CommentariesPrevalence, Incidence and Duration of Braak’s Stages in the General Population: Can We Know?
Section snippets
Hypotheses
These difficulties of the study may, however, not keep the reader’s attention as long as other questions that could all come under the title: “If all these data were fully valid, what would they mean?” Could we take the values, which are presented here, and treat them as if they were standard epidemiological data? In so doing, we would implicitly accept a set of hypotheses that we will try to state here explicitly to facilitate further discussion.
Interaction of the Lesions: Their General Incidence and Prevalence in the Study
The cases have been considered by groups of 5 years. The prevalence of cases with lesions (i.e., the number of cases with at least one lesion, all stages together, compared to the total number) is 78% for the neurofibrillary tangles and 43% for the amyloid lesions. The interaction between the two lesions is obvious: if they were independent, only 43% of the individuals with tangles should have had amyloid lesions; they were 54%. Only 78% of the cases with amyloid should have had tangles; they
A Model of the Progression of the Lesions
Prevalence varies from zero to one and, confirming hypothesis 2 of lesion irreversibility, increases with age. This increase suggests the use of a regression analysis to describe the relationship between time and pathology. Because prevalence values are ratios varying from 0 to 1, direct linear approximation is unsatisfactory and a logistic transformation has to be performed before the calculation of the linear regression; the results may be converted from logit of prevalence to prevalence
Delay Between Neurofibrillary and Amyloid Stages
To evaluate the time delay between the occurrence of neurofibrillary changes and amyloid deposits, we compared the age at which the same prevalence was found for both types of lesions. The point of prevalence 0.5 was chosen because it is easily identified on the curve, often corresponds to its linear portion and is easily calculated with the logistic equation. This value will be referred to later as the age of mid-prevalence. If we were studying longitudinally one cohort of people, the age of
Duration of the Stages
Estimating the chronology of individual stages is more difficult. The number of cases included in each stage except in the last ones (VI and C) is indeed the balance between two flows: the occurrence of new cases (increasing the prevalence) and the passage to more severe stages (decreasing the prevalence). The dynamic equilibrium is impossible to reconstruct if each stage is studied separately. In the last stages, C and V–VI, on the contrary, there is no outflow in the hypothesis of lesion
Incidence of the Stages
The first derivative of the prevalence curve as a function of age is an estimate of the number of new cases per brain-year, i.e., of the incidence. (The derivative of the logistic regression equation is: (a.e−at + b)/(1 + e−at + b)2 where a and b are coefficients of the equation; t = age.) The incidence curves had the shape of a Gaussian function (Fig. 1, right). All the peaks of incidence (Table 2) were of the same order of magnitude (from 2.1 to 2.8% according to the stages). The curves had a
Prevalence of the Stages
The difference between the first curve (all stages) and the second curve (all stages minus the first) gives the prevalence curve, in relation to age, of the first stage (I–II for the neurofibrillary pathology or A for the amyloid changes) and the difference between the third and second curves estimates the prevalence of the second stages (III–IV and B) (Fig. 2). Those curves, also shaped as Gaussian functions, reached a maximum value 10 years after the peak of incidence, except for stage I–II
Relation with Dementia
The clinical data of the autopsied cases were not available. However, the pathological data provided by Heