Abstract 183P
Background
We aimed to develop imaging biomarkers to predict radiation pneumonitis (RP) in non-small cell lung cancer (NSCLC) patients undergoing thoracic radiotherapy. We hypothesized that measuring morphometric complexity in the lung using simulation computed tomography (CT) may provide objective imaging biomarkers for lung parenchyma integrity, potentially forecasting the risk of RP.
Methods
A retrospective study on medical records of 175 patients diagnosed with NSCLC who had received thoracic radiotherapy. Three indices were utilized to measure the morphometric complexity of the lung parenchyma: box-counting fractal dimension, lacunarity, and minimum spanning tree (MST) fractal dimension. Patients were dichotomized into two groups at median values. Cox proportional hazard models were constructed to estimate the hazard ratios for grade ≥2 or grade ≥3 RP.
Results
We found significant associations between lung parenchymal morphometric complexity and RP incidence. In univariate Cox-proportional hazard analysis, patients with lower MST fractal dimension had a significantly higher hazard ratio of 2.296 (95% CI: 1.348–3.910) for grade ≥2 RP. When adjusted for age, sex, smoking status, histology, and DLCO, patients with a lower MST fractal dimension showed a significantly higher hazard ratio of 3.292 (95% CI: 1.722–6.294) for grade ≥2 RP and 7.952 (95% CI: 1.722 36.733) for grade ≥3 RP than those with a higher MST fractal dimension. Patients with lower lacunarity exhibited a significantly lower hazards ratio of 0.091 (95% CI: 0.015–0.573) for grade ≥3 RP in the adjusted model. We speculated that the lung tissue integrity is captured by morphometric complexity measures, particularly by the MST fractal dimension.
Conclusions
We showed a significant association between morphological complexity measures of NAA in simulated CT scans and RP incidence. Specifically, we suggest an MST fractal dimension as an imaging biomarker for predicting the occurrence of symptomatic RP after thoracic radiotherapy.
Legal entity responsible for the study
The author.
Funding
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2022R1A2C109170011.
Disclosure
The author has declared no conflicts of interest.