Effects of hypertension and aging on brain function in spontaneously hypertensive rats: a longitudinal resting-state functional magnetic resonance imaging study

Abstract

To investigate the dynamic evolution of brain function under the comorbidities of hypertension and aging. Resting-state functional magnetic resonance imaging scans were longitudinally acquired at 10, 24, and 52 weeks in spontaneously hypertensive rats (SHRs) and Wistar-Kyoto rats. We computed the mean amplitude of low-frequency fluctuation (mALFF), mean regional homogeneity (mReHo), and functional connectivity (FC). There was no interaction between hypertension and aging on brain function. The main effect of aging reflects primarily the cumulative increase of brain activity, especially the increase of mALFF in amygdala and mReHo in cingulate cortex, accompanied by the decrease of brain activity. The main effect of hypertension reflects primarily decreased brain activity in default modal network, accompanied by increased brain activity. The main effect of aging shows reduced brain FC as early as 24 weeks, and the main effect of hypertension shows higher brain FC in SHRs. The novel discovery is that 1 brain FC network increased linearly with age in SHRs, in addition to the linearly decreasing FC. Hypertension and aging independently contribute to spatiotemporal alterations in brain function in SHRs following ongoing progression and compensation. This study provides new insight into the dynamic characteristics of brain function.

Introduction

The prevalence of hypertension is increasing globally due to population aging and unhealthy lifestyles. Hypertension and aging associate with similar influences on cerebrovascular and neurovascular insults, and both are considered primary risk factors for cognitive impairment (Hay et al. 2020). There appear to be strong interrelationship between hypertension, aging, and brain damage, whereas, hypertension is not a necessary consequence of biological aging. The combined effect of hypertension and aging on brain function is therefore of particular interest. Resting-state functional magnetic resonance imaging (rs-fMRI) provides an experimental approach to exploring intrinsic brain activity without any stimuli or tasks (Raichle 2015). In rs-fMRI studies, the mean amplitude of low-frequency fluctuation (mALFF) measures the intensity of spontaneous brain activity (Zang et al. 2007), mean regional homogeneity (mReHo) reflects the synchronous activities in a particular region (Zang et al. 2004), and functional connectivity (FC) refers to the network connections contribution to brain function. Briefly, mALFF and mReHo could identify the regions of brain activity changes due to hypertension and aging, and region of interest (ROI) based FC could reflect communication for information between these brain regions. Previous studies have shown that hypertension is associated with abnormal brain activity and FC alterations (Carnevale et al. 2020; Zhuang et al. 2020). It should not be ignored that brain function also changes with aging (Hu et al. 2014; Malagurski et al. 2022). Therefore, the combined effects of hypertension and aging on brain function require further investigation (Jennings et al. 2020). By characterizing the spatiotemporal dynamics of brain function, rs-fMRI can provide a wealth of information for understanding the progression of hypertension over aging.

The spontaneously hypertensive rats (SHRs) are a valuable model of essential hypertension, which are extensively investigated for evaluating hypertensive brain damage and its treatment. SHRs are normotensive at birth and progress to sustained hypertension gradually by 24 weeks of age. Moreover, brain atrophy, loss of nerve cells, and reactive gliosis in SHRs resembled those in the human hypertensive brain (Li et al. 2021). It has been reported that SHRs show cognitive impairment with age (Gao et al. 2019). An observational magnetic resonance imaging (MRI) experiment revealed that neurodegeneration was associated with age-related brain damage in SHRs (Zhao et al. 2016). Pathological studies have indicated that both hypertension and aging enhanced hippocampal damage in SHRs (Li et al. 2016). Ischemia–reperfusion experimental research confirmed that acute brain injury in aged SHRs was increased compared with that in their young counterparts (Chan et al. 2018). These previous studies provide clues to aid in the detection of cerebral aging progression in SHRs. Rs-fMRI is not only noninvasive, but also requires no cooperation from the subject, which has inspired considerable enthusiasm for exploring the brain function in elderly people with hypertension (Chen et al. 2018; Carnevale et al. 2020; Shah et al. 2021). These clinical studies provide significant findings, but are susceptible to interference from antihypertensive therapy. To avoid this confounding, SHRs are attractive. Moreover, it is possible to apply longitudinal rs-fMRI to characterize the dynamic changes of brain function in SHRs during aging.

