| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Endocrine Care |
INSERM U 349, Bone Pathology (C.B., M.E.C.-S., J.B., M.C.D.V.) and Fédération de Rhumatologie (M.E.C.-S., J.B., M.C.D.V.), Hôpital Lariboisière, 75475 Paris, France
Address all correspondence and requests for reprints to: Marie-Christine de Vernejoul, M.D., INSERM U349, Hôpital Lariboisière, 2 rue Ambroise Paré, 75425 Paris Cedex 10, France. E-Mail: . christine.devernejoul{at}inserm.lrb.ap-hop-paris.fr
Abstract
Our aim was to assess the relative impacts of genetics and environment in the families of osteoporotic patients and identify the best subgroup of patients to investigate the genes associated with osteoporosis. We recruited 36 men and 47 women with osteoporosis (probands), median age of 52 and 68 yr, and all their siblings (90) and offspring (83). The families were classified as young or old on the basis of the median age of the probands. We measured the bone mineral density at the femoral neck (FN) and lumbar spine (LS) adjusted for age and weight and standardized (Z-score). Physical activity, nutritional calcium, and alcohol and tobacco consumption were investigated. We compared the mean Z-score using linear mixed model and assessed the familial resemblance using intraclass correlation. The mean Z-scores of the families of osteoporotic patients were significantly negative at FN and LS, with no intergeneration or intergender differences. At FN, but not at LS, the mean Z-score was independently lower in the families of male probands (mean ± SD: -0.57 ± 0.96, female: -0.18 ± 0.85, P = 0.012) and in young families (-0.58 ± 0.94, old families: -0.11 ± 0.83, P = 0.006). This suggested that the lower Z-score in the families of men with osteoporosis was related to their younger age. There was significant phenotypic resemblance among members in the families. In the families of female probands, the correlation between the probands and her siblings was weak and disappeared after adjustment on environment, and a resemblance appeared within their children (FN: r = 0.61) suggesting that different environment had masked the resemblance in this subgroup. In the families of male probands, a strong resemblance persisted after adjusting for environment, (proband-offspring at FN: r = 0.46 and within offspring at FN: r = 0.66, at LS: r = 0.61). This showed that resemblance was independent of a common measurable environment in these families of men with osteoporosis. In conclusion, mainly young osteoporotic patients, most of whom were male in our study, are affected by the genetic component.
BONE DENSITY EXHIBITS familial resemblance in normal subjects. Twin studies have shown that there is much closer concordance of bone mineral density (BMD) in monozygotic twins than dizygotic twins (1, 2). Some resemblance of bone density has also been shown in normal families (3, 4). The familial resemblance of bone density is acquired quite early; the bone density of prepubertal offspring is correlated to that of their mother (5). It is therefore generally accepted that peak bone mass is strongly influenced by heredity. Subsequently, it is not clear whether bone loss is also dependent on genetic factors because the insidious effects of lifestyle may be cumulated throughout life and become dominant in later life (6).
In other complex traits, such as breast cancer or Parkinson’s disease, there exists mendelian subgroups. Indeed, even the phenotype of mendelian disorder has complex traits (7). Our aim was therefore to investigate, among osteoporotic families, the possible existence of subgroups with high heredity. Familial resemblance has been demonstrated in the daughters of osteoporotic women (8, 9): there was a weak relationship between bone density of daughters and their osteoporotic mothers, and their bone density decrease was half that of their mothers (8). The pathogenesis of male osteoporosis is obscure, and it is not certain that any analogy can be drawn with female osteoporosis (10, 11, 12). We previously studied the families of male patients with idiopathic osteoporosis and showed that bone density was usually low in all the members of their family (13). In that study we found that the mean decrease in the Z-score of these relatives was greater than that reported by Seeman et al. (8) for the daughters of osteoporotic women.
We therefore decided to study the variance components of BMD in the nuclear families of osteoporotic men and women. Our aim was to estimate the respective impacts of genetics and environment on the femoral neck and lumbar spine in osteoporotic subjects to identify the best subgroup of patients for investigating the genes involved in osteoporosis. We analyzed BMD rather than fractures for several reasons: although there is a higher risk of fracture and low BMD in the offspring of parents with fracture (14, 15), fractures obviously depend in part on an environmental factor, the fall. In addition, twins have been shown to exhibit high heritability for bone density (1, 2, 16) but not for fractures (17), suggesting that fractures may not be an appropriate marker to use when investigating the genetics of osteoporosis.
