Pharmacogenetic and microRNA mechanisms of beta blocker use on bone

Abstract

Motivated by studies showing an association between beta blocker (BB) use and positive bone outcomes, a pilot randomized control trial was performed at the Mayo Clinic which randomized postmenopausal women to placebo, propranolol (40 or 80 mg twice daily), atenolol (50 mg/d), or nebivolol (5 mg/d) to determine changes in bone turnover markers (BTMs) and in BMD over 20 wk. Pharmacogenetic effects and microRNA-mediated mechanisms involving beta adrenergic receptor and related genes have previously been found. We sought to validate these effects and discover new candidates in an ancillary study to the pilot clinical trial. We genotyped all participants and performed microRNA (miRNA) sequencing at baseline and at 20 wk for 24 participants from the atenolol or placebo groups. We discovered several variants in ADRB1, ADRB2, and HDAC4 which showed significant pharmacogenetic effects with BMD at multiple sites and with BTMs. Our miRNA results showed a significant treatment effect for miR-19a-3p over time with atenolol use in the low-responder group compared to placebo. Overall, the longitudinal miRNA analysis showed a large number of miRNAs which were up-regulated over the trial in the low responders but not the high responders compared to placebo, of which miR-19a-3p was one example. Finally, we compared the response to atenolol treatment for cardiovascular traits (pulse and blood pressure) with the response for the bone resorption marker, C-terminal telopeptide, and found a largely independent effect. Our results have implications for personalized therapy and for understanding mechanisms of BB treatment effect on bone.

Keywords: BMD; beta adrenergic signaling; beta blocker; bone mineral density; miRNA; microRNA; osteoporosis; pharmacogenetics; pharmacogenomics.

Graphical Abstract

Figure 1

Consort diagram for pharmacogenetics (n = 164) and microRNA-seq (n = 24) studies. 164 postmenopausal women were randomized to one of five treatment groups: Placebo, propranolol (20 mg or 40 mg twice daily), atenolol (50 mg/d), or nebivolol (5 mg/d), and all these women were included in the pharmacogenetics study. A subset of 24 women were included in the microRNA-seq study with miRNA collected from serum at baseline and 20 wk (6 from the placebo group and 18 from the atenolol group, 9 each from the high-responder (HR) or low-responder (LR) groups determined by change in C-terminal telopeptide (CTX) over the trial).

Figure 2

Box plots of pharmacogenetic associations for previously discovered and new variants in ADRB1 and ADRB2. Note a box plot is a descriptive summary only and does not show results of statistical tests. For statistical test results of additive genetic models, please see Table S1. Each dot represents a single data point and the boxplot is a five number summary of the data points for a given group, which includes the minimum value or the median minus 1.5 × the interquartile range (IQR), whichever is larger (lower whisker), the first quartile (lower box boundary), the median (box centerline), the third quartile (upper box boundary), and the maximum value or the median plus 1.5 × IQR, whichever is smaller (upper whisker). In the case of only two data points, the median is the average of the two points, the whiskers are the smaller and larger of the two values, respectively, and the box boundaries are at one half the distance between the median and the lower or upper whisker for the lower and upper box boundaries, respectively. (A) rs1801252 A allele in ADRB1 for Ultradistal Radius for propranolol 80*. (B) rs1801252 A allele in ADRB1 for femoral neck BMD for propranolol 40*. (C) rs1042714 G allele in ADRB2 for total body BMD for propranolol 80. (D) rs1042720 G allele in ADRB2 for propranolol 40. *Note: frequency of 0 allele group too low for box plot so excluded from plot (not analysis).

Figure 3

Volcano plot for differentially expressed microRNA comparing atenolol-treated high responder (HR), atenolol-treated low responder (LR), or placebo-treated individuals at baseline, where atenolol response was defined based on the change in C-terminal telopeptide (CTX) over the trial. Labeled miRNA are significant at FDR < 0.05.

Figure 4

Box plot for individual microRNA values over time and heatmap of significant longitudinal results. Note a box plot is a descriptive summary only and does not show results of statistical tests. For statistical test results of additive genetic models, please see Table S1. Each dot represents a single data point and the boxplot is a five number summary of the data points for a given group, which includes the minimum value or the median minus 1.5 × the interquartile range (IQR), whichever is larger (lower whisker), the first quartile (lower box boundary), the median (box centerline), the third quartile (upper box boundary), and the maximum value or the median plus 1.5 × IQR, whichever is smaller (upper whisker). In the case of only two data points, the median is the average of the two points, the whiskers are the smaller and larger of the two values, respectively, and the box boundaries are at one half the distance between the median and the lower or upper whisker for the lower and upper box boundaries, respectively. (A) Boxplot for miR-4467 at baseline and endpoint for atenolol high responder, low responder, and placebo. (B) Boxplot for miR-19a-3p at baseline and endpoint for same treatment groups. (C) Heatmap of median expression across treatment groups for baseline (T1) and endpoint (T2) for the set of microRNA found to be significant in longitudinal analysis.

Figure 5

Box plots for change in bone or cardiovascular outcomes between baseline and endpoint for atenolol higher-responder, low-responder, and placebo groups. Note a box plot is a descriptive summary only and does not show results of statistical tests. For statistical test results of additive genetic models, please see Table S1. Each dot represents a single data point and the boxplot is a five number summary of the data points for a given group, which includes the minimum value or the median minus 1.5 × the interquartile range (IQR), whichever is larger (lower whisker), the first quartile (lower box boundary), the median (box centerline), the third quartile (upper box boundary), and the maximum value or the median plus 1.5 × IQR, whichever is smaller (upper whisker). In the case of only two data points, the median is the average of the two points, the whiskers are the smaller and larger of the two values, respectively, and the box boundaries are at one half the distance between the median and the lower or upper whisker for the lower and upper box boundaries, respectively. (A) Boxplot for change in C-terminal telopeptide (CTX) which was used to define treatment groups (placebo and atenolol high and low responders). (B) Boxplot for change in systolic blood pressure (p > .05 for both contrasts). (C) Boxplot for change in diastolic blood pressure (p < .1 for atenolol low responder compared with placebo, not significant for high responder). (D) Boxplot for change in pulse (p < .001 for both atenolol high and low responder compared with placebo).