The data from the UK Biobank resource were used for the purpose of this study. The UK Biobank study design and population had previously been detailed (Sudlow et al., 2015). In brief, UK Biobank is a large community-based prospective study in the United Kingdom that recruited 502,299 participants aged 40-85 years old with the aim of improving prevention, diagnosis, and treatment of a plethora of illnesses. The study collected detailed phenotype and genotype data, including sociodemographic, lifestyle, clinical diagnosis, treatment genetic, imaging and physiological parameters. UK Biobank has approval from the institutional review boards, namely, the North West Multi-centre Research Ethics Committee for the United Kingdom, from the National Information Governance Board for Health and Social Care for England and Wales, and from the Community Health Index Advisory Group for Scotland¹. All participants gave informed consent for the study via a touch-screen interface that required agreement for all individual statements on the consent form as well as the participant’s signature on an electronic pad². The present study is restricted to a subsample of United Kingdom participants with information of ASI, blood pressure, diffusion tensor imaging (DTI) and white matter hyperintensities volumes (WMHV). Those with self-reported conditions listed in SM1 were excluded which led to a total of 20,742 participants included in the study. The research reported here was conducted using the UK Biobank Resource under Application Number 54531.

Prior to analysis, we tested the presence of outliers in each variable of interest and removed 2,758 participants in total ASI, WMHV, FA in CC, IC and CR data using the 1.5 interquartile rule. Therefore in this study, we only considered the remaining 17984 participants. The association between vascular risk factors (e.g., increased in peripheral PP or in ASI) and changes in the white matter and cognitive decline is complex, and largely mediated by mixed cerebrovascular and neurodegenerative lesions (Qiu and Fratiglioni, 2015; Wang et al., 2017). These associations have been found to be stronger when multiple vascular risk factors are present in mid-life (40–59 years of age), especially if left untreated (Launer et al., 1995; Kivipelto et al., 2001; Yaffe et al., 2014; Qiu and Fratiglioni, 2015). These associations are in contrast less certain when these vascular risk factors occur in later life (>75 years of age) (Qiu et al., 2005; Novak and Hajjar, 2010; Anstey et al., 2011; Tolppanen et al., 2012). For that reason, we divided our participants into 4 groups based on their age G1 (45–55 years), G2 (55–65 years), G3 (65–75 years), and G4 (≥75 years).

To test the hypothesis that ASI is a better predictor of white matter integrity than peripheral PP, a stepwise linear model was performed. For this analysis, each MRI metric was treated as a dependent variable and all the following variables were considered as independent variables: age, sex, peripheral PP, and ASI (Table 2). To better understand the relationship between vascular aging and white matter damage across the age span, we then looked at the relationship between either peripheral PP or ASI with our MRI markers of white matter integrity across different age brackets. Scatter plots between age and peripheral PP, ASI, FA in the CC, and WMHV, respectively, further help to interpret the results (SM2).

All statistical tests and figures except the stepwise linear regression were done with R version 3.5.2. The stepwise linear regression was done with SPSS (IBM SPSS 25, Statistics, Chicago, IL, United States).

To promote reproducibility, the data and code to perform all statistical analyses and figures are publicly available at https://github.com/atefbadji/ASI_PP_MRI_UKBiobank.

Several studies have previously compared different markers of arterial stiffness and vascular aging on the incidence of cardiovascular events, in particular cfPWV and PP. cfPWV has been established as the gold standard measure of arterial stiffness and a stronger predictor of cardiovascular events than peripheral PP, not only in older adults but also in individuals with different conditions such as renal failure, atrial fibrillation, and heart failure (Laurent et al., 2001; Rahman, 2017; Niiranen et al., 2019). Peripheral PP has also been associated with cardiovascular outcomes (Madhavan et al., 1994; Benetos et al., 1997; Mitchell et al., 1997), however, such observation has been attributed to the interplay between peripheral PP and arterial stiffness.

Beyond its effects on cardiovascular events, cfPWV has been also shown to better predict cognitive decline than measures of BP (Marfella and Paolisso, 2016). The emerging MRI literature has often highlighted a relationship between arterial stiffness, WMHs and white matter microstructure (Shrestha et al., 2009; Rosano et al., 2013; Singer et al., 2014; Badji et al., 2019). In particular, DTI studies looking at the effect of cfPWV on the white matter microstructure showed consistently that several white matter tracts are particularly vulnerable to increased arterial stiffness, among which is the CC, a key white matter tract altered in AD (Badji et al., 2019). Interestingly, cfPWV was recently found to better predict the microstructural integrity of the CC when compared with measures of 24 h SBP, 24 h PP and AIx in participants between 65 and 75 years old (Badji et al., 2020). However, to the best of our knowledge, no other study has yet compared different markers of arterial stiffness on brain integrity, nor has included measures of ASI before the present study.

In the emerging literature, the correlation between ASI and cfPWV has been found convincing (r = 0.65, p < 0.001) (Millasseau et al., 2002). The relationship between ASI, age and BP was even found similar to that between cfPWV, age and BP which further supported the concept that ASI and cfPWV are influenced by similar factors (Millasseau et al., 2002). Studies looking at the effect of ASI on brain integrity would therefore add value to the already published literature in terms of similitude and differences of such markers. This will in turn hopefully help clinicians in their choice of markers of vascular aging with respect to the outcome of interest.

Our study presents the first evidence that peripheral PP better predicts white matter integrity compared to ASI in participants of the UK Biobank younger than 75 years old. This finding implies that ASI from plethysmography may not be a reliable measure of vascular aging in older adults, and may therefore not be used to estimate cerebrovascular integrity in older adults.