Subjects and Methods
We recruited 47 female and 36 male subjects with osteoporosis among the patients referred to our clinic for osteoporosis management. They were defined as probands, and were selected on the basis of the willingness of all their offspring and siblings to be investigated. We did not select the families in which the siblings or offspring did not all agree to participate; however, we did not include subjects younger than 18 yr. The Ethics Committee of the hospital approved the study, and all subjects gave written informed consent.
All probands were osteoporotic as defined by a T-score at one site less than -2.5, except five women who had a T-score less than -2 and vertebral crushed (from two to five vertebrae). The 47 proband women aged (mean ± SD) 67 ± 8 yr (range 43–79 yr) were all postmenopausal. Thirty-two women had at least one vertebral crush fracture, 12 had a peripheral fracture (three Colles fractures, three rib fractures, two femoral neck fracture, and four fractures at other sites), and 3 had no fracture. The 36 proband men were 49 ± 12 yr of age (range 26–74 yr old). Twenty-four had at least one crushed vertebra, six had a peripheral fracture (four rib fractures, one humerus, one Colles fractures), and five patients had no fractures. The proband without fracture (five men and three women) had a Z-score lower than -2 at femoral neck (FN) or lumbar spine (LS). The values of BMD and Z-score was comparable between probands with and without vertebral crushed and between their respective family members (data not shown).
All probands underwent extensive clinical, biochemical, and radiological assessment to exclude secondary causes of osteoporosis, including malignant disease, multiple myeloma, malabsorption, osteomalacia, hypogonadism, hemochromatosis, hyperthyroidism, and endogenous or exogenous hypercortisolism. None reported any history of urolithiasis. We excluded patients with alcoholic cirrhosis. Patients had no plasma hormone abnormalities (free T, TSH, or PTH). We did not exclude osteoporotic patients who were heavy drinkers or had any other putative environmental risk factor such as a high tobacco consumption.
We measured the bone mineral content (grams) of all the subjects at FN and LS (L2-4), using dual-energy x-ray absorptiometry (DPX-L densitometer, Lunar Corp., Madison, WI). The areal BMD (grams per square centimeter) was calculated by dividing the bone mineral content by the projected area of the region scanned. These values were adjusted on age and gender and standardized in a Z-score: for each bone site, we subtracted from the individual value the mean value of a normal French population of the same gender and age referenced by the manufacturer. These differences were divided by the SD of the reference population. Reference BMD values were primarily corrected for the patient’s weight. Z-score was also adjusted on hormonal replacement therapy for the women who had received it for more than 10 yr.
During the visit, a physician, using a structured questionnaire about lifestyle factors that can influence bone density, questioned each participant. The usual dietary intake of calcium was assessed by a questionnaire about the consumption of usual foods: milk, bread, cheese, and biscuits. It had been shown previously that estimates of calcium intake obtained using this questionnaire correlate well with those obtained using food frequency methods (18). An alcohol-related questionnaire investigated the number of glasses of wine, beer, and spirits consumed per day. We converted the replies into grams per day, using the usual conversion factors (19). We assessed the usual physical activity expended in leisure, occupational, and domestic activities. The energy expenditure of each activity was assessed in terms of metabolic equivalent, depending on gender and weight, using the Durnin and Passmore scale (20). This score was multiplied by the time usually spent in each regularly performed activity during the past year. Menopausal status, including whether estrogen replacement therapy was taken and, if so, for how long, was investigated. Personal and family fracture histories were collected.
Statistical methods
The familial data, which are repeated measurements on the same unit, the family, define two independent random effects, the family effect and the member effect within families, with their respective variances (the inter- and intrafamily variance) and correlations among family members. These correlations define the familial resemblance, which measures how much greater the variance is among families than within families. We estimated both variance components by maximum likelihood (21). The resemblance among offspring or among siblings was based on the intraclass correlation, and its variance was estimated by Smith’s formula (22). The resemblance between offspring and siblings was based on the interclass correlations (pairwise individual estimator) (23), and that between probands and offspring or siblings used a simplified form of the interclass correlation (pairwise estimator) (24).
