High Incidence of Osteoporosis in a Natural History Cohort of Patients with Moderate to Severe Chronic Graft-Versus-Host Disease and Risk Factor Analysis

The NIH chronic GVHD (cGVHD) Consensus Project Ancillary and Supportive Care Guidelines prescribe yearly monitoring of bone mineral density (BMD) based on practices in other chronic diseases. We hypothesize that patients with cGVHD have a high incidence of osteoporosis due to additive risk factors including inflammation, hypogonadism, malnutrition, and chronic steroid use.  A large group of cGVHD patients was analyzed to determine the prevalence of osteoporosis and characterize risk factors.

258 cGVHD patients were enrolled in the NIH/NCI Natural History protocol (NCT00092235), a cross-sectional study where patients undergo a one-week comprehensive evaluation including DEXA. T-scores were calculated at the femoral neck (FN), lumbar spine (LS), and total hip (TH). Osteoporosis was defined by BMD (T-score ≤-2.5). Patients were placed into 2 groups, osteoporosis and non-osteoporosis and analyzed for factors associated with osteoporosis initially by standard univariate statistical tests; p values<0.005 were considered statistically significant while tests with 0.005<p<0.05 exhibited strong trends towards statistical significance. Multivariable analyses were performed using multiple logistic regression analysis.

Patients (145 M, 113 F) with median age of 48 years (20-71), enrolled a median of 692 days (0-6670) after cGVHD diagnosis, with a median of 5 affected organs (1-8), received a median of 4 prior systemic therapies (0-9), and a median equivalent prednisone dose of 0.078mg/kg (0-3.38). 183 patients (71%) had severe, 73 (28%) moderate, and 2 (1%) mild cGVHD by NIH Global score. 42 (16%) patients had osteoporosis in at least 1 location with incidence highest at the FN (33; 13%) followed by LS (24; 9%) and TH (15; 6%). Female gender (FN: p=0.04), lower weight (FN: p<0.0001; LS: p=0.0002; TH: p<0.0001), higher platelet count (FN: p=0.007; TH: p=0.003), higher NIH average organ score (FN: p=0.04), higher prednisone dose (LS: p=0.03), and lower C3 (LS: p=0.007) were associated with osteoporosis. Factors not associated included age, vitamin D3, estradiol, FSH, LH, testosterone, number of prior systemic therapies, among others. Multiple logistic regression analyses determined that BMI, prednisone dose, and number of prior regimens were potentially predictive for spinal osteoporosis, while FN and TH osteoporosis were potentially predictable by weight and platelet counts.

Prevalence of osteoporosis was high (16%) which supports current recommendations of regular BMD monitoring.  Classic risk factors such as age, hypogonadism, and steroid use were not found to be significantly associated, factors reflecting more severe disease and inflammation such as low weight and elevated platelets were found to be associated. This suggests a need to focus on cGVHD control as a means to prevent osteoporosis, a hypothesis which needs further testing in prospective studies.