Normal tissue complication probability (NTCP) models for predicting temporal lobe injury after intensity-modulated radiotherapy in nasopharyngeal carcinoma: A large registry-based retrospective study from China

Development and validation of Prediction models for radiation-induced hypoglossal neuropathy in patients with nasopharyngeal carcinoma

BACKGROUND AND PURPOSE

To establish predictive models for radiation-induced hypoglossal neuropathy (RIHN) in patients with nasopharyngeal carcinoma (NPC) after intensity-modulated radiotherapy (IMRT).

MATERIALS AND METHODS

Data from 423 NPC patients receiving IMRT-based treatment were retrospectively reviewed. They were randomly (3:2) divided into a training set (n = 256) and a testing set (n = 167). Dosimetric variables were selected by penalized regression and machine learning, with area under the receiver operating curve (AUC) calculated. Clinical variables were selected by the competitive risk analysis. A competitive risk model including clinical and dosimetric variables was performed, and a nomogram was generated as a visualization of the model to predict the incidence of RIHN.

RESULTS

During a median follow-up of 102 months (IQR: 89.5 to 112.9 months), the cumulative incidence of RIHN at 3, 5, and 8 years were 2.1 %, 5.4 %, and 10.5 %, respectively. D1cc and aV75 were the most predictive dosimetric variables. The dose-effect curve plotted with D1cc indicated the tolerance dose for a 5 % probability of developing RIHN in 8 years (TD5/8) was 77.3 Gy (EQD2). The restricted cubic spline between aV75 and RIHN indicated a volume threshold of 0.81 cm3. A competitive risk model including hypoglossal canal involvement, concurrent chemotherapy, D1cc, and aV75 was established, with the C-index of the training set and testing set being 0.726 and 0.691, respectively. The nomogram-defined high-risk group had the higher RIHN incidences in the training and testing sets.

CONCLUSIONS

This study identified the most critical dosimetric predictors, which is expected to become a feasible dose constraint for hypoglossal nerves in radiation plan. Combining dosimetric and clinical predictors, we further proposed and validated the first nomogram model to quantify the risk of RIHN, contributing to identifying high-risk patients and early intervention. Further multicenter studies are needed to validate or complement our findings.

Systematic review of MRI alterations in the brain following proton and photon radiation therapy: Towards a uniform European Particle Therapy Network (EPTN) definition

Magnetic resonance imaging (MRI) often demonstrates alterations following cranial radiotherapy (RT), which may result in clinical symptoms and diagnostic uncertainty, and thus potentially impact treatment decisions. The potential differences in MRI alterations after proton and photon RT, has raised concerns regarding the relative biological effectiveness of proton therapy. To provide an overview of MRI alterations in the brain post-RT and to explore differences between photon and proton RT, a systematic review adhering to the PRISMA guidelines was conducted, focusing on the assessment methods and definitions across studies. A systematic search of three electronic databases was performed using the concepts 'normo-fractionated radiotherapy ', 'MRI alterations' and 'brain, skull base or head and neck tumours in adult and paediatric populations'. Data extraction and quality assessment was performed on articles meeting the predefined criteria by two independent reviewers. Out of 5887 screened studies, 94 met the inclusion criteria. These studies were categorized based on confinement of the MRI alterations to temporal lobe, brainstem, or across the entire brain. Additional subclassification was performed based on MRI sequences evaluated or by the nature of the alterations, with pseudoprogression generally reserved for glioma patients. While many papers exist on MRI alterations in the brain after RT, this review highlights significant inconsistencies in the terminology and definitions, limiting the comparability of findings across studies. Our results highlight the need for and facilitate the development of a standardized framework for describing MRI alterations after RT.

Reduced-volume radiotherapy versus conventional-volume radiotherapy after induction chemotherapy in nasopharyngeal carcinoma: An open-label, noninferiority, multicenter, randomized phase 3 trial

BACKGROUND

Nearly 90% locoregionally advanced nasopharyngeal carcinoma (LANPC) responds to induction chemotherapy (IC) with significant tumor volume shrinkage. Radiotherapy always follows IC, and reduced volume has been proposed. However, the efficacy and safety of reduced-volume radiotherapy is uncertain.

METHODS

In this multi-center, noninferiority, randomized, controlled trial, patients with LANPC who completed IC were randomly assigned (1:1) to receive reduced-volume radiotherapy based on post-IC tumor volume (Post-IC group) or conventional volume radiotherapy based on pre-IC tumor volume (Pre-IC group). The primary endpoint was locoregional relapse-free survival, with a noninferiority margin of 8%. Secondary endpoints comprised adverse events, and quality of life (QoL).

RESULTS

Between August 7, 2020, and May 27, 2022, 445 patients were randomly assigned to Post-IC (n = 225) or Pre-IC (n = 220) groups. The average volume receiving radical dose was 66.6 cm3 in Post-IC group versus 80.9 cm3. After a median follow-up of 40.4 months, the 3-year locoregional relapse-free survival was 91.5% in the Post-IC group versus 91.2%, with a difference of 0.3% (95% confidence interval -4.9% to 5.5%). The incidence of grade 3-4 radiation-related toxicity was lower in the Post-IC group including: acute mucositis (19.8% vs 34.1%), late otitis media (9.5% vs 20.9%) and late dry month (3.6% vs 9.5%). The Post-IC group had better QoL for global health status, physical functioning, emotional functioning, dry mouth and sticky saliva.

