# External Validation of PathFx 3.0 Model on a Single-Centre Portuguese Population

## TL;DR

This study tested the accuracy of the PathFx 3.0 model in predicting survival for patients with bone metastases in a Portuguese population.

## Contribution

The paper provides an external validation of PathFx 3.0 in a new geographic and healthcare context.

## Key findings

- PathFx 3.0 showed poor early survival prediction but improved accuracy after three months.
- Calibration was good except in the early post-operative period.
- Observed survival rates were consistent with predictions in later time periods.

## Abstract

Purpose

PathFx 3.0 (Jonathan A. Forsberg, Memorial Sloan Kettering Cancer Center, New York, NY, USA; Rikard Wedin, Karolinska Institutet, Stockholm, Sweden) is a prognostic model designed to predict survival in patients with bone metastatic disease. While it has been widely validated, concerns remain regarding its application across different healthcare systems. This study aimed to externally validate the performance of PathFx 3.0 in a Portuguese cohort.

Methods

A retrospective cohort of patients who underwent surgery for skeletal metastases between January 2018 and April 2023 was analysed. Survival probabilities at one, three, six, 12 and 24 months were generated for each patient. Performance was assessed using the area under the receiver operating characteristic curve (AUC-ROC) with 95% confidence intervals, Brier scores and calibration.

Results

A total of 49 patients were included, with a median follow-up of 19 months (interquartile range (IQR) 5-35). Observed survival was 89.8% at one month, 81.6% at three months, 69.4% at six months, 65.3% at 12 months and 42.9% at 24 months. Discrimination was poor at one month (area under the curve (AUC) 0.495) but improved in subsequent time periods, with AUC values of 0.760 at three months, 0.773 at six months, 0.791 at 12 months and 0.752 at 24 months. Brier scores ranged from 0.172 to 0.246. Calibration analysis showed good agreement between predicted and observed outcomes, except in the early post-operative period.

Conclusion

PathFx 3.0 demonstrated good performance in the studied population. While early survival prediction remains limited, the model provides valuable prognostic support for operative decision-making in patients with metastatic bone disease in later time periods. Continued external validation in larger, multicentric Portuguese cohorts is warranted.