Background and purpose Incontinence, hematuria, voiding frequency and pain during voiding are possible side effects of radiotherapy among patients treated for prostate cancer. The objective of this study was to develop multivariable NTCP models for these side effects. Material and methods This prospective cohort study was composed of 243 patients with localized or locally advanced prostate cancer (stage T1-3). Genito-urinary (GU) toxicity was assessed using a standardized follow-up program. The GU toxicity endpoints were scored using the Common Terminology Criteria for Adverse Events version 3.0 (CTCAE 3.0) scoring system. The full bladder and different anatomical subregions within the bladder were delineated. A least absolute shrinkage and selection operator (LASSO) logistic regression analysis was used to analyze dose volume effects on the four individual endpoints. Results In the univariable analysis, urinary incontinence was significantly associated with dose distributions in the trigone (V55-V75, mean). Hematuria was significantly associated with the bladder wall dose (V40-V75, mean), bladder dose (V70-V75), cardiovascular disease and anticoagulants use. Pain during urinating was associated with the dose to the trigone (V50-V75, mean) and with trans transurethral resection of the prostate (TURP). In the final multivariable model urinary incontinence was associated with the mean dose of the trigone. Hematuria was associated with bladder wall dose (V75) and cardiovascular disease, while pain during urinating was associated with trigone dose (V75) and TURP. No significant associations were found for increase in voiding frequency. Conclusions Radiation-induced urinary side effects are associated with dose distributions to different organs as risk. Given the dose effect relationships found, decreasing the dose to the trigone and bladder wall may reduce the incidence of incontinence, pain during voiding and hematuria, respectively.

Objectives: Establish a method to determine and convey lifetime radiation risk from FFDM screening. Methods: Radiation risk from screening mammography was quantified using effective risk (number of radiation induced cancer cases/million). For effective risk calculations, organ doses and examined breast MGD were used. Screening mammography was simulated by exposing a breast phantom for cranio-caudal and medio-lateral oblique for each breast using 16 FFDM machines. An ATOM phantom loaded with TLD dosimeters was positioned in contact with the breast phantom to simulate the client's body. Effective risk data were analysed using SPSS software to establish a regression model to predict the effective risk of any screening programme. Graphs were generated to extrapolate the effective risk of all screening programmes for a range of commencement ages and time intervals between screens. Results: The most important parameters controlling clients' total effective risk within breast screening are the screening commencement age and number of screens (correlation coefficients were-0.865 and 0.714, respectively). Since the tissue radio-sensitivity reduces with age, the end age of screening does not result in noteworthy effect on total effective risk. Conclusions: The regression model can be used to predict the total effective risk for clients within breast screening but it cannot be used for exact assessment of total effective risk. Graphical representation of risk could be an easy way to represent risk in a fashion which might be helpful to clients and clinicians.

Background and purpose Curative radiotherapy for prostate cancer may lead to anorectal side effects, including rectal bleeding, fecal incontinence, increased stool frequency and rectal pain. The main objective of this study was to develop multivariable NTCP models for these side effects. Material and methods The study sample was composed of 262 patients with localized or locally advanced prostate cancer (stage T1–3). Anorectal toxicity was prospectively assessed using a standardized follow-up program. Different anatomical subregions within and around the anorectum were delineated. A LASSO logistic regression analysis was used to analyze dose volume effects on toxicity. Results In the univariable analysis, rectal bleeding, increase in stool frequency and fecal incontinence were significantly associated with a large number of dosimetric parameters. The collinearity between these predictors was high (VIF > 5). In the multivariable model, rectal bleeding was associated with the anorectum (V70) and anticoagulant use, fecal incontinence was associated with the external sphincter (V15) and the iliococcygeal muscle (V55). Finally, increase in stool frequency was associated with the iliococcygeal muscle (V45) and the levator ani (V40). No significant associations were found for rectal pain. Conclusions Different anorectal side effects are associated with different anatomical substructures within and around the anorectum. The dosimetric variables associated with these side effects can be used to optimize radiotherapy treatment planning aiming at prevention of specific side effects and to estimate the benefit of new radiation technologies.