CONCLUSIONS

In this trial, reduced-volume radiotherapy was noninferior to conventional volume radiotherapy in locoregional relapse-free survival, and was associated with lower toxicities and improved QoL. (ClinicalTrials.gov identifier NCT04384627).

Potential Risk of Cognitive Impairment Due to Irradiation of Neural Structures in Locally Advanced Nasopharyngeal Cancer Treated by Curative Radiotherapy

Background and Objectives: Brain radionecrosis is an under-recognized but potentially life-altering late complication of radiotherapy in patients with locally advanced nasopharyngeal cancer. Temporal lobe radionecrosis and high-dose exposure to the hippocampus are strongly associated with cognitive decline and radiation-induced dementia, negatively impacting patients' long-term quality of life (QoL). This study aimed to evaluate and compare radiation dose distributions to critical brain structures across three radiotherapy techniques-3D conformal radiotherapy (3D-CRT), intensity-modulated radiotherapy (IMRT), and volumetric-modulated arc therapy (VMAT)-in order to assess potential neurocognitive risks and support hippocampal-sparing protocols. Materials and Methods: Ten patients previously treated with 3D-CRT were retrospectively replanned using IMRT and VMAT techniques on the Eclipse v13.3 (VARIAN) planning system. Bilateral hippocampi and temporal lobes were delineated as organs at risk (OARs) according to the RTOG atlas, and dosimetric parameters including D_max, D_mean, and D_min were recorded. V7.3 values were evaluated for hippocampal avoidance regions. Results: While IMRT and VMAT provided improved target volume coverage and reduced high-dose exposure to many standard OARs, both techniques were associated with increased D_mean and D_min to the hippocampus and temporal lobes compared to 3D-CRT. The highest D_max values to the temporal lobes were observed in 3D-CRT plans, indicating a potential risk of radionecrosis. VMAT plans showed hippocampal mean doses exceeding 10 Gy in some cases, with V7.3 > 40%, breaching established neurocognitive risk thresholds. Conclusions: These findings support the routine delineation of the hippocampus and temporal lobes as OARs in radiotherapy planning for nasopharyngeal cancer. The implementation of hippocampal-sparing strategies, particularly in IMRT and VMAT, is recommended to reduce the risk of radiation-induced cognitive toxicity and preserve long-term QoL in survivors.

Artificial Intelligence in Head and Neck Surgery

This article explores artificial intelligence's (AI's) role in otolaryngology for head and neck cancer diagnosis and management. It highlights AI's potential in pattern recognition for early cancer detection, prognostication, and treatment planning, primarily through image analysis using clinical, endoscopic, and histopathologic images. Radiomics is also discussed at length, as well as the many ways that radiologic image analysis can be utilized, including for diagnosis, lymph node metastasis prediction, and evaluation of treatment response. The study highlights AI's promise and limitations, underlining the need for clinician-data scientist collaboration to enhance head and neck cancer care.

Non-complete recovery of temporal lobe white matter diffusion metrics at one year Post-Radiotherapy: Implications for Radiation-Induced necrosis risk

BACKGROUND

Temporal lobe (TL) white matter (WM) injuries are often seen early after radiotherapy (RT) in nasopharyngeal carcinoma patients (NPCs), which fail to fully recover in later stages, exhibiting a "non-complete recovery pattern". Herein, we explored the correlation between non-complete recovery WM injuries and TL necrosis (TLN), identifying dosimetric predictors for TLN-related high-risk WM injuries.

METHODS

We longitudinally examined 161 NPCs and 19 healthy controls employing multi-shell diffusion MRI. Automated fiber-tract quantification quantified diffusion metrics within TL WM tract segments. ANOVA identified non-complete recovery WM tract segments one-year post-RT. Cox regression models discerned TLN risk factors utilizing non-complete recovery diffusion metrics. Normal tissue complication probability (NTCP) models and dose-response analysis further scrutinized RT-related toxicity to high-risk WM tract segments.

RESULTS

Seven TL WM tract segments exhibited a "non-complete recovery pattern". Cox regression analysis identified mean diffusivity of the left uncinate fasciculus segment 1, neurite density index (NDI) of the left cingulum hippocampus segment 1, and NDI of the right inferior longitudinal fasciculus segment 1 as TLN risk predictors (hazard ratios [HRs] with confidence interval [CIs]: 1.45 [1.17-1.81], 1.07 [1.00-1.15], and 1.15 [1.03-1.30], respectively; all P-values < 0.05). In NTCP models, D10cc.L, D20cc.L and D10cc.R demonstrated superior performance, with TD50 of 37.22 Gy, 24.96 Gy and 37.28 Gy, respectively.

CONCLUSIONS

Our findings underscore the significance of the "non-complete recovery pattern" in TL WM tract segment injuries during TLN development. Understanding TLN-related high-risk WM tract segments and their tolerance doses could facilitate early intervention in TLN and improve RT protocols.

