This study found a high prevalence of low BMD, fractures, and bone pain in both β and α thalassemia syndromes. It showed a strong association between bone mass and fractures, and it identified factors that may contribute to the pathogenesis of bone disease in thalassemia. Bone disease would be expected to be less severe with mild or moderate hemolysis, as is generally found in patients with β TI, HbH disease, or HbH/CS. Indeed, bone mass was lower among β TM and E-β and higher among β TI, HbH disease, and HbH/CS. However, the effect of thalassemia syndrome seemed to be caused by underlying differences in growth, hypogonadism, and bone turnover, so that hypogonadal and lighter patients with high bone turnover had a higher likelihood of low BMD among all thalassemia syndromes.
Bone mass was reduced even in children-only 45% of 6- to 10-yr-old participants had normal BMD. We found a strong negative association between bone mass and age in the 11- to 19-yr age group. Peak bone mass was suboptimal. Short stature in some patients may have influenced these results, which are based on areal BMD measurements. However, the mean height Z-score was close to the population mean, and correlations with volumetric BMD did not suggest an important effect of bone size on our findings. Lower weight and higher bone turnover were the only two independent predictors of bone mass in patients <20 yr of age, whereas hypogonadism, transfusion and chelation parameters, and family history of osteoporosis were not found to play roles. These data indicate that adolescence is a critical period for the development of low bone mass in thalassemia and suggest that bone turnover plays a central role in this process. Similar to other reports,(23
) we found increased bone turnover in adults with thalassemia, regardless of hypogonadism, and our limited pediatric normative data suggest that the same is true during adolescence. Altogether, it is possible that the increased bone turnover during this time does not allow for positive bone accrual and attainment of optimal peak bone mass. Neither the etiology of increased bone turnover in thalassemia nor a possible disassociation between bone formation and resorption can be addressed in this study. One may hypothesize, however, that the increased resorption in thalassemia relates to a state of increased oxidative stress,(25
) which has been shown to be present in these patients, and which can also result in increased TNF production and bone loss.(26
) Because current chelation and transfusion regimens have remained largely unchanged in this cohort over the last 20 yr, it seems that current thalassemia therapies do not prevent or retard the development of low bone mass.
Hypogonadism was a strong independent predictor of low bone mass. Surprisingly, we did not find higher BMD among hypogonadal subjects who received gonadal steroid replacement compared with those untreated. It is possible that this is related to the cross-sectional design of the study, the delayed or inadequate replacement with gonadal steroids, or intermittent compliance with the prescribed treatment. Gonadal steroid replacement schemes in females of this study were frequently similar to the regimens used in postmenopausal women, which may be inadequate for optimal bone accrual.(28
) Nonetheless, the potential beneficial effect of gonadal steroid replacement in thalassemia should not be dismissed based on this study. An improvement of BMD in thalassemia with gonadal steroid replacement has been reported in some studies but not others.(11
) Thus, this subject needs to be rigorously evaluated by longitudinal randomized studies testing various types of gonadal replacement. Further longitudinal studies are also needed to determine the optimal serum 25 vit D concentrations for normal bone mass accrual and maintenance in thalassemia,(32
) because we found high rates of vitamin D deficiency and insufficiency, as well as a negative and nonlinear relationship between spine BMD and 25 vit D concentrations. These findings may have significant clinical implications because both screening and therapy of vitamin D deficiency are easy, affordable, and beneficial. Information about fluctuations in 25 vit D concentrations over the course of years and their effect on the bone is not available in this study. Overall, the cross-sectional design of this study constitutes a limiting factor in our ability to infer the etiology of bone disease in thalassemia. This limitation, as well as sample size considerations, may also explain the lack of association between bisphosphonate treatment and bone mass. In addition, because there was no control group, our data were compared with standard databases, and the results should be interpreted with this in mind.
