Differentiation between normal and metastatic lymph nodes in patients with skin melanoma: Preliminary findings using a DIXON-based whole-body MRI approach

Metastatic melanoma lymph nodes (MMLns) might be challenging to detect on MR-WBI, as both MMLns and normal lymph nodes (NLns) can show restricted water diffusion. Our purpose is to assess the potential contribution of the DIXON sequence in differentiating MMLns from NLns. We followed a cohort of 107 patients with stage IIIb/c and IV skin melanoma for 32 months using MR-WBI with DIXON, STIR, and DWI/ADC sequences. We compared signal intensity (SI) values of MMLns and NLns in the four series of the DIXON sequence (in/out-of-phase, fat_only, and water_only series). The fat fraction (SIfat_only/SIin) and the long:short axis ratio of MMLns were calculated. The fat fraction was also calculated in the fatty hila of NLns. All MMLns (8 from 7 patients) showed SIout>SIin with a mean fat fraction of 10%. In 40 normal fatty hila (25 patients), the proportion of SIout>SIin was 100% and mean fat fraction was 89% (pSIin pattern was identified in 41/113 cases from 19/40 patients. The median long:short axis ratio in MMLns was 1.13 (range 1.03–1.25). The combination of three features of MMLns (SIout>SIin, low-fat fraction and rounded shape) might hold promise in differentiating NLns from MMLns in patients with skin melanoma. Further research is warranted due to the small number of MMLns in our cohort. •MR-WBI-based surveillance for melanoma lymph node metastases typically includes three sequences (DIXON, STIR and DWI).•DIXON could improve upon MR-WBI performance in differentiating metastatic melanoma (MMLns) from normal (NLns) lymph nodes.•On DIXON images, MMLns showed a drop-in effect (SI out-of-phase > SI in-phase) in 100% of cases.•Our preliminary findings suggest that the DIXON sequence may play a role in early MMLns detection.