A pretreatment multiparametric MRI-based radiomics-clinical machine learning model for predicting radiation-induced temporal lobe injury in patients with nasopharyngeal carcinoma

BACKGROUND

To establish and validate a machine learning model using pretreatment multiparametric magnetic resonance imaging-based radiomics data with clinical data to predict radiation-induced temporal lobe injury (RTLI) in patients with nasopharyngeal carcinoma (NPC) after intensity-modulated radiotherapy (IMRT).

METHODS

Data from 230 patients with NPC who received IMRT (130 with RTLI and 130 without) were randomly divided into the training (n = 161) and validation cohort (n = 69) with a ratio of 7:3. Radiomics features were extracted from pretreatment apparent diffusion coefficient (ADC) map, T2-weighted imaging (T2WI), and CE-T1-weighted imaging (CE-T1WI). T-test, spearman rank correlation, and least absolute shrinkage and selection operator (LASSO) algorithm were employed to identify significant radiomics features. Clinical features were selected with univariate and multivariate analyses. Radiomics and clinical models were constructed using multiple machine learning classifiers, and a clinical-radiomics nomogram that combined clinical with radiomics features was developed. Receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) were drawn to compare and verify the predictive performances of the clinical model, radiomics model, and clinical-radiomics nomogram.

RESULTS

A total of 5064 radiomics features were extracted, from which 52 radiomics features were selected to construct the radiomics signature. The AUC of the radiomics signature based on multiparametric MRI was 0.980 in the training cohort and 0.969 in the validation cohort, outperforming the radiomics signature only based on T2WI and CE-T1WI (p < 0.05), which highlighted the significance of the DWI sequence in the prediction of temporal lobe injury. The area under the curve (AUC) of the clinical model was 0.895 in the training cohort and 0.905 in the validation cohort. The nomogram, which integrated radiomics and clinical features, demonstrated an impressive AUC value of 0.984 in the validation set; however, no statistically significant difference was observed compared to the radiomics model. The calibration curve and decision curve analysis of the nomogram demonstrated excellent predictive performance and clinical feasibility.

CONCLUSIONS

The clinical-radiomics nomogram, integrating clinical features with radiomics features derived from pretreatment multiparametric MRI, exhibits compelling predictive performance for RTLI in patients diagnosed with NPC.

Dosimetric parameters predict radiation-induced temporal lobe necrosis in nasopharyngeal carcinoma patients: A systematic review and meta-analysis

This systematic review examines the role of dosimetric parameters in predicting temporal lobe necrosis (TLN) risk in nasopharyngeal carcinoma (NPC) patients treated with three-dimensional conformal RT (3D-CRT), intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT). TLN is a serious late complication that can adversely affect the quality of life of NPC patients. Understanding the relationship between dosimetric parameters and TLN can guide treatment planning and minimize radiation-related complications. A comprehensive search identified relevant studies published up to July 2023. Studies reporting on dosimetric parameters and TLN in NPC patients undergoing 3D-CRT, IMRT, and VMAT were included. TLN incidence, follow-up duration, and correlation with dosimetric parameters of the temporal lobe were analyzed. The review included 30 studies with median follow-up durations ranging from 28 to 110 months. The crude incidence of TLN varied from 2.3 % to 47.3 % and the average crude incidence of TLN is approximately 14 %. Dmax and D1cc emerged as potential predictors of TLN in 3D-CRT and IMRT-treated NPC patients. Threshold values of >72 Gy for Dmax and >62 Gy for D1cc were associated with increased TLN risk. However, other factors should also be considered, including host characteristics, tumor-specific features and therapeutic factors. In conclusion, this systematic review highlights the significance of dosimetric parameters, particularly Dmax and D1cc, in predicting TLN risk in NPC patients undergoing 3D-CRT, IMRT, and VMAT. The findings provide valuable insights that can help in developing optimal treatment planning strategies and contribute to the development of clinical guidelines in this field.

Neurological complications of modern radiotherapy for head and neck cancer

Radiotherapy is one of the mainstay treatment modalities for the management of non-metastatic head and neck cancer (HNC). Notable improvements in treatment outcomes have been observed in the recent decades. Modern radiotherapy techniques, such as intensity-modulated radiotherapy and charged particle therapy, have significantly improved tumor target conformity and enabled better preservation of normal structures. However, because of the intricate anatomy of the head and neck region, multiple critical neurological structures such as the brain, brainstem, spinal cord, cranial nerves, nerve plexuses, autonomic pathways, brain vasculature, and neurosensory organs, are variably irradiated during treatment, particularly when tumor targets are in close proximity. Consequently, a diverse spectrum of late neurological sequelae may manifest in HNC survivors. These neurological complications commonly result in irreversible symptoms, impair patients' quality of life, and contribute to a substantial proportion of non-cancer deaths. Although the relationship between radiation dose and toxicity has not been fully elucidated for all complications, appropriate application of dosimetric constraints during radiotherapy planning may reduce their incidence. Vigilant surveillance during the course of survivorship also enables early detection and intervention. This article endeavors to provide a comprehensive review of the various neurological complications of modern radiotherapy for HNC, summarize the current incidence data, discuss methods to minimize their risks during radiotherapy planning, and highlight potential strategies for managing these debilitating toxicities.

A deep learning-based method for the prediction of temporal lobe injury in patients with nasopharyngeal carcinoma

PURPOSE

To establish a deep learning-based model to predict radiotherapy-induced temporal lobe injury (TLI).

