Texture analysis of liver NET liver metastases on Ga-68 DOTATATE PET candidate to PRRT: non-invasive histological and prognostic stratification

Ziel/Aim Liver metastases (LM) from well-differentiated (G1/2) neuro-endocrine tumor (NET) can be managed by peptide-receptor radiotherapy (PRRT). Grading determination requires a biopsy; however, biological features can be heterogenous across metastases. Moreover, time-to-progression (TTP) after PRRT can be widely variable. In this study, we tested whether texture analysis (TA) can identify features associated with histology and TTP.

Methodik/Methods The institutional database was searched to identify NET patients with LM and treated with PRRT (Jan 2015 to Dec 2018). VOIs were constructed on each LM on the pre-PRRT Ga-68 DOTATATE PET (PET1) and on their counterparts on the same examination after two PRRT cycles (PET2); TTP was computed. TA was carried out with LifeX®.

Ergebnisse/Results Fifty patients (17 females, age 64 ± 11, range 30-80) were identified. At PET1, G3 LM metastases showed a slightly higher SUV_mean than G1/G2 (13.4 ± 3.7 vs 10.8 ± 4.7, p < 0.01). The strongest predictors of a G3 status were, however, run-length and size zone matrices, such as GLRLM_LGRE/HGRE/SRLGE/SRHGE/LRLGE (p < 0.001) and GLZLM_LGZE/HGZE/SZLGE (p < 0.001). Likewise, SUV_mean could tell apart midgut-NET (10.4 ± 4.2) from pancreatic NET (12.3 ± 5.1, p < 0.01); however, TA provided the strongest differentiation. In particular, GLCM Contrast and Dissimilarity were markedly higher in pancreatic NETs (p < 0.001), as well as NGLDM_Contrast, GLZLM_SZE, and GLZLM_ZP (p < 0.001).

No PET1 factor could tell apart LR/SR. However, when comparing PET1 and PET2, LR had a lesser drop of SUV_mean (3.4 ± 4.4 vs. 1.4 ± 4.6, p < 0.05), as well as of GLRLM_HRRE/SRHGE (p < 0.01), GLZLM_SZE/HGZE/SZHGE/ZP (p < 0.01) after two PRRT cycles.

Schlussfolgerungen/Conclusions TA can non-invasively identify characteristics associated to NET grading/histotype at baseline Ga-68 DOTATATE PET.

Modifications of textural features, and in particular reduction of homogeneity and tracer uptake is associated with a more aggressive biological behavior, with faster disease progression.

Publication History

Article published online:
08 April 2021