ESTRO 2022

Session Item

Monday

May 09

10:30 - 11:30

Mini-Oral Theatre 2

20: Breast

Co-Chair: Nienke Hoekstra, The Netherlands;

Chair: Wilfried Budach, Germany

Session Code: 3260

Session Type: Mini-Oral

Track: Clinical

External Validation of NTCP-models for Acute Coronary Events after Breast Cancer Radiotherapy

, The Netherlands

Presentation Number: MO-0803

Abstract

Abstract Title:

External Validation of NTCP-models for Acute Coronary Events after Breast Cancer Radiotherapy

Authors:

Daan Spoor¹, Floor van Leeuwen², Nicola Russell³, Naomi Boekel², Sophie Jacob⁴, Stephanie Combs⁵, Kai Borm⁵, Rozemarijn Vliegenthart⁶, Gert Sikkema⁷, Marianna Sijtsema¹, Arjen van der Schaaf⁸, John Maduro¹, Hans Langendijk¹, Ewoud Schuit⁹, Anne Crijns¹

¹University Medical Center Groningen, University of Groningen, Radiation Oncology, Groningen, The Netherlands; ²Netherlands Cancer Institute, Psychological Research and Epidemiology, Amsterdam, The Netherlands; ³Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands; ⁴Institute for Radiation Protection and Nuclear Safety, Epidemiology, Fontenay Aux Roses, France; ⁵Technical University of Munich, School of Medicine, Radiation Oncology, Munich, Germany; ⁶University Medical Center Groningen, University of Groningen, Radiology, Groningen, The Netherlands; ⁷University Medical Center Groningen, University of Groningen, radiation Oncology, Groningen, The Netherlands; ⁸University Medical Center Groningen, University of Groningen, Radiation ONcology, Groningen, The Netherlands; ⁹Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, The Healthcare Innovation Center , Utrecht, The Netherlands

Show Affiliations

Purpose or Objective

Radiotherapy (RT) for breast cancer (BC) may put survivors at increased risk of acute coronary events (ACE). NTCP-models for ACE can be used to identify BC patients at high risk to optimize primary and/or secondary prevention strategies. As part of the MEDIRAD-BRACE study (H2020-Euratom-1.4 / 755523) we developed two NTCP-models for ACE in 4,305 BC patients treated with adjuvant RT from 2005 to 2015. Both models performed equally well but differed regarding the DVH parameters included (table 1). Model 1 included patient characteristics combined with the left ventricle V5 and the whole heart Dmean, and Model 2 included the whole heart and right ventricle D1 and left ventricle V6 as DVH parameters. The aim of the current study was to externally validate both models in a multi-centre cohort of BC patients.

Material and Methods

The validation cohort consisted of 1,520 BC patients from 3 other centres treated with adjuvant radiotherapy in the same period. After application of both models to the patients included in the validation cohort, model performance was assessed by discrimination in terms of a c-statistic and calibration in terms of calibration plots and the observed/expected ratio. If needed, models were updated.

Results

Mean follow up was 7.6 years and the cumulative ACE incidence at 10 years was 1.7% for the validation cohort. This cumulative ACE incidence at 10 years was significantly lower than observed in the development set (4.8%). Model performance after external validation was similar for both models (table 2). External validation showed good discrimination: c-statistic 0.72 (0.61-0.83) and 0.71 (0.60-0.82) for model 1 and 2, respectively. There was an overestimation of the ACE risk by a factor 2 in the validation cohort. But after recalibration (adjustment of the mean linear predictor and baseline risk, but same coefficients) the prediction models performed well in the validation cohort.

Conclusion

We externally validated two models based on different cardiac DVH parameters for ACE risk in the first 10 years after radiotherapy. We prefer the model including whole heart Dmean and left ventricle V5 as these DVH-parameters are more suitable to apply in dose optimization strategies.