External validation of the PATHFx decision-support tool on Turkish patients with skeletal metastasis.

Abstract

Accurate determination of life expectancy becomes very important when determining the treatment of patients with pathologic fractures. We aimed to investigate the predictive role of the PATHFx model in Turkish patients by estimating the area under curve (AUC) of the receiver operator characteristic (ROC) and externally validating the results of PATHFx on the Turkish population.

The data of 122 patients who presented to one of four orthopaedic oncology referral centres in Istanbul (2010-2017) and underwent surgical management of pathologic fractures were retrospectively collected. Patients were evaluated according to age, sex, type of pathologic fracture, presence of organ metastasis, presence of lymph node metastasis, haemoglobin concentration at presentation, primary oncologic diagnosis, number of bone metastases, and Eastern Cooperative Oncology Group (ECOG) status. Estimations of the PATHFx program by months were statistically evaluated using ROC analysis.

In our study population (122 patients), all survived the first month, 102 survived the third month, 89 were alive at 6 months, and 58 patients survived at 12 months. At 18 and 24 months, 39 and 27 patients were alive, respectively. The AUC value was 0.677 at 3 months, 0.695 at 6 months, 0.69 at 12 months, 0.674 at 18 months, and 0.693 at 24 months. The 3-, 6-, 12-, 18-, and 24-month survival rates were statistically significant (P < 0.01 and P < 0.05). ECOG performance status was 0-2 points in 33 patients (Memorial Sloan-Kettering Cancer Center (MSKCC) data set: 93 cases, our data set: 33 cases). ECOG performance status was 3-4 points in 89 patients (MSKCC data set: 96 cases, our data set: 89 cases).

The objective data used by PATHFx for prediction provided statistically accurate estimates on Turkish patients, who are presumed to have mixed genomes through history from both Europe and Asia, and demonstrates its applicability to the Turkish population.