A seventeen-year-old black boy sustained a hyperextension injury of the left knee while playing in a high-school football game in September 1994. He was able to continue playing and did not notice any swelling about the knee until three days after the injury. He completed the football season, but he reported intermittent swelling and repeated episodes of giving-way of the knee associated with pain in the proximal part of the tibia. The pain was severe enough to awaken him at night periodically. When he was first seen in December 1994, the findings on radiographs of the knee were unremarkable. When he was seen again in March 1995, the pain had worsened progressively and continued to awaken him at night. However, he still had been able to participate in weight-training. Aspirin relieved most, but not all, of the pain for short periods. He reported occasional epistaxis and rhinorrhea, but he had no systemic symptoms. Computerized tomography scans and repeat radiographs of the knee were made, and the patient was referred for orthopaedic oncological evaluation with a presumptive diagnosis of osteosarcoma.
Physical examination revealed a robust boy who was six feet (183 centimeters) tall, weighed 240 pounds (109 kilograms), and walked with a slightly antalgic gait. Examination of the left knee revealed a full, pain-free active range of motion. There was no effusion, tenderness at the joint line, or laxity, but there was tenderness over the medial aspect of the proximal part of the tibia between the region of the pes anserinus and the tibial tubercle. There was no cervical, axillary, popliteal, or inguinal adenopathy; no hepatosplenomegaly; and no palpable abdominal mass.
Laboratory studies revealed a white blood-cell count of 5100 per cubic millimeter (5.1 x 109 per liter) (normal, 4500 to 13,000 per cubic millimeter [4.5 to 13.0 x 109 per liter]), with 44 per cent neutrophils (normal, 28 to 68 per cent), 41 per cent lymphocytes (normal, 13 to 52 per cent), 9 per cent eosinophils (normal, 0 to 5 per cent), 1 per cent basophils (normal, 0 to 2 per cent), and 5 per cent monocytes (normal, 0 to 10 per cent). A mild normochromic, normocytic anemia was noted, with a hemoglobin of 125 grams per liter (normal, 135 to 180 grams per liter) and a hematocrit of 0.37 (normal, 0.41 to 0.53). Electrolyte levels were all within normal limits, with 139 millimoles of sodium per liter (normal, 137 to 145 millimoles per liter); 4.3 millimoles of potassium per liter (normal, 3.6 to 5.0 millimoles per liter); ninety-nine millimoles of chloride per liter (normal, ninety-eight to 107 millimoles per liter); twenty-nine millimoles of bicarbonate per liter (normal, twenty-two to thirty-one millimoles per liter); 4.7 milliequivalents of calcium (2.4 millimoles) per liter (normal, 4.2 to 5.2 milliequivalents [2.1 to 2.6 millimoles] per liter); and 4.2 milligrams of phosphorous per deciliter (1.36 millimoles per liter) (normal, 2.5 to 4.5 milligrams per deciliter [0.8 to 1.45 millimoles per liter]). The results of the liver function tests were also normal, with levels of aspartate aminotransferase of twenty-seven international units per liter (normal, twenty to fifty-seven international units per liter); alanine aminotransferase, twenty-two international units per liter (normal, twenty-one to seventy-two international units per liter); alkaline phosphatase, 111 international units per liter (normal, thirty-eight to 126 international units per liter); and total bilirubin, 0.5 milligram per deciliter (8.6 micromoles per liter) (normal, 0.2 to 1.3 milligrams per deciliter [3.4 to 22.2 micromoles per liter]). The erythrocyte sedimentation rate was elevated at thirty-three millimeters per hour (normal, zero to fifteen millimeters per hour). The C-reactive protein also was elevated to fifty-two milligrams per liter (normal, zero to ten milligrams per liter). Antinuclear antibody testing had negative findings. Serum protein electrophoresis showed the beta and gamma fractions to be slightly elevated to 13 per cent (normal, 6 to 12 per cent) and 20 per cent (normal, 8 to 18 per cent), respectively.
