One hundred and thirty male and female participants between the ages of 18 and 45 years were recruited from the McMaster University and Hamilton communities through posters, online advertisements, and word of mouth. Participants were recruited from July 2021 to February 2023. A sample size calculation estimated that 50 participants (25 males, 25 females) would be needed to find sex differences in scaled %FMD, based on previous work in our lab (14) (%FMD: male: 9.0 ± 2.6%, female: 6.5 ± 2.1%, α ≤ 0.05, power: 95%; independent t-test in G*Power). However, given that no study had previously examined gender expression in a young, healthy cohort to base power calculations on, and given that a 1% change in %FMD is clinically relevant (11) and the robust change in clinical risk observed previously (34), we considered a 2% scaled FMD difference between gender expression groups (masculine, feminine, androgynous, undifferentiated; Δ2% change between feminine and masculine groups) to be meaningful. We determined that 120 participants were required to detect a significant difference assuming an SD of 2%, an alpha level of 0.05, and a power of 90% based on a one-way analysis of variance (ANOVA) in G*Power. This study was approved by the Hamilton Integrated Research Ethics Board (#14884) and conforms to the standards set by the Declaration of Helsinki, apart from registration of this study in a database. Participants completed a medical screening form to determine eligibility and provided written informed consent before participating in the study.

Participants were included if they were between the ages of 18 and 45 years, self-reported as healthy, non-smoking, and were located near Hamilton, Ontario, Canada. Participants were excluded if they reported cardiovascular or metabolic diseases, were pregnant or within the last year, were taking vasoactive medication (e.g., beta-blockers, ACE inhibitors, diuretics), or were smokers. Participants who had previously participated in a study in our laboratory within the last year and agreed to be contacted for future studies were recruited (n = 149), of which 130 agreed to participate. Potential participants who had been recruited using existing trials were re-consented to have their data included for this specific research question. Most participants (n = 114) were recruited prospectively through five existing studies that had embedded the demographic questionnaire into their study design and used lab-wide standardized methods for the relevant outcomes. A small number of participants (n = 16) were recruited retrospectively after completion of a study and asked to complete the questionnaire through a virtual medium (Zoom, San Jose, CA, USA).

All study visits took place in the Vascular Dynamics Laboratory at McMaster University in a temperature- (23.3°C) and humidity-controlled (15%), quiet room. Prior to commencing the study, participants attended a familiarization visit to become familiar with the lab environment, assess eligibility, and complete consent forms and medical screening forms. Following recruitment, anthropometric information, including age, height, and weight to assess body mass index (kg/m²), was collected during the familiarization visit. Participants were also familiarized to the FMD test as described below.

Following the familiarization visit, participants took part in a vascular assessment session and a cardiorespiratory fitness exercise test. The cardiorespiratory fitness test was completed in close temporal proximity to the vascular assessment session, either in the week prior to the vascular assessment visit during the familiarization session (n = 64), immediately after on the same day (n = 46), or within 3 days following the vascular assessment visit (n = 20). Prior to the vascular assessment session, participants completed at least a 6 h overnight fast, 12 h without alcohol or caffeine, 24 h without moderate to vigorous physical activity, and 12 h without the use of prescription or non-prescription medications (i.e., anti-inflammatory and pain medications). Participants were tested during the morning hours to control for diurnal variation in endothelial function (40).

All statistical analyses were performed using SPSS (version 22, IBM, Chicago, IL, USA) and R Statistical Software (v.4.1.0, R Core Team, 2023, Vienna, Austria) for allometric scaling analysis. All data were reported using descriptive statistics, including means and standard deviations for normally distributed continuous variables, medians and interquartile ranges (IQRs) for non-normally distributed continuous variables, and frequencies (percentages) for categorical variables (i.e., race, ethnicity, sex category, gender identity category, gender expression). Statistical significance was set as p  <0.05.