MATERIALS AND METHODS

Spatial features of dose distribution within the temporal lobe were extracted using both the three-dimensional convolution (C3D) network and the dosiomics method. The Minimal Redundancy-Maximal-Relevance (mRMR) method was employed to rank the extracted features and select the most relevant ones. Four machine learning (ML) classifiers, including logistic regression (LR), k-nearest neighbors (kNN), support vector machines (SVM) and random forest (RF), were used to establish prediction models. Nested sampling and hyperparameter tuning methods were applied to train and validate the prediction models. For comparison, a prediction model base on the conventional D0.5cc of the temporal lobe obtained from dose volume (DV) histogram was established. The area under the receiver operating characteristic (ROC) curve (AUC) was utilized to compare the predictive performance of the different models.

RESULTS

A total of 127 nasopharyngeal carcinoma (NPC) patients were included in the study. In the model based on C3D deep learning features, the highest AUC value of 0.843 was achieved with 5 features. For the dosiomics features model, the highest AUC value of 0.715 was attained with 1 feature. Both of these models demonstrated superior performance compared to the prediction model based on DV parameters, which yielded an AUC of 0.695.

CONCLUSION

The prediction model utilizing C3D deep learning features outperformed models based on dosiomics features or traditional parameters in predicting the onset of TLI. This approach holds promise for predicting radiation-induced toxicities and guide individualized radiotherapy.

Disparate Radiation-Induced Microstructural Injuries in Whole-Brain White Matter of Patients With Nasopharyngeal Carcinoma: A Longitudinal Study Using Multishell Diffusion MRI

BACKGROUND

Evidence for prevention strategies of radiotherapy (RT)-related injury in patients with nasopharyngeal carcinoma (NPC) was lacking. Understanding the dynamic alterations in the cerebral white matter (WM) microstructure after RT may be helpful.

PURPOSE

To investigate the dynamic alterations in the whole brain WM microstructure in patients with NPC in the 12 months after RT using multishell diffusion MRI (MS-dMRI).

STUDY TYPE

Single-center longitudinal study.

POPULATION

A total of 28 treatment-naïve patients with pathologically confirmed NPC (age: 39.68 ± 8.93 years, 11 female) and 20 healthy controls (age: 40.65 ± 9.76 years, 7 female).

FIELD STRENGTH/SEQUENCES

A 3 T, MS-dMRI using a single-shot echo planar imaging sequence.

ASSESSMENT

MS-dMRI was acquired at baseline for the NPC patients and healthy controls, at 0-3 (acute, AC), 6 (early delayed, ED) and 12 months (late delayed, LD) after RT for the NPC patients. The mean and maximum radiation doses to the temporal lobe were acquired. The quality of images was reviewed. MS-dMRI was analyzed using tract-based spatial statistics (TBSS). The presentations of injury were defined by the findings of TBSS.

STATISTICAL TESTS

Chi-square, t tests, repeated ANOVA, and Spearman-rank correlation analysis were used. P < 0.05 was considered to be statistically significant.

RESULTS

TBSS showed two WM injuries (injuries 1 and 2). Injury 1 emerged in the ED phase in the bilateral temporal poles and persisted throughout the ED and LD phases. Injury 2 developed from the AC to ED phase in the bilateral hemisphere and partially recovered in the LD phase. In the ED and LD phases, the multiple diffusion metrics were well correlated (r > 0.5 or <-0.5) with the RT dose, especially in the WM tracts in the temporal lobes.

DATA CONCLUSION

Disparate WM injuries were observed in NPC patients after RT. The injuries may be primarily or secondarily induced by radiation. Injury 1 may be irreversible, while injury 2 seems to partially recover.

EVIDENCE LEVEL

2.

TECHNICAL EFFICACY

Stage 4.

A comprehensive predictive model for radiation-induced brain injury in risk stratification and personalized radiotherapy of nasopharyngeal carcinoma

BACKGROUND AND PURPOSE

Radiation-induced brain injury (RBI) is a severe radiotoxicity for nasopharyngeal carcinoma (NPC) patients, greatly affecting their long-term life quality and survival. We aim to establish a comprehensive predictive model including clinical factors and newly developed genetic variants to improve the precision of RBI risk stratification.

MATERIALS AND METHODS

By performing a large registry-based retrospective study with magnetic resonance imaging follow-up on RBI development, we conducted a genome-wide association study and developed a polygenic risk score (PRS) for RBI in 1189 NPC patients who underwent intensity-modulated radiotherapy. We proposed a tolerance dose scheme for temporal lobe radiation based on the risk predicted by PRS. Additionally, we established a nomogram by combining PRS and clinical factors for RBI risk prediction.

RESULTS

The 38-SNP PRS could effectively identify high-risk individuals of RBI (P = 1.42 × 10-34). Based on genetic risk calculation, the recommended tolerance doses of temporal lobes should be 57.6 Gy for individuals in the top 10 % PRS subgroup and 68.1 Gy for individuals in the bottom 50 % PRS. Notably, individuals with high genetic risk (PRS > P50) and receiving high radiation dose in the temporal lobes (D0.5CC > 65 Gy) had an approximate 50-fold risk over individuals with low PRS and receiving low radiation dose (HR = 50.09, 95 %CI = 24.27-103.35), showing an additive joint effect (Pinteraction < 0.001). By combining PRS with clinical factors including age, tumor stage, and radiation dose of temporal lobes, the predictive accuracy was significantly improved with C-index increased from 0.78 to 0.85 (P = 1.63 × 10-2).

