Comparison between model-predicted tumor oxygenation dynamics and vascular-/flow-related Doppler indices

Antonella Belfatto, Ailyn M.Vidal Urbinati, Delia Ciardo, Dorella Franchi, Federica Cattani, Roberta Lazzari, Barbara A. Jereczek-Fossa, Roberto Orecchia, Guido Baroni, Pietro Cerveri

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Purpose: Mathematical modeling is a powerful and flexible method to investigate complex phenomena. It discloses the possibility of reproducing expensive as well as invasive experiments in a safe environment with limited costs. This makes it suitable to mimic tumor evolution and response to radiotherapy although the reliability of the results remains an issue. Complexity reduction is therefore a critical aspect in order to be able to compare model outcomes to clinical data. Among the factors affecting treatment efficacy, tumor oxygenation is known to play a key role in radiotherapy response. In this work, we aim at relating the oxygenation dynamics, predicted by a macroscale model trained on tumor volumetric data of uterine cervical cancer patients, to vascularization and blood flux indices assessed on Ultrasound Doppler images. Methods: We propose a macroscale model of tumor evolution based on three dynamics, namely active portion, necrotic portion, and oxygenation. The model parameters were assessed on the volume size of seven cervical cancer patients administered with 28 fractions of intensity modulated radiation therapy (IMRT) (1.8 Gy/fraction). For each patient, five Doppler ultrasound tests were acquired before, during, and after the treatment. The lesion was manually contoured by an expert physician using 4D View (General Electric Company -Fairfield, Connecticut, United States), which automatically provided the overall tumor volume size along with three vascularization and/or blood flow indices. Volume data only were fed to the model for training purpose, while the predicted oxygenation was compared a posteriori to the measured Doppler indices. Results: The model was able to fit the tumor volume evolution within 8% error (range: 3-8%). A strong correlation between the intrapatient longitudinal indices from Doppler measurements and oxygen predicted by the model (about 90% or above) was found in three cases. Two patients showed an average correlation value (50-70%) and the remaining two presented poor correlations. The latter patients were the ones featuring the smallest tumor reduction throughout the treatment, typical of hypoxic conditions. Moreover, the average oxygenation value predicted by the model was close to the average vascularization-flow index (average difference: 7%). Conclusions: The results suggest that the modeled relation between tumor evolution and oxygen dynamics was reasonable enough to provide realistic oxygenation curves in five cases (correlation greater than 50%) out of seven. In case of nonresponsive tumors, the model failed in predicting the oxygenation trend while succeeded in reproducing the average oxygenation value according to the mean vascularization-flow index. Despite the need for deeper investigations, the outcomes of the present work support the hypothesis that a simple macroscale model of tumor response to radiotherapy is able to predict the tumor oxygenation. The possibility of an objective and quantitative validation on imaging data discloses the possibility to translate them as decision support tools in clinical practice and to move a step forward in the treatment personalization.

Original languageEnglish
Pages (from-to)2011-2019
Number of pages9
JournalMedical Physics
Volume44
Issue number5
DOIs
Publication statusPublished - Jan 1 2017

Fingerprint

Blood Vessels
Neoplasms
Radiotherapy
Doppler Ultrasonography
Tumor Burden
Uterine Cervical Neoplasms
Oxygen
Reproducibility of Results
Therapeutics
Physicians
Costs and Cost Analysis

Keywords

  • Cervical cancer
  • Doppler
  • LQ model
  • Mathematical model
  • Radiosensitivity
  • Radiotherapy
  • Ultrasound

ASJC Scopus subject areas

  • Biophysics
  • Radiology Nuclear Medicine and imaging

Cite this

Comparison between model-predicted tumor oxygenation dynamics and vascular-/flow-related Doppler indices. / Belfatto, Antonella; Urbinati, Ailyn M.Vidal; Ciardo, Delia; Franchi, Dorella; Cattani, Federica; Lazzari, Roberta; Jereczek-Fossa, Barbara A.; Orecchia, Roberto; Baroni, Guido; Cerveri, Pietro.

In: Medical Physics, Vol. 44, No. 5, 01.01.2017, p. 2011-2019.