Of the 130 participants recruited, 50 identified as male (38%) while 80 identified as female (62%). Gender identity groups included 49 men (38%), 79 women (61%), and 2 non-binary participants (2%). The proportion of participants categorized into gender expression groups is outlined in Figure 1. Ethnicity and race of the participants are detailed in full in Supplementary Table S2. Briefly, the ethnicity of the participants included in this study is as follows: African Origins (n = 5; 4%), Asian Origins (n = 54; 42%), European Origins (n = 38; 29%), Mixed Origins (n = 30; 23%), and North American Origins (n = 3; 2%). The race of the participants included in this study is as follows: Asian (n = 48; 37%), Black or African American (n = 2; 2%), Middle Eastern or North African (n = 10; 8%), Mixed (n = 8; 6%), Prefer to Self-Describe (n = 2; 2%), and White or Caucasian (n = 60; 46%).

While the average age of the participants across sex and gender identity category groups were not different, all other characteristics measured were higher in males compared to females, and men compared to women, including height, weight, BMI, and V̇O₂peak (all p < 0.01; Table 1). There was no difference across gender expression groups for any participant characteristics (Table 2).

BSRI-Feminine scores were higher on average in females compared to males [female: 5.4 ± 0.8, 95% confidence interval (CI): 5.2–5.8; male: 5.0 ± 0.9, 95% CI: 4.7–5.2; p < 0.01, r = 0.50], and also higher in women compared to men (women: 5.4 ± 0.8, 95% CI: 5.2–5.6; men: 5.0 ± 0.9, 95% CI: 4.7–5.2; p = 0.01, r = 0.47; non-binary: 5.4 and 3.7). In contrast, there was no difference for the BSRI-Masculine scores across sex categories (female: 4.4 ± 0.7, 95% CI: 4.2–4.5; male: 4.4 ± 0.7, 95% CI: 4.2–4.5, p = 0.85, r = 0.03) or gender identity category groups (women: 4.4 ± 0.7, 95% CI: 4.2–4.5l men: 4.4 ± 0.7, 95% CI: 4.2–4.6, p = 0.93, r = 0.02; non-binary: 4.4 and 3.8).

BSRI-Feminine scores were higher in feminine (5.9 ± 0.5) and androgynous (5.8 ± 0.5) gender expression groups, compared to masculine (4.5 ± 0.6) and undifferentiated (4.4 ± 0.5) gender expression groups (p < 0.01 for all, η² = 0.65), with no difference between the feminine and androgynous (p = 1.00) and the masculine and undifferentiated (p = 1.00) gender expression groups (Figure 1). BSRI-Masculine scores were lower in the feminine (3.8 ± 0.3) and undifferentiated (3.8 ± 0.4) groups compared to the masculine (5.1 ± 0.4) and androgynous (4.8 ± 0.4) gender expression groups (p < 0.01 for all, η² = 0.71), with no difference between the feminine and undifferentiated groups (p = 1.00; Figure 1). However, BSRI-Masculine was also higher in the masculine compared to the androgynous gender expression group (p = 0.03; Figure 1).

To assess the internal consistency of each subscale (feminine, masculine, neutral) of the BSRI, Cronbach's alphas were computed: BSRI-Feminine α = 0.87, BSRI-Masculine α = 0.75, and BSRI-Neutral α = 0.40; BSRI-Feminine and BSRI-Masculine are within acceptable scores for internal consistency (66–68).

Resting SBP and MAP were higher on average in males compared to females (d = 1.58 and 0.87 respectively; both p < 0.01; Table 1) and higher in men compared to women (d = 1.57 and 0.83 respectively; both p < 0.01; Table 1). There were no sex or gender identity category differences for DBP or HR (Table 1). There were no differences across gender expression groups for any hemodynamic measure (SBP, DBP, MAP, and HR; Table 2).

Central PWV was higher on average in males compared to females [6.4 (1.8) vs. 5.8 (2.2) m/s, p = 0.02; r = 0.20; Figure 2A], and higher in men compared to women [6.4 (1.9) vs. 5.8 (2.1) m/s, p = 0.02; r = 0.20; Figure 2D]. There were no differences in peripheral leg PWV (Figures 2B,E) or peripheral arm PWV (Figures 2C,F) between sex or gender identity category groups. Similarly, there were no differences across gender expression groups for any PWV measures (Figures 2G–I). None of these findings were altered when the independent association of MAP or V̇O₂peak was added as covariates.

