Alzheimer's Disease (AD) is a neurodegenerative illness with frequency of occurrence increasing with old age (1). If memory loss setting progressively in is here an element associated upfront, other neuro-psycho-cognitive troubles are also associated, as, for example, language impairment which can degenerate into aphasia (2). Language dysfunction is one of the most significant symptoms in terms of incidence on narrativity in AD (3). In the earliest stages, it is slight, but it increases in time through foreseeable phases. Memory dysfunction also interferes with language possibilities (4). Knowing the importance of both these components could help to prepare as well as possible for the consequences on the patient's social life. We have tried, in this monocentric prospective observation and study to evaluate semantic and narrative dysfunction in AD.
The cohort studied were old demented patients living at home with AD followed since at least 3 months by gerontopsychiatrics for cognitive, behavioral and memory dysfunction. During a gerontopschiatric interview, patients and accompanying family were informed completely and loyally and in understandable language how the study would be carried out. With the agreement of the patient and assistant, the results of the neuropsychological tests were computer recorded anonymously. The study protocol was ethically reviewed and approved by Hospital of Brest ethics committee. Computing procedures for data capture and processing followed the recommendations of the French National Commission on Computer Technology and Freedom.
Criteria were determined for inclusion based from AD patients, (MMSE 10) (5), having given written consent together with their family assistants, to take part in this study validated by the Hospital of Brest ethics committee. Only those patients still capable of understanding the instructions for the Barbizet's test, showing symptoms of Alzheimer dementia, i.e. without massive memory impairment and capable of asserting their consent and assent during the test and fit and able in the eyes of the Law, were included. Patients inability to understand information relating to the study, their inability to choose whether or not to take part, or the possible opposition of their family accompanying person as a criterion were decisive for non-inclusion of such patients. Population number reflect the demented persons follow-up during the period of one year of this study.
Semantic memory is the memory necessary for language manipulation and not over vulnerable to amnesia (6). To evaluate semantic and textual disorders in a cohort of AD patients, tests were routinely recorded with memory consultation. Apart from a few sociodemographic data (sex, age, cultural level estimated by the number years of schooling: 1= illiterate ; 2 = can read and write ; 3 = six years schooling ; 4= 11 years schooling ; 5 = more than 11 years schooling). The following are the tests customarily carried out under consultation : Folstein Test (MMSE) (5), Dubois's 5-word test (D5WT) (7), formal fluency (Number of animals mentioned in one minute) semantic lexical fluency (number of words beginning with L named in one minute) (7), DFB (Dubois's Frontal Battery) (8), Cornell's scale (Depression) (9), Barbizet's test (10) or “The lion's tale”.
The « Lion's tale » (presented Table 1) must be read by the examiner slowly. The subject must be informed that it is a long story the point of which is to evaluate his memory capacity. He must be told it is not a matter of repeating the story word for word but just recalling its main lines and as many details as possible. Barbizet recommended an immediate application, an immediate recall, then a delayed recall. In practice, in the present framework, the text was read out at the beginning of the consultation, followed by an immediate recall and then a delayed recall towards the end of the consultation, that is, about one hour later.
So, this short text with 22 sub-items learned sequential, is recalled immediately and repeated later. The items can be organized according to the author of the test into five episodes: escape, crowd movement, woman and child, deal, resolution. Barbizet shows up things from the tale forgotten, intrusions and inversions to be noted. These data will be compared with that of the neurosychological tests. Statistics were established with a SAS program by a bio-statistician, Dr R. Billon. Tests selected were Chi2, Student (un-paired), Pearson's correlations, Partial Least Squares regression, chosen to reduce risks of collinearity (11). With a classic linear model, collinearity between explanatory variables does make the model unstable and influence the results, making them uncertain.
Hundred and fifty-one patients, living at home, on their own or with their families, were included in the study. No cases were recorded of refusal to participate in the study or to having the results observed computerized anonymously. The cohort of demented patients is made up of 102 females and 49 males of average age 80.3 years 6.91. Table 2 resumes the main results observed. Average characteristics did not differ statistically between the males and the females. The formal fluency on demented patients is less impaired than the semantical lexical fluency test (scored respectively 5.74 1. 09 versus 4.41 2. 19; t= 5.60, p 0.01).
