A latent clinical-anatomical dimension relating metabolic syndrome to brain structure and cognition

  1. Marvin Petersen  Is a corresponding author
  2. Felix Hoffstaedter
  3. Felix L Nägele
  4. Carola Mayer
  5. Maximilian Schell
  6. D Leander Rimmele
  7. Birgit-Christiane Zyriax
  8. Tanja Zeller
  9. Simone Kühn
  10. Jürgen Gallinat
  11. Jens Fiehler
  12. Raphael Twerenbold
  13. Amir Omidvarnia
  14. Kaustubh R Patil
  15. Simon B Eickhoff
  16. Goetz Thomalla
  17. Bastian Cheng
  1. Department of Neurology, University Medical Center Hamburg-Eppendorf, Germany
  2. Institute for Systems Neuroscience, Medical Faculty, Heinrich-Heine University Düsseldorf, Germany
  3. Institute of Neuroscience and Medicine, Brain and Behaviour (INM-7), Research Center Jülich, Germany
  4. Midwifery Science-Health Services Research and Prevention, Institute for Health Services Research in Dermatology and Nursing (IVDP), University Medical Center Hamburg-Eppendorf, Germany
  5. Department of Cardiology, University Heart and Vascular Center, Germany
  6. German Center for Cardiovascular Research (DZHK), partner site Hamburg/Kiel/Luebeck, Germany
  7. University Center of Cardiovascular Science, University Heart and Vascular Center, Germany
  8. Department of Psychiatry and Psychotherapy, University Medical Center Hamburg-Eppendorf, Germany
  9. Department of Diagnostic and Interventional Neuroradiology, University Medical Center Hamburg-Eppendorf, Germany
  10. Epidemiological Study Center, University Medical Center Hamburg-Eppendorf, Germany
12 figures, 9 tables and 4 additional files

Figures

Methodology.

(a) Illustration of the partial least squares correlation analysis. Starting from two input matrices containing per-subject information of regional morphological measures as well as clinical data …

Figure 2 with 4 supplements
Partial least squares correlation analysis (PLS).

(a) Explained variance and p-values of latent variables. (b) Scatter plot relating subject-specific clinical and imaging PLS scores. Higher scores indicate higher adherence to the respective …

Figure 2—figure supplement 1
Partial least squares correlation analysis – Latent variable 2.

The figure presents the results of latent variable 2 of the partial least squares correlation analysis. (a) Scatter plot relating subject-specific clinical and imaging PLS scores. (b) Clinical …

Figure 2—figure supplement 2
Partial least squares correlation analysis – UK Biobank (including cognitive test results).

Partial least squares correlation analysis of the UK Biobank subsample including cognitive test results. (a) Explained variance and p-values of latent variables. (b) Scatter plot relating …

Figure 2—figure supplement 3
Partial least squares correlation analysis – Hamburg City Health Study (HCHS) (including cognitive test results).

Partial least squares correlation analysis of the HCHS subsample including cognitive test results. (a) Explained variance and p-values of latent variables. (b) Scatter plot relating subject-specific …

Figure 2—figure supplement 4
Spatial correlation of effect size maps.

Spatial correlation matrix of all Schaefer 400-parcellated metabolic syndrome effect maps (bootstrap ratio and t-statistic). The upper triangle of the matrix displays Spearman correlations with dot …

Mediation analysis results.

Mediation effects of subject-specific imaging PLS scores on the relationship between metabolic syndrome (MetS) represented by the clinical PLS score and cognitive test performances. Path plots …

Figure 4 with 2 supplements
Virtual histology analysis.

The regional correspondence between metabolic syndrome (MetS) effects (bootstrap ratio) and cell type-specific gene expression profiles was examined via an ensemble-based gene category enrichment …

Figure 4—figure supplement 1
Virtual histology analysis of latent variables 2 and 3.

Virtual histology analysis of the bootstrap ratio maps of latent variables 2 and 3 from the partial least squares (PLS) main analysis. Barplots display spatial correlation results of the bootstrap …

Figure 4—figure supplement 2
Sensitivity virtual histology analysis based on t-statistic map from group comparison.

Virtual histology analysis of the t-statistic map derived from group comparison between individuals with metabolic syndrome and controls. (a) Barplot displaying spatial correlation results of the …

Figure 5 with 2 supplements
Brain network contextualization.

Spatial correlation results derived from relating Schaefer 400×7-parcellated maps of metabolic syndrome (MetS) effects (bootstrap ratio) to network topological indices (red: functional connectivity, …

Figure 5—figure supplement 1
Sensitivity network contextualization analysis based on t-statistic map derived from group comparison.

