Comparison of clinical data between T-T2DM and L-T2DM
In this study, we compared the clinical data of 100 T-T2DM and 100 L-T2DM patients. There were no significant differences between T-T2DM and L-T2DM patients in terms of sex distribution, body mass index (BMI) height, sex or weight. However, the average age in the T-T2DM group was lower than that in the L-T2DM group (P < 0.001). The biochemical test results showed that alanine aminotransferase (ALT) (P < 0.001), aspartate aminotransferase (AST) (P = 0.001) and hemoglobin A1 (HbA1C) (P = 0.014) levels were significantly higher in the T-T2DM group than in the L-T2DM group. High-density lipoprotein cholesterol (HDL-C) levels were significantly lower in the T-T2DM group than in the L-T2DM group (P = 0.010). Other than that, there were no significant differences in the rest of the metabolites, including 2 h glucose (Table 1). Thus, in the population we investigated, the high-altitude environment did not play a role in lowering blood sugar. In contrast, the T-T2DM patients tended to be younger, have a higher HbA1C level, and have more severe symptoms of diabetes than L-T2DM patients. Therefore, we speculated that the DCD in high-altitude was more severe than that in low-altitude.
Table 1
Clinical characteristics of T2DM between high-altitude and low-altitude
Characteristic
|
T-T2DM
|
L-T2DM
|
P-value
|
Age, years
|
51.54 ± 9.14
|
55.64 ± 6.73
|
P<0.001
|
Gender
|
|
|
0.083
|
Male
|
59
|
63
|
|
Female
|
41
|
37
|
|
Height, cm
|
168.71 ± 8.35
|
165.93 ± 7.72
|
0.676
|
Weight, kg
|
73.55 ± 11.75
|
64.98 ± 9.40
|
0.040
|
BMI, kg/m2
|
25.80 ± 3.44
|
23.57 ± 2.83
|
0.082
|
Systolic blood pressure (mm Hg)
|
120.40 ± 14.60
|
123.08 ± 14.67
|
0.733
|
Diastolic blood pressure (mm Hg)
|
79.69 ± 9.68
|
80.87 ± 10.22
|
0.399
|
LDL-cholesterol (mmol/l)
|
2.62 ± 0.79
|
2.44 ± 0.85
|
0.408
|
HDL-cholesterol (mmol/l)
|
1.15 ± 0.25
|
1.31 ± 0.72
|
0.010
|
ALT, median(range), U/l
|
29.99 ± 15.95
|
20.08 ± 11.67
|
P<0.001
|
AST, median(range), U/l
|
20.58 ± 11.72
|
17.40 ± 6.62
|
0.001
|
Triacylglycerides (mmol/l)
|
1.52 ± 0.77
|
1.62 ± 0.90
|
0.130
|
Total cholesterol (mmol/l)
|
4.47 ± 0.92
|
4.33 ± 1.08
|
0.205
|
HbA1C (%)
|
10.32 ± 2.55
|
8.46 ± 2.11
|
0.014
|
Fasting glucose (mmol/l)
|
7.87 ± 2.10
|
7.00 ± 2.00
|
0.334
|
2 h glucose, OGTT (mmol/l)
|
16.91 ± 3.97
|
16.32 ± 4.52
|
0.313
|
Note: T2DM, Type 2 diabetes mellitus; LDL, low density lipoprotein; HDL, high density lipoprotein; ALT, alanine aminotransferase; AST, aspartate aminotransferase; HbA1c, hemoglobin A1. Data are mean ± SD. T-test in continuous variables and chi-square test in categorical data were performed as appropriate. Results were considered significant when P < 0.05. |
Comparison of cognitive function between T-T2DM and L-T2DM
To confirm whether the DCD in high-altitude was more severe than that in low-altitude, we administered questionnaires to assess cognitive function. Scores on the Patient Health Questionnaire-9 (PHQ-9) (P = 0.014), 7-item Generalized Anxiety Disorder Scale (GAD-7) (P = 0.042) and Self-Rating Depression Scale (SDS) (P = 0.038) were significantly higher in the T-T2DM group than in the L-T2DM group (P < 0.05), suggesting that depression and anxiety were more severe in the T-T2DM group than in the L-T2DM group. The mean Memory_Immediately_Score and Memory_Delay_Score were lower in the T-T2DM group than in the L-T2DM group, revealing that the memory ability of T-T2DM patients was lower than that of L-T2DM patients. Moreover, the average MMSE score in the T-T2DM group (26) was lower than that in the L-T2DM group; a mean MMSE score less than 26 was previously considered the threshold for cognitive dysfunction [12], suggesting that the risk of DCD in high- altitude was higher than that in low-altitude. In addition, the mean time of Visual Memory (VM)_Immediately_Date and VM_Delay_Date were longer in the T-T2DM group than in the L-T2DM group, indicating that visual memory ability in the T-T2DM group was worse than in the H-T2DM group (Table 2). Thus, the memory and cognitive function of T-T2DM patients were more impaired than those of L-T2DM patients, and T-T2DM patients exhibited more severe anxiety.
