A Guide to Outcome Modeling In Radiotherapy and Oncology. Listening to the Data

 

Description:

This book explores outcome modeling in cancer from a data-centric perspective to enable a better understanding of complex treatment response, to guide the design of advanced clinical trials, and to aid personalized patient care and improve their quality of life. It contains coverage of the relevant data sources available for model construction (panomics), ranging from clinical or preclinical resources to basic patient and treatment characteristics, medical imaging (radiomics), and molecular biological markers such as those involved in genomics, proteomics and metabolomics. It also includes discussions on the varying methodologies for predictive model building with analytical and data-driven approaches. This book is primarily intended to act as a tutorial for newcomers to the field of outcome modeling, as it includes in-depth how-to recipes on modeling artistry while providing sufficient instruction on how such models can approximate the physical and biological realities of clinical treatment. The book will also be of value to seasoned practitioners as a reference on the varying aspects of outcome modeling and their current applications. Features: Covers top-down approaches applying statistical, machine learning, and big data analytics and bottom-up approaches using first principles and multi-scale techniques, including numerical simulations based on Monte Carlo and automata techniques Provides an overview of the available software tools and resources for outcome model development and evaluation, and includes hands-on detailed examples throughout Presents a diverse selection of the common applications of outcome modeling in a wide variety of areas: treatment planning in radiotherapy, chemotherapy and immunotherapy, utility-based and biomarker applications, particle therapy modeling, oncological surgery, and the design of adaptive and SMART clinical trials

 

 

Table of Contents:

 

Section I Multiple Sources of Data

Chapter 1 ■ Introduction to data sources and outcome models

1.1 Introduction to Outcome Modeling

1.2 Model Definition

1.3 Types of Outcome Models

1.3.1 Prognostic versus predictive models

1.3.2 Top-down versus bottom-up models

1.3.3 Analytical versus data-driven models

1.4 Types of Data Used in Outcome Models

1.5 The Five Steps Towards Building an Outcome Model

1.6 Conclusions

Chapter 2 ■ Clinical data in outcome models

2.1 Introduction

2.2 Collagen Vascular Disease

2.3 Genetic Studies

2.4 Biological Factors Impacting Toxicity After Sbrt

2.4.1 Chest wall toxicity after SBRT

2.4.2 Radiation-induced lung toxicity (RILT) after SBRT

2.4.3 Radiation-induced liver damage (RILD) after SBRT

2.5 Big Data

2.6 Conclusions

Chapter 3 ■ Imaging data (radiomics)

3.1 Introduction

3.2 Image Features Extraction

3.2.1 Static image features

3.2.2 Dynamic image features

3.3 Radiomics Examples from Different Cancer Sites

3.3.1 Predicting local control in lung cancer using PET/CT

3.3.2 Predicting distant metastasis in sarcoma using PET/MR

3.4 Conclusions

Chapter 4 ■ Dosimetric data

4.1 Introduction

4.2 Dose Volume Metrics

4.3 Equivalent Uniform Dose

4.4 Dosimetric Model Variable Selection

4.4.1 Model order based on information theory

4.4.2 Model order based on resampling methods

4.5 A Dosimetric Modeling Example

4.5.1 Data set

4.5.2 Data exploration

4.5.3 Multivariate modeling with logistic regression

4.5.4 Multivariate modeling with machine learning

4.5.5 Comparison with other known models

4.6 Software Tools for Dosimetric Outcome Modeling

4.7 Conclusions

Chapter 5 ■ Pre-clinical radiobiological insights to inform modelling of radiotherapy outcome

5.1 Variability in Response to Highly Standardized Radiotherapy

5.2 Variation in Sensitivity to Radiation

5.3 Understanding Dose-Response of Tissues and Organs

5.4 Animal Models to Study Radiation Response

5.5 Processes Governing Outcome

5.6 Patient-Individual Factors / Co-Morbidity

5.7 Use in Models

5.8 Conclusion

Chapter 6 ■ Radiogenomics

6.1 Introduction

6.2 Biomarkers and the World of “-Omics”

6.2.1 Structural variations

6.2.1.1 Single nucleotide polymorphisms (SNPs)

6.2.1.2 Copy number variations (CNVs)

6.2.2 Gene expression: mRNA, miRNA, lncRNA

6.2.3 Protein expression

6.2.4 Metabolites

6.3 Resources for Biological Data

6.4 Examples of Radiogenomic Modeling

6.4.1 Prostate cancer

6.4.2 Breast cancer

6.4.3 Lung cancer

6.5 Conclusions

Section II Top-down Modeling Approaches

Chapter 7 ■ Analytical and mechanistic modeling

7.1 Introduction

7.2 Track Structure and DNA Damage

7.3 Linear-Quadratic Model

7.4 Kinetic Reaction Rate Models

7.4.1 Repair-misrepair and lethal-potentially-lethal models

7.4.2 Refined models

7.4.3 The Giant LOop Binary LEsion (GLOBE)

7.4.4 Local Effect Model (LEM)

7.4.5 Microdosimetric-kinetic model (MKM)

7.4.6 The Repair-misrepair-fixation model

7.5 Mechanistic Modeling of Stereotactic Radiosurgery (SRS) and Stereotactic Body Radiotherapy (SBRT)

7.5.1 LQ limitations and alternative models

7.6 Incorporating Biological Data to Describe and Predict Biological Response

7.7 Conclusions

Chapter 8 ■ Data driven approaches I: conventional statistical inference methods, including linear and logistic regression

