SummaryIn a series of 84 head and neck patients, a statistically significant correlation was observed between high serum soluble interleukin (IL)-2 receptor alpha (sIL-2Ra) (P = 0·034) and metalloproteinase-9 (MMP-9) concentrations (P = 0·036) at diagnosis and a shorter survival of these patients. As MMP-9 has been shown to mediate cleavage of IL-2Ra (CD25) by preactivated T cells, we looked for a relationship between MMP-9 expression and soluble IL-2Ra serum concentrations in these cancer patients. We did not find any correlation between intratumoral expression of MMP-9 or serum MMP-9 concentrations and serum sIL-2Ra levels. These results led us to reassess the role of MMP-9 in the release of sIL-2Ra. Treatment of Kit225 leukaemic cells with recombinant MMP-9 slightly decreased membrane CD25 expression and was associated with an increased concentration of sIL-2Ra in the supernatants. However, using a selective inhibitor of MMP-9 we did not succeed in specifically inhibiting the release of sIL-2Ra by the Kit225 cell line or by phytohaemagglutinin (PHA)-activated peripheral blood mononuclear cells. In addition, in a preclinical mouse model, basal serum sIL-2Ra concentrations and sIL-2Ra production by activated cells were not altered in MMP-9-deficient mice compared to wild-type mice. Interestingly, a broad spectrum metalloproteinase inhibitor inhibited the release of sIL-2Ra by PHAactivated peripheral blood mononuclear cells, suggesting that in contrast with current views concerning the major role of MMP-9 in the cleavage of membrane IL-2Ra, other proteases are involved in the shedding of sIL-2Ra. MMP-9 and sIL-2Ra appear therefore as independent prognostic markers in head and neck cancers.