 Well, good afternoon everybody. My name is Marco Ramis and I'm one of the associate scientific directors for the Canadian Longitudinal Study on Aging. Thank you very much for joining us for our next installment of our regular CLSA webinar series. I'm very, very pleased to present our presenter today, Dr. Bill Leslie. Dr. Leslie is a professor of medicine and radiology at the University of Manitoba. He has over 300 peer-reviewed publications to his credit and his research interests relate to fracture risk assessment, osteoporosis testing, and other nuclear diagnostic techniques. He's the president of the International Society for Clinical Densitometry, past chair of the Osteoporosis Canada Scientific Advisory Council, and he co-led the 2010 Clinical Practice Guidelines in Osteoporosis Working Group. He's a director of the Manitoba Bone Density Program and co-director of the Winnipeg PET Imaging Center. So Dr. Leslie joins us from Winnipeg today and he's going to be presenting advances in fracture risk assessment. So just for those of you who've never attended a webinar of CLSA before, what we're going to do is everyone is muted and that is just so that we don't get too much feedback on the microphones, which would happen if we were unmuting everyone. So when it comes time for the question and answer period, I would please ask you to type your questions into the chat box, which is located at the lower left of the webinar screen, and I will read your questions to Dr. Leslie and he will respond to them orally. So it'll be approximately 40 to 45 minute presentation and then we'll have time for questions at the end. So without further ado, welcome Dr. Leslie and Bill, I turn the reins over to you. Thank you very much, Mark, and thank you folks wherever you are for joining this session. I will be speaking about a subject near and dear to my heart, fracture risk assessment, and hope to have plenty of discussion time at the end. So with that, I will take you through the first part of this presentation. This is a challenging situation for many when we see patients, the question arises to treat or not to treat. So the case scenario here, a 60-year-old woman, smoker, recurrent falls with a hip T score of minus 2.4, and I can't ask for a show of hands across the country, but this is the kind of situation that crops up regularly in your offices and where the conundrum is that this individual is not quite osteoporotic by her T score but has additional risk factors. How do we factor that into the decision-making process? We're aware that the BMD has been an integral part of risk assessment, osteoporosis diagnosis, and management well on 20 years now. And for individuals over age 50, we focus on the T score, which is the number of standard deviations that bone density is above or below. Peak bone mass is defined at age 20 to 30 years. For individuals younger than 50, we don't actually have a good framework for diagnosis of osteoporosis, and so we tend to focus on whether bone density is normal or below average for age. But for the older individual, we have historically used the T score to define normal bone density, low bone mass, formerly called osteopenia or osteoporosis, and that has been the framework around the diagnosis of osteoporosis and treatment guidelines for a couple of decades. The problems with that approach, though, are outlined in this slide. This is data from Manitoba that Dan Cranney published showing that if you look at the fracture rate for 1,000-person years and you can go from an individual with a bone density that's greater than zero, so better than average, better than my bone density, down to an individual who's well down in the osteoporotic range with very low bone density, that sure enough there is a gradient of increasing fracture risk as one goes from normal to very low bone density measurements. And that's highlighted here when we go from normal to osteopenia to osteoporosis. There's no specific fracture threshold at which a fracture suddenly takes off, but it's a continuous gradient of risk. That's good for bone density, but the problem with bone density is if you actually superimpose the number of fractures in the population as shown in the bars here, you see that the numbers of fractures in the population tend to cluster in the area of low bone mass or osteopenia because there just are not that many individuals with extremely low bone density, so they contribute a relatively small number of fractures to the overall population. So if one is wedded to a T score of minus 2.5 for treatment, one winds up missing the majority of individuals who will actually experience a major osteoporotic fracture because they actually happen to have T scores above that osteoporosis cut off. The focus then moved in 2010 with the Osteoporosis Canada Clinical Practice Guidelines to getting beyond a T score approach to therapy until a more holistic look at fracture risk assessment. I encourage readers to consult that review, which was evidence-based and was many, many individuals contributed to that particular document. Fundamental to this is the idea of 10-year fracture risk assessment. This has been part of the Osteoporosis Canada approach since 2005 that was formalized and integrated into the management model in the 2010 guidelines. And fundamental to this are two closely related fracture risk assessment systems. One is the FRAC system, which I will be speaking about in great detail. And the other is the Canadian Association of Radiologists, Osteoporosis