Table 2
Comparing of cognitive function between T_T2DM and L_T2DM
Characteristic
|
T_T2DM
|
L_T2DM
|
P-value
|
PSQI
|
6.63 ± 3.22
|
6.07 ± 4.38
|
0.196
|
SDS
|
34.89 ± 6.41
|
29.53 ± 3.11
|
0.038
|
PHQ9
|
3.78 ± 3.97
|
1.40 ± 1.45
|
0.014
|
GAD7
|
3.89 ± 3.76
|
1.53 ± 2.00
|
0.042
|
Anteroposterior_Score
|
8.33 ± 1.49
|
8.27 ± 1.79
|
0.867
|
Reverse_digit_Score
|
3.93 ± 1.00
|
3.80 ± 1.15
|
0.659
|
Memory_Immediately_Score
|
9.44 ± 3.43
|
11.67 ± 3.74
|
0.501
|
MMSE
|
26.30 ± 2.71
|
27.40 ± 2.29
|
0.496
|
Memory_Delay_Score
|
8.00 ± 4.03
|
10.33 ± 3.85
|
0.858
|
VM_Immediately_Date
|
4.52 ± 2.14
|
3.40 ± 2.50
|
0.294
|
VM_Delay_Date
|
4.07 ± 2.18
|
2.73 ± 1.94
|
0.435
|
Metabolic profiles of different groups
We then explored biomarkers associated with severe cognitive impairment in T2DM patients at high altitude. We used LC‒MS/MS to profile serum samples from 100 T-T2DM, 100 T-HC, 100 L-T2DM, and 100 L-HC individuals using the criteria of FC > 1.2 or < 0.833, P < 0.05 and VIP > 1.
In total, we identifified 412 differentially metabolites (DMs) expressed (32 up-regulated and 380 down-regulated) were related with the T-T2DM (Fig. 2A). Moreover, we also identified 308 metabolites associated with L-T2DM, and 51 DMs were up-regulated and 257 DMs were down-regulated (Fig. 2B). Then, we compared the 412 DMs association with T-T2DM to the 308 DMs related with L-T2DM. We identified 26 differentially expressed (8 up-regulated and 18 down-regulated) endogenous metabolites (Fig. 2C).
To comprehensively investigate the levels and regulation process of these 26 DMs, we applied the Z score transformation (a value converted based on the relative level of metabolite) and conducted cluster analysis. The absolute value of the Z score distribution in the L-T2DM group was smaller than that in the T-T2DM group, indicating that the levels of these metabolites were generally lower in the L-T2DM group than in the T-T2DM group (Fig. 2D). The clustering results showed that 4-(2-thienyl) benzoic acid, N'1-(2-chlorobenzoyl)-2-[(2-methyl-1H-indol-3-yl)thio]ethanohydrazide, 2-hydroxy-6-aminopurine, and guanosine had the most significant differences between the T-T2DM and L-T2DM groups (Fig. 2E). In addition, correlation analysis of the levels of the top 20 metabolites showed that N'1-(2-chlorobenzoyl)-2-[(2-methyl-1H-indol-3-yl)thio]ethanohydrazide and guanosine levels were both strongly positively correlated with 2-hydroxy-6-aminopurine levels (Fig. 2F). This suggests that N'1-(2-chlorobenzoyl)-2-[(2-methyl-1H-indol-3-yl)thio]ethanohydrazide, guanosine and 2-hydroxy-6-aminopurine are closely related to T-T2DM group.