Brain network contextualization analysis of group statistics results. Results are presented for t-statistics maps derived from group statistics considering the pooled sample of UK Biobank subjects …

Figure 5—figure supplement 2
Sensitivity network contextualization analysis based on group-consensus connectomes from the Hamburg City Health Study.

Brain network contextualization analysis of partial least squares correlation results (bootstrap ratio) based on group-consensus connectomes from the Hamburg City Health Study. Results are presented …

Graphical abstract.
Appendix 1—figure 1
Flowchart sample selection procedure.
Appendix 1—figure 2
Correlation matrix of metabolic syndrome-related risk factors.

The upper triangle of the matrix displays Pearson correlations with dot size and color representing the magnitude of the coefficients. The diagonal shows kernel density plots. The lower triangle …

Appendix 2—figure 1
Matching - UK Biobank.
Appendix 2—figure 2
Matching – Hamburg City Health Study.
Appendix 2—figure 3
Proportion of metabolic syndrome criteria.

Barplots indicate the percentage amount of metabolic syndrome (MetS) criteria that apply by group for the pooled sample. Significant group differences in χ2-tests are highlighted by asterisks.

Appendix 2—figure 4
Vertix-wise group comparison of cortical thickness.

Vertex-level group comparison between individuals with metabolic syndrome and matched controls. Resulting surface maps of standardized t-statistic estimates encode the group-differences between …

Tables

Table 1
Descriptive statistics UKB and HCHS.
MetricStat*
Age (years)63.55±7.59 (40087)
Sex (% female)46.47 (40087)
Education (ISCED)2.62±0.73 (39944)
Metabolic syndrome components
Waist circumference (cm)88.47±12.71 (38800)
Hip circumference (cm)100.90±8.79 (38801)
Waist-hip ratio0.88±0.09 (38800)
Body mass index26.47±4.37 (38701)
RRsystolic (mmHg)138.30±18.57 (31234)
RRdiastolic (mmHg)78.88±10.09 (31238)
Antihypertensive therapy (%)6.96 (39976)
HDL (mg/dL)61.76±23.69 (34468)
LDL (mg/dL)137.38±36.29 (37456)
Cholesterol (mg/dL)211.29±56.42 (37531)
Triglycerides (mg/dL)148.90±83.84 (37510)
Lipid lowering therapy (%)14.44 (39976)
HbA1c (%)5.37±0.48 (37284)
Blood glucose (mg/dL)90.29±17.58 (34432)
Antidiabetic therapy (%)0.45 (39976)
Imaging
Mean cortical thickness (mm)2.40±0.09 (40087)
Cognitive variables of the UK Biobank
Fluid Intelligence6.63±2.06 (36510)
Matrix Pattern Completion7.99±2.13 (25771)
Numeric Memory Test6.69±1.52 (26780)
Paired Associate Learning6.92±2.63 (26048)
Prospective Memory1.07±0.39 (37192)
Reaction Time (sec)594.16±109.08 (37015)
Symbol Digit Substitution18.96±5.25 (25810)
Tower Rearranging Test9.91±3.23 (25555)
Trail Making Test A (sec)223.03±86.51 (26048)
Trail Making Test B (sec)550.01±270.09 (26048)
Cognitive variables of the Hamburg City Health Study
Animal Naming Test24.78±6.92 (2416)
Clock Drawing Test6.43±1.12 (2479)
Trail Making Test A (sec)40.09±14.33 (2290)
Trail Making Test B (sec)90.05±37.30 (2264)
Multiple-Choice Vocabulary Intelligence Test31.27±3.58 (2026)
Word List Recall7.75±1.84 (2342)
  1. *

    Presented as mean ± SD (N).