Canada System, or CAROX for short. And they are actually built from the same building blocks and are very closely aligned as a result. Under either of these frameworks, we look at 10-year fracture risk, and that's categorized as low risk, less than 10% 10-year fracture risk defined as clinical spine, hip, forearm, or humerus fractures. And these are individuals who are generally be considered unlikely to benefit from pharmacotherapy, but where we would continue to monitor their fracture risk. At the other end of the spectrum are those individuals considered to be at high risk, and that means a 10-year fracture is greater than 20%. Or they could also be defined as high risk based upon clinical features, specifically fractures involving the hip or spine that are low trauma fractures, or more than one low trauma fracture episode. That also defines a group of individuals who would be considered high risk and where there is our best evidence to date on the benefits of pharmacotherapy to prevent future fractures. And in between, in the range of 10-20%, a moderate risk group, where guidelines are still evolving to provide guidance to managing physicians, but where one of the focuses is on vertebral imaging, because if we can identify a previously unrecognized vertebral fracture, either on x-ray or vertebral fracture assessment, off of DEXA, if you happen to have that available, then of course a vertebral fracture automatically makes you high risk, and so that would be a population that would be recommended for therapy. So let's look through those risk assessment systems and then go into the advances that have occurred since 2010. So the Canadian Frax System is hosted through the University of Sheffield website. You can see that there are many tools from different parts of the world. You select a continent, you select a country, and the Canadian Frax Calculator joined this family of over 60 calculators in 2010, just in time for the launch of the Canadian guidelines. If you were to go to that website and have a look at what the calculator entails, you would see that, well, I'll look at that in one second, but this slide shows the importance of selecting the correct calculator. As I said, there are over 60 calculators, and with identical inputs, changing the country that one is looking at can have a dramatic effect on the calculated fracture risk. This looks at Canada, which happens to be in the relatively high risk countries for osteoporotic fractures, and so that a man or woman with a T score of minus 2.5, a prior fracture at age 65, would be, if they were a woman, close to the 20% high risk for fracture, a little less than that if it was a man. You can see that getting the right calculator is critical, so any person managing patients in Canada should use the Canadian calculator, which has been specifically built to reflect fracture epidemiology in Canada. The development of these systems is a two-stage process. One of them is to reflect the fracture and mortality in the population, and we had access to cross-Canada hip fracture data and mortality statistics for the purpose of calibrating the Canadian Frax Tool. And then there are relative risk components to this. These are the clinical risk factors that I'll show you on the next slide that are known to have similar effects in different populations. So if you're a smoker, it increases your risk of fracture by a certain amount, whether you're in Canada, the United States, or Asia. And so these are the actual inputs into the calculator. This is the Canadian calculator, but they all have the same look. One enters age, sex, height, and weight for derivation of body mass index, and a few key clinical risk factors which are entered as yes, no, previous fracture, parental hip fracture, large smoking, prolonged corticosterouse, rheumatoid arthritis. There's a list of secondary osteoporosis causes, high alcohol use, and if you have it available to you, and hopefully you do, BMD measured at the femoral neck, either as the BMD itself or as a T-score. And then entering that information, pressing Calculate, will lead to calculation of the 10-year risk of major osteoporotic fractures, which is critical in the Canadian guidelines. Hip fracture risk is also calculated, but that does not have a specific role in the osteoporosis Canada guidelines. This fractious calculator for Canada was developed and validated in many, using two large cohorts. One was the Canadian Multicenter Osteoporosis Study, and the other was the Manitoba BMD cohort, which together comprised almost 45,000 individuals. And it was possible to take the FRAX calculations and to look at the performance of that calculator to see whether the predicted risk of fracture actually agreed with the observed risk of fracture in samples that were independent from the national fracture data used to develop the model. And indeed, you can see that the dotted line of identity, which shows perfect agreement between prediction and observation, shows a very good agreement for men and women across a range of risk categories. So I'm pleased to say that the Canadian FRAX calculator has been well validated in the Canadian population. Mostly aligned to the FRAX calculator is the tarot system that I referred to. This is a simplified version, which still emphasizes the same three risk categories, low risk, moderate risk, and high risk. And that is based upon age, sex, and the hip bone density measurement. So very much like the FRAX calculator. But it also considers