Sex‐specific differences and predictors of echocardiographic measures of diastolic dysfunction in rhesus macaques (Macaca mulatta)

Diastolic dysfunction in humans is an age‐related process with an overrepresentation in women. In rhesus macaques (Macaca mulatta), the incidence and predictors of diastolic dysfunction have yet to be reported.

(LV) relaxation and diastolic filling. 7The primary cause of reduced LV relaxation is attributed to LV stiffening as a result of increased canonical myocardial fibrosis.In the case of severe diastolic dysfunction (DD), overtime, the dysfunction can progress to left-sided congestive heart failure (CHF) causing a patient to develop clinical signs (fatigue, shortness of breath, exercise intolerance, and/or arrythmias).In the human population, DD is a common pathophysiologic finding on echocardiography in elderly individuals that have been cleared from any known cardiac disease. 8In younger populations, echocardiographic evidence of DD serves as an early clinical finding for common cardiomyopathies afflicting the left side of the heart which can lead to sinister outcomes, such as sudden cardiac death (SCD), if left untreated. 9One of the most common and reliable ways of defining DD is by interrogating transmittal blood flow parameters (E/A ratio) on spectral Doppler echocardiography, although other methods exist. 10With this echocardiographic parameter, spectral Doppler echocardiography measures the magnitude (in cm/sec) of blood flow speed as it fills the LV during diastole.A passive (E) and active (A) wave are shown which have specific relationships defining a normal or compliant versus a stiff ventricle.In the case of assessing DD, the speed of passive filling (E) when compared to the speed of active filling (A) is reported as a ratio.Smaller ratios, in particular, those less than 0.9, represent a ventricle that is abnormally stiff and predominantly relies on active rather than passive filling waves. 11sessing evidence of DD on echocardiography serves as a powerful tool for clinicians to improve patient prognosis and quality of life via early clinical intervention strategies.
In humans, the incidence of asymptomatic DD is overrepresented in women for reasons that have yet to be fully understood, although age, endocrinologic, and exercise activity level differences between sexes are thought to play a role. 12,13Women are overrepresented in cases of heart failure with preserved ejection fraction (HFpEF). 14In such cases, women have stiff and noncompliant heart muscle which is consistent with DD.Variables like body weight, age, and sex have previously been shown to contribute to the incidence of DD and serve as predictor variables for E/A ratios in humans. 10,15,16The incidence of DD between sexes and predictors of DD measures in rhesus macaques have yet to be reported.
Here we report the results of a single-center observational study that aimed to determine predictors and sex-specific differences in echocardiographic measures of DD in rhesus macaques which should be considered for future study designs, as well as to determine relationships between other commonly used methods of DD assessment, such as LV isovolumetric relaxation time (msec) (LV-IVRT) and IVRT normalized to E wave velocity (E:IVRT), to E/A ratios.We hypothesized that the incidence of DD is overrepresented in clinically healthy females, and that body weight, age, and sex will serve as predictors of E/A ratio measures in rhesus macaques.We report sex-specific differences in the incidence of DD, the effects of sex, body weight, and age on E/A ratio as an index of DD, and the relationships between common echocardiographic indices of diastology in a large juvenile, adult, and geriatric population of 631 clinically healthy rhesus macaques.

| Humane care guidelines
Complete echocardiographic assessments were conducted over a five-year period (2014-2019) at the California National Primate Research Center (CNPRC).8][19] The use of animals in this study were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at the UC Davis CNPRC.

| Animal care and husbandry
This is a retrospective study where no animals where used, albeit, data was obtained from previously performed IACUC approved projects where all animals were humanely handled and cared for according to standard regulations and CNPRC protocols as previously defined. 20Briefly, echocardiographic measurements were obtained for monkeys housed in either indoor, outdoor, or mixed configurations.Monkeys of the same housing configuration were exposed to the same environmental conditions, including diet (LabDiet Monkey Diet 5047 [Purina Mills International, St Louis, MO] and fresh mixed vegetables/fruits), lighting exposure, and environmental enrichments.Routine wellness evaluations, clinicopathological assessments, and disease testing (tuberculosis, herpes B virus, simian Type D retrovirus, simian immunodeficiency virus, and simian Tlymphotropic virus) were performed.

