Abstract
The etiology of systemic lupus erythematosus (SLE) still remains unknown and the disease entity named SLE is not definitely but empirically constructed by the combination of the signs and symptoms (SS), and laboratory findings which are frequently observed in typical SLE and SLE-like diseases. Using the mathematical clustering technique, internal structure of SLE has been studied as to whether it can be divided into smaller groups. The Ward's method was adopted because the preliminary clustering experiments using SLE and progressive systemic sclerosis patients revealed : that the Ward's method was most preferable to treat them among seven hierarchical cluster analyses included in CLUSTAN I program written by Wishart. Each SLE case was represented in the form of multi-dimensional vector shown by existence or non-existence of 57 SS. The temporal relation of SS was not taken into consideration. The results were as follows : 1. When the original SLE group (54 cases) was divided into two subgroups, one (A) contained 30 cases, and the other (B), 24 cases. 2. Incidences of palmar erythema, general edema, psychosis, arthralgia, hypercholesteremia, hypoalbuminemia, proteinuria, hematuria and cellular casts in urine in Group B were considerably higher than those in Group A (statistically significant at the 5 % level). Relation of clustering method to the results of classification and clinical significance of the mathematically derived subgroups have not been fully discussed.