8.1 What is a Regression

8.2 Linear Regression

8.2.1 Mathematical formalism

8.2.2 Estimation of regression coefficients

8.2.3 Accuracy of coefficient estimates

8.2.4 Rejecting the null hypothesis

8.2.5 Accuracy of the model

8.2.6 Qualitative predictors

8.2.7 Including interactions between variables

8.2.8 Linear regression: example

8.3 Logistic Regression

8.3.1 Modelling of qualitative (binary) response

8.3.2 Mathematical formalism

8.3.3 Estimation of regression coefficients

8.3.4 Accuracy of coefficient estimates

8.3.5 Rejecting the null hypothesis, testing the significance of a model

8.3.6 Accuracy of the model

8.3.7 Qualitative predictors

8.3.8 Including interaction between variables

8.3.9 Statistical power for reliable predictions

8.3.10 Time consideration

8.4 Model Validation

8.4.1 Apparent validation

8.4.2 Internal validation

8.4.3 External validation

8.5 Evaluation of an Extended Model

8.6 Feature Selection

8.6.1 Classical approaches

8.6.2 Shrinking and regularization methods: LASSO

8.6.3 Bootstrap methods

8.6.4 Logistic regression: example

8.7 Conclusions

Chapter 9 ■ Data driven approaches II: Machine Learning

9.1 Introduction

9.2 Feature Selection

9.2.1 Principal Component Analysis (PCA)

9.2.1.1 When should you use them?

9.2.1.2 Who has already used them?

9.3 Flavors of Machine Learning

9.3.1 Artificial Neural Networks

9.3.1.1 The basics

9.3.1.2 When should you use them?

9.3.1.3 Who has already used them?

9.3.2 Support Vector Machine

9.3.2.1 The basics

9.3.2.2 When should you use them?

9.3.2.3 Who has already used them?

9.3.3 Decision Trees and Random Forests

9.3.3.1 The basics

9.3.3.2 When should you use them?

9.3.3.3 Who has already used them?

9.3.4 Bayesian approaches

9.3.4.1 The basics

9.3.4.2 When should you use them?

9.3.4.3 Who has already used them?

9.4 Practical Implementation

9.4.1 The data

9.4.2 Model fitting and assessment

9.5 Conclusions

9.6 Resources

Section III Bottom-up Modeling Approaches

Chapter 10 ■ Stochastic multi-scale modeling of biological effects induced by ionizing radiation

10.1 Introduction

10.2 Particle Tracks: Physical Stage

10.3 Particle Tracks: Physico-Chemical and Chemical Stage

10.4 Multi-Scale DNA and Chromatin Models

10.5 Induction of DNA and Chromatin Damage

10.6 DNA Damage Response

10.7 Modeling Beyond Single-Cell Level

10.8 Conclusions

Chapter 11 ■ Multi-scale modeling approaches: application in chemo– and immuno–therapies

