Urine Formaldehyde Predicts Cognitive Impairment in Post-Stroke Dementia and Alzheimer’s Disease

INTRODUCTION

Alzheimer’s disease (AD) is characterized by progressive deterioration in cognitive function [1]. Also, stroke patients often suffer from post-stroke cognitive impairment [2], or even from post-stroke dementia (PSD) [3]. However, the precise molecular pathogenesis of PSD and AD remains unclear. Substantial evidence indicates that a decline in hippocampus norepinephrine (NE) is related with cognitive deficits in aged rats [4], in particular in AD patients [5]. Moreover, accumulated endogenous formaldehyde had been found to contribute to NE depletion in vitro and in vivo [6, 7]. Interestingly, an abnormal overexpression of semicarbazide-sensitive amine oxidase (SSAO, a blood formaldehyde-generating enzyme [8, 9]), has been observed in both blood and brain, which contribute significantly to cognitive impairment in PSD [10] and AD patients [11].

Recently, excess formaldehyde accumulated in the hippocampi has been found in both AD patients and several transgenic AD-like animal models [12]. Surprisingly, excess formaldehyde not only induces the aggregation of amyloid-β (Aβ) and tau proteins in vitro [13, 14] and in vivo [15, 16], but also leads to vascular damage in patients with stroke and AD [13, 17]. These data suggest that accumulated formaldehyde plays a critical role in cognitive decline. Remarkably, urine levels of formaldehyde were shown to be inversely correlated with cognition in healthy aging individuals [18], and patients with dementia [12]. Herein, we investigated the relationship between blood SSAO, urine formaldehyde, and cognitive impairment in patients with PSD or AD, and tested whether endogenous formaldehyde could predict cognitive impairment inthese patients.

MATERIALS AND METHODS

RESULTS

We first investigated the relationship between formaldehyde level in the saliva, blood, or urine and cognitive ability of 195 participants in the first stage of this survey (Supplementary Figure 1). The results showed that formaldehyde concentrations in the urine, blood, and saliva were neither negatively nor positively correlated with ADL, CDR, and MMSE scores (Fig. 1a-g). However, urine formaldehyde levels were negatively correlated with the ability of cognition (MMSE) in 162 participants(Fig. 1h). In the second stage of this study, we then included patients who previously suffered from stroke (n = 61), and who suffered from post-stroke dementia (n = 65), or AD (n = 220), together with control participants (n = 231). The control was defined as a participant who had no clinical history of stroke, post-stroke dementia, and AD. A  number of participants still complained of age-related cognitive impairment (MMSE score≤20) (Supplementary Figure 2), since aging is associated with a significant decline in cognition in humans [18].

As shown in Table 1, the results revealed no significant differences in cognitive impairment among the groups assigned according to gender, level of education, diabetes, hypertension, coronary heart disease, and hyperlipidemia (p > 0.05). As expected, single-factors, such as age, stroke, AD, and urine levels of formaldehyde were correlated significantly with cognitive impairment in all 577 participants, respectively (p < 0.01). In particular, the multi-factors of education, sex, age, and stroke synergistically increased risk of cognitive impairment related to urine formaldehyde concentration (odd ratio = 7.204) (Supplementary Table 3).

Furthermore, we found that abnormally increased formaldehyde levels in the urine were associated with significantly decreased MMSE scores in patients with stroke, post-stroke dementia, and AD, when compared with controls (Fig. 2a, b). To determine the optimal concentration of urine formaldehyde for predicting cognitive impairment (≤20 MMSE scores), we analyzed the means of the ROC curve. As shown in Table 2 and Supplementary Figure 3, the results suggested that the best Cut-Off value of formaldehyde concentration for predicting cognitive impairment was 0.0418 mM in patients with PSD (Sensitivity: 92.3%; Specificity: 77.1%), and 0.0449 mM in patients with AD (Sensitivity: 94.1%; Specificity: 81.8%), respectively. Therefore, urine formaldehyde at 0.042 mM represents a risk threshold for cognitive impairment inthese patients.

To investigate why endogenous formaldehyde is accumulated in urine, we measured the expression levels of SSAO, a blood formaldehyde-generating enzyme, in the blood of these gender- and age-matched subjects selected from above four groups (n = 42 per group). The results showed that a marked decline in MMSE scores correlated with an elevation in SSAO protein, urine formaldehyde, and blood formaldehyde levels in patients with stroke, PSD, and AD (Table 3).

Furthermore, we found that the SSAO levels in the blood were negatively correlated with MMSE scores in these subjects (R = –0.790) (Supplementary Figure 4a-c), and positively correlated with formaldehyde levels in blood (R = 0.8078) and urine (R = 0.9230) (Supplementary Figure 4d, e). Urine formaldehyde levels were positively correlated with blood formaldehyde levels (R = 0.587) (Supplementary Figure 4f-h). Similarly, urine formaldehyde levels were correlated negatively with the MMSE scores in these subjects (R = –0.947) (Supplementary Figure 4i). Moreover, in the autopsied samples from dementia patients, a high level of expression of hippocampal SSAO was found to be associated with the concentration of formaldehyde in the hippocampi (Supplementary Figure 5a, b). These results indicate that the increases of SSAO in both the blood and the hippocampus contribute to the elevated level of urine formaldehyde.

