Table 1—

Relation between diabetes and WMLs

Study (ref.)Study populationSubjects (total/diabetic) (n)Mean age (years)Diabetes typeImagingRating scaleOutcome and resultsP valueAdjustments/matching
General cohorts
    Dejgaard et al. (16)Case/control60/20401MRIDichOR 15.4 (3.8–63.2)<0.05Age
    Yousem et al. (37)Case/ control35/25311MRIDichNo WML in either groupNSAge, sex
    Longstreth et al. (7)Population based3,301/369>652MRIOrd (0–9)NSAge, sex
    den Heijer et al. (12)Population based506/41>602MRIOrd (0–9) PVHPVH: diff +0.4 (−0.2 to 1.1), d = 0.4NSAge, sex
Int WMLWML: diff +0.01 ml (−1.0 to 1.0), d = 0.2NS
    Schmidt et al. (13)Population based1,252/114692MRIIntPVH: median diff 00.1Sex, age, edu, BP, smoking, BMI, HL, PVD
WML: median diff +0.20.2
    Jerekathil et al. (9)Population based1814/9154NDMRIVolumetryWML: β = 0.020.8Age, sex
    Jorm et al. (15)Population based475/?>602MRIVolumetryWML: r = 0.03NSNo
Vascular cohorts
    Hijdra et al. (19)Stroke376/59>652CTOrd (0–4)*OR 2.0 (1.1–3.5)0.02No
    Schmidt et al. (17)Stroke/vascular risk factors and control234/3855NDMRIOrd (0–3)β = 0.2 (SE 0.1)<0.001Age, BP, sex, ICD, PVD
    Manolio et al. (10)Stroke and control303/76>652MRIOrd (0–8)Diabetes: mean diff 0.40.1No
    Fukuda and Kitani (63)Hypertension238/52>40NDMRIOrd (0–5)Diabetes: mean diff 0.1, d = 0.20.3No
    Jorgensen et al. (64)Stroke or TIA1,084/203>702CTDichOR 0.8 (0.5–2.3)0.3No
    Awada and Omojola (39)Stroke398/131>602CTDichOR 0.9 (0.6–1.5)No
    Henon et al. (65)Stroke610/83642CTOrd (0–3)Estimated coefficient 1.2 (SE 0.8)0.1Age, sex, BP, alcohol, HL, ICD, cerebral atrophy
    Padovani et al. (34)Stroke and control100/20>602MRIOrd (0–3)β = 0.02 (SE 0.1)0.9Age, sex, BP, PVD, ICD, ventricular index
    Coskun et al. (38)Stroke288/57>652CTOrd (0–4)*OR 0.6 (0.06–0.3)NSNo
    Streifler et al. (18)Stroke596/110>602CTOrd*OR 1.6 (1.0–2.6)0.04No
    Kario et al. (47)Hypertension20/20692MRIOrd*OR 2.2 (0.5–9.0)NSAge, sex, BP
Outpatient cohorts
    Raiha et al. (41)Geriatric department204/55742CTDichOR 0.9 (0.5–1.7)NSNo
    Araki et al. (21)MRI for any indication2,725/159602MRIDichOR 1.4 (0.7–2.8)NSNo
    Fukuda and Kitani (66)Neurologically normal253/50662MRIOrd (0-4)d = 0.10.6Age, sex, BP, smoking, HL, ICD
    Kobayashi et al. (22)Neurologically normal933/6658NDMRIOrd (PVH: 0–4)*PVH: OR 2.0 (0.8–4.8)0.2No
Dich (WML)DWML: OR 0.8 (0.3–2.8)0.6
    Hogervorst et al. (24)Alzheimer’s disease and control subjects414/29742CTOrd (0–3)*OR 1.4 (0.5–4.3)0.5Age, sex, diagnosis, smoking, diabetes, BP, apoE4
    Masana and Motozaki (23)Mild headache and vertigo1,674/8751NDMRIOrd (0–3)OR 1.6 (0.8–3.0)0.2Age, sex, BP, HL, family history, smoking, alcohol, ICD
    Biessels et al. (11)Memory clinic347/29732MRIOrd (0–6)PVH: mean diff 0 (−0.5 to 0.5)NSAge, sex
Ord (0–24)DWML: mean diff −0.5 (−2.0 to 1.5)NS
    Lazarus et al. (20)Memory clinic177/20>652MRIOrd (0–4)*PVH: OR 1.6 (0.6–5.2)NSAge, BP, AF, ICD
DWML: OR 2.9 (1.0–7.8)?
    Taylor et al. (14)Depressive disorder and control subjects399/34702MRIVolumetryWML: t = −1.30.2Age, sex, BP
GML in BG: t = −1.080.3
  • Studies are listed in chronological order. Study populations: general cohorts, population-based or case-control studies; vascular cohorts, cohort with stroke or other cardiovascular risk factors; outpatient cohorts, neurological, or psychiatric outpatients. Diabetes type: 2 (except for the Study by Kario et al. [47]), population type classified as predominantly type 2 diabetes (see research design and methods). ND, not determined. Rating scale: Dich, dichotomous scale; Int, interval scale; Ord, ordinal scale.

  • *

    * Dichotomization was performed for analysis. Outcome and results (diabetic patients vs. control subjects): BG, basal ganglia; DWML, deep WMLs; GML, gray matter lesions; PVH, periventricular hyperintensities. ORs are presented with 95% CIs in parentheses.

  • Where possible, we calculated OR or effect sizes (d) if they were not provided in the original article. Mean differences (diff; with 95% CIs in parentheses), d, β, and t values >0 reflect more severe WMLs in the diabetic group relative to control subjects. Studies marked with

  • were included in the meta-analysis presented in Table 4. P value: NS, not significant; ?, not specified. Adjustments/matching: apo, apolipoprotein; AF, atrial fibrillation; BP, blood pressure (including hypertension, mean arterial pressure, systolic blood pressure, use of antihypertensive drugs, left ventricular hypertrophy, and ankle-to-arm index); edu, education; HL, hyperlipidemia; ICD, ischemic cerebrovascular disease (including transient ischemic attack [TIA], stroke, leukoaraiosis, WMLs, and ultrasound examination of carotid or intracranial arteries); PVD, peripheral vascular disease (including peripheral artery disease, coronary artery disease, cardiac disease, congestive heart failure, and electrocardiogram changes).