Can episodic memory tasks differentiate semantic dementia from Alzheimer's Disease?

Abstract

The performance of two groups of patients with semantic dementia (SD), with predominant right (SDR) and left temporal lobe atrophy (SDL), was contrasted with that of cases with probable Alzheimer's disease (AD) on a range of standard episodic memory tasks. While the SDL group achieved a good score on a composite visual, but not a verbal, episodic memory measure, the AD and SDR groups were equivalently impaired at visual and verbal memory. The SD, but not the AD, groups were, by definition, impaired on simple tests of semantic memory. Standard verbal episodic memory tests, therefore, failed to discriminate patients with SD from those with probable AD and even visual memory tests may result in misclassification of SDR cases.

Acknowledgments

We thank Tina Emery, Joanna Drake and Sharon Erzinçlioglu for help with data collection. This research was supported by the Alzheimer's Research Trust, UK and the Medical Research Council, UK.

Notes

^aFor this test, data was available for 4 SDR patients, 7 SDL patients, and 8 AD patients.

^aFor these tasks, n = 8 for each group.

^bFor these tasks, n = 9 for each group.

^cFor these tasks, n = 7 for each group.

^dFor these tasks, score is number generated.

^eSDR group did not differ significantly from any other group.

^fSDL group did not differ significantly from any other group.

^gAD group did not differ significantly from any other group.

In two cases (one SDR and one SDL) the Logical Memory task was taken from the Wechsler Memory Scale – Third edition (Wechsler, 1997), in which the second story is repeated and a second recall required. In these two cases, only the percentage retained of the first recall of the two stories was used, in order to provide a more suitable comparison with the other patients.

In four cases (one SDR and three AD), the short version of the RMT (Warrington, 1996), with 25 items in each subtest, had been administered. Thus percentage correct was used in the analysis as a measure of overall performance, in order to allow comparison between the two versions.