Semi-quantitative analysis of visually normal (123)I-FP-CIT across three large databases revealed no difference between control and patients

BACKGROUND: To show the equivalence between the specific binding ratios (SBR) of visually normal (123)I-FP-CIT SPECT scans from patients to those from healthy volunteers (Hv) or patients without dopaminergic degeneration to allow their use as a reference database. METHODS: The SBR values of visually normal SPECT scans from 3 groups were studied: (1) suspected Parkinsonism and no diagnostic follow-up (ScanOnlyDB: n = 764, NM/CT 670 CZT, GE Healthcare), (2) no degenerative dopaminergic pathology after a 5-year follow-up (NoDG5YearsDB: n = 237, Symbia T2, Siemens Medical Solutions), and 3) Hv (HvDB: n = 118, commercial GE database). A general linear model (GLM) was constructed with caudate, putamen, and striatum SBR as the dependent variables, and age and gender as the independent variables. Following post-reconstruction harmonization of the data, DB were combined in pairs, ScanOnlyDB&NoDG5yearsDG and ScanOnlyDB&HvDB before performing GLM analysis. Additionally, ScanOnlyDB GLM estimates were compared to those published from Siemens commercial DB (SiemensDB) and ENC-DAT. RESULTS: The dispersion parameters, R(2) and the SBR coefficients of variation, did not differ between databases. For all volumes of interest and all databases, SBR decreased significantly with age (e.g., decrease per decade for the striatum: − 4.94% for ScanOnlyDB, − 4.65% for NoDG5YearsDB, − 5.69% for HvDB). There was a significant covariance between SBR and gender for ScanOnlyDB (P < 10(–5)) and NoDG5YearsDB (P < 10(–2)). The age-gender interaction was significant only for ScanOnlyDB (P < 10(–2)), and the p-value decreased to 10(–6) after combining ScanOnlyDB with NoDG5YearsDB. ScanOnlyDB GLM estimates were not significantly different from those from SiemensDB or ENC-DAT except for age-gender interaction. CONCLUSION: SBR values distribution from visually normal scans were not different from the existing reference database, enabling this method to create a reference database by expert nuclear physicians. In addition, it showed a rarely described age-gender interaction related to its size. The proposed post-reconstruction harmonization method can also facilitate the use of semi-quantitative analysis. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13550-023-00983-6.