Article Text

Download PDFPDF
Broken hearts and bones: new insights or falling for unmeasured confounding?
  1. Joshua R Lewis1,2,3,
  2. Pawel Szulc4
  1. 1 School of Medical and Health Sciences, Edith Cowan University, Joondalup, Western Australia, Australia
  2. 2 School of Public Health, The University of Sydney, Centre for Kidney Research, Sydney Medical School, Children’s Hospital at Westmead, Westmead, New South Wales, Australia
  3. 3 Medical School, University of Western Australia, Perth, Australia
  4. 4 INSERM UMR 1033, University of Lyon, Hospices Civils de Lyon, Lyon, France
  1. Correspondence to Dr Joshua R Lewis, School of Medical and Health Sciences, Edith Cowan University, Joondalup, Western Australia, Australia ; joshua.lewis{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Most minimal trauma or osteoporotic fractures are due to low bone mass and/or quality combined with a fall from standing height or less. Osteoporosis is common in older men and women and combined with increased falls propensity during ageing leads to the dramatic rise in the risk of minimal trauma fracture in older men and women. Causes of falls in older men and women are often complex and can be due to loss of skeletal muscle mass and strength, acute and chronic illnesses, cardiovascular causes, medications, nutritional deficiencies and environmental hazards, either individually, or in combination.

While cardiovascular disease and minimal trauma fractures are generally thought of as independent diseases of old age, low bone mass and osteoporotic fracture are associated with cardiovascular events and deaths.1 The mechanism(s) underpinning this relationship are yet to be fully explained. However, there are shared genetic and modifiable risk factors, comorbid conditions and common signalling pathways regulating bone homeostasis and vascular wall repair. In particular, low bone mass is associated with greater abdominal aortic calcification, which is a marker of advanced atherosclerosis and future cardiovascular disease risk.2

The evidence linking cardiovascular disease to the long-term risk of minimal trauma fracture is conceptually similar to that of low bone mass and fracture with cardiovascular disease. However, there are subtle differences: (i) the skeleton is dependent on the vasculature for blood and nutrient supply with vascular disease and occlusion potentially leading to increased bone loss, (ii) the skeletal muscles are also dependent on the vasculature for blood and nutrient supply with vascular disease and occlusion potentially leading to decreased muscle function, (iii) cardiovascular disease can cause falls and increased falls propensity by mechanisms such as neurally mediated disorders and cardiac abnormalities of structure and function,3 (iv) abdominal aortic calcification is associated with decreased bone mass and increased fracture risk.2 To date, few longitudinal studies have investigated whether cardiovascular biomarkers predict long-term fracture risk.

In this regard, Johansson and colleagues report in their Heart manuscript on the relationship between four biomarkers of neuroendocrine activation and endothelial dysfunction, measured at a single time-point, and the risk of future fracture hospitalisation in a randomly selected subset of the Malmö Preventive Project that attended a rescreening visit in 2002–20064. The 5415 (30%) community-dwelling individuals (mean age, 68.9±6.2 years, 30.3% female) had Midregional proadrenomedullin (MR-proADM), midregional proatrial natriuretic peptide (MR-proANP), C-terminal pro-arginine vasopressin (CT-proAVP) and C-terminal proendothelin-1 (CT-proET-1) measured at baseline with up to 18 years of follow-up of fracture-hospitalisation through the Swedish National Hospital Discharge Register. MR-proADM is a precursor of adrenomedullin and a possible surrogate marker of adrenomedullin system that exerts multiple cardioprotective effects.5 MR-proANP has been studied mainly as a prognostic marker in cardiovascular diseases. CT-proET-1 is released from proendothelin-1, stoichiometrically in a 1:1 ratio with endothelin-1. CT-proET-1 has a longer half-life than endothelin-1, is quantified more easily and is considered to be a surrogate marker of endothelin-1. CT-proAVP is a surrogate marker of vasopressin, which contributes to the development and progression of chronic kidney disease. Elevated levels of these biomarkers have been reported in patients with syncope,6 and peripheral arterial disease in older men for MR-proADM, MR-proANP and CT-proET-1.7

Over a mean follow-up period of 8.1±2.9 years, 1030 older men and women suffered a fracture-hospitalisation. In unadjusted Cox regression, per SD increase in log-transformed MR-proADM, MR-proANP and CT-proET-1 were associated with a 30%–37% higher relative hazard for fracture-hospitalisation. Importantly, the association between MR-proADM and MR-proANP remained significant even after adjusting for age, gender, body mass index, systolic blood pressure, heart rate, antihypertensive, treatment, smoking, diabetes, prevalent fractures, history of cardiovascular/cerebrovascular disease and physical activity. Men and women in the highest quartile for all four biomarkers had a twofold to threefold increase in the relative hazard for fracture-hospitalisation, even after adjusting for the above-mentioned confounders. The reported associations were most evident for spine and femur fracture-hospitalisation. An important consideration when interpreting these findings is that significant interactions between genders with all four biomarkers for fracture-hospitalisations were identified, with the most apparent associations in older men.

