Developing Novel Prognostic Biomarkers for Multivariate Fracture Risk Prediction Algorithms

Abstract

Multivariate prediction algorithms such as FRAX ® and QFractureScores provide an opportunity for new prognostic biomarkers to be developed and incorporated, potentially leading to better fracture prediction. As more research is conducted into these novel biomarkers, a number of factors need to be considered for their successful development for inclusion in these algorithms. This review paper describes two well-known multivariate prediction algorithms for osteoporosis fracture risk applicable to the UK population, FRAXand QFractureScores, and comments on the current prognostic tools available for fracture risk, dual x-ray assessment (DXA), quantitative ultrasound (QUS), genomic and biochemical markers and highlights the factors that need to be considered in the development of new biomarkers. These factors include the requirement for prospective data, collected in new cohort studies or using archived samples, the need for adequate stability data to be provided and appropriate storage methods to be used when retrospective data is required. AUC measures have been found to have limited utility in assessing the impact of the addition of new risk factors on the predictive performance multivariate algorithms. New performance evaluation measures, such as net reclassification index (NRI) and integrated discrimination improvement (IDI) are increasingly important in the evaluation of the impact of the addition of new markers to multivariate algorithms and these are also discussed.

Riesgo de fractura en la cohorte FRODOS: estudio comparativo de la aplicación del modelo FRAX® español, francés, inglés y sueco

“…Ideally, in order for a model for the prediction of clinical risk such as FRAX ® to be used with confidence in daily clinical practice it ought to comply with at least two conditions: having demonstrated its validity in other population groups similar to the original ones; and helping to resolve problems for users less experienced in the field of osteoporosis, be they general practitioners or health care planners [20][21][22] . It is evident that if the first point is not complied with there should not be a move towards the generalisation of its use since we would be doing this on an inappropriate basis 2,8,17,[20][21][22] . The Spanish FRAX ® model has been evaluated in three cohort studies 5-7 which clearly differ, but which, in having a sufficient number of participants and events, agree in their conclusions: the Spanish FRAX ® model clearly predicts a lower number of POFs than are observed, while the prediction of FFs is somewhat closer to what actually happens; however, the predictive power measured by the area under the curve of the ROC curves is no higher than 70%.…”

“…It is for these reasons that for more than two decades various research groups have created different tools to optimise the diagnosis, calculate the risk of associated and secondary fracture, and improve the prevention and treatment of osteoporosis 2 . One of the most widely-used tools is the FRAX ® index 3 .…”

Fracture risk prediction with FRAX in Slovak postmenopausal women

“…When compared to other risk assessment methods, FRAX performs similarly or better than other simpler methods (based on age, previous fracture or BMD), outcomes on comparison with other calculators are inconsistent [11][12][13][14][15][16][17][18]. However, construction on multiple population cohorts, external validation, applicability to various countries, output as absolute risk and inclusion of risk factors amenable to treatment are major advantages of FRAX compared to other risk assessment tools [19,20].…”

Osteoporosis – a current view of pharmacological prevention and treatment

“…FRAX ® has been validated for use in many countries; however, there is some debate about whether these risk calculators actually provide better prediction of osteoporotic fracture than assessing age and BMD alone, and the importance of calibration of the algorithms to local cohorts, time period for risk assessment, and inclusion of risk of mortality have been highlighted and discussed in detail 22–24. It is likely that other non-BMD assessments in addition to DXA are required to improve osteoporosis detection 25. Active shape and appearance modeling were developed based on previous observations that differences in femoral geometry and bone quality might increase susceptibility to hip fracture 26,27.…”