Identifying key metabolites
To identify key metabolites associated with DCD in high-altitude, KEGG enrichment analysis was performed with the 26 DMs. The results showed that the main KEGG pathways were lysine degradation, fatty acid biosynthesis, purine metabolism and metabolic pathways and focused on four metabolites, including pipecolic acid, lauric acid, guanosine and kaempferol (but not 2-hydroxy-6-aminopurine) (Fig. 3A). This means that these 4 metabolites can be used as early biomarkers of DCD in high-altitude.
The random forest results showed that guanosine, pipecolic acid and lauric acid were three of the top 20 biomarkers (Fig. 3B). Moreover, the ROC results showed that the area under the curve (AUC) reached 0.818, suggested that the random forest model had high predictive accuracy (Fig. 3C).
Subsequently, we analyzed the levels of guanosine, pipecolic acid and lauric acid in each group. The results showed that compared with the T-HC, L-HC and L-T2DM groups, the T-T2DM group had the lowest levels of pipecolic acid and lauric acid and the highest levels of guanosine (Fig. 3D-F). Combined this finding and the random forest results, we believe that guanosine, pipecolic acid and lauric acid could be as potential markers of DCD in high-altitude.
To further verify the reliability of these biomarkers, we matched the top 20 biomarkers in the mzCloud (https://www.mzcloud.org/) and mzVault databases (https://mytracefinder.com/tag/mzvault/) (the match between the actual detected mass-to-charge ratio and the mass-to-charge ratio included in the database; the highest score was 100 points) and found that only guanosine and pipecolic acid had high matches in the three databases, and the match scores were both up to 88 points (Table 3). This illustrated that only guanosine and pipecolic acid levels had high predictive accuracy regarding.
Table 3
The matching degree of the top 20 metabolites in three database
Metabolite Name
|
mzCloud
|
mzVault
|
MassList
|
PC (16:2e/4:0)
|
No results
|
79.7
|
full match
|
LPC 17:1
|
No results
|
87.3
|
No results
|
LPC 12:0
|
No results
|
88.1
|
No results
|
Lysope 16:0
|
No results
|
No results
|
full match
|
Lysopc 14:0
|
No results
|
No results
|
full match
|
Feruloyl Putrescine
|
No results
|
No results
|
full match
|
FAHFA (18:3/8:0)
|
No results
|
53.4
|
No results
|
LPC 14:0
|
No results
|
89.4
|
No results
|
LPC 20:0
|
No results
|
76.8
|
No results
|
(±)13-HODE
|
No results
|
73
|
full match
|
Lauric acid
|
86.9
|
No results
|
full match
|
Guanosine
|
88.7
|
92
|
full match
|
Methyl 3-indolyacetate
|
No results
|
No results
|
full match
|
N'1-(2-chlorobenzoyl)-2-[(2-methyl-1H-indol-3-yl)thio]ethanohydrazide
|
41
|
No results
|
No results
|
all-cis-4,7,10,13,16-Docosapentaenoic acid
|
82.4
|
No results
|
full match
|
LPC 19:1
|
No results
|
85.5
|
full match
|
2-(Formylamino)Benzoic Acid
|
No results
|
No results
|
full match
|
(±)9-HpODE
|
88.1
|
No results
|
No results
|
Pipecolic acid
|
91.2
|
89.9
|
full match
|
Moreover, a study reported that guanosine levels were strongly associated with oxidative brain damage and were negatively correlated with cognitive dysfunction [13]. Coincidentally, in our studies, the memory and cognitive function of T-T2DM patients were more impaired than that of the L-T2DM group, and the T-T2DM group had more severe anxiety. In addition, the level of guanosine in the T-T2DM group was higher than that of the L-T2DM, T-HC and L-HC groups. Therefore, we believe that guanosine is a potential biomarker of cognitive function.