The aim of our study was threefold: (i) to determine whether the effects of hypertension and aging on brain function are interactive; (ii) to investigate the spatiotemporal dynamics of brain activity over aging in SHRs and WKY rats; and (iii) to identify the brain FC impairment under the comorbidities of hypertension and aging. We designed a prospective longitudinal experiment to examine the differences in brain function between SHR and WKY rats during aging using rs-fMRI. It might be able to predict the pace of cognition impairment progression by detecting the combined contribution of hypertension and aging to brain function. Furthermore, the neuroimaging technique that provides potential biomarkers of brain function in SHRs can provide new insights into the pathogenesis underpinning hypertension.

Material and methods

Animals

Thirteen male SHRs and 10 Wistar-Kyoto (WKY) rats aged 8 weeks were purchased. One WKY rat was excluded, since it died at 34 weeks of age. All rats were housed in an air-conditioned room (constant temperature of 22–24°C, relative humidity 50–60%) under a 12-h light/dark cycle. They were maintained on a standard pellet diet, and tap water was available ad libitum. Blood pressure was measured at 20 weeks of age, and body weight was recorded weekly (Fig. S1, see online supplementary material for a color version of this figure).

MRI scanning protocol

All rats were scanned longitudinally at 10, 24, and 52 weeks. They were initially anesthetized with 3% isoflurane and then intramuscularly injected with 0.015 mg/kg dexmedetomidine. Rats inhaled a mixture of isoflurane and pure oxygen during MRI acquisition. The isoflurane concentration was adjusted between 0.6% and 1.2% to maintain the breathing rate at 50–60 breaths/min during T2-weighted imaging (T2WI) and then reduced to 0.2–0.8% to keep the breathing rate close to 70 breaths/min before rs-fMRI acquisition. A pulse oximeter was used to ensure oxygen saturation was above 95%. Body temperature was maintained at 37°C using a water circulation heating system. This experiment was approved by the Experimental Animal Ethics Committee.

Whole-brain T2WI and rs-fMRI scans were performed on a 7.0 T Bruker scanner (Pharma Scan 70/16 US) using a surface array coil (Fig. S2, see online supplementary material for a color version of this figure). T2WI were acquired using a rapid acquisition with relaxation enhancement (RARE) sequence. Scan parameters: time repetition, TR = 10,700 ms, effective time echo, TE = 36 ms, RARE factor = 8, field of view = 35 × 35 mm², matrix size = 256 × 256, slice thickness = 0.3 mm. Rs-fMRI were acquired using a single-shot gradient echo echo-planar imaging sequence. Scan parameters: TR = 2,000 ms, TE = 10 ms, flip angle = 90°, field of view = 25 × 20 mm², matrix size = 80 × 64, slice thickness = 0.8 mm, number of repetitions = 200.

Data preprocessing

Data preprocessing was performed by the SPM12 toolbox within the MATLAB2013b (The MathWorks Inc., MA, United States). (i) After the first 10 images of functional scans were discarded, the voxel size of all images was multiplied by a factor of 10 to approximate the size of human brain (Diaz-Parra et al. 2017). (ii) Functional images were slice-timing corrected and then spatially realigned with a 6 parameter rigid body transformation. All rats had <1.0 mm of translation in the x, y, and z axes, and <1.0° rotation in 3 axes. (iii) The rs-fMRI data were spatially normalized into a customized SHRs template space using individual T2WI as intermediaries (Yang et al. 2022; Fig. S3, see online supplementary material for a color version of this figure). (iv) Spatial smoothing was performed with a 6-mm full-width half-maximum Gaussian kernel. (v) After removing linear trend and regressing covariates of the 6 motion parameters, extract the low-frequency from the time curve and calculate the value of ALFF. The ALFF of each voxel was divided by the global mean ALFF, termed the standardized mALFF map. (vi) Based on the step (iii), the normalized images were subjected to linear detrending, motion parameter regression, and 0.01–0.08 Hz band-pass filtering. After calculating the mReHo maps, the smoothed mReHo maps were obtained by spatial smoothing. (vii) For ROI-based FC analyses, the normalized images were linearly detrended, and nuisance covariates of the 6 motion parameters and the mean signal of white matter and cerebral spinal fluid were regressed out. Finally, band-pass filtered between 0.01 and 0.08 Hz.

