Multivariable modeling of radiotherapy outcomes, including dose–volume and clinical factors

A dosiomics approach to treatment outcome modeling in carbon ion radiotherapy for skull base chordomas

PURPOSE

To investigate the role of dosiomics features extracted from physical dose (DPHYS), RBE-weighted dose (DRBE) and dose-averaged Linear Energy Transfer (LETd), to predict the risk of local recurrence (LR) in skull base chordoma (SBC) treated with Carbon Ion Radiotherapy (CIRT). Thus, define and evaluate dosiomics-driven tumor control probability (TCP) models.

MATERIALS AND METHODS

54 SBC patients were retrospectively selected for this study. A regularized Cox proportional hazard model (r-Cox) and Survival Support Vector Machine (s-SVM) were tuned within a repeated Cross Validation (CV) and patients were stratified in low/high risk of LR. Models' performance was evaluated through Harrell's concordance statistic (C-index), and survival was represented through Kaplan-Meier (KM) curves. A multivariable logistic regression was fit to the selected feature sets to generate a dosiomics-driven TCP model for each map. These were compared to a reference model built with clinical parameters in terms of f-score and accuracy.

RESULTS

The LETd maps reached a test C-index of 0.750 and 0.786 with r-Cox and s-SVM, and significantly separated KM curves. DPHYS maps and clinical parameters showed promising CV outcomes with C-index above 0.8, despite a poorer performance on the test set and patients stratification. The LETd-based TCP showed a significatively higher f-score (0.67[0.52-0.70], median[IQR]) compared to the clinical model (0.4[0.32-0.63], p < 0.025), while DPHYS achieved a significatively higher accuracy (DPHYS: 0.73[0.65-0.79], Clinical: 0.6 [0.52-0.72]).

CONCLUSION

This analysis supports the role of LETd as relevant source of prognostic factors for LR in SBC treated with CIRT. This is reflected in the TCP modeling, where LETd and DPHYS showed an improved performance with respect to clinical models.

NRG Oncology White Paper on the Relative Biological Effectiveness in Proton Therapy

This position paper, led by the NRG Oncology Particle Therapy Work Group, focuses on the concept of relative biologic effect (RBE) in clinical proton therapy (PT), with the goal of providing recommendations for the next-generation clinical trials with PT on the best practice of investigating and using RBE, which could deviate from the current standard proton RBE value of 1.1 relative to photons. In part 1, current clinical utilization and practice are reviewed, giving the context and history of RBE. Evidence for variation in RBE is presented along with the concept of linear energy transfer (LET). The intertwined nature of tumor radiobiology, normal tissue constraints, and treatment planning with LET and RBE considerations is then reviewed. Part 2 summarizes current and past clinical data and then suggests the next steps to explore and employ tools for improved dynamic models for RBE. In part 3, approaches and methods for the next generation of prospective clinical trials are explored, with the goal of optimizing RBE to be both more reflective of clinical reality and also deployable in trials to allow clinical validation and interpatient comparisons. These concepts provide the foundation for personalized biologic treatments reviewed in part 4. Finally, we conclude with a summary including short- and long-term scientific focus points for clinical PT. The practicalities and capacity to use RBE in treatment planning are reviewed and considered with more biological data in hand. The intermediate step of LET optimization is summarized and proposed as a potential bridge to the ultimate goal of case-specific RBE planning that can be achieved as a hypothesis-generating tool in near-term proton trials.

Normal tissue complication probability models of hypothyroidism after radiotherapy for breast cancer

Purpose

We aimed to develop Lyman-Kutcher-Burman (LKB) and multivariable normal tissue complication probability (NTCP) models to predict the risk of radiation-induced hypothyroidism (RIHT) in breast cancer patients.

Materials and methods

A total of 1,063 breast cancer patients who underwent whole breast irradiation between 2009 and 2016 were analyzed. Individual dose-volume histograms were used to generate LKB and multivariable logistic regression models. LKB model was fit using the thyroid radiation dose-volume parameters. A multivariable model was constructed to identify potential dosimetric and clinical parameters associated with RIHT. Internal validation was conducted using bootstrapping techniques, and model performance was evaluated using the area under the curve (AUC) and Hosmer-Lemeshow (HL) goodness-of-fit test.

Results

RIHT developed in 4 % of patients with a median follow-up of 77.7 months. LKB and multivariable NTCP models exhibited significant agreement between the predicted and observed results (HL P values > 0.05). The multivariable NTCP model outperformed the LKB model in predicting RIHT (AUC 0.62 vs. 0.54). In the multivariable model, systemic therapy, age, and percentage of thyroid volume receiving ≥ 10 Gy (V10) were significant prognostic factors for RIHT. The cumulative incidence of RIHT was significantly higher in patients who exceeded the cut-off values for all three risk predictors (systemic therapy, age ≥ 40 years, and thyroid V10 ≥ 26 %, P < 0.005).

Conclusions

Systemic therapy, age, and V10 of the thyroid were identified as strong risk factors for the development of RIHT. Our NTCP models provide valuable insights to clinicians for predicting and preventing hypothyroidism by identifying high-risk patients.

Comparing Performances of Predictive Models of Toxicity after Radiotherapy for Breast Cancer Using Different Machine Learning Approaches

PURPOSE

Different ML models were compared to predict toxicity in RT on a large cohort (n = 1314).

METHODS

The endpoint was RTOG G2/G3 acute toxicity, resulting in 204/1314 patients with the event. The dataset, including 25 clinical, anatomical, and dosimetric features, was split into 984 for training and 330 for internal tests. The dataset was standardized; features with a high p-value at univariate LR and with Spearman ρ>0.8 were excluded; synthesized data of the minority were generated to compensate for class imbalance. Twelve ML methods were considered. Model optimization and sequential backward selection were run to choose the best models with a parsimonious feature number. Finally, feature importance was derived for every model.

RESULTS

The model's performance was compared on a training-test dataset over different metrics: the best performance model was LightGBM. Logistic regression with three variables (LR3) selected via bootstrapping showed performances similar to the best-performing models. The AUC of test data is slightly above 0.65 for the best models (highest value: 0.662 with LightGBM).

CONCLUSIONS

No model performed the best for all metrics: more complex ML models had better performances; however, models with just three features showed performances comparable to the best models using many (n = 13-19) features.

Fine Motor Skill Decline After Brain Radiation Therapy-A Multivariate Normal Tissue Complication Probability Study of a Prospective Trial

PURPOSE

Brain radiation therapy can impair fine motor skills (FMS). Fine motor skills are essential for activities of daily living, enabling hand-eye coordination for manipulative movements. We developed normal tissue complication probability (NTCP) models for the decline in FMS after fractionated brain radiation therapy (RT).

METHODS AND MATERIALS

On a prospective trial, 44 patients with primary brain tumors received fractioned RT; underwent high-resolution volumetric magnetic resonance imaging, diffusion tensor imaging, and comprehensive FMS assessments (Delis-Kaplan Executive Function System Trail Making Test Motor Speed [DKEFS-MS]; and Grooved Pegboard dominant/nondominant hands) at baseline and 6 months postRT. Regions of interest subserving motor function (including cortex, superficial white matter, thalamus, basal ganglia, cerebellum, and white matter tracts) were autosegmented using validated methods and manually verified. Dosimetric and clinical variables were included in multivariate NTCP models using automated bootstrapped logistic regression, least absolute shrinkage and selection operator logistic regression, and random forests with nested cross-validation.

