An Exercise in Geometry

“To get back my youth I would do anything in the world, except take exercise… . ” Oscar Wilde The Picture of Dorian Gray, Chapter 19 IN THIS ISSUE of the Journal, Petit et al.1 report the structural changes accompanying exercise during growth.1 During 7 months, 43 pre- and 43 early-pubertal girls, mean age 10-10.5 years, were exercised for 10 minutes three times weekly using a jumping program of gradually increasing intensity. Changes in these two intervention groups were compared with changes in 25 pre- and 63 early-pubertal controls matched by age, sitting height, leg length, and lean and fat mass from a different school. Structural changes occurred in all groups. However, in the prepubertal group (Tanner stage 1), the changes in the intervention group were no greater than in controls. In the early pubertal group (Tanner stages 2 and 3), the changes in femoral neck bone cross-sectional area (CSA), cortical thickness, and section modulus (a measure of bending and torsional strength) in the intervention group were greater than in controls. The study raises many issues regarding the rationale, hypothesis, design, execution, analysis, and interpretation of clinical trials of exercise during growth. About 30-50% of women and 15-20% of men will suffer the consequences of fractures related to osteoporosis.2 Drugs reduce fracture risk by about 40-50% in the subgroup of women at highest risk, that is, women with osteoporosis.3 However, at least 50-75% of all fractures in the community come from the larger segment of the population at mild or moderate risk due to modest deficits in areal bone mineral density (aBMD) or younger age.4 This distribution of event rates is well recognized in the cardiovascular field but underemphasized and poorly appreciated in the field of bone metabolism because of our obsession with fracture “thresholds.” Although persons with aBMD reduced by more than 2.5 SD below the young normal mean are at highest risk, this represents only a small proportion of the whole population, most of whom have aBMD above −2.5 SD. Thus, the population burden of fractures and events like stroke or myocardial infarction come from the larger population, not from the tail end of the population distribution containing the high risk group but smaller numbers of individuals.5 The public health burden of fractures cannot be solved with drugs, because drug trials have been done largely in women at highest fracture risk due to osteoporosis—those women with aBMD in the tail end of the population distribution (aBMD < −2.5 SD below the young normal mean). The risk of events such as fractures, strokes, and myocardial infarction is low in the larger segment of the population; to prevent one morbid event, large numbers need to be exposed to the costs of medical care, inconveniences, and side effects of drug therapy at no benefit whatsoever. The cost of treatment using this approach is greater than the cost of fractures.6 We can reduce fracture risk in the subgroup of high risk individuals (case-finding), but they contribute only 25-50% of all fractures in the population. The solution to the public health burden of fractures remains an enigma, particularly as more and more of us are living longer to enjoy the fruits of old age. Thus, nondrug related interventions are needed to reduce the population burden of fractures. For these population-based approaches to be successful, several criteria must be fulfilled. The interventions must be efficacious, safe, accessible to all, easily carried out, and inexpensive to implement. If it were possible to move the population distribution of the structural determinants of bone strength just a few percentage points using an intervention applied to the whole population, there is likely to be a profound benefit on the population burden of fractures.5 Of all the modifiable lifestyle factors that influence the skeleton, such as nutrition, tobacco use, and exercise, it is exercise during growth that has the potential to fulfill all these criteria and therefore could reduce the public health burden of fractures. The answer to this question is unknown. Inferences regarding the anti-fracture efficacy of exercise will never be based on the highest level of experimental evidence. There will obviously never be a double-blind trial of exercise. There will almost certainly never be a randomized open trial of exercise during growth or aging with fragility fractures in old age as an endpoint; the sample sizes needed to show a biologically worthwhile fracture risk reduction are similar to those needed in industry funded drug trials. Issues of design, compliance and funding almost certainly make these trials too difficult to execute successfully. The firm belief held by many that exercise reduces fracture risk is derived from lower levels of evidence—retrospective and prospective observational cohort and case-control studies that suggest active persons have fewer fractures than less active persons.7-9 This inference may be correct, but these studies are subject to systematic healthy user bias and should be evaluated with skepticism and interpreted cautiously.7 Healthier individuals may choose to be more active and are less prone to falls and fractures. Persons inheriting a larger musculo-skeletal size have higher bone mass and lower fracture risk before starting exercise. The exercise may be a consequence of the larger musculo-skeletal mass rather than the larger musculo-skeletal mass being a consequence of the exercise. There is no strength in numbers. Meta-analyses of observational studies examining efficacy should not even be under taken, much less interpreted. At very best, they are hypothesis generating, not hypothesis testing. Recall the iconoclastic impact of the first randomized double-blind trial examining the effect of hormone replacement therapy (HRT) on cardiac events, and the more recently reported negative results concerning cerebrovascular morbidity and