Four women, aged 63 to 90 years old, presented with mildly painful shoulders of decreased mobility or stability. Radiographic evidence of a complete tear of the fibrous rotator cuff was present in 7 of 8 shoulder joints. Microspheroids containing hydroxyapatite crystals were seen by scanning electronmicroscopy in 12 of 13 synovial fluid samples. All synovial fluids showed activated collagenase and neutral protease activity. This constellation of findings represents a heretofore undescribed syndrome which we have designated "Milwaukee shoulder."The shoulder is the most mobile joint in the human body. The glenohumeral joint contributes most to this remarkable range of motion (1). Such mobility has been achieved at the sacrifice of joint stability, which depends largely on the integrity of the fibrous conjoint tendons of the surrounding muscles+ommonly called the rotator cuff (1).

Hydroxyapatite crystals in spheroid-shaped masses 1.9-15.6~ in diameter were found in 12 of 13 synovial fluids obtained from the shoulder joints of 4 patients with rotator cuff tears and glenohumeral osteoarthritis. Two of 16 control joint fluids also showed these particles. Collagen types I, 11, and 111 were identified in the joint fluid pellets from 3 of the 4 patients, and fibers with typical collagen periodicity were also seen on transmission electronmicroscopy. Collagenase and neutral protease activities were found in fluids from 5 joints in 3 patients, whereas active collagenase was found in only 1 of 10 fluids from rheumatoid arthritis patients and in none of 3 fluids from patients with osteoarthritis. Neutral protease activities were present in several rheumatoid joint fluids. These findings are compatible with the hypothesis of an enzymatic release of hydroxyapatite crystals from the synovium and endocytosis by synovial macrophage-like cells with subsequent crystalstimulated release of collagenase and neutral protease into the joint fluid, completing a pathogenetic cycle. The clinical and radiographic characteristics of 4 women patients with bilateral shoulder joint problems are detailed in the preceding report (1). Rupture of the rotator cuff was proved by arthrography of both shoulders in 2 patients (KH and CG) and was present by inference in 3 shoulder joints in 2 other patients (JS and MB) because of marked superior displacement of the humeral head on standard roentgenograms. KH and CG both showed multiple focal periarticular calcifications, but JS and MB did not. Synovial fluids were obtained on at least one occasion from at least 1 shoulder joint in all 4 patients; 12 of 13 fluids contained hydroxyapatite crystals within microspheroids. Active collagenase* and neutral protease were found in the synovial fluids from these patients. Details relative to these morphologic and biochemical studies are given in this communication. MATERIALS AND METHODS14C EHDP binding assay. All fluids were studied initially using the I4C ethane-I-hydroxy-1, I-diphosphate (EHDP) binding assay described previously (2). In addition to the method previously described, henceforth called protocol I, a second method of sample preparation was used to increase the sensitivity of this assay. This method, henceforth called protocol 11, is as follows: freshly aspirated synovial fluid was collected in plastic tubes calibrated to 5 or 10 ml and containing 250 units of heparin and 5 mg hyaluronidase as previously described (2); after 30 minutes incubation at room temperature and centrifugation at 10,OOOg for 20 minutes at 4"C, the supernatant was removed; the original volume was reconstituted with 0.1M Tris chloride buffer, pH 8.0, and the sediment was resuspended by using a vortex mixer. After centrifugation as before, all but 1.0 ml of supernatant was removed and discarded; 100 p l of freshly prepared 1% trypsin (Worthington * The term "collagenase" is used herein to indicate "collagenolytic activity."

The initial transient [Ca2+]i increase probably serves as a second messenger leading to activation of early cellular responses such as c-fos expression which is important in BCP crystal-induced mitogenesis. The second, slower and more sustained rise of [Ca2+]i probably initiates other cellular processes needed for fibroblast mitogenesis.

Hydroxyapatite (HA) and calcium pyrophosphate dihydrate (CPPD) crystals were phagocytosed when added to cultured human rheumatoid or normal canine synovial cells. Collagenase and neutral protease secretion into the culture medium was increased 5-to 8-fold over control values in the presence of HA and increased 3-fold in the presence of CPPD crystals. HA but not CPPD crystals induced a 300-fold increase in human rheumatoid synovial cell culture fluid prostaglandin (PG) E2 levels and an 8-fold increase in PGFa levels. This mechanism may be important in the pathogenesis of the destructive arthropathies associated with HA and CPPD crystals.Recently we described a syndrome affecting the shoulder joints of 4 patients ("Milwaukee shoulder syndrome"). Bilateral rotator cuff defects, glenohumeral osteoarthritis, and joint stiffness or instability were accompanied in each case by the finding of hydroxyapatite (HA) crystal clumps (microspheroids), activated collagenase, neutral protease, and particulate collagen types I, 11, and 111 in a nearly acellular

obtained for crystal identification and radiographs were available. Synovial fluids were examined by compensated polarised light microscopy, and leucocyte counts were determined as described elsewhere.9 BCP crystals were identified by the binding of radiolabelled diphosphonate, followed by scanning electron microscopy with energy dispersive analysis to determine the calcium to phosphorus molar ratio as described previously.' Anteroposterior and lateral radiographs of the knees were read blindly by both authors. The presence of joint space narrowing and osteophytes in each of the three (medial and lateral tibiofemoral and patellofemoral)