two clinical risk factors, so it's simplified from FRAX. The two clinical risk factors are fragility fracture after age 40, or in the last year, prolonged corticosteroid use of 7.5 milligram prednisone equivalent or greater. These are to show how the system works then to actually describe it. And this is an example for a 65-year-old woman. Let's assume that her femoral neck T score is minus 2.8, and she has no other clinical risk factors. Then based upon the lookup table that accompanies this, the CAROCK tool, which also has a tabular version, you would see that her fracture was good plot in the moderate risk category. So again, this is a categorical or semi-quantitative method of risk assessment. If she actually happened to have either of those clinical risk factors that we referred to prior fragility fracture or corticosteroid, that bumps her risk category up one for each clinical risk factor. So she was at risk to start with, and the presence of either of these would shift her up to the high risk category. And so this generally gives the kind of clinical guidance that is necessary to use the integrated management framework from the osteoporosis candidate 20 time guidelines. So for those people that don't want to carry around a lookup table or a graph, there's an app for that. You can go to the Osteoporosis candidate website, and there's a downloadable application that gives you exactly the same results and doesn't require you to remember any of these details. The 2010 guidelines were an update to the CAROCK system that was originally launched in 2005. Canada was the first country to embrace absolute fractures based upon 10-year risk prediction for risk assessment. That was a precursor to the current system. Again, it had the same layout, the same clinical risk factors. We did not have accurate Canadian fracture data at the time for calibrating the system, so we had to be reliant on somewhat old outdated Swedish fracture data. And we know that fractures have actually been going down globally. So in conjunction with some other technical factors, this tended to overestimate risk assessment in the Canadian population. It also was not directly validated in the Canadian population. In contrast, the 2010 system uses the same layout, the same graphical display that people have come to know and love, same clinical risk factors, but because it's now derived from Canadian data and we have also made some additional technical corrections to the risk calculator and been able to do the validation work, we know that this actually does a very good job at assessing risk fracture in the Canadian population. So if we go back to a sample here, 65-year-old woman, T score minus 2.8. Again, this is the same one that I showed you before. She was moderate risk under the CAROCK system, under the FRACS calculator. Her 10-year risk is 13 percent, again between 10 and 20 percent, so in the moderate risk range. If she had an additional risk factor with the FRACS calculator, a prior fracture of steroids would push up to 21 percent, which would have her in the high risk category. And again, as I said, either of those risk factors under the CAROCK system would also make her high risk. So we have a good agreement between the FRACS and the CAROCK calculators for this particular individual. This is some of the validation work that was published around the FRACS versus the CAROCK system, showing that for low, moderate, or high, that the observed risk of fracture, 10-year risk of fracture on the Y-axis, in fact is less than 10 percent, and 20 percent or greater than 20 percent with both the FRACS and with the CAROCK system. So again, the kind of stratification, the kind of calibration that you would like to see. The nice thing about these two systems is that they were designed to give the same risk category in the great majority of individuals. 90 percent of individuals actually have the same risk categorization under CAROCK and FRACS. So the question that arises, can we do better than these risk calculators? What's new on the horizon that might appear in future guidelines? Well, part of some of this was addressed at a joint position development conference between the International Society for Clinical Antisatometry and the International Osteoporosis Foundation that was held in Bucharest, Romania in 2010 and was published by those two organizations in 2011. I'm not going to take you through the recommendations, but for those of you that are interested in seeing this, they are published and there are 28 specific recommendations on how to use the FRACS calculator in clinical practice. Examples of enhancements to the original FRACS system is stratification by corticosteroid dose. So the original risk calculator on the FRACS website says corticosteroid use, yes, no, we know that there's going to be a dose response effect with higher doses giving higher risk and lower doses giving lower risk. And so John Canis was able to do some additional analysis and showed that you could, in fact, modify the FRACS output. So after you have the result, to reflect the difference between a low corticosteroid dose versus a high corticosteroid dose, so that if you were a low dose user, you would reduce the FRACS calculated risk by 20%. High dose user, you would increase it by about 15% for major fractures and if you're in the intermediate dose range, of course, you would just use the FRACS output directly with a similar adjustment applied for hip fractures. One of the conundrums is that the