Predicting Toxicities and Survival Outcomes in De Novo Metastatic Hormone-Sensitive Prostate Cancer Using Clinical Features, Routine Blood Tests and Their Early Variations
by
, , , , , , , , and
Giuseppe Salfi
1,2,†
,
Martino Pedrani
1,2,†,
Amos Colombo
3,4,
Lorenzo Ruinelli
3,4,
Daniele Brenna
1,
Chiara Maria Agrippina Clerici
1,
Giovanna Pecoraro
1,2,5,
Sara Merler
6,
Caroline-Claudia Erhart
1,2,7,
Marialuisa Puglisi
1,8,
Fabio Turco
1,
Luigi Tortola
1,
Ursula Vogl
1,
Silke Gillessen
1,7 and
Ricardo Pereira Mestre
1,2,7,* 1
Oncology Institute of Southern Switzerland (IOSI), Ente Ospedaliero Cantonale (EOC), 6500 Bellinzona, Switzerland
2
Institute of Oncology Research (IOR), 6500 Bellinzona, Switzerland
3
ICT (Information and Communication Technology), Ente Ospedaliero Cantonale (EOC), 6500 Bellinzona, Switzerland
4
CTU (Clinical Trial Unit), Ente Ospedaliero Cantonale (EOC), 6500 Bellinzona, Switzerland
5
Department of Oncology, University of Naples Federico II, 80138 Naples, Italy
6
Oncology Department, San Maurizio Central Hospital of Bolzano, South Tyrolean Health Service, 39100 Bolzano, Italy
7
Faculty of Biomedical Sciences, Università della Svizzera Italiana, 6900 Lugano, Switzerland
8
Department of Human Pathology “G. Barresi”, School of Specialization in Medical Oncology, University of Messina, 98125 Messina, Italy
*
Author to whom correspondence should be addressed.
†
These authors contributed equally to this work.
Cancers 2025, 17(23), 3806; https://doi.org/10.3390/cancers17233806
Submission received: 16 October 2025
/
Revised: 21 November 2025
/
Accepted: 26 November 2025
/
Published: 27 November 2025
(This article belongs to the Special Issue Application of Artificial Intelligence-Based Approaches in Cancer Diagnosis, Treatment and Prognosis)
Simple Summary
In metastatic hormone-sensitive prostate cancer (mHSPC), prognostic stratification still relies on a limited set of clinical factors, mostly collected before treatment. With 7-month PSA emerging as a prognostic marker, our retrospective study aimed at investigating whether routinely collected laboratory and vital sign data, monitored during the same early treatment period, could identify significant toxicities; whether these toxicities could be predicted using baseline variables; and whether early variations in these parameters could be employed to improve outcome prediction. We confirmed, among 363 patients with de novo mHSPC, that we could automatically detect early adverse events from electronic medical records, that these correlated with survival outcomes in Cox analyses, and that their detection improved a machine-learning-based prediction of poor progression-free survival, beyond standard clinical prognostic factors. Early monitoring of routinely collected data can therefore serve as a tool for prognostic stratification in mHSPC.
Abstract
Background: Conventional prognostic factors are typically assessed at diagnosis in metastatic hormone-sensitive prostate cancer (mHSPC). However, variations in vital signs and laboratory parameters occur during systemic treatment and may predict patients’ prognosis and anticipate organ-specific toxicity development. Methods: This single-center retrospective study included 363 patients with de novo mHSPC treated between 2014 and 2023. Clinical and laboratory data were systematically collected from the hospital data warehouse, from treatment initiation through the following seven months. Variations in vital parameters and blood test results were graded using CTCAE V5.0 (dynamic variables). Cox regression analyses were performed to explore the impact of dynamic variables on progression-free survival (PFS) and overall survival (OS). Machine learning (ML) models (Support Vector Classifier, Random Forest, and LGBM Classifier) were developed to predict single organ-specific toxicities and to identify good and poor responders based on 7-month PSA levels, PFS and OS. We compared ML model performance when trained only on baseline factors (static models) with those integrating variables generated by vital sign and blood test monitoring within 3 and 7 months from treatment start (dynamic models). Results: Dynamic model failed to improve the prediction of single organ-specific toxicities. Univariable Cox analysis revealed that the development of hematological, liver, and kidney-related toxicity, as well as the development of electrolyte disturbances within 3 or 7 months, was associated with shorter PFS (p = 0.011, 0.007, 0.174, and 0.02, respectively) and/or OS (p = 0.001, 0.099, 0.012, and 0.001, respectively). In multivariable Cox analysis, increasing alkaline phosphatase levels (HR = 1.93, p = 0.009), decreasing albumin (HR = 1.92, p = 0.008) and development of hyponatremia (HR = 1.79, p = 0.033) were associated with a shorter OS. The combination of static and dynamic variables significantly improved the ability of ML models to identify poor responders (shorter PFS: AUC range 0.91–0.94 vs. 0.79–0.89). Conclusions: The integration of conventional prognostic factors with the detection of significant changes in vital signs and blood tests occurring early during systemic treatment in patients with de novo mHSPC may enhance patient stratification and improve prediction of survival outcomes. Multicenter validation studies are needed to confirm these results.
Keywords:
prostate cancer; hormone-sensitive; machine learning; prognostic; blood tests; toxicity; mHSPC; ARPI; adverse events