Research output: Contribution to journalArticle

@article{59df372047694685a4f01d266f511124,
title = "Comparison between model-predicted tumor oxygenation dynamics and vascular-/flow-related Doppler indices",
abstract = "Purpose: Mathematical modeling is a powerful and flexible method to investigate complex phenomena. It discloses the possibility of reproducing expensive as well as invasive experiments in a safe environment with limited costs. This makes it suitable to mimic tumor evolution and response to radiotherapy although the reliability of the results remains an issue. Complexity reduction is therefore a critical aspect in order to be able to compare model outcomes to clinical data. Among the factors affecting treatment efficacy, tumor oxygenation is known to play a key role in radiotherapy response. In this work, we aim at relating the oxygenation dynamics, predicted by a macroscale model trained on tumor volumetric data of uterine cervical cancer patients, to vascularization and blood flux indices assessed on Ultrasound Doppler images. Methods: We propose a macroscale model of tumor evolution based on three dynamics, namely active portion, necrotic portion, and oxygenation. The model parameters were assessed on the volume size of seven cervical cancer patients administered with 28 fractions of intensity modulated radiation therapy (IMRT) (1.8 Gy/fraction). For each patient, five Doppler ultrasound tests were acquired before, during, and after the treatment. The lesion was manually contoured by an expert physician using 4D View (General Electric Company -Fairfield, Connecticut, United States), which automatically provided the overall tumor volume size along with three vascularization and/or blood flow indices. Volume data only were fed to the model for training purpose, while the predicted oxygenation was compared a posteriori to the measured Doppler indices. Results: The model was able to fit the tumor volume evolution within 8{\%} error (range: 3-8{\%}). A strong correlation between the intrapatient longitudinal indices from Doppler measurements and oxygen predicted by the model (about 90{\%} or above) was found in three cases. Two patients showed an average correlation value (50-70{\%}) and the remaining two presented poor correlations. The latter patients were the ones featuring the smallest tumor reduction throughout the treatment, typical of hypoxic conditions. Moreover, the average oxygenation value predicted by the model was close to the average vascularization-flow index (average difference: 7{\%}). Conclusions: The results suggest that the modeled relation between tumor evolution and oxygen dynamics was reasonable enough to provide realistic oxygenation curves in five cases (correlation greater than 50{\%}) out of seven. In case of nonresponsive tumors, the model failed in predicting the oxygenation trend while succeeded in reproducing the average oxygenation value according to the mean vascularization-flow index. Despite the need for deeper investigations, the outcomes of the present work support the hypothesis that a simple macroscale model of tumor response to radiotherapy is able to predict the tumor oxygenation. The possibility of an objective and quantitative validation on imaging data discloses the possibility to translate them as decision support tools in clinical practice and to move a step forward in the treatment personalization.",
keywords = "Cervical cancer, Doppler, LQ model, Mathematical model, Radiosensitivity, Radiotherapy, Ultrasound",
author = "Antonella Belfatto and Urbinati, {Ailyn M.Vidal} and Delia Ciardo and Dorella Franchi and Federica Cattani and Roberta Lazzari and Jereczek-Fossa, {Barbara A.} and Roberto Orecchia and Guido Baroni and Pietro Cerveri",
year = "2017",
month = "1",
day = "1",
doi = "10.1002/mp.12192",
language = "English",
volume = "44",
pages = "2011--2019",
journal = "Medical Physics",
issn = "0094-2405",
publisher = "John Wiley and Sons Ltd",
number = "5",

}

TY - JOUR

T1 - Comparison between model-predicted tumor oxygenation dynamics and vascular-/flow-related Doppler indices

AU - Belfatto, Antonella

AU - Urbinati, Ailyn M.Vidal

AU - Ciardo, Delia

AU - Franchi, Dorella

AU - Cattani, Federica

AU - Lazzari, Roberta

AU - Jereczek-Fossa, Barbara A.