| Echocardiographic assessments and data acquisition
Complete echocardiographic data were stored and retrieved from our offline echocardiographic database.Sedation (ketamine) was administered for ease of animal handling; no animal(s) was euthanized for the completion of this study.Complete echocardiographic evaluations on sedated rhesus macaques were performed by a boardcertified cardiologist (J.A.S.) or an experienced sonographer (Y.U.) under the direct supervision of a board-certified cardiologist (J.A.S.) with the use of an Affinity 50 echocardiographic ultrasound (Phillips, Amsterdam, Netherlands).Ketamine has been previously shown to alter cardiovascular function 21 ; however, the dose and route of administration was standardized in this study to ensure the effects of this confounding variable is minimized.Echocardiographic evaluations were performed with either a 4-to 12mHZ or 1-to 5-mHZ sector array transducer depending on animal size.Animals were placed in left and right recumbency, and echocardiographic images were obtained via 2D, M-mode, color, and spectral Doppler modalities.
Echocardiographic images were subsequently analyzed, and standard echocardiographic variables measured with the use of the Syngo Dynamics offline software (Siemens, Erlangen, Germany) according to adult comprehensive transthoracic echocardiographic measurement guidelines. 22ere multiple studies were present for a single monkey, the most recent echocardiographic measurements were retained for our analyses.Animals with apparent evidence of any acquired or congenital cardiac disease on echocardiography, including presence of ventricular septal defects, left ventricular hypertrophy (LVH; IVSd and/or LVPWd >6.5 mm for animals <9 years-old, or IVSd >8.8 mm and/or LVPWd >7.4 mm for animals >9 years-old), or dilated cardiomyopathy (LVEF% <50%); DD was not excluded and was diagnosed when the E/A ratio was <0.9. 4,20,23,24A final curated data set including patient demographic data (predictor variables; body weight [kg], age [yrs], and sex) and index of DD (outcome variable; E/A ratio) was obtained.All animals in this study were considered to be apparently healthy representatives of their species and ages group based upon echocardiographic, electrocardiographic, biochemical profile, and physical examinations.

| Statistical analyses
Univariate analyses were conducted for the sex and DD categorical variables, and for the following continuous variables: body weight (BW), age (yrs), and E/A ratio.Normality testing was carried out to assess parametric versus nonparametric data characteristics.
Variables with a Shapiro-Wilk test p < .05,and visual identification of data points not falling within the best fit line in their corresponding quantile-quantile (QQ) plot, was considered non-normally distributed (nonparametric).Variables that failed normality testing were subject to log-transformation of their raw values; summary statistics of raw measurements from univariate analysis (range, standard deviation, and IQR) and results from Shapiro-Wilk tests before and after log-transformation are reported for descriptive purposes; however, the log-transformed values of these non-normally distributed variables were used in the analyses for approximation of normal to the greatest extent.
Because the predominant objective of this study was to ascertain sex-differences between echocardiographic measures of DD, a bivariate analysis between Log_E/A and Sex was performed using a Mann-Whitney U test, where median, ranges, standard deviations, and IQRs of the raw values are reported.Additional Mann-Whitney U tests were performed to determine differences in Log_LV-IVRT and Log_E:IVRT between individuals with and without DD as previously characterized by their corresponding E/A ratios.Spearman's correlation tests of Log_E/A with Log_BW and Log_Age was performed; additional Spearman's correlation tests were performed to determine the relationship between Log_E/A values with Log_LV-IVRT and Log_E:IVRT.Hypothesis testing was performed to evaluate whether DD incidence between male and female rhesus macaques differ using a two-sided 2 × 2 contingency chi-square test.A 4 × 2 contingency table was constructed and a chi-square test performed to identify differences between housing configurations (indoor, outdoor, mixed-indoor, and mixed-outdoor) and the incidence of DD. these numerical bins were selected to mirror the increasing activity level of monkeys with greater outdoor exposure.Animals that were housed >75% of either indoor or outdoor configurations were assigned as strictly indoor or outdoor monkeys.Monkeys housed >50% indoor, but <75% indoor at the time of the latest echocardiographic evaluation, were defined as mixed-indoor monkeys.
Similarly, animals with >50% outdoor, but <75% outdoor exposure at the time of the latest echo, where classified as mixed-outdoor mon-

| Univariate analyses
Initial univariate analysis of sex and DD revealed that there are 346 female and 285 male rhesus macaques in our data set (Table 1; Figure 1); excluding missing values for 37 individuals, a total of 59 (9.93%) rhesus macaques showed evidence of DD in the population (Table 1; Figure 2).Univariate analysis on the raw values of  1; Figures 3-5).Subjective interpretation of the resulting QQ plots of the log-transformed values displayed improved data points falling within the best fit line for all variables and followed either a moderately left skewed or an approximately normal distribution, therefore, their log-transformed values were used for all subsequent statistical analyses.Given that all animals used in this study are apparently healthy at the time of echocardiography, no outliers were excluded from the study as they represent true variation within echocardiographic measurements for rhesus macaques within the CNPRC population.F I G U R E 1 Number of female and male rhesus macaques.The total number of female (gold) and male (dark blue) rhesus macaques in the echocardiographic dataset comprised of a total of 631 subjects are displayed.