RESULTS

Half of the patients showed a decline on grooved pegboard test of nondominant hands, 17 of 42 (40.4%) on grooved pegboard test of -dominant hands, and 11 of 44 (25%) on DKEFS-MS. Automated bootstrapped logistic regression selected a 1-term model including maximum dose to dominant postcentral white matter. The least absolute shrinkage and selection operator logistic regression selected this term and steroid use. The top 5 variables in the random forest were all dosimetric: maximum dose to dominant thalamus, mean dose to dominant caudate, mean and maximum dose to the dominant corticospinal tract, and maximum dose to dominant postcentral white matter. This technique performed best with an area under the curve of 0.69 (95% CI, 0.68-0.70) on nested cross-validation.

CONCLUSIONS

We present the first NTCP models for FMS impairment after brain RT. Dose to several supratentorial motor-associated regions of interest correlated with a decline in dominant-hand fine motor dexterity in patients with primary brain tumors in multivariate models, outperforming clinical variables. These data can guide prospective fine motor-sparing strategies for brain RT.

Radiation-induced lung injury after breast cancer treatment: incidence in the CANTO-RT cohort and associated clinical and dosimetric risk factors

Purpose

Radiation-induced lung injury (RILI) is strongly associated with various clinical conditions and dosimetric parameters. Former studies have led to reducing radiotherapy (RT) doses to the lung and have favored the discontinuation of tamoxifen during RT. However, the monocentric design and variability of dosimetric parameters chosen have limited further improvement. The aim of our study was to assess the incidence of RILI in current practice and to determine clinical and dosimetric risk factors associated with RILI occurrence.

Material and methods

Data from 3 out of the 10 top recruiting centers in CANTO-RT, a subset of the CANTO prospective longitudinal cohort (NCT01993498), were retrospectively analyzed for RILI occurrence. This cohort, which recruited invasive cT0-3 cN0-3 M0 breast cancer patients from 2012 to 2018, prospectively recorded the occurrence of adverse events by questionnaires and medical visits at the end of, and up to 60 months after treatment. RILI adverse events were defined in all patients by the association of clinical symptoms and compatible medical imaging.

Results

RILI was found in 38/1565 (2.4%) patients. Grade II RILI represented 15/38 events (39%) and grade III or IV 2/38 events (6%). There were no grade V events. The most frequently used technique for treatment was 3D conformational RT (96%). In univariable analyses, we confirmed the association of RILI occurrence with pulmonary medical history, absence of cardiovascular disease medical history, high pT and pN, chemotherapy use, nodal RT. All dosimetric parameters were highly correlated and had close predictive value. In the multivariable analysis adjusted for chemotherapy use and nodal involvement, pulmonary medical history (OR=3.05, p<0.01) and high V30 Gy (OR=1.06, p=0.04) remained statistically significant risk factors for RILI occurrence. V30 Gy >15% was significantly associated with RILI occurrence in a multivariable analysis (OR=3.07, p=0.03).

Conclusion

Our study confirms the pulmonary safety of breast 3D RT in CANTO-RT. Further analyses with modern radiation therapy techniques such as IMRT are needed. Our results argue in favor of a dose constraint to the ipsilateral lung using V30 Gy not exceeding 15%, especially in patients presenting pulmonary medical history. Pulmonary disease records should be taken into account for RT planning.

Parameters of the Lyman Model for Calculation of Normal-Tissue Complication Probability: A Systematic Literature Review

PURPOSE

The Lyman model is one of the most used radiobiological models for calculation of normal-tissue complication probability (NTCP). Since its introduction in 1985, many authors have published parameter values for the model based on clinical data of different radiotherapeutic situations. This study attempted to collect the entirety of radiobiological parameter sets published to date and provide an overview of the data basis for different variations of the model. Furthermore, it sought to compare the parameter values and calculated NTCPs for selected endpoints with sufficient data available.

METHODS AND MATERIALS

A systematic literature analysis was performed, searching for publications that provided parameters for the different variations of the Lyman model in the Medline database using PubMed. Parameter sets were grouped into 13 toxicity-related endpoint groups. For 3 selected endpoint groups (≤25% reduction of saliva 12 months after irradiation of the parotid, symptomatic pneumonitis after irradiation of the lung, and bleeding of grade 2 or less after irradiation of the rectum), parameter values were compared and differences in calculated NTCP values were analyzed.

RESULTS

A total of 509 parameter sets from 130 publications were identified. Considerable heterogeneities were detected regarding the number of parameters available for different radio-oncological situations. Furthermore, for the 3 selected endpoints, large differences in published parameter values were found. These translated into great variations of calculated NTCPs, with maximum ranges of 35.2% to 93.4% for the saliva endpoint, of 39.4% to 90.4% for the pneumonitis endpoint, and of 5.4% to 99.3% for the rectal bleeding endpoint.

CONCLUSIONS

The detected heterogeneity of the data as well as the large variations of published radiobiological parameters underline the necessity for careful interpretation when using such parameters for NTCP calculations. Appropriate selection of parameters and validation of values are essential when using the Lyman model.

Cardiac Substructures Dosimetric Predictors for Cardiac Toxicity After Definitive Radiotherapy in Esophageal Cancer

PURPOSE

To investigate the predictive value of the cardiac substructures (CSs) dosimetric parameters for cardiac toxicity after definitive radiation therapy in locally advanced esophageal cancer.

METHODS AND MATERIALS

Between August 2010 and January 2016, 716 patients with stage 2-3 esophageal cancer receiving definitive radiation therapy at 2 institutions were divided into training (n = 432) and external validation (n = 284) cohorts. Dose-volume histogram parameters for the whole heart (WH) and CSs were extracted. Competing risks and Cox regressions analyses were performed. The predictive performance of the models was evaluated using the area under the receiver operating characteristic curve (AUC) and the Brier score.

RESULTS

With a median follow-up of 93 months, 68 patients (15.7%) developed grade ≥3 cardiac events (G3+ CEs), with a median of 13.5 months to the first event. Multivariable analysis showed left ventricle, left anterior descending (LAD), and mean left circumflex (LCX) variables were significantly associated with G3+ CEs. The AUCs and Brier scores demonstrated favorable predictive accuracies of the models integrating these CS variables when predicting G3+ CEs in the training and validation cohorts. However, compared with the WH variables, the CS variables did not significantly improve the prediction of G3+ CEs. Nevertheless, when G3+ acute coronary syndrome and/or congestive heart failure (ACS/CHF) CE was the outcome of interest, models based on the LAD or LCX variables were superior to the WH variable models in training and validation cohorts.

CONCLUSIONS

Models based on CS variables showed favorable predictive accuracy for G3+ CEs. The LAD and LCX variables significantly improved the prediction of G3+ ACS/CHF events compared with the WH variables. Radiation doses to CSs, such as LCX and LAD, should be monitored to help reduce the occurrence of significant CEs in patients with esophageal cancer undergoing definitive radiation therapy.

