Osteoporosis in men: findings from the Osteoporotic Fractures in Men Study (MrOS)

Abstract

The lifespan of men is increasing and this is associated with an increased prevalence of osteoporosis in men. Osteoporosis increases the risk of bone fracture. Fractures are associated with increased disability and mortality, and public health problems. We review here the study of osteoporosis in men as obtained from a longitudinal cohort of community-based older men, the Osteoporotic Fractures in Men Study (MrOS).

Introduction

Much is known about osteoporosis in women but osteoporosis in men has been less well studied. This may be because the incidence of the disease is lower in men and because there have not been many well characterized longitudinal cohort studies in community-based men to determine the risk factors for fractures in men. Similar to women, bone loss in men is related to aging, although at a lower rate and magnitude due to lack of equivalent menopause seen in women. Differences in skeletal size, mechanical loading and muscle mass may also play role in the patterns of bone loss in men and women [Seeman, 2002]. Nevertheless, bone loss takes place in both trabecular and cortical compartments with increased cortical porosity with age [Seeman, 2002; Sundh et al. 2015]. These changes cause dramatic increases in fracture in men after the age of 70 [Donaldson et al. 1990].

To determine the epidemiology of osteoporotic fractures in men, the National Institutes of Health (NIH) funded a large community-based cohort study of men in 1999, the Osteoporotic Fractures in Men Study, also referred to as MrOS. After nearly 15 years of follow up, the study has identified significant risk factors for osteoporosis in men. In this review, we describe the overall study design of MrOS, and its major findings regarding osteoporosis and fracture including determinants of bone loss, risk factors for falls and fractures, advanced imaging of bone, strategies for screening for osteoporosis and the genetics of osteoporosis.

Study design

MrOS was initially designed to describe the epidemiology of osteoporosis and fractures in older men [Blank et al. 2005; Orwoll et al. 2005], including the identification of risk factors for fracture and bone loss. MrOS has described diagnostic methods for fracture assessment including identification of vertebral fractures [Ferrar et al. 2007; Cawthon et al. 2014b]. To allow gender comparisons, many of the protocols and procedures used in MrOS are similar to the Study of Osteoporotic Fractures (SOF), a large prospective cohort study in women [Cummings et al. 1995].

The scope of the study has expanded over time and the cohort has been extensively phenotyped in a number of other health conditions, including dental health [Phipps et al. 2009], sleep [Stone et al. 2014], falls and physical performance[Chan et al. 2007], and cardiovascular disease [Mehra et al. 2009]. MrOS has also contributed information regarding the genetics of osteoporosis [Eriksson et al. 2015]. The main US study has also collaborated with colleagues in Sweden [Mellstrom et al. 2006] and Hong Kong [Lau et al. 2006] to allow for international comparisons and to take advantage of analyses involving the combined cohorts (a total of 11,000 study participants); MrOS Sweden and MrOS Hong Kong were conducted using the same design as US MrOS to facilitate those collaborations. Although they have been very productive, here we concentrate on results from the US MrOS cohort.

Men (~6000, 90% white, mean age 73 ± 6 years) were recruited at six US academic medical centers between March 2000 and April 2002, and completed measures in study Visit 1 including hip and spine dual energy X-ray absorptiometry (DXA) and quantitative computed tomography (QCT), and assessment of bone-related markers and hormones. To be eligible for inclusion, men must have been aged ⩾65 years, been able to walk without assistance, and not have had bilateral hip replacements. Although MrOS is not strictly a representative cohort (it depended on volunteers) and thus its participants were relatively healthy at baseline, the recruitment strategy was based primarily on widespread community mailings and the characteristics of MrOS men are very similar to representative cohorts such as the US National Health and Nutrition Examination Survey (NHANES). Ongoing follow up has continued since enrollment. Every 4 months since the baseline exam, men have completed a questionnaire about recent falls and fractures; fractures are centrally adjudicated by physician review of radiology reports. Approximately 3000 men participated in an ancillary study evaluating sleep disorders, titled the MrOS Sleep Study, between December 2003 and March 2005. The entire cohort returned to the clinical centers for repeat measures from the first study visit at Visit 2 between March 2005 and April 2006. The entire cohort then again returned to the clinical centers for study Visit 3 between March 2007 and March 2009. The sleep cohort returned for repeat sleep assessment between November 2009 and March 2012. Study Visit 4 is currently underway, with participants returning for repeat assessments starting in April 2014 and continuing through April 2016. Thus MrOS has substantial longitudinal data on its participants and excellent retention in the cohort.

MrOS findings

Epidemiology of osteoporosis and fracture in men

MrOS is a large observational study describing the epidemiology of fracture and osteoporosis in older men. The proportion of men identified as having osteoporosis at baseline was 2% using the World Health Organization (WHO) reference female-specific T-score and 7% using the US National Osteoporosis Foundation (NOF) male-specific T-score [Ensrud et al. 2014].

MrOS was the first US study to demonstrate that lower bone mineral density (BMD) is associated with higher fracture risk in men; each standard deviation (SD) decrease in hip BMD increased the risk of hip fracture 3.2-fold [Cummings et al. 2006]. The BMD/fracture association was stronger in older men [Cummings et al. 2006] than in older women reported in SOF [relative risk (RR): 2.1] [Cummings et al. 1998]. Only a few other prospective studies have reported the relationship between BMD and fracture in men [Schuit et al. 2004; Johnell et al. 2005, Nguyen et al. 2005]. Moreover, we have shown that bone loss increases with advancing age in men; for example, the average BMD loss in femoral neck during a 4.5-year follow up was estimated at 0.008, 0.014 and 0.021 g/cm² for men aged 65, 75 and 85 years, respectively [Cawthon et al. 2009]. Acceleration of bone loss in men after the age of 70 was reported in the Rotterdam and MINOS studies [Burger et al. 1998; Szulc and Delmas, 2007]. In addition, MrOS investigations found that bone loss accelerated exponentially in older men and continued even in very old men [Cawthon et al. 2009]. The subjects with accelerated bone loss had an increased risk of hip and nonspine fracture [Cawthon et al. 2012]. This observation has clinical relevance given the increasing life expectancy in men and rapidly increasing population of very old men.

In parallel with SOF, we showed that ‘high trauma’ fractures in older men are ‘osteoporotic’ by their association with low BMD and increased risk of future fracture [Mackey et al. 2007]. Each SD decrease in hip BMD was associated with a 1.5-fold increased risk of high-trauma fracture in MrOS. Other studies support these findings, as seemingly high trauma distal forearm fractures in men were reported to be ‘osteoporotic’ with lower BMD at femoral neck, total hip and lumbar spine than healthy age-matched controls [Tuck et al. 2002]. These data suggest that all fractures should prompt evaluation and perhaps preventive therapy, regardless of the circumstances of trauma surrounding the fracture event. Our findings also suggest that BMD is a strong predictor of fractures in older men and may serve as a useful clinical tool in the evaluation of osteoporosis and fracture risk.

We have investigated the epidemiology of fractures at different anatomical locations. For example, we reported that vertebral fractures identified in community clinical settings were most commonly the result of a fall from standing height or less, were relatively common in older men, and increased with increasing age [Freitas et al. 2008]. Fractures were also more common in men with lower BMD at the total hip and lumbar spine. Of 5994 men followed for an average of 4.7 years, 1% (n = 61) sustained incident clinical vertebral fractures (2.2 per 1000 person-years) [Freitas et al. 2008] and 1.2% (n = 72) had hip fractures (2.6 per 1000 person-years). The age-adjusted annual rate of hip and nonvertebral fracture was 2.4 and 14 per 1000 person-years, respectively, in 4.4 years of follow up [Cummings et al. 2006]. Rib fractures were the most common fracture in MrOS, with an incidence of 3.5 per 1000 person-years in a ~6 year follow-up study [Barrett-Connor et al. 2010], and were associated with traditional risk factors for osteoporosis including old age, low BMD and history of fracture. Nearly half of fractures occurred after a fall from a standing height. Men with rib fracture were less likely to have a history of arthritis, and their calcium and vitamin D intake and medication use was not different from that in men with no rib fractures [Barrett-Connor et al. 2010].

