Osteoporosis in β-thalassaemia major patients: analysis of the genetic background


Achille Iolascon, Dipartimento di Biomedicina dell'Età Evolutiva, Universita' di Bari, Piazza G. Cesare 11, 70124 Bari, Italy. E-mail: a.iolascon@bioetaev.uniba.it


Regular blood transfusions from infancy until adulthood in β-thalassaemia major patients have substituted severe bone deformities with less marked skeletal lesions as osteoporosis. Osteoporosis is characterized by low bone mass and disruption of bone architecture, resulting in reduced bone strength and increased risk of fractures. Genetic factors have an important role in determining bone mineral density (BMD). We have investigated the possible association between BMD and two polymorphisms in 135 β-thalassaemic patients: (i) a substitution G→Τ in a regulatory region of the COLIA1 gene encoding for the major protein of bone (type 1 collagen), and (ii) a one-base deletion in intron 4 (713–8del C) of transforming growth factor beta 1 (TGF-β1) gene. We have found a remarkable incidence (90%) of osteopenia and osteoporosis among regularly transfused patients. Bone mass was lower in men than in women (P = 0·0023), with a more prevalent osteopenia/osteoporosis of the spine in men than in women (P = 0·001). The sample was stratified on the basis of BMD expressed as Z-score, i.e. normal, osteopenic and osteoporotic patients, and genotype frequencies of each group were evaluated. TGF-β1 polymorphism failed to demonstrate a statistical difference in BMD groups. However, subjects with heterozygous or homozygous polymorphism of the COLIA1 gene showed a lower BMD than subjects without the sequence variation (P = 0·012). The differences among genotypes were still present when the BMD was analysed as adjusted Z-score and when men and women were analysed separately (P = 0·022 and 0·004 respectively), with men more severely affected. Analysis of COLIA1 polymorphism could help to identify those thalassaemic patients at risk of osteoporosis and fractures.

The new transfusion regimens and early iron-chelating therapy have improved the survival of thalassaemia major patients (Olivieri, 1999) and have substituted the marked bone abnormalities previously described (Cooley et al, 1927) with less severe skeletal lesions. In fact, osteopenia with cortical thinning, increased trabeculation of the spine and severe osteoporosis remain serious complications, even in well-transfused and iron-chelated patients (Vichinsky, 1998). Several factors may affect bone metabolism and turnover in β-thalassaemia patients, such as hormone deficiency, vitamin deficiency, iron overload and chelation therapy (Wonke, 1998; Singer & Vichinsky, 1999).

Osteoporosis is a common disease characterized by reduced bone mass and disruption of bone architecture (Kanis et al, 1994). Bone mineral density (BMD) is determined by a variety of genetic and environmental factors and its inheritance is thought to be under polygenic control (Pocock et al, 1987; Slemenda et al, 1991; Soroko et al, 1994). Polymorphisms of several genes have been associated with BMD (Morrison et al, 1994; Grant et al, 1996; Kobayashi et al, 1996; Masi et al, 1998). However, conflicting results have so far been obtained and genetic susceptibility to osteoporosis is not fully understood.

Transforming growth factor beta 1 (TGF-β1) in bone matrix has been implicated as a possible mediator of coupling between bone resorption and formation (Derynck et al, 1985; Bonewald & Mundy, 1990). Langdahl et al (1997) and Bertoldo et al (2000) found that a one-base deletion in intron 4 (713–8delC) of TGF-β1 was associated with severe osteoporosis and increased bone turnover in normal and osteoporotic women. At the same time, the genes encoding for collagen types Iα1 and Iα2 (COLIA1 and COLIA2 respectively) have also been implicated in bone mass defects, and some studies showed that COLIA1 polymorphism was associated with reduced BMD and predisposed men and women to osteoporotic fractures (Grant et al, 1996; Langdahl et al, 1998; Uitterlinden et al, 1998).

Being aware of the constant interaction between genetic and environmental factors in bone mass determination, the aim of this study was to investigate the allelic distribution for 713–8delC and COLIA1 polymorphisms in a β-thalassaemia transfusion-dependent Italian population and its relationship with bone mass.


Subjects One hundred and thirty-five patients with β-thalassaemia major were studied: 78 women (age 18–34 years) and 57 men (age 19–33 years). To reduce the impact of bone growth, only subjects older than 18 years were included in the study. A sexual assessment was made on all patients. Spontaneous puberty occurred in 66 (49%) subjects (43 out of 78 women and 23 out of 57 men). Hypogonadotrophic hypogonadism was present in 95 (70%) patients. Eleven (8%) patients had insulin-dependent diabetes.