Years on regular transfusions, DFO treatment, and transfusion rates were used to assess the long-term effects of thalassemia management and iron overload on the bone. No association was found. Current degree of iron overload (assessed by serum ferritin concentration) was also not associated with bone disease. It is possible that the role of iron in the pathogenesis of the disease was missed because we lacked a definitive marker of iron overload. Ferritin is a poor marker in heavily overloaded individuals. Other measures of iron excess, for example, liver biopsy, superconductive quantum interference device (SQUID), or MRI, were not performed in this study. Finally, ineffective erythropoiesis (assessed by serum transferrin receptor concentrations) was not associated with bone mass, with the exception of an association with femur BMD in univariate analysis of unclear biological significance.
The fracture prevalence based on both patient history and medical record review was 36%. This is substantially lower than the rates of 50% reported in 1960s and 1970s.(34
) Our present results differ from a recent TCRN report that observed an overall fracture prevalence of 12%.(35
) The latter report(35
) most likely underestimated the fracture prevalence because of its reliance on chart review of a TCRN established database that was generic in nature and not comprehensive for prior bone problems. In this study, E-β and α thalassemias were found to have lower rates of fractures compared with β TM. However, as in the case of our BMD results, neither thalassemia syndrome nor chelation and transfusion parameters were significant independent predictors of fracture rates in multivariate analysis. Instead, a strong association between BMD and fracture rate was observed. We also found that, in these patients, as in osteoporotic patients, the cumulative risk for fractures increased with age in an almost linear fashion. A bimodal fracture distribution with one peak related to sports activities among adolescent males and a second rise related to osteoporosis among postmenopausal females has been described in the general population.(36
) Such a bimodal fracture risk was not observed among our patients. Similar to the previous retrospective TCRN study,(35
) children and adolescents did not seem to be at increased risk for fractures. This finding may be related to the cross-sectional design of our study or to the decreased level of physical activity in this patient population.
Whereas long bone metaphyseal irregularities and abnormal vertebral bodies resembling bone dysplasias have been described in β TM and attributed to DFO toxicity,(39
) the presence and etiology of these abnormalities given today's standard of care are largely unknown. This is the first study in thalassemia to use vertebral morphometry to evaluate a large number of patients for such abnormalities. Indeed, a large number of deformities were found, mostly VGDs characterized by decreased height of multiple vertebrae. VGD occurred exclusively in regularly transfused and chelated patients with β TM and E-β, were associated with early use of DFO (before age 6), and occurred regardless of age, with the youngest affected patient being 18 yr old. A rather small number of vertebral fractures were documented, found primarily in patients with β TM. Vertebral fractures increased with age and were related to low bone mass and hypogonadism. These data suggest different etiologies between vertebral fractures and deformities, and in particular, the role of early treatment with DFO in the development of vertebral deformities in thalassemia.
Despite its high prevalence, the etiology of bone pain in our patients remains obscure. There was no correlation with history of fractures or vertebral deformities, except in those with documented vertebral fractures by morphometry, who almost always reported bone pain. Bone pain increased with age and was more frequent among females, despite fractures rates similar to those of males. Patients with β TM also reported bone pain more frequently than those with other thalassemia syndromes, although we were unable to establish any transfusion or chelation related parameter as a possible contributor to pain. Neither endocrinopathy, other thalassemia complications, nor vitamin D deficiency were independent predictors of bone pain. Why females with β TM are more prone to bone pain remains uncertain.
In summary, in this large cohort of patients across all thalassemia syndromes, we observed a high prevalence of fractures and a strong association between bone mass and fractures. Therefore, strategies to improve BMD are very important in thalassemia management. We showed that bone disease in thalassemia is an adolescent problem with adult manifestations. Current transfusion and chelation practices seem insufficient to prevent the development of low bone mass. Our data highlight the need for randomized trials to determine the appropriate form of gonadal steroid replacement and vitamin D supplementation as well as additional strategies to optimize bone accrual in this disease. Further longitudinal studies are needed to address changes of bone mass during puberty. Finally, changes in bone turnover seem to be involved in the development of bone disease, although the factors that lead to increased bone resorption in thalassemia remain unclear and warrant further study.