11.1 Introduction

11.2 Medical Oncology Treatments

11.2.1 From chemotherapy to molecular targeted agents

11.2.2 Immunotherapy

11.3 Modeling Types

11.3.1 Continuum tumor modeling

11.3.2 Discrete tumor modeling

11.3.3 Hybrid tumor modeling

11.4 Modeling Examples

11.4.1 Modeling of chemotherapy

11.4.2 Modeling of immunotherapy

11.5 Software Tools for Multi-Scale Modeling

11.6 Conclusions

Section IV Example Applications in Oncology

Chapter 12 ■ Outcome modeling in treatment planning

12.1 Introduction

12.1.1 Review of the history and dose-volume based treatment planning and its limitations

12.1.2 Emerging dose-response modeling in treatment planning and advantages

12.2 Dose-Response Models

12.2.1 Generalized equivalent uniform dose (gEUD)

12.2.1.1 Serial and parallel organ models

12.2.2 Linear-Quadratic (LQ) Model

12.2.3 Biological effective dose (BED)

12.2.4 Tumor control probability (TCP) models

12.2.5 Normal Tissue Complication Model (NTCP) models

12.2.5.1 Lyman-Kutcher-Burman (LKB) model

12.2.5.2 Relative seriality (RS) model

12.2.5.3 Model parameters and Quantitative Analysis of Normal Tissue Effects in the Clinic (QUANTEC)

12.2.6 Combined TCP/NTCP models –Uncomplicated tumor control model (UTCP or P+)

12.3 Dose-Response Models for Stereotactic Body Radiotherapy (SBRT)

12.3.1 Linear-Quadratic (LQ) model applied to SBRT

12.3.2 Universal survival curve (USC) model

12.3.3 Linear-Quadratic-Linear (LQL) model

12.3.4 Regrowth model

12.3.5 Dose limits for SBRT treatments

12.4 Biological Models in Treatment Planning

12.4.1 Plan evaluation

12.4.2 Plan optimization

12.4.3 Dose summation using biological models

12.4.4 Selection of outcome models and model parameters

12.5 Commercially Available Treatment Planning Systems (TPS) Employing Outcome Models

12.5.1 Elekta Monaco system (Maryland Heights, MO)

12.5.2 Philips Pinnacle system (Andover, MA)

12.5.2.1 Sensitivity of model parameters

12.5.3 Varian Eclipse system (Palo Alto, CA)

12.5.3.1 Objective functions in plan optimization

12.5.3.2 Plan evaluation

12.5.3.3 Sensitivity of model parameters

12.5.4 RaySearch RayStation (Stockholm, Sweden)

12.5.4.1 Plan evaluation tools

12.5.4.2 Plan optimization tools

12.5.5 MIM (MIM Software Inc., Cleveland, OH)

12.5.5.1 Plan summation

12.5.5.2 Plan evaluation

12.6 Conclusions

Chapter 13 ■ A utility based approach to individualized and adaptive radiation therapy

13.1 Introduction

13.2 Background

13.2.1 Treatment planning in radiation therapy

13.2.2 Biomarkers in RT

13.3 Utility Approach to Plan Optimization

13.3.1 In phase I trials

13.3.2 In RT treatment planning

13.3.3 Choice of the tradeoff parameter

13.3.4 Virtual clinical trial

13.4 Conclusions

Chapter 14 ■ Outcome modeling in Particle therapy

14.1 How are Particles Different from Photons?

14.2 Linear Energy Transfer (LET)

14.2.1 Dose averaging, track averaging and limitations

14.3 Relative Biological Effectiveness

14.3.1 The 1.1 conundrum in proton therapy

14.3.2 LET based RBE models

14.3.3 Non-LET based

14.3.3.1 Track structure (δ-ray) model

14.3.4 Uncertainties

14.4 The Role of Monte Carlo

14.4.1 Understanding dose and LET distributions

14.4.1.1 Range uncertainties

14.4.1.2 Considerations for dose and DVH

14.4.1.3 LET

14.4.2 RBE modeling

14.4.3 Example MC simulations using TOPAS

14.4.3.1 2-spot pencil setup

14.4.3.2 Expansion to include patient setup, dose, LET and one RBE scorer

14.5 Implications of Particle Therapy for Outcome Models

14.5.1 Target effects

14.5.2 Normal Tissue effects

14.6 Application in Treatment Planning

14.6.1 Vision for the future

Chapter 15 ■ Modeling response to oncological surgery

15.1 Introduction to Oncological Surgery

15.1.1 Clinical and surgical factors modifying patients’ outcomes

15.1.2 Complementary therapies to oncological surgery

15.2 Modeling of Oncological Surgery

15.2.1 Computational oncology models

15.2.2 Mechanistic models from physical oncology

15.2.2.1 Relevant variables

15.2.2.2 Implemented models

15.3 Example: A Bidimensional Oncological Surgery Simulation Model

15.3.1 Step 1: diffusion of nutrients

15.3.2 Step 2: CA rules and tumor growth constrained by the nutrients concentration and immune system response

15.3.3 Step 3: surgery

15.4 Discussion

15.5 Conclusions and Perspectives

15.6 Appendix 1: R Code

Chapter 16 ■ Tools for the precision medicine era: developing highly adaptive and personalized treatment recommendations using SMARTs

16.1 Introduction

16.2 Studying Treatments in Sequence

16.2.1 Adaptive treatment strategies

16.2.2 Decision rules

16.2.3 Tailoring variables are key for personalized recommendations

16.2.4 Machine learning “teaches” us the optimal ATS

16.3 Comparison to Traditional Methods

16.3.1 Why might RCTs fail to identify good treatment sequences?

16.3.2 Why can’t we combine results from separate, single-stage RCTs?

16.3.3 What are the advantages of SMARTs?

16.3.4 Motivating example

16.4 Validating a Proposed ATS

16.4.1 If we find an optimal ATS with a SMART, do we still need an RCT?

16.4.2 Are SMARTs used in cancer?

16.5 Challenges and Opportunities

Bibliography

Index

 

 

An aparitie	2020
Autor	Issam El Naqa
Dimensiuni	17.53 x 2.29 x 25.4 cm
Editura	CRC Press
Format	Softcover
ISBN	9780367572082
Limba	Engleza
Nr pag	392