Propensity Score Matched Analysis to Compare the Effectiveness of Proton Therapy Versus Photon-Based Radiation for Non-Small Cell Lung Cancer Patients Based on National Cancer Database Open Access

OConnell, Kelli A. (2016)

Permanent URL:


Introduction: Lung cancer is the leading cause of cancer death among adults in the United States. Theory suggests that proton beam therapy may provide benefits over conventional photon-based radiotherapies (external beam radiation, 3D Conformal radiotherapy, intensity-modulated radiation therapy, etc.) due to the lower dose of radiation delivered to healthy tissue. It is not known whether these dosimetric benefits translate into clinical benefits.

Methods: Cases with Stage I-IV non-small cell lung cancer (NSCLC) who received some form of external beam radiation directed to the lung or chest and who were not missing data on primary outcomes were extracted from the National Cancer Database (NCDB). Descriptive and univariate statistics were calculated for all variables. A multivariable Cox proportional hazards model was used to compare overall survival in two cohorts--one receiving proton therapy and one receiving photon-based radiation therapy--after controlling for socio-demographic, facility-level, and disease characteristics. Propensity score matching was performed to reduce confounding. A univariate Cox proportional hazards model was used to compare overall survival in the two matched cohorts. Stratified analyses were performed to investigate possible interaction between covariates and treatment group.

Results: There were 243,822 NSCLC (348 proton and 243,474 non-proton) cases included in the analysis. Patients who were treated at academic centers, in the west, on government insurance, and were wealthier were more likely to receive proton therapy (all p < 0.001). The multivariate Cox proportional hazards model suggested that non-proton patients were at a significantly higher risk of death compared to proton patients (HR = 1.21, p = 0.005). Propensity score matching yielded two cohorts of 308 patients each. A univariate Cox proportional hazards model comparing the two matched cohorts indicated to no significant differences in risk of death (HR = 1.16, p = 0.12). Stratified analyses suggested that proton therapy may be more beneficial for those who had more comorbidities, had adenocarcinoma histology, received chemotherapy, and were treated at academic centers.

Conclusion: Results suggest that there may be some clinical benefit to receiving proton therapy compared to conventional forms of radiation therapy. Randomized, controlled clinical trials are still needed to further confirm the advantages of proton therapy.

Table of Contents

Introduction. 1

Methods. 7

Data Source and Study Design. 7

Study Population. 8

Variables/Measurement 8

Statistical Analyses. 10

Results. 15

Descriptive Statistics. 15

Univariate Association with Reception of Proton Therapy. 16

Predicting Reception of Proton Therapy. 16

Univariate Association with Overall Survival 17

Multivariable Association with Overall Survival 18

Propensity Score Matching. 19

Overall Survival in Matched Sample. 19

Stratified Analyses. 21

Discussion. 21

References. 26

Appendix. 30

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
Subfield / Discipline
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Partnering Agencies
Last modified

Primary PDF

Supplemental Files