CONCLUSIONS

The PRS, together with clinical factors, could improve RBI risk stratification and implies personalized radiotherapy.

Proton Bragg Peak FLASH Enables Organ Sparing and Ultra-High Dose-Rate Delivery: Proof of Principle in Recurrent Head and Neck Cancer

Proton pencil-beam scanning (PBS) Bragg peak FLASH combines ultra-high dose rate delivery and organ-at-risk (OAR) sparing. This proof-of-principle study compared dosimetry and dose rate coverage between PBS Bragg peak FLASH and PBS transmission FLASH in head and neck reirradiation. PBS Bragg peak FLASH plans were created via the highest beam single energy, range shifter, and range compensator, and were compared to PBS transmission FLASH plans for 6 GyE/fraction and 10 GyE/fraction in eight recurrent head and neck patients originally treated with quad shot reirradiation (14.8/3.7 CGE). The 6 GyE/fraction and 10 GyE/fraction plans were also created using conventional-rate intensity-modulated proton therapy techniques. PBS Bragg peak FLASH, PBS transmission FLASH, and conventional plans were compared for OAR sparing, FLASH dose rate coverage, and target coverage. All FLASH OAR V40 Gy/s dose rate coverage was 90-100% at 6 GyE and 10 GyE for both FLASH modalities. PBS Bragg peak FLASH generated dose volume histograms (DVHs) like those of conventional therapy and demonstrated improved OAR dose sparing over PBS transmission FLASH. All the modalities had similar CTV coverage. PBS Bragg peak FLASH can deliver conformal, ultra-high dose rate FLASH with a two-millisecond delivery of the minimum MU per spot. PBS Bragg peak FLASH demonstrated similar dose rate coverage to PBS transmission FLASH with improved OAR dose-sparing, which was more pronounced in the 10 GyE/fraction than in the 6 GyE/fraction. This feasibility study generates hypotheses for the benefits of FLASH in head and neck reirradiation and developing biological models.

Clinical and genome-wide association analysis of chemoradiation-induced hearing loss in nasopharyngeal carcinoma

Chemoradiation-induced hearing loss (CRIHL) is one of the most devasting side effects for nasopharyngeal carcinoma (NPC) patients, which seriously affects survivors' long-term quality of life. However, few studies have comprehensively characterized the risk factors for CRIHL. In this study, we found that age at diagnosis, tumor stage, and concurrent cisplatin dose were positively associated with chemoradiation-induced hearing loss. We performed a genome-wide association study (GWAS) in 777 NPC patients and identified rs1050851 (within the exon 2 of NFKBIA), a variant with a high deleteriousness score, to be significantly associated with hearing loss risk (HR = 5.46, 95% CI 2.93-10.18, P = 9.51 × 10^-08). The risk genotype of rs1050851 was associated with higher NFKBIA expression, which was correlated with lower cellular tolerance to cisplatin. According to permutation-based enrichment analysis, the variants mapping to 149 hereditary deafness genes were significantly enriched among GWAS top signals, which indicated the genetic similarity between hereditary deafness and CRIHL. Pathway analysis suggested that synaptic signaling was involved in the development of CRIHL. Additionally, the risk score integrating genetic and clinical factors can predict the risk of hearing loss with a relatively good performance in the test set. Collectively, this study shed new light on the etiology of chemoradiation-induced hearing loss, which facilitates high-risk individuals' identification for personalized prevention and treatment.

Dosiomics Risk Model for Predicting Radiation Induced Temporal Lobe Injury and Guiding Individual Intensity-Modulated Radiation Therapy

PURPOSE

We aimed to assess the value of dose distribution-based dosiomics and planning computed tomography-based radiomics to predict radiation-induced temporal lobe injury (TLI) and guide individualized intensity modulated radiation therapy.

METHODS AND MATERIALS

A total of 5599 nasopharyngeal carcinoma patients were enrolled, including 2503, 1072, 988, and 1036 patients in the training, validation, prospective test, and external test cohorts, respectively. The concordance index (C-index) was used to compare the performance of the radiomics and dosiomics models with that of the quantitative analyses of normal tissue effects in the clinic and Wen's models. The predicted TLI-free survival rates of redesigned simulated plans with the same dose-volume histogram but different dose distributions for same patient in a cohort of 30 randomly selected patients were compared by the Wilcoxon matched-pairs signed-rank test.

RESULTS

The radiomics and dosiomics signatures were constructed based on 30 selected computed tomography features and 10 selected dose distribution features, respectively, which were important predictors of TLI-free survival (all P <.001). However, the radiomics signature had a low C-index. The dosiomics risk model combining the dosiomics signature, D_1cc, and age had favorable performance, with C-index values of 0.776, 0.811, 0.805, and 0.794 in the training, validation, prospective test, and external test cohorts, respectively, which were better than those of the quantitative analyses of normal tissue effects in the clinic model and Wen's model (all P <.001). The dosiomics risk model can further distinguish patients in a same risk category divided by other models (all P <.05). Conversely, the other models were unable to separate populations classified by the dosiomics risk model (all P > .05). Two simulated plans with the same dose-volume histogram but different dose distributions had different TLI-free survival rates predicted by dosiomics risk model (all P ≤ .002).