AU - Orecchia, Roberto

AU - Baroni, Guido

AU - Cerveri, Pietro

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Purpose: Mathematical modeling is a powerful and flexible method to investigate complex phenomena. It discloses the possibility of reproducing expensive as well as invasive experiments in a safe environment with limited costs. This makes it suitable to mimic tumor evolution and response to radiotherapy although the reliability of the results remains an issue. Complexity reduction is therefore a critical aspect in order to be able to compare model outcomes to clinical data. Among the factors affecting treatment efficacy, tumor oxygenation is known to play a key role in radiotherapy response. In this work, we aim at relating the oxygenation dynamics, predicted by a macroscale model trained on tumor volumetric data of uterine cervical cancer patients, to vascularization and blood flux indices assessed on Ultrasound Doppler images. Methods: We propose a macroscale model of tumor evolution based on three dynamics, namely active portion, necrotic portion, and oxygenation. The model parameters were assessed on the volume size of seven cervical cancer patients administered with 28 fractions of intensity modulated radiation therapy (IMRT) (1.8 Gy/fraction). For each patient, five Doppler ultrasound tests were acquired before, during, and after the treatment. The lesion was manually contoured by an expert physician using 4D View (General Electric Company -Fairfield, Connecticut, United States), which automatically provided the overall tumor volume size along with three vascularization and/or blood flow indices. Volume data only were fed to the model for training purpose, while the predicted oxygenation was compared a posteriori to the measured Doppler indices. Results: The model was able to fit the tumor volume evolution within 8% error (range: 3-8%). A strong correlation between the intrapatient longitudinal indices from Doppler measurements and oxygen predicted by the model (about 90% or above) was found in three cases. Two patients showed an average correlation value (50-70%) and the remaining two presented poor correlations. The latter patients were the ones featuring the smallest tumor reduction throughout the treatment, typical of hypoxic conditions. Moreover, the average oxygenation value predicted by the model was close to the average vascularization-flow index (average difference: 7%). Conclusions: The results suggest that the modeled relation between tumor evolution and oxygen dynamics was reasonable enough to provide realistic oxygenation curves in five cases (correlation greater than 50%) out of seven. In case of nonresponsive tumors, the model failed in predicting the oxygenation trend while succeeded in reproducing the average oxygenation value according to the mean vascularization-flow index. Despite the need for deeper investigations, the outcomes of the present work support the hypothesis that a simple macroscale model of tumor response to radiotherapy is able to predict the tumor oxygenation. The possibility of an objective and quantitative validation on imaging data discloses the possibility to translate them as decision support tools in clinical practice and to move a step forward in the treatment personalization.

AB - Purpose: Mathematical modeling is a powerful and flexible method to investigate complex phenomena. It discloses the possibility of reproducing expensive as well as invasive experiments in a safe environment with limited costs. This makes it suitable to mimic tumor evolution and response to radiotherapy although the reliability of the results remains an issue. Complexity reduction is therefore a critical aspect in order to be able to compare model outcomes to clinical data. Among the factors affecting treatment efficacy, tumor oxygenation is known to play a key role in radiotherapy response. In this work, we aim at relating the oxygenation dynamics, predicted by a macroscale model trained on tumor volumetric data of uterine cervical cancer patients, to vascularization and blood flux indices assessed on Ultrasound Doppler images. Methods: We propose a macroscale model of tumor evolution based on three dynamics, namely active portion, necrotic portion, and oxygenation. The model parameters were assessed on the volume size of seven cervical cancer patients administered with 28 fractions of intensity modulated radiation therapy (IMRT) (1.8 Gy/fraction). For each patient, five Doppler ultrasound tests were acquired before, during, and after the treatment. The lesion was manually contoured by an expert physician using 4D View (General Electric Company -Fairfield, Connecticut, United States), which automatically provided the overall tumor volume size along with three vascularization and/or blood flow indices. Volume data only were fed to the model for training purpose, while the predicted oxygenation was compared a posteriori to the measured Doppler indices. Results: The model was able to fit the tumor volume evolution within 8% error (range: 3-8%). A strong correlation between the intrapatient longitudinal indices from Doppler measurements and oxygen predicted by the model (about 90% or above) was found in three cases. Two patients showed an average correlation value (50-70%) and the remaining two presented poor correlations. The latter patients were the ones featuring the smallest tumor reduction throughout the treatment, typical of hypoxic conditions. Moreover, the average oxygenation value predicted by the model was close to the average vascularization-flow index (average difference: 7%). Conclusions: The results suggest that the modeled relation between tumor evolution and oxygen dynamics was reasonable enough to provide realistic oxygenation curves in five cases (correlation greater than 50%) out of seven. In case of nonresponsive tumors, the model failed in predicting the oxygenation trend while succeeded in reproducing the average oxygenation value according to the mean vascularization-flow index. Despite the need for deeper investigations, the outcomes of the present work support the hypothesis that a simple macroscale model of tumor response to radiotherapy is able to predict the tumor oxygenation. The possibility of an objective and quantitative validation on imaging data discloses the possibility to translate them as decision support tools in clinical practice and to move a step forward in the treatment personalization.

KW - Cervical cancer

KW - Doppler

KW - LQ model

KW - Mathematical model

KW - Radiosensitivity

KW - Radiotherapy

KW - Ultrasound

UR - http://www.scopus.com/inward/record.url?scp=85032511110&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85032511110&partnerID=8YFLogxK

U2 - 10.1002/mp.12192

DO - 10.1002/mp.12192

M3 - Article

AN - SCOPUS:85032511110

VL - 44

SP - 2011

EP - 2019

JO - Medical Physics

JF - Medical Physics

SN - 0094-2405

IS - 5

ER -