The probands’ Z-scores were compared between gender and age classes using a two-way ANOVA. Relatives’ Z-scores were analyzed as a function of four factors: two of them had a fixed effect, which differentiated the families (the proband’s gender and the youth of the family), and the other two characterized the individuals within families with random effects (the relatives’ gender and generation: offspring or siblings). The analysis was based on an adjustment of the Z-score to a model, which includes these four factors and their six combined effects two by two (first-order interactions terms). FN and LS were analyzed successively. Linear mixed-effect regression (25) was used to carry out the adjustment. This differs from multiple linear regression by taking into account the fixed and random effects and the resemblance between members of the same family. The assumptions underlying this model, which have been discussed in detail elsewhere (26), were checked, including diagnostic tests for detecting nonnormality and outliers. The model was fitted according to the maximum likelihood criteria. The best-fitting and most parsimonious subsets of factors were selected using the lack-of-fit technique, which compares the maximum likelihood observed in two nested models. The comparison was assessed by the likelihood ratio test.
BMD is usually adjusted on environment using a multiple linear regression, but this fitting using straight lines or their transformations may be questionable, especially when the data include both genders and a wide range of ages. For instance, a possible complex interaction among bone status, dietary calcium, physical activity, and genetics has been suggested (27), and alcohol consumption data show that both beneficial and detrimental effects may be dose dependent (19). To take account of more general interactions between predictor variables and nonadditive behaviors, we used a regression tree method that matches subject in subgroups with similar environmental effect on Z-score at FN and LS (28). We analyzed the influence of environmental risk factors on BMD using a two-stage approach. First, the fitted value was defined by these subgroups.
Second, residual values were used to fit models of inheritance. Data were adjusted on gender. Statistical analyses were performed using the S-PLUS 4.5 statistics package for Windows (29). Linear mixed-effect model adjustment was performed using the NLME software package for S-PLUS (30). The threshold of significance was set at 0.05. The probability values used to compare the means or the correlations to zero value were one-tailed. Results were expressed as mean ± SD.
Results
We analyzed 173 relatives of 83 probands (47 women and 36 men). The relatives of proband women consisted of 41 siblings of probands (30 sisters and 11 brothers) and 49 offspring (30 daughters and 19 sons). For proband men, there were 49 siblings (29 sisters and 20 brothers) and 34 offspring (17 daughters and 17 sons). The sex ratio among relatives (women/men) was 1.6 without statistical difference between gender distribution in relatives of proband women and men (odds ratio = 1.61, P = 0.13) or inside the generation (siblings or offspring odds ratio = 1.46, P = 0.23). At the FN, the mean Z-scores of probands was (mean ± SD) -0.90 ± 0.88, that of their siblings was -0.37 ± 0.94, and that of their offspring was -0.35 ± 0.90. All were significantly different from 0. At the LS, the mean Z-scores were, respectively, -1.78 ± 1.22, -0.49 ± 1.18, and -0.67 ± 1.02 and were also significantly different from 0. The demographic characteristics and bone status of the probands and their relatives are shown in Table 1
, according to the gender of the subjects and of the probands.
|
Z-score of the probands.
No significant difference between the Z-scores of proband men and women (Table 1
) was found at the FN (men: -1.07, women: -0.77, P = 0.38) or LS (men: -1.91, women: -1.67, P = 0.64). Table 1 shows that the proband men, and subsequently their relatives, were much younger than the proband women. The difference between the mean ages of the proband men and women was 18 yr, and those for their respective offspring and siblings were 16 and 12 yr, respectively. Furthermore, proband gender was confounded with the youth of the families. To take the youth of the family into account, all the subjects were classified on the basis of their proband’s age (Table 2). Because of the age difference between men and women probands, we chose different young/old thresholds for men and women: 55 and 65 yr old, respectively, which was close to their respective median ages (52 and 68 yr old).
|
shows that this difference was observed in both men (young -1. 28, old -0.66) and women (young -0.91, old -0.68) and did not depend on gender (interaction age x gender P = 0.34). A similar trend was observed at LS (young: -2.05 ± 1.03, old: -1.47 ± 1.34, P = 0.063), with no significant difference between women (young -2.03, old -1.39) and men (young -2.06, old -1.63, interaction age x gender P = 0.71). Z-scores of the relatives
Femoral neck. The best and most parsimonious adjustment of Z-score at FN involved only two significant factors: the proband’s gender and the family age. The results for the different terms included in the model were as follows:
Effect of relative’s gender.