Radiographs revealed a sclerotic, poorly defined lesion in the left proximal tibial metaphysis and extending into the epiphysis (Figs. 1-A and 1-B). Magnetic resonance imaging of the knee showed an area of low signal intensity indicative of the sclerotic lesion in the proximal part of the tibia extending from the tibial tubercle posteriorly and proximally across the physis into the subchondral region (Fig. 2). A computerized tomography scan of the proximal part of the tibia (Fig. 3), performed before the patient was referred to us, revealed an osteosclerotic area with scattered lytic foci, which corresponded anatomically with the findings on magnetic resonance imaging. No cortical erosion, periosteal reaction, or soft-tissue extension was evident.
The findings on radiographs and computerized tomography scans of the chest were negative for metastatic disease and lymphadenopathy. A bone scan revealed intense uptake in the proximal part of the left tibia, which corresponded to the area of the lesion on the plain radiographs, and two areas of less intense uptake were seen in the right orbit and the left paranasal sinus area. No clear abnormalities were revealed on subsequent computerized tomography scans of the skull.
In April 1995, an open biopsy of the proximal part of the left tibia yielded fragments of sclerotic cancellous bone. The initial results of the biopsy were consistent with florid chronic osteomyelitis. Three of the four cultures of material obtained on biopsy were negative, and one culture showed growth of two colonies of a corynebacterium species. No fungal organisms or mycobacteria grew on culture. Intravenous administration of cephalexin was started to cover for the usual pathogens found in osteomyelitis as well as the corynebacterium that grew on culture. Three weeks postoperatively, the patient continued to have severe pain. As the cultures were negative and the patient had not had improvement, we suspected that the diagnosis of chronic osteomyelitis was incorrect, and the slides of the biopsy specimens were re-examined. The pathological differential diagnosis included Rosai-Dorfman disease and malignant lymphoma as well as the original consideration, osteomyelitis.
Histological examination showed prominent deposition of new bone with considerable thickening of the trabeculae and prominent cement lines, a cellular infiltrate in the marrow, and regions of infarct-like necrosis (Fig. 4-A). The cellular infiltrate included mature lymphocytes intermixed with large aggregates of plasma cells without atypia (Fig. 4-B). Scattered through the background were occasional histiocytes containing abundant pale pink cytoplasm. Engulfed cells, which were seen within the cytoplasm of the histiocytes, usually consisted of lymphocytes, but occasionally plasma cells, erythrocytes, and neutrophils were seen (Fig. 4-C). Immunohistochemical studies revealed reaction of the histiocytic cells with antibody to S-100 protein (Fig. 4-D). The findings on T-cell receptor and immunoglobulin gene-rearrangement studies, performed to exclude a non-Hodgkin lymphoma, were negative.
The final diagnosis of osseous involvement by Rosai-Dorfman disease was made. Pathologists from other institutions who reviewed the histological specimens agreed with this diagnosis. Because of the persistent symptoms, the patient was managed with curettage of the lesion in the proximal part of the tibia and insertion of autogenous and allogenic bone graft. After the procedure, histological examination revealed that the tissue was identical to that obtained at the first procedure, with negative findings on repeat culture. In addition, the pain in the proximal part of the tibia as well as the epistaxis and the rhinorrhea resolved. Three months postoperatively, the patient was pain-free and the erythrocyte sedimentation rate as well as the C-reactive protein values were normal. During the ensuing three months, the tibial pain and epistaxis returned twice for short periods. Unilateral axillary adenopathy was first noted at the six-month postoperative evaluation, but it resolved within the next three months. The patient was offered a trial of chemotherapy, but he chose continued observation instead. The symptoms were quiescent two years postoperatively.
