Serum high sensitivity C-reactive protein poorly predicts bone mineral density: A NHANES 2017-2020 analysis

A reliable, widely available method to detect osteoporosis prior to fracture is needed. Serum levels of C-reactive protein may independently predict low bone mineral density (BMD) and high fracture risk. Existing empirical data focus on sexually and/or racially homogenous populations. This study tests the hypotheses that: C-reactive protein (1) negatively correlates with BMD and (2) fracture history, and (3) independently predicts BMD and fracture history in a diverse population. NHANES 2017-2020 pre-pandemic cycle data were analyzed in R studio. Strength and direction of relationships (-1 to +1) between variables were determined using Kendall's rank correlation coefficient (tau). Linear models were optimized to predict femoral neck or lumbar spine BMD. C-reactive protein positively correlated with femoral (tau = 0.09, p<0.0001) and spine BMD (tau = 0.10, p<0.0001). Individuals identifying as female demonstrated more robust, but still weak, correlations between C-reactive protein and femoral neck (tau = 0.15, p<0.0001; male, tau = 0.06, p = 0.051) and spine BMD (tau = 0.16, p<0.0001; male, tau = 0.06, p = 0.04). C-reactive protein positively correlated with fracture history (tau = 0.083, p = 0.0009). C-reactive protein significantly predicted femoral neck (R-2 = 0.022, p = 0.0001) and spine BMD (R-2 = 0.028, p<0.0001) and fracture history (R-2 = 0.015, p<0.0001). Exploratory analyses identified weight was the single best predictor for femoral neck (R-2 = 0.24, p<0.0001) and spine BMD (R-2 = 0.21, p<0.0001). In sum, C-reactive protein statistically correlates with and predicts femoral neck and spine BMD, but the magnitude is too low to be biologically meaningful. While weight is a more robust predictor, individuals who are overweight or obese account for nearly half of all osteoporotic fractures, limiting the predictive power of this variable at identifying individuals at risk for osteoporosis. Identification of a robust predictor of fracture risk in a diverse population and across of range of body weights and compositions is needed.