Clinical Radiobiology for Radiation Oncology

This chapter is focused on radiobiological aspects at the molecular, cellular, and tissue level which are relevant for the clinical use of ionizing radiation (IR) in cancer therapy. For radiation oncology, it is critical to find a balance, i.e., the therapeutic window, between the probability of tumor control and the probability of side effects caused by radiation injury to the healthy tissues and organs. An overview is given about modern precision radiotherapy (RT) techniques, which allow optimal sparing of healthy tissues. Biological factors determining the width of the therapeutic window are explained. The role of the six typical radiobiological phenomena determining the response of both malignant and normal tissues in the clinic, the 6R’s, which are Reoxygenation, Redistribution, Repopulation, Repair, Radiosensitivity, and Reactivation of the immune system, is discussed. Information is provided on tumor characteristics, for example, tumor type, growth kinetics, hypoxia, aberrant molecular signaling pathways, cancer stem cells and their impact on the response to RT. The role of the tumor microenvironment and microbiota is described and the effects of radiation on the immune system including the abscopal effect phenomenon are outlined. A summary is given on tumor diagnosis, response prediction via biomarkers, genetics, and radiomics, and ways to selectively enhance the RT response in tumors. Furthermore, we describe acute and late normal tissue reactions following exposure to radiation: cellular aspects, tissue kinetics, latency periods, permanent or transient injury, and histopathology. Details are also given on the differential effect on tumor and late responding healthy tissues following fractionated and low dose rate irradiation as well as the effect of whole-body exposure.

Longitudinal study of irradiation-induced brain functional network alterations in patients with nasopharyngeal carcinoma

BACKGROUND

To investigate radiotherapy (RT)-related brain network changes in patients with nasopharyngeal carcinoma (NPC) over time and develop least absolute shrinkage and selection operator (LASSO)-based multivariable normal tissue complication probability (NTCP) models to predict RT-related brain network changes.

METHODS

36 NPC patients were followed up at four timepoints: baseline, within 3 months (acute), 6 months (subacute), and 12 months (delayed) post-RT. 15 comparable healthy controls (HCs) were finally included and followed up in parallel. Functional neuroimaging data, dose-volume parameters of bilateral temporal lobes and Montreal Cognitive Assessment (MoCA) were acquired. Graph theoretical analysis and mixed-design analysis of variance were performed to investigate how the brain global and nodal changes were affected by RT. Multivariate logistic regression NTCP models were developed. LASSO with nested cross-validation strategy was used to select features. The relationships between network changes and MoCA changes were also examined.

RESULTS

Significant changes were detected in nodal efficiency (NE) in NPC patients but not in HCs over time. Altered NE was distributed in the bilateral frontal, temporal lobes and the right insula, which showed a "decrease-increase/recovery" pattern over time. Among all models, the model for predicting NE changes of STG.R showed a relatively good performance (area under the receiver operating curve: 0.68), and D_20cc and V₂₀ to right temporal lobe outperformed in this model.

CONCLUSION

Our findings indicate that RT-induced brain injury begin at the acute period and follow a recovery over time. Furthermore, our study presents prediction models for brain dysfunction based on the dosimetric and clinical parameters.

Predictive performance of different NTCP techniques for radiation-induced esophagitis in NSCLC patients receiving proton radiotherapy

This study aimed to compare the predictive performance of different modeling methods in developing normal tissue complication probability (NTCP) models for predicting radiation-induced esophagitis (RE) in non–small cell lung cancer (NSCLC) patients receiving proton radiotherapy. The dataset was composed of 328 NSCLC patients receiving passive-scattering proton therapy and 41.6% of the patients experienced ≥ grade 2 RE. Five modeling methods were used to build NTCP models: standard Lyman–Kutcher–Burman (sLKB), generalized LKB (gLKB), multivariable logistic regression using two variable selection procedures-stepwise forward selection (Stepwise-MLR), and least absolute shrinkage and selection operator (LASSO-MLR), and support vector machines (SVM). Predictive performance was internally validated by a bootstrap approach for each modeling method. The overall performance, discriminative ability, and calibration were assessed using the Negelkerke R², area under the receiver operator curve (AUC), and Hosmer–Lemeshow test, respectively. The LASSO-MLR model showed the best discriminative ability with an AUC value of 0.799 (95% confidence interval (CI): 0.763–0.854), and the best overall performance with a Negelkerke R² value of 0.332 (95% CI: 0.266–0.486). Both of the optimism-corrected Negelkerke R² values of the SVM and sLKB models were 0.301. The optimism-corrected AUC of the gLKB model (0.796) was higher than that of the SVM model (0.784). The sLKB model had the smallest optimism in the model variation and discriminative ability. In the context of classification and probability estimation for predicting the NTCP for radiation-induced esophagitis, the MLR model developed with LASSO provided the best predictive results. The simplest LKB modeling had similar or even better predictive performance than the most complex SVM modeling, and it was least likely to overfit the training data. The advanced machine learning approach might have limited applicability in clinical settings with a relatively small amount of data.

Assessment of Radiation-Induced Optic Neuropathy in a Multi-Institutional Cohort of Chordoma and Chondrosarcoma Patients Treated with Proton Therapy

Simple Summary

Proton therapy is an effective therapeutic option for the treatment of skull-base tumors that require high radiation doses to be controlled. On rare occasions, patients suffer from radiation-induced optic neuropathy (RION) to the detriment of their post-treatment quality-of-life. We have collected multi-institutional data of 289 skull-base patients having received high doses to the optic apparatus from proton therapy or proton–photon mixed treatments and have observed a RION incidence rate (all grades) of 4.2% (12). We have furthermore confirmed older age and hypertension as risk factors for the onset of this side effect, with tumor involvement or its proximity to the optic apparatus and repeated surgical procedures showing moderate association. Our findings were consolidated into a NTCP model that can support pre-treatment patient segmentation into risk groups and the planning of necessary treatment countermeasures. However, further data and validation are necessary to confirm validity of the model.

Abstract

Radiation-induced optic neuropathy (RION) is a rare side effect following radiation therapy involving the optic structures whose onset is, due to the low amount of available data, challenging to predict. We have analyzed a multi-institutional cohort including 289 skull-base cancer patients treated with proton therapy who all received >45 Gy_RBE to the optic apparatus. An overall incidence rate of 4.2% (12) was observed, with chordoma patients being at higher risk (5.8%) than chondrosarcoma patients (3.2%). Older age and arterial hypertension, tumor involvement, and repeated surgeries (>3) were found to be associated with RION. Based on bootstrapping and cross-validation, a NTCP model based on age and hypertension was determined to be the most robust, showing good classification ability (AUC-ROC 0.77) and calibration on our dataset. We suggest the application of this model with a threshold of 6% to segment patients into low and high-risk groups before treatment planning. However, further data and external validation are warranted before clinical application.

Assessment of the confidence interval in the multivariable normal tissue complication probability model for predicting radiation-induced liver disease in primary liver cancer

We developed a confidence interval-(CI) assessing model in multivariable normal tissue complication probability (NTCP) modeling for predicting radiation-induced liver disease (RILD) in primary liver cancer patients using clinical and dosimetric data. Both the mean NTCP and difference in the mean NTCP (ΔNTCP) between two treatment plans of different radiotherapy modalities were further evaluated and their CIs were assessed. Clinical data were retrospectively reviewed in 322 patients with hepatocellular carcinoma (n = 215) and intrahepatic cholangiocarcinoma (n = 107) treated with photon therapy. Dose-volume histograms of normal liver were reduced to mean liver dose (MLD) based on the fraction size-adjusted equivalent uniform dose. The most predictive variables were used to build the model based on multivariable logistic regression analysis with bootstrapping. Internal validation was performed using the cross-validation leave-one-out method. Both the mean NTCP and the mean ΔNTCP with 95% CIs were calculated from computationally generated multivariate random sets of NTCP model parameters using variance-covariance matrix information. RILD occurred in 108/322 patients (33.5%). The NTCP model with three clinical and one dosimetric parameter (tumor type, Child-Pugh class, hepatitis infection status and MLD) was most predictive, with an area under the receiver operative characteristics curve (AUC) of 0.79 (95% CI 0.74-0.84). In eight clinical subgroups based on the three clinical parameters, both the mean NTCP and the mean ΔNTCP with 95% CIs were able to be estimated computationally. The multivariable NTCP model with the assessment of 95% CIs has potential to improve the reliability of the NTCP model-based approach to select the appropriate radiotherapy modality for each patient.