Local characteristics of brain activity

A flexible factorial design was constructed within SPM12 for the interaction between hypertension and aging as well as for the main effect of aging and hypertension. As no interaction was found in either mALFF or mReHo, the main effects of aging and hypertension with the post hoc test are reported. We performed cluster-level familywise error (P < 0.05) with a cluster-extent threshold of 20 voxels for multiple comparisons correction.

ROI-based FC

Pearson correlation and Fisher's z-transformation were performed using DPARSF V4.4 (data processing assistant for resting-state fMRI). Based on our results of mALFF and mReHo, as well as prior brain regions in the literature (Wiesmann et al. 2017; Carnevale et al. 2020), we selected 20 ROIs, including 12 cortical and 8 subcortical regions (Table S1). Group comparisons of FC matrices were performed using network-based statistics (NBS) with a statistic threshold of 2.1 and P < 0.05 by 5,000 permutations (Zalesky et al. 2010). In detail, 2-way repeated measures analysis of variance was applied to detect the interaction of hypertension and aging as well as the main effect of aging. A 2-sample T-test was used to analyze the main effect of hypertension after averaging the FC values across the 3 time points. One-way repeated measures analysis of variance was used to detect the age-dependent changes in FC in each strain, and the 2-sample T-test was used to assess the difference in FC between SHRs and WKY rats at each time point. Notably, if significant changes in FC were found, the average FC strength within these regions were post hoc analyzed. Furthermore, we explored the linear tendency of brain FC changes with aging using NBS.

Results

Local characteristics of brain activity

There was no interaction between hypertension and aging on mALFF or mReHo. Hypertension and aging have a significant main effect on mALFF and mReHo (Fig. S4, see online supplementary material for a color version of this figure). Figure 1 shows the post-hoc analysis of the main effect of aging on mALFF and mReHo. Both mALFF and mReHo outcomes showed age-dependent differences at 3 time points, with the most significant differences between 10 and 52 weeks. The colored regions indicate the T values of comparisons between any 2 time points. The blue–green regions indicate decreased brain activity, and the red–yellow regions indicate increased brain activity (Tables S2 and S3). Figure 2 shows the post-hoc analysis of the main effect of hypertension on mALFF and mReHo. Compared with the WKY rats, both mALFF and mReHo values of some regions were lower in SHRs, and the values of other regions were higher in SHRs (Tables S4 and S5).

Fig. 1.

Age-dependent difference distribution map of the mALFF and mReHo values. There was no interaction between hypertension and aging on mALFF or mReHo. The post-hoc analysis of the main effect of aging showed changes in mALFF A–C) and mReHo D–F) between 10 and 24 weeks, 24 and 52 weeks, and 10 and 52 weeks. Both mALFF and mReHo outcomes show significant age-dependent differences, with the most significant differences between 10 and 52 weeks of age. Significantly decreased brain activities are highlighted in the blue–green heat bar scales; significantly increased brain activities are highlighted in the red–yellow heat bar scales. The left side of the images corresponds to the left side of the brain, and vice versa.

Fig. 2.

A significant main effect of hypertension on mALFF A) and mReHo B) with post hoc comparisons. The blue–green color scale indicates the T values of significantly decreased mALFF and mReHo in SHRs compared with the WKY rats, whereas the red–yellow color scale indicates the T values of significantly increased mALFF and mReHo in SHRs compared with the WKY rats.