CONCLUSIONS

The dosiomics risk model was superior to traditional models in predicting the risk of TLI. This is a promising approach to precisely predict radiation-induced toxicities and guide individualized intensity modulated radiation therapy.

Deep learning-based precise prediction and early detection of radiation-induced temporal lobe injury for nasopharyngeal carcinoma

Background

Radiotherapy is the mainstay of treatment for nasopharyngeal carcinoma. Radiation-induced temporal lobe injury (TLI) can regress or resolve in the early phase, but it is irreversible at a later stage. However, no study has proposed a risk-based follow-up schedule for its early detection. Planning evaluation is difficult when dose-volume histogram (DVH) parameters are similar and optimization is terminated.

Methods

This multicenter retrospective study included 6065 patients between 2014 and 2018. A 3D ResNet-based deep learning model was developed in training and validation cohorts and independently tested using concordance index in internal and external test cohorts. Accordingly, the patients were stratified into risk groups, and the model-predicted risks were used to develop risk-based follow-up schedules. The schedule was compared with the Radiation Therapy Oncology Group (RTOG) recommendation (every 3 months during the first 2 years and every 6 months in 3-5 years). Additionally, the model was used to evaluate plans with similar DVH parameters.

Findings

Our model achieved concordance indexes of 0.831, 0.818, and 0.804, respectively, which outperformed conventional prediction models (all P < 0.001). The temporal lobes in all the cohorts were stratified into three groups with discrepant TLI-free survival. Personalized follow-up schedules developed for each risk group could detect TLI 1.9 months earlier than the RTOG recommendation. According to a higher median predicted 3-year TLI-free survival (99.25% vs. 99.15%, P < 0.001), the model identified a better plan than previous models.

Interpretation

The deep learning model predicted TLI more precisely. The model-determined risk-based follow-up schedule detected the TLI earlier. The planning evaluation was refined because the model identified a better plan with a lower risk of TLI.

Funding

The Sun Yat-sen University Clinical Research 5010 Program (2015020), Guangdong Basic and Applied Basic Research Foundation (2022A1515110356), Medical Scientific Research Foundation of Guangdong Province (A2022367), and Guangzhou Science and Technology Program (2023A04J1788).

Increased relative biological effectiveness and periventricular radiosensitivity in proton therapy of glioma patients

PURPOSE

Currently, there is an intense debate on variations in intra-cerebral radiosensitivity and relative biological effectiveness (RBE) in proton therapy of primary brain tumours. Here, both effects were retrospectively investigated using late radiation-induced brain injuries (RIBI) observed in follow-up after proton therapy of patients with diagnosed glioma.

METHODS

In total, 42 WHO grade 2-3 glioma patients out of a consecutive patient cohort having received (adjuvant) proton radio(chemo)therapy between 2014 and 2017 were eligible for analysis. RIBI lesions (symptomatic or clinically asymptomatic) were diagnosed and delineated on contrast-enhanced T1-weighted magnetic resonance imaging scans obtained in the first two years of follow-up. Correlation of RIBI location and occurrence with dose (D), proton dose-averaged linear energy transfer (LET) and variable RBE dose parameters were tested in voxel- and in patient-wise logistic regression analyses. Additionally, anatomical and clinical parameters were considered. Model performance was estimated through cross-validated area-under-the-curve (AUC) values.

RESULTS

In total, 64 RIBI lesions were diagnosed in 21 patients. The median time between start of proton radio(chemo)therapy and RIBI appearance was 10.2 months. Median distances of the RIBI volume centres to the cerebral ventricles and to the clinical target volume border were 2.1 mm and 1.3 mm, respectively. In voxel-wise regression, the multivariable model with D, D × LET and periventricular region (PVR) revealed the highest AUC of 0.90 (95 % confidence interval: 0.89-0.91) while the corresponding model without D × LET revealed a value of 0.84 (0.83-0.86). In patient-level analysis, the equivalent uniform dose (EUD₁₁, a = 11) in the PVR using a variable RBE was the most prominent predictor for RIBI with an AUC of 0.63 (0.32-0.90).

CONCLUSIONS

In this glioma cohort, an increased radiosensitivity within the PVR was observed as well as a spatial correlation of RIBI with an increased RBE. Both need to be considered when delivering radio(chemo)therapy using proton beams.

Added value of histogram analysis of ADC in predicting radiation-induced temporal lobe injury of patients with nasopharyngeal carcinoma treated by intensity-modulated radiotherapy

BACKGROUND

This study evaluated the predictive potential of histogram analysis derived from apparent diffusion coefficient (ADC) maps in radiation-induced temporal lobe injury (RTLI) of nasopharyngeal carcinoma (NPC) after intensity-modulated radiotherapy (IMRT).

