Alzheimer disease: can the exam predict the pathology?

Neurology 2012;78:374–375 Memory testing beyond the 3-word recall of brief mental status examinations is not routinely performed on patients with memory complaints. This is due to time constraints and uncertainty over how to conduct and interpret a valid learning and memory examination. Failure to evaluate memory complaints consistently with objective assessments has no doubt contributed to the impression that memory testing lacks specificity for Alzheimer disease (AD). Memory disorders specialists who frequently diagnose AD recognize the value of a targeted memory examination. In this issue of Neurology, Wagner et al.1 reinforce this by using memory tests that specifically probe mesiotemporal function in elderly individuals with memory deficits, demonstrating that these tests predict CSF A 42/tau levels. The authors found that subjects who performed poorly on tests of mesiotemporal function had low CSF A 42/tau ratios. Other studies have shown low CSF A 42/tau ratios, in turn, associate with AD.2 By establishing a neuroanatomic localization for the patient’s complaint, therefore, an appropriately applied memory examination essentially changes AD from a diagnosis of exclusion to one of inclusion. For elderly individuals with memory complaints, bilateral mesiotemporal dysfunction is usually due to AD. The authors studied subjects with mild cognitive impairment (MCI), and analyzed data from CSF biomarker testing, the logical memory paragraph recall test from the Wechsler Memory Scale–Revised (LM), the word list learning task from the Consortium to Establish a Registry for Alzheimer’s Disease neuropsychological battery (CERAD-NP), and the Free and Cued Selective Reminding test (FCSRT).1,3 The A 42/tau ratio was reduced in 40% of the 185 evaluated subjects. According to new diagnostic criteria, these 40% qualified for a diagnosis of prodromal AD4 or MCI due to AD.5 All 3 memory tests correlated with a diagnosis of MCI due to AD, because poor test performance predicted a low CSF A 42/tau ratio. The FCSRT discriminated between the different MCI groups better than the LM and CERAD-NP. Why did the FCSRT outperform the LM and CERAD-NP? The FCSRT uses a cued recall paradigm to assess memory.6 A card containing 4 pictures is presented to the subject. A semantic cue is provided for each picture. The card is removed and the cues are used to ensure each picture had an adequate chance to be encoded. This process is repeated with 3 more cards. Over just a few minutes the subject is exposed to 16 pictures and 16 semantic cues. After the last card, the patient counts backwards by 3s for 20 seconds; this distraction task cleanses the pictures from the subject’s prefrontal lobe–mediated working memory. The subject then states, without prompting, as many pictures as possible. This results in that trial’s “free recall” score. The previously presented semantic cues are then used to elicit the remaining pictures and generate a “cued recall” score for that trial. The sum of the free and cued items that were successfully remembered constitutes the “total recall” score. Two more trials are immediately performed and the free, cued, and total recall results from the 3 trials are summed. Approximately 15 minutes later free, cued, and total recall are probed to provide “delayed recall” scores. The FCSRT delayed total recall score effectively predicted CSF biomarker profiles, while the FCSRT delayed free recall score was far less accurate.1 This suggests cuing paradigms interrogate functional or anatomic substrates that are minimally engaged by non-cue-based memory tests. In particular, the authors speculate that semantic cuing changes the test from one of hippocampal function to one of hippocampal plus entorhinal cortex function. This is based on evidence that argues the entorhinal cortex mediates associations between contextually related information fragments.7 In other words, semantic cuing strategically engages the entorhinal cortex, a mesiotemporal region affected in AD.8