In their review of 81 cases, Keller and colleagues [3] found that all patients with localized Castleman disease had a benign clinical course. Complete local surgical excision totally eliminated the systemic symptoms present at diagnosis, with only occasional local recurrences. About 80% of cases of localized Castleman disease were hyaline vascular, and 20% were plasma-cell; only an occasional case of mixed disease was reported. Most patients with localized plasma-cell disease were children or young adults, and they presented with abdominal or mediastinal masses (peripheral masses were rare). Some had constitutional symptoms and laboratory abnormalities, such as anemia, hypoalbuminemia, hypergammaglobulinemia, and elevated erythrocyte sedimentation rates [3, 5].

Gaba and associates [6] reported the first case of multicentric Castleman disease in 1978. Most multicentric cases subsequently described have been of the plasma-cell type, although hyaline vascular and mixed cases have occasionally been reported [7-9]. The multicentric plasma-cell form of the disease is histologically similar to the localized plasma-cell type, but the two have major clinical differences. In contrast to the localized form, multicentric Castleman disease is a systemic illness with disseminated lymphadenopathy; its aggressive and usually fatal course is associated with infectious complications and risk for malignant tumors, such as lymphoma or Kaposi sarcoma [10, 11]. Multicentric Castleman disease has also been described in association with HIV infection [11]. Because limited information is available about therapy for the multicentric form of the disease, we analyzed the clinical features, management, and outcome of patients in whom Castleman disease was diagnosed at our institutions.

Methods

Top Methods Results Discussion Author & Article Info References

We reviewed the medical reports on 17 consecutive patients who 1) received a histologic diagnosis of Castleman disease at Texas Medical Center, Houston, Texas, between March 1977 and October 1995 and 2) had information on clinical follow-up available. The histologic material was reviewed by two pathologists who were blinded to clinical findings and the clinical course of the patients, and agreement was reached for each patient. In 15 of the 17 patients, the diagnosis of Castleman disease was confirmed; the other 2 patients had nonspecific histologic changes and were excluded from our analysis. Eight of the patients were identified from the files of the University of Texas M.D. Anderson Cancer Center, and 7 were identified from the files of hospitals affiliated with Baylor College of Medicine: Methodist Hospital, St. Luke's Episcopal Hospital, Ben Taub Hospital, and the Veterans Affairs Medical Center. Clinical information was obtained from a review of patient charts and from the attending pathologists and clinicians. In this paper, we describe the clinical features at presentation, clinicopathologic correlations, treatments, and outcomes of the 15 patients.

As earlier investigators have done, we classified the patients as having one of two distinct clinical presentations: localized or multicentric. Patients were defined as having localized disease if they presented with only one group of lymph nodes histologically involved by Castleman disease; they were defined as having multicentric disease if they presented with involvement of two or more lymph node groups. The definition of extranodal involvement required a biopsy result that showed the features of Castleman disease in the involved organ. Systemic symptoms were attributed to Castleman disease if no other cause was found. Because the histologic features of plasma-cell Castleman disease may be found in patients with many reactive lymph node conditions, such as rheumatoid arthritis, syphilis, or skin diseases, and in patients with nodes draining malignancies and patients with the Wiskott-Aldrich syndrome [12], we excluded these possibilities before diagnosing plasma-cell Castleman disease.

Results

Top Methods Results Discussion Author & Article Info References

Localized Castleman Disease

Diagnosis

Of the 15 patients in our series, 7 met the criteria for localized Castleman disease. These patients were analyzed according to the histologic type of disease present (Table 1). No erythrocyte sedimentation rate at diagnosis was available for any of these patients.

Four (57%) of the patients with localized disease had histologic evidence of the hyaline vascular type of disease. None of these four patients had constitutional symptoms at diagnosis. This histologic type of disease was not associated with mediastinal node involvement in any patient. Two patients had histologic evidence of the plasma-cell type of disease; both presented with constitutional symptoms. One patient had histologic evidence of mixed-type disease.

Initial Treatment

The primary treatment for all seven patients with localized disease was complete surgical resection. All of these patients are currently alive with no evidence of recurrence. The location of the lymphadenopathy did not influence outcome.

Multicentric Castleman Disease

Diagnosis

Eight patients met the criteria for multicentric disease. These patients were analyzed according to the histologic type of disease present (Table 2). All of these patients, regardless of the histologic type of disease, had constitutional symptoms and laboratory abnormalities at presentation. One patient had hyaline vascular disease, four (50%) had plasma-cell disease, and three (37%) had mixed disease.

Initial Treatment

The initial therapy for patients 10, 13, and 15, all of whom had multicentric disease, was combination chemotherapy (Table 2). Patient 10 received cyclophosphamide, vincristine, doxorubicin, and dexamethasone; patient 15 received cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP); and patient 13 received chlorambucil and prednisone. Patient 10 attained a partial radiologic response but still had stable residual retroperitoneal adenopathy. Because no biopsy was done, it is not known whether this represented residual tumor or scar tissue. Patient 10 developed mediastinal adenopathy 3 years after diagnosis of Castleman disease, and a biopsy result was diagnostic of sarcoidosis.