Rosai-Dorfman disease was first described in 1969 by Rosai and Dorfman as a distinct clinicopathological entity that they termed sinus histiocytosis with massive lymphadenopathy12. By founding a registry of patients who have Rosai-Dorfman disease and by continuing to analyze the accumulated cases, they have elucidated the clinical spectrum of the disease to a great extent2. The most common clinical presentation of Rosai-Dorfman disease is a painless, bilateral cervical lymphadenopathy with fever, an elevated erythrocyte sedimentation rate, leukocytosis, mild anemia, and hypergammaglobulinemia2,7,12,13. The disease may affect patients of any age, but the mean age of onset was 20.6 years among the 395 patients in the registry for whom age was designated2. Of the 305 registry patients for whom race was designated, 43.6 per cent were black, 43.6 per cent were white, 4.6 per cent were Oriental, and 8.2 per cent belonged to another racial group or were of mixed race2.
The diagnosis is made by identification of characteristic histological features of the involved tissue, which is typically a lymph node2,12,13. On histological analysis, the sinuses of the lymph node are expanded by a large number of histiocytes that often show lymphocytophagocytosis, which is characteristic of the disease but not pathognomonic16. Mature plasma cells often are seen with occasional neutrophils within the affected lymph nodes. Capsular and pericapsular fibrosis is often present. In extranodal sites, the histological diagnosis of Rosai-Dorfman disease may be difficult because fewer of the characteristic histiocytic cells are present, the lymphocytophagocytosis is less prominent, and the lesions are fibrotic2.
Extranodal involvement has been reported in 182 of the 423 patients in the registry2. The most common extranodal sites are the skin, nasal cavity, paranasal sinuses, soft tissue, eyelids, orbits, and bones2,7,12-14,16. The frequency of lymphadenopathy accompanying extranodal disease varies and depends on the site of involvement2.
Nine patients in the registry had osseous involvement by Rosai-Dorfman disease in the absence of adenopathy2. Three of the nine patients lacked other sites of extranodal involvement and had isolated osseous disease on presentation. Only four such patients, including those three, have been reported on in the literature as far as we know1,2. The four patients ranged in age from eighteen months to fourteen years at the time of presentation. The radius was involved in two patients, and a metacarpal bone and the temporal bone were involved in one each. Radiographs showed that all of the lesions were lytic1.
The clinical differential diagnosis of Rosai-Dorfman disease includes malignant histiocytosis, Langerhans-cell histiocytosis, lymphoma, metastatic melanoma or carcinoma, and a variety of infections, such as leprosy and histoplasmosis2. When bone is involved, the typical multicentric lytic lesions easily can be confused with Langerhans-cell histiocytosis, although a metastatic malignant tumor, sarcoidosis, storage disorders, and the bone lesions of neurofibromatosis also may be indistinguishable from Rosai-Dorfman disease2,16.
Analysis of the Rosai-Dorfman disease registry revealed that thirty-three of the 423 patients had findings suggesting involvement of bone2,4,11. With the addition of four other patients with Rosai-Dorfman disease who were reported on in the literature1,3,7,8 but were not included in the registry, a total of thirty-seven patients have had involvement of bone. Twenty-one of those patients had involvement of more than one bone, and sixteen had involvement of a single bone. The skull, tibia, and femur, in that order, are involved most frequently, but nearly every bone of the body has been affected by Rosai-Dorfman disease2,4,11. Almost all of the lesions are medullary and lytic, although some surrounding sclerosis has been noted on occasion1-4,7,8,11. The borders vary from well defined to permeative. Purely sclerotic lesions are rare, with the registry including only three patients who had such lesions2.
Of the sixteen patients who had apparently isolated involvement of bone similar to that in our patient, five had secondary erosive changes associated with a contiguous soft-tissue lesion1-3,7,8. Solitary bone lesions caused by Rosai-Dorfman disease were identified in the remaining eleven patients. The skull has been reported as the most common site of involvement in patients who have a solitary bone lesion1-3. Solitary involvement of the femur, radius, ulna, and metacarpal have been reported2,8. Eleven of the thirty-seven patients who had Rosai-Dorfman disease with osseous involvement had no evidence of lymphadenopathy on presentation1-3,7,8.