Another study showed that changes in certain plasma metabolites of pipecolic acid can mediate gut-brain signaling and that levels of these metabolites are positively correlated with cognitive function [14]. Interestingly, our study found that the level of pipecolic acid in the T-T2DM group was lower than that in the L-T2DM, T-HC and L-HC groups. This result suggests that the differences in cognitive function between the T-T2DM and L-T2DM groups were also related to pipecolic acid levels.
Overall, both guanosine and pipecolic acid may be the key biomarkers that predict the DCD in high-altitude.
Correlation of DMs with cognitive function
To further verify the correlations of guanosine and pipecolic acid levels with DCD in high-altitude, we analyzed the correlation between levels of all 26 metabolites and participant cognitive function scores. The results showed that guanosine levels were indeed negatively correlated with Reverse_Digit_Score, Memory_Immediately_Score, MMSE score, Memory_Delay_Score, VM_Immediately_Date, and VM_Delay_Date, all values closely related to cognitive function. Pipecolic acid levels were positively correlated with these scores, but the correlations were weaker than those of guanosine levels (Fig. 4A). These findings are both consistent with Petronilh’s [13] and Kong’s reports [14]. These results also showed that levels of both guanosine and pipecolic acid could be used as biomarkers of DCD in high-altitude.
To further verify whether guanosine and pipecolic acid levels are affected by T2DM, we also performed correlation analysis on the 26 DMs and biochemical indicators. The results showed that only guanosine levels were not correlated with biochemical indicators, especially age, fasting glucose, and HbAc1 (which are closely related to diabetes) (Fig. 4B). This suggests that changes in guanosine levels were not affected by diabetic and were only associated with changes in cognitive function.
Therefore, we believe that guanosine levels are the most promising biomarker for DCD in high-altitude.
Multicenter analysis
To confirm the validity of guanosine levels as a biomarker of DCD in high-altitude, we further recruited 30 diabetics live in high altitude (TDM) and 30 diabetics live in low-altitude (LDM). These new participants completed the cognitive function test at Hospital C.T. after application of the inclusion and exclusion criteria, and their serum was collected to perform LC‒MS/MS analysis. The cognitive function results also showed that scores on the PHQ-9 (P = 0.015), GAD-7 (P = 0.046) and SDS (P = 0.049) in the TDM group were significantly higher than those in the LDM group (P < 0.05) (Table 4), which was consistent with the results obtained in the comparison of the T-T2DM and L-T2DM groups. The memory and cognitive dysfunction of the TDM group were also impaired to a greater extent than those of the LDM group, and the TDM group exhibited more severe anxiety. The LC‒MS/MS results showed that 800 and 792 metabolites were detected in samples from the TDM and LDM groups, respectively, and there were 790 metabolites in common (Fig. 5A). The results of partial least squares discrimination analysis (PLS-DA) showed that the R2Y was greater than the Q2Y, and the model was well established (Fig. 5B). Then, KEGG enrichment analysis was also performed, and the results revealed that the DMs were mainly concentrated in lipid metabolism (Fig. 5C). The network structure analysis results showed that the top 20 DMs still contained guanosine (Fig. 5D). Thus, we belived that the guanosine is the biomarker with the most potential to become an early biomarker of DCD in high-altitude.
Table 4
Comparing of cognitive function between TDM and LDM
Characteristic
|
TDM
|
HDM
|
P-value
|
PSQI
|
6.8±3.21
|
6. 71±5.03
|
0.051
|
SDS
|
35.00±6.23
|
30.11±3.71
|
0.049
|
PHQ9
|
3.77±3.78
|
1.47±1.42
|
0.015
|
GAD7
|
4.27±3.83
|
1.71±2.20
|
0.046
|
Anteroposterior_Score
|
8.37±1.45
|
8.24±1.68
|
0.939
|
Reverse_digit_Score
|
3.93±0.98
|
3.82±1.07
|
0.965
|
Memory_Immediately_Score
|
9.43±3.26
|
11.67±3.74
|
0.619
|
MMSE
|
26.37±2.62
|
27.65±2.26
|
0.329
|
Memory_Delay_Score
|
8.00±3.81
|
10.32±3.58
|
0.552
|
VM_Immediately_Date
|
4.65±2.11
|
3.43±2.42
|
0.272
|
VM_Delay_Date
|
4.03±2.11
|
2.63±1.87
|
0.556
|