Few clinical tests have had such a meteoric rise in popularity as the Mini-Mental State Examination (MMSE). Originally developed in the early 1970s by a group of psychiatrists with the aim of “serial testing of the cognitive mental state in patients on a neurogeriatric ward”, only a few decades later it has become the most widely used tool for cognitive assessment in a wide range of neurological diseases. As such, it was introduced to the readers of Practical Neurology by Ridha and Rossor.¹ However, as the authors point out in their article, the MMSE has several serious limitations (table 1), such as its over-reliance on verbal cognitive function at the expense of non-dominant hemisphere skills and executive functions. In fact, the MMSE is based almost entirely on verbal assessment of memory and attention. It is insensitive to frontal-executive dysfunction and visuospatial deficits. The assessment of memory and language is very superficial. Moreover, it does not provide qualitative information about the patient’s cognitive profile and hence cannot be used to differentiate between different diseases.

In some patient groups, such as Parkinson’s disease and related disorders, the MMSE as the only method of cognitive assessment might not only be inappropriate, but even misleading; a normal MMSE is likely to lead to the false assumption of preserved cognitive status while in fact the patient might have severe frontal-dysexecutive or visuospatial symptoms.

The limitations of the MMSE are hardly surprising, given the fact that it was created at a time when the routine brain imaging technique was pneumencephalography and CT had not yet entered clinical practice. It was based on a concept of dementia as a unitary syndrome of “a global deterioration of intellect”, irrespective of the underlying brain pathology. In the original sample, all 29 patients with “dementia syndromes due to a variety of brain diseases” were evaluated as a single group, and compared to patients with affective illness, schizophrenia and neurosis.

However since then our understanding of dementia has dramatically changed. It is now widely recognised that, rather than constituting a single entity, it is a label for a variety of neurological conditions associated with different patterns of cognitive deterioration, such as Alzheimer’s disease, frontotemporal dementia, dementia with Lewy bodies, semantic dementia and many others. Specific neuropsychological deficits can be related to atrophy or hypoactivity of circumscribed brain regions on neuroimaging, and characteristic distribution of pathological changes on postmortem examination. An up-to-date cognitive assessment must take these important developments into account.

ADDENBROOKE’S COGNITIVE EXAMINATION

The growing awareness of the insufficient assessment provided by the MMSE has lead to different versions of an “extended MMSE”, in which the core questions have been complemented by additional cognitive tasks, covering a wider spectrum of mental functions. One of the most popular versions is the Addenbrooke’s Cognitive Examination (ACE), designed to give the clinician a brief and simple cognitive screening tool, including the MMSE, but extending it to encompass important areas not covered by it, such as frontal-executive function and visuospatial skills. Indeed, the first version of the ACE, published in 2000,² was demonstrated to be superior to the MMSE in both the detection of dementia and in the differentiation between Alzheimer’s disease and frontotemporal dementia, discussed in more detail below. The ACE has comparable sensitivity to the Dementia Rating Scale,³ a well-established dementia screening tool, widely used in research, but, because of its length and difficulty in administration (for example, needing special stimulus cards etc) rarely applied in the clinical setting.

In the years following the publication of the first version of the ACE extensive clinical and research experience in Cambridge and other centres has highlighted its strengths and weaknesses and lead to several modifications. The revised version, ACE-R,⁴ maintains the overall structure of the test, such as the division into five domains. However, it has better designed individual test items, it is easier to administer, and detailed scoring instructions are provided as well as three parallel versions of the memory test, in order to prevent practice effects.

STRUCTURE

The ACE-R is presented on six pages and no additional stimuli are required. It takes about 15 minutes to administer and consists of 26 tasks, divided into five domains: attention and orientation, memory, verbal fluency, language and visuospatial skills. The maximum total score is 100 (the maximum scores of the individual domains are given in brackets below).

Attention and orientation (18 points)

This domain consists of items adapted from the MMSE: orientation in time and space, registration of three words, and serial subtraction/backward spelling.

Memory (26 points)

The memory assessment goes well beyond the recall of the three items from the MMSE. Not only does it include additional material but the encoded information is recalled after a longer delay (about 10 min, filled with other tasks), making it more sensitive to mild memory impairment. The patient is asked to learn the name and address of a fictitious person (For example, Harry Barnes, 73 Orchard Close, Kingsbridge, Devon) and then to recall it at the end of the tests. If he or she fails to recall one or more items, the recognition part offers three options from which the correct one can be selected. The memory domain also contains four general knowledge questions (for example, current Prime Minister, US president assassinated in the 1960s, etc).

Verbal fluency (14 points)

Although not specific (it can be influenced by many other factors) verbal fluency is probably the most sensitive cognitive screening test that can be completed within 60 seconds, a kind of “cognitive equivalent of the erythrocyte sedimentation rate”. The patient is asked to produce as many different words as possible in a minute, according to specified rules (for example, no proper names). The domain consists of two parts: letter fluency (as many words as possible starting with the letter “p”) and category fluency (as many different animals as possible).

Language (26 points)

In addition to the language-related items from the MMSE, this domain contains repetition of single words and phrases, naming 10 pictures of low frequency items (for example, anchor, rhinoceros), answering four questions related to these items (for example, point to the item with a nautical connection) and reading five irregular words (for example, pint, height).