The examined population had been treated with monthly blood transfusions with a pretransfusion haemoglobin ranging from 8·3 and 9·9 g/dl. The mean blood transfused during the last 3 years was noted. Chelation therapy was obtained by subcutaneous administration of desferrioxamine (DFX) 30–50 mg/kg with a compliance of 5–6 d a week. The age at starting chelation therapy, the number of years of chelation and the mean DFX dosage of the 3 years before the BMD assay were recorded. The mean serum ferritin level of the last three measurements was assessed.

Detailed clinical histories on smoking, exercise and calcium intake were taken for each patient. Patients who had taken drugs such as vitamin D supplementation and bisphosphonates, and patients with hypoparathyroidism, were excluded. In order to avoid adulterated results on BMD analysis, patients with scoliosis, vertebral deformity and fractures detected with standard radiographic examination were excluded from the study. Blood was available for DNA isolation from all patients. Informed consent was obtained and the Local Ethic Committee approved the study protocol.

Bone mass measurement BMD expressed as g/cm2 was assessed at the neck of the femur and at the lumbar spine (L1-L4) by dual energy X-ray absorptiometry (DXA) using a QDR-2000 (Hologic, Waltham, MA, USA). The coefficient of variance (CV) was 1·2% for the femoral neck (FN-BMD) and 2·2% for vertebrae (TH-BMD). The patients were divided into normal (Z-score between 0 and −1), osteopenic (Z-score between −1 and −2·5) and osteoporotic (Z-score below −2·5) groups on the basis of the Z-score values according to the WHO classification (World Health Organization, 1994). The Z-score (the value of the SD obtained when the average of the data was adjusted to 0) was calculated by using the data of BMD obtained from up to 200 sex-, ethnic- (Lombardia and Campania regions) and body mass index (BMI)-matched subjects.

Genotyping DNA was isolated from peripheral blood leucocytes using conventional methods.

For the assessment of COLIA1 gene, we studied a G→T polymorphism at the first base of a binding site for the transcription factor Sp1 in the first intron of the gene. This polymorphism was detected by polymerase chain reaction (PCR) with a mismatched primer that introduces a diallelic restriction site, i.e. MluNI, as described elsewhere (Fig 1) (Grant et al, 1996). The test discriminates two alleles, S and s, that correspond to the presence of guanine and thymidine respectively.

Figure 1.

COLIA1 and TGF-β1 gene polymorphism. A. 1, homozygous subject (ss) shows only one fragment of 148 bp. 2, heterozygous subject (Ss) shows three fragments of 148, 128 and 20 (not shown) bp. B. 1, normal subject (TT) shows three fragments of 108, 81 and 32 (not shown) bp. 2. heterozygous subject (Tt) shows four fragments of 140, 108, 81 and 32 (not shown) bp. MW, molecular weight.

For the TGF-β1 polymorphism, PCR primers were designed to amplify a 221-bp product including the 713–8delC sequence variation. The PCR product was cut with two different restriction enzymes: AluI and BsiYI. Two fragments of 140 bp and 81 bp were obtained after the AluI digestion in both normal and mutated subjects. In normal individuals, only the fragment of 140 bp was further cut into two smaller fragments of 108 bp and 32 bp after the BsiYI digestion. Therefore, homozygotes for the presence of the polymorphism showed two fragments (140 bp and 81 bp), whereas homozygotes for the normal sequence showed three bands (108 bp, 81 bp and 32 bp). An intermediate pattern with four bands was detectable in heterozygous subjects (Fig 1). The polymorphism was coded as T-t. The uppercase letter stands for the presence and the lowercase letter for the absence of the restriction site (Langdahl et al, 1997).

One hundred and twelve out of 200 BMD controls were screened to establish the frequency of two polymorphisms in our population.

Statistical analysis All data are expressed as means ± SEM. Differences in all continuous clinical variables and BMD among the three COLIA1 and TGF-β1 genotypes were tested using one-way analysis of variance and Scheffe's multiple range test. The frequency distribution of COLIA1 and TGF-β1 genotypes in osteoporotic, osteopenic and normal subjects were compared using cross-tabulation and standard χ2 and Mann–Whitney U-tests were used for non-parametric data. The comparisons of spinal BMD among the different COLIA1 and TGF-β1 genotypes were carried out after adjusting mean BMD Z-score values (adj. Z-score) for potential confounding factors, such as height, weight and haematological factors (blood transfused, serum ferritin, pretransfusion haemoglobin, DFX dosage, age at starting chelation therapy and number of years of chelation), by using analysis of covariance and multiple regression analysis. Finally, sex, spontaneous puberty, hypogonadotrophic hypogonadism, diabetes and gene polymorphisms were entered into a general linear model to test their association with the bone mass. A P-value of < 0·05 was considered statistically significant. All statistical analyses were performed using statgraphics version 5 (Manugistic, Rockville, MA, USA).