DISCUSSION

Our present study made two major findings. First, in patients with PSD or AD, urine formaldehyde was inversely correlated with their cognitive ability, thus it could predict cognitive impairment. Second, as shown in analysis of these gender- and age-matched samples, the high-level expression of SSAO in blood is most likely one contributing factor for the abnormally increased level of urine formaldehyde. We also provide direct evidence that stroke is a pathological factor for developing PSD.

Growing evidence in the literature shows that aging and stroke contribute to the accumulation of endogenous formaldehyde and the occurrence of cognitive impairment. For example, overexpression and elevated activity of SSAO has been reported in blood samples taken from the elderly and stroke patients, and in the hippocampal samples from aging individuals [8, 26], particularly, in the hippocampus/cortex of AD patients [27]. In agreement with those observations, abnormally increased formaldehyde levels were observed in the blood, urine, and hippocampi of AD patients and rat models in our previous studies [28]. Moreover, SSAO can promote the expression of the amyloid-β protein precursor, resulting in the subsequent formation of amyloid oligomers, including Aβ_1-40, and Aβ_1-42 [29–31]. The overexpression of SSAO has been reported in patients with stroke, suggesting that dysregulation of SSAO and the subsequent imbalance of formaldehyde metabolism play a critical role in stroke, and thus contribute to cognitive impairment in patients with PSDand AD.

Our recent studies have shown that an abnormal high level of formaldehyde accumulates in the hippocampi of aged mice and rats [28], in the brains of three transgenic mouse models for AD, as well as in the autopsied hippocampi from AD patients [12]. Furthermore, injection of formaldehyde into hippocampus induced the death of hippocampal neurons, and interfered with the formation of spatial memory [28]. Another potential mechanism is that accumulated formaldehyde induces the aggregation of Aβ [13, 14], as well as the hyperphosphorylation and aggregation of tau proteins [15]. The deposition of Aβ as well as the phosphorylation and aggregation of tau proteins are widely considered to be responsible for initiating a cascade of pathological events that ultimately leads to AD [32]. The present study suggests that SSAO-derived formaldehyde contributes to cognitive impairment in PSD and AD (Supplementary Figure 6).

In this study, we found that measuring the levels of formaldehyde in overnight fasting urine is a feasible and noninvasive approach for monitoring the degree of cognitive impairment in PSD and AD patients, as this assay showed both high sensitivity and specificity. The optimal Cut-Off value of the level of urine formaldehyde as a  predictive concentration for cognitive impairment was approximately 0.042 mM. To rule out of the urine proteins interfering with the fluorescence signal of the formaldehyde-derivative, the speed of centrifugation of all urine samples in this study was increased from 4,000×g to 8,000×g (4,000×g was used in our previous study [12]), resulting in the precipitation of urine proteins from the supernatant. Using this method, urine formaldehyde levels were detected in the range from 0.009 to 0.565 mM, which is consistent with previously reported data [33]. Recent years, some fluorescence probes for detecting formaldehyde have been established [34–38], and these simple and quick methods will be used to measure urine formaldehyde in our further study. Intriguingly, the correlation coefficient of blood formaldehyde versus MMSE scores of the 42 participants (R = –0.425) is higher than that of urine formaldehyde versus MMSE scores (R = –0.947) (Supplementary Figure 4e, i). This observation implies that blood formaldehyde was deemed unsuitable to predict cognitive impairment. As formaldehyde is prone to react with serum proteins, the blood levels of formaldehyde are more variable than the formaldehyde levels in urine. Because urine contains very little residual proteins (Supplementary Table 2), the formaldehyde levels in urine probably reflect more accurately the status of endogenous formaldehyde metabolism. Our cross-sectional survey was limited by the small sample size (n = 577). First, we cannot state categorically whether there is indeed a causal relationship between the change in the urine level of formaldehyde and the degree of cognitive impairment during the progression of PSD and AD. A longitudinal (long-term follow-up) study is required to confirm such a relationship. Alternatively, an invasive detection of formaldehyde in the cerebrospinal fluid from dementia patients might be used to verify the results obtained for urine formaldehyde. Second, MMSE is commonly and reliably used to assess cognitive ability of both healthy control and the patients with dementia [21]. Usually, CDR is used to examine cognitive impairment for patients with dementia, but seldom for healthy humans [20]. In this study, we need to explore the relationship between urine formaldehyde and cognitive impairment of normal, MCI, and AD patients. Therefore, MMSE is more suitable to this study. Third, urine formaldehyde was also elevated in stroke patients. This seems to affect the sensitivity and specificity of urine formaldehyde for predicting dementia. However, acute stroke can be easily diagnosed clinically by using computed tomography, and magnetic resonance imaging. Remarkably, the higher levels of urine formaldehyde associated with the more severe cognitive impairment in stroke patients was found in this study (Supplementary Figure 4a, f). This result supports the notion that stroke is a high risk for developing PSD [3]. In addition, although we found that the specificity of formaldehyde (94.1%) for predicting cognitive impairment in patients with AD was higher than that in patients with PSD (77.1%), the cut-off value of formaldehyde (0.0449 mM) in AD was not markedly higher than that in PSD (0.0418 mM). This means that measuring urine formaldehyde is not a  suitable method for distinguishing PSD and AD, but the level of urine formaldehyde can predict the degree of cognitive decline in patients with these two diseases.

In summary, although we found that measuring formaldehyde content in the urine may be a reliable and sensitive, but also a simple and noninvasive method for predicting cognitive decline in patients with PSD or AD, whether urine formaldehyde acts as a biomarker for AD should be investigated in a longitudinal study.