A matter of timing: Identifying significant multi-dose radiotherapy improvements by numerical simulation and genetic algorithm search

Simon Angus, Monika Joanna Piotrowska

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Multi-dose radiotherapy protocols (fraction dose and timing) currently used in the clinic are the product of human selection based on habit, received wisdom, physician experience and intra-day patient timetabling. However, due to combinatorial considerations, the potential treatment protocol space for a given total dose or treatment length is enormous, even for relatively coarse search; well beyond the capacity of traditional in-vitro methods. In constrast, high fidelity numerical simulation of tumor development is well suited to the challenge. Building on our previous single-dose numerical simulation model of EMT6/Ro spheroids, a multi-dose irradiation response module is added and calibrated to the effective dose arising from 18 independent multi-dose treatment programs available in the experimental literature. With the developed model a constrained, non-linear, search for better performing candidate protocols is conducted within the vicinity of two benchmarks by genetic algorithm (GA) techniques. After evaluating less than 0.01 of the potential benchmark protocol space, candidate protocols were identified by the GA which conferred an average of 9.4 (max benefit 16.5 ) and 7.1 (13.3 ) improvement (reduction) on tumour cell count compared to the two benchmarks, respectively. Noticing that a convergent phenomenon of the top performing protocols was their temporal synchronicity, a further series of numerical experiments was conducted with periodic time-gap protocols (10 h to 23 h), leading to the discovery that the performance of the GA search candidates could be replicated by 17-18 h periodic candidates. Further dynamic irradiation-response cell-phase analysis revealed that such periodicity cohered with latent EMT6/Ro cell-phase temporal patterning. Taken together, this study provides powerful evidence towards the hypothesis that even simple inter-fraction timing variations for a given fractional dose program may present a facile, and highly cost-effective means of significantly improving clinical efficacy.
Original languageEnglish
Article numbere114098
Number of pages28
JournalPLoS ONE
Volume9
Issue number12
DOIs
Publication statusPublished - 2014

Cite this

@article{928bf1d9f65a47c9adb177771273141c,
title = "A matter of timing: Identifying significant multi-dose radiotherapy improvements by numerical simulation and genetic algorithm search",
abstract = "Multi-dose radiotherapy protocols (fraction dose and timing) currently used in the clinic are the product of human selection based on habit, received wisdom, physician experience and intra-day patient timetabling. However, due to combinatorial considerations, the potential treatment protocol space for a given total dose or treatment length is enormous, even for relatively coarse search; well beyond the capacity of traditional in-vitro methods. In constrast, high fidelity numerical simulation of tumor development is well suited to the challenge. Building on our previous single-dose numerical simulation model of EMT6/Ro spheroids, a multi-dose irradiation response module is added and calibrated to the effective dose arising from 18 independent multi-dose treatment programs available in the experimental literature. With the developed model a constrained, non-linear, search for better performing candidate protocols is conducted within the vicinity of two benchmarks by genetic algorithm (GA) techniques. After evaluating less than 0.01 of the potential benchmark protocol space, candidate protocols were identified by the GA which conferred an average of 9.4 (max benefit 16.5 ) and 7.1 (13.3 ) improvement (reduction) on tumour cell count compared to the two benchmarks, respectively. Noticing that a convergent phenomenon of the top performing protocols was their temporal synchronicity, a further series of numerical experiments was conducted with periodic time-gap protocols (10 h to 23 h), leading to the discovery that the performance of the GA search candidates could be replicated by 17-18 h periodic candidates. Further dynamic irradiation-response cell-phase analysis revealed that such periodicity cohered with latent EMT6/Ro cell-phase temporal patterning. Taken together, this study provides powerful evidence towards the hypothesis that even simple inter-fraction timing variations for a given fractional dose program may present a facile, and highly cost-effective means of significantly improving clinical efficacy.",
author = "Simon Angus and Piotrowska, {Monika Joanna}",
year = "2014",
doi = "10.1371/journal.pone.0114098",
language = "English",
volume = "9",
journal = "PLoS ONE",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "12",

}

A matter of timing: Identifying significant multi-dose radiotherapy improvements by numerical simulation and genetic algorithm search. / Angus, Simon; Piotrowska, Monika Joanna.

