Comparing RECIST, Choi, modified Choi (mChoi) and viable tumor volume (VTV) in predicting clinical benefit of vascular endothelial growth factor receptor inhibitors (VEGFRi) in patients (pts) with recurrent/metastatic (R/M) adenoid cystic carcinoma (ACC).

Abstract

e18145 Background: RECIST may underestimate the clinical activity of systemic therapies for ACC, especially in pts with ACC-II molecular subtype receiving VEGFRi, which can cause tumor necrosis without significant tumor shrinkage. Choi and mChoi, which consider size and density (attenuation), and VTV, which incorporates volume and density, may be better indicators of VEGFRi efficacy. Methods: Retrospective study conducted at MD Anderson Cancer Center. Pts with the following criteria were included: 1) biopsy-proven R/M ACC, 2) received at least one cycle (28 days) of lenvatinib or axitinib monotherapy in any line of treatment, 3) available baseline and at least one re-staging scan. Best overall response (BOR) (partial response [PR], stable disease [SD], and progressive disease [PD]) was evaluated according to RECIST 1.1, Choi, mChoi, and VTV. Time to next treatment (TTNT) and overall survival (OS) according to BOR were compared for each of the four response criteria using the Kaplan-Meier method. Results: Of 50 R/M ACC pts treated with lenvatinib or axitinib, 27 had available scans for analysis. Median age at VEGFRi initiation was 53.2 years. 70.4% were female, 81.5% had head and neck salivary gland primary tumors, and 77.8% received lenvatinib. Among 21 pts with known histology and 22 pts with molecular profiling, 81% had either non-solid or mixed solid/non-solid histology, and 72.7% were NOTCH wild-type, respectively. BOR according to the four response criteria were as follows: RECIST: 0 PR, 21 SD, 5 PD; Choi: 17 PR, 4 SD, 5 PD; mChoi: 5 PR, 16 SD, 5 PD; VTV: 5 PR, 15 SD, 5 PD. Median TTNT for the total cohort was 13.3 months. There was a significant difference in TTNT according to RECIST BOR (median TTNT was 17.6 months for SD and 8.3 months for PD, p =.010), Choi BOR (median TTNT was 24 months for PR, 7.2 months for SD, and 8.3 months for PD, p =.005) and mChoi BOR (median TTNT was 5.4 months for PR, 17.6 months for SD, and 8.3 months for PD, p =.028), but not according to VTV BOR. Median OS from VEGFRi initiation for the total cohort was 22.3 months. There was a significant difference in OS according to mChoi BOR (median OS was 14.2 months for PR, 39.6 months for SD, and 16.8 months for PD, p =.002), but not according to BOR for any of the other response criteria. Conclusions: Although RECIST was associated with TTNT,Choi was better able to identify responders with the longest real-world clinical benefit from VEGFRi. However, Choi was not predictive of treatment outcome among non-responders (SD or PD). Patients with SD per mChoi had the longest survival, but this may be related to confounding from ACC molecular subtype or other factors.