Baseline brachial artery diameter and peak brachial artery diameter were both larger on average in males compared to females (p < 0.01 both; r = 0.63 and 0.65 respectively; Table 1) and men compared to women (p < 0.01 both; r = 0.63 and 0.65 respectively; Table 1). AbsFMD was not different between sex or gender identity category groups (Table 1). %FMD was initially higher on average in females compared to males (8.5 ± 3.7 vs. 6.9 ± 3.5%; p = 0.01, d = 0.45; Figure 3A) and in women compared to men (8.6 ± 3.7 vs. 7.0 ± 3.5%; p = 0.02, d = 0.44; Figure 3C); this result remained significant when V̇O₂peak or SRAUC were added as covariates in the model (both p = 0.02). After allometric scaling to consider differences in artery size, %FMD_scaled was higher on average in males compared to females (8.8 ± 3.3 vs. 7.2 ± 3.1%, p = 0.03, r = 0.42; Figure 3B) and in men compared to women (8.9 ± 3.3 vs. 7.2 ± 3.1, p = 0.02, r = 0.47; Figure 3D).

Baseline MBV was higher on average in males compared to females (p = 0.01, r = 0.22; Table 1) and men compared to women (p = 0.02, r = 0.22; Table 1). There were no sex or gender identity category group differences for baseline SR, SRAUC, or time to peak diameter (Table 1). There was no difference in any endothelial function or blood velocity outcome across gender expression groups (Table 2; Figure 3E); this result remained non-significant when %FMD was allometrically scaled (p = 0.39; Figure 3F) or when V̇O₂peak or SRAUC was considered as a covariate.

Due to a low sample size for gender-diverse participants, non-binary participants were qualitatively, instead of statistically, examined. Compared to men and women, non-binary participants may have higher BMI (25–30 and 35–40 kg/m²) and lower cardiorespiratory fitness (27.5 and 32.4 ml/kg/min; Table 1). Non-binary participants also may have elevated blood pressure (SBP: 110 and 138, MAP: 78 and 97 mmHg; Table 1) and resting heart rate (66 and 92 bpm; Table 1) compared to women, impaired endothelial function (AbsFMD: 0.18 and 0.09 mm; %FMD: 4.69% and 2.16%, Table 1 and Figures 3C,D), including after differences in artery size were considered using allometric scaling (%FMD_scaled: 4.7 ± 2.8%). However, non-binary participants had lower central (4.2 and 5.3 m/s) and peripheral arm PWV (4.8 and 4.5) but elevated peripheral leg PWV (9.2 and 9.3 m/s).

There was no significant relationship between gender expression scores (BSRI-Feminine, BSRI-Masculine) and endothelial function or central PWV (Supplementary Figures S1A–D). Similarly, there was no relationship between cardiorespiratory fitness and endothelial function (Supplementary Figure S1E) or central PWV (Supplementary Figure S1F). Finally, there was a negative relationship between baseline diameter and %FMD (r = −0.55, p < 0.001; Supplementary Figure S1G).

The present study found that males had elevated blood pressure (SBP and MAP) compared to females, with no difference in DBP or HR. Using an electronic tool (sexdifference.org) to estimate the overlap between the normal distributions in males and females, we also observed that the overlap between sexes was ∼46% for SBP and ∼68% for MAP (69). Taken alongside the large effect sizes for SBP (d = 1.58) and MAP (d = 0.87), sex differences in BP are evident. These findings are in line with most previous research on sex differences in hemodynamics, finding higher SBP, MAP, and occasionally DBP in males compared to females (20, 21, 36, 70–74). Research by Harris et al. found that males had higher SBP compared to females (70). Similarly, research in our lab also found elevated SBP and MAP in males compared to two groups of premenopausal females: natural cycling and combined oral contraceptive pill users (20) Finally, a recent narrative review discussed that resting blood pressure is elevated in males compared to females and that this may be a critical factor in the increased risk for CVD in males earlier in life than females (36). The mechanisms underlying elevated BP in males compared to females include sex hormones that may involve testosterone increasing blood pressure and 17β-estradiol decreasing blood pressure (and the ratio of testosterone/estradiol) (75–78), and anatomical differences in height resulting in increased pulse wave propagation that must be accompanied by a heightened SBP in males (18). While sex hormones were not measured in this study, males were taller than females, which could be in part responsible for these differences in blood pressure. In addition, sex and gender factors are challenging to separate, and gender-related lifestyle factors [i.e., smoking, alcohol consumption, sodium intake, sleep (79)] may also play a role influencing observed sex differences in blood pressure. However, the finding that BP is higher in males compared to females is not always true; BP is reported to be the same or elevated in females compared to males in populations with elevated CVD risk factors, including elevated BMI and low fitness levels (22, 80). Therefore, it is plausible that the “female advantage” observed with blood pressure previously may be outweighed by the influence of additional risk factors for CVD, such as obesity and low fitness.

