PV QA 1 - Poster Viewing Q&A 1
Purpose/Objective(s): To determine the effect of proton vs. photon therapy and dose-volume factors on the increase in Child-Pugh (CP) scores for hepatobiliary malignancies, using a DVH estimation model.
Materials/Methods: We searched PubMed for eligible studies containing ≥20 patients treated either with SBRT or proton beam therapy (PBT) for hepatobiliary malignancies. Clinical outcome of interest was liver toxicity assessed by an increase in Child-Pugh (CP) score by 2+ points or decrease in CP class. Additionally, we developed a model based on in-house patients treated with SBRT and PBT, to derive an average DVH based on a given dosimetric parameter. This enables us to estimate dose-volume parameters that were not provided by the studies evaluated. Dose distributions for uninvolved liver were converted into EQD2 by the linear quadratic model (α/β=3 Gy). Linear regression analysis was performed on the probit-transformed outcome variable to correlate the percentage of CP score increase in each study to the predictor variables.
Results: We found 20 eligible studies, of these five pro- and 15 retrospective, 14 using SBRT (n=1114) and six using PBT (n=486). Fraction sizes ranged from 3-15Gy, and four studies reported CP score increase separately depending on patients’ initial CP class A and B. Results of univariate analyses reveal a significantly higher risk for an increase in CP score of 2+ points for patients with an initial CP class B compared to initial CP class A, and receiving SBRT compared to PBT (p= 0.012 and p=0.048, respectively). We evaluated the DVH-estimation model by testing it on data from external studies that provided MLD and at least one more dose-volume parameter. The model’s error estimation was determined for SBRT and PBT separately. Given the MLD as input, the estimated error of parameters derived from the model was 9% for SBRT and 7% for PBT, based on 4 studies each. Using the DVH-estimation model, regression testing for correlation of dosimetric parameters with increase in CP score reveals no significant result, but shows increasingly trending associations for the lower dose-volume parameters. Restricting the analysis to studies with fraction sizes of ≤8Gy per fraction (13 studies) further indicates that low dose volumetric factors correlate better with CP score increase.
Conclusion: Initial CP score is a powerful confounding factor that needs to be controlled for to allow valid analysis of the predictive value of dose-volume factors for models of liver toxicity. Our results indicate that PBT results in less CP score increase for similar MLD and that low dose parameters (V10-V20) might show better correlation with toxicity.
The asset you are trying to access is locked. Please enter your access key to unlock.