A number of reasons can explain this result. The contour of the pulse waveform is essentially determined by characteristics of the heart such as ventricular ejection and those of large arteries which are altered with increasing age (Avolio et al., 1983; Millasseau et al., 2002; Alty et al., 2007; Russo et al., 2011). In older individuals, the systolic and diastolic peaks in the contour of the arterial waveform become difficult to identify (McVeigh et al., 1999; Alty et al., 2007). This was already demonstrated by McVeigh et al. (1999) who measured the arterial waveform shapes in a group of people of different ages. Increased arterial stiffness in older adults elevates arterial pressure wave propagation, causing the reflected wave to arrive back at the aorta during the systolic rather than the diastolic phase of the cardiac cycle, thereby escalating the SBP and contributing to a widening of PP (Laurent et al., 2005; Vlachopoulos et al., 2011; Pase et al., 2012). The time between the peaks of forward and reflected waves assessed using plethysmography appeared thereby reduced in older subjects (Millasseau et al., 2002). In addition, studies looking at age-related changes in shape characteristics in individual fingers and toes using frequency spectrum analysis (Oliva et al., 1976; Sherebrin and Sherebrin, 1990) showed that there is a general reduction in the harmonic components of the pulse in older subjects (Shirouzu et al., 1998) which makes it further difficult to estimate ASI using plethysmography technique in older adults.

As mentioned in the introduction, peripheral PP is determined by brachial SBP and DBP, and the relative contribution of these is a function of age. In young adults, both DBP and SBP increase, whereas in the elderly SBP increases and DBP reduces with age, which escalates PP (Wills et al., 2011). The association among higher SBP, mean BP, DBP, peripheral PP, and WMH is consistent in many previous studies during adulthood (Singer et al., 2014; Badji et al., 2019). Likewise, the association between high BP in middle age (age 40–64 years) and increased risk for vascular dementia is well established (Kennelly et al., 2009; Launer et al., 2010; Ninomiya et al., 2011). However, studies looking at these associations at an older age are sparse and there is no consensus on the association between BP and dementia in individuals aged 75 and above (Peng et al., 2017). In addition, information about the impact of BP and arterial stiffness measures on white matter integrity in people over 75 years old is still lacking.

Nevertheless, a recent study investigated the association between BP measures longitudinally, the internal carotid arteries (ICA) blood flow velocity parameters, and age-related WMH in a well-characterized 694 community-dwelling cohort of older adults over 70 years old (Aribisala et al., 2014). Results from this study showed that no association was found between peripheral PP and WMH measures (both WMH volume or Fazekas), agreeing well with the results of the present study. Looking at the results in detail, Aribisala et al. (2014) also showed that the size of WMH at the age of 73 years was weakly associated with mean BP and DBP at the age of 70 years old. Similar but even weaker associations were seen between WMH and BP at the age of 73 years (Aribisala et al., 2014). In addition, Peng et al. (2017) reported no association between BP levels nor peripheral PP at the age of 70–75 years and the incidence of vascular dementia.

Altogether, these results could be explained by different reasons. Indeed, both SBP and DBP can reduce at an older age. This phenomenon has a multifactorial etiology such as frailty, polymedication, improper/non-adjusted medication, heart failure, kidney failure, and so on and so forth. The integrity of the brain has a critical dependence on a nearly impeccable vascular supply, it is therefore not excluded that a decline in BP happens as a result of brain damage in the pathogenesis of dementia, just as much as the multifactorial-related decline in BP can further damage cerebrovascular integrity leading to dementia. Such a decline in BP could weaken the association between increased BP levels, white matter changes, and the incidence of vascular dementia later in life (Skoog et al., 1996). Although individuals with any dementia diagnosis or neurological condition were excluded from the present study, some of our participants over 75 years old may be in the early stage of dementia, before clinical symptoms appear. This could explain the significant association found between decreased peripheral PP and increased WMHs in individuals between 65 and 75 years, and the non-significant association between peripheral PP and either FA or WMHs in participants over 75 years old.

Another issue with BP measures in general in late life is the lack of information regarding the dose of exposure (and untreated exposure) in most studies, including this one. Those with a new-onset PP increase may not have an immediate risk of brain injury but those with prolonged PP elevation that continues into late life are likely at very high risk. Effects of treatment only compound these dose-dependent associations that are typically unmeasured.

The strength of this study lies in the use of a large community-dwelling cohort from the United Kingdom and well-validated imaging-derived data used across all researchers accessing the UK Biobank data. To the best of our knowledge this is the first study comparing different markers of arterial stiffness on brain integrity including ASI from plethysmography. Including comparisons of drug-naive participants and those on antihypertensive treatments would have been interesting. Unfortunately, none of the women included in the present study presented information about antihypertensive medication history, therefore we could not have included it. Likewise, excluding participants with heart conditions such as valve diseases, cardiac arrhythmias or heart failure and participants with renal or hepatic failure was not possible due to the lack of information in the UK Biobank database. Finally, although we did not find any relationship between either peripheral PP or ASI and white matter integrity after 75 years, we must acknowledge that we had less participants (n = 788) compared to other groups (n = 3,015–17,984). Fewer subjects may imply less power to detect such association. In this study, we looked for the first time at the relationship between peripheral PP, ASI and white matter microstructure after the age of 75 years. Although our finding may question the nature of the relationship between markers of vascular aging and white matter damage in individuals over 75 years old, it certainly needs to be replicated in a larger study before being able to draw an accurate conclusion.