Patients 8 and 11 had surgery as initial therapy. Patient 8 underwent debulking surgery, and systemic symptoms improved. Disease recurred in the same area 2 years later and was treated with prednisone for a few weeks without response; this treatment was discontinued after an episode of severe upper gastrointestinal bleeding. At that time, patient 8 developed new weakness in the left hand that failed to resolve despite several courses of intravenous immunoglobulin. He became progressively weaker and continued to lose weight. He then received four cycles of CHOP without response. His last course was complicated with pneumonia and sepsis, and he received no further chemotherapy. He died 34 months after initial diagnosis and 1 year after his relapse. Patient 11 underwent complete surgical resection for abdominal and axillary adenopathy. He had a recurrence of Castleman disease in the submaxillary area 8 years later, which was treated with complete surgical excision. Ten months after his relapse, he developed peripheral T-cell non-Hodgkin lymphoma that was resistant to several chemotherapy regimens. The patient died 13 years (156 months) after initial diagnosis and 5 years after his relapse. In both patient 8 and patient 11, the initial disease and the relapses were of the plasma-cell type.

Patients 12 and 14 received prednisone, 1 mg/kg of body weight per day, as initial treatment. Although both patients achieved complete clinical remission while receiving steroids, lymphadenopathy recurred after tapering. Maintenance therapy with prednisone, 20 to 30 mg/d, was required. Both patients remain alive 8 and 35 months after diagnosis.

Patient 9 received no specific treatment during observation and had spontaneous regression of symptoms and palpable lymphadenopathy several months after initial diagnosis. No results of follow-up imaging procedures were available for this patient, and she remains alive and symptom-free 57 months after initial diagnosis.

Discussion

Top Methods Results Discussion Author & Article Info References

Clinical Findings

Castleman disease is a rare disorder. Only three series of the localized form, describing a total of 174 patients [2, 3, 5], and three series of the multicentric form, describing a total of 38 cases [7-9], have been published. After comparing our data with those from other series [2, 3, 5, 7-9], we conclude that the localized and multicentric forms of Castleman disease are two distinct conditions that have similar histologic features but different clinical features, natural history, and responses to treatment.

In the seven patients with localized Castleman disease in our series, the median age at diagnosis was 35 years (range, 15 to 45 years). This is not significantly different from data in previous reports [3, 5]. Similarly, no sex predilection was seen. Constitutional symptoms and laboratory abnormalities were seen almost exclusively in patients with plasma-cell disease; this finding is consistent with previous reports [3, 5]. The systemic manifestations of localized disease seemed to be less florid than those in multicentric disease, suggesting that the number of areas of involved lymph nodes may be related to the intensity of symptoms. Three patients in our series presented with either plasma-cell or mixed localized disease.

In our series, the distribution of lymphadenopathy in patients with localized disease differs from that in previous reports. Keller and colleagues [3] reported that of 81 patients with localized Castleman disease, 86% had disease limited to the mediastinum, 3% had disease limited to the abdomen, and 11% had disease limited to the periphery. In a combined series of 78 patients with localized disease, Frizzera [5] showed a distribution of 46% in the mediastinum, 39% in the abdomen, and 15% in the periphery. Only 1 (14%) of our 7 patients had disease localized to the mediastinum, 2 (28%) had it localized to the abdomen, and 4 (58%) had it localized to peripheral lymph node areas. This difference in the distribution of nodal sites probably does not reflect the more frequent use of routine computed tomography in the evaluation of these patients because the use of computed tomography should have revealed more rather than fewer intraabdominal presentations. This difference is probably due to chance, given the small size of our series.

Multicentric Castleman disease is associated with an increased frequency of infection [7-9], which is the most common cause of death in patients with this condition. Immunosuppression caused by chemotherapy or the long-term use of steroids may also play a role in these occurrences. In some patients, multicentric disease has been associated with HIV infection; in these cases, it has been linked to human herpesvirus-8, also known as Kaposi sarcoma-associated virus [13]. Multicentric disease has also been reported in association with second cancers, particularly lymphoma [10] and Kaposi sarcoma [11]. The association with human herpesvirus-8 may help explain why patients with Castleman disease who do not have HIV infection have developed Kaposi sarcoma. In our series, one patient (patient 2) had a simultaneous diagnosis of Hodgkin disease and Castleman disease (Table 1). Another patient received a diagnosis of non-Hodgkin peripheral T-cell lymphoma 9 years after the initial diagnosis of multicentric Castleman disease. These associations have been noted previously [10, 14].

An interesting clinical observation is the association of multicentric Castleman disease with the POEMS (polyneuropathy, organomegaly, endocrinopathy, monoclonal protein, and skin changes) syndrome [15, 16], which was present in patient 8 (Table 2) [17, 18]. Not every patient with the POEMS syndrome has all of the symptoms associated with the syndrome, and the acronym does not include every manifestation of this rare disorder. Thus, patient 9, who had neuropathy, hepatomegaly, splenomegaly, and skin rash, could have had an incomplete form of the POEMS syndrome (Table 2). Multicentric Castleman disease has also been reported to be associated with amyloidosis [19] and pemphigus vulgaris [20]. Although the cause of these entities is unclear, they probably represent different parts of a spectrum of immune dysregulation. In the case of multicentric Castleman disease, experimental and clinical evidence support a central role for interleukin-6 in pathogenesis [21-25].