Radiomic and radiogenomic modeling for radiotherapy: strategies, pitfalls, and challenges

The power of predictive modeling for radiotherapy outcomes has historically been limited by an inability to adequately capture patient-specific variabilities; however, next-generation platforms together with imaging technologies and powerful bioinformatic tools have facilitated strategies and provided optimism. Integrating clinical, biological, imaging, and treatment-specific data for more accurate prediction of tumor control probabilities or risk of radiation-induced side effects are high-dimensional problems whose solutions could have widespread benefits to a diverse patient population-we discuss technical approaches toward this objective. Increasing interest in the above is specifically reflected by the emergence of two nascent fields, which are distinct but complementary: radiogenomics, which broadly seeks to integrate biological risk factors together with treatment and diagnostic information to generate individualized patient risk profiles, and radiomics, which further leverages large-scale imaging correlates and extracted features for the same purpose. We review classical analytical and data-driven approaches for outcomes prediction that serve as antecedents to both radiomic and radiogenomic strategies. Discussion then focuses on uses of conventional and deep machine learning in radiomics. We further consider promising strategies for the harmonization of high-dimensional, heterogeneous multiomics datasets (panomics) and techniques for nonparametric validation of best-fit models. Strategies to overcome common pitfalls that are unique to data-intensive radiomics are also discussed.

Long-term follow-up of a phase I/II trial of radiation dose escalation by simultaneous integrated boost for locally advanced esophageal squamous cell carcinoma

BACKGROUND

To observe the long-term survival and late adverse events in a phase Ⅰ/Ⅱ trial (NCT01843049) of dose escalation for thoracic esophageal squamous cell carcinoma (ESCC) with simultaneous integrated boost (SIB) technique.

METHODS

Patients with ESCC were treated with escalating radiation dose of four predefined levels. Dose of 62.5-64 Gy/25-32 fractions was delivered to the gross tumor volume (GTV), with (Level 3&4) or without (Level 1&2) a SIB up to 70 Gy for pre-treatment 50% SUVmax area of GTV. Patients also received 2 cycles of chemotherapy of cisplatin and fluorouracil concurrently and 2 more cycles after radiotherapy.

RESULTS

Median follow-up duration was 17.2 (2.5-83.4) months for all 44 patients and 47.2 (3.9-83.4) months for 25 survivors. The 3-year overall survival and progression-free survival rates were 57.6% and 41.0%, respectively. One, one, four and twelve severe (grade≥3) esophageal late adverse events (SEAE) occurred in patients of Level 1/2/3/4 (n = 5/10/16/13), with median occurrence time of 6.5 months. In univariable and multivariable competing risk models, maximal dose of the esophagus (Dmax) was found to have significant impact on the incidence of SEAE, and the cutoff distinguishing patients who developed SEAE or not was 77 Gy.

CONCLUSION

Boosting the gross tumor to 63 Gy while delivering 50.4 Gy to subclinical diseases in 28 fractions in locally advanced ESCC is well tolerated with promising long-term survival. Intenser dose regimen should be considered with caution before further toxicity assessment. Esophageal Dmax was significantly associated with severe late esophageal injury, while more findings of dose-volume predictors need larger-sample investigation.

Comparison of machine learning methods for prediction of osteoradionecrosis incidence in patients with head and neck cancer

OBJECTIVES

Mandible osteoradionecrosis (ORN) is one of the most severe toxicities in patients with head and neck cancer (HNC) undergoing radiotherapy (RT). The existing literature focuses on the correlation of mandible ORN and clinical and dosimetric factors. This study proposes the use of machine learning (ML) methods as prediction models for mandible ORN incidence.

METHODS

A total of 96 patients (ORN incidence ratio of 1:1) treated between 2011 and 2015 were selected from the local HNC toxicity database. Demographic, clinical and dosimetric data (based on the mandible dose-volume histogram) were considered as model variables. Prediction accuracy (measured using a stratified fivefold nested cross-validation), sensitivity, specificity, precision and negative predictive value were used to evaluate the prediction performance of a multivariate logistic regression (LR) model, a support vector machine (SVM) model, a random forest (RF) model, an adaptive boosting (AdaBoost) model and an artificial neural network (ANN) model. The different models were compared based on their prediction accuracy and using the McNemar's hypothesis test.

RESULTS

The ANN model (77% accuracy), closely followed by the SVM (76%), AdaBoost (75%) and LR (75%) models, showed the highest overall prediction accuracy. The RF model (71%) showed the lowest prediction accuracy. However, based on the McNemar's test applied to all model pair combinations, no statistically significant difference between the models was found.

CONCLUSION

Based on our results, we encourage the use of ML-based prediction models for ORN incidence as has already been done for other HNC toxicity end points.

ADVANCES IN KNOWLEDGE

This research opens a new path towards personalised RT for HNC using ML to predict mandible ORN incidence.

A Risk Prediction Model by LASSO for Radiation-Induced Xerostomia in Patients With Nasopharyngeal Carcinoma Treated With Comprehensive Salivary Gland-Sparing Helical Tomotherapy Technique

Objective

This study aimed to develop a least absolute shrinkage and selection operator (LASSO)-based multivariable normal tissue complication probability (NTCP) model to predict radiation-induced xerostomia in patients with nasopharyngeal carcinoma (NPC) treated with comprehensive salivary gland–sparing helical tomotherapy technique.

Methods and Materials

LASSO with the extended bootstrapping technique was used to build multivariable NTCP models to predict factors of patient-reported xerostomia relieved by 50% and 80% compared with the level at the end of radiation therapy within 1 year and 2 years, R50-1year and R80-2years, in 203 patients with NPC. The model assessment was based on 10-fold cross-validation and the area under the receiver operating characteristic curve (AUC).

Results

The prediction model by LASSO with 10-fold cross-validation showed that radiation-induced xerostomia recovery could be predicted by prognostic factors of R50-1year (age, gender, T stage, UICC/AJCC stage, parotid Dmean, oral cavity Dmean, and treatment options) and R80-2years (age, gender, T stage, UICC/AJCC stage, oral cavity Dmean, N stage, and treatment options). These prediction models also demonstrated a good performance by the AUC.

Conclusion

The prediction models of R50-1year and R80-2years by LASSO with 10-fold cross-validation were recommended to validate the NTCP model before comprehensive salivary gland–sparing radiation therapy in patients with NPC.

Prediction of Radiation Pneumonitis With Machine Learning in Stage III Lung Cancer: A Pilot Study

Background:

Radiation pneumonitis (RP) is a dose-limiting toxicity in lung cancer radiotherapy (RT). As risk factors in the development of RP, patient and tumor characteristics, dosimetric parameters, and treatment features are intertwined, and it is not always possible to associate RP with a single parameter. This study aimed to determine the algorithm that most accurately predicted RP development with machine learning.

Methods:

Of the 197 cases diagnosed with stage III lung cancer and underwent RT and chemotherapy between 2014 and 2020, 193 were evaluated. The CTCAE 5.0 grading system was used for the RP evaluation. Synthetic minority oversampling technique was used to create a balanced data set. Logistic regression, artificial neural networks, eXtreme Gradient Boosting (XGB), Support Vector Machines, Random Forest, Gaussian Naive Bayes and Light Gradient Boosting Machine algorithms were used. After the correlation analysis, a permutation-based method was utilized for as a variable selection.

