PT  - JOURNAL ARTICLE
AU  - Ryan Reddy
AU  - Charles Schmidtlein
AU  - Romina Grazia Giancipoli
AU  - Audrey Mauguen
AU  - Heiko Schoder
AU  - Lisa Bodei
TI  - Functional tumor volume measurement in Ga-DOTATATE PET-CT: Which method is correct?
DP  - 2020 May 01
TA  - Journal of Nuclear Medicine
PG  - 1306--1306
VI  - 61
IP  - supplement 1
4099  - http://jnm.snmjournals.org/content/61/supplement_1/1306.short
4100  - http://jnm.snmjournals.org/content/61/supplement_1/1306.full
SO  - J Nucl Med2020 May 01; 61
AB  - 1306Introduction: The use of PET/CT with 68Ga-labelled somatostatin analogs has substantially impacted the diagnosis and management of neuroendocrine tumors (NETs) due to its superior sensitivity and resolution. Unlike FDG PET/CT where intensity of uptake correlates well with prognosis, functional tumor volume (FTV) is a better correlate for prognosis in NETs with 68Ga-somatostatin-analog avidity. However, no validated segmentation approach currently exists, and each method of measuring functional tumor volume often results in substantially different volumes. Our aim was to test the concordance of four different functional tumor volume measurement approaches compared to traditional anatomic volume measurement. Methods: The clinical and imaging data of 327 patients with NETs was collected at the MolecularImaging and Therapy Service at MSKCC in New York. Of these, 249 patients had a baseline 68Ga-DOTATATE scan between December 2016 and April 2018 with a concurrent CT or MRI scan between January 2017 and May 2018. Scans were excluded if there was no contrast on the correlative CT, patient had surgery or other medical therapies between imaging studies, or no corresponding lesions were seen when comparing CT and Ga-DOTATATE PET findings. The resulting 167 lesions on the included scans were evaluated with the dedicated CT and MRI scans to ensure an ellipsoid shape, lack of necrosis, and accurate measurements in three dimensions. Final evaluations resulted in 96 lesions with precise correlative findings on 68Ga-DOTATATE PET and dedicated CT/MRI exams. The FTV for each lesion was then measured with hand-drawn segmentation from SUVmax using the AW server VolumeShare 7 software by General Electric Healthcare. Additionally, three automated thresholds were used to measure volume from the PET acquisition -- volume determined by a 50% threshold from SUVmax, volume determined by a 42% threshold from SUVmax, and a background-subtracted lesion histogram-based method that attempts to determine the tumor activity by estimating and then subtracting the background volume surrounding the lesion. These measurements were then compared to lesion volume calculated from traditional measurements obtained from CT/MRI imaging. Pearson’s correlation coefficient via a linear model with a random intercept to account for the intra-patient correlation was used for statistical analysis. Results: Overall, the correlation for each method of measuring FTV from the PET acquisition was as follows: 0.673 correlation for volume measured with hand-drawn segmentation, 0.699 correlation for volume determined by a 42% threshold from SUVmax, 0.635 correlation for volume determined by a 50% threshold from SUVmax, and 0.261 correlation for volume determined by background-subtracted lesion histogram-based method. Conclusions: Of these approaches, volume determined with a 42% threshold from SUVmax segmentation best measures Functional Tumor Volume and is consistent with volumes obtained using hand-drawn segmentation.