In the present study, males had elevated central PWV, but not peripheral arm or leg PWV compared to females, which was not explained by controlling for blood pressure elevations in males. Again, using an electronic tool detailed above (69), the overlap between sexes for central PWV was ∼86% with a small effect size (r = 0.20), suggesting that while sex differences in central PWV were significant, the magnitude of difference was more subtle than BP. This is in line with some (20–22, 36, 71), but not all (73), studies that have observed sex differences in local (carotid artery) and/or systemic arterial stiffness. For example, research by Baldo et al. found that central PWV was higher in males compared to females, across the aging lifespan (21). The same was true in a population of patients with pre-hypertension to stage 1 hypertension, with males having higher central PWV than females, despite females having a higher SBP compared to males (22). Marlatt et al. identified that sex differences in carotid artery compliance became present in early adulthood (∼late 30s), but not in childhood (71). This study, alongside previous research, points to the role of sex hormones like 17β-estradiol to play a role in these apparent sex differences and/or the influence of sex differences in growth patterns between males and females (17).

Similarly, recent work found that β-stiffness of the carotid artery was higher in males compared to females who were naturally cycling or used oral contraceptive pills (20). Despite finding a sex difference in local arterial stiffness in the carotid artery, no differences were observed in central PWV, although it is possible that they were underpowered to detect a sex difference in central PWV (20). Our study, with 130 participants found a significant sex difference in central PWV between males and females (p = 0.02), but arguably the ∼0.6 m/s difference may not be clinically significant. Prior research has determined that a 1 m/s increase in arterial stiffness is associated with a 15% (95% CI: 9%–21%) increased risk for CVD mortality (81). While a 0.6 m/s difference in central PWV may still increase the risk for CVD marginally, it is unlikely to be solely responsible for differential rates in CVD in males and females. Furthermore, the small effect size and substantial overlap in normal distribution curves between sexes suggests that sex differences in central PWV are subtle.

We found that unscaled %FMD was elevated in female compared to male participants but %FMD may have been artificially inflated given that baseline brachial artery diameter was also smaller in females. After performing allometric scaling analysis to account for differences in baseline artery diameter, we found that scaled %FMD was elevated in males compared to females. Using an electronic tool detailed above (69), the overlap between sexes for %FMD_scaled was ∼80% with a medium effect size (d = 0.45), suggesting that while sex differences in %FMD_scaled were significant the magnitude of difference between sexes was only moderately pronounced. These findings are aligned with most prior %FMD (unscaled) research finding greater %FMD in females (12, 14, 70, 82, 83), and prior research from our lab finding %FMD (scaled) is greater in males (14).

The present study also aligns with a consistent finding that male arteries are, on average, larger than female arteries (12, 70, 80, 82–84). Early research identified sex differences in reductions in %FMD with aging, alongside sex differences in %FMD and in artery diameter (85). Furthermore, this study reported a negative relationship between %FMD and resting arterial diameter, which has been extensively replicated in healthy and clinical populations (12, 83, 85). Recent research by Holder et al. observed marked sex differences in %FMD (higher in females), developing reference ranges in a large population of both healthy and clinical participants (12), following FMD guidelines (12, 56). We observed similar %FMD values compared to the reference intervals (12): present study data in the ∼50th percentile of reported references: male baseline diameter: 4.09 ± 0.49 mm; male %FMD: 6.86 ± 3.50%; female baseline diameter: 3.30 ± 0.46 mm, female %FMD: 8.51 ± 3.70%.