ROI-based FC

Figure 3 shows the results of ROI-based FC analysis. The FC matrix for both SHRs and WKY rats at different ages (A) and significantly different FCs are plotted across ages (B) and strains (C). There were age-related FC declines among some nodes in both strains, with post-hoc tests showing that the FC values were weaker at 52 weeks than at 10 weeks. No interaction effect of hypertension and aging was found, and a significant main effect of aging and hypertension is shown (D). The main effect of aging shows a significant decline in FC as early as 24 weeks. The main effect of hypertension shows stronger FC in SHRs compared with WKY rats. The FC difference between SHRs and WKY rats was most significant at 10 weeks of age. Figure 4 indicates the age-dependent linear changes in interregional FC. Two networks of brain FC were decreased with age in SHRs and 1 in WKY rats, yet the nodes of the decreased brain networks in the 2 strains were similar, including bilateral olfactory bulbs, amygdala, motor cortex, right auditory cortex, and visual cortex. We also found 1 network that increased linearly with age in SHRs. Table S6 details the entire FC analysis proposal.

Fig. 3.

The composite FC matrix for both SHRs and WKY rats across different ages. The matrix indicates the value of Fisher's z-transformed correlation coefficients, with red–yellow indicating high positive correlations, green–blue indicating low correlations, and dark blue–purple indicating inverse correlations A). The matrix illustrates the F values in regions with significant changes in FC across ages, and the corresponding error bar chart represents the mean FC values in these regions B). The mean FC values among these nodes were lower at 52 weeks than at 10 weeks in both strains. The matrix illustrates the T values in regions with significant intergroup differences, and the difference is significant only at 10 weeks C). No significant interaction effect of hypertension and aging was found D). The matrix represents the F value in regions with a significant main effect of aging (down). The corresponding error bar chart shows that the main effect of aging on FC significantly declined as early as 24 weeks. The matrix represents the T value in regions with a significant main effect of hypertension (up), which shows stronger FC among some nodes in SHRs than in WKY rats.

Fig. 4.

The age-dependent linear changes in interregional FC. The edges between nodes represent age-dependent linear alterations in the brain FC network. Two networks of brain FC decreased linearly with age A,B), whereas one increased linearly with age in SHRs C). Only 1 network of brain FC decreased linearly with age in WKY rats D). The nodes of the decreased brain networks were similar between the 2 strains.

Discussion

Our study explored longitudinal changes in brain function during aging in SHRs. Our results confirmed no interaction between hypertension and aging on brain function. The current study revealed that the spatiotemporal characteristics of brain function show the ongoing progression and compensation as SHRs age.

Local characteristics of brain activity

No interaction effects of hypertension and aging on mALFF or mReHo were found in our study. In this cohort of adult to aged rats, hypertension did not accelerate the aging process of brain function. The joint associations of hypertension and aging with brain function remain unclear (Jennings et al. 2020). Previous studies have shown that the effects of hypertension on cerebral blood volume were independent of aging in rhesus monkeys, which supports our finding from another perspective (Farris et al. 2021). As shown by an earlier investigation, hypertension accelerates gray matter atrophy during aging in SHRs (Yang et al. 2021), and a clinical study observed that hypertension accelerates white matter integrity decline in association fibers (Sabisz et al. 2019). Perhaps there are different brain aging patterns between the structure and function, and hypertension does not seem to make SHRs older in terms of brain function. It should be noted that the situation may become more complicated when other comorbidities are present. Previous study reported hypertension accelerates the impairments of cerebral structure and function in aged Alzheimer's disease mice (Wiesmann et al. 2017).

The main effect of aging on mALFF and mReHo reflects the chronic and cumulative burden. As rats aged, brain activity reflects mainly the cumulative increase, especially the increase of mALFF in amygdala and mReHo in cingulate cortex, accompanied by the decrease of brain activity including cerebellum. The increase in mReHo was distributed extensively. The relationship between aging and ALFF has been poorly studied in animals, and clinical studies are still unclear. Although it have been reported age-related increases in ALFF in thalamic and language regions (Mather and Nga 2013; Zhang et al. 2021), others have reported age-related decreases in ALFF in precuneus, which is recognized as a major hub of the default-mode network (DMN; Yin et al. 2015). These previous literatures have relied mainly on cross-sectional differences between older and younger population. We investigated the association between mALFF and aging longitudinally across the lifespan from adult to aged rats. Furthermore, our findings reveal this complex age association, including both increasing and decreasing mALFF with age. A study of healthy aging population indicated that the spatial distribution of ReHo is a predictor of brain glucose uptake, and ReHo decreased with age, particularly in the DMN and frontal areas (Bernier et al. 2017). In addition to age-related reduction of mReHo, our findings demonstrate that mReHo is increased in aged rats, especially in the DMN. Longitudinal rs-fMRI allows us to visualize the dynamic evolution of brain activity in aging rats. Nevertheless, the brain activity measured by mALFF and mReHo complicate their biological interpretation. The possible underlying explanation lies in the disruption of blood–brain barrier and neurovascular unit injury, affecting neural homeostasis (Presa et al. 2020; Li et al. 2021).

