In this issue of The American Journal of Medicine Israel and Grossman1 report an interesting observation based on their analysis of the Systolic Blood Pressure Intervention Trial (SPRINT) and the National Health and Nutrition Examination Survey (NHANES) datasets. They used gradient boosting machines to build predictive models for hyponatremia based on SPRINT participant high-density lipoprotein cholesterol (HDL-C) levels, along with other baseline characteristics. They used SPRINT’s intensive and standard arms for training and validation models, respectively. They found HDL-C levels >62 mg/dL to be an independent predictor of future development of hyponatremia (serum sodium ≤130 mEq/L). They further studied the NHANES dataset between the years 2005 and 2010 and confirmed this association.
We find this work to provide an important and novel discovery that potentially can impact the treatment and follow-up of subjects with hypertension at high risk for hyponatremia. Several points are worth mentioning when looking at the details of this work. First and foremost, data on drug therapies, especially antihypertensives and diuretics, were lacking owing to the authors’ limited access to the SPRINT dataset. Such data could have significantly impacted the results, especially that thiazide diuretics would likely be highly prescribed in SPRINT participants and knowing its high lifetime risk for hyponatremia. Second, although results were significant in both datasets, it is important to point out that in the SPRINT analysis, HDL-C was a predictor of future development of hyponatremia that was not present at baseline, whereas the NHANES analysis suggested rather an association based on the nature of the analysis in this dataset being cross-sectional. Moreover, and as expected on the basis of the different characteristics of participants in the 2 datasets, hyponatremia was more prevalent in SPRINT. Finally, characteristics and types of hyponatremia could not be further elucidated on the basis of the materials presented.
With the current available evidence, these results cannot be extended beyond an association, and further research is needed in which some element of causation can be dissected. The idea of arginine vasopressin receptor sensitivity and arginine vasopressin 1b gene polymorphism is of interest to this work and can explain some of its findings. Alternatively, knowing its involvement in oxidative stress and its associations with hyponatremia, uric acid can also be of interest to this association. Uric acid has been shown to mediate proinflammation in adipose tissues, and its lowering with a xanthene oxidoreductase inhibitor can improve this trend.2 In a NHANES III analysis, HDL-C was found to be significantly inversely associated with serum uric acid level.3 Negative uric acid balance and uricosuria along with hypouricemia are consistent findings in euvolemic hyponatremia, even if diuretics are used.4 Taken all together, hypouricemia can be associated and explains both elevated HDL-C as well as euvolemic hyponatremia. It would be interesting to know whether uric acid levels have anything to do with this association reported by Israel and Grossman, but unfortunately such levels were not available in SPRINT.
Until further clarification is made to this association, we agree with the authors’ conclusion and stress the point that subjects with elevated HDL-C levels need to be closely followed and monitored for potential development of hyponatremia, especially if they harbor other risk factors.
We urge researchers to pay attention to this association in future large prospective trials.
To read this article in its entirety please visit our website.
-Zeid J. Khitan, MD, Joseph I. Shapiro, MD
This article originally appeared in the November 2017 issue of The American Journal of Medicine.