mortality using HRT.10, 11 One or two properly conducted studies shatter the dogma created by meta-analysis after meta-analysis of observational studies that claim HRT reduces cardiovascular events. Thus, we are forced to make inferences within the constraints of the uncertainty imposed by even lower levels of evidence—the effect of exercise on surrogates of anti-fracture efficacy such as changes in bone mass, structure and derived measures of bone strength. These endpoints also have limitations that should be acknowledged. For example, a change in aBMD in response to drug therapy is a poor surrogate of the reduction in fracture risk.12 What then are the effects of exercise on the surrogates of anti-fracture efficacy? There is little replicated and methodologically sound evidence to suggest that exercise during young adulthood, peri-menopause, late adulthood, or old age modifies bone size, prevents bone loss, or restores bone mass, architecture, or strength. Consistency in results is lacking; some studies suggest bone loss is prevented by exercise and others suggest bone loss is not prevented or is increased.7-9, 13 The increase in aBMD of a few percentage points reported in some studies is probably due to a reduction in the size of the reversible remodeling space.14 There is little, if any, evidence of changes in bone tissue mass beyond that produced by reducing the remodeling space. There is no evidence that exercise in adults increases cortical thickness by increasing periosteal apposition, reducing endocortical resorption, or increasing endocortical bone formation. Whether exercise reduces intracortical porosity, increases trabecular thickness, connectivity or the mineral content of the matrix tissue mass (regrettably called “true” density), is not known. When exercise is stopped, bone loss accelerates,15 probably due to an increase in the size of the reversible remodeling space. Exercise during adulthood may reduce the risk and severity of falls, but evidence that this translates into fewer fractures is lacking.7 It is during growth that exercise produces its most beneficial effects. Growth is the single most opportune time to modify the mass and geometry of the skeleton.16-24 The studies of vigorous exercise during growth provide evidence that large increments in aBMD can be achieved in loaded compared with unloaded regions of the skeleton. High impact loading in gymnastics, weight lifting, and racket sports is associated with high regional aBMD. Differences of 1-3 SD are often reported in the playing arm compared with nonplaying arm in racket sports, increments that are an order of magnitude higher than produced by exercise in adults. These changes are the result of distinct structural changes in bone mass, size, and geometry, not just small changes in the remodeling space. The region specific in playing nonplaying arm cannot be the result of bias or factors that this one of the few in the bone field that is not loading exercise during growth is for of this is The study bone mass and structure of the playing and nonplaying of racket sports about The results have never been on and have been replicated several The to loading during growth by its size, architecture, and These changes are likely to biologically worthwhile in the and torsional strength of bone that may well be the solution to the of fractures in old age, the changes can be during the or of There are issues in study that need to be before the greater changes reported in an exercise compared with a group can be to be the result of the group in exercise than being due to group in many Petit et al.1 an intervention group of prepubertal and early pubertal matched with a group by age, pubertal sitting height, leg length, and lean and fat The are to be for this to These may influence the of the exercise. small in in an exercise and even 2 to in age or pubertal can large If controls are more in and are about to move into an of then benefit of exercise in the less will be by the growth in controls. The will that exercise no If the exercise group is more then the changes in the exercise group will be greater than the changes in the controls due to the and benefit exercise may This may have occurred in a study of exercise reported by et mineral content and bone area to more in the exercise than large in growth by a of in and exercise, than in exercise in the are not to a single or group also in and of the age by If they are matched by they will by age. The of may to About of the in the study by Petit et al.1 were Although the were not to in an study of and and these reported that the increase in and aBMD greater in the than and greater in the than In study by these exercise levels were lower in than Whether the or specific aBMD to exercise and specific in in exercise may influence the response to or and specific to loading is not The in in and of different are not solved by that are of is by small sample sizes of and different and in age, height, and weight in the exercise and groups. The of stage is by a in the of the study by Petit et al.1 age of all groups no about 10-10.5 years, there about SD in sitting height, leg length, and lean and fat mass in the prepubertal cohort than early pubertal The is an exercise group with controls only by age in studies of growth is The above some issues in studies an exercise group with a group no exercise. There are greater the question of exercise is more in prepubertal than early-pubertal more in than in or in one group compared with These evidence that there is loading in these groups. Exercise in pre- early-pubertal may less loading in the prepubertal if loading is and is in two in the of of the and pre- and and groups in study that cannot be easily In most sample sizes will be too small for and are likely to be 2 The or will be to on and will no the of We are in the and effects of Thus, the many that may influence in it may be more to the of these studies by single groups with with bone age rather