Use of a Claims Database to Characterize and Estimate the Incidence of Castleman's Disease

Castleman's Disease (CD), first described in 1956 by Benjamin Castleman and also known as angiofollicular or giant lymph node hyperplasia, is characterized by lymphadenopathy resulting from abnormal proliferation of B lymphocytes and plasma cells. The clinical manifestations of the disease range from asymptomatic discrete lymphadenopathy, Unicentric Castleman's Disease (UCD), to a more severe form with recurrent episodes of diffuse lymphadenopathy and multi-organ involvement, Multicentric Castleman's Disease (MCD). In a discrete subpopulation, MCD is associated with HIV and Human herpesvirus-8 (HHV-8) infection. Incidence and prevalence of this rare disease is generally unknown but the hypothesized prevalence of CD ranges between 30,000 and 100,000 in the US. There are no codes to accurately identify CD, UCD or MCD in national databases.

The objective of this effort is to propose an algorithm identifying and characterizing potential CD patients to estimate related incidences based on available information and patient characteristics from a national claims database.

The diagnostic and procedure codes that are used in diagnosing lymphadenopathy patients in a longitudinal commercial claims database were used to define and characterize a cohort of likely CD patients. Included patients were required to be continuously enrolled in the database for at least three years. Patients with a history of rheumatoid arthritis (RA), Lupus, Hodgkin's disease (HD), Non-Hodgkin's Lymphoma (NHL) or other malignancies were excluded. Patients must have a diagnosis code of ICD-9 = 785.6 indicating lymphadenopathy followed by a lymph node biopsy (procedure code = 38505) on or within two years after the initial ICD-9 diagnosis date. Patients who had a diagnosis, within 1 year after the initial ICD-9 diagnosis date, of RA, Lupus, HD, NHL or other malignancies were excluded. Once the pool of potential CD patients was identified, it was further characterized according to age, gender, prescription medication use and comorbidities, particularly the development of NHL. To estimate incidence of the disease in the US, the number of patients identified in the pool adjusted for the total patient-time in the database was applied to US census population.

In a claims database of nearly 27 million patients (N=26,982,399), 16,967 patients were identified with an ICD-9 diagnosis of 785.6 and a biopsy code of 38505. After the additional exclusion criteria based on a three year continuous enrollment and a biopsy code on of within two years after the index diagnosis date and excluding patients with prior RA, Lupus, HD, NHL or other malignancies, there were 2,094 patients. After further excluding patients who developed RA, Lupus, HD, NHL other malignancies or HIV within one year after the diagnosis date, 1,153 patients remained with potential diagnosis of CD. These patients were mostly female (64.4%, N=742) and the mean and median age of this cohort were 43 and 45 years, respectively. The most common comorbid conditions were swelling in head/neck, followed by malaise and fatigue and pain in limb. Among these patients, only a quarter were taking a steroid or chemotherapy. Steroid use (dexamethasone and prednisone) was the more common treatment; rituxan usage was documented for less than 1% of the cohort. Based on this sample the incidence rate of CD is 21 cases per million person-years. Applying this rate to the 25 years and older US population, assumed to be 207,301,600 in 2011, the incidence of CD in the US is 4,353 patients.

To date, a thorough investigation of the incidence of CD in the US has not been done. Within the limitations of this effort, a plausible strategy using a US claims database to identify and characterize patients with CD was developed and confirms the very low incidence of CD. Furthermore this effort characterizes potential CD patients as young adults, and only very few of them receive treatment with steroids or chemotherapy. The low chemotherapy usage may suggest that there are very few MCD patients in this cohort. The value of this effort is to demonstrate a relatively quick and cost-effective strategy to characterize the epidemiology and medical need for a rare disorder.

Mehra:Janssen Global Services, LLC: Employment. Cossrow:Janssen Global Services, LLC: Employment. Stellhorn:Janssen Global Services, LLC: Employment. Vermeulen:Janssen Biologics Europe: Employment. Desai:Janssen Global Services, LLC: Employment. Munshi:Janssen Global Services, LLC: Consultancy.