The antrophometric, haematological and densitometric data are shown in Table I. The BMD of the examined subjects was expressed as the Z-score according to the WHO classification (World Health Organization, 1994). A large majority of the patients showed bone mass below one standard deviation from the mean value of the healthy subjects (Table I). Together, 121 out of 135 (90%) subjects with thalassaemia major showed osteopenia or osteoporosis of the spine and 105 out of 135 (78%) had low bone mass at the femoral neck (data not shown). Bone mass was lower in men than in women (P = 0·0023) (Table I) with a more prevalent osteopenia/osteoporosis of the spine in men than in women (χ2 = 13·4; P = 0·001) (data not shown).

Table I.  Antrophometric, haematological and densitometric data in thalassaemia major patients. The data are expressed as mean (± SEM).
 Pooled (n = 135)Men (n = 57)Women (n = 78)
  • *

    P =  0·0023 men vs. women (Mann–Whitney U-test).

Age (years)24·0 ± 0·425·4 ± 0·322·5 ± 0·4
Weight (kg)55·8 ± 0·6561·4 ± 0·551·8 ± 0·4
Height (cm)159·3 ± 0·78166 ± 0·52154·5 ± 0·54
Blood (g/kg/year)130·5 ± 0·5129·9 ± 0·5131·0 ± 0·6
Ferritin (μg/l)1892·1 ± 1·51692·0 ± 1·32038·4 ± 1·6
Haemoglobin (g/dl)9·07 ± 0·039·06 ± 0·039·08 ± 0·03
DFX (g/kg/year)20·56 ± 0·1720·97 ± 0·1920·26 ± 0·16
Age started DFX (years)5·81 ± 0·226·12 ± 0·295·88 ± 0·21
Years on DFX15·9 ± 0·2816·22 ± 0·3215·76 ± 0·25
Lumbar BMD (Z-score) −2·19 ± 0·09 −2·59 ± 0·16* −1·93 ± 0·10

Table II reports the distribution of TGF-β1 and COLIA1 genotypes in the sample under study and in the normal reference population. Proportions among different genotypes were as expected according to the Hardy–Weinberg equilibrium. In the thalassaemic group, there were no differences in the genotype frequencies between men and women for the two polymorphisms studied (χ2 = 1·34, P = 0·246 and χ2 = 2·75, P = 0·25 respectively). Comparison between thalassaemia major and normal subjects showed no differences in the TGF-β1 and Sp1 genotype frequencies (χ2 = 1·45, P = 0·483 and χ2 = 2·21, P = 0·330 respectively).

Table II.  Genotype frequencies for COLIA1 and TGF-β1 polymorphisms in thalassaemia major patients (% in brackets).
(n = 135)
(n = 57)
(n = 78)
(n = 112)
 SS74 (55)35 (61)39 (50)74 (66)
 Ss55 (41)20 (35)35 (45)34 (30)
 ss6 (4)2 (4)4 (5)4 (4)
 TT112 (83)50 (88)62 (79)92 (82)
 Tt23 (17)7 (12)16 (21)19 (17)
 tt0001 (1)

COLIA1 Sp1 polymorphism and BMD

The mean Z-score values of lumbar spine BMD and clinical data divided by COLIA1 Sp1 polymorphism are reported in Table III.

Table III.  Antrophometric, haematological and densitometric data according to genotypes of COLIA1 polymorphism. The data are expressed as mean (± SEM)
  • *

    P   =  0·0006 men vs. women (Mann–Whitney U-test).