than and or of different Although there are many cross-sectional studies that have higher bone mass than the needed to this us is not is in for The question is, moderate exercise produces biologically worthwhile changes in bone What is When is the time during growth to What is the structural increase in bone strength produced by exercise during the exercise is Petit et al.1 the effects of moderate exercise, an intervention that can easily be into a exercise This study is of a larger study in the report the results of of by to be the of intervention groups and to be This is a approach because it controls exercise. However, the not the were to exercise or no exercise. to one of the groups have occurred and a For example, a at the randomized to be the of controls but may if a randomized to be the of the intervention more active may for the intervention group but to as a an effect that may result in a the exercise of the sample in an it that there were no in exercise in the intervention and groups. of exercise to some of or cannot be in the that influence growth may also a of individuals rather than is one of that are in groups. is The in produced by may be if or poor compliance in exercise The were a program of jumping 10 minutes three times The reported compliance in the only of the study due to loss of or reported an compliance prepubertal and during 7 In this from were into the carried during three times In study of prepubertal 10 years, et an exercise program of of three times weekly during with all but 2 compliance reported by et events were in all these trials. Thus, moderate exercise is accessible to all, by the of safe, and inexpensive to but it can be from the in the study by Petit et the the increase in most during 7 not and were in all groups femoral in the prepubertal However, the increases in prepubertal intervention group were no greater than in the prepubertal controls. the increases in the early pubertal intervention group in femoral neck bone cortical thickness and section modulus were greater than in early pubertal controls. and similar in aBMD the suggest that the most of to exercise is during early (Tanner stages 2 and 3), rather than during the prepubertal There are several to suggest that this is the have not the increments in bone mass, or section modulus in the early pubertal intervention group are greater than the increments in the prepubertal intervention group and most after for the large in these two different groups. at the the mean changes this is the inference that one of growth is more than cannot be the in the prepubertal group is not by the results of several In the study by et femoral neck bone mass and bone area in prepubertal and of the exercise group more than in the controls. Whether there were specific changes is In a study of prepubertal the increase in aBMD greater than controls and the result of less of the In with the of Petit et et suggest that the not is a to exercise. However, this a cross-sectional study and the group for a longer of a small increase in periosteal to exercise may have been for rather than The numbers of controls in the early pubertal group larger than in the prepubertal group and from stages and 2 occurred in of controls and of the intervention more of the intervention group in stage Thus, in greater numbers of controls than could the benefit of the exercise. In periosteal and endocortical were at one the and bone may result in periosteal in the change in the a change in bone and in the of the by Petit et suggest that periosteal in prepubertal to a the as using et report greater periosteal and endocortical in the playing compared with nonplaying arm in prepubertal some measure of loading such as of a or in the that loading produced by the jumping less in the pre- than early-pubertal cannot be lean mass to weight no less in the prepubertal than in early pubertal Thus, of growth are likely to be opportune of to a skeleton. growth is more than growth in the prepubertal most of the growth is growth then growth then but its growth one of growth is more opportune than is the it possible for a of regional the response of the may be by loading before the response of the be by loading during exercise to in and be carried in the prepubertal at the be and at a different age in and girls, to for the in The greater changes in the intervention than groups were by Petit et al.1 in the early-pubertal The higher aBMD in the early-pubertal intervention may be the result of one or of two greater bone than in controls or less or greater than in controls. a small change in produced by periosteal bone a large change in the cross-sectional because it is to the of the However, the increase in in the early pubertal intervention group less than the increase in the controls not It the increases in femoral neck bone cortical thickness, and section that were greater in the intervention group than controls. The greater increase in femoral cortical bone is a of bone mineral derived as if the tissue were not there is not the of the femoral The increase in and femoral cortical thickness is likely to be more the result of events on the endocortical than periosteal This is reported in exercise et in a prospective study of high impact exercise in reported changes on the not periosteal in the study of et reported that the greater cortical in the playing compared with the nonplaying arm the result of greater endocortical in the and greater periosteal in the prepubertal et report that the greater increase in aBMD in prepubertal in a moderate exercise program at the result of greater endocortical not greater periosteal than in controls. is needed to the effects of exercise on the on the endocortical If exercise reduces remodeling this may reduce the of during growth compared with a and a may the of bone and in by the or of the and This may result in less or a exercise may bone of the remodeling The effects of exercise on of matrix are unknown. In all of these the of bone were