Age and sex differences in %FMD relate to differences in baseline diameter and further allude to structural influences on artery function between sexes (12). One evident reason for sex differences in artery diameter is a positive relationship between baseline diameter and height. Recent work by this same group also found that a “…10 cm increase in height is associated with a 0.16 mm increase in baseline diameter and a 0.28% decrease in FMD…” and this finding was independent of sex (12). In our study, there was a 14 ± 7 cm difference in height between males and females, which would have been attributed to an ∼0.23 mm increase in baseline diameter and an ∼0.39% decrease in %FMD. While sex differences in height do not fully explain the sex differences in baseline diameter and %FMD in the present study, artery size differences cannot be ignored in examining sex differences in endothelial function through allometric scaling of %FMD to baseline diameter. Overall, while unscaled %FMD may initially suggest that females have improved endothelial function compared to males, accounting for artery size differences result in males having elevated %FMD (scaled) compared to females. Researchers should consider using allometric scaling when comparing between sexes to consider baseline differences in artery size and ensure valid interpretation of sex difference findings.

While there is evidence of alterations in CVD and CVD risk factors associated with gender and gender-related factors, the present study did not observe any variation in novel and traditional CVD risk factors across gender expression groups. Prior research has observed the association between gender or gender-related factors and CVD; for example, research by Pelletier et al. found that recurrent acute coronary syndrome was associated with “femininity” as a composite score of gender roles and expression (assessed by the BSRI) in middle-aged (aged ∼48 years) individuals, after adjusting for sex (34). Similarly, previous research has found that gender-related roles including caregiver burden (29), role strain (e.g., workplace-home life role stress), and other psychosocial stressors are predominant in women compared to men and attributed to increased risk factors and development of CVD (30–32). The present study did not see impaired cardiovascular health (i.e., blunted %FMD or increased PWV) in young individuals (aged ∼22 years) with higher femininity (gender expression) scores; however, it is possible that gender expression and other gender-related factors such as gender roles may not manifest until later in life. For example, caregiving burden may not be present until middle- and older-adulthood where parenting roles and care for aging relatives commonly occurs (86); therefore, any influence of these gender-related factors may not have yet progressed to the stage of negative remodeling in the vasculature. These factors may also intersect with known increases in CVD risk associated with age (9), though further research is necessary. It is also possible that gender roles have a stronger relationship with CVD health outcomes and that gender expression is less associated; further research investigating these constructs is needed in young adults. In addition, given the limitations of the BSRI in only examining gender expression, further examining of gender roles in professional and home life, along with exposure to gendered expression and roles in family and friend circles, may have added further depth to this analysis. A newly developed questionnaire by Nielsen et al., examining seven gender-related variables across domains of gender norms, gender-related traits (gender expression), and gender relations in American undergraduate students and younger adults, may be useful in extending this research (87).

In examining gender identity category groups, we observed the same sex differences detailed above in gender identity groups of men and women. The finding that ∼99% of participant sex aligned with their gender identity (male = man, female = woman) substantiates this overlap. Considering that sex and gender identity category are highly interrelated, the same conclusions detailed for sex may also be attributed to gender identity differences; furthermore, there may be gender identity influences on seemingly sex differences. As a result, separating categorical sex assigned at birth and biological factors from gender identity and sociocultural factors is not possible given this overlap in this study. However, while sex differences and gender identity differences aligned in this study, these are two different identity constructs and should be represented separately, especially to allow for the representation of gender-diverse participants.

In this study, we recruited a small number of gender-diverse individuals (n = 2 non-binary; comprising ∼1.5% of the total study population). While this is not representative of the diverse groups of gender queer participants that could have been recruited (i.e., transgender, genderqueer, two-spirit, etc.), this number of participants is proportionally representative of the number of gender-diverse individuals on average. For example, in Canada, 0.13% of the population report being non-binary (88), though this may be an underestimation. When considering gender-diverse individuals in analyzing for differences in cardiovascular outcomes across gender identity groups, we removed non-binary individuals from the analysis due to this low sample size and instead reported the data qualitatively in hopes of stimulating further needed research in this population. Qualitatively, it appears that the two non-binary participants in this study may have some elevated risk factors for CVD, including elevated blood pressure (above 90th percentile for SBP in the present study), impaired endothelial function (below 15th percentile for %FMD in the present study), and elevated leg PWV (above 75th percentile in the present study), though lower central and arm PWV. This could be in part due to higher BMI and lower cardiorespiratory fitness in these participants. While these findings are limited and must be explored further with a larger sample size, they do align with current literature indicating increased CVD risk factors in gender-diverse populations (89, 90), in part attributed to gender minority stress (89, 91) and increased allostatic load, or the accumulation of stress and life events, in some gender and sexual minority groups (92).

