Evidence for a bimodal relation between serum lysozyme and prognosis in 232 patients with acute myeloid leukaemia.

Lysozyme values are sometimes used as an aid for diagnostic subtyping of acute myeloid leukaemia (AML), since monocytic forms often show high levels. We wanted to study if pretreatment serum lysozyme has any relation to prognosis in AML. For this purpose, 232 adult AML patients who had received remission induction therapy at two hospitals were reviewed retrospectively. Their median age was 65.5 yr. Sixty-three patients were FAB classified as "monocytic" AML (M4, M5) and 169 as "non-monocytic" AML (M0, M1, M2, M3, M6). A linear relation was rejected, and a bimodal relation was found between lysozyme and prognosis where values below 20 or above 80 mg L-1 were indicative of better outcome than values in the range 20-80 mg L-1. Analysed in three categories with cut-off levels at 20 and 80 mg L-1, lysozyme showed an independent effect on complete remission (CR) frequency (P = 0.0003), overall survival (P < 0.0001), and CR duration (P = 0.0005) in multivariate analysis. The hazard ratios (HR) for lysozyme <20, 20-80, and >80 mg L-1 regarding overall survival were 1.0, 3.3, and 0.7. Influence of lysozyme on survival was bimodal both in "non-monocytic" AML (HR 1.0, 3.0, and 0.1) and M4-M5 (HR 1.0, 10.1, and 1.2). Our finding of a bimodal relation between serum lysozyme and prognosis in AML should be regarded as a new hypothesis and controlled in other studies.