Abstract

Objective The goal of this study was to explore conceptual benefits of characterizing delineated target volumes based on surface area and to utilize the concept for assessing risk of therapeutic toxicity in radiosurgery. Methods and materials Four computer-generated targets, a sphere, a cylinder, an ellipsoid and a box, were designed for two distinct scenarios. In the first scenario, all targets had identical volumes, and in the second one, all targets had identical surface areas. High quality stereotactic radiosurgery plans with at least 95% target coverage and selectivity were created for each target in both scenarios. Normal brain volumes V12Gy, V14Gy and V16Gy corresponding to received dose of 12 Gy, 14 Gy and 16 Gy, respectively, were computed and analyzed. Additionally, V12Gy and V14Gy volumes and values for seven prospective toxicity variables were recorded for 100 meningioma patients after Gamma Knife radiosurgery. Multivariable stepwise linear regression and best subset linear regression analyses were performed in two statistical software packages, SAS/STAT and R, respectively. Results In a phantom study, for the constant volume targets, the volumes of 12 Gy, 14 Gy and 16 Gy isodose clouds were the lowest for the spherical target as an expected corollary of the isoperimetric inequality. For the constant surface area targets, a conventional wisdom is confirmed, as the target volume increases the corresponding volumes V12Gy, V14Gy and V16Gy also increase. In the 100-meningioma patient cohort, the best univariate model featured tumor surface area as the most significantly associated variable with both V12Gy and V14Gy volumes, corresponding to the adjusted R2 values of 0.82 and 0.77, respectively. Two statistical methods converged to matching multivariable models. Conclusions In a univariate model, target surface area is a better predictor of spilled dose to normal tissue than target largest dimension or target volume itself. In complex multivariate models, target surface area is an independent variable for modeling radiosurgical normal tissue toxicity risk.

Original languageEnglish (US)
Article numbere0224047
JournalPloS one
Volume14
Issue number10
DOIs
StatePublished - Jan 1 2019

Fingerprint

Radiosurgery
Toxicity
surface area
Meningioma
toxicity
Linear Models
Linear regression
Software
Regression Analysis
Tissue
Brain
Software packages
Tumors
Neoplasms
Statistical methods
knives
dosage
Therapeutics
statistical analysis
brain

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)
  • General

Cite this

@article{9662ec4fe50047b8941390ee6d969380,
title = "Considerations of target surface area and the risk of radiosurgical toxicity",
abstract = "Objective The goal of this study was to explore conceptual benefits of characterizing delineated target volumes based on surface area and to utilize the concept for assessing risk of therapeutic toxicity in radiosurgery. Methods and materials Four computer-generated targets, a sphere, a cylinder, an ellipsoid and a box, were designed for two distinct scenarios. In the first scenario, all targets had identical volumes, and in the second one, all targets had identical surface areas. High quality stereotactic radiosurgery plans with at least 95{\%} target coverage and selectivity were created for each target in both scenarios. Normal brain volumes V12Gy, V14Gy and V16Gy corresponding to received dose of 12 Gy, 14 Gy and 16 Gy, respectively, were computed and analyzed. Additionally, V12Gy and V14Gy volumes and values for seven prospective toxicity variables were recorded for 100 meningioma patients after Gamma Knife radiosurgery. Multivariable stepwise linear regression and best subset linear regression analyses were performed in two statistical software packages, SAS/STAT and R, respectively. Results In a phantom study, for the constant volume targets, the volumes of 12 Gy, 14 Gy and 16 Gy isodose clouds were the lowest for the spherical target as an expected corollary of the isoperimetric inequality. For the constant surface area targets, a conventional wisdom is confirmed, as the target volume increases the corresponding volumes V12Gy, V14Gy and V16Gy also increase. In the 100-meningioma patient cohort, the best univariate model featured tumor surface area as the most significantly associated variable with both V12Gy and V14Gy volumes, corresponding to the adjusted R2 values of 0.82 and 0.77, respectively. Two statistical methods converged to matching multivariable models. Conclusions In a univariate model, target surface area is a better predictor of spilled dose to normal tissue than target largest dimension or target volume itself. In complex multivariate models, target surface area is an independent variable for modeling radiosurgical normal tissue toxicity risk.",
author = "Strahinja Stojadinovic and Yulong Yan and Andrew Leiker and Chul Ahn and Zabi Wardak and Tu Dan and Lucien Nedzi and Robert Timmerman and Toral Patel and Samuel Barnett and Bruce Mickey and Jeffrey Meyer",
year = "2019",
month = "1",
day = "1",
doi = "10.1371/journal.pone.0224047",
language = "English (US)",
volume = "14",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "10",

