Noorazrul Yahya

Dosimetry, clinical factors and medication intake influencing urinary symptoms after prostate radiotherapy: An analysis of data from the RADAR prostate radiotherapy trial

PURPOSE/OBJECTIVE To identify dosimetry, clinical factors and medication intake impacting urinary symptoms after prostate radiotherapy. MATERIAL AND METHODS Data describing clinical factors and bladder dosimetry (reduced with principal component (PC) analysis) for 754 patients treated with external beam radiotherapy accrued by TROG 03.04 RADAR prostate radiotherapy trial were available for analysis. Urinary symptoms (frequency, incontinence, dysuria and haematuria) were prospectively assessed using LENT-SOMA to a median of 72months. The endpoints assessed were prevalence (grade ⩾1) at the end of radiotherapy (representing acute symptoms), at 18-, 36- and 54-month follow-ups (representing late symptoms) and peak late incidence including only grade ⩾2. Impact of factors was assessed using multivariate logistic regression models with correction for over-optimism. RESULTS Baseline symptoms, non-insulin dependent diabetes mellitus, age and PC1 (correlated to the mean dose) impact symptoms at >1 timepoints. Associations at a single timepoint were found for cerebrovascular condition, ECOG status and non-steroidal anti-inflammatory drug intake. Peak incidence analysis shows the impact of baseline, bowel and cerebrovascular condition and smoking status. CONCLUSIONS The prevalence and incidence analysis provide a complementary view for urinary symptom prediction. Sustained impacts across time points were found for several factors while some associations were not repeated at different time points suggesting poorer or transient impact.

Modeling Urinary Dysfunction After External Beam Radiation Therapy of the Prostate Using Bladder Dose-Surface Maps: Evidence of Spatially Variable Response of the Bladder Surface

PURPOSE We assessed the association of the spatial distribution of dose to the bladder surface, described using dose-surface maps, with the risk of urinary dysfunction. METHODS AND MATERIALS The bladder dose-surface maps of 754 participants from the TROG 03.04-RADAR trial were generated from the volumetric data by virtually cutting the bladder at the sagittal slice, intersecting the bladder center-of-mass through to the bladder posterior and projecting the dose information on a 2-dimensional plane. Pixelwise dose comparisons were performed between patients with and without symptoms (dysuria, hematuria, incontinence, and an International Prostate Symptom Score increase of ≥10 [ΔIPSS10]). The results with and without permutation-based multiple-comparison adjustments are reported. The pixelwise multivariate analysis findings (peak-event model for dysuria, hematuria, and ΔIPSS10; event-count model for incontinence), with adjustments for clinical factors, are also reported. RESULTS The associations of the spatially specific dose measures to urinary dysfunction were dependent on the presence of specific symptoms. The doses received by the anteroinferior and, to lesser extent, posterosuperior surface of the bladder had the strongest relationship with the incidence of dysuria, hematuria, and ΔIPSS10, both with and without adjustment for clinical factors. For the doses to the posteroinferior region corresponding to the area of the trigone, the only symptom with significance was incontinence. CONCLUSIONS A spatially variable response of the bladder surface to the dose was found for symptoms of urinary dysfunction. Limiting the dose extending anteriorly might help reduce the risk of urinary dysfunction.

Urinary symptoms following external beam radiotherapy of the prostate: Dose–symptom correlates with multiple-event and event-count models

BACKGROUND AND PURPOSE This study aimed to compare urinary dose-symptom correlates after external beam radiotherapy of the prostate using commonly utilised peak-symptom models to multiple-event and event-count models which account for repeated events. MATERIALS AND METHODS Urinary symptoms (dysuria, haematuria, incontinence and frequency) from 754 participants from TROG 03.04-RADAR trial were analysed. Relative (R1-R75 Gy) and absolute (A60-A75Gy) bladder dose-surface area receiving more than a threshold dose and equivalent uniform dose using exponent a (range: a ∈[1 … 100]) were derived. The dose-symptom correlates were analysed using; peak-symptom (logistic), multiple-event (generalised estimating equation) and event-count (negative binomial regression) models. RESULTS Stronger dose-symptom correlates were found for incontinence and frequency using multiple-event and/or event-count models. For dysuria and haematuria, similar or better relationships were found using peak-symptom models. Dysuria, haematuria and high grade (⩾ 2) incontinence were associated to high dose (R61-R71 Gy). Frequency and low grade (⩾ 1) incontinence were associated to low and intermediate dose-surface parameters (R13-R41Gy). Frequency showed a parallel behaviour (a=1) while dysuria, haematuria and incontinence showed a more serial behaviour (a=4 to a ⩾ 100). Relative dose-surface showed stronger dose-symptom associations. CONCLUSIONS For certain endpoints, the multiple-event and event-count models provide stronger correlates over peak-symptom models. Accounting for multiple events may be advantageous for a more complete understanding of urinary dose-symptom relationships.

Statistical-learning strategies generate only modestly performing predictive models for urinary symptoms following external beam radiotherapy of the prostate: A comparison of conventional and machine-learning methods

PURPOSE Given the paucity of available data concerning radiotherapy-induced urinary toxicity, it is important to ensure derivation of the most robust models with superior predictive performance. This work explores multiple statistical-learning strategies for prediction of urinary symptoms following external beam radiotherapy of the prostate. METHODS The performance of logistic regression, elastic-net, support-vector machine, random forest, neural network, and multivariate adaptive regression splines (MARS) to predict urinary symptoms was analyzed using data from 754 participants accrued by TROG03.04-RADAR. Predictive features included dose-surface data, comorbidities, and medication-intake. Four symptoms were analyzed: dysuria, haematuria, incontinence, and frequency, each with three definitions (grade ≥ 1, grade ≥ 2 and longitudinal) with event rate between 2.3% and 76.1%. Repeated cross-validations producing matched models were implemented. A synthetic minority oversampling technique was utilized in endpoints with rare events. Parameter optimization was performed on the training data. Area under the receiver operating characteristic curve (AUROC) was used to compare performance using sample size to detect differences of ≥0.05 at the 95% confidence level. RESULTS Logistic regression, elastic-net, random forest, MARS, and support-vector machine were the highest-performing statistical-learning strategies in 3, 3, 3, 2, and 1 endpoints, respectively. Logistic regression, MARS, elastic-net, random forest, neural network, and support-vector machine were the best, or were not significantly worse than the best, in 7, 7, 5, 5, 3, and 1 endpoints. The best-performing statistical model was for dysuria grade ≥ 1 with AUROC ± standard deviation of 0.649 ± 0.074 using MARS. For longitudinal frequency and dysuria grade ≥ 1, all strategies produced AUROC>0.6 while all haematuria endpoints and longitudinal incontinence models produced AUROC<0.6. CONCLUSIONS Logistic regression and MARS were most likely to be the best-performing strategy for the prediction of urinary symptoms with elastic-net and random forest producing competitive results. The predictive power of the models was modest and endpoint-dependent. New features, including spatial dose maps, may be necessary to achieve better models.

Is Diagnostic Performance of Quantitative 2D-Shear Wave Elastography Optimal for Clinical Classification of Benign and Malignant Thyroid nodules? A Systematic Review and Meta-Analysis

RATIONALE AND OBJECTIVE This study is a dedicated 2D-shear wave elastography (2D-SWE) review aimed at systematically eliciting up-to-date evidence of its clinical value in differential diagnosis of benign and malignant thyroid nodules. METHODS PubMed, Web of Science, and Scopus databases were searched for studies assessing the diagnostic value of 2D-SWE for thyroid malignancy risk stratification published until December 2016. The retrieved titles and abstracts were screened and evaluated according to the predefined inclusion and exclusion criteria. Methodological quality of the studies was assessed using the Quality Assessment of Studies of Diagnostic Accuracy included in Systematic Review 2 (QUADAS-2) tool. Extracted 2D-SWE diagnostic performance data were meta-analyzed to assess the summary sensitivity, specificity, and area under the receiver operating characteristic curve. RESULTS After stepwise review, 14 studies in which 2D-SWE was used to evaluate 2851 thyroid nodules (1092 malignant, 1759 benign) from 2139 patients were selected for the current study. Study quality on QUADAS-2 assessment was moderate to high. The summary sensitivity, specificity and area under the receiver operating characteristic curve of 2D-SWE for differential diagnosis of benign and malignant thyroid nodules were 0.66 (95% confidence interval [CI]: 0.64-0.69), 0.78 (CI: 0.76-0.80), and 0.851 (Q* = 0.85), respectively. The pooled diagnostic odds ratio, negative likelihood ratio, and positive likelihood ratio were 12.73 (CI: 8.80-18.43), 0.31 (CI: 0.22-0.44), and 3.87 (CI: 2.83-5.29), respectively. CONCLUSION Diagnostic performance of quantitative 2D-SWE for malignancy risk stratification of thyroid nodules is suboptimal with mediocre sensitivity and specificity, contrary to earlier reports of excellence.

OC-0254: Dosimetric predictors for urinary symptoms using longitudinal endpoint and multiple events models

Purpose/Objective: As urinary symptoms tend to recur throughout follow-up, conventional method of analysis using cumulative peak event and time-to-event may not be optimal to uncover the dosimetric-symptom correlates. We assessed the bladder dosimetry and urinary symptom correlates using longitudinally-defined endpoints and using recurrent event models which are contrasted to the conventional analysis methods. Materials and Methods: In this study, 754 dose-surface information and their corresponding specific urinary symptoms (dysuria (D), haematuria (H), incontinence (I) and frequency (F)) from a cohort of patients who received prostate radiotherapy in the RADAR TROG 03.04 trial were analysed. The dosimetric-symptom correlates were analysed using; 1) conventional methods (cumulative incidence(peak) and time-to-event(Cox) analysis), 2) longitudinally-defined endpoint (mean symptoms), 3) recurrent event models using the Andersen-Gill extension of the Cox regression model for counting process (AG) & generalised estimating equation (GEE) models. Dosimetric-symptom correlates were contrasted for the different analytic methods. Results: For dysuria and haematuria, stronger relationships were found to the dose indices using peak and Cox models compared to mean symptom, AG and GEE models. Despite the different strength of relationship, dose-surface of the bladder receiving higher than 65 Gy (S65) and S70 consistently show strong relationship to dysuria. S60 to S65 are the most significant for Hpeak , HGEE, HCox and Hmean. None of the dosimetric indices satisfy the proportional hazard assumption for HAG. For urinary incontinence and frequency, stronger relationships for dosimetric indices were found for AG, GEE and to lesser extent mean score model while both peak and Cox models do not result in significant or show trend towards significance. S35 to S40 were found to be the most significant for FGEE, Fmean and FAG while S20 to S25 for IAG and Imean. Conclusions: The use of peak or time-to-event model alone is not optimal in assessing dose-volume correlates for certain urinary symptoms endpoints. Dosimetric-symptom correlates analysis should be supplemented by longitudinally-defined endpoints and/or using recurrent event models to account for multiple events per patient.