Overall, while gender expression differences were not observed in this study, this does not mean that differences do not exist or should not be studied by researchers. On the contrary, given that this study was in a young, healthy population, we may have yet to observe CVD risk factor elevations or disease manifestation later in the time course of CVD development. Similarly, it is possible that cardiovascular risk may only be apparent during periods in which gender expression or roles are challenged. For example, research by Kramer et al. found that men presented with low masculinity feedback experienced an exaggerated vagal withdrawal response during a speech task compared to those who received higher feedback (93). Overall, it is critical for researchers to continue to examine gender identity, expression, and gender-related factors, alongside sex and sex-related factors, and how they influence novel and traditional CVD risk factors.

While this study had several strengths, including its inclusion of several novel and traditional CVD risk factor outcomes, recruitment of participants with a wide range of cardiorespiratory fitness levels, compilation of outcomes collected using the same standardized methodologies, and a reasonably large sample size, there are some limitations to consider. First, the findings of this study are only generalizable to young, healthy adults who were primarily university students, and where the majority had alignment between their sex and gender identity. As a result, further research is necessary, particularly in middle-aged and older adults when gender expression changes alongside critical gender milestones and other contextual factors (94–96). Similarly, though representative of the proportions of gender-diverse populations in Canada, the number of non-binary participants was too low to make conclusions; further research is needed in this population adequately powered to draw conclusions about CV health. Second, the gender assessment in this study was limited to gender identity and expression, measured by the BSRI. While other gender assessment tools exist (48, 87), their utility in a university population is challenged as many questions ask about gender-related roles in familial structures, financial status in a family or income, caregiver strain, workplace role and environment, among others, many of which are less applicable in a university student context. The BSRI, in contrast, was created using data from a university population and has applicability in this context (45), albeit a limited assessment of gender expression. However, the BSRI also may use some outdated gender stereotypical traits to explore gender expression, as it was constructed in the 1970s and gender norms have shifted since (28, 97–99); though at the time of the study's design, it was the only validated and widely used gender scoring system available recommended by experts (25). Further research is necessary to create additional relevant gender assessment methods in young adults. Another limitation of the questionnaire was that a small number of participants (n = 11) required follow-up prompting to complete one question on the BSRI-30; this is unlikely to alter the results of the study given the lower number of participants (∼8%) requiring follow-up and the high internal consistency scores of the BSRI-30 questionnaire. Finally, the type of contraceptive or hormonal cycle phase was not controlled for in this study. Previous research has found conflicting results on the influence of contraceptives and the hormonal cycle on endothelial function and arterial stiffness (14, 52, 70, 100–103). However, recent work from our lab and others have observed no impact of contraceptives, contraceptive cycle or menstrual cycle on arterial stiffness (20), and a small influence of the menstrual cycle on endothelial function (104). Therefore, any effects of contraceptive type or hormonal cycle phase are unlikely to have changed the findings from the current study.

The present study found that sex and gender identity category, but not gender expression groups, influenced novel and traditional risk factors for CVD in a young, healthy cisgender population. Specifically, blood pressure (SBP, MAP) and central PWV were elevated in males compared to females, but %FMD, once larger artery diameter in males was controlled for, was improved in males compared to females. While young otherwise healthy males appear to have elevated measures of central stiffness, this may be compensated by elevated vasodilatory capacity of a major conduit artery (brachial artery). Given that this study is only generalizable to a young, healthy population of primarily university students of Asian and European/Caucasian racial and ethnic origins, further research is necessary to examine sex and gender considerations in other more ethnically diverse and representative groups of young adults, older adults, and those with CVD. Similarly, further research on populations of gender-diverse adults, including non-binary and transgender populations is warranted. Finally, further research is also needed examining participants in conditions where gender expression is challenged (i.e., masculinity stressors) or in populations that may experience gender-related stress (i.e., non-binary and transgender participants) and their influence on novel and traditional risk factors for CVD.