Table 1 shows the similarity between the mean Z-scores of the male and female relatives of the probands. There was no interaction between the gender of the relatives and the gender of the probands (P = 0.64) or the generation (offspring or siblings P = 0.4). The overall mean Z-scores of male and female relatives were -0.39 ± 0.91 and -0.34 ± 0.92, respectively (P = 0.40).
Effect of generation (offspring, siblings). There was also no substantial difference between offspring and siblings related to the gender of the probands (interaction test: P = 0.43). The overall mean Z-score of the offspring was similar to that of the siblings (offspring: -0.35 ± 0.90, siblings: -0.37 ± 0.94, P = 0.17).
Effect of proband’s gender.
Table 1 shows that in each subgroup the mean Z-score in the families of male probands was lower than that in families of female probands. The overall mean Z-scores were -0.57 ± 0.96 -0.18 ± 0.85 respectively (P = 0.012).
Effect of the age of the family.
Table 2 shows that the mean Z-scores of the offspring and siblings of osteoporotic patients were lower in young families (-0.96, -0.88, -0.40, -0.17) than in old families (-0.16, -0.11, -0.21, +0.15), regardless of generation or proband gender. The overall mean Z-scores of the relatives in young and old families were -0.58 ± 0.94 and -0.11 ± 0.83, respectively (P = 0.006). The effect of family age was more obvious in the relatives of male osteoporotic patients (young: -1.03 ± 0.92, old: -0.28 ± 0.90) than in those of female osteoporotic patients (young: -0.45 ± 0.96, old: -0.31 ± 0.80), although this difference was not statistically significant (interaction test: P = 0.10).
Lumbar spine
The occurrence of arthrosis and osteoporotic fractures at the LS in 28/87 probands could have influenced the data of the Z-score at that site, but none of the children or siblings had LS fractures. At LS, none of the factors included in the model was significant. Familial Z-score was not substantially different in young and old families (young: -0.70 ± 1.15, old: -0.42 ± 1.03, P = 0.10) or in the male or female probands (men: -0.63 ± 1.12, women: -0.53 ± 1.09, P = 0.54). There was no effect of the relatives’ gender (sons and brothers: -0.71 ± 1.15, daughters and sisters: -0.49 ± 1.07, P = 0.20) or generation (offspring: -0.67 ± 1.02, siblings: -0.49 ± 1.18, P = 0.29).
Family resemblance
Initially, we included family correlations in this statistical model without adjusting for environmental factors (Table 3). The FN Z-scores were closely correlated among the relatives of proband men. A particularly close resemblance was observed among offspring (r = 0.81) and among offspring and siblings (r = 0.62). The probands were correlated both to their offspring (r = 0.47) and their siblings (r = 0.41). Paradoxically, there was no resemblance among siblings (r = 0.28). At LS, there was no resemblance among relatives, except between the offspring of proband men, who were closely correlated among themselves (r = 0.78). The families of proband women showed no resemblance at any bone site other than the correlations between probands and their siblings at both LS (r = 0.26) and FN (r = 0.40) and at LS only between the probands and their offspring (r = 0.23). The correlations were of borderline significance at LS.
|
Table 4 shows the main lifestyle characteristics known to be bone risk factors. Half of all the subjects (55.4%) drank alcohol and half (51.1%) smoked at least one cigarette per day, 31.8% taking both, 24.8% taking neither. Also, 9.3% of the subjects, mainly men, were heavy alcohol drinkers (consumption 
30 g/d). Overall, the men tended to drink more alcohol than the women, the consumption ratio being 4.3 for the probands, 3.3 for the siblings and 1.5 for the offspring. The median tobacco consumption was 15 cigarettes/d, with a maximum of 60 cigarettes/d. Men, both relatives and probands, regardless of age, smoked twice as much as women. The overall mean calcium intake in the diet was 0.91 g/d ± 0.4, with a range of 0.22–2.91 g/d.