Weight (kg)55·5 ± 0·655·9 ± 0·359·5 ± 0·80·352
Height (cm)158·8 ± 0·8159·5 ± 0·7164·8 ± 0·90·186
Blood (g/kg/year)129·7 ± 0·5131 ± 0·6128·8 ± 0·50·078
Ferritin (μg/l)1759 ± 1·42023 ± 1·52323 ± 0·90·478
Haemoglobin (g/dl)9·07 ± 0·039·05 ± 0·039·1 ± 0·040·734
DFX (g/kg/year)20·6 ± 0·220·2 ± 0·221·1 ± 0·20·071
Age started DFX (years)5·91 ± 0·245·33 ± 0·255·7 ± 0·180·751
Years on DFX15·7 ± 0·315·5 ± 0·316·3 ± 0·30·489
Lumbar BMD (Z-score) −2·06 ± 0·13 −2·21 ± 0·15 −3·23 ± 0·460·014
Adj lumbar BMD (Z-score)    
 Pooled2·0 ± 0·11 −2·19 ± 0·13 −3·12 ± 0·260·012
 Men2·20 ± 0·18 −3·0 ± 0·24* −4·70 ± 0·300·022
 Women −1·80 ± 0·13 −1·68 ± 0·14 −2·28 ±0·630·004

The ‘Ss’ and ‘ss’ genotypes had a lower bone mass than subjects with ‘SS’ genotype (P = 0·014); the homozygous patients for ‘s’ allele had the lowest bone mass. There were no differences among genotypes in all potential confounding clinical factors. When polymorphism, spontaneous puberty, hypogonadotrophic hypogonadism and diabetes were tested together in a general linear model to verify their association with the level of bone mass, the COLIA1 polymorphism continued to maintain a significant association (P = 0·04, P = 0·610, P = 0·203 and P = 0·389 respectively). The differences among genotypes were still present when the BMD was analysed as the adjusted Z-score and when the two sexes were analysed separately (men P = 0·022 and women P = 0·004). Moreover, in male thalassaemic patients, the presence of the ‘s’ allele was associated with more severe osteoporosis of the spine than in female patients (P = 0·0006 in ‘Ss’ patients) (Table III). While spine BMD values were still lower in ‘ss’ subjects, the difference between men and women was not significant because of the small number of homozygotes. When the study population was stratified on the basis of Z-score values at the lumbar spine BMD in normal (Z-score > −1) and osteopenic/osteoporotic (Z-score < −1) groups, a different allelic distribution was found (P = 0·047) (Table IV). The ‘ss’ genotype (5%) was present only in the osteopenic/osteoporotic group.

Table IV.  Genotypes frequencies of COLIA1 polymorphism in normal (vertebral BMD > −1) and osteopenic/osteoporotic patients (vertebral BMD < −1) (% in brackets).
  1. * χ 2 6·098; P = 0·047.

Normal12 (86)2 (14)0
Osteopenic/osteoporotic62 (51)53 (44)6 (5)

TGF-β1 polymorphism and BMD

There were no differences among genotypes in all examined clinical data (data not shown). When the mean Z-score and the adjusted Z-score of lumbar spine BMD of the subjects with ‘TT’ genotype was compared with those with the ‘Tt’ genotype there was a lower bone mass in ‘Tt’ genotype, but the difference was not significant (Table V and data not shown). When men and women were analysed separately, no significant differences were found in Z-scores between the two genotypes. Moreover, in male thalassaemic patients, the presence of more severe osteoporosis of the spine than female patients was not associated with the TGF-β1 polymorphism, although ‘Tt’ men showed a trend to a lower bone mass than ‘Tt’ women (Table V). When the study population was divided by Z-score values at the lumbar spine BMD in normal and osteopenic/osteoporotic subjects, we found no differences in genotype frequencies in thalassaemic subjects with Z-score < −1 compared with thalassaemic subjects with Z-score > −1 (Table VI).

Table V.  Vertebral BMD (expressed as Z-score) according to genotypes of TGF-β1 polymorphism. The data are expressed as mean (± SEM).
  • *

    P   =  0·782 men vs. women (Mann–Whitney U-test).

Pooled −2·15 ± 0·10 −2·24 ± 0·230·74
Men −2·61 ± 0·17 −2·41 ± 0·46*0·68
Women −1·80 ± 0·11 −2·16 ± 0·230·10
Table VI.  Genotypes frequencies of TGF-β1 polymorphism in normal (vertebral BMD > −1) and osteopenic/osteoporotic patients (vertebral BMD < −1) (% in brackets).
Normal12 (86)2 (14)
Osteopenic/osteoporotic100 (83)21 (17)