only bone density input to FRACS is the T-score from the femoral neck and we know that in clinical practice the lumbar spine is often used. So what you do when the lumbar spine BMD is significantly different, specifically lower than the femoral neck T-score. And so a methodology for adjusting the FRACS score based upon the number of standard deviations difference between the T-score at the femoral neck and lumbar spine was developed. We actually developed that in Manitoba, but it's been subsequently validated in the Canadian population and in international meta-analysis and was adopted by ISCD and IOF as a procedure that could be used to accommodate the lumbar spine in the risk assessment framework under FRACS. And so how big of a difference does it make? Well, this is an example of that. 70-year-old asthmatic woman with a T-score of minus 2.0 with a spine T-score of minus 3.5 on relatively high-dose glucocorticoid. Using the FRACS calculator, which does not consider the lumbar spine or the steroid dose, her risk comes out at 16%, which would be in the moderate risk range. But if you incorporate the adjustment based upon the spine T-score as well as the higher steroid dose, in fact, her risk gets recalculated as 21%, so she would be designated high risk. And so this is an example of how these additional modifiers can be considered in FRACS. Another advance that occurred is in the assessment of risk in individuals on treatment because there is some concern that in individuals who are receiving treatment for osteoporosis that maybe these risk calculators cannot be used. We did an analysis in a large number of women, some of whom were on therapy. We were able to link this through pharmacy records. And this graph shows the calibration in the untreated individuals in the gray area versus individuals on treatment, looking at different levels of adherence and on in the lines. And you can see they all fall within the area of 95% confidence for the calibration of FRACS. And in fact, in our detailed analysis, we could not show any adverse effective treatment on the FRACS predictions in our analysis. So it's suggesting that maybe there is a possibility of using the FRACS calculator to predict risk in individuals on therapy. Now, does that mean that we're predicting the risk reduction on therapy? And the answer is absolutely not. We have to distinguish from that. We did a follow-up study where we looked at individuals on therapy to see whether the change in their risk on treatment was useful with the FRACS calculator. We looked at over 11,000 individuals who had two FRACS risk calculations. Again, they were able to link their pharmacy records, so we knew who was initiating therapy and what their adherence was. And unfortunately, you can see that in the figure here, those people that did not initiate therapy, their risk of a fracture went up over time. This is the change in their risk assessment to actually increase. People that started therapy and took a little bit of drug, but less than 50% of the time, their risk still went up. Those people that took 50% to 80% of their medication, their risk went up. And even those people that took the highest level of medication over 80% compliant, their risk still went up. And if you think about it for a moment, you realize that, of course, age is a component of the FRACS calculation. Even if your bone density stabilizes or improves, and that should stabilize or improve your risk assessment, aging does not stop. And so your risk of fractional contingency increase may be a little slower on therapy, but it means that the FRACS risk calculator is not going to be a good indicator of on-treatment risk reduction. Well, we've been very dexa-focused on to this point, but we know that dexa measures bone density, but there's an awful lot of other stuff that goes on in the skeleton that contributes to a little strength that dexa does not directly measure or has not been used in clinical practice. And so the International Society for Clinical Densitometry earlier this year held a position development conference where we looked at the use of dexa and CT to assess fracture risk, but looking beyond bone density measurements, can we use other metrics that are derived from dexa and CT to assess fracture risk that would be complementary to the BMDT score? And so these are the three primary topics addressed. We looked at trabecular bone score, which I'll talk about in some detail, a measure of bone texture, geometric measures derived from the hip, typically hip structure analysis and hip axis length. And there were also detailed discussions on three-dimensional quantitative CT, finite element analysis and opportunistic screening based upon CT scans, which I will not be touching on today. So let's talk to trabecular bone score. This has been very topical in the last few years. TBS is basically a texture measurement that can be derived for lumbar spine dexa images. It's specialized software. It's approved by the FDA. You can get a license for it and allows you to analyze the region of interest in the lumbar spine that's used for the BMD measurement. And then it comes out with a measure of the bone texture, which is a reflection of bone structure that correlates with the bone strength. A higher TBS measurement implying better preserved bone structure and a low TBS measurement implying worse or in the parlance of TBS degraded bone structure. And the principles are very simplified here, but a well-structured bone has lots of variation pixel