Results:

RP was seen in 51 of the 193 cases. Parameters affecting RP were determined as, total(t)V5, ipsilateral lung D_max, contralateral lung D_max, total lung D_max, gross tumor volume, number of chemotherapy cycles before RT, tumor size, lymph node localization and asbestos exposure. LGBM was found to be the algorithm that best predicted RP at 85% accuracy (confidence interval: 0.73-0.96), 97% sensitivity, and 50% specificity.

Conclusion:

When the clinical and dosimetric parameters were evaluated together, the LGBM algorithm had the highest accuracy in predicting RP. However, in order to use this algorithm in clinical practice, it is necessary to increase data diversity and the number of patients by sharing data between centers.

Mini review: Personalization of the radiation therapy management of prostate cancer using MRI-based radiomics

Decisions on how to treat prostate cancer with radiation therapy are guideline-based but as such guidelines have been developed for populations of patients, this invariably leads to overly aggressive treatment in some patients and insufficient treatment in others. Heterogeneity within prostate tumors and in metastatic sites, even within the same patient, is believed to be a major cause of treatment failure. Radiomics biomarkers, more commonly referred to as radiomics 'features", provide readily available, cost-effective, non-invasive tools for screening, detecting tumors and serial monitoring of patients, including assessments of response to therapy and identification of therapeutic complications. Radiomics offers the potential to analyse whole tumors in 3D, as well as sub-regions or 'habitats' within tumors. Combining quantitative information from imaging with pathology, demographic details and other biomarkers will pave the way for personalised treatment selection and monitoring in prostate cancer. The aim of this review is to consider if MRI-based radiomics can bridge the gap between population-based management and personalised management of prostate cancer.

Clinical implication in the use of the AAA algorithm versus the AXB in nasopharyngeal carcinomas by comparison of TCP and NTCP values

Purpose

Retrospective analysis of volumetric modulated arc therapy treatment plans to investigate qualitative, possible, clinical consequences of the use of AAA versus AXB in nasopharyngeal cancer (NPC) cases.

Methods

The dose distribution of 26 treatment plans, produced using RapidArc technique and AAA algorithm, were recalculated using AXB and the same number of monitor units provided by AAA and clinically delivered to each patient. The potential clinical effect of dosimetric differences in the planning target volume (PTV) and in organs at risk (OAR) were evaluated by comparing TCP and NTCP values. The Wilcoxon Signed Rank test was used for statistical comparison of all results obtained from the use of the two algorithms.

Results

The poorer coverage of the PTV, with higher prescribed dose, was reflected in the TCP, which was significantly lower when AXB was used, the median value was 81.55% (range: 74.90, 88.60%) and 84.10% (range: 77.70, 89.90%) for AAA (p < 0.001). OAR mean dose was lower in the AXB recalculated plan than the AAA plan and the difference was statistically significant for all the structures. The NTCP for developing mandible necrosis showed the largest median percentage difference between AAA and AXB (56.6%), the NTCP of risk for larynx edema of Grade ≥ 2 followed with 12.2%.

Conclusions

Differences in dose distribution of NPC treatment plans recalculated with AXB are of clinical significance in those situations where the PTV and OAR involve air or bone, media in which AXB has been shown to more accurately represent the true dose distribution. The availability of AXB algorithm could improve patient dose estimation, increasing the data consistency of clinical trials.

External Validation of a Predictive Model of Urethral Strictures for Prostate Patients Treated With HDR Brachytherapy Boost

Purpose: For prostate cancer treatment, comparable or superior biochemical control was reported when using External-Beam-Radiotherapy (EBRT) with High-Dose-Rate-Brachytherapy (HDRB)-boost, compared to dose-escalation with EBRT alone. The conformal doses produced by HDRB could allow further beneficial prostate dose-escalation, but increase in dose is limited by normal tissue toxicity. Previous works showed correlation between urethral dose and incidence of urinary toxicity, but there is a lack of established guidelines on the dose constraints to this organ. This work aimed at fitting a Normal-Tissue-Complication-Probability model to urethral stricture data collected at one institution and validating it with an external cohort, looking at neo-adjuvant androgen deprivation as dose-modifying factor.

Materials and Methods: Clinical and dosimetric data of 258 patients, with a toxicity rate of 12.8%, treated at a single institution with a variety of prescription doses, were collected to fit the Lyman–Kutcher–Burman (LKB) model using the maximum likelihood method. Due to the different fractionations, doses were converted into 2 Gy-equivalent doses (α/β = 5 Gy), and urethral stricture was used as an end-point. For validation, an external cohort of 187 patients treated as part of the TROG (Trans Tasman Radiation Oncology Group) 03.04 RADAR trial with a toxicity rate of 8.7%, was used. The goodness of fit was assessed using calibration plots. The effect of neo-adjuvant androgen deprivation (AD) was analyzed separating patients who had received it prior to treatment from those who did not receive it.

Results: The obtained LKB parameters were TD50 = 116.7 Gy and m = 0.23; n was fixed to 0.3, based on numerical optimization of the likelihood. The calibration plot showed a good agreement between the observed toxicity and the probability predicted by the model, confirmed by bootstrapping. For the external validation, the calibration plot showed that the observed toxicity obtained with the RADAR patients was well-represented by the fitted LKB model parameters. When patients were stratified by the use of AD TD50 decreased when AD was not present.

Conclusions: Lyman–Kutcher–Burman model parameters were fitted to the risk of urethral stricture and externally validated with an independent cohort, to provide guidance on urethral tolerance doses for patients treated with a HDRB boost. For patients that did not receive AD, model fitting provided a lower TD50 suggesting a protective effect on urethra toxicity.

Machine Learning-Based Models for Prediction of Toxicity Outcomes in Radiotherapy

In order to limit radiotherapy (RT)-related side effects, effective toxicity prediction and assessment schemes are essential. In recent years, the growing interest toward artificial intelligence and machine learning (ML) within the science community has led to the implementation of innovative tools in RT. Several researchers have demonstrated the high performance of ML-based models in predicting toxicity, but the application of these approaches in clinics is still lagging, partly due to their low interpretability. Therefore, an overview of contemporary research is needed in order to familiarize practitioners with common methods and strategies. Here, we present a review of ML-based models for predicting and classifying RT-induced complications from both a methodological and a clinical standpoint, focusing on the type of features considered, the ML methods used, and the main results achieved. Our work overviews published research in multiple cancer sites, including brain, breast, esophagus, gynecological, head and neck, liver, lung, and prostate cancers. The aim is to define the current state of the art and main achievements within the field for both researchers and clinicians.

Validation of irtGPUMCD, a GPU-based Monte Carlo internal dosimetry framework for radionuclide therapy

PURPOSE

Monte Carlo (MC) simulations are highly desirable for dose treatment planning and evaluation in radiation oncology. This is true also in emerging nuclear medicine applications such as internal radiotherapy with radionuclides. The purpose of this study is the validation of irtGPUMCD, a GPU-based MC code for dose calculations in internal radiotherapy.

METHODS

The female and male phantoms of the International Commission on Radiological Protection (ICRP 110) were used as benchmarking geometries for this study focused on ¹⁷⁷Lu and including ^99mTc and ¹³¹I. Dose calculations were also conducted for a real patient. For phantoms, twelve anatomical structures were considered as target/source organs. The S-values were evaluated with irtGPUMCD simulations (10⁸ photons), with gamma branching ratios of ICRP 107 publication. The ¹⁷⁷Lu electrons S-values were calculated for source organs only, based on local deposition of dose in irtGPUMCD. The S-value relative difference between irtGPUMCD and IDAC-DOSE were evaluated for all targets/sources considered. A DVHs comparison with GATE was conducted. An exponential track length estimator was introduced in irtGPUMCD to increase computational efficiency.