A variety of skeletal disorders including rickets, scoliosis, spinal deformities, nerve compression, fractures and severe osteoporosis have been reported in thalassaemia major patients (Giardina et al, 1995; Vichinsky, 1998). In particular, osteoporosis and bone fractures seem to be present in a variable percentage of the patients from 33% to 20% (Giardina et al, 1995). Recently, Jensen et al (1998) showed a prevalence of low and severely low bone mass in 45% and 51% of thalassaemia major patients respectively. Our findings in a group of adult transfusion-dependent patients show that the bone loss is a common feature in well-treated thalassaemic patients. The prevalence of low bone mass is very high with 90% of the thalassaemic patients having a lower Z-score than controls. According to Jensen et al (1998), the prevalence of osteopenia and osteoporosis is higher in men than in women and is more prevalent in the spine than the femoral neck. These differences between two skeletal sites are as a result of different bone architecture: the lumbar spine has prevalent trabecular bone and the femoral neck has a mixture of trabecular and cortical bone.

Evidence from family and twin studies has clearly shown that genetic factors play an important role in the pathogenesis of osteoporosis. An estimated 46–80% of the total variance in adult bone mass is attributed to genetic determinants (Pocock et al, 1987; Krall & Dawson-Hughes, 1993). Recently, great interest has been generated regarding the relationship between some polymorphisms, e.g. vitamin D receptor and oestrogen receptor genes, and bone density (Morrison et al, 1994; Kobayashi et al, 1996). However, the consistency of this effect has not yet been established and controversy surrounding the relationship between these polymorphisms and BMD demands further investigation (Peacock, 1995; Han et al, 1997).

The association between 713-8delC of the TGF-β1 gene and osteoporosis has been demonstrated in different populations (Langdahl et al, 1997; Yamada et al, 1998; Bertoldo et al, 2000). Moreover, other studies have demonstrated that a novel polymorphism in the collagen type Iα1 (COLIA1) gene is over-represented in osteoporotic patients, and is associated with reduced bone density (Grant et al, 1996) and high risk of fractures (Langdahl et al, 1998; Uitterlinden et al, 1998). Recently, it has been shown (Wonke, 1998, Wonke et al, 1998) that male thalassaemia patients who are heterozygous or homozygous at the polymorphic Sp1 site have a lower BMD than females and had no improvement in spinal osteoporosis in response to treatment with bisphosphonates. In both male and female patients, we found a consistent association between Sp1 polymorphism and differences in bone density at the spine, with evidence of gene-dose effects. Furthermore, BMD values were higher in women than men in each genotype. Therefore, we suggest that this polymorphism may represent a genetic ‘background’ influencing the development of secondary osteoporosis in thalassaemia major, as demonstrated in other secondary osteoporosis conditions such as in diabetes mellitus (Hampson et al, 1998), and potentially increasing per se the high risk of vertebral fractures documented in thalassaemia major.

The association between the gene polymorphisms and bone mass in thalassaemia major needs to be considered, bearing in mind that the disease is a well-documented type of secondary osteoporosis with multiple negative influences, also exogenous, on bone metabolism that could appear to be as a result of the effect of the polymorphisms. In fact, many factors in thalassaemia major can lead to an unbalanced bone remodelling and increased bone resorption. These factors include hormonal deficiency, bone marrow expansion, iron overload and desferrioxamine toxicity (Vichinsky, 1998). In our study, the impact of antrophometric and haematological factors on lumbar bone mass values was negligible when tested in a multiple regression analysis model. Only blood and DFX cumulative amounts reached a relationship with bone mass near to the statistical significance (P = 0·078 and P = 0·071 respectively).

The molecular mechanism underlying the association between both polymorphisms and the BMD remain to be defined. A possible explanation of the different association pattern between the two polymorphisms and BMD could be that the TGF-β1 polymorphism appears to influence bone mass by mainly increasing bone turnover (Langdahl et al, 1997; Bertoldo et al, 2000). Bone turnover is also dramatically affected by thalassaemia and its treatment per se, and this may represent the cause of the lack of a strong association with bone mass in thalassaemic patients. On the contrary, COLIA1 polymorphism can affect bone mass by influencing other aspects of bone metabolism, i.e. collagen structure. It therefore appears as an independent risk factor for genetic susceptibility to osteoporosis.

This study provides strong evidence of an association between the Sp1 polymorphism of the COLIA1 gene and vertebral osteoporosis in a sample of Italian β-thalassaemia major patients, with men more severely affected. Our results raise the possibility that genotyping at the Sp1 site could be of clinical value in identifying the thalassaemic patients at risk of osteoporosis and fractures.


This work was partially supported by the Associazione Giambrone per la Lotta alla Thalassemia, Italy, and by MURST 1998 and by University of Bari, Italy.