Northam et al.Central nervous system function in youth with type 1 diabetes 12 years after disease onset.

Table A1Regression of IQ and metabolites.

Outcome / VIQ / PIQ / FSIQ
Overall model / Adjusted R2=0·50, F=18·08, P<0·001 / Adjusted R2=0·46, F=15·79,
P<0·001 / Adjusted R2=0·58, F=25·24,
P<0·001
Predictor / Β* (95% CI) / P- value / Β* (95% CI) / P- value / Β* (95% CI) / P- value
diabetes onset age / –3·12 (–7·33, 1·10) / 0·15 / –12·01 (–16·68, –7·34) / <0·001 / –7·9 (–11·81, –3·98) / <0·001
Hypoglycemia / –4·51 (–7·99, –1·04) / 0·01 / –0·21 (–4·06, 3·64) / >0·9 / 2·85 (–6·07, 0·38) / 0·08
Metabolic control / –0·07 (–0·72, 0·59) / 0·8 / –0·25 (–0·98, 0·48) / 0·5 / –0·18 (–0·79, 0·43) / 0·6
FSIQbaseline / 3·95 (2·65, 5·25) / <0·001 / 4·67 (3·23, 6·12) / <0·001 / 4·89 (3·68, 6·10) / <0·001
SES / –3·34 (–5·19, –1·50) / 0·001 / –2·31 (–4·35, –0·27) / 0·03 / –3·04 (–4·75, –1·33) / 0·001
Time b’n assessments / –0·02 (–1·90, 1·85) / >0·9 / –1·12 (–3·20, 0·96) / 0·3 / –0·64 (–2·39, 1·10) / 0·5
Outcome / mI basal ganglia / mI frontal lobe / mI temporal lobe
Overall model / Adjusted R2=0·08, F=3·13,
P=0·03 / Adjusted R2=0·03, F=1·57,
P=0·2 / Adjusted R2=0·00, F=1·04,
P=0·4
Predictors / Β* (95% CI) / P- value / Β* (95% CI) / P- value / Β* (95% CI) / P- value
Age / 0·03 (–0·00, 0·06) / 0·06 / 0·01 (–0·02, 0·04) / 0·5 / 0·03 (–0·01, 0·07) / 0·13
Hypoglycemia / 0·06 (–0·18, 0·29) / 0·6 / 0·21 (–0·05, 0.47) / 0·12 / 0·17 (–0·15, 0·49) / 0·3
Metabolic control / 0·05 ( 0·00, 0·09) / 0·04 / 0·04 (–0·01, 0·09) / 0·12 / –0·01 (–0·07, 0·05) / 0·7
Outcome / Cr basal ganglia / Cr frontal lobe / Cr temporal lobe
Overall model / Adjusted R2=-0·01, F=0·87,
P=0·5 / Adjusted R2=-0·01, F=0·71,
P=0·6 / Adjusted R2=0·01, F=1·19,
P=0·3
Predictors / Β* (95% CI) / P- value / Β* (95% CI) / P- value / Β* (95% CI) / P- value
Age / 0·01 (–0·02, 0·05) / 0·5 / –0·01 (–0·03, 0·01) / 0·3 / –0·02 (–0·05, 0·02) / 0·3
Hypoglycemia / –0·20 (–0·50, 0·10) / 0·20 / –0·06 (–0·20, 0·09) / 0·5 / –0·07 (–0·35, 0·21) / 0·6
Metabolic control / –0·02 (–0·07, 0·04) / 0·6 / 0·01 (–0·02, 0·04) / 0·4 / 0·04 (–0·01, 0·09) / 0·11
Outcome / Glx basal ganglia / Glx frontal lobe / Glx temporal lobe
Overall model / Adjusted R2=–0·02, F=0·54,
P=0·7 / Adjusted R2=0·08, F=2·79,
P=0·05 / Adjusted R2=0·04, F=1·94,
P=0·13
Predictors / Β* (95% CI) / P- value / Β* (95% CI) / P- value / Β* (95% CI) / P- value
Age / 0·04 (–0·04, 0·12) / 0·3 / –0·09 (–0·15, –0·02) / 0·007 / –0·05 (–0·13, 0·03) / 0·2
Hypoglycemia / –0·04 (–0·69, 0·61) / 0·9 / –0·24 (–0·76, 0·28) / 0·4 / 0·31 (–0·35, 0·97) / 0·4
Metabolic control / 0·02 (–0·10, 0·14) / 0·7 / 0·04 (–0·06, 0·14) / 0·4 / 0·12 (–0·01, 0·24) / 0·06
Outcome / NAA basal ganglia / NAA frontal lobe / NAA temporal lobe
Overall model / Adjusted R2=0·02, F=1·54,
P=0·2 / Adjusted R2=0·01, F=1·12,
P=0·3 / Adjusted R2=0·10, F=3·66,
P=0·02