Direct comparison with other studies of fracture rate in men is difficult because of the differences in study design and methodologies. The European Prospective Osteoporosis Study (EPOS) [European Prospective Osteoporosis Study Group, 2002] in men 50–79 years old reported age-adjusted incident vertebral fracture was 5.7 (morphometric) and 6.8 (clinical) per 1000 person-year (3.8 year follow up). In a male population of Rochester, Minnesota, the age-adjusted incidence of clinical vertebral fractures was reported to be 0.7 per 1000 person-years studied over a 4-year period (all ages, median age at diagnosis 69) [Cooper et al. 1992]. In an Australian longitudinal study lasting 3.2 years in men ⩾60 years-old, total radiographic vertebral and hip fracture incidence was reported to be 0.8 and 2.4 per 1000 person-years, respectively [Jones et al. 1994]. Another Australian study in men aged ⩾35 years [Sanders et al. 1999] reported an age-adjusted clinical vertebral and hip fracture incidence of 0.7 and 0.9 per 1000 person-years, respectively, during a 2-year follow up. Environmental factors in the etiology of fracture and secular trends, and variations in assessment of vertebral deformities, may also explain differences in fracture reported in these studies. The overall data in MrOS support that men who fractured were older, had lower BMD of the hip and spine, and were more likely to have a history of fracture.

Risk factors for fracture

Numerous risk factors for fracture have been evaluated in MrOS. Many of these analyses have investigated specific predictors by testing directed hypotheses. We have also conducted a single analysis that screened many potential risk factors (assessed at the MrOS baseline visit) for their association with incident nonspine fractures. Many measures (tricyclic antidepressant use, history of previous fracture, poor dynamic balance, history of falls, depressed mood and lower hip BMD) were associated with nonspine fracture, and the greatest risk was seen in those men who had both low BMD and several risk factors [Lewis et al. 2007]. In the next several paragraphs, we describe the associations evaluated in individual reports.

Weight and health habits

Weight and body size have been extensively and repeatedly evaluated in MrOS. Lower weight was a risk factor for a new rib fracture and nonspine fracture [Barrett-Connor et al. 2010]. Weight loss, regardless of intention and category of body mass index (BMI) (obese and overweight), was associated with greater bone loss (hip bone loss of 1.4% per year with mean weight loss of 7.7% in a 1.8-year mean follow up) [Ensrud et al. 2005] and bone loss was worse in the presence of a sex steroid insufficiency [Ensrud et al. 2006]. A previous study found that risk of hip fracture in older white men increased with loss in body weight from age 50 years to old age [Langlois et al. 1998]. Weight loss can decrease muscle mass, resulting in bone loss, by decreased mechanical forces on bone [Frost, 1997]. Other studies in women have shown that weight loss, even voluntary, is associated with increased hip fracture risk [Ensrud et al. 2003]. Overall, these data support the consideration of weight history in assessment of fracture risk in older men. The association between weight loss and fracture risk in MrOS is currently being evaluated.

We have found that the relation between BMI and fracture risk is complex. Without accounting for BMD, overweight men (but not obese men) have a slightly reduced risk of nonspine factures; without accounting for BMD, BMI is not related to hip fracture [Nielson et al. 2011]. However, after accounting for BMD, those with higher BMI are at a higher risk of both hip and nonspine fracture. In other words, for individuals with the same BMD, those with a higher BMI have a greater risk of fracture. However, the thickness of the tissue around the hip (specifically the trochanteric soft tissue thickness) was not associated with hip fracture risk in men [Nielson et al. 2009]. In a study of postmenopausal women, greater trochanteric soft tissue thickness decreased hip fracture risk [Bouxsein et al. 2007]; the thickness observed in MrOS was lower than that found in women and may not be sufficient in most men to provide significant mechanical protection during a fall.