}

TY - JOUR

T1 - Considerations of target surface area and the risk of radiosurgical toxicity

AU - Stojadinovic, Strahinja

AU - Yan, Yulong

AU - Leiker, Andrew

AU - Ahn, Chul

AU - Wardak, Zabi

AU - Dan, Tu

AU - Nedzi, Lucien

AU - Timmerman, Robert

AU - Patel, Toral

AU - Barnett, Samuel

AU - Mickey, Bruce

AU - Meyer, Jeffrey

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Objective The goal of this study was to explore conceptual benefits of characterizing delineated target volumes based on surface area and to utilize the concept for assessing risk of therapeutic toxicity in radiosurgery. Methods and materials Four computer-generated targets, a sphere, a cylinder, an ellipsoid and a box, were designed for two distinct scenarios. In the first scenario, all targets had identical volumes, and in the second one, all targets had identical surface areas. High quality stereotactic radiosurgery plans with at least 95% target coverage and selectivity were created for each target in both scenarios. Normal brain volumes V12Gy, V14Gy and V16Gy corresponding to received dose of 12 Gy, 14 Gy and 16 Gy, respectively, were computed and analyzed. Additionally, V12Gy and V14Gy volumes and values for seven prospective toxicity variables were recorded for 100 meningioma patients after Gamma Knife radiosurgery. Multivariable stepwise linear regression and best subset linear regression analyses were performed in two statistical software packages, SAS/STAT and R, respectively. Results In a phantom study, for the constant volume targets, the volumes of 12 Gy, 14 Gy and 16 Gy isodose clouds were the lowest for the spherical target as an expected corollary of the isoperimetric inequality. For the constant surface area targets, a conventional wisdom is confirmed, as the target volume increases the corresponding volumes V12Gy, V14Gy and V16Gy also increase. In the 100-meningioma patient cohort, the best univariate model featured tumor surface area as the most significantly associated variable with both V12Gy and V14Gy volumes, corresponding to the adjusted R2 values of 0.82 and 0.77, respectively. Two statistical methods converged to matching multivariable models. Conclusions In a univariate model, target surface area is a better predictor of spilled dose to normal tissue than target largest dimension or target volume itself. In complex multivariate models, target surface area is an independent variable for modeling radiosurgical normal tissue toxicity risk.

AB - Objective The goal of this study was to explore conceptual benefits of characterizing delineated target volumes based on surface area and to utilize the concept for assessing risk of therapeutic toxicity in radiosurgery. Methods and materials Four computer-generated targets, a sphere, a cylinder, an ellipsoid and a box, were designed for two distinct scenarios. In the first scenario, all targets had identical volumes, and in the second one, all targets had identical surface areas. High quality stereotactic radiosurgery plans with at least 95% target coverage and selectivity were created for each target in both scenarios. Normal brain volumes V12Gy, V14Gy and V16Gy corresponding to received dose of 12 Gy, 14 Gy and 16 Gy, respectively, were computed and analyzed. Additionally, V12Gy and V14Gy volumes and values for seven prospective toxicity variables were recorded for 100 meningioma patients after Gamma Knife radiosurgery. Multivariable stepwise linear regression and best subset linear regression analyses were performed in two statistical software packages, SAS/STAT and R, respectively. Results In a phantom study, for the constant volume targets, the volumes of 12 Gy, 14 Gy and 16 Gy isodose clouds were the lowest for the spherical target as an expected corollary of the isoperimetric inequality. For the constant surface area targets, a conventional wisdom is confirmed, as the target volume increases the corresponding volumes V12Gy, V14Gy and V16Gy also increase. In the 100-meningioma patient cohort, the best univariate model featured tumor surface area as the most significantly associated variable with both V12Gy and V14Gy volumes, corresponding to the adjusted R2 values of 0.82 and 0.77, respectively. Two statistical methods converged to matching multivariable models. Conclusions In a univariate model, target surface area is a better predictor of spilled dose to normal tissue than target largest dimension or target volume itself. In complex multivariate models, target surface area is an independent variable for modeling radiosurgical normal tissue toxicity risk.

UR - http://www.scopus.com/inward/record.url?scp=85073655314&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85073655314&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0224047

DO - 10.1371/journal.pone.0224047

M3 - Article

C2 - 31634366

AN - SCOPUS:85073655314

VL - 14

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 10

M1 - e0224047

ER -