Table 3.

Prediction of cerebral pathologies by retinal pathologies in spontaneously hypertensive stroke-prone rats.

Cerebral pathology	Threshold (weakly vs. strongly affected)	Retinal pathology	Cutoff value	Sensi-tivity	Speci-ficity	J	AUC
Erythrocyte thrombi	≥81 (n = 12 vs. n = 12)	Erythrocyte thrombi	6.5	83.3%	66.7%	0.500	0.816
		AA	21	50.0%	41.7	0.083	0.448
Small perivascular bleeds	≥1 (n = 18 vs. n = 6)	Erythrocyte thrombi	7.5	83.3%	72.2%	0.555	0.546
		AA	17.5	50%	38.9%	−0.111	0.398
CAA	≥30 (n = 9 vs. n = 15)	Erythrocyte thrombi	6.5	55.6%	40%	−0.044	0.489
		AA	21	88.9%	66.6%	0.556	0.793

This table displays the prediction values for cerebral erythrocyte thrombi, small perivascular bleeds and CAA by retinal arteriolar erythrocyte thrombi, small perivascular bleeds and AA. According to a certain threshold value, all animals were grouped into “weakly” and “strongly” affected by the cerebral pathologies; the thresholds and numbers of weakly and strongly affected animals are given in the table. The following retinal pathologies showed the highest combined sensitivity and specificity correlations for the cerebral pathologies: (i) retinal erythrocyte thrombi for cerebral erythrocyte thrombi, (ii) retinal erythrocyte thrombi for cerebral small perivascular bleeds, and (iii) AA for CAA; the respective values are marked in orange.

AA, amyloid angiopathy; AUC, area under the curve; CAA, cerebral amyloid angiopathy; J, Youden Index.