Visuospatial skills (16 points)

This domain consists of two parts: the first is the drawing from the MMSE of overlapping pentagons, and also a cube and a clock. The second part consists of counting dots (4 arrays) and naming four incomplete letters. It requires only simple motor responses and can, therefore, be used in patients whose drawing is impeded by different types of motor impairment, such as paresis, rigidity or apraxia.

INTERPRETATION

Total score

The ACE-R provides age and education dependent norms for the total score as well as for the individual subtests.⁴ Based on a validation study two cut-off scores were suggested for clinical practice. The higher cut-off score of 88 has a better sensitivity for the detection of dementia (94%) but a lower specificity (89%). The lower cut-off score of 82 misses some dementia cases (sensitivity 84%), but is 100% accurate in classifying normal controls as non-demented.

Individual domains

When interpreting the results of the individual domains it is important not to draw conclusions from isolated subtests but to compare the overall pattern of performance across all five domains. In this respect, cognitive bedside assessment is very much like the neurological examination; it might be at times difficult to decide whether an isolated reflex should be considered abnormally brisk, suppressed or within the normal range. However, a clear side-to-side difference, or a difference between upper and lower limbs, might provide valuable diagnostic information.

Two cognitive domains, which often demonstrate a contrasting pattern of impairment, are memory and verbal fluency.⁵ This can be illustrated by the comparison of the cognitive profiles in patients with Alzheimer’s disease and progressive supranuclear palsy. Although both groups have the same level of overall impairment, the pattern of deficits is radically different: the most pronounced impairment in Alzheimer patients is in memory, in the progressive supranuclear palsy group it is in verbal fluency (fig).

EXAMPLES OF THE ACE IN DIFFERENTIAL DIAGNOSIS

Frontotemporal dementia, Alzheimer’s disease, and mild cognitive impairment

The original description of the ACE² not only established it as a sensitive screening test for dementia, but also demonstrated that it could successfully distinguish between Alzheimer’s disease and frontotemporal dementia (table 2). The best differentiation was reached using the ratio of verbal fluency + language/orientation + memory (VLOM): a ratio >3.2 is in favour of Alzheimer’s, a ratio <2.2 suggests, in contrast, frontotemporal dementia. Using the same criteria as Mathuranath et al,² Mioshi et al⁴ found that the sensitivity and specificity of the VLOM ratio with the ACE-R was virtually identical to that of the first ACE version.

The ACE can also be used to predict the likelihood of conversion of mild cognitive impairment into Alzheimer’s disease; the ACE subtests of category fluency and naming are the best predictors of subsequent conversion.⁶ The authors attributed this sensitivity to the fact that the ACE contains measures of semantic as well as episodic memory, both of which are known to be impaired early in the course of Alzheimer’s disease. Indeed, a recent study found that mild cognitive impairment patients had difficulty not only with episodic memory but also with verbal fluency and language.⁴

Atypical parkinsonian syndromes

An area in which the ACE has been particularly useful is that of atypical parkinsonian syndromes, such as progressive supranuclear palsy, multiple system atrophy, corticobasal degeneration and dementia with Lewy bodies. The main cognitive changes in these patients affect frontal-executive and visuospatial functions—exactly the kind of problems not evaluated by the MMSE (table 1). In a longitudinal follow-up study of pathologically confirmed progressive supranuclear palsy patients in Cambridge, we observed many cases with severe cognitive impairment, in particular an extreme reduction in verbal fluency, but with an entirely normal MMSE score of 29–30/30.³ In multiple system atrophy the cognitive deficits tend to be similar in pattern to that of progressive supranuclear palsy, but less pronounced. The cognitive picture of corticobasal degeneration and dementia with Lewy bodies is more complex and characterised, in addition to a reduction in verbal fluency, by deficits in attention/orientation (dementia with Lewy bodies), language (corticobasal degeneration) and visuospatial functions (corticobasal degeneration and dementia with Lewy bodies).⁵

Dementia versus affective disorder

The ACE can also be helpful in distinguishing progressive degenerative dementia from transient cognitive dysfunction associated with affective illness. Dudas et al compared patients with major depression and affective illness not reaching DSM criteria for depression with patients with a documented neurodegenerative dementia (Alzheimer’s and frontotemporal dementia).⁷ Although the affective group did have mild deficits in memory and letter fluency, their mean overall score (89) was only slightly lower than normal controls (94) and significantly higher than Alzheimer’s (62) and frontotemporal dementia (74) patients. Interestingly, in 15 out of 16 patients with a questionable dementia and an ACE total score <88, the two-year follow-up confirmed the diagnosis of a progressive degenerative dementia.

HOW TO GET THE ACE-R

The ACE-R is available, free of charge, together with brief administration and scoring instructions, from Eneida Mioshi (eneida.mioshi{at}mrc-cbu.cam.ac.uk) and also on the Practical Neurology website (http://pn.bmj.com/supplemental). Eneida can also be contacted to obtain the most up-to-date information on ACE and ACE-R versions in languages other than English (table 3).