Purpose: To establish dose volume-effect relationships predicting late rectal morbidity in cervix cancer patients treated with concomitant chemoradiation and MRI-guided adaptive brachytherapy within the prospective longitudinal multicenter EMBRACE study. Material and Method: Data from all EMBRACE patients who completed their treatment at least one year earlier were extracted from the study database in August 2015. All patients were treated with curative intent according to institutional protocols. Reported doses followed the GEC-ESTRO recommendations (D0.1cm 3 , D2 cm 3 ). The follow-up included assessments of rectal morbidity 3, 6, 9, 12, 18, 24, 30, 36, 48, and 60 months after treatment completion with prospective assessment of morbidity according to the CTC-AE 3.0 scale. Late morbidity was defined as any event occurring or lasting over 90 days after treatment initiation. Patients with missing D2cm3, or follow-up less than 90 days from treatment initiation were excluded. All doses were converted in 2 Gy equivalent using the linear quadratic model and a/b53 Gy. For analyzes, 4 events types were considered: proctitis, bleeding, stenosis, and fistula. Dose effect-relationships were assessed using comparisons of mean doses, log rank tests on actuarial estimates of event-free periods, and the probit model. Results: Of the 1,129 patients identified, 960 patients fulfilled the inclusion criteria. Their median follow-up was 25.4 months. A total of 193 patients had grade 1 events (20.1%), 58 grade 2 (6.0%), 14 grade 3 (1.6%), 1 patient a grade 4 fistula (0.1%). Proctitis represented 50.4% of the events, rectal bleeding 43.9%, stenosis 3.1%, and fistula, 2.5%. The mean DICRU, D0.1cm 3 , and D2cm 3 were 66.2AE9.1Gy, 72.9AE11.9 Gy, and 62.8AE7.6 Gy, respectively. The mean DICRU, D0.1cm 3 , and D2cm 3 increased progressively with the severity of the events except stenosis (p ranging from 0.018 to 0.001). The probit model showed significant correlations between D2cm 3 , D0.1cm 3 , and the DICRU and the probability of grade 1-4, 2-4, and 3-4 rectal events. Significant correlations were observed between the D2cm 3 and the probability of occurrence of proctitis, bleeding, and fistula but not for stenosis. The effective D2cm 3 for a 10% probability of grade 2-4 events (ED10) were 82.3, 80.4, and 102 Gy for proctitis, bleeding, and fistula, respectively. The overall rectal ED10 was 69.5 Gy. When related to length of followup, this value decreased towards 60.0 Gy in patients with follow-up O 36 months (n5324, 33.8%). After sorting patients according to 6 D2cm3 levels (! 55 Gy,(55)(56)(57)(58)(59)(60)(60)(61)(62)(63)(64)(65)(65)(66)(67)(68)(69)(70)(70)(71)(72)(73)(74)(75) and $ 75 Gy), log rank test showed significant differences in event free periods for grade 1-4, 2-4, and 3-4 bleeding, fistula, and all rectal events (Figure). For proctitis, the D2cm 3 was discriminant for grade 2-4 events. Conclusions: Significant dose volume-effect correlations were established between late rectal morbidity, including specific events such as bl...
