Introduction: Patient selection in lung volume reduction (LVR) plays a pivotal role in achieving meaningful clinical outcomes. Currently, LVR patients are selected based on three established criteria: heterogeneity index, percentage of low attenuation area (%LAA), and fissure integrity score. Quantitative computed tomography (QCT) has been developed to quantify lung physiological indices at the lobar level and could potentially revolutionise patient selection in LVR procedures. We developed an in-house QCT software, LungSeg, and used its radiological indices for the purposes of this thesis. The aim of this thesis is to discover potential physiological and radiological indices that could serve as predictors for superior LVR outcomes for better patient selection.
Methods: This thesis took two studies and analysed them using LungSeg. The first study was the long-term coil study, a randomised controlled study that had the control group crossing over to the treatment arm at 12 months. At 12 months post-procedure the baseline measurements were assessed against the 12-months post-procedural measurements. The second study was the short-term valve study which was another randomised controlled study that compared the primary and secondary endpoints between the control and the valve-treated group at three months post-procedure.
Results: In the long-term coil study, we found that the best statistically significant combination of predictors for change in target lobar volume at inspiration was found to be the combination of baseline target LV at inspiration, -950HU EI at inspiration, and TLCabs with a model adjusted R2 of 0.407 (p = 0.0001). In a subsequent multivariate analysis using ≥45% LAA on the -950HU at Inspiration, the R2 of the same prediction model did improve to 0.493 (P-value = 0.002). Meanwhile, the best statistically significant combination of predictors for change in target lobar volume at inspiration following valve treatment was found to be the combination of baseline target LV at inspiration, target lobar fissure integrity and baseline FEV1abs with a model adjusted R2 of 0.193 (p = 0.105).
Conclusion: Using QCT, we have improved the proposed patient selection algorithm for LVR procedures based on the best QCT and lung function predictors.