to pixel to pixel, and that's with TBS measurements. Whereas in a osteoporotic bone, there's loss in the variation at small distances. So you lose that complex structure and texture, and that results in a lower TBS measurement. The correlation between BMD and TBS is actually quite poor. So you can see individuals like in this example that have identical bone density measurements, but where the TBS measurements can be quite different. So it has the potential to provide complementary information. So the questions addressed by the ISD panel was the use of TBS in clinical practice for initiation of treatment for monitoring purposes and whether there were special conditions where TBS might have greater value. This is a summary slide that looks at the longitudinal studies that have assessed prediction of osteoporotic fractures in post-menopausal women. The studies varied from the largest study from Manitoba with 29,000 women to a study from Japan that had 600 women or a study from France that had just over 500. But the important point is not the specific results from the assessments, but the fact that all of them, after adjustment for risk factors, showed a significant difference from the null effect or has a ratio of one in terms of predicting an increased risk in both vertebral fractures, hip fractures, any or major osteoporotic fractures. So that there was predictive information in the TBS measurement in terms of any of those types of fractures. Recently, in fact, just within the last few weeks, a large meta-analysis appeared using 14 prospective studies. You can see now a large number of men, a large number of women in this cohort, contributing a large number of hip fractures and major osteoporotic fractures. And the objective here was to develop something that would be complementary to the use of frax. So the frax was one of the adjusting covariates. And the important thing is that for major osteoporotic fractures and hip fractures in both men and women with equal strength, each standard deviation reduction in TBS led to an increase in fracture probability between 27 and 35%, which may not sound like a lot, but it's very comparable to the other clinical risk factors in frax. And this is now information that is complementary or independent of the frax calculator. The question of whether TBS might be useful as a monitoring tool has been addressed. Unfortunately, TBS seems to be very unresponsive to currently available treatments, especially the bisphosphonates. In gray is the change in lumbar spine BMD that you see with the bisphosphonate or terraparitide in one of these slides, bars versus in blue, the change in TBS. And you can see that across the board, the change on therapy in TBS is considerably smaller than the change in BMD, suggesting that TBS may actually not be a useful tool for monitoring purposes. Diabetes has been of great interest because we know that diabetes is not a risk factor in the frax tool and data from several studies, including the study from Laura Jean Gregorio, have shown that the frax calculator, although well-calibrated, as seen by the alignment between the dotted line and the calibration curve in non-diabetic individuals, underestimates fracture risk in those individuals who are diabetic, as seen in the blue line, and that applies to both major osteoporotic fractures and hip fractures. So is TBS useful in diabetes? And the answer is, it just may be. And if you look at the BMD measurements, we know that BMD tends to be higher in individuals with type 2 diabetes, and so this large study from Anatobe found indeed that all of the BMD measurement sites we looked at were significantly higher in the diabetic versus the non-diabetic individuals, even when you adjusted for BMI and other risk factors. Interestingly, lumbar spine TBS was significantly lower in those people that had diabetes, and because diabetes acts as a risk factor for fracture, this suggests we may be able to capture some of that information. And so when we looked at the prediction of fracture with TBS, adjusted for other covariates, we showed that lumbar spine TBS, in fact, did predict fractures in individuals with diabetes with a hazard ratio of 1.27, virtually identical to the individuals without diabetes with a hazard ratio of 1.31, and that about half of the information attributed to diabetes as a risk factor in the model was captured by TBS. So it captures some, although not all, of the diabetes-associated risk. So the recommendations from the ISCB was that TBS was associated with the fracture risk in older women and in men should not be used along to determine treatment which was not useful for monitoring. What had value in individuals with type 2 diabetes, but the most important thing was the final point that TBS can be used in association with frax to adjust fracture probability in older women and men. And so this is the study that supported that statement, that when you look in red at the association between TBS, adjusted for frax-concurrent risk factors in bone density, that it was a significant contributor for hip fractures, for other osteoporotic fractures, and for mortality, which, of course, as you will recall, is part of the frax framework because competing mortality does protect against future risk of death. And when the folks in Sheffield put all that together, they came up with the TBS adjustment to frax that's published in the McCloskey paper and that is reproduced here so that you can program this on to Excel if you wanted. But you