RESULTS

The relative S-value differences between irtGPUMCD and IDAC-DOSE were <5% while this comparison with GATE was <1%. The DVHs dosimetric indices comparison between GATE and irtGPUMCD for the patient led to an excellent agreement (<2%). The time required for the simulation of 10⁸ photons was 1.5 min for the female phantom, and one minute for the real patient (<1% uncertainty). These results are promising and let envision the use of irtGPUMCD for internal dosimetry in clinical applications.

Machine learning helps identifying volume-confounding effects in radiomics

PURPOSE

Highlighting the risk of biases in radiomics-based models will help improve their quality and increase usage as decision support systems in the clinic. In this study we use machine learning-based methods to identify the presence of volume-confounding effects in radiomics features. Methods 841 radiomics features were extracted from two retrospective publicly available datasets of lung and head neck cancers using open source software. Unsupervised hierarchical clustering and principal component analysis (PCA) identified relations between radiomics and clinical outcomes (overall survival). Bootstrapping techniques with logistic regression verified features' prognostic power and robustness. Results Over 80% of the features had large pairwise correlations. Nearly 30% of the features presented strong correlations with tumor volume. Using volume-independent features for clustering and PCA did not allow risk stratification of patients. Clinical predictors outperformed radiomics features in bootstrapping and logistic regression. Conclusions The adoption of safeguards in radiomics is imperative to improve the quality of radiomics studies. We proposed machine learning (ML) - based methods for robust radiomics signatures development.

Artificial Intelligence: reshaping the practice of radiological sciences in the 21st century

Advances in computing hardware and software platforms have led to the recent resurgence in artificial intelligence (AI) touching almost every aspect of our daily lives by its capability for automating complex tasks or providing superior predictive analytics. AI applications are currently spanning many diverse fields from economics to entertainment, to manufacturing, as well as medicine. Since modern AI's inception decades ago, practitioners in radiological sciences have been pioneering its development and implementation in medicine, particularly in areas related to diagnostic imaging and therapy. In this anniversary article, we embark on a journey to reflect on the learned lessons from past AI's chequered history. We further summarize the current status of AI in radiological sciences, highlighting, with examples, its impressive achievements and effect on re-shaping the practice of medical imaging and radiotherapy in the areas of computer-aided detection, diagnosis, prognosis, and decision support. Moving beyond the commercial hype of AI into reality, we discuss the current challenges to overcome, for AI to achieve its promised hope of providing better precision healthcare for each patient while reducing cost burden on their families and the society at large.

Normal tissue complication probability (NTCP) models for modern radiation therapy

Mathematical models of normal tissue complication probability (NTCP) able to robustly predict radiation-induced morbidities (RIM) play an essential role in the identification of a personalized optimal plan, and represent the key to maximizing the benefits of technological advances in radiation therapy (RT). Most modern RT techniques pose, however, new challenges in estimating the risk of RIM. The aim of this report is to schematically review NTCP models in the framework of advanced radiation therapy techniques. Issues relevant to hypofractionated stereotactic body RT and ion beam therapy are critically reviewed. Reirradiation scenarios for new or recurrent malignances and NTCP are also illustrated. A new phenomenological approach to predict RIM is suggested.

Developing a multivariable normal tissue complication probability model to predict late rectal bleeding following intensity-modulated radiation therapy

Purpose: To develop a multivariable normal tissue complication probability (NTCP) model to predict moderate to severe late rectal bleeding following intensity-modulated radiation therapy (IMRT).

Methods and materials: Sixty-eight patients with localized prostate cancer treated by IMRT from 2008 to 2011 were enrolled. The median follow-up time was 56 months. According to the criteria of D'Amico risk classifications, there were 9, 20 and 39 patients in low, intermediate and high-risk groups, respectively. Forty-two patients were combined with androgen deprivation therapy. Fifteen patients had suffered from grade 2 or more (grade 2+) late rectal bleeding. The numbers of predictors for a multivariable logistic regression NTCP model were determined by the least absolute shrinkage and selection operator (LASSO).

Results: The most important predictors for late rectal bleeding ranked by LASSO were platelet count, risk group and the relative volume of rectum receiving at least 65 Gy (V₆₅). The NTCP model of grade 2+ rectal bleeding was as follows: S = -17.49 + Platelets (1000/μL) * (-0.025) + Risk group * Corresponding coefficient (low-risk group = 0; intermediate-risk group = 19.07; high-risk group = 20.41) + V₆₅ * 0.045.

Conclusions: A LASSO-based multivariable NTCP model comprising three important predictors (platelet count, risk group and V₆₅) was established to predict the incidence of grade 2+ late rectal bleeding after IMRT.

Combining handcrafted features with latent variables in machine learning for prediction of radiation-induced lung damage

PURPOSE

There has been burgeoning interest in applying machine learning methods for predicting radiotherapy outcomes. However, the imbalanced ratio of a large number of variables to a limited sample size in radiation oncology constitutes a major challenge. Therefore, dimensionality reduction methods can be a key to success. The study investigates and contrasts the application of traditional machine learning methods and deep learning approaches for outcome modeling in radiotherapy. In particular, new joint architectures based on variational autoencoder (VAE) for dimensionality reduction are presented and their application is demonstrated for the prediction of lung radiation pneumonitis (RP) from a large-scale heterogeneous dataset.

METHODS

A large-scale heterogeneous dataset containing a pool of 230 variables including clinical factors (e.g., dose, KPS, stage) and biomarkers (e.g., single nucleotide polymorphisms (SNPs), cytokines, and micro-RNAs) in a population of 106 nonsmall cell lung cancer (NSCLC) patients who received radiotherapy was used for modeling RP. Twenty-two patients had grade 2 or higher RP. Four methods were investigated, including feature selection (case A) and feature extraction (case B) with traditional machine learning methods, a VAE-MLP joint architecture (case C) with deep learning and lastly, the combination of feature selection and joint architecture (case D). For feature selection, Random forest (RF), Support Vector Machine (SVM), and multilayer perceptron (MLP) were implemented to select relevant features. Specifically, each method was run for multiple times to rank features within several cross-validated (CV) resampled sets. A collection of ranking lists were then aggregated by top 5% and Kemeny graph methods to identify the final ranking for prediction. A synthetic minority oversampling technique was applied to correct for class imbalance during this process. For deep learning, a VAE-MLP joint architecture where a VAE aimed for dimensionality reduction and an MLP aimed for classification was developed. In this architecture, reconstruction loss and prediction loss were combined into a single loss function to realize simultaneous training and weights were assigned to different classes to mitigate class imbalance. To evaluate the prediction performance and conduct comparisons, the area under receiver operating characteristic curves (AUCs) were performed for nested CVs for both handcrafted feature selections and the deep learning approach. The significance of differences in AUCs was assessed using the DeLong test of U-statistics.

RESULTS

An MLP-based method using weight pruning (WP) feature selection yielded the best performance among the different hand-crafted feature selection methods (case A), reaching an AUC of 0.804 (95% CI: 0.761-0.823) with 29 top features. A VAE-MLP joint architecture (case C) achieved a comparable but slightly lower AUC of 0.781 (95% CI: 0.737-0.808) with the size of latent dimension being 2. The combination of handcrafted features (case A) and latent representation (case D) achieved a significant AUC improvement of 0.831 (95% CI: 0.805-0.863) with 22 features (P-value = 0.000642 compared with handcrafted features only (Case A) and P-value = 0.000453 compared to VAE alone (Case C)) with an MLP classifier.