Numerous other health habits have been assessed in MrOS, including objective assessment of sleep and physical activity. Men with sleep disturbances had a higher risk of falling [Stone et al. 2014]. Falls were higher in men who slept 5 hours or less (odds ratio = 1.79) than for those who slept 7–8 hours. Also, hypoxia during sleep (arterial oxygen saturation of less than 90% in 10% or more of sleep time, from in-home overnight polysomonography), was associated with a 40% increased risk of recurrent falls and a 30-40% increased risk of nonspine fractures [Cauley et al. 2014]. Other health habits and social conditions have been also evaluated. For example, older men with lower levels of physical activity had lower bone strength as estimated by pQCT measurements of the tibia and radius [Cousins et al. 2010]. We also found that stressful life events were associated with greater bone loss and an increased likelihood of incident falling, but this did not translate into a greater risk of fracture [Fink et al. 2014a, 2014b]. In addition, men with recent alcohol use had higher BMD, and light alcohol use was associated with a reduction in fall risk, but the association between alcohol use and fracture was not consistent [Cawthon et al. 2006].

Physical activity and physical performance

Falls are a critical factor in the etiology of fracture in men. In MrOS, higher self-reported physical activity was associated with a greater risk of falls [Chan et al. 2007] and there was a tendency for men with higher reported activity levels but lower strength to have a higher rate of falls. Moreover, when activity was measured objectively using an armband device, we found that age modified the association between activity and fall risk, such that in men younger than 80, those with the lowest energy expenditure (<1944 kcal/day) had a 25% reduced risk of falls compared with those with the highest energy expenditure (>2661 kcal/day), while in men older than 80, those with the lowest energy expenditure had a 40% increased risk of falls compared to those with higher energy expenditure [Cauley et al. 2013]. Low objectively measured activity levels were also associated with a higher risk of nonspine fracture [Cauley et al. 2013], but physical activity by self-report was not associated with hip fracture risk [Mackey et al. 2011]. Thus, it may be that men who are active but less physically capable, as well as men who are frailer, are those most likely to fall. However, the association between activity and falls and fractures is complicated, and the association varies depending on the method used to assess activity level.

Poor physical performance, in particular inability to rise from a chair, is a strong risk factor for falls [Chan et al. 2007; Karlsson et al. 2012], hip fractures and radiographic vertebral fracture [Cawthon et al. 2008, 2014a; Rosengren et al. 2012]. Worse frailty status is also associated with a greater risk of falls and fractures [Ensrud et al. 2009a]. We have also investigated the inter-relationship of bone and muscle, and how this may affect fracture risk. Bone–muscle indices, expressed as bone-to-muscle ratios for strength, mass and area (derived from pQCT and DXA scans) were risk factors for nonspine and clinical vertebral fractures in men; however, these measures were correlated with areal BMD (aBMD) from DXA and the associations between the bone–muscle indices and fracture were not independent of aBMD [Wong et al. 2014].

Medication use and medical conditions

In MrOS, we have evaluated the relation between numerous medical conditions and medications with BMD, falls and fractures. The use of several medications was associated with lower BMD, such as selective serotonin reuptake inhibitors (SSRIs) [Haney et al. 2007] or loss of BMD over time, such as antiepileptic drugs [both nonenzyme-inducing antiepileptic drugs (NEIAEDs) and enzyme-inducing antiepileptic drugs (EIAEDs)] [Ensrud et al. 2008]. However, we found no association between warfarin [Woo et al. 2008] and BMD, BMD loss or fracture risk. We also found no association between the use of the general class of diuretic medications and BMD or BMD loss [Lim et al. 2005], although our results suggested that men who used loop diuretics had increased BMD loss over time [Lim et al. 2008]. Nonbenzodiazepine sedative hypnotics used in the treatment of sleep disturbances or anxiety-related disorders were associated with increased risk of falls and benzodiazepines were related to increased falls by increased disability, depressive symptoms and poorer physical performance [Diem et al. 2014].

