Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Twin data support a genetic contribution to fracture risk
- Alex J MacGregor (13 December 1999)
-
On the genetic contribution of osteoporosis and osteoporotic fractures
- Amado Salvador Pena (23 December 1999)
-
Authors' reply
- Pekka Kannus (3 March 2000)
|
|
|||
|
Alex J MacGregor
Send response to journal:
|
Kannus et al (1) suggest from prospective data collected on Finnish twins that genetic factors are only of minor importance in explaining the population occurrence of osteoporotic fracture, particularly in females. The evidence given to support this is the relatively small excess in concordance in monozygotic (MZ) twins when compared to dizygotic (DZ)twins. However, it is well recognised that twin concordances may be misleading unless the underlying prevalence of a disease is taken into account (2). For example, a small absolute difference in MZ vs. DZ concordance is more suggestive of a genetic effect for a trait that is relatively rare (such as fracture) than for one that is common. The data thus warrant closer scrutiny. We have estimated the relative contribution of genetic, shared environmental and unique environmental components to the variation in susceptibility to fracture in these twins from the data provided. The analysis was conducted using a variance components approach using the statistical software Mx (3). The method assumes that fracture risk is determined by a continuous underlying liability and is a plausible assumption for this trait (4). As expected, the results show significant evidence of familial resemblance in fracture risk in both males and females. In females, despite the nationwide sampling frame, there is insufficient statistical power in this study to distinguish between models containing components in which this clustering is attributed to genetic factors alone, the shared family environment of the twins or the combination of the two. In a model in which the only contribution is from genetic and unique environmental factors, genetic factors account for 36% of the variance in the liability to fracture at any body site. In males, the familial resemblance is explained by a significant contribution from genetic factors but not by the shared family environment, with genetic factors accounting for 35% of the variation in liability to fracture. A greater genetic contribution is also suggested at the spine in Table A ( www.bmj.com/cgi/content/full/319/7221/1334/DC1), although inference is limited by the small numbers of concordant pairs, and the lack of data on sex differences in fracture rate. In contrast to the conclusion reached by the authors, these data show that genetic factors contribute to a third of the liability to osteoporotic fracture in males and are entirely compatible with the hypothesis that genetic factors contribute to a similar extent in females. The data suggest there may be differences in the nature of the genetic risk in males and females and at different body sites that merit further study. Alex J MacGregor, Arthritis Research Campaign Senior Fellow
References 1. Kannus P, Palvanen M, Kaprio J, Parkkari J, Koskenvuo M. Genetic factors and osteoporotic fractures in elderly people: prospective 25 year follow up of a nationwide cohort of elderly Finnish twins. BMJ 1999; 319(7221):1334-1337. 2. Smith C. Concordance in twins: methods and interpretation. Am.J Hum.Genet. 1974;26:454-466. 3. Neale MC. Mx: Statistical modeling. 4th Edition. Box 126 MCV, Richmond, VA 23298: Department of Psychiatry. 1997. 4. Falconer DS. Introduction to quantitative genetics. Harlow: Longman Scientific and Technical, 1989. |
|||
|
|
|||
|
Amado Salvador Pena, Professor of Gastrointestinal Immunology Vrije Universiteit Amsterdam
Send response to journal:
|
Editor, We very much appreciated the large, long-term prospective study of Kannus et al. (1) on analysing the genetic predisposition of osteoporotic fractures in elderly Finnish twins. However, it is important to note that whereas the concordance rate for fractures were indeed not strikingly different between monozygotic and dyzygotic twin pairs in women, there was a fourfold difference in men. The estimation on the relative contribution of genetic components performed by McGregor et al. (2) based on the data provided has indeed shown a strong role fore genetic factors in men, while not in women. Previous studies in the literature pointed to possible differences between the main regulatory factors in the development of low BMD and osteoporotic fractures between the two genders. Therefore, we suggest that data by Kannus et al. (1) could be interpreted as further evidence supporting this hypothesis. Bone density is suggested to be multifactorially regulated and polygenically determined. Underlying diseases such as inflammatory bowel disease and several drugs – like glucocorticosteroids can modulate and enhance the effect of genetic factors on the development of osteoporotic fractures. For example, our studies have shown that the allele 2 at the AvaI polymorphism in the interleukin-1 beta gene – that is related with a higher production of the cytokine –is associated with a subgroup of patients, the non-fistulising form of Crohn’s disease (3). Additionally, we have found a strong correlation between this polymorphism and bone mineral density in patients with inflammatory bowel diseases but not in healthy controls. There was also in our data a difference between the two genders: the association at the lumbar spine was only present in men, not in women (4; full article submitted). Bone loss at the cortical and trabecular bones seem to have differently influenced by several physiologic and other elements such as menopause or corticosteroid usage. Therefore, analysing data after screening for the often "silent" vertebral fractures could further modify the results of Kannus et al. concerning the importance of the genetic background. Bone mineral density is a key predictor of osteoporotic fractures. To judge the weight of genetic influence it would be important to compare concordance rates in the groups of patients with low respectively high Z– score values. On the whole we suggest that studies in well defined subgroups of patients may help to define those conditions where genetic background has a high prognostic value on osteoporotic fractures risk.
