![Epidemiology and Biostatistics Department](../../assets/md5images/366950619e08c64786b4a19b498f8183.png)
Epidemiology and Biostatistics Publications
Prediction of Radiation Pneumonitis by Dose-volume Histogram Parameters in Lung Cancer--A Systematic Review
Document Type
Article
Publication Date
5-1-2004
Journal
Radiotherapy and Oncology
Volume
71
Issue
2
First Page
127
Last Page
138
URL with Digital Object Identifier
10.1016/j.radonc.2004.02.015
Abstract
BACKGROUND AND PURPOSE: To perform a systematic review of the predictive ability of various dose-volume histogram (DVH) parameters (V(dose), mean lung dose (MLD), and normal tissue complication probability (NTCP)) in the incidence of radiation pneumonitis (RP) caused by external-beam radiation therapy.
METHODS AND MATERIALS: Studies assessing the relationship between CT-based DVH reduction parameters and RP rate in radically treated lung cancer were eligible for the review. Synonyms for RP, lung cancer, DVH and its associated parameters (NTCP, V(20), V(30), MLD) were combined in a search strategy involving electronic databases, secondary reference searching, and consultation with experts. Individual or group data were abstracted from the various reports to calculate operating characteristics and odds ratios for the different DVH metrics.
RESULTS: A total of 12 published studies and two abstracts were identified. Eleven studies assessed V(dose), seven assessed MLD, and eight assessed NTCP. Nine studies exclusively analyzed the association between various DVH metrics and RP risk. Five studies also analyzed other patient, tumor, and treatment variables in conjunction with standard DVH metrics. A direct comparison between studies and the generation of summary statistics (i.e. meta-analysis) could not be achieved due to significant predictive and outcome variable heterogeneity. Most studies did show an association between DVH parameters and RP risk. However, overall accuracy, sensitivity, specificity, and positive predictive value were generally poor to fair for all three classes of DVH metrics.
CONCLUSIONS: An association between DVH parameters and RP risk has been demonstrated in the literature. However, the ideal DVH metric with excellent operating characteristics, either alone or in a model with other predictive variables, for RP risk prediction has not yet been identified. Several recommendations for reporting and conduct of future research into the association between DVH metrics and RP risk are provided.