We have also evaluated the association between several medical conditions with low BMD, greater BMD loss or increased fracture risk. We found that abdominal aortic calcification (AAC) assessed on lateral spine radiographs was associated with increased risk of hip fracture [Szulc et al. 2014]. The link between AAC and fracture risk has been studied mainly in women [Bagger et al. 2006] and our studies expand on the limited prospective data available in men [Samelson et al. 2007; Szulc et al. 2008a] by establishing the association between severe AAC and hip fracture. Depressive symptoms were also associated with BMD loss, although this association was at least partially explained by poorer physical functioning in those exhibiting depressive symptoms [Diem et al. 2013]. The association between depression and fracture has not yet been evaluated. A subset of MrOS men had a periodontal exam, and we found little evidence for a cross-sectional association between periodontal disease and skeletal BMD [Phipps et al. 2007]. Men with radiographic hip osteoarthritis (OA) have higher BMD at the hip and spine than men without OA [Chaganti et al. 2010]. MrOS used volumetric BMD (vBMD) (QCT), in addition to aBMD (DXA), and found that cortical BMD in the hip was increased in OA and the trabecular BMD was not [Chaganti et al. 2010]. Older men with Parkinson’s disease may benefit from fall prevention and osteoporosis measures as MrOS found that these patients had lower BMD [Fink et al. 2005] a three-fold higher risk of multiple falls, and higher bone loss and fracture [Fink et al. 2008]. Poor renal function was associated with a greater rate of hip bone loss [Ishani et al. 2008], but it was not associated with fracture risk after accounting for the large comorbidity burden of those with poor renal function [Ensrud et al. 2014]. Chronic obstructive pulmonary disease (COPD) was associated with lower spine BMD cross-sectionally and an increased risk of nonspine and spine fractures, but was not associated with loss of spine BMD over time [Dam et al. 2010]. MrOS showed that diabetics have an increased risk of fracture [Napoli et al. 2014], similar to findings in SOF that older women with diabetes had an increased risk of nonvertebral fractures [Schwartz et al. 2001]. In addition, MrOS data have been used in meta-analyses, including one that found that fracture risk amongst those with diabetes was higher for a given T-score and age, or for a given Fracture Risk Assessment (FRAX) score, than the risk amongst those without diabetes at the same T-score and age, or FRAX score [Schwartz et al. 2011]. This suggested that diabetics should be identified for treatment at different BMD or FRAX risk levels than those without diabetes. Diffuse idiopathic skeletal hyperostosis (DISH), a sclerosing bone disorder, was associated with a greater likelihood of prevalent vertebral fracture [Diederichs et al. 2011]. Men with moderate to severe lower urinary tract symptoms had a greater risk of falls than men free of such symptoms [Parsons et al. 2009]. Altogether, these findings suggest that complete evaluation of fracture risk in older men should include a complete review of medications and medical conditions.

Race, BMD and fracture risk

Using data from MrOS and other cohorts, we have shown that men of African descent have higher BMD compared with Caucasian men, and that the lower BMD observed among men of Asian descent compared with Caucasian men is largely explained by differences in body size [Nam et al. 2010]. Rates of BMD loss with aging appear to be similar between Caucasian and African American but slower in Asian men [Sheu et al. 2011]. The association between BMD and fracture risk appear similar across race and ethnic groups [Shin et al. 2014]. Most of the fractures occurred in white men. Structural analysis using QCT showed that black and Asian men had thicker cortices in the femoral neck and that higher trabecular vBMD that may explain their lower hip fracture rate [Marshall et al. 2008].

Hormones and circulating markers

We have extensively evaluated sex hormones and their relation to BMD, falls and fractures in MrOS. A low level of sex hormones (total testosterone <200 ng/dl, or total estradiol <10 pg/ml, using radioimmunoassay) was associated with higher prevalence of osteoporosis at baseline and greater loss of BMD over time [Fink et al. 2006]. This cross-sectional analyses in 2450 men (mean age 73 years) revealed 3% of men as testosterone deficient, 3.2% as estradiol deficient and 0.7% deficient in both. Estradiol deficiency was a stronger predictor of osteoporosis, whereas testosterone deficiency was a stronger predictor of rapid hip bone loss (loss ⩾3% per year) [Fink et al. 2006]. More accurate sex steroid measures (using chromatography–mass spectrometry) in MrOS clearly established that lower bioavailable estradiol or higher sex hormone binding globulin (SHBG) related to greater hip bone loss [Cauley et al. 2010a] and fracture [Leblanc et al. 2009]; it showed a threshold level of estradiol for fracture [Mellstrom et al. 2008] but found no association, independently between bioavailable or total testosterone, and fracture [Mellstrom et al. 2008; Leblanc et al. 2009] suggesting that estradiol, and not testosterone, may be the major sex hormone with an impact on fracture risk in older men. Men with low testosterone levels had worse physical functioning and a greater risk of falls [Orwoll et al. 2006]. The association of testosterone level to the risk of falling persisted regardless of physical performance, but the association was more robust in the younger men. In a study with potentially wide implications, US, Swedish, Hong Kong and Japanese investigators reported that levels of testosterone, estradiol, SHBG and other steroid hormones appear to vary considerably by race and geography [Orwoll et al. 2010], raising the concern that the relationship between sex steroids and skeletal health may have to be considered independently as a function of these factors.