at one the bone and in the The of periosteal and endocortical and remodeling the whole of a loading will modify of the geometry of the bone in with the imposed In racket sports, the greater cortical area of the playing nonplaying side is the result of greater periosteal and greater endocortical The of periosteal and endocortical to the greater cortical thickness in the playing than nonplaying arm in a study by et were at the and at and In the study by et the of greater periosteal and greater to the greater cortical thickness were in the and in the in men and Thus, loading the periosteal and endocortical but the magnitude of the effects to the is or and the region is or the These are not with a single and cortical thickness Thus, as by Petit et inferences about or of must be only one and of cortical and whole bone strength were then We need safe, and that measure the size and of the that us Although greater periosteal in the playing arm than nonplaying arm is reported by several periosteal at bone be less in the intervention than group in the above et reported greater increase in bone size in prepubertal and can this be In the studies by Petit et al.1 and et the changes in the periosteal were for changes in This is If the change in in an exercise and group is the in the change in periosteal can be to the exercise. If increases less in an exercise group than group due to reduced growth of the or skeleton, then this may be by periosteal of the region of loading that may be This must be into in the and of the of the bone that is for may if growth is region specific of it In one study by et and girls, increase less in the exercise than sitting and leg not in the study by Petit et and the changes in these reported in the also not in intervention or groups in the prepubertal or early pubertal Although not leg by about SD less in the prepubertal intervention than Exercise may be associated with reduced growth in and and may be to one The is that the of the is on the effects of the exercise as well as possible growth and factors that and growth of the The study by et to changes in the In the aBMD of the more in the early pubertal exercise than of the changes in not probably because of the in the of this This is needed if are to be about the fracture efficacy of exercise. and aBMD increase during growth because bone size the in a the increase in may be the result of an increasing bone size and not an increase in the mass of would the to be if it can to greater well by its will increase during growth if the increase in the of bone within the periosteal is greater than the increase in the size of the The results of changes in were reported in the study by et at the femoral with the that there change on the periosteal of the bone endocortical remodeling of the by bone or less endocortical there more bone the not increase at the In the study by et in the exercise group in controls that the increase in area were no different in the two groups. of the by similar in groups with no change in in et reported that the in aBMD in a playing a nonplaying arm due to in bone size produced by greater periosteal apposition, not due to an increase in These studies all suffer from the we measure bone size and we If all the of exercise achieved during growth are exercise is stopped, This question is The answer to this question or not can be regarding exercise as a public health There is evidence that exercise bone loss Whether this at the endocortical and trabecular periosteal is is not known. In such as are for many years, but these are young The are about lower than the SD in active but this may changes in the may have achieved a lower bone size and mass than the In years, to be or largely rates were no less in the than in but the study may have been These are cross-sectional but as as it is to We will not have prospective studies of young they the to the There is no evidence of a benefit of exercise during growth on bone structure or strength in old age falls and fracture risk The evidence regarding no or is based on studies in not based on studies of moderate exercise. There is no to or the that the modest achieved by moderate exercise during growth are into adulthood are There is only one by et in that the derived from 7 of exercise were 7 after the jumping This study must before the of the results can be is the most achieved during growth will be by a less or less exercise If the are is it to never have exercised at There is we are from such as and aBMD and a greater of The is not an that can be into a single or single like aBMD. is an its and during growth because they are The is not a single The growth of the and are not single that at end like an the segment of the bone more than the The at the time being exposed to the of There is no one cortical in bone size, and cortical thickness is at its and its and these are changes to region of the and skeleton, on the and a and is by and into a structure to The effect of exercise may on the stage of growth exercise is as well as the and of we need of all of these at the I We must get from and aBMD we about the of growth and We have recognized that the effects of loading are region specific and the changes are the result of the and of the periosteal and measure and the that in space. we that moderate exercise during growth a of with a a and or more If do we that this will just a little less and a little less after of a in the of and than would have occurred moderate exercise not been during but I is There are many and of the studies reported by Petit et al.1 and et and these should be to study is in all but the needed are particularly in studies of growth. We need studies that are in a single and single with pubertal bone age. We need that measure bone size and its and bone If the is the studies will be to and the or will be to and difficult to on When can do we will not need clinical trials. we need study rather than that by in small sample sizes to the with the to us by a in a