|
Adjustment of the family resemblance for environment
Table 3 shows the changes in family resemblance after adjusting for environment. This analysis showed that the resemblance among the offspring of male probands was still quite high at both FN (r = 0.66) and LS (r = 0.61), suggesting that there was a high genetic component in the bone density of the offspring of osteoporotic males. Quite a high correlation was also found between the osteoporotic male and his offspring (r = 0.46) or siblings (r = 0.51) but only at FN. The resemblance between the osteoporotic mother and her offspring persisted at the same level at LS (r = 0.30). Interestingly, a resemblance between the offspring of the osteoporotic mother at FN (r = 0.61) emerged once the data had been adjusted for environment, suggesting that a differing familial environment masked the genetic component. The correlation between the osteoporotic female and her siblings was no longer significant at LS and FN when adjusted for environment.
Discussion
We analyzed the families of osteoporotic women and men and found that both had in average a low bone density. We confirmed previous studies (3, 8, 9, 15) that had shown that the offspring of both men and women with a vertebral or hip fracture have a low BMD. We extended this observation to the sisters and brothers of the osteoporotic subjects. Our goal in this study was to compare the Z-scores of the families of osteoporotic patients of both genders and find out in which subgroup and at which site the genetic component was significant.
When we analyzed the Z-score of the relatives within families, there was no gender-related or generation effect at LS or FN: the sons, daughters, brothers, and sisters all exhibited a similar reduction in the Z-score. Also, the resemblance was not dependent on the gender of the relative. In contrast, when we analyzed the families on the basis of the proband gender, the relatives of osteoporotic patients of both genders had identical Z-scores for LS. By contrast, the relatives of osteoporotic men had lower Z-scores at the FN, although there was no significant difference between Z-scores of osteoporotic men and women at that site. Moreover, we found higher phenotype resemblance among the families of osteoporotic men than those of the women. The offspring-proband correlation we found in the families of osteoporotic women was low and similar to that reported between daughters and their osteoporotic mothers (8, 31) or among offspring and their normal parents (3, 4). In contrast, the correlation coefficient observed among the offspring of osteoporotic men was extremely high, close to the values reported in twin studies (1, 2, 16). One explanation for this could be those osteoporotic men and women do not have the same disease, and therefore their bone density family resemblance patterns are not the same. It is also possible that the postmenopausal bone loss that the osteoporotic women experienced is under the control of specific genes that do not affect the Z-score of their premenopausal daughters and of the male members of their family, therefore decreasing the resemblance within the families of postmenopausal female probands.
However, another hypothesis is that the difference between the relatives of male and female probands could be attributable to the younger age of the proband men. The osteoporotic men and indeed their relatives were younger, and the lower FN Z-score of their relatives could also be associated to the age rather than to the gender of the proband. This observation was to be expected because the FN Z-score of the younger proband themselves, regardless of gender, was lower and was related to that of the relatives. However, among the families of proband men and women of the same age (the oldest male proband and youngest female proband were both around 60 yr in age), the Z-scores of their offspring were similar. This strongly suggests that it is the young age rather than the male genders of the probands that is associated with the lower BMD in the families. The closer resemblance of the offspring to their younger osteoporotic father is similar to what has been reported in breast cancer, in which the risk was found to be higher for the daughter if the mother had developed cancer and she was under age 40 yr (32). Indeed, it can be suggested that in both diseases aging itself, as a result of accumulation of noninherited factors (environment), increases the risk of the occurrence of the disease and masks the genetic component. Therefore, those heritable factors are usually most obvious if the disease occurs in young subjects. Also, the lower Z-score observed in younger probands, independent of their gender, equates to a more severe disease when there is a younger age at onset. The lower mean Z-score and the higher genetic resemblance in younger osteoporotic subjects support the received hypothesis that the heritable component affects peak bone mass rather than bone loss (7, 33, 34).