RESULTS

Pretreatment diffusion-weighted imaging (DWI) of the temporal lobes of 214 patients with NPC was retrospectively analyzed to obtain ADC histogram parameters. Of the 18 histogram parameters derived from ADC maps, 7 statistically significant variables in the univariate analysis were included in the multivariate logistic regression analysis. The final best prediction model selected by backward stepwise elimination with Akaike information criteria as the stopping rule included kurtosis, maximum energy, range, and total energy. A Rad-score was established by combining the four variables, and it provided areas under the curve (AUCs) of 0.95 (95% confidence interval [CI] 0.91-0.98) and 0.89 (95% CI 0.81-0.97) in the training and validation cohorts, respectively. The combined model, integrating the Rad-score with the T stage (p = 0.02), showed a favorable prediction performance in the training and validation cohorts (AUC = 0.96 and 0.87, respectively). The calibration curves showed a good agreement between the predicted and actual RTLI occurrences.

CONCLUSIONS

Pretreatment histogram analysis of ADC maps and their combination with the T stage showed a satisfactory ability to predict RTLI in NPC after IMRT.

Nomogram Based on Clinical and Radiomics Data for Predicting Radiation-induced Temporal Lobe Injury in Patients with Non-metastatic Stage T4 Nasopharyngeal Carcinoma

AIMS

To use pre-treatment magnetic resonance imaging-based radiomics data with clinical data to predict radiation-induced temporal lobe injury (RTLI) in nasopharyngeal carcinoma (NPC) patients with stage T4/N0-3/M0 within 5 years after radiotherapy.

MATERIALS AND METHODS

This study retrospectively examined 98 patients (198 temporal lobes) with stage T4/N0-3/M0 NPC. Participants were enrolled into a training cohort or a validation cohort in a ratio of 7:3. Radiomics features were extracted from pre-treatment magnetic resonance imaging that were T1-and T2-weighted. Spearman rank correlation, the t-test and the least absolute shrinkage and selection operator (LASSO) algorithm were used to select significant radiomics features; machine-learning models were used to generate radiomics signatures (Rad-Scores). Rad-Scores and clinical factors were integrated into a nomogram for prediction of RTLI. Nomogram discrimination was evaluated using receiver operating characteristic analysis and clinical benefits were evaluated using decision curve analysis.

RESULTS

Participants were enrolled into a training cohort (n = 139) or a validation cohort (n = 59). In total, 3568 radiomics features were initially extracted from T1-and T2-weighted images. Age, D_max, D_1cc and 16 stable radiomics features (six from T1-weighted and 10 from T2-weighted images) were identified as independent predictive factors. A greater Rad-Score was associated with a greater risk of RTLI. The nomogram showed good discrimination, with a C-index of 0.85 (95% confidence interval 0.79-0.92) in the training cohort and 0.82 (95% confidence interval 0.71-0.92) in the validation cohort.

CONCLUSION

We developed models for the prediction of RTLI in patients with stage T4/N0-3/M0 NPC using pre-treatment radiomics data and clinical data. Nomograms from these pre-treatment data improved the prediction of RTLI. These results may allow the selection of patients for earlier clinical interventions.

Magnetic resonance imaging-based radiomics model for predicting radiation-induced temporal lobe injury in nasopharyngeal carcinoma after intensity-modulated radiotherapy

BACKGROUND

To develop a model based on magnetic resonance imaging (MRI) radiomics and clinical features for predicting radiation-induced temporal lobe injury (RTLI) in patients with nasopharyngeal carcinoma (NPC) after intensity-modulated radiotherapy (IMRT).

METHODS

Two hundred and sixteen patients with NPC were retrospectively included. Radiomics features were extracted and selected. The logistic regression analysis was performed for prediction models construction. The area under the receiver operating characteristic curve (AUC) was calculated for performance evaluation.

RESULTS

Three radiomics features were selected to construct the radiomics signature (AUC of 0.94 and 0.92). The clinical-radiomics model, integrating radiomics signature with T classification, achieved higher predictive performance in the training and validation cohorts (AUC of 0.95 and 0.93), as well as improved accuracy of the classification of RTLI outcomes (net reclassification improvement: 0.711; 95% CI: 0.57-0.86; p < 0.001).

CONCLUSIONS

The clinical-radiomics model and radiomics signature both showed great performance in predicting RTLI in patients with NPC.

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.

Impact of RBE variations on risk estimates of temporal lobe necrosis in patients treated with intensity-modulated proton therapy for head and neck cancer

BACKGROUND

Temporal lobe necrosis (TLN) is a potential late effect after radiotherapy for skull base head and neck cancer (HNC). Several photon-derived dose constraints and normal tissue complication probability (NTCP) models have been proposed, however variation in relative biological effectiveness (RBE) may challenge the applicability of these dose constraints and models in proton therapy. The purpose of this study was therefore to investigate the influence of RBE variations on risk estimates of TLN after Intensity-Modulated Proton Therapy for HNC.

MATERIAL AND METHODS

Seventy-five temporal lobes from 45 previously treated patients were included in the analysis. Sixteen temporal lobes had radiation associated Magnetic Resonance image changes (TLIC) suspected to be early signs of TLN. Fixed (RWD_Fix) and variable RBE-weighed doses (RWD_Var) were calculated using RBE = 1.1 and two RBE models, respectively. RWD_Fix and RWD_Var for temporal lobes were compared using Friedman's test. Based on RWD_Fix, six NTCP models were fitted and internally validated through bootstrapping. Estimated probabilities from RWD_Fix and RWD_Var were compared using paired Wilcoxon test. Seven dose constraints were evaluated separately for RWD_Fix and RWD_Var by calculating the observed proportion of TLIC in temporal lobes meeting the specific dose constraints.