Appendix 1—table 1
Partial least squares analysis - latent variables.
Latent variableExplained variance (%)p-value
071.200.0002
122.330.0002
22.120.0002
31.840.0006
41.030.0026
50.520.0266
60.380.0100
70.230.0032
80.180.1178
90.160.2122
100.000.3137
110.000.0608
120.001
130.001
140.001
150.001
Appendix 1—table 2
Partial least squares analysis-Cross-validation.
CV foldrsp
00.17
10.21
20.22
30.16
40.15
50.18
60.23
70.13
80.20
90.22
Appendix 1—table 3
Virtual histology analysis - Bootstrap ratio (partial least squares, PLS).
Cell typeZrsppFDR
Endo0.1900.016
Micro0.2710.016
Ex80.1650.016
In10.3630.036
Ex60.1460.034
Oligo0.2070.057
In70.0790.083
Ex10.1220.144
In20.0580.179
In30.0470.208
Astro0.0710.259
In80.0550.299
Ex70.0440.336
In50.0370.388
Ex4–0.0200.776
Ex5–0.0550.924
In4–0.0560.949
In6–0.0990.949
Ex2–0.1020.967
Ex3–0.2890.999
Appendix 1—table 4
Virtual histology analysis - t-statistic (group comparison).
Cell typeZrsppFDR
Endo0.2080.020
Micro0.3210.040
Ex80.2080.040
Oligo0.2330.055
In10.4320.108
Ex60.1450.123
Ex10.1560.229
In30.0580.233
Astro0.1200.233
In70.0590.233
In20.0630.233
Ex70.0890.263
In50.0630.300
In80.0660.317
Ex40.0150.585
Ex5–0.0070.690
In6–0.0780.861
Ex2–0.0700.861
In4–0.0870.901
Ex3–0.3410.997
Appendix 1—table 5
UK Biobank field IDs.
Age21003
Sex31
Education6133*
Waist circumference48
Hip circumference49
Body mass index21001
RRsystolic4080
RRdiastolic4079
HDL30760
LDL30780
Cholesterol30690
Triglycerides30870
HbA1c30750
Blood glucose30740
Medication for cholesterol, blood pressure, diabetes6153
Fluid Intelligence20191
Matrix Pattern Completion6373
Numeric Memory Test20240
Paired Associate Learning20197
Prospective Memory20018
Reaction Time20023
Symbol Digit Substitution20159
Tower Rearranging Test21004
Trail Making Test A6348
Trail Making Test B6350
  1. Abbreviations: RR = blood pressure.

  2. *

    Converted to International Standard Classification of Education (ISCED) via the UKBB parser (https://github.com/USC-IGC/ukbb_parser; Zhu et al., 2019).

Appendix 1—table 6
Metabolic syndrome Criteria of the International Diabetes Federation (IDF) (Alberti et al., 2006).
Metabolic syndrome = obesity + two further criteria
Obesitywaist circumference ♀:≥80 cm; ♂:≥94 cm
Dyslipidemia (raised triglycerides)≥150 mg/dL (1.7 mmol/L) or lipid lowering medication
Dyslipidemia (reduced HDL cholesterol)♀:<50 mg/dL (1.29 mmol/L); ♂:<40 mg/dL (1.03 mmol/L) in males
Arterial hypertension (raised blood pressure)systolic BP ≥130 or diastolic BP ≥85 mm Hg or antihypertensive medication or diagnosis of hypertension
Insuline resistanceFasting plasma glucose ≥100 mg/dL (5.6 mmol/L) or antidiabetic therapy or diagnosis of diabetes mellitus type 2*
  1. *

    Measurements of fasting plasma glucose were not available for the study sample. Consequently, the criterion of insulin resistance was only based on the diagnosis of diabetes mellitus and administration of antidiabetic therapy.

Appendix 2—table 1
Descriptive group statistics - UK Biobank.
Metric*Individuals with MetSMatched controlsPuncorrPFDRStat
Age (years)64.73±7.42 (6746)64.51±7.27 (6746)0.0950.154–0.03
Sex (% female)18.81 (6746)18.81 (6746)>0.99>0.990
Education (ISCED)2.63±0.73 (6746)2.67±0.71 (6746)0.0360.0690.04
Metabolic syndrome criteria
Waist circumference (cm)97.39±10.21 (6726)88.22±10.59 (6595)<0.001<0.001 –0.88
RRsystolic (mmHg)146.41±15.38 (6213)135.57±17.71 (5397)<0.001<0.001 –0.66
RRdiastolic (mmHg)82.26±9.39 (6214)77.58±9.79 (5397)<0.001<0.001 –0.49
Antihypertensive therapy (%)9.96 (6746)9.68 (6746)<0.001<0.001 7.07
HDL (mmol/L)1.18±0.26 (6225)1.49±0.32 (6332)<0.001<0.001 1.08
Triglycerides (mmol/L)2.43±1.13 (6617)1.30±0.59 (6543)<0.001<0.001 –1.25
Lipid-lowering therapy (%)39.05 (6746)7.07 (6746)<0.001<.001 2446.5
Blood glucose (mmol/L)5.18±1.41 (6219)4.92±0.68 (6325)<0.001<.001 –0.23
Antidiabetic therapy (%)0.06 (6746)0.19 (6746)0.0520.0973.77
Cognitive scores
Fluid Intelligence6.66±2.10 (6221)6.82±2.09 (6241)<0.001<0.001 0.08
Matrix Pattern Completion8.02±2.13 (4283)8.14±2.06 (4355)0.0550.0960.06
Numeric Memory Test6.64±1.61 (4419)6.84±1.53 (4505)<0.001<0.001 0.12
Paired Associate Learning6.45±2.60 (4337)6.73±2.61 (4392)<0.001<0.001 0.10
Prospective Memory1.05±0.40 (6362)1.06±0.39 (6349)0.2210.3390.02
Reaction Time590.75±108.27 (6331)590.15±111.13 (6325)0.7920.858–0.005
Symbol Digit Substitution18.47±5.12 (4292)19.00±5.16 (4353)<0.001<0.001 0.10
Tower Rearranging Test10.00±3.28 (4255)10.08±3.20 (4325)0.7470.8450.02
Trail Making Test A (sec)226.83±86.26 (4337)224.06±83.06 (4392)0.6430.836–0.03
Trail Making Test B (sec)561.81±271.39 (4337)553.62±282.55 (4392)0.6110.756–0.03
Imaging
Mean cortical thickness (mm)2.392±0.09 (6746)2.397±0.09 (6746)0.0350.0710.05
  1. Abbreviations: cm = centimeter, dL = deciliter, HDL = high-density lipoprotein, ISCED = International Standard Classification of Education, MetS = metabolic syndrome, mg = milligram, mm = millimeter, mmHg = millimeters of mercury, mmol/L = millimole perliter, PC = principal component, Puncor = uncorrected p-values, PFDR = false-discovery rate-corrected p-values, RR = Blood pressure, sec = seconds.