CONCLUSION

The potential for combination of traditional machine learning methods and deep learning VAE techniques has been demonstrated for dealing with limited datasets in modeling radiotherapy toxicities. Specifically, latent variables from a VAE-MLP joint architecture are able to complement handcrafted features for the prediction of RP and improve prediction over either method alone.

Potential of serum microRNAs as biomarkers of radiation injury and tools for individualization of radiotherapy

Due to tremendous technological advances, radiation oncologists are now capable of personalized treatment plans and deliver the dose in a highly precise manner. However, a crucial challenge is how to escalate radiation doses to cancer cells while reducing damage to surrounding healthy tissues. This determines the probability of achieving therapeutic success whilst safeguarding patients from complications. The current dose constraints rely on observational data. Therefore, incidental toxicity observed in a minority of patients limits the admissible dose thresholds for the whole population, theoretically narrowing down the curative potential of radiotherapy. Future tools for measurements of individual's radiosensitivity before and during treatment would allow proper treatment personalization. Variation in tissue tolerance is at least partially genetically-determined and recent progress in the field of molecular biology raises the possibility that novel assays will allow to predict the response to ionizing radiation. Recently, microRNAs have garnered interest as stable biomarkers of tumor radiation response and normal-tissue toxicity. Preclinical studies in mice and nonhuman primates have shown that serum circulating microRNAs can be used to accurately distinguish pre- and postirradiation states and predict the biological impact of high-dose irradiation. First reports from human studies are also encouraging, however biology-driven precision radiation oncology, which tailors treatment to individual patient's needs, still remains to be translated into clinical studies. In this review, we summarize current knowledge about the potential of serum microRNAs as biodosimeters and biomarkers for radiation injury to lung and hematopoietic cells.

Machine learning and modeling: Data, validation, communication challenges

With the era of big data, the utilization of machine learning algorithms in radiation oncology is rapidly growing with applications including: treatment response modeling, treatment planning, contouring, organ segmentation, image-guidance, motion tracking, quality assurance, and more. Despite this interest, practical clinical implementation of machine learning as part of the day-to-day clinical operations is still lagging. The aim of this white paper is to further promote progress in this new field of machine learning in radiation oncology by highlighting its untapped advantages and potentials for clinical advancement, while also presenting current challenges and open questions for future research. The targeted audience of this paper includes newcomers as well as practitioners in the field of medical physics/radiation oncology. The paper also provides general recommendations to avoid common pitfalls when applying these powerful data analytic tools to medical physics and radiation oncology problems and suggests some guidelines for transparent and informative reporting of machine learning results.

Multivariate normal tissue complication probability modeling of vaginal late toxicity after brachytherapy for cervical cancer

PURPOSE

To explore the best variables combination for a predictive model of vaginal toxicity in cervical cancer patients undergoing brachytherapy (BT).

METHODS AND MATERIALS

Clinical and 3-dimensional dosimetric parameters were retrospectively extracted from an institutional database of consecutive patients undergoing intracavitary BT after external beam radiation therapy from 2006 to 2013 for a cervical cancer. A least absolute shrinkage and selection operator selection procedure in Cox's proportional hazards regression model was performed to select a set of relevant predictors for a multivariate normal tissue complication probability model of Grade ≥2 vaginal late toxicity. Outcomes reliability was internally assessed by bootstrap resampling method.

RESULTS

One hundred sixty-nine women were included in the present study with a median followup time of 3.8 years (interquartile range [IQR], 1.9-5.6 years). The 2 years and 5 years cumulative incidence rates of Grade ≥2 late vaginal toxicity were 19.9% and 27.5%, respectively. Among 31 metrics and six clinical factors extracted, the optimal model included two dosimetric variables: V70_Gy and D5_% (the percentage volume that received a dose greater or equal to 70 Gy and the minimum dose given to the hottest 5% volume, respectively). Area under the ROC curve at 2 and 5 years of followup were 0.85 and 0.91, respectively. Regarding internal validation, median area under the ROC curve of bootstrap predictions was 0.83 (IQR, 0.78-0.88) and 0.89 (IQR, 0.85-0.93) at 2 and 5 years of followup, respectively.

CONCLUSIONS

A multivariate normal tissue complication probability model for severe vaginal toxicity based on two dosimetric variables (V70_Gy and D5_%) provides reliable discrimination capability in a cohort of cervical cancer treated with external beam radiation therapy and BT.

Investigating rectal toxicity associated dosimetric features with deformable accumulated rectal surface dose maps for cervical cancer radiotherapy

BACKGROUND

Better knowledge of the dose-toxicity relationship is essential for safe dose escalation to improve local control in cervical cancer radiotherapy. The conventional dose-toxicity model is based on the dose volume histogram, which is the parameter lacking spatial dose information. To overcome this limit, we explore a comprehensive rectal dose-toxicity model based on both dose volume histogram and dose map features for accurate radiation toxicity prediction.

METHODS

Forty-two cervical cancer patients treated with combined external beam radiotherapy (EBRT) and brachytherapy (BT) were retrospectively studied, including 12 with Grade ≥ 2 rectum toxicity and 30 patients with Grade 0-1 toxicity (non-toxicity patients). The cumulative equivalent 2-Gy rectal surface dose was deformably summed using the deformation vector fields obtained through a recent developed local topology preserved non-rigid point matching algorithm. The cumulative three-dimensional (3D) dose was flattened and mapped to a two-dimensional (2D) plane to obtain the rectum surface dose map (RSDM). The dose volume parameters (DVPs) were calculated from the 3D rectum surface, while the texture features and the dose geometric parameters (DGPs) were extracted from the 2D RSDM. Representative features further computed from DVPs, textures and DGPs by principle component analysis (PCA) and statistical analysis were respectively fed into a support vector machine equipped with a sequential feature selection procedure. The predictive powers of the representative features were compared with the GEC-ESTRO dosimetric parameters D_0.1/1/2cm³.

RESULTS

Satisfactory predictive accuracy of sensitivity 74.75 and 84.75%, specificity 72.67 and 79.87%, and area under the receiver operating characteristic curve (AUC) 0.82 and 0.91 were respectively achieved by the PCA features and statistical significant features, which were superior to the D_0.1/1/2cm³ (AUC 0.71). The relative area in dose levels of 64Gy, 67Gy, 68Gy, 87Gy, 88Gy and 89Gy, perimeters in dose levels of 89Gy, as well as two texture features were ranked as the important factors that were closely correlated with rectal toxicity.

CONCLUSIONS

Our extensive experimental results have demonstrated the feasibility of the proposed scheme. A future large patient cohort study is still needed for model validation.

Method for Automatic Selection of Parameters in Normal Tissue Complication Probability Modeling

PURPOSE

To present a fully automatic method to generate multiparameter normal tissue complication probability (NTCP) models and compare its results with those of a published model, using the same patient cohort.

METHODS AND MATERIALS

Data were analyzed from 345 rectal cancer patients treated with external radiation therapy to predict the risk of patients developing grade 1 or ≥2 cystitis. In total, 23 clinical factors were included in the analysis as candidate predictors of cystitis. Principal component analysis was used to decompose the bladder dose-volume histogram into 8 principal components, explaining more than 95% of the variance. The data set of clinical factors and principal components was divided into training (70%) and test (30%) data sets, with the training data set used by the algorithm to compute an NTCP model. The first step of the algorithm was to obtain a bootstrap sample, followed by multicollinearity reduction using the variance inflation factor and genetic algorithm optimization to determine an ordinal logistic regression model that minimizes the Bayesian information criterion. The process was repeated 100 times, and the model with the minimum Bayesian information criterion was recorded on each iteration. The most frequent model was selected as the final "automatically generated model" (AGM). The published model and AGM were fitted on the training data sets, and the risk of cystitis was calculated.