Purpose/Objective(s)Model selection, i.e., choosing predictor variables, is a crucial step in the development of Normal Tissue Complication Probability models. Conventional methods use available knowledge, from biological insight or literature, and fit a single selected model to the data; however, this approach discards the possibility to improve the model by finding better predictors from the data. Data-driven model selection, on the other hand, is inherently unstable, especially if candidate predictors are collinear and if the number of events in the dataset is low, leading potentially to spurious and overoptimistic models. This study compares knowledge-based and data-driven model selection for rectal toxicity using simulation.Materials/MethodsWe used prospective data of 262 prostate cancer patients treated with intensity modulated radiation therapy (39 x 2 Gy) in our hospital. Four-year incidences of 3 rectal toxicity outcomes were scored (see Table 1). Independent variables included mean dose (MD) and dose-volume parameters (Vx) of structures in the rectal region. For each endpoint, a knowledge-based (H0) model and multiple data-driven (H1) models were fitted to the data (see Table 1). The performance of each H1 model compared to H0 was measured as the difference in Akiake Information Criterion (ΔAIC). This procedure was subsequently repeated many times with simulated datasets where the actual outcomes were replaced with random values with probabilities matching the actual H0 model. For each simulation, the maximum ΔAIC (ΔAICmax) was obtained, resulting in the expected distribution of ΔAICmax under the assumption that H0 is correct. The actual ΔAIC values were compared with the ΔAICmax distribution to give a P value, i.e., the probability, to find a value as extreme as the actual ΔAIC if H0 is true. We rejected H0 in favor of H1 if P < .05. Note that such a hypothesis test cannot be made with common resampling methods.ResultsePoster Abstracts 1062; Table 1Endpointrectal bleeding (CTCAE 3 Grade>2)incontinence(CTCAE 3 Grade>2)increased stool frequency(>baseline+3)Observations254254252Events121929H0 ModelR V65ES MD + PR MDA MDAUC H0 Model0.880.820.58No. of H1 models2885108ΔAIC threshold for P < 0.05-4.82-6.37-7.99Significant H1 modelsAR V70 (P = 0.001)ES V15 (P = 0.03)IC V45 (P < 0.01)R V70 (P = 0.001)LA V45 (P < 0.01)max AUC H1 Model0.910.840.79AUC = area under the ROC curve; R = rectum; AR = anorectum; A = anal canal; ES = external sphincter; PR = puborectal muscle; LA = levator ani muscles; IC = iliococcygeal muscle Open table in a new tab ConclusionThe presented method enables us to discern genuine from spurious data-driven model improvements, taking the actual selection procedure and collinearity of the data into account. Purpose/Objective(s)Model selection, i.e., choosing predictor variables, is a crucial step in the development of Normal Tissue Complication Probability models. Conventional methods use available knowledge, from biological insight or literature, and fit a single selected model to the data; however, this approach discards the possibility to improve the model by finding better predictors from the data. Data-driven model selection, on the other hand, is inherently unstable, especially if candidate predictors are collinear and if the number of events in the dataset is low, leading potentially to spurious and overoptimistic models. This study compares knowledge-based and data-driven model selection for rectal toxicity using simulation. Model selection, i.e., choosing predictor variables, is a crucial step in the development of Normal Tissue Complication Probability models. Conventional methods use available knowledge, from biological insight or literature, and fit a single selected model to the data; however, this approach discards the possibility to improve the model by finding better predictors from the data. Data-driven model selection, on the other hand, is inherently unstable, especially if candidate predictors are collinear and if the number of events in the dataset is low, leading potentially to spurious and overoptimistic models. This study compares knowledge-based and data-driven model selection for rectal toxicity using simulation. Materials/MethodsWe used prospective data of 262 prostate cancer patients treated with intensity modulated radiation therapy (39 x 2 Gy) in our hospital. Four-year incidences of 3 rectal toxicity outcomes were scored (see Table 1). Independent variables included mean dose (MD) and dose-volume parameters (Vx) of structures in the rectal region. For each endpoint, a knowledge-based (H0) model and multiple data-driven (H1) models were fitted to the data (see Table 1). The performance of each H1 model compared to H0 was measured as the difference in Akiake Information Criterion (ΔAIC). This procedure was subsequently repeated many times with simulated datasets where the actual outcomes were replaced with random values with probabilities matching the actual H0 model. For each simulation, the maximum ΔAIC (ΔAICmax) was obtained, resulting in the expected distribution of ΔAICmax under the assumption that H0 is correct. The actual ΔAIC values were compared with the ΔAICmax distribution to give a P value, i.e., the probability, to find a value as extreme as the actual ΔAIC if H0 is true. We rejected H0 in favor of H1 if P < .05. Note that such a hypothesis test cannot be made with common resampling methods. We used prospective data of 262 prostate cancer patients treated with intensity modulated radiation therapy (39 x 2 Gy) in our hospital. Four-year incidences of 3 rectal toxicity outcomes were scored (see Table 1). Independent variables included mean dose (MD) and dose-volume parameters (Vx) of structures in the rectal region. For each endpoint, a knowledge-based (H0) model and multiple data-driven (H1) models were fitted to the data (see Table 1). The performance of each H1 model compared to H0 was measured as the difference in Akiake Information Criterion (ΔAIC). This procedure was subsequently repeated many times with simulated datasets where the actual outcomes were replaced with random values with probabilities matching the actual H0 model. For each simulation, the maximum ΔAIC (ΔAICmax) was obtained, resulting in the expected distribution of ΔAICmax under the assumption that H0 is correct. The actual ΔAIC values were compared with the ΔAICmax distribution to give a P value, i.e., the probability, to find a value as extreme as the actual ΔAIC if H0 is true. We rejected H0 in favor of H1 if P < .05. Note that such a hypothesis test cannot be made with common resampling methods. ResultsePoster Abstracts 1062; Table 1Endpointrectal bleeding (CTCAE 3 Grade>2)incontinence(CTCAE 3 Grade>2)increased stool frequency(>baseline+3)Observations254254252Events121929H0 ModelR V65ES MD + PR MDA MDAUC H0 Model0.880.820.58No. of H1 models2885108ΔAIC threshold for P < 0.05-4.82-6.37-7.99Significant H1 modelsAR V70 (P = 0.001)ES V15 (P = 0.03)IC V45 (P < 0.01)R V70 (P = 0.001)LA V45 (P < 0.01)max AUC H1 Model0.910.840.79AUC = area under the ROC curve; R = rectum; AR = anorectum; A = anal canal; ES = external sphincter; PR = puborectal muscle; LA = levator ani muscles; IC = iliococcygeal muscle Open table in a new tab ConclusionThe presented method enables us to discern genuine from spurious data-driven model improvements, taking the actual selection procedure and collinearity of the data into account. The presented method enables us to discern genuine from spurious data-driven model improvements, taking the actual selection procedure and collinearity of the data into account.