can see that this has a specific term in here for TBS by age. The TBS by age term is very important because it captures the fact that TBS has a greater importance in younger than older individuals. So if you look at the magnitude of the adjustment that would occur to an individual whose baseline risk of fracture was, say, 12.5 percent major osteoporotic fracture, 2.5 percent for hip fracture, if their TBS measurement is at the 10th percentile that increases the risk for fracture, if it's at the 90th percentile that reduces the risk for fracture, but that adjustment is much stronger in individuals who are younger than individuals who are older and that effect is captured by that age by TBS interaction term. So if you've been to the frax website recently, you may have noticed that there is a new button that's appeared called Adjust with TBS. So if you have a TBS measurement, you can click that button and it will take you to a page. You enter in the TBS measurement and now you can get a frax adjusted or a TBS adjusted frax probability that does all of these calculations for you. The next thing that the EICD panel looked at was bone geometry. This has been available for over 20 years and goes by the term structural analysis or advanced hip analysis on different genders and it looks at things like dimensions around the femoral neck, the hip axis length shown in green, the neck shaft angle is seen in magenta. The parameters have been around for a long time but we haven't actually known if they were useful for risk prediction. So the panel addressed those questions. Can these geometry measures predict fractures? Can they be used for treatment purposes? Can they be used for monitoring purposes? There was a positive data in men and limited data in women. This is a summary of some of the adjusted odds ratios which went from very weak effects to very strong effects and it was really all over the map. Probably because many of the studies were underpowered. So to inform this discussion we did a large scale analysis in the manitoba cohort where we identified 50,000 women, 10,000 men incident hip fractures and we were able to simultaneously look at the risk of all of these geometric parameters adjusted for the frax probabilities and showed that in fact all of them as seen in bold phase here were predictive of hip fractures. If you adjusted for BMD in the model some of them no longer were important because BMD was such a strong contributor to those risk measurements like the cross-sectional area and cross-sectional moment of inertia. Basically it's all captured by the BMD measurement. But interestingly, three of them continued to be quite interesting. Hip axis length, neck check, dangle, buckling ratio and when we evaluated them simultaneously we found of those the most robustly associated with hip fracture risk was the hip axis length. So hip axis length is interesting because it's so easily measured off of a DEXA image and this looks at the degree of stratification that we saw from the shortest hip axis length, smallest quintile to the longest hip axis length the longest quintile showing really a more than two-fold variation in hip fracture risk across the physiologic range of hip axis length. And so the ISCD positions basically highlighted the importance of hip axis length as a parameter that can predict risk in women that the other geometric parameters were actually not useful for risk assessment. None of these are going to be treatment responsive because of course it's geometry and so you can't initiate treatment on them solely or use them for monitoring purposes. Now that was then and this is now. More data just published where we were able to actually look at men and we showed that in black compared to orange that the hip axis length actually gives you the same risk for hip fracture in men as it does in women which is nice because now we have something to use in both genders. And we were able to develop a methodology that would allow you to integrate that into the frax risk assessment based upon how far an individual's hip axis length differs from average for their sex and refer you to the publication in the interest of time but to show you that for every millimeter increase or decrease in hip axis length then you can apply a relatively simple adjustment to the frax probability, 4% to 5% per millimeter and that can actually have quite a large effect on the hip fracture risk in an individual between the 10th percentile and the 90th percentile almost doubling the hip fracture risk. I won't touch on finite element analysis but that is probably the future of risk assessment in bone density geometry. This is historically been an engineering technique that we used to build bridges but of course the analogy with bone is fairly obvious. This has been widely used in quantitative CT but there is now interest in developing this technology for DEXA so we may be seeing something useful in that line in the next year or two. The group in Sheffield has been very forward in that work. I'll close with some crystal ball gazing about what the next-gen risk assessment might look like and where the CLSA may be contributory. We have a huge dataset of bone density measurements including hip DEXA from CLSA and so this may be helpful to address questions like frax versus other calculators that have been developed. One of those from Australia, the Garvin calculator, incorporates falls and numbers of fractures and many people think that that is very attractive but it hasn't