The studied population showed more cases of intrusion (n=36) than inversion in the later narration, whether items or episodes combining the two (n=19). Nearly all the inversions were by patients whose MMSE was below 19 and the intrusions by patients whose MMSE was below 15 (Table 3). Admittedly, intrusions are specific to AD pathology, where we find their greatest occurrence. However, these results could suggest that intrusions are essentially linked to executive problems concerning response control. When referring to patients' dysexecutive syndrome (DFT 13; n=77) it is to be remarked that, overall, inversions occur in the deficient group (Table 4). It is also to be remarked that inversions are observed to be more precocious when disease deteriorates textuality and prevents the pragmatics of communication. MMSE for the executive syndrome sub-group (n=77) was 11.9 2.9 against 20.9 2.6 for the less seriously affected group (n=74)
The correlation matrix for the various parameters studied show that many parameters are correlated whence the choice of PLS for regression analysis. The factor studied is the overall score for Barbizet's test with the model including the variables of the previous figure. The formal fluency remains the one that most influences the overall score with a coefficient of 0.224. The higher it is, the greater the success in the test and this influence is almost three times that of the lexical memory. Then, we will retain cultural level (Coefficient: 0.12) and, more accessorily, age (Coefficient: -0.11), but all the others are less interesting.
We are presenting here the results of a preliminary study on textuality in AD based on consultation with the patients in an old-age psychiatric department. The very nature of the department explains the old age of those recruited. MMSE worsening in certain patients is an approach making their understanding of the text difficult and even the value of their consent. The aid of their family accompanying person was then precious and all the tests were those habitually carried out regularly on this type of patient under the specialized consultation of our establishment.
The executive disorders studied by the DFB and cognitive disorders estimated by the MMSE are strongly correlated, also it is difficult to know which verbal consequences result specifically from one of these factors or from the other one. The threshold of 13 was however chosen on one hand so that the sub-populations were balanced, and on the other hand for a clear cut-off with the existence or not of a dysexecutive syndrome connected to frontal disorders. The presence of a such dysexecutive syndrome, in case of late depression (with or without dementia), impacts on the cognitive status (12). The depressive elderly present frequently changes of the verbal fluencies and the deficits of planning connected to frontal disorders (13). The scores on the scale of Cornell are weak in the population of the study presented here. The depression is frequent in particular when the dementia progresses, in particular when the elderly live in nursing home (9). The elderly included in this study lived at home. The scores on the scale of Cornell presented here are low (Table 2 : 11.3 4,5, depression's threshold value 8).
We shall notice in this study, the important progress of the narrative disorders observed through the score in the delayed recall in « the lion's tale », all the more degraded that the cognitive disorders are installed. Intrusions and inversions can result from cognitive disorders as well as from related executive disorders. Formal fluencies weigh in a significant way but in a modest way on the studied model of regression (CR=0.224), so with a lesser scale the cultural level (CR = 0.117), the cognitive disorders of the dementia as well as the linguistic disorders being classically modulated by this one (5, 14).
Even if the PLS model PLS is more influenced by the formal fluency the lexical semantic fluencies are more altered than the formal fluency (Table 2). Find a word beginning with a letter (“L” in French, “S” in English, for reasons of frequency of use of this letter in the beginning of a word) require a strategy of recovery based on the abstract lexical representation (15). The degradation of the formal fluency testifies of a more important disorganization of the functions of the language
The concrete items with strong emotional connotation in “the lion's tale” (A woman, a child, a lion ...) are better memorized than more secondary (For a long time, fixedly ...): even there the emotion appears clearly as the "power steering" of the reason and the memory.
The lexical semantic fluency is more altered than the formal fluency in the Alzheimer's disease. The degradation of the formal fluency testifies of a more important disorganization of the functions of the language. Formal fluencies are less impaired (Imaging or Gist memory process) than Semantic lexical Fluency (Verbatim process) (16). The fluencies impairments have an impact on the memorization of a text, in parallel with developments of cognitive disorders. The cultural level also influences the narrative skills of the patient which remain sensitive to the emotional aspect of a story when assessed in a memory testing. This emotional mobilization is probably an interesting way of motivation in speech therapy, to lead preventively or to maintain lexical and textual capacities in the Alzheimer's disease.