doi: 10.1111/j.1365-2796.2011.02413.x Maternal and paternal transmission of type 2 diabetes Dear Sir, Abbasi et al. [1] published an interesting article esti- mating the contribution of diet, lifestyle and adiposity to the familial risk of type 2 diabetes (T2D). They also reported data on the risk of T2D in offspring when both parents were diabetic. Besides our recently re- ported work, few other previous studies have investi- gated such effects [2, 3]. The interactions between the underlying genetic pathways can be better under- stood by comparing the risks between unilineal and bilineal transmission of disease. An additive model of parental transmission is interpreted to imply inheri- tance of two non-interactive variants or sets of vari- ants influencing different pathways whereas a multi- plicative model would imply interactive variants or sets of variants [3, 4].To test for the interaction, we have used the Dutch data as reported by Abbasi et al. [1] with their fully adjusted Model 4. Our own dataset included Swedish patients hospitalised for T2D between 1964 and 2007, as previously described [2]. For the Swedish data, we calculated familial relative risks (RRs) using Poisson regression for T2D in offspring (n = 22 654; aged 40–75 years) depending on the number of affected parents, also aged 40–75 years [2]. Analyses were adjusted for age (5-year groups), sex, period (5-year groups), region and socioeconomic status. Table 1 shows the results from both datasets of familial risks from one and two affected parents with the data fitted to the additive (2·[RR ) 1.0] + 1.0) and multiplicative (RR · RR) models [5]. In the Swedish dataset the RRs were 2.34 for one affected parent and 4.28 for two affected parents; in the Dutch dataset these values were 2.20 and 3.92, respectively. For both datasets, the additive model fitted better than the multiplicative model, and in the Swedish dataset the multiplicative model was rejected (95% confidence intervals did not overlap between empirical and modelled RR values). The small Framingham study reporting 11 bilineal fami- lies also supported an additive model [3]. Additive effects have also been favoured when low-penetrance data on T2D have been modelled; however, with a caveat that modelling may not be accurate as long as the functions of many low-penetrance loci remain unknown [6, 7]. Acknowledgements This work was supported by the Swedish Council for Working Life and Social Research and Deutsche Krebshilfe. Conflict of interest No conflicts of interest to declare. K. Hemminki 1,2,3 , X. Li 3 , J. Sundquist 3,4 & K. Sundquist 2,3 From the 1 Division of Molecular Genetic Epidemiology, German Cancer Research Center (DKFZ), Heidelberg, Germany; 2 Center for Family and Community Medicine, Karolinska Institute, Huddinge, Sweden; 3 Center for Primary Health Care Research, Lund University, Malmo ¨ , Sweden; and 4 Stanford Prevention Research Center, Stanford University School of Medicine, Stanford, CA, USA Table 1 Modelling of familial risk of type 2 diabetes based on the number of affected parents Proband Observed Additive Multiplicative No. RR 95% CI RR 95% CI RR 95% CI Risk to offspring, one or two parents affected One parent 2284 2.34 2.24–2.45 Two parents 79 4.28 3.43–5.34 3.68 3.46–3.91 5.48 5.26–5.70 Abbasi et al. (Model 4) One parent (mother) 201 2.20 1.87–2.60 Two parents 36 3.92 2.78–5.50 3.40 2.41–4.79 4.84 3.43–6.82 RR, relative risks. ª 2011 The Association for the Publication of the Journal of Internal Medicine 293 Letter to the Editor |