Session Item

To evaluate systemic recurrence within the prospective observational multi-centre IntErnational study on MRI-guided BRAchytherapy in locally advanced CErvical Cancer (EMBRACE) cohort.

Data from 1416 patients with locally advanced cervical cancer (LACC) treated from 2008-2015 in 24 centres were collected. Treatment consisted of pelvic +/- para-aortic external beam radiotherapy with 45-50Gy (1.8-2.0 Gy per fraction), concurrent chemotherapy (Cisplatin 40mg/m² weekly) (RCHT) and MRI based image-guided adaptive brachytherapy (IGABT). Follow-up (FU) visits were scheduled at 3 months intervals for the first year, 6 months intervals for the second and third year and yearly intervals thereafter. In case of recurrent disease only the first site of recurrence was registered. A systemic recurrence (SR) was defined as any first recurrent disease outside the pelvis. Descriptive statistics, uni- and multivariate analyses (MVA) were performed to describe patterns of recurrence and risk factors using the following variables: age, haemoglobin and white blood cell count at diagnosis, WHO performance score, hydronephrosis, smoking status, FIGO₂₀₀₉ stage, histology, maximum tumour dimension, involved lymph node regions (N0 vs pelvic vs common iliac vs para-aortic (PAN), no. of chemotherapy cycles, CTV_HR volume at IGABT, D90 for CTV_HR, overall treatment time. Risk factors were evaluated for the overall cohort and for initially node negative and node positive patients (NNP/NPP), separately.

1318 patients were eligible for analysis. FIGO₂₀₀₉ stages were IB1 (8%), IB2 (9%), IIA (5%), IIB (52%), IIIA (1%), IIIB (14%), IVA (3%) and IVB (PAN only, 7%), N0 = 48%, N1 = 52%. After a median FU of 51 months, SR was reported in 243 patients (18%). Of these, 31 (13%) had a simultaneous local and 77 (32%) had a simultaneous pelvic nodal recurrence. Median time to SR was 12 months (1-97).  Most common sites for SR were PAN (n=106), lungs (n=61), unknown (n=41), mediastinal lymph nodes (n=36), bones (n=23), peritoneum(n=19) and liver (n=17). Risk factors for SR in the overall cohort are summarized in table 1. Nodal status (with risk increasing the more cranial the nodal region) and histology were the strongest risk factors for SR in the overall cohort. Additional risk factors in NNP were dose to the D90 of CTV_HR at IGABT and presence of tumour necrosis on MRI. Additional risk factors in NPP were CTV_HR >45cm³ at IGABT and haemoglobin level at diagnosis. FIGO₂₀₀₉ stage was not significant in any analysis. 5-year systemic control based on these factors is shown in table 2.

Para-aortic lymph nodes and lungs are the most common sites of SR. Nodal status at diagnosis, histologic subtype and large CTV_HR at IGABT are the most important risk factors for the occurrence of SR. Patients with presence of one or more of these factors should be considered in future trials investigating the role of adjuvant treatment after primary chemoradiation.