Therapy

Localized Castleman disease seems to be almost uniformly curable with surgery [2, 3, 5], but the clinical course of the multicentric form varies [7-9]. A summary of the literature on initial treatment for 44 patients with multicentric Castleman disease is shown in Table 3. Although the effectiveness of treatment is difficult to evaluate retrospectively, most patients receiving steroids or chemotherapy had some clinical response [7-917, 18, 21, 26-30].

Localized Castleman Disease

All three of our patients with plasma-cell or mixed localized Castleman disease were managed by surgical resection, and all three have remained well for periods ranging from 22 to 34 months. Complete surgical resection has been reported to be curative and remains the treatment of choice for localized disease, including plasma-cell and mixed disease [3, 5]. Recurrence of localized cases, even 9 years after initial diagnosis, is rare and has been associated with incomplete surgical removal [3, 31]. Although radiation therapy was previously cited as ineffective in localized disease [3], subsequent reports [32, 33] have described successful outcomes in patients with unresectable lesions.

Multicentric Castleman Disease

In Table 4, patient data from our series are compared with data from past series of patients with multicentric disease.

Although complete surgical excision is clearly the preferred therapy for localized Castleman disease, limited information is available about management of the multicentric form of the disease. The evaluation of various treatment regimens is complicated by the variable course of the disease, which may occasionally include spontaneous remission [34]. As in the two patients in our series who received prednisone as initial therapy, the results of treatment with high-dose steroids have been variable and the treatment must often be prolonged [35, 36]. The contribution of steroids as part of the cytotoxic chemotherapy combinations used in our series is difficult to measure. Nevertheless, when used in these regimens, steroids were usually given for only 5 days. Antibodies against interleukin-6 [37] and other biological immunomodulators, such as interferon- [38], have also been used to treat multicentric disease with transient success. Chemotherapy has included combinations of the kind used for lymphoma; several instances of sustained response have been seen with this approach (Table 3) [8, 9]. Of the eight patients with multicentric disease in our series, three received chemotherapy as initial therapy. All three achieved a prolonged response and are still alive. In contrast to data on localized disease, the available data on multicentric disease are too scarce to allow us to draw any conclusions about the timing or type of chemotherapy best suited to this condition. Although surgery is curative in localized disease, its role in multicentric disease seems to be limited. In one case of multicentric disease, radiation therapy was used as the initial treatment and produced sustained complete remission [39]; in another, no response to radiation therapy was seen [7].

Other investigators have shown steroids to have variable effectiveness, ranging from no response [8, 9] to transient response [7, 8], with respect to disease control while patients are receiving continuous maintenance therapy [7, 8, 17, 29]. None of the 7 patients who did not respond to steroids are alive; of the 16 patients who did have a clinical response, 9 (39% of the total group receiving steroids) are alive.

Chemotherapy (in the form of a single agent, an agent together with steroids, or a combination regimen) produced clinical responses in 14 (93%) reported patients [7, 8, 26-2830, 36]. At the time of analysis, seven patients (47%) were reportedly alive and 4 (27%) had sustained complete remission. Of the 4 who had sustained remission, 2 received chlorambucil alone [35] or with prednisone [27] at the time of diagnosis, 1 received cyclophosphamide plus prednisolone after initial therapy with azathioprine plus prednisolone [28], and 1 received high-dose melphalan with bone marrow transplantation after failure of CHOP [30].

Although all three patients with plasma-cell and mixed multicentric disease in our series achieved a durable response after chemotherapy given at the time of diagnosis (58, 39, and 8 months, respectively [all three patients were still alive at the time of analysis]), it is still not clear which treatment is most effective. Data shown in Table 3 and our own experience suggest that steroids, although effective in producing an antitumor response, have no sustained effects after therapy is discontinued. Chemotherapy, on the other hand, seemed to induce sustained responses in 27% of reported patients. Combining our findings with those reported in the literature would produce a total of 39% of patients treated with chemotherapy who had a sustained response.

In summary, it is not possible to draw definitive conclusions about therapy from the few experiences reported with the many different regimens used in the management of patients with multicentric Castleman disease. Nevertheless, combination chemotherapy might be effective for these patients. A better understanding of the pathogenesis of this rare but remarkable disorder may help determine the best therapeutic approach.

Dr. Cabanillas: Lymphoma Section, Department of Hematology, University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030.

Dr. Rice: Department of Medicine, Section of Hematology-Oncology, Baylor College of Medicine, Methodist Hospital, 6565 Fannin, MS 902, Houston, TX 77030.

Drs. Manning and Pugh: Section of Hematopathology, Department of Pathology, University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030.