Familial resemblance might be owing to a common environment inside families. However, we could detect very few correlations of these environmental factors within the families of osteoporotic patients. When we adjusted for environment, the resemblance that persisted is generally classified as genetic. However, although we estimated all the well-known environmental factors, we are aware that common unmeasured environmental factors could have accounted for this genetic component (35). Most of the phenotype resemblance at FN persisted in the families of osteoporotic males after adjustment on environment. In the families of osteoporotic females, a resemblance appeared for FN among the offspring, suggesting that different environmental factors had masked the genetic resemblance. By contrast, the phenotypic resemblance among older subjects (sibling/proband) disappeared, suggesting that common environmental factors were responsible for the resemblance, including estrogen deficiency in women and in both genders, physical activity, alcohol, tobacco, and calcium intake.
As previously reported, in populations with common osteoporosis, vertebral (36, 37) and femoral (38) osteoporotic fractures occurred at an earlier age in males than in females. Our population of male patients with idiopathic osteoporosis matches the conventional picture of this condition: the clinical presentation is usually symptomatic with fractures. All the known etiological factors of male osteoporosis had been excluded by clinical examination and biochemical measurements (11, 39). We included some patients with a high alcohol and tobacco consumption, but we have previously shown that in this subpopulation of patients, the bone density of the other members of the family was as low as in the families of patients without risk factors (13). The physiopathology of this idiopathic osteoporosis in males is still obscure, but our data strongly suggest that low peak bone mass is the main determinant of this highly heritable condition. All the osteoporotic women included in the study were postmenopausal and were quite different from the recently reported patients with "idiopathic osteoporosis" who were under 35 yr of age (40). We could not collect data from the families of young premenopausal osteoporotic women because this condition is quite rare in clinical practice and is distinct from common osteoporosis, but it is highly likely that there is also a strong familial resemblance in premenopausal women with low bone mass (41).
Osteoporosis is a multigenic disease, and several strategies can be adopted to identify the genes responsible (42). Our results suggest that the best situation in which to investigate the genetic component of osteoporosis is not in postmenopausal osteoporosis, even though it is the most common form of the disease, but in osteoporosis in younger subjects, mainly males in our study, and using femoral neck density as a phenotype marker.
Acknowledgments
We thank Dr. Nicole Feingold for advisable commentaries and nurse Florence Trion for help in recruiting and interviewing the patients and their families.
Footnotes
This work was supported by a grant from the European Community "Genospora" program.
Abbreviations: BMD, Bone mineral density; FN, femoral neck; LS, lumbar spine.
Received October 17, 2001.
Accepted February 7, 2002.
References
This article has been cited by other articles:
 |
|
 |
|
  B. Lapauw, Y. Taes, S. Goemaere, K. Toye, H.-G. Zmierczak, and J.-M. Kaufman Anthropometric and Skeletal Phenotype in Men with Idiopathic Osteoporosis and Their Sons Is Consistent with Deficient Estrogen Action during Maturation J. Clin. Endocrinol. Metab., November 1, 2009; 94(11): 4300 - 4308. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J.-M. Kaufman, A. Ostertag, A. Saint-Pierre, M. Cohen-Solal, A. Boland, I. Van Pottelbergh, K. Toye, M.-C. de Vernejoul, and M. Martinez Genome-Wide Linkage Screen of Bone Mineral Density (BMD) in European Pedigrees Ascertained through a Male Relative with Low BMD Values: Evidence for Quantitative Trait Loci on 17q21-23, 11q12-13, 13q12-14, and 22q11 J. Clin. Endocrinol. Metab., October 1, 2008; 93(10): 3755 - 3762. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Balemans and W. Van Hul The Genetics of Low-Density Lipoprotein Receptor-Related Protein 5 in Bone: A Story of Extremes Endocrinology, June 1, 2007; 148(6): 2622 - 2629. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Szulc Low Free 17{beta}-estradiol Level in Men with Idiopathic Osteoporosis and Their Sons with Low Bone Mineral Density IBMS BoneKEy, January 1, 2005; 2(1): 20 - 22. [Full Text] [PDF]
|
|
|
|
|
|
J. L. Belsky, J. S. Hamer, J. E. Hubert, K. Insogna, and W. Johns Torus Palatinus: A New Anatomical Correlation with Bone Density in Postmenopausal Women J. Clin. Endocrinol. Metab., May 1, 2003; 88(5): 2081 - 2086. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
