# The diagnostic value of inflammatory markers in bone metastasis of prostate cancer at initial prostate biopsy

## TL;DR

This study shows that combining inflammatory markers with clinical data improves the prediction of bone metastasis in prostate cancer patients at diagnosis.

## Contribution

A novel nomogram integrating inflammatory markers with clinical parameters is developed for accurate bone metastasis prediction in prostate cancer.

## Key findings

- The comprehensive model with inflammatory markers achieved an AUC of 0.874, significantly higher than the clinical model.
- The model showed a net reclassification improvement of 8.96% and an overall IDI of 10.3%.
- The model is well-calibrated and provides a positive net benefit for clinical decision-making.

## Abstract

Accurate prediction of bone metastasis at diagnosis is crucial for optimizing management in prostate cancer (PCa) patients. While clinical parameters like PSA and Gleason score are established predictors, their accuracy is suboptimal. Systemic inflammation, reflected in biomarkers like the high-sensitivity C-reactive protein-to-albumin ratio (HAR), fibrinogen (FIB), and hemoglobin (HB), has emerged as a key player in cancer progression, yet its integration into clinical predictive tools remains underexplored.

In this retrospective study of 803 newly diagnosed PCa patients, we developed and validated two nomograms for predicting bone metastasis. A baseline clinical model was constructed using total prostate-specific antigen (TPSA) and biopsy Gleason grade groups. An enhanced comprehensive model integrated these clinical parameters with inflammatory markers (HAR, FIB, HB). Model performance was rigorously assessed through discrimination (ROC analysis, AUC), calibration (calibration curves, Hosmer-Lemeshow test), and clinical utility (Decision Curve Analysis). Internal validation was performed via bootstrapping.

Multivariate analysis confirmed TPSA, FIB, HB, HAR, and Gleason grade groups as independent predictors of bone metastasis. The comprehensive model demonstrated significantly superior discriminative ability, achieving an AUC of 0.874 (95% CI: 0.845–0.902) compared to 0.830 (95% CI: 0.798–0.863) for the clinical model (Delong’s test, P < 0.01). This translated to a net improvement in reclassification (NRI: 8.96%) and overall predictive performance (IDI: 10.3%). The model was well-calibrated and provided a positive net benefit across a wide range of clinical threshold probabilities.

We present a novel, internally validated nomogram that synergistically combines inflammatory and clinical markers to accurately predict bone metastasis in PCa at initial diagnosis. This practical and cost-effective tool has the potential to aid clinicians in risk stratification, guide personalized diagnostic imaging decisions, and ultimately help reduce unnecessary bone scans, particularly in resource-conscious settings. Our findings underscore the pivotal role of the systemic inflammatory response in PCa metastasis.