Tumori Journal Abstract Book - October 2020 - 157

N - Head and Neck Tumours	
89 sites). Following the Phase 1 dose escalation portion of
the trial, patients received the recommended dose of 160 mg
orally twice daily. Each cycle was 28 days. The primary
endpoint was objective response rate (ORR) per RECIST
1.1. Secondary endpoints included duration of response
(DoR) and safety. Per health authority agreement, the primary analysis set was defined as the first 55 consecutively
enrolled patients previously treated with multikinase inhibitors cabozantinib and/or vandetanib. Patients naïve to cabozantinib and vandetanib treatment were analyzed separately.
All analyses were based on a 16-Dec-2019 data cutoff date.
Results: In the primary analysis set of prior cabozantinib
and/or vandetanib-treated patients with MTC (n=55), the
ORR by investigator assessment was 62% (95%CI 47.7-
74.6, n=34/55) and the median DoR was not reached
(95%CI 18.4 months-not estimable) despite a median follow-up of 14.8 months. In cabozantinib/vandetanib treatment-naïve patients (n=88), the ORR by investigator
assessment was 69% (95%CI 58.6-78.7, n=61/88, including 2 responses pending confirmation). Of the 59 confirmed
responding patients, with a median follow-up of 8 months,
responses were ongoing for 57 responders at the time of the
analysis. In the safety analysis set consisting of all selpercatinib dosed patients (N=702), the most common treatment-related adverse events (TRAEs) that occurred in
⩾15% of patients were dry mouth (33.3%), increased AST
(24.5%), increased ALT (23.8%), hypertension (23.2%),
diarrhea (19.7%), and fatigue (16.8%). Only 2% (14 of
702) of patients discontinued selpercatinib for TRAEs.
Conclusions: Selpercatinib use was associated with
marked and durable antitumor activity in prior cabozantinib and/or vandetanib-treated patients and in cabozantinib/vandetanib-naïve patients with RET-mutant MTC,
with the majority of responses ongoing in both cohorts.
Selpercatinib was well tolerated. Efficacy data assessed by
independent review committee based on the 16-Dec-2019
data cutoff date will be presented. ©2020 ASCO, Inc.
Reused with permission.

N02
Radiomics of pre-treatment
magnetic resonance imaging
(MRI) can predict recurrence in
effectively cured head and neck
squamous cell carcinoma (HNSCC)
patients
Alfieri S.1, Romanò R.1, Bologna M.2, Corino V.2, Calareso G.3,
Mirabile A.4, Ferri A.5, Marcantoni A.6, Grosso E.7, Tarsitano A.8,
Valerini S.9, Vecchio S.10, Deantonio L.11, Blengio F.12, Ibrahim T.13,
Locati L.D.14, Bergamini C.14, Bossi P.15, Mainardi L.2, Licitra L.16
1

S.C. Oncologia Medica 3 Tumori della Testa e del Collo, Fondazione IRCCS
Istituto Nazionale dei Tumori (INT) di Milano, Milano; 2Dipartimento di
Elettronica, Informatica and Bioingegneria (DEIB), Politecnico di Milano,

157
Milano; 3Dipartimento di Radiologia, Fondazione IRCCS Istituto Nazionale
dei Tumori (INT) di Milano, Milano; 4Dipartimento di Oncologia, Divisione
di Medicina Sperimentale, Fondazione IRCCS Ospedale San Raffaele,
Milano; 5Divisione di Chirurgia Maxillo-Facciale, Dipartimento di Chirurgia,
Azienda Ospedaliera Universitaria di Parma, Parma; 6Dipartimento di
Otorinolaringoiatria, Ospedale Santa Chiara di Trento, Trento; 7Istituto
Europeo di Oncologia (IEO), Milano; 8Unità di Chirurgia Maxillo-facciale,
Dipartimento di Scienze Biomediche e Neuromotorie, Ospedale S. OrsolaMalpighi, Alma Mater Studiorum Università di Bologna, Bologna; 9Unità di
Otorinolaringoiatria, Dipartimento di Neuroscienze, Azienda Ospedaliero
Universitaria di Modena, Modena; 10Fondazione IRCCS Istituto Nazionale
dei Tumori (IST) di Genova e Università di Genova, Ospedale San Martino,
Genova; 11Radiation Oncology Clinic Oncology, Institute of Southern
Switzerland (IOSI), Bellinzona-Lugano, Switzerland, Bellinzona-Lugano
(Switzerland); 12Dipartimento di Oncologia Medica, A.O. SS Antonio e Biagio
e Cesare Arrigo, Alessandria, Alessandria; 13Centro di Osteoncologia and
Tumori Rari, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori
(IRST) IRCCS, Meldola, Meldola; 14Fondazione IRCCS Fondazione IRCCS
Istituto Nazionale dei Tumori (INT) di Milano, Milano; 15Università di Brescia
e Ospedale Civili di Brescia, Brescia; 16S.C. Oncologia Medica 3 Tumori della
Testa e del Collo, Fondazione IRCCS Istituto Nazionale dei Tumori (INT) di
Milano, Università di Milano, Milano

Background: Emerging data suggest that radiomics can
be used to predict outcomes in HNSCC. At present, only
few data are available for baseline MRI (b-MRI)
Methods: The training set (TS) was retrieved from an
ongoing multicenter, randomized, prospective trial
(HETeCo) and consisted of effectively cured stage III-IV
(VIII TNM ed.) HNSCC. Patients (pts) with both pre- and
post-contrast enhancement T1 and T2-weighted b-MRI
and at least 2 years (2y)-FUP were selected. The validation
set (VS) was based on a multicenter H2020-funded project
(BD2Decide) and selected by same inclusion criteria of
TS. In total, 1608 radiomic features (RF) were extracted
by b-MRI. The best RF combination was used to classify
risk of disease recurrence: high risk (HR) and low risk
(LR). In HR vs. LR, Overall Survival (OS) and 3y-disease
free survival (DFS) Kaplan-Meier (KM) curves were compared. The radiomic model was evaluated using 10-fold
cross-validation in the TS and external validation in the
VS. Sensitivity, specificity and area under the curve (AUC)
were evaluated. The radiomic-based risk class was used in
a multivariate Cox model with well established prognostic
factors in HNSCC (TNM, subsite, HPV).
Results: In total, 57 pts of TS and 72 of VS met the eligibility criteria. Baseline TS and VS were also well balanced
(Table).
Sensitivity, specificity and AUC of the classifier were
82%, 70%, and 79% on TS, and 70%, 68% and 69% on
VS, respectively.
The radiomic risk class (HR/LR) was found to be an independent prognostic factor for 3y-DFS (TS, p=0.02; VS,
p=0.02) and OS (TS, p=0.042; VS, p=0.004).
The KM curves for HR and LR groups differed for both
3y-DFS (TS, p=0.019; VS, p=0.024) and OS (TS, p=0.045;
VS, p=0.002). In the TS, within HR vs LR comparison,
3y-DFS was 63% [47-85%] vs 95% [87-100%] and OS
was 88% [75-100%] vs 96% [88-100%], respectively. In
Tumori Journal Abstract Book - October 2020

Table of Contents for the Digital Edition of Tumori Journal Abstract Book - October 2020
