Derivation of dose/volume constraints for the anorectum from clinician and patient-reported outcomes in the CHHiP trial of radiotherapy fractionation.

Abstract

BACKGROUND:The CHHiP trial randomised 3216 men with localised prostate cancer (1:1:1) to three radiotherapy fractionation schedules: 74Gy/37 fractions (f) over 7.4 weeks, 60Gy/20f/4 weeks and 57Gy/19f/3.8 weeks. Literature-based dose constraints were applied with arithmetic adjustment for the hypofractionated arms. This study aimed to derive anorectal dose constraints using prospectively-collected clinician-reported outcomes (CRO) and patient-reported outcomes (PRO) and to assess the added predictive value of spatial dose metrics. METHODS:A case-control study design was used, seven CRO and five PRO bowel symptoms were evaluated. Cases experienced a moderate or worse symptom 1-5 years post-radiotherapy, and did not have the symptom pre-radiotherapy. Controls did not experience the symptom at baseline, or between 1-5 years post-radiotherapy. The anorectum was re-contoured from the anal verge to the recto-sigmoid junction; dose/volume parameters were extracted. Univariate logistic regression, atlases of complication indices and bootstrapped receiver-operating-characteristic (ROC) analysis (1000 replicates, balanced outcomes) were used to derive dose constraints for the whole cohort (hypofractionated schedules were converted to 2Gy equivalent schedules using α/β=3Gy) and separate hypofractionated/conventional fractionation cohorts. Only areas under the curve (AUC) with 95% confidence interval lower limits >0.5 were considered statistically significant. Any constraint derived in <95-99% of bootstraps was excluded. RESULTS:Statistically significant dose constraints were derived for CRO, but not PRO. Intermediate to high doses were important for rectal bleeding whereas intermediate doses were important for increased bowel frequency, faecal incontinence and rectal pain. Spatial dose metrics did not improve prediction of CRO or PRO. A new panel of dose constraints for hypofractionated schedules to 60Gy or 57Gy are V20Gy<85%, V30Gy<57%, V40Gy<38%, V50Gy<22% and V60Gy<0.01%. CONCLUSIONS:Dose constraints differed between symptoms, indicating potentially different pathogenesis of radiation-induced side effects. Derived dose constraints were stricter than those used in CHHiP and may reduce bowel symptoms post-radiotherapy.