Inter- and intra-tumoral heterogeneity on [68Ga]Ga-DOTA-TATE/[68Ga]Ga-DOTA-TOC PET/CT predicts response to [177Lu]Lu-DOTA-TATE PRRT in neuroendocrine tumor patients

Camila Gadens Zamboni; Ayca Dundar; Sanchay Jain; Marc Kruzer; Bradley T Loeffler; Stephen A Graves; Janet H Pollard; Sarah L Mott; Joseph S Dillon; Michael M Graham; Yusuf Menda; Ahmad Shariftabrizi

doi:10.1186/s41824-024-00227-3

Inter- and intra-tumoral heterogeneity on [⁶⁸Ga]Ga-DOTA-TATE/[⁶⁸Ga]Ga-DOTA-TOC PET/CT predicts response to [¹⁷⁷Lu]Lu-DOTA-TATE PRRT in neuroendocrine tumor patients

EJNMMI Rep. 2024 Nov 30;8(1):39. doi: 10.1186/s41824-024-00227-3.

Affiliations

¹ Division of Nuclear Medicine, Department of Radiology-Room 3820-AJPP, University of Iowa Carver College of Medicine, 200 Hawkins Drive, Iowa City, IA, 52242, USA.
² MIM Software Inc, Beachwood, OH, USA.
³ Biostatistics Core, Holden Comprehensive Cancer Center, Iowa City, IA, USA.
⁴ Department of Medicine-Hematology/Oncology, University of Iowa Carver College of Medicine, Iowa City, IA, USA.
⁵ Division of Nuclear Medicine, Department of Radiology-Room 3820-AJPP, University of Iowa Carver College of Medicine, 200 Hawkins Drive, Iowa City, IA, 52242, USA. ashariftabrizi@healthcare.uiowa.edu.

^# Contributed equally.

Abstract

Background: Indices of tumor heterogeneity on somatostatin receptor PET/CT scans may potentially serve as predictive biomarkers of treatment efficacy in neuroendocrine tumor (NET) patients undergoing [¹⁷⁷Lu]Lu-DOTA-TATE PRRT.

Methods: NET patients who underwent [¹⁷⁷Lu]Lu-DOTA-TATE therapy at the University of Iowa from August 2018 to February 2021 were retrospectively evaluated. Radiomic features on the pre-PRRT somatostatin receptor PET/CT were evaluated using a custom MIM Software® LesionID workflow. Conventional PET/CT metrics of tumor burden, such as somatostatin receptor expression and tumor volume, were calculated in addition to the indices of tumor heterogeneity for each lesion (intra-lesional) and then summarized across all lesions throughout the body (inter-lesional). Endpoints included post-PRRT 24-month time to progression (TTP) and overall survival (OS). Cox regression models were used to assess the predictive ability of the imaging factors on post-PRRT 24-month TTP and OS. LASSO-penalized Cox regression was used to build a multivariable model for each outcome.

Results: Eighty patients with a mean age of 65.1 years were included, with most (71.3%) completing 4 cycles of PRRT. Median TTP was 19.1 months, and OS at 60 months was 50%. A large degree of variability between patients was evidenced for imaging features related to somatostatin receptor expression. On multivariable analysis, total receptor expression and mean liver-corrected SUVmean were selected for 24-month TTP. The model was not able to significantly predict progression (C-statistic = 0.58, 95% CI 0.50-0.62). Total receptor expression and mean skewness were selected for OS. The resulting model was able to significantly predict death (C-statistic = 0.62, 95% CI 0.53-0.67), but the predictive ability was limited, as evidenced by the low C-statistic.

Conclusions: Our exploratory analysis provides preliminary results showing that imaging indices of inter- and intra-tumor heterogeneity from pretreatment PET/CT images may potentially predict treatment efficacy in NET patients undergoing [¹⁷⁷Lu]Lu-DOTA-TATE therapy. However, prospective evaluation in a larger cohort is needed to further assess whether a comprehensive characterization of tumor heterogeneity within a patient can help guide treatment decisions.