Horm Metab Res 2010; 42(10): 758-759
DOI: 10.1055/s-0030-1261966
Letter to the Editor

© Georg Thieme Verlag KG Stuttgart · New York

Primary Aldosteronism and Diabetes Mellitus

O. Steichen1 , 2
  • 1AP-HP, Georges Pompidou Hospital, Clinical Investigation Centre, Paris, France
  • 2Paris Descartes University, Faculty of Medicine, Paris, France
Further Information

Publication History

received 03.06.2010 accepted 03.06.2010

Publication Date:
20 July 2010 (online)

I enjoyed reading the paper by Prof. Reincke et al., which reports an increased prevalence of diabetes mellitus in primary aldosteronism (PA) compared to essential hypertension (EH) [1]. We recently published a study investigating the same topic with similar methods, but did not find a significant difference of diabetes prevalence between these 2 groups [2].

Prof. Reincke et al. appropriately highlight that we did not match our patients or adjust our analyses on body mass index (BMI), antihypertensive treatment, or diastolic blood pressure (DBP). The question is whether these characteristics may have confounded our results toward the null hypothesis (no difference between patients with PA and EH) and could explain the discrepancy between the German and the French studies.

The association between a higher BMI and increased prevalence of diabetes is well documented and was also observed within both groups of patients in the German study. In the French study, however, PA patients had a similar prevalence of diabetes as EH patients despite a higher BMI. In the German study, conversely, PA patients had a higher prevalence of diabetes than EH patients despite a lower BMI. The lack of matching or adjustment on BMI therefore does not obscure the results of either study and cannot explain their discrepant results.

The German study also showed an association between a higher number of antihypertensive drugs and increased prevalence of diabetes within both groups. We did not try to account for treatment exposure in the French study because current drugs only partly reflect the whole treatment history. However, PA patients from the French study had a similar prevalence of diabetes as EH patients despite a higher number of drugs and a longer history of treatment. Treatment exposure is therefore unlikely to have biased the results toward the null hypothesis in the French study. On the opposite, the higher treatment load in PA patients from the German study may partly explain the higher prevalence of diabetes compared to EH patients. Thiazide diuretics, for example, were prescribed 7 times more often in PA patients than in EH patients.

Unadjusted analysis in the German study showed an association between lower DBP and increased prevalence of diabetes. I would guess that lower diastolic blood pressure is simply a marker of older age. The fact that the association with DBP was no longer significant after adjustment for age in PA patients supports this hypothesis. Nevertheless, and although PA patients and EH patients did not significantly differ regarding DBP in our study, I analysed the French data anew with DBP forced in the conditional logistic regression model. The results remained largely unchanged: no statistically significant difference in diabetes prevalence (17% in PA vs. 14% in EH, p=0.12), a significantly lower prevalence of impaired fasting glucose in PA patients (26% vs. 33%, p=0.008) leading to an overall prevalence of hyperglycemia similar in both groups (45% in PA vs. 48% in EH, p=0.29).

Besides the difference in treatment exposure between PA and EH patients, 2 other points are worth mentioning regarding the German study. Firstly, only 338 among 638 patients with PA were included in the matched analysis. The prevalence of diabetes in PA patients was 118/638=18% in the whole sample, 79/338=23% in those included in the matched study, and (118–79)/(638–338)=39/300=13% in those not included in the matched study. The difference in diabetes prevalence between PA patients included in the matched study and those not included is highly significant (Chi square p=0.005). Matched analyses were therefore performed on a sub-sample of PA patients with a nonrepresentatively high prevalence of diabetes. Secondly, if I correctly understood the methods, diabetes was ascertained in PA patients after a full workup in a specialized centre, whereas it was diagnosed by interview (and a nonfasting blood sample?) in patients with EH. If this is indeed the case, the higher prevalence of diabetes in PA patients from the German study might partly be explained by the fact that it was more thoroughly looked for than in EH patients.

Considering the inherent limits of retrospective studies, the association of primary aldosteronism with diabetes mellitus still awaits a well powered prospective study with appropriate control patients (including similar treatment exposure) in order to be adequately investigated.

References

  • 1 Reincke M, Meisinger C, Holle R, Quinkler M, Hahner S, Beuschlein F, Bidlingmaier M, Seissler J, Endres S. Participants of the German Conn's Registry. Is primary aldosteronism associated with diabetes mellitus? Results of the German Conn's Registry.  Horm Metab Res. 2010;  42 435-439
  • 2 Matrozova J, Steichen O, Amar L, Zacharieva S, Jeunemaitre X, Plouin PF. Fasting plasma glucose and serum lipids in patients with primary aldosteronism: a controlled cross-sectional study.  Hypertension. 2009;  53 605-610

Correspondence

O. Steichen

Centre d’Investigations Cliniques

Hôpital Georges Pompidou

rue Leblanc 20–40

75015 Paris

France

Phone: +33/156/0929 13

Fax: +33/156/0929 29

Email: olivier.steichen@tnn.aphp.fr

    >