F I G U R E 2 Incidence of diastolic dysfunction in rhesus macaques.
The total number of rhesus macaques with (gold) and without (dark blue) echocardiographic evidence of DD defined by an E/A < 0.9 is presented (n = 594); animals with missing E/A ratios (gray) were excluded from further statistical analyses.DD, diastolic dysfunction.

| Incidence of diastolic dysfunction between sexes and housing configurations
A total of 46 out of 324 females (14.2%) and 13 out of 270 males (4.81%) rhesus macaques displayed DD in our dataset.Female rhesus macaques displayed a statistically significant increase in DD incidence compared to that of their male counterparts (p = .0001,χ2 = 14.49) (Table 3; Figure 9).Out of the total monkeys with an obtained E/A ratio (n = 594), DD was identified in 18 out of 285 (6.32%), 9 out of 45 (20%), 19 out of 179 (10.61%), and 13 out of 85 (15.29%) in outdoor, indoor, mixed-indoor, or mixed-outdoor housed monkeys, respectively.Statistically significant differences in the incidence of DD were identified by housing configuration (indoor, outdoor, mixed-indoor, and mixed-outdoor; p = .0071,χ2 = 12.09).Post hoc testing identified a single comparison that withstood Bonferroni correction and demonstrated that strictly indoor versus strictly outdoor housed monkeys had a higher incidence of DD (p Adjusted = .01).supported by their observed differences of mean E/A ratios. 25,26milar to humans, age and sex are significant predictors of E/A ratios in rhesus macaques. 27When comparing other measures of DD (i.e., LV-IVRT and E:IVRT) to our DD criteria (E/A), statistically significant differences were not observed in LV-IVRT between rhesus macaques with and without DD; however, as observed in human populations, differences between DD status groups were found when standardized to passive filling of the LV (E:IVRT).The results from this analysis support that E:IVRT values serve as a more reliable predictor of DD in rhesus macaques as opposed to LV-IVRT alone; the authors suggest that E:IVRT may represent a second valuable index for assessment of diastology in rhesus macaques.

| Stepwise and multiple linear regression
Similar to human reports, rhesus macaque E/A measurements are negatively correlated with age and body weight.
Post hoc testing for each possible housing configuration comparison was performed via sequential 2 × 2 contingency table constructs, chi-square testing, and Bonferroni adjustment of p values.To determine whether BW, age, and sex serve as predictors of Log_E/A values, a stepwise multiple linear regression (SMLR) analysis was performed considering Log_Age, Sex (Female[1],Male [0]), Log_BW, and the interactions between Log_Age*Sex, Log_Age*Log_BW, and Log_BW*Sex.It was of particular interest to investigate the impact of potential environmental factors on the index of DD; therefore, we introduced Housing Configuration and Specific Pathogen-Free (SPF) Status, as well as the interaction between these two variables in our model as proxies of activity levels and/or potential pathogen-mediated changes that might help explain Log_E/A patterns.Housing configuration was converted to numerical values for SMLR and subsequent multiple linear regression (MLR) analyses.Monkeys were numerically binned as either of indoor (1), mixed-indoor (2), mixed-outdoor (3), and outdoor (4) configuration; keys.SPF monkeys of both level 1 (negative for simian Type D retrovirus, simian immunodeficiency virus, simian T-lymphotropic virus 1, and herpes B) and level 2 (negative for simian Type D retrovirus, simian immunodeficiency virus, simian T-lymphotropic virus 1, herpes B, cytomegalovirus, simian foamy virus, and rhesus monkey rhadinovirus) were included in this study and grouped as a single cohort for purposes of analyses in contrast to their conventional (non-SPF) counterparts.Variables were entered into the model with an entry significance level of .25;variables that did not produce an F statistic significant at a .15level were excluded from the model.To determine the percent variability of Log_E/A values explained by variables deemed to be significant predictors of Log_E/A in our SMLR model, a subsequent MLR analysis was performed.Statistical analyses were first performed using SAS Studio Software and GraphPad Prism.For all statistical analyses, missing values were excluded and a p value of <.05 was considered significant.

F I G U R E 3
Body weight distribution of rhesus macaques.A violin plot of the distribution of raw BW values (A), frequency distribution of BW values before (B) and after (D) log-transformation, and QQ plots of the raw (C) and log-transformed (E) BW values in a total of 631 clinically healthy rhesus macaques are presented.Shaded best fit curves of the data are provided for values before (B: blue) and after (D: yellow) log-transformations.BW, body weight; QQ, quantile-quantile.F I G U R E 4 Age distribution of rhesus macaques.A violin plot of the distribution of raw age values (A), frequency distribution of age values before (B) and after (D) log-transformation, and QQ plots of the raw (C) and log-transformed (E) age values in a total of 631 clinically healthy rhesus macaques are presented.Shaded best fit curves of the data are provided for values before (B: blue) and after (D: yellow) logtransformations.QQ, quantile-quantile.