The main effect of hypertension on mALFF and mReHo shows disequilibrium in spontaneous brain activity in SHRs. The mALFF values of the cingulate cortex, motor cortex, visual cortex, and olfactory bulb in SHRs were lower than those in WKY rats, and the mALFF values of the cerebellum in SHRs were higher than those in WKY rats. The size of the brain regions with mReHo differences between the 2 strains was more extensive. Compared with WKY rats, the mReHo values of the cingulate cortex, motor cortex, hippocampus, and periaqueductal gray in SHRs were lower, and the mReHo values of the retrosplenial cortex, cerebellum, amygdala, and olfactory bulb in SHRs were higher. The changes of brain activity in SHRs are mostly overlapped with the DMN, and the most significant reduction was in the cingulate cortex. Clinical studies have demonstrated that metabolic syndrome, including hypertension, is associated with disrupted DMN FC (Rashid et al. 2019). We speculate that changes in the intensity and synchrony of brain activity might contribute to impaired information communication in DMN. It has been documented that the aggregation of vascular risk factors, including hypertension, can modulate ALFF (Zhuang et al. 2020). Similar to our outcomes, they also found increasing ALFF values in the cerebellum and inferred the cerebellum may be activated to compensate for the functional deficits of the cerebral cortex. The increase and decrease of brain activity in SHRs may indicate that hypertension impaired brain activity and initiated a compensatory mechanism. Our study may account for the global feature of brain function, where hypertension impairs brain activity while other regions compensate for the deficit.

Functional connectivity

Assessing the dynamic evolution of brain FC in SHRs during aging can help disentangle the relationship of hypertension, aging, and brain function. The effects of hypertension and aging on brain FC may not be interactive but rather independent in SHRs. When analyzing the main effect of aging, we detected reduced brain FC as early as 24 weeks. When analyzing the main effect of hypertension, we found that brain FC in SHRs was higher than that in WKY rats. Our findings are similar to those of a study of functional reorganization at the early stages of hypertension in middle-aged patients (Naumczyk et al. 2017). In the early course of hypertension, FC changes and functional reorganization may occur without direct macrostructural damage, which might suggest a time window for therapy. Therefore, we speculate that early intervention for hypertension is critical. We found explicit FC reorganization but did not explore the mechanism of these changes. One previous study confirmed neurovascular dysfunction using MRI and immunohistology in SHRs (Li et al. 2021). Moreover, they suggested that multiparametric MRI might be useful to characterize disease pathogenesis.

We evaluated the changes in brain FC with aging in both strains separately and found a gradually decreasing FC. After assuming that brain FC changes linearly with age, we identified 2 brain FC networks in SHRs and 1 in WKY rats that linearly decreased during aging. Of note, the brain regions involved in the decreased FC networks were similar between the 2 strains, including the olfactory bulb, amygdala, motor cortex, auditory cortex, and visual cortex. Another novel discovery is the existence of a linearly increasing FC network during aging in SHRs. Our results confirmed that hypertension affects brain FC, leading to ongoing changes over aging. Imbalance alterations of brain FC have also been reported in the previous literature on hypertension (Son et al. 2015; Gu et al. 2019). Chen et al. have propose a dynamic compensatory neural processes that fluctuated along with variations of vascular risk factors loading (Chen et al. 2018). As SHRs age, the changes in brain FC suggest the possibility that hypertension impairs brain FC, initiating a compensatory process. A previous investigation reported alterations in network activity and connectivity when perturbing 1 node in the DMN (Tu et al. 2021). Our findings and the previous literature support that the brain FC network might be a well-organized integrated network with coordinated activity.

