Tesi etd-10052025-221037
Link copiato negli appunti
Tipo di tesi
Corso Ordinario Ciclo Unico 6 Anni
Autore
TROMBELLO, LIDIA
URN
etd-10052025-221037
Titolo
DNADX ctDNA-Based Classification for Predicting Outcomes in HR+/HER2- Advanced Breast Cancer: Insights from the PARSIFAL Trial
Struttura
Classe Scienze Sperimentali
Corso di studi
SCIENZE MEDICHE - SCIENZE MEDICHE
Commissione
Tutor Prof. EMDIN, MICHELE
Relatore Prof. PRAT ALEIX
Relatore Dott.ssa BRASÓ MARISTANY FARA
Presidente Prof. PASSINO, CLAUDIO
Membro Prof. RECCHIA, FABIO ANASTASIO
Membro Dott.ssa durante, angela
Membro Dott.ssa TOGNINI, PAOLA
Relatore Prof. PRAT ALEIX
Relatore Dott.ssa BRASÓ MARISTANY FARA
Presidente Prof. PASSINO, CLAUDIO
Membro Prof. RECCHIA, FABIO ANASTASIO
Membro Dott.ssa durante, angela
Membro Dott.ssa TOGNINI, PAOLA
Parole chiave
- Breast cancer
- CDK4/6 inhibitors
- ctDNA
- DNADX
- endocrine therapy
- HR+/HER2-
- liquid biopsy
- machine learning
- molecular subtypes
- overall survival
- PARSIFAL trial
- precision oncology
- predictive biomarkers
- progression-free survival
- shallow whole genome sequencing
Data inizio appello
15/12/2025;
Disponibilità
parziale
Riassunto analitico
DNADX is a novel machine learning-based approach that uses tumor DNA or plasma ctDNA to identify clinically relevant phenotypic tumor features and classify breast cancer into four molecular subtypes. In advanced HR+/HER2- breast cancer, the identification of predictive biomarkers remains a major challenge, particularly following endocrine therapy and CDK4/6 inhibition. This thesis evaluates the prognostic and predictive value of ctDNA-based DNADX classification in patients with HR+/HER2- advanced breast cancer treated within the randomized phase II PARSIFAL trial, and explores its potential to guide the choice of endocrine therapy. Baseline plasma ctDNA samples from patients enrolled in the PARSIFAL study (NCT02491983) were analyzed using shallow whole genome sequencing. DNADX classified patients into four molecular subtypes (Clusters 1–4) when ctDNA tumor fraction (TF) was ≥3%. Progression-free survival (PFS) and overall survival (OS) were compared among DNADX subtypes using uni- and multivariable Cox regression analyses adjusted for clinical-pathological variables. Among 122 evaluable patients (25.1% of the total population), DNADX identified 56.6% with TF <3%, 14.8% with Cluster 1, 19.7% with Cluster 2, 5.7% with Cluster 3, and 3.3% with Cluster 4. Patients with TF <3% had significantly better PFS and OS compared to DNADX clusters (global log-rank p=0.010 and p=0.003, respectively). A numerical benefit from fulvestrant over letrozole was observed in Cluster 1 and Cluster 4 subgroups (interaction p=0.037). In conclusion, ctDNA-based DNADX subtyping predicts clinical outcomes in HR+/HER2- advanced breast cancer and may help identify the optimal endocrine therapy for each patient, highlighting the potential of integrating machine learning and liquid biopsy to refine treatment personalization in oncology.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|