  2. *

    Presented as mean ± SD (N).

  3. Presented as χ2 for categorical and Cohen’s d for continuous data.

  4. Denotes statistical significance at FDR-corrected p<0.001

Appendix 2—table 2
Descriptive group statistics Hamburg City Health Study (HCHS).
Metric*Individuals with MetSMatched controlsPuncorrPFDRStat
Age (years)65.77±7.40 (759)65.97±7.52 (759)0.6130.6470.03
Sex (% female)33.4736.500.2360.2811.4
Education (ISCED)2.37±0.58 (759)2.42±0.60 (759)0.090.1140.09
Metabolic syndrome criteria
Waist circumference (cm)103.38±11.23 (754)91.45±11.36 (747)<0.001<0.001–1.06
RRsystolic (mmHg)145.66±18.54 (740)140.17±21.10 (746)<0.001<0.001–.28
RRdiastolic (mmHg)83.75±10.11 (740)82.00±10.40 (746)0.0010.002–.17
Antihypertensive therapy (%)52.60%26.22%<0.001<0.001108.89
HDL (mg/dL)54.60±16.13 (751)67.63±17.46 (759)<0.001<0.0010.78
Triglycerides (mg/dL)161.53±92.61 (751)91.23±30.62 (759)<0.001<0.001–1.02
Lipid lowering therapy (%)40.85%7.64%<0.001<0.001225.29
Blood glucose (mg/dL)107.47±28.59 (742)90.99±10.87 (753)<0.001<0.001–0.76
Antidiabetic therapy (%)14.42%1.45%<0.001<0.00185.47
Cognitive scores
Animal Naming Test23.71±6.46 (712)24.77±6.75 (711)0.0050.0090.16
Clock Drawing Test6.36±1.17 (730)6.39±1.16 (733)0.7740.7740.02
Trail Making Test A (sec)41.26±14.28 (685)40.42±14.54 (675)0.3210.359–0.06
Trail Making Test B (sec)93.74±37.30 (675)89.89±37.69 (671)0.0860.114–0.10
Multiple-Choice Vocabulary Intelligence Test31.18±3.43 (603)31.71±3.22 (619)0.0190.0340.16
Word List Recall7.42±1.89 (691)7.64±1.84 (673)0.0570.0830.12
Imaging
Mean cortical thickness (mm)2.327±0.08 (759)2.334±0.08 (757)0.0450.0710.1
  1. Abbreviations: cm = centimeter, dL = deciliter, HDL = high-density lipoprotein, ISCED = International Standard Classification of Education, MetS = metabolic syndrome, mg = milligram, mm = millimeter, mmHg = millimeters of mercury, Puncor = uncorrected p-values, PFDR = false-discovery rate-corrected p-values, RR = Blood pressure, sec = seconds.

  2. *

    Presented as mean ± SD (N).

  3. Denotes statistical significance at FDR-corrected p<0.001.

Additional files

Supplementary file 1

Schaefer 400×7-parcellated maps of metabolic syndrome (MetS)-related brain morphological abnormalities (bootstrap ratio from PLS, t-statistic from group comparison).

https://cdn.elifesciences.org/articles/93246/elife-93246-supp1-v1.csv
Supplementary file 2

Schaefer 100×7-parcellated bootstrap ratio map.

https://cdn.elifesciences.org/articles/93246/elife-93246-supp2-v1.csv
Supplementary file 3

t-statistic from group comparison on fsLR.

https://cdn.elifesciences.org/articles/93246/elife-93246-supp3-v1.csv
MDAR checklist
https://cdn.elifesciences.org/articles/93246/elife-93246-mdarchecklist1-v1.pdf

Download links