F I G U R E 6
SMLR analysis with all aforementioned variables included in the model revealed that, in stepwise order, only Log_Age, Housing Configuration, Sex, and SPF Status were included in the model, albeit, in the last step, SPF Status was removed from the model (p < .0001,p = .0026,p = .0030,and p = .1786,respectively).When these variables were interrogated via MLR, Log_Age, Housing Configuration, and Sex remained to be statistically significant at predicting Log_E/A values (p < .0001,p = .0008,and p = .0018,respectively) while explaining 32.05% of the variability in Log_E/A measurement (Table4).Log_E/A ratio can be calculated under this model via the following equation: 4 | DISCUSS ION This work capitalizes on a large dataset comprised of juvenile, adult, and geriatric clinically healthy rhesus macaques.Female rhesus macaques are more likely to develop DD as opposed to males, Log _ E ∕ A = 0.60 + ( − 0.20 × Log _ Age − 0.04 × Housing Configuration − 0.07 × Sex) Difference of E/A ratios between rhesus macaque sexes.Box plots depicting differences of E/A ratios between 324 females (gold) and 270 males (dark blue) rhesus macaques are displayed.The top and bottom of each box plot depicts the 75th and 25th percentiles, respectively; whiskers represent the minimum and maximum values.****p < .0001.F I G U R E 7 Differences between measures of diastolic function in rhesus macaques.A box plot (A) illustrating no statistical difference between LV-IVRT values of 59 DD (gold) and 535 no DD (dark blue) rhesus macaques.The top and bottom of each box plot depicts the 75th and 25th percentiles, respectively; whiskers represent the minimum and maximum values.Violin plots (B) depicting statistically significant differences in E:IVRT ratios of rhesus macaques with DD and without echocardiographic evidence of DD. ****p < .0001.DD, diastolic dysfunction; IVRT, isovolumetric relaxation time; LV, left ventricular.

F I G U R E 9 F I G U R E 8
Distribution of diastolic function between male and female rhesus macaques.Bar graphs illustrating the total number of male (dark blue) and female (gold) rhesus macaques discriminated by echocardiographic presence or absence of DD are presented.Note the overrepresentation of female subjects with DD.DD, diastolic dysfunction.Correlation matrices of demographic variables and indices of diastolic function.Results from a Spearman's correlation test for demographic variables and Log_E/A values in 594 animals (A) and other measures of diastolic function (B) are illustrated.Where comparisons include IVRT a total of 590 observations are included.Strength of correlation (dark blue = strong positive correlation, gold = strong negative correlation) is displayed via heat map, r values for each comparison are contained within each corresponding box.***p = .002,****p < .0001.IVRT, isovolumetric relaxation time; LV, left ventricular.

TA B L E 1
Univariate analysis of sex, incidence of diastolic function, body weight, age, and E/A ratio.Descriptive statistics of categorical (sex and diastolic function) and raw continuous (BW, age, and E/A) variables are presented at the top and bottom of the table, respectively.Log-transformed values were used for all statistical analyses.

of females No of males χ2 p Value DF
28,29Because our final MLR model could only explain 32.05% of Log_E/A values, there are likely other covariables that could drive a better model for predicating E/A in rhesus macaques that were not included in this study, therefore, an even more robust dataset is warranted for future studies.We report that SPF status is not a significant predictor for measures of diastology, thus, it is unlikely that a role for infectious disease (such as viral myocarditis secondary to common pathogens) in DD of rhesus macaques exist; we confirm that an index of exercise activity (housing configuration) is a notable variable when assessing diastology in rhesus macaques and should be considered in future cardiovascular study designs as increased exercise in the outdoor housing configuration imparts a reduced hazard ratio for diastolic dysfunction.TA B L E 3 Differences in incidence of diastolic dysfunction between male and female rhesus macaques.The results from a 2 × 2 contingency chi-square test for incidence of DD between females and males in a total population of 594 clinically healthy rhesus macaques are depicted.VariableNoAbbreviations: DD, diastolic dysfunction; DF, degrees of freedom.
Stepwise and multiple linear regression analyses of Log_E/A values.The results from a stepwise and multiple linear regression analysis for predicting Log_E/A values are presented.
TA B L E 4Abbreviations: DF, degrees of freedom; HR, hazard ratio.