Andrea Nemetz, M.D., Amado Salvador Pena, MD, FRCP References: 1. Kannus P, Palvanen M, Kaprio J, Parkkari J, Koskenvuo M. Genetic factors and osteoporotic fractures in elderly people: prospective 25 year follow up of a nation-wide cohort of elderly Finnish twins. BMJ 1999; 319:1334-1337. 2. A. J MacGregor, H. Snieder, T. D Spector. Twin data support a genetic contribution to fracture risk. BMJ 1999; 319:electronic letter 3. A. Nemetz, M.P. Nosti-Escanilla, T. Molnár, A. Köpe, Á. Kovács, J. Fehér et al. IL1B gene polymorphisms influence the course and severity of inflammatory bowel disease - Immunogenetics 1999; 49:527-531. 4. A. Nemetz, T. Zágoni, M. Tóth, Á. Kovács, M.P. Nosti-Escanilla, M.A. García-González et al. Interleukin-1 gene polymorphisms stimulate bone loss in inflammatory bowel diseases. Gut 1999; 45 (Suppl.V), A15 |
|||
|
|
|||
|
Pekka Kannus
Send response to journal:
|
EDITOR- Based on our published study data, MacGregor et al have computed estimates of genetic variance in liability to osteoporotic fractures. It is of interest that our original manuscript already had the same modelling analysis (Table below), but after a clear recommendation of the first reviewer and the BMJ editorial committee, the modelling part of the study was omitted from the revised final paper. MacGregor et al.'s interpretation of the modelling results somewhat differs from ours. Based on the results and discussion of our original paper and the table below, we want to emphasise three points. First, no matter what model is used to examine our data, it is clear that in both women and men there is a large unshared environmental component (always 60% and over) to explain the liability to osteoporotic fracture. This fact is not clearly pointed out by MacGregor et al. Second, in our interpretation of the data we do not want to say that "genetic factors are only of minor importance in explaining the population occurrence of osteoporotic fracture", but that they are not very strongly related to it and thus not the best explanation for fracture occurrence. We thus want to draw attention to the fact that although genetic factors are known to have a dominant role in explaining inter-individual variation in bone mass and density, the result is quite different (ie, unshared environmental effects become dominant) when the end point is changed from bone tissue to fractures, the true end point of the entire osteoporosis problem. Third, in women MacGregor et al seem to highlight the AE- model, in which genetic factors account for about 36% of the variance in liability to fracture, but ignore that fact that the CE- model below (no genetic effects to explain twin similarity in liability to fracture) gave a better fit to the data than the AE-model. In this context, it has to be remembered, however, that the statistical power of the study was not sufficient to make a statistically significant differentiation between these competing explanations. As clearly said in the Discussion of our paper and also noted by MacGregor et al, more incident cases are needed for more definitive conclusions. It is our intention to follow our twin cohort further to clarify the fracture development in the coming years. Finally, we fully agree with MacGregor et al's last statement that there can well be differences in the nature of the genetic risk of osteoporotic fracture in males and females and at different body sites. Further follow up of our cohort and examination of the situation in other populations are thus needed and most welcome. As mentioned above, we agree with Nemetz and Pena that the genetic risk for fracture can be different in men and women and at different body sites. However, analysis of "silent" vertebral fractures will not change the result of the clinically more important nonvertebral fractures. We do not fully agree with Nemetz and Pena's statement that "bone mineral density is a key predictor of osteoporotic fractures". As mentioned in our Discussion, many recent epidemiological studies have shown that falling is clearly the strongest single predictor of these fractures and BMD is a moderate-level independent predictor only. When this fact becomes more largely recognized, then the entire framework of fracture prevention will become shifted more properly, ie, towards prevention of falls in elderly people. Pekka Kannus Jaakko Kaprio Markku Koskenvuo Mika Palvanen Jari Parkkari
-------------------------------------------------------------------------
Proportion of variance
accounted for by:
---------------------------
(A) (D) (C) (E) chi P value AIC*
-------------------------------------------------------------------------
Men
Model
E 1.000 15.28 0.009 5.28
CE 0.216 0.784 7.22 0.13 -0.78
AE 0.347 0.653 3.28 0.51 -4.72
ACE 0.347 0.000 0.653 3.28 0.35 -2.72
ADE 0.000 0.398 0.602 1.83 0.61 -4.17
Women
Model
E 1.000 26.68 0.00 8.44
CE 0.260 0.740 0.57 0.97 -7.44
AE 0.367 0.633 1.35 0.85 -6.45
ACE 0.146 0.168 0.686 0.10 0.99 -5.91
ADE 0.367 0.000 0.633 1.35 0.72 -4.65
---------------------------------------------------------------------------
A=additive genetic effects.
D=non-additive genetic effects.
C=shared environmental effects.
E=unique environmental effects.
*AIC or the Akaike's information criteria is a statistics which combines
information on the goodness of fit statistics (the lower the chi squared
value the better the fit) and the simplicity of the model, the best model
usually being the one with the lowest AIC value.
|
|||