This study and its findings are notable for a number of reasons. First, it investigated less well-studied vascular biomarkers of endothelial dysfunction with a long-term fracture-hospitalisation risk. Second, it was a large, well-powered study, with a large number of fracture-hospitalisations in both older men and women. Third, cohorts investigating fracture risk factors in older men are rare and the findings are novel and potentially clinically important, given the high fracture risk in this population and the paucity of identified risk factors. Fourth, all the fractures were serious enough to warrant hospitalisations and were identified independent of self-report. Finally, the cohort measured and adjusted for a number of potentially important confounders suggesting the findings are robust and independent of the confounders adjusted included in the models.

However, a number of important limitations of the study need to be considered. First, as this was an observational study, causality cannot be established. Second, while some fracture and falls risk factors were assessed, important factors, such as bone mineral density, muscle mass and function, dizziness and balance, and medications known to affect falling were not. Similarly, measures of atherosclerosis and newer cardiac biomarkers such as high-sensitivity cardiac troponins were not assessed. As such, it remains unclear what pathophysiological aspect(s) of fracture risk these biomarkers may capture, whether the associations are independent of established clinical measures of muscle and bone and whether this increased risk may be reduced with treatment. Third, the fractures captured in this study were only those that resulted in a hospitalisation, and as such other more common fractures that do not generally result in a hospitalisation were not captured. Therefore, the relationship with less serious fractures not requiring hospitalisation remains uncertain.

The findings that MR-proADM predicts fracture-hospitalisations are supported by a recent study in the prospective Malmö Diet and Cancer study of 5291 older men and women (mean age, 57 years; 59% women). Härstedt and colleagues reported that per SD increase in MR-proADM was associated with a 23% increase in the relative hazard of fracture-hospitalisations in older men over a median of 21 years.8 Importantly, again this association was only observed in men and no measures of bone mineral density, muscle mass or function were available.8 In contrast to the finding reported by Johansson and colleagues4, the study by Härstedt and colleagues found no association between MR-proANP and CT-proAVP with long-term fracture-hospitalisations.8 These differences may be due to a younger cohort, substantially longer follow-up between biomarker assessment and fracture-hospitalisations and fewer older men.

When evaluating whether novel biomarkers are clinically useful, it is important to carefully consider whether they are diagnostic or prognostic, can be used to monitor disease severity or progression, and can be used to guide treatment or whether they are in themselves a potential therapeutic target. New clinical biomarkers assays must also be affordable, precise and accurate and meet an unmet clinical need. The association between the marker and the outcome should be biologically plausible. Thus, given the link between cardiovascular disease with (i) bone mineral density, (ii) skeletal muscle mass and function, (iii) falls and falls propensity, and (iv) vascular calcification for fracture risk, it is interesting to analyse their association between the four reported biomarkers with these outcomes. Unfortunately, published data are scarce and inconclusive.

To our knowledge, there is no evidence to support MR-proADM, MR-proANP, CT-proET and CT-proAVP as biomarkers of bone mineral density, osteoporosis, muscle mass or function in healthy older individuals. Similarly, there are no data on these biomarkers for long-term falls risk, although their relationship with syncope and structural vascular disease conceptually supports this link. Animal studies in rats consistently demonstrate adrenomedullin inhibits vascular calcification,9 however no association between plasma MR-proADM with coronary calcification was observed in humans.10 An experimental study in a chronic kidney disease animal model in rats suggests that endothelin-1 may also be involved in vascular calcification with blockade of the endothelin-1 receptor inhibiting vascular calcification,11 but there are no human data. To our knowledge, no experimental or clinical studies have assessed the association between MR-proANP or CT-proAVP with vascular calcification. While there is some evidence linking two of the biomarkers with vascular calcification, data are very limited.

In conclusion, while these findings are interesting they should be interpreted cautiously until they have been externally replication in other studies with established measures of atherosclerosis and musculoskeletal phenotypes. The suggestion that these biomarkers should be considered to identify individuals at a high fracture risk who may benefit from the more restrictive use of antihypertensive and psychopharmacological drugs is premature until the mechanism(s) for increased fracture risk is elucidated. However, while the current findings are unlikely to change clinical practice, they are interesting and should stimulate further research in this important research area. In particular, further research in this area may identify novel non-bone-related treatments for reducing the burden of minimal trauma fractures in older men, an often overlooked and undertreated group.



  • Contributors JRL and PS wrote the editorial.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Commissioned; internally peer reviewed.

  • Data sharing statement Not applicable.

Linked Articles