Predictors / Β* (95% CI) / P- value / Β* (95% CI) / P- value / Β* (95% CI) / P- value
Age / –0·03 (–0·08, 0·01) / 0·14 / –0·03 (–0·06, 0·01) / 0·17 / –0·08 (–0·13, –0·03) / 0·002
Hypoglycemia / –0·22 (–0·60, 0·16) / 0·2 / –0·17 (–0·47, 0·14) / 0·3 / –0·17 (–0·58, 0·24) / 0·4
Metabolic control / 0·05 (–0·02, 0·12) / 0·19 / –0·02 (–0·08, 0·04) / 0·5 / 0·00 (–0·08, 0·07) / >0·9
Outcome / Cho basal ganglia / Cho frontal lobe / Cho temporal lobe†
Overall model / Adjusted R2=0·02, F=1·59,
P=0·2 / Adjusted R2=-0·03, F=0·18,
P>0·9
Predictors / Β* (95% CI) / P- value / Β* (95% CI) / P- value
Age / 0·01 (0·00, 0·02) / 0·04 / –0·00 (–0·01, 0·01) / 0·7
Hypoglycemia / –0·01(–0·10, 0·09) / 0·9 / 0·02 (–0·06, 0·11) / 0·6
Metabolic control / 0·00 (–0·02, 0·02) / >0·9 / 0·00 (–0·02, 0·02) / 0·9
Outcome / GM volume thalamus / T2 thalamus
Overall model / Adjusted R2=0·10, F=3·94,
P=0·12 / Adjusted R2=0·49, F=24·78, P0·001
Predictors / Β* (95% CI) / P- value / Β* (95% CI) / P- value
Age / –0·003 (–0·005, –0·001) / 0·016 / –5·09 (–6·40, –3·79) / <0·001
Hypoglycemia / –0·019 (–0·037,–0·002) / 0·03 / –6·58 (–17·49, 4·33) / 0.2
Metabolic control / –0·002 (–0·005, 0·002) / 0·3 / –2·055 (–4·126, 0·016) / 0·05
Outcome / GM volume lentiform / T2 lentiform
Overall model / Adjusted R2=0·26, F=10·32,
P<0·001 / Adjusted R2=0.62, F=41.45, P<0.001
Predictors / Β* (95% CI) / P- value / Β* (95% CI) / P- value
Age / –0·004 (–0·005, –0·002) / <0·001 / –5·06 (–6·01, –4·10) / <0·001
Hypoglycemia / –0·002 (–0·014, 0·01) / 0·7 / –3·27 (–11·25, 4·72) / 0·4
Metabolic control / 0·001 (0·004, 0·001) / 0·3 / –0·836 (–2·352, 0·679) / 0·3

β and CI values for metabolic control and FSIQbaseline are multiplied by 10.

Non-parametric univariate procedures were used to examine the relationship between Cho in the temporal lobe and the three predictors because of an extreme value in the original data set.. Two of these relationships were very weak: Cho and metabolic control (Spearman’s ρ=·03. p=·8), Cho and hypoglyceamia (median Cho for those with and without a history of hypoglycemia is 1·7, p=·8 from the Mann Whitney U test). Spearman’s ρ for Cho and age was ·29, (p=·01).

Figure A1. Voxel-based (VB) analysis of volume and T2 differences between type 1 diabetes and C. The glass brain in the upper row depicts the whole-brain picture of significant voxels, while the lower row shows these results overlayed onto an averaged brain in standard space in an area of interest. Results are shown according to neurological convention and for p<·0005 (non-corrected for multiple comparisons).

The left panel (A) shows decreased GM volume (in particular, highlighting the thalamus) in type 1 diabetes. The centre panel (B) indicates regions of decreased white matter volume (highlighting the mesial temporal area). The right panel (C) shows areas of decreased tissue T2 relaxation time (highlighting the lentiform nuclei).

1