RESULTS

RWD_Var were significantly higher than RWD_Fix (p < 0.01). NTCP model performance was good (AUC:0.79-0.84). The median difference in estimated probability between RWD_Fix and RWD_Var ranged between 5.3% and 20.0% points (p < 0.01), with V_60GyRBE and D_Max at the smallest and largest differences, respectively. The proportion of TLIC was higher for RWD_Fix (4.0%-13.1%) versus RWD_Var (1.3%-5.3%). For V_65GyRBE ≤ 0.03 cc the proportion of TLIC was less than 5% for both RWD_Fix and RWD_Var.

CONCLUSION

NTCP estimates were significantly influenced by RBE variations. D_max as model predictor resulted in the largest deviations in risk estimates between RWD_Fix and RWD_Var. V_65GyRBE ≤ 0.03 cc was the most consistent dose constraint for RWD_Fix and RWD_Var.

Application of Artificial Intelligence for Nasopharyngeal Carcinoma Management - A Systematic Review

INTRODUCTION

Nasopharyngeal carcinoma (NPC) is endemic to Eastern and South-Eastern Asia, and, in 2020, 77% of global cases were diagnosed in these regions. Apart from its distinct epidemiology, the natural behavior, treatment, and prognosis are different from other head and neck cancers. With the growing trend of artificial intelligence (AI), especially deep learning (DL), in head and neck cancer care, we sought to explore the unique clinical application and implementation direction of AI in the management of NPC.

METHODS

The search protocol was performed to collect publications using AI, machine learning (ML) and DL in NPC management from PubMed, Scopus and Embase. The articles were filtered using inclusion and exclusion criteria, and the quality of the papers was assessed. Data were extracted from the finalized articles.

RESULTS

A total of 78 articles were reviewed after removing duplicates and papers that did not meet the inclusion and exclusion criteria. After quality assessment, 60 papers were included in the current study. There were four main types of applications, which were auto-contouring, diagnosis, prognosis, and miscellaneous applications (especially on radiotherapy planning). The different forms of convolutional neural networks (CNNs) accounted for the majority of DL algorithms used, while the artificial neural network (ANN) was the most frequent ML model implemented.

CONCLUSION

There is an overall positive impact identified from AI implementation in the management of NPC. With improving AI algorithms, we envisage AI will be available as a routine application in a clinical setting soon.

Prediction and verification of survival in patients with non-small-cell lung cancer based on an integrated radiomics nomogram

AIM

To develop and validate a nomogram to predict 1-, 2-, and 5-year survival in patients with non-small-cell lung cancer (NSCLC) by combining optimised radiomics features, clinicopathological factors, and conventional image features extracted from three-dimensional (3D) computed tomography (CT) images.

MATERIALS AND METHODS

A total of 172 patients with NSCLC were selected to construct the model, and 74 and 72 patients were selected for internal validation and external testing, respectively. A total of 828 radiomics features were extracted from each patient's 3D CT images. Univariable Cox regression and least absolute shrinkage and selection operator (LASSO) regression were used to select features and generate a radiomics signature (radscore). The performance of the nomogram was evaluated by calibration curves, clinical practicability, and the c-index. Kaplan-Meier (KM) analysis was used to compare the overall survival (OS) between the two subgroups.

RESULT

The radiomics features of the NSCLC patients correlated significantly with survival time. The c-indexes of the nomogram in the training cohort, internal validation cohort, and external test cohort were 0.670, 0.658, and 0.660, respectively. The calibration curves showed that the predicted survival time was close to the actual survival time. Decision curve analysis shows that the nomogram could be useful in the clinic. According to KM analysis, the 1-, 2- and 5-year survival rates of the low-risk group were higher than those of the high-risk group.

CONCLUSION

The nomogram, combining the radscore, clinicopathological factors, and conventional CT parameters, can improve the accuracy of survival prediction in patients with NSCLC.

Cognitive Decline following Radiotherapy of Head and Neck Cancer: Systematic Review and Meta-Analysis of MRI Correlates

Radiotherapy for head and neck cancers exposes small parts of the brain to radiation, resulting in radiation-induced changes in cerebral tissue. In this review, we determine the correlation between cognitive deterioration in patients with head and neck cancer after radiotherapy and magnetic resonance imaging (MRI) changes. Systematic searches were performed in PubMed, Scopus, and Cochrane databases in February 2021. Studies of head and neck cancer patients treated with radiotherapy and periodical cognitive and MRI assessments were included. Meta-analysis was performed to analyse the correlation of Montreal Cognitive Assessment (MoCA) scores to MRI structural and functional changes. Seven studies with a total of 404 subjects (irradiated head and neck patients, n = 344; healthy control, n = 60) were included. Most studies showed the significance of MRI in detecting microstructural and functional changes in association with neurocognitive function. The changes were seen in various brain areas, predominantly the temporal region, which also shows dose dependency (6/7 studies). An effect size (r = 0.43, p < 0.001) was reported on the correlation of MoCA scores to MRI structural and functional changes with significant correlations shown among patients treated with head and neck radiotherapy. However, the effect size appears modest.