Abstract
<title>Abstract</title> <p>Background Sentinel lymph node excision (SLNE) is the standard staging procedure in intermediate- and high-risk cutaneous melanoma, but in the head and neck it is technically demanding, carries a risk of facial nerve injury, and may fail to retrieve a preoperatively marked node, leaving the patient with the operative risk but no staging information. Non-invasive predictors of sentinel lymph node (SLN) status are therefore needed. We investigated whether quantitative computed tomography (CT) parameters from the low-dose CT component of preoperative single-photon emission computed tomography/computed tomography (SPECT/CT), already performed for SLN mapping, differ between metastatic and tumor-free SLNs. Methods In this retrospective single-center study, 42 SLN regions of interest (ROIs) from patients with head and neck cutaneous melanoma treated between 2014 and 2025 were analyzed: 18 with histologically confirmed metastasis and 24 tumor-free controls, selected by simple random sampling. SLN ROIs were segmented by a radiologist blinded to histopathology. Five parameters were compared using the Mann–Whitney U test, with receiver operating characteristic analysis and bootstrap confidence intervals (CIs): minimum, maximum, and mean attenuation, the standard deviation (SD) of attenuation, and mask volume. Both significant parameters were entered into a multivariable logistic regression. Results Metastatic ROIs were larger and showed higher mean attenuation. Mask volume discriminated best (median 397.9 versus 257.7 mm³; P = 0.004; area under the curve [AUC] 0.77, 95% CI 0.61–0.90), followed by mean attenuation (22.6 versus 8.2 HU; P = 0.046; AUC 0.68, 95% CI 0.51–0.84). The two were only weakly correlated (ρ = 0.24; P = 0.129). Entered together, mask volume remained borderline significant (odds ratio [OR] 5.26 per 1-SD increase, 95% CI 1.00–27.55; P = 0.050) while mean attenuation retained a comparable effect size but lost significance (OR 1.88, 95% CI 0.86–4.15; P = 0.116). Minimum and maximum attenuation and the SD of attenuation did not differ significantly. Conclusions Quantitative low-dose CT parameters distinguish metastatic from tumor-free SLN regions without additional imaging, contrast, or radiation, with nodal volume the most robust discriminator and a concordant density signal whose independence a larger sample should be able to confirm. These hypothesis-generating findings require validation in larger cohorts.</p>