Pretreatment PET dissemination features as biomarkers for survival outcomes in melanoma patients treated with immune-checkpoint inhibitors.

Abstract

The primary aim of this study was to assess whether there is an association between pretreatment fluorine-18-labelled fluorodeoxyglucose([¹⁸F]FDG) positron-emission tomography (PET) dissemination features and survival outcomes: overall survival (OS) and progression-free survival (PFS), in metastatic melanoma patients treated with immune-checkpoint inhibitors. The secondary aim was to assess the added value of dissemination features to conventional PET metrics and clinical characteristics with respect to survival outcome prediction. We included 80 consecutive patients with metastatic/unresectable melanoma treated with first-line immunotherapy. On a patient level, eighteen dissemination features and five conventional PET metrics: maximum standardized uptake value (SUVmax), mean standardized uptake value (SUVmean), peak standardized uptake value (SUVpeak), total metabolically active tumour volume (MATV) and total total lesion glycolysis (TLG) were extracted. From the eighteen dissemination features, five provided quantitative measures on spatial distribution, ten on SUV variability and three on tumour volume variability. Furthermore, clinical characteristics were collected. Univariable and multivariable Cox regression analyses were used to assess the association of OS and PFS with the conventional PET metrics, dissemination features and clinical characteristics. The final models were internally validated using bootstrapping. The spatial dissemination features DmaxPatient, DmaxBulk and SpreadBulk and volume dissemination feature DvolPatient on pretreatment [¹⁸F]FDG PET/CT were associated with OS. DmaxBulk and SpreadBulk were also associated with PFS. The spatial dissemination feature DmaxBulk, which measures the distance between the largest lesion and any other lesion within the studied metastatic melanoma patients, showed an independent association with OS (hazard ratio = 2.1) and PFS (hazard ratio = 1.9). Conventional PET metrics were not associated with OS nor PFS. The validated dissemination model, clinical model and combination model showed similar low discriminative ability with an Area Under the Curve (AUC) of 0.61-0.66 for OS and an AUC of 0.59-0.66 for PFS. Limitations of our study results were the lack of an available external cohort for further validation and the retrospective design. Future research could include features from other imaging modalities, non-tumour characteristics and other clinical characteristics for model improvement.