Rs-fMRI allows us to visualize the brain FC network dynamically. Aedo-Jury et al demonstrated that changes in brain FC are directly related to the generation of slow waves (Aedo-Jury et al. 2020). Decreased brain FC and effective connectivity have been revealed previously in hypertensive patients using wavelet phase coherence and near-infrared spectroscopy techniques (Bu et al. 2018). Of note, the relationship between hypertension and brain FC remains to be determined when transferring the changes in FC from rats to patients. Clinical studies have documented that hypertension is associated with impaired brain FC in the DMN, dorsal attention network, and frontal parietal executive system (Li et al. 2017; Carnevale et al. 2020; Shah et al. 2021). These inconsistent regions may arise from differences in the duration and severity of hypertension as well as measurement techniques. In addition, neuroanatomical differences between animals and humans are not negligible (Schaeffer et al. 2020). Nonetheless, rs-fMRI studies in SHRs can provide an initial examination to identify spatiotemporal characteristics of brain FC associated with hypertension and aging.

Methodological considerations

The main innovation of our study is the longitudinal design with multiple time points. A major benefit of longitudinal study is the evaluation of the causation of the variables. Longitudinal follow-up can potentially improve the sensitivity and accuracy of the rs-fMRI parameters by reducing the confounding variabilities between subjects. Continuous dynamics of brain function over the whole disease course will help disentangle the relationship of hypertension, aging, and brain function. Although statistical power can be improved, one of the drawbacks of our longitudinal design is the missing data due to premature deaths of elderly rats. Aging rats become fragile, and the risk of heart failure increases with aging in SHRs. It has been reported that cardiac damage is associated with lower cognitive performance independent of brain damage in hypertensive patients (Uiterwijk et al. 2018). The follow-up time point of 52 weeks may not have been late enough to initiate more pronounced brain dysfunction. It is worth mentioning that the SHRs brain template set improved the accuracy of the image segmentation (Yang et al. 2022). A customized brain template set provides an advantage in reliably interpreting neuroimaging data. Given continuous controversy, we did not regress out the global signal. It has been reported that global signal regression sharply reduces the sensitivity of brain activity and connectivity despite increasing specificity in rodents (Chuang et al. 2019).

Limitations

This study has some limitations. First, only male SHRs were used. The female SHRs showed lower levels of blood pressure than the males. Early research confirmed that estrogen levels affect brain function in SHRs (Pietranera et al. 2016). In our future study, we will explore the effects of sex on brain function in SHRs by expanding the number of rats and stratifying by hypertension severity. Second, anesthesia was repeatedly used in our study. Anesthesia affects the physiological state of the brain and may lead to potential confusion. Our study was performed using low-dose isoflurane combined with dexmedetomidine, which is a viable option for longitudinal MRI studies in rats (Brynildsen et al. 2017). Blood oxygenation, respiration rates and body temperature were maintained at a stable level throughout our scanning to ensure validity and repeatability (Steiner et al. 2020). It is feasible to perform longitudinal rs-fMRI in awake rats. However, both acclimation and physical restraint are likely to add extra stress. Hence, MRI was performed in rats under anesthesia in our study. Third, we did not perform behavior tests on the rats. Thus, we could not identify whether changes in brain function were related to cognitive outcomes. Long-term longitudinal rs-fMRI studies with multiple time points of cognition measurement have unique potential to link brain function to cognitive ability. More research is needed to search for sufficiently predictive, sensitive MRI-based biomarkers.

Conclusion

Our longitudinal rs-fMRI experiment in SHRs provides preliminary evidence that hypertension and aging can independently contribute to brain dysfunction. The ongoing progression of brain function in aging SHRs highlights the importance of the continuous assessment of brain function in hypertensive population. Rs-fMRI is a unique tool for detecting brain function in rats. This study provides new insight into the dynamic characteristics of brain function, which is expected to become potential dynamic biomarkers of brain function damage. The major issue to be investigated is the mechanism underlying brain functional evolution in aging SHRs. We speculate that the changes in brain function are the consequence of numerous factors, which need to be investigated across multiple disciplines.

Funding

This research was supported by the National Natural Science Foundation of China (8177070094) and the Technology Research and Development Program of Qinhuangdao (202101A070).

Conflict of interest statement: None declared.