The pathogenesis of Rosai-Dorfman disease is unknown5. Proposed mechanisms include an occult chronic infection, an aberrant exaggerated immune response to an infectious agent or another antigen, or a neoplastic proliferation of histiocytes. Some authors have advocated inclusion of Rosai-Dorfman disease with the histiocytoses10. Two hystiocytic disorders, Langerhans-cell histiocytosis (eosinophilic granuloma and the more aggressive forms) and Erdheim-Chester disease (a rare type of histiocytosis affecting adults), also have an obscure pathogenesis and bear some resemblance to Rosai-Dorfman disease. The spectrum of clinical findings produced by those disorders is similar. Most patients have mild, localized, and self-limited lesions, and a smaller proportion have progressive disease that causes considerable morbidity or death. Variation in the anatomical distribution of the lesions and in the characteristic findings associated with these disorders may be related to the properties of the type of histiocytes involved.
Few patients who have Rosai-Dorfman disease need specific treatment, and almost all patients eventually have spontaneous regression6. Incisional or excisional biopsy has been the sole operative procedure in most patients. From an orthopaedic standpoint, isolated or unusual bone lesions usually have only been biopsied by curettage6, although in unusual circumstances resection of the cuboid8, amputation of a ray9, and decompressive laminectomy16 have been performed. The only guidelines for systemic treatment are derived from retrospective data, as there have been no prospective, controlled trials to our knowledge. Most physicians manage patients who have extensive or progressive disease with regimens known to be successful for treatment of malignant hematological disease and histiocytosis6. Radiation therapy typically has been reserved for treatment of compression of the spinal cord or respiratory compromise6.
The course of Rosai-Dorfman disease varies2, but it has a good prognosis over-all. At the most recent follow-up examination, after a minimum duration of one year, forty-nine (21 per cent) of 238 registry patients were alive without disease, thirty-six (15 per cent) possibly had persistent disease, and 129 (54 per cent) had stable disease. Three additional patients had progressive disease at that follow-up evaluation. The remaining twenty-one patients had died. The disease may persist, progress, or follow a course of exacerbation and remission. Four of the 423 patients in the registry who had sinus histiocytosis with massive lymphadenopathy died of progressive disease. In addition, thirteen of the patients in the registry, including three of the thirty-three who had osseous involvement, died but the cause of death was not definitely associated with sinus histiocytosis and massive lymphadenopathy. Death usually occurred in association with immunological disease and involvement of unusual sites, such as the kidneys and the lower respiratory tract2. Radiographically, these lesions frequently resolve completely or they may persist as an area of sclerosis15.
We reported the rare case of a patient who had a sclerotic osseous lesion caused by Rosai-Dorfman disease with involvement limited to bone on clinical presentation. The unique clinicoradiographic presentation was important because the sclerotic appearance of the lesion on radiographs resembled osteosarcoma or the chronic sclerosing form of osteomyelitis and the initial results on histological examination were mistaken for osteomyelitis. Cervical lymphadenopathy, which nearly always accompanies this disease, never developed. We cannot exclude the possibility that the paranasal sinuses were a second extranodal site of involvement, as there was clinical and radiographic evidence of a lesion, but biopsy of the region showed none of the characteristic histological features. It is of interest that nineteen (58 per cent) of the thirty-three patients in the registry who had osseous Rosai-Dorfman disease had additional sites of extranodal involvement, with the nasal cavity and paranasal sinuses being the most common extranodal sites2.
NOTE: The authors thank Dr. Krishnan Unni and Dr. David C. Dahlin of the Department of Pathology, Mayo Clinic, Rochester, Minnesota, and Dr. Juan Rosai of the Department of Pathology, Mount Sinai Medical Center, New York, N.Y., for their review of the histological tissue.