Predicting Survival in Lung Cancer with CT and Inflammation Data
Author Information
Author(s): Balbi Maurizio MD, Mazzaschi Giulia MD, Leo Ludovica PhD, Moron Dalla Tor Lucas PhD, Milanese Gianluca MD, PhD, Marrocchio Cristina MD, Silva Mario MD, PhD, Mura Rebecca MD, Favia Pasquale MD, Bocchialini Giovanni MD, Trentini Francesca MD, Minari Roberta MD, PhD, Ampollini Luca MD, PhD, Quaini Federico MD, Roti Giovanni MD, PhD, Tiseo Marcello MD, PhD, Sverzellati Nicola MD, PhD
Primary Institution: University Hospital of Parma, Italy
Hypothesis
Longitudinal changes in CT radiomic features and systemic inflammatory indices can better predict survival in advanced non-small cell lung cancer patients treated with immune checkpoint inhibitors than single-time-point assessments.
Conclusion
Longitudinal changes in radiomic features improved overall survival prediction in advanced lung cancer patients treated with immunotherapy.
Supporting Evidence
- Δ-RAD outperformed baseline-RAD for predicting overall survival.
- Integrating longitudinal changes of systemic inflammatory indices with clinical data led to the best model performance.
- Patients with high-risk PI had nearly sevenfold increased risk of death compared to low-risk patients.
Takeaway
Doctors can better predict how long lung cancer patients will live by looking at changes in their CT scans and blood tests over time instead of just one test.
Methodology
The study analyzed CT radiomic features and systemic inflammatory indices from stage IV NSCLC patients at two time points to develop predictive models for overall survival.
Potential Biases
The study's retrospective nature and variability in imaging acquisition may introduce bias.
Limitations
The sample size was small, and the study did not test the models in an independent population, which may affect generalizability.
Participant Demographics
Median age of participants was 70 years, with 70% male and a mix of current and former smokers.
Statistical Information
P-Value
<0.01
Confidence Interval
95% CI: 3.9 to 13.94
Statistical Significance
p<0.01
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website