RESULTS

The 2 models had no significant differences in predictive performance, both for the training and test data sets (P value > .05) and found similar clinical and dosimetric factors as predictors. Both models exhibited good explanatory performance on the training data set (P values > .44), which was reduced on the test data sets (P values < .05).

CONCLUSIONS

The predictive value of the AGM is equivalent to that of the expert-derived published model. It demonstrates potential in saving time, tackling problems with a large number of parameters, and standardizing variable selection in NTCP modeling.

Data Analysis in Radiotherapy Treatments

Radiotherapy is one of the main cancer treatments available today, together with chemotherapy and surgery. Radiotherapy treatments have to be planned for each patient in an individualized manner. The knowledge acquired from one single treatment can be used to improve the treatment planning and outcome of several other patients. In the last years, attention has been drawn to the added value of using data analysis for radiotherapy treatment planning, prediction of treatment outcomes, survival analysis and quality assurance. In this article, existing literature is reviewed.

Artificial intelligence in radiation oncology: A specialty-wide disruptive transformation?

Artificial intelligence (AI) is emerging as a technology with the power to transform established industries, and with applications from automated manufacturing to advertising and facial recognition to fully autonomous transportation. Advances in each of these domains have led some to call AI the "fourth" industrial revolution [1]. In healthcare, AI is emerging as both a productive and disruptive force across many disciplines. This is perhaps most evident in Diagnostic Radiology and Pathology, specialties largely built around the processing and complex interpretation of medical images, where the role of AI is increasingly seen as both a boon and a threat. In Radiation Oncology as well, AI seems poised to reshape the specialty in significant ways, though the impact of AI has been relatively limited at present, and may rightly seem more distant to many, given the predominantly interpersonal and complex interventional nature of the specialty. In this overview, we will explore the current state and anticipated future impact of AI on Radiation Oncology, in detail, focusing on key topics from multiple stakeholder perspectives, as well as the role our specialty may play in helping to shape the future of AI within the larger spectrum of medicine.

Development of a normal tissue complication probability (NTCP) model for radiation-induced hypothyroidism in nasopharyngeal carcinoma patients

BACKGROUND

The objectives of this study were to build a normal tissue complication probability (NTCP) model of radiation-induced hypothyroidism (RHT) for nasopharyngeal carcinoma (NPC) patients and to compare it with other four published NTCP models to evaluate its efficacy.

METHODS

Medical notes of 174 NPC patients after radiotherapy were reviewed. Biochemical hypothyroidism was defined as an elevated level of serum thyroid-stimulating hormone (TSH) value with a normal or decreased level of serum free thyroxine (fT4) after radiotherapy. Logistic regression with leave-one-out cross-validation was performed to establish the NTCP model. Model performance was evaluated and compared by the area under the receiver operating characteristic curve (AUC) in our NPC cohort.

RESULTS

With a median follow-up of 24 months, 39 (22.4%) patients developed biochemical hypothyroidism. Gender, chemotherapy, the percentage thyroid volume receiving more than 50 Gy (V50), and the maximum dose of the pituitary (Pmax) were identified as the most predictive factors for RHT. A NTCP model based on these four parameters were developed. The model comparison was made in our NPC cohort and our NTCP model performed better in RHT prediction than the other four models.

CONCLUSIONS

This study developed a four-variable NTCP model for biochemical hypothyroidism in NPC patients post-radiotherapy. Our NTCP model for RHT presents a high prediction capability.

TRIAL REGISTRATION

This is a retrospective study without registration.

Impact of different biologically-adapted radiotherapy strategies on tumor control evaluated with a tumor response model

Motivated by the capabilities of modern radiotherapy techniques and by the recent developments of functional imaging techniques, dose painting by numbers (DPBN) was proposed to treat tumors with heterogeneous biological characteristics. This work studies different DPBN optimization techniques for virtual head and neck tumors assessing tumor response in terms of cell survival and tumor control probability with a previously published tumor response model (TRM). Uniform doses of 2 Gy are redistributed according to the microscopic oxygen distribution and the density distribution of tumor cells in four virtual tumors with different biological characteristics. In addition, two different optimization objective functions are investigated, which: i) minimize tumor cell survival (OF_surv) or; ii) maximize the homogeneity of the density of surviving tumor cells (OF_std). Several adaptive schemes, ranging from single to daily dose optimization, are studied and the treatment response is compared to that of the uniform dose. The results show that the benefit of DPBN treatments depends on the tumor reoxygenation capability, which strongly differed among the set of virtual tumors investigated. The difference between daily (fraction by fraction) and three weekly optimizations (at the beginning of weeks 1, 3 and 4) was found to be small, and higher benefit was observed for the treatments optimized using OF_surv. This in silico study corroborates the hypothesis that DPBN may be beneficial for treatments of tumors which show reoxygenation during treatment, and that a few optimizations may be sufficient to achieve this therapeutic benefit.

Machine Learning in Radiation Oncology: Opportunities, Requirements, and Needs

Machine learning (ML) has the potential to revolutionize the field of radiation oncology, but there is much work to be done. In this article, we approach the radiotherapy process from a workflow perspective, identifying specific areas where a data-centric approach using ML could improve the quality and efficiency of patient care. We highlight areas where ML has already been used, and identify areas where we should invest additional resources. We believe that this article can serve as a guide for both clinicians and researchers to start discussing issues that must be addressed in a timely manner.

Radiomics in precision medicine for lung cancer

With the improvement of external radiotherapy delivery accuracy, such as intensity-modulated and stereotactic body radiation therapy, radiation oncology has recently entered in the era of precision medicine. Despite these precise irradiation modalities, lung cancers remain one of the most aggressive human cancers worldwide, possibly because of diverse genotypic alterations that drive and maintain lung tumorigenesis. It has been long recognized that imaging could aid in the diagnosis, tumor delineation, and monitoring of lung cancer. Moreover, accumulating evidence suggests that imaging information could be further used to tailor treatment type and intensity, as well as predict treatment outcomes in radiotherapy. However, these imaging tasks have been carried out either qualitatively or using simplistic metrics that doesn't take advantage of the full scale of imaging knowledge. Radiomics, which is a recent field of research that aims to provide a more quantitative representation of imaging information relating tumor phenotypes to clinical and genotypic endpoints by embedding extracted image features into predictive mathematical models. These predictive models can be a key component in the clinician decision making and treatment personalization. This review provides an overview of the radiomics application and its methodology for radiation oncology studies in lung cancer.

Radiogenomics and radiotherapy response modeling

Advances in patient-specific information and biotechnology have contributed to a new era of computational medicine. Radiogenomics has emerged as a new field that investigates the role of genetics in treatment response to radiation therapy. Radiation oncology is currently attempting to embrace these recent advances and add to its rich history by maintaining its prominent role as a quantitative leader in oncologic response modeling. Here, we provide an overview of radiogenomics starting with genotyping, data aggregation, and application of different modeling approaches based on modifying traditional radiobiological methods or application of advanced machine learning techniques. We highlight the current status and potential for this new field to reshape the landscape of outcome modeling in radiotherapy and drive future advances in computational oncology.