ESTRO 2022

Session Item

Saturday

May 07

10:30 - 11:30

Room D5

Photon radiotherapy planning

Chair: , The Netherlands;

Co-Chair: Marcel Nachbar, Germany

Session Type: Proffered Papers

Track: Physics

Journey:

11:00 - 11:10

multi-institutional real-world validation of autoplanning for breast cancer

Christian Fiandra, Italy

Presentation Number: OC-0128

Abstract

Abstract Title:

multi-institutional real-world validation of autoplanning for breast cancer

Authors:

Christian Fiandra¹, Stefania Zara², Alessandro Alparone², Gianfranco Loi³, Antonella Roggio⁴, Alberto Ciarmatori⁵, Ilaria Benvenuto⁶, Angela Poggiu⁷, Anna Di Dio⁸, Elisabetta Verdolino⁹, Federica Rosica¹⁰, Stefano Ren Kaiser¹¹, Lidia Strigari¹², Luca Reversi¹³, Elena Pierpaoli¹⁴, Paolo Ferrari¹⁵, Lorenzo Placidi¹⁶, Stefania Comi¹⁷, Erminia Infusino¹⁸, Manuela Coeli¹⁹, Eva Gino²⁰, Tiziana Licciardello²¹, Nando Romeo²², Nunzia Ciscognetti²³, Gian Marco Deotto²⁴, Stefania Cora²⁵, Silvia Pesente²⁶, Linda Rossi²⁷, Umberto Ricardi¹, Ben Heijmen²⁸, Maristella Marrocco²⁹

¹University of Torino, Department of Oncology, Turin, Italy; ²Tecnologie Avanzate, Research and Development, Turin, Italy; ³'Maggiore della Carità’ University Hospital, Medical Physics, Novara, Italy; ⁴Veneto Institute of Oncology IOV-IRCCS, Medical Physics, Padova, Italy; ⁵AORMN, Medical Physics, Pesaro, Italy; ⁶IRCCS-CROB, Radiotherapy, Rionero in Vulture, Italy; ⁷AOU, Fisica Sanitaria, Sassari, Italy; ⁸Città della Salute e della Scienza, Medical Physics, Turin, Italy; ⁹Ospedale S. G. MOSCATI, Medical Physics, Taranto, Italy; ¹⁰ASL Teramo, Medical Physics, Teramo, Italy; ¹¹Azienda Sanitaria Universitaria Giuliano Isontina, Fisica Sanitaria, Trieste, Italy; ¹²IRCCS Azienda Ospedaliero-Universitaria di Bologna, Medical Physics, Bologna, Italy; ¹³Ospedali Riuniti di Ancona, Medical Physics, Ancona, Italy; ¹⁴Area Vasta 5 Asur P.O. Mazzoni, Medical Physics, Ascoli, Italy; ¹⁵Azienda Sanitaria dell'Alto Adige, Medical Physics, Bolzano, Italy; ¹⁶Fondazione Policlinico Universitario A. Gemelli IRCCS, Medical Physics, Rome, Italy; ¹⁷European Institute of Oncology IRCCS, Medical Physics, Milan, Italy; ¹⁸IFO Istituti Fisioterapici Ospitalieri, Medical Physics, Rome, Italy; ¹⁹Azienda ULSS 9 Scaligera del Veneto, Medical Physics, Legnago, Italy; ²⁰AO Ordine Mauriziano di Torino, Fisica Sanitaria, Turin, Italy; ²¹IRCCS Istituto Romagnolo per lo Studio dei Tumori, Medical Physics, Meldola, Italy; ²²S. Vincenzo Hospital, Medical Physics, Taormina, Italy; ²³ASL2 Savonese, Medical Physics, Savona, Italy; ²⁴University Hospital of Siena, Medical Physics, Siena, Italy; ²⁵San Bortolo Hospital, Medical Physics, Vicenza, Italy; ²⁶Tecnologie Avanzate, Research and Development, Udine, Italy; ²⁷Erasmus MC Cancer Institute, Medical Physics, Rotterdam, The Netherlands; ²⁸Erasmus MC Cancer centre, Radiation Oncology, Rotterdam, The Netherlands; ²⁹Campus Biomedico University, Radiation Oncology, Rome, Italy

Show Affiliations

Purpose or Objective

Investigate whether autoplanning could replace current clinical planning in Italy for whole breast irradiation after breast-sparing surgery, with/without boost, and with/without inclusion of regional lymph nodes.

Material and Methods

The 24 participating centers can include a selection of 10 IMRT/VMAT patients treated in their practice (240 patients in total). There were no restrictions regarding delivery units or clinical TPS and no limitation in tumor location, extension, treatment machine, prescribed dose and fractionation.

Autoplanning was performed with an optimizer implemented in a commercial TPS. A single algorithm configuration was used for all participating centers. For each patient, the autoplan was generated for the clinically applied delivery machine.

Autoplans were compared to clinical plans by clinician blind scoring, and by dosimetrical comparisons for PTV and OARs. In the blind scoring, the clinical plan and the autoplan were simultaneously loaded and the treating clinician could score High, Medium or Low impact advantage for one of the plans, or Parity.

Results

So far 175 patients have been included from 18 centers; 116 (66%) with left-sided breast cancer and 59 with right-sided. 40 patients (23%) also had a boost dose. Supraclavicular nodes, internal mammary nodes and axillary nodes were treated in 54 patients (31%), 12 patients (7%) and 7 patients (4%), respectively. Delivered doses were between 40.05 Gy and 50 Gy (from 15 to 25 fractions), with concomitant boost doses of 55-60 Gy. All C-arm linacs and Tomotherapy were used for delivery. All clinical plans were generated manually, and all major commercial TPS were represented.

For 82/175 patients (47%), the treating clinicians preferred the autoplan, while the clinical plan was favoured in 75 (43%) cases. For 18 (10%) the clinicians scored Parity. See Fig. 1 for details. Statistically significant differences (p<0.05) in favor of autoplans were found for PTV coverage and conformity, while D_0,03ccand Homogeneity index were favorable to Manual (Figure 2a). For OARs , following differences were found to be statistically significative: contralateral breast average and D_0.03ccwere reduced by 23% and 23,6% respectively; for heart, average dose and V_8Gy were reduced by 19.6% and 3.8%; for ipsilateral lung, all dosimetric parameters (V_16Gy, V_8Gy, V_4Gy and Average) were favorable to autoplans (2,7%, 2,6%, 2.8% and 11,6%)

Conclusion

Even with the large variation in included patients, delivery machines, treatment techniques and clinical TPSs, and no institution-specific configurations of the autoplanning algorithm, in 57% of the 175 patients from 18 centers, clinicians scored higher or equal quality for autoplans compared to manually generated clinical plans. Dose- volume parameters were mainly in favor of autoplans. This study points at an opportunity to substantially reduce treatment planning workload for breast cancer.