In: PLoS ONE, Vol. 9, No. 12, e114098, 2014.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - A matter of timing: Identifying significant multi-dose radiotherapy improvements by numerical simulation and genetic algorithm search

AU - Angus, Simon

AU - Piotrowska, Monika Joanna

PY - 2014

Y1 - 2014

N2 - Multi-dose radiotherapy protocols (fraction dose and timing) currently used in the clinic are the product of human selection based on habit, received wisdom, physician experience and intra-day patient timetabling. However, due to combinatorial considerations, the potential treatment protocol space for a given total dose or treatment length is enormous, even for relatively coarse search; well beyond the capacity of traditional in-vitro methods. In constrast, high fidelity numerical simulation of tumor development is well suited to the challenge. Building on our previous single-dose numerical simulation model of EMT6/Ro spheroids, a multi-dose irradiation response module is added and calibrated to the effective dose arising from 18 independent multi-dose treatment programs available in the experimental literature. With the developed model a constrained, non-linear, search for better performing candidate protocols is conducted within the vicinity of two benchmarks by genetic algorithm (GA) techniques. After evaluating less than 0.01 of the potential benchmark protocol space, candidate protocols were identified by the GA which conferred an average of 9.4 (max benefit 16.5 ) and 7.1 (13.3 ) improvement (reduction) on tumour cell count compared to the two benchmarks, respectively. Noticing that a convergent phenomenon of the top performing protocols was their temporal synchronicity, a further series of numerical experiments was conducted with periodic time-gap protocols (10 h to 23 h), leading to the discovery that the performance of the GA search candidates could be replicated by 17-18 h periodic candidates. Further dynamic irradiation-response cell-phase analysis revealed that such periodicity cohered with latent EMT6/Ro cell-phase temporal patterning. Taken together, this study provides powerful evidence towards the hypothesis that even simple inter-fraction timing variations for a given fractional dose program may present a facile, and highly cost-effective means of significantly improving clinical efficacy.

AB - Multi-dose radiotherapy protocols (fraction dose and timing) currently used in the clinic are the product of human selection based on habit, received wisdom, physician experience and intra-day patient timetabling. However, due to combinatorial considerations, the potential treatment protocol space for a given total dose or treatment length is enormous, even for relatively coarse search; well beyond the capacity of traditional in-vitro methods. In constrast, high fidelity numerical simulation of tumor development is well suited to the challenge. Building on our previous single-dose numerical simulation model of EMT6/Ro spheroids, a multi-dose irradiation response module is added and calibrated to the effective dose arising from 18 independent multi-dose treatment programs available in the experimental literature. With the developed model a constrained, non-linear, search for better performing candidate protocols is conducted within the vicinity of two benchmarks by genetic algorithm (GA) techniques. After evaluating less than 0.01 of the potential benchmark protocol space, candidate protocols were identified by the GA which conferred an average of 9.4 (max benefit 16.5 ) and 7.1 (13.3 ) improvement (reduction) on tumour cell count compared to the two benchmarks, respectively. Noticing that a convergent phenomenon of the top performing protocols was their temporal synchronicity, a further series of numerical experiments was conducted with periodic time-gap protocols (10 h to 23 h), leading to the discovery that the performance of the GA search candidates could be replicated by 17-18 h periodic candidates. Further dynamic irradiation-response cell-phase analysis revealed that such periodicity cohered with latent EMT6/Ro cell-phase temporal patterning. Taken together, this study provides powerful evidence towards the hypothesis that even simple inter-fraction timing variations for a given fractional dose program may present a facile, and highly cost-effective means of significantly improving clinical efficacy.

U2 - 10.1371/journal.pone.0114098

DO - 10.1371/journal.pone.0114098

M3 - Article

VL - 9

JO - PLoS ONE

JF - PLoS ONE

SN - 1932-6203

IS - 12

M1 - e114098

ER -