been compared with frax in a large cohort and CLSA captures all of the essential ingredients needed for both the frax calculation and the garvin calculation so a head-to-head comparison to those algorithms or whatever else appears on the horizon would be very attractive. Body composition, sarcopenia clearly contribute to frail teeth, falls and fracture and so again CLSA is the world's largest cohort of total body DEXA measurements which allows you to derive body composition and appendicular lean mass which is part of the definition for sarcopenia and so again a tremendous opportunity for CLSA. Vertuber fracture assessment to look at a cult for tuber fractures has been part of CLSA and again this will be the largest dataset of that methodology in the world. In the next wave, let's hope that the spine DEXA will be incorporated in the CLSA measurements and so this creates opportunities both to look at the spine hip discordance adjustments that I spoke about and also to look at the spine TBS because as you can see that really is a very exciting new development in the field of DEXA. So with that I'll close with one of my favorite quotes the stone age was marked by men's clever use of crude tools. The information age to date has been marked by men's crude use of clever tools. There are a lot of clever tools out there we're hoping to be able to make better use of them in clinical management and I think that hopefully I've convinced you that that is the direction that osteoporosis Canada has been charting over the last decade. Thank you very much. Thank you very much Bill for a very excellent and informative presentation. Not being from the clinical world myself I always find these types of presentations very interesting because these days research is so multi-disciplinary so non-clinical people like me really have to understand the clinical perspective. So we have about 15 or so minutes and we will entertain questions from the audience as I said at the outset. Please type your questions into the chat feature and I will then read them out to the rest of us and Bill will be able to answer. To get the ball rolling I have a couple of questions and first question coming from an epidemiologic perspective we like to think about external and internal validity and when I think of external validity I think of the extent to which the results from CLSA could be generalized to other populations so to Americans or people in the United Kingdom and so I guess Bill thinking back to what you concluded with with your crystal ball do you think that the CLSA's results could easily apply to populations in other countries outside of Canada? So I'll speak based on my experience with the Frax and Karak tools and there's no question that there are population differences in the epidemiology of fractures we see that dramatically in the developed versus the developing world but at the same time I see many commonalities and similarities and especially in North America the 49th parallel really is just a line on a map it does not reflect any intrinsic differences between the populations so when we've looked at the performance of the Canadian Frax calculator we actually also looked at the US Frax calculator in the Canadian population it was virtually identical in terms of its calibration and performance those calculators are around for hip fracture risk which is built upon very solid epidemiology showed virtually identical results for the Australia, New Zealand, France and UK calculators so I think that there are certainly pretty good argument that there are more similarities than differences in these historically European derived populations the evidence for other similarities to populations are maybe more relative similarities than absolute similarities so the meta-analyses that have been done with Frax for the clinical risk factors and these clinical measurements have not to date shown that there is a substantial difference in the relative risks in the North American, European or Asian populations so there were actually Asian populations in the TBS meta-analysis and they did not look any different in that population than in the Canadian population so I think that the population differences although they clearly exist are not to preclude generalization from CLSA to other countries great thanks, one other question and then we have a question in the chat future that I'll read out what are the, again thinking epidemiologically about longitudinal studies I'm not sure if you're familiar with KMOS if you are, what do you think are the major differences in how CLSA can inform osteoporosis research versus KMOS? I have been involved with KMOS, the Canadian Multicenter Osteoporosis Study since the mid-1990s so I think KMOS is clearly an enormous contribution to our understanding of osteoporosis and fracture epidemiology in Canada but it's also been an international contributor in the TBS meta-analysis that I referred to, KMOS contributed to that and was the largest single cohort contributing to that analysis the advantage to KMOS is that it's very specific to the areas of osteoporosis and fracture it's got an enormous lead time now, we're 17 years of follow-up data so it's got, although not as large as CLSA, it does have a head start and it's very granular in terms of the detail around osteoporosis risk factors whereas CLSA is a much broader framework for health assessment what I think CLSA offers tremendous opportunity in the osteoporosis field is the body composition again, sarcopenia