure our community would love to chat with you again.