MrOS also examined the relationship between total 25-hydroxy vitamin D [25(OH)D] and fractures in older men. Vitamin D deficiency was common in MrOS in one cross-sectional study (26% of men deficient, 25(OH)D <20 ng/ml, and 72% insufficient, 25(OH)D <30 ng/ml) [Orwoll et al. 2009b]. We found that low vitamin D levels are associated with higher rates of bone loss and higher fracture risk (mean of 21.5 ± 7.9 ng/ml in hip fracture subjects versus 25.2 ± 7.8 ng/ml in controls, p < 0.0001) [Ensrud et al. 2009b; Cauley et al. 2010b], suggesting potential use of serum 25(OH)D in identifying men at high risk of hip fracture. The presence of low vitamin D in the presence of low bioavailable estradiol and high SHBG was associated with greater bone loss and higher fracture risk, suggesting that men with this combination may be at particular risk [Barrett-Connor et al. 2012]. Measures of 1,25-dihydroxyvitamin D [1,25(OH)2D] did not appear to add to the predictive value of 25(OH)D levels [Swanson et al. 2015], but higher parathyroid hormone (PTH) can contribute to the relationship between 25(OH)D and bone loss or hip fracture. Higher intact PTH was associated with greater rates of BMD loss, irrespective of vitamin D status [Curtis et al. 2012]. A number of other hormones and circulating factors have been evaluated as potential risk factors for fracture. For example, we found that higher serum uric acid levels (presumably acting as an antioxidant) were associated with higher BMD and a lower risk of nonspine (but not hip fractures) [Lane et al. 2014]. We also measured thyroid hormone levels. Neither thyroid-stimulating hormone (thyrotropin) (TSH) nor free thyroxine (FT4) was associated with bone loss [Waring et al. 2013]. Only TSH, but not FT4 or categories of thyroid function, was associated with hip fracture risk; no measure of thyroid function or hormones was associated with risk of any nonspine fracture [Waring et al. 2013]. Higher blood lead levels were associated with lower hip BMD [Khalil et al. 2014], but the association with fracture has not yet been investigated [Bauer et al. 2009]. Finally, fibroblast growth factor 23 (FGF23) was only associated with fracture risk in men with poor renal function [Lane et al. 2013]. These findings suggest that laboratory testing may be useful to diagnose the causes of bone loss in men especially in the absence of apparent etiology.

We have measured several markers of bone turnover [type I collagen N-propeptide [PINP]; β-C-terminal cross-linked telopeptide of type I collagen (β-CTX) and tartrate-resistant acid phosphatase (TRACP5b)] in MrOS. Higher levels of these markers were associated with greater hip bone loss in older men, but not fracture risk after accounting for baseline BMD [Bauer et al. 2009]. Some studies in men have found that elevated bone turnover markers were associated with modestly increased fracture risk [Meier et al. 2005], others found no such association [Szulc et al. 2008b]. In MrOS, elevated bone turnover markers did not independently predict fracture risk in older men.

Ultrasound and imaging

MrOS participants have had heel ultrasound and numerous imaging studies completed and metrics from these assessments have been extensively evaluated. MrOS generated the first prospective demonstration, for men or women, that measurements of bone structure by QCT, and of bone strength by finite element analysis (FEA), predict hip and other fractures, often independently of aBMD. For example, for incident hip fracture, measures of structure and strength of the proximal femur were associated with hip fracture risk in MrOS, but only measures of strength (femoral strength and load-to-strength ratio) derived from FEA were risk factors independent of aBMD from hip DXA [Black et al. 2008; Orwoll et al. 2009a]. For incident vertebral fractures, we also found that vertebral compressive strength and vertebral load-to-strength ratio (from FEA) were associated with clinically identified vertebral fractures independent of aBMD of the spine from DXA [Wang et al. 2012]. Finally, we found that quantitative ultrasound at the heel predicted fracture risk in men, but this association was not independent from aBMD from DXA [Bauer et al. 2007]. Thus, important structural and biomechanical measures may improve fracture prediction in older men, although their feasibility in clinical practice has not been established.

