[home] [Personal Program] [Help]

---

Session: Onco
Session starts: Friday 27 January, 11:30
Presentation starts: 12:15
Room: Room 558

---

Esther Visser (TU Eindhoven, Department of Biomedical Engineering; Catharina Ziekenhuis Eindhoven; Máxima Medisch Centrum Veldhoven)
Sylvia Genet (TU Eindhoven, Department of Biomedical Engineering; Catharina Ziekenhuis Eindhoven)
Remco de Kock (TU Eindhoven, Department of Biomedical Engineering; Catharina Ziekenhuis Eindhoven; Máxima Medisch Centrum Veldhoven)
Ben van den Borne (Catharina Ziekenhuis Eindhoven)
Maggy Youssef-El Soud (Máxima Medisch Centrum Veldhoven)
Huub Belderbos (Amphia Ziekenhuis Breda)
Gerben Stege (Anna Ziekenhuis Geldrop)
Marleen de Saegher (Sint Jans Gasthuis Weert)
Susan van 't Westeinde (Maasstad Ziekenhuis Rotterdam)
Luc Brunsveld (TU Eindhoven, Department of Biomedical Engineering)
Maarten Broeren (Máxima Medisch Centrum Veldhoven)
Daan van de Kerkhof (Catharina Ziekenhuis Eindhoven)
Birgit Deiman (Catharina Ziekenhuis Eindhoven)
Federica Eduati (TU Eindhoven, Department of Biomedical Engineering)
Volkher Scharnhorst (TU Eindhoven, Department of Biomedical Engineering; Catharina Ziekenhuis Eindhoven)

Abstract:
Optimal treatment decisions of lung cancer (LC) patients are based on histological and/or molecular profiling of the tumor. Nowadays, this information is retrieved by pathologic subtyping of tissue biopsies. However, tissue biopsies could be inadequate for analysis or unavailable in e.g. fragile patients. Diagnosis could be supported by measurement of protein tumor markers (TMs) and circulating tumor DNA (ctDNA) in minimally invasively obtained liquid biopsies, i.e. a venous blood draw. In this multicenter prospective study, we evaluate the performance of liquid-biopsy based decision-support algorithms for diagnosis of LC and identification of the two main histological subtypes small and non-small-cell lung cancer (SCLC and NSCLC). For 1096 patients with suspected LC, eight protein TMs (CA125, CA15.3, CEA, CYFRA 21-1, HE4, NSE, proGRP and SCCA) and ctDNA mutations in EGFR, KRAS and BRAF were analyzed in blood. Individual and combined TMs were used to train logistic regression models to identify LC, NSCLC or SCLC. To give better insight in the clinical applicability of the models, the performance of these models was evaluated at pre-specified positive predictive values (PPV) of ≥95% or ≥98%. For the combined TM models, only the most informative protein TMs selected by recursive feature elimination were included. The best performing individual TMs allowed for identification of LC, NSCLC and SCLC patients with 46%, 25% and 40% sensitivity, respectively, at pre-specified PPVs. Combining multiple protein TMs and ctDNA resulted in significantly increased sensitivities of 65%, 67% and 50%, respectively. The LC model could identify a high fraction of stage IV LC patients (80%), but also allowed for identification of earlier stage patients (23% stage I and 44% stage II). In conclusion, for a subset of patients suspected of LC, the diagnosis could be supported in a minimally invasive manner using liquid biopsy-based decision-support algorithms. In the future, these models may even help in the diagnosis of patients for whom pathologic subtyping is impossible or incomplete yet.