Dynamic contrast enhancement in magnetic resonance imaging (DCE-MRI) is a promising tool for the clinical diagnosis of tumors, whose implementation may be improved through the use of suitable hemodynamic models. If one prefers to avoid assumptions about the tumor physiology, empirical fitting functions may be adopted. For this purpose, in this paper we discuss the exploitation of a recently proposed phenomenological universalities (PUN) formalism. In fact, we show that a novel PUN class may be used to describe the time-signal intensity curves in both healthy and tumoral tissues, discriminating between the two cases and thus potentially providing a convenient diagnostic tool. The proposed approach is applied to analysis of the DCE-MRI data relative to a study group composed of ten patients with spine tumors.