being a very hot topic in the field of fracture risk assessment right now and also vertebral fracture assessment, which was not systematically part of KMOS so I think those are both the size of CLSA as well as those unique contributions to these little measurements will complement KMOS. A question from the audience, Dean Wright, thank you for presenting on this topic Bill when looking at hip structure measurements such as CSMI were such measures derived using a mass or density weighted computation? Well, Dean, thank you for such a specific detailed question and I can't remember exactly how that was calculated in our analysis we use the GE Lunar Advanced Hip Analysis AHA routine when we get off the call I'll send you Ken Faulkner's paper where they describe that and hopefully you'll find the answer there. So I think that Dean, it says that he's typing up another question entering a chat message, I'm not seeing it here so I think that he's typing something else so give him a minute to type if there are any other questions from the audience then by all means go ahead one question that I can ask you in CLSA we're not sending our participants to get in-depth physical exams by specialist physicians so no quote unquote bone doctor is going to look at any of our participants to assess the presence of osteoarthritis or osteoporosis or anything like that instead we're using algorithms we're taking bits and bobs of CLSA data and combining them together to come out with a ruling in or a ruling out as to whether or not someone might have a disease so are you familiar with any literature on the extent to which self-report versus clinical diagnosis the extent to which they agree or disagree? I'm not and it'll be interesting actually to see how those algorithms work I was part of the group providing some input into that but it was more intuitive than evidence-based there's no question so I think that would be actually an early analysis that would be quite interesting to look at other studies that have asked people their recollection on for example bone density reports suggest that they may not do a great job that their recollection for osteoporotic bone density is better but if they say normal osteoporosis generally they don't do as good a job okay and certainly that's what I my understanding is that even for very serious chronic disorders people tend to misclassify themselves another question from Dean how important do you think accounting for differences in B.M.D. and hence material strength are in deriving Okka measures that can be used to augment instruments such as Frax? okay just read that again how important Okka let's just look at this differences in B.M.D. in material strength so I'm not sure if this is the intent of the question that B.M.D. as a crude measure of skeletal strength not material properties but skeletal strength clearly is a very strong risk factor for fracture but a lot of the clinical risk factors in the Frax tool are also very strong including age prior fracture and interestingly parental hip fracture independent of bone density is a very strong contributor so we have actually done a comparison of the Frax calculator when used without B.M.D. versus when it is used with B.M.D. to see how important B.M.D. is as a contributor to that risk assessment based upon the 20% cut off for defining high risk we showed that the area under the curve using Frax without B.M.D. to predict 20% risk or greater for Frax used with B.M.D. the AUC was greater than 0.9 the same thing for using the HIP Frax calculator to predict a 3% risk so that would apply that B.M.D. is important but for the majority of the population you can actually do a fairly good job in risk assessment using clinical risk factors without B.M.D. or it makes a big difference for treatment purposes is when the B.M.D. is close to the or your risk is close to the treatment threshold so those individuals that if you are going to treat it 20% but you're in the 15 to 25% range B.M.D. makes a big difference if you're outside of that range much less of an effect. I'm sorry. No, I was just going to say so I answered a question whether that was a question or not I have no idea if there's any further queries. Oh, he says thanks so maybe you did answer it. Well, great I think we're coming up to the end of our time and I don't see any other questions being entered. Thank you Bill very much for this excellent presentation. Well, I appreciate taking the time to present this material to us. Very interesting once again and again thank you very much. Great, so we do try and have webinars every month except for July and August when people tend to be on vacation so just a couple of plugs for the next couple of webinars. They're both going to be international. In December, December the 9th from 12 to one eastern time Dr. Paul Loprinzi from the United States is going to talk about factors that influence physical activity and the effects of physical activity on cardiovascular risk factors and health outcomes in middle to old age adults. And that actually segues nicely into another international based webinar. It's going to be on January 21st at noon eastern time Diana Kuh from the United Kingdom is going to be talking about the life course perspective in epidemiology and she is one of the seminal thinkers in the area of studying epidemiology from a life course. So that will be very interesting these next two webinars and so we hope to see you there and thanks again Bill and I wish everyone a nice afternoon.