Screening and risk prediction

MrOS data support that older men with risk factors (e.g. low body weight, history of fracture as an adult, low testosterone, hyperparathyroidism and medical causes present) may benefit from fracture risk assessment including BMD testing. MrOS BMD and fracture data were used in a simulation cost–benefit analysis. This US-based analysis concluded that BMD assessment followed by bisphosphonate therapy for those with osteoporosis may be cost-effective for men aged 65 years or older with a self-reported prior clinical fracture and for men aged 80–85 years with no prior fracture [Schousboe et al. 2007]. Models examining populations from Europe also concluded that cost-effective intervention can be offered even for men at moderate risk [Kanis et al. 2005]. Further simulations using data from MrOS found that body weight can be used to select men for whom bone density screening would be cost-effective [Schousboe et al. 2013].

We have also investigated the proportion of men who would be recommended for treatment or screening based on clinical guidelines, such as those by NOF [Orwoll et al. 2005] or WHO [WHO, 1994; Kanis et al. 2008]. Overall, in MrOS we found that different guidelines would result in widely divergent estimates of the number of men who would be recommended for treatment or screening [Donaldson et al. 2010]. Fracture risk is high in men who meet WHO criteria for osteoporosis (that is, femoral neck BMD T-score ⩽–2.5, based on young female normal reference database), in men with existing vertebral fractures, or in men with a recent history of hip fracture, and MrOS data support offering treatment to these individuals [Ensrud et al. 2014]. However, the value of expanding indications for treatment beyond these groups is unclear given lower fracture rates in other groups, such as those with a high FRAX likelihood of fracture without osteoporosis, prevalent vertebral fracture or hip fracture history [Ensrud et al. 2014].

MrOS included only men aged 65 years and older. The results of changes in BMD and incident fracture cannot be generalized to younger men. MrOS participants are also predominantly white, so the study is limited in examining the risk factors for prediction of fracture for non white older men. The relatively low frequency of fractures has limited, in some cases, assessment of the relationship between incident fractures, bone loss and other risk factors. However, the prospective nature of MrOS, its large cohort size and its origins from several US communities have offered a unique opportunity to study the nature of osteoporosis, its relation to many different risk factors and its health implications in the broad population of aging men.

Skeletal genetics

Skeletal characteristics are strongly influenced by genetic factors; it is estimated that the heritability of BMD is ~60–80% [Nguyen et al. 2000; Karasik et al. 2002]. A variety of analyses from MrOS have contributed to an increased understanding of the genes that may be important. Candidate gene association studies have suggested chromosomal loci that are associated with BMD [Yerges et al. 2009, 2010; Zmuda et al. 2011] and a detailed evaluation of variants in the alkaline phosphatase gene revealed that they influenced circulating alkaline phosphatase levels and BMD [Nielson et al. 2012]. In collaborations with other cohorts in meta-analyses, MrOS has contributed to the discovery of additional chromosomal loci that are associated with BMD and fracture, including in a recent report that revealed an interesting and novel regulatory variant affecting a gene (EN1) not previously known to be involved in bone metabolism [Zheng et al. 2015]. Ongoing MrOS studies involving exome and whole genome sequencing will further add to the field.

Conclusion

The number of elderly men is increasing worldwide. Through large longitudinal cohort studies such as MrOS we have identified a number of risk factors that are associated with low bone mass and fractures, thus improving our understanding of osteoporosis and fracture risk in older men. In addition, we have developed excellent clinical tools for detecting men with low bone strength and/or increased fall risk who might benefit from treatment to prevent fractures. Lastly, a number of clinical trials have demonstrated the effectiveness of several pharmacological agents to treat and prevent osteoporosis in men. Our major task now is to educate both the public and the medical community about utilizing these tools to improve the detection, prevention and treatment of osteoporosis in men.