Anatomical Localization of Late Rectal Toxicity Predictors in Prostate Radiotherapy
K Foo1*, MA Ebert2, MG Carolan3, A Haworth4, M Bulsara5, D Joseph6, JW Denham7, (1) Royal Prince Alfred Hospital, Sydney, Australia, (3) University of Western Australia, Perth, Australia, (3) Illawarra Cancer Care Centre, Wollongong, Australia, (4) Peter MacCallum Cancer Centre, Melbourne, Australia, (5) University of Notre Dame, Fremantle, Australia, (6) Sir Charles Gairdner Hospital, Perth, Australia, (7) University of Newcastle, Newcastle, AustraliaTU-G-108-6 Tuesday 4:30PM - 6:00PM Room: 108
Purpose: To determine dosimetric predictors of late rectal toxicity after prostate radiotherapy, differentiating between dose to rectum and anal canal.
Methods: Clinical and 3D dosimetric data for 738 patients were prospectively collected in a clinical trial (TROG 03.04 "RADAR") of variable-duration hormone therapy with conformal radiotherapy using rectal DVH constraints of V65<40%, V70<30%, V75<5%. Dose-volume histograms were generated for whole anorectum, anal canal, rectum and matched to longitudinal toxicity (measured 6-monthly). Outcome measures were: bleeding, urgency/tenesmus (grade 1+, 2+, 3), pooled Common Toxicity Criteria version 2 proctitis (grade 1+, 2+), incontinence, bowel bother, diarrhea, and anorectal pain. Both prevalence (at 30 months post radiotherapy) and actuarial incidence (time-to-failure) endpoints were analyzed. Rectal mean dose and V5-V70 were used as factors in multivariate and univariate logistic regression (prevalence) and Cox proportional hazards (incidence) analyses.
Results: Median follow-up was 6.5 years post-treatment. Multivariate models produced erratic effect estimates due to multicollinearity arising from constrained DVH shape as mandated by the trial protocol. Univariate analysis identified that pain, pooled proctitis, incontinence were predicted most by anal canal dose. Bleeding and diarrhea were not subvolume dependent. Bowel bother was not associated with any factor. Urgency/tenesmus association with subvolume depended on the outcome grade used.
Conclusion: Doses to different subvolumes of the rectum are responsible for different types of toxicity. Different grades of the same toxicity are predicted by different subvolumes and DVH points. For datasets planned with dose constraints, multicollinearity of DVH points limits the usefulness of multivariate analysis. Further work will estimate DVH constraints to limit toxicity, and Lyman model parameters from the same data.
Funding Support, Disclosures, and Conflict of Interest: Funding: National Health and Medical Research Council (Australian Government), Hunter Prostate Cancer Alliance, Trans-Tasman Radiation Oncology Group. Conflicts of interest: none declared