Optimisation and validation of line-of-therapy advancement algorithms in advanced/metastatic non-small-cell lung cancer: an I-O Optimise study.
OpenAlex 토픽 ·
Lung Cancer Research Studies
Advanced Radiotherapy Techniques
Lung Cancer Diagnosis and Treatment
[BACKGROUND] Defining lines of therapy (LoTs) in real-world data is challenging, but it is essential for understanding the evolving treatment landscape for metastatic non-small-cell lung cancer (NSCLC
APA
S. Lay-Flurrie, A. Greystoke, et al. (2026). Optimisation and validation of line-of-therapy advancement algorithms in advanced/metastatic non-small-cell lung cancer: an I-O Optimise study.. ESMO real world data and digital oncology, 12, 100701. https://doi.org/10.1016/j.esmorw.2026.100701
MLA
S. Lay-Flurrie, et al.. "Optimisation and validation of line-of-therapy advancement algorithms in advanced/metastatic non-small-cell lung cancer: an I-O Optimise study.." ESMO real world data and digital oncology, vol. 12, 2026, pp. 100701.
PMID
42004489
Abstract
[BACKGROUND] Defining lines of therapy (LoTs) in real-world data is challenging, but it is essential for understanding the evolving treatment landscape for metastatic non-small-cell lung cancer (NSCLC). Validated algorithms are critical for determining LoTs; however, the performance of these algorithms across different data sources remains uncertain. This retrospective study aimed to optimise and evaluate the performance of several LoT algorithms for patients with metastatic NSCLC in Germany and the UK.
[MATERIALS AND METHODS] Six LoT algorithms were assessed in two real-world data sources: Frankfurt University Hospital, Germany, and the Real World Evidence Alliance at Leeds-Oncology, UK. The algorithms used various combinations of information on treatment (dispensation time, drug type, cycle) and progression of disease. Accuracy and positive predictive value were measured for each data source.
[RESULTS] Across both data sources, all algorithms showed similar accuracy. On average, the algorithms correctly identified the full treatment pathway for >75% of patients (≥90% excluding the worst-performing algorithm). The most accurate algorithms were time- and drug based (94% accuracy), and time- and drug-cycle based (95% accuracy).
[CONCLUSIONS] Algorithms using both drug type and timing information demonstrated high accuracy and consistent performance in defining LoTs. Adding disease progression information did not improve LoT definition. This study can help guide future real-world evidence generation.
[MATERIALS AND METHODS] Six LoT algorithms were assessed in two real-world data sources: Frankfurt University Hospital, Germany, and the Real World Evidence Alliance at Leeds-Oncology, UK. The algorithms used various combinations of information on treatment (dispensation time, drug type, cycle) and progression of disease. Accuracy and positive predictive value were measured for each data source.
[RESULTS] Across both data sources, all algorithms showed similar accuracy. On average, the algorithms correctly identified the full treatment pathway for >75% of patients (≥90% excluding the worst-performing algorithm). The most accurate algorithms were time- and drug based (94% accuracy), and time- and drug-cycle based (95% accuracy).
[CONCLUSIONS] Algorithms using both drug type and timing information demonstrated high accuracy and consistent performance in defining LoTs. Adding disease progression information did not improve LoT definition. This study can help guide future real-world evidence generation.