PURPOSE:To determine the impact of late radiation-induced toxicity on health-related quality of life (HRQoL) among patients with prostate cancer.PATIENTS AND METHODS:The study sample was composed of 227 patients, treated with external beam radiotherapy. Common Terminology Criteria for Adverse Events version 3.0 were used to grade late genitourinary and gastrointestinal toxicity. The European Organization for Research and Treatment of Cancer Quality of life Questionnaire C30 (EORTC QLQ-C30) was used to assess HRQoL at baseline, and 6, 12 and 24 months after completion of radiotherapy. Statistical analysis was performed using a multivariate analysis of variance (MANOVA).RESULTS:Urinary incontinence and rectal discomfort significantly affected HRQoL. The impact of urinary incontinence on HRQoL was most pronounced 6 months after radiotherapy and gradually decreased over time. The impact of rectal discomfort on HRQoL was predominant at 6 months after radiotherapy, decreased at 12 months and increased again 2 years after radiotherapy. No significant impact on HRQoL was observed for any of the other toxicity endpoints, or non-toxicity related factors such as hormonal therapy, radiotherapy technique or age.CONCLUSION:Urinary incontinence and rectal discomfort have a significant impact on HRQoL. Prevention of these side effects may likely improve quality of life of prostate cancer patients after completion of treatment.

Quality of life (QoL) of cancer patients after treatment may not only be affected by the treatment itself, but also by non-treatment related factors such as co-morbidity. As co-morbidity among prostate cancer patients is prevalent it is relevant to understand its impact on QoL. In this reagard, a comparison with a normal reference population may help to understand the relevance of the impact of treatment and nontreatment related factors on QoL.

PurposeTo investigate the course of quality of life (QoL) among prostate cancer patients treated with external beam radiotherapy and to compare the results with QoL of a normal age-matched reference population.Patients and methodsThe study population was composed of 227 prostate cancer patients, treated with radiotherapy. The EORTC QLQ-C30 was used to assess QoL before radiotherapy and six months, one year, two years and three years after completion of radiotherapy. Mixed model analyses were used to investigate longitudinal changes in QoL. QoL of prostate cancer patients was compared to that of a normative cohort using a multivariate analysis of covariance.ResultsA significant decline in QoL was observed after radiotherapy (p<0.001). The addition of hormonal therapy to radiotherapy was associated with a lower level of role functioning. Patients with coronary heart disease and or chronic obstructive pulmonary disease or asthma had a significantly worse course in QoL. Although statistically significant, all differences were classified as small or trivial.ConclusionProstate cancer patients experience a small worsening of QoL as compared with baseline and as compared with a normal reference population. As co-morbidity modulates patients’ post-treatment QoL, a proper assessment of co-morbidity should be included in future longitudinal analyses on QoL.

Radiotherapy offers the opportunity to treat prostate cancer in an effective manner, considering high tumor control and survival. The organs around the prostate receive unwanted, but unavoidable irradiation as a consequence of this treatment. Particularly the intestines and bladder can be damaged due to this radiation and may result in side effects such as incontinence, blood loss and pain. When it is clear which dose causes which side effect, this information can be used to change future treatment plans. This research has shown that dose to the pelvic floor muscles, dose to the last part of the large intestine and dose to specific parts of the bladder are predictive of the aforementioned side effects. This thesis also shows that these side effects influence a patients’ quality of life. To optimize the quality of life of a prostate cancer survivor more attention should be given to side effects that have the largest impact on quality of life. A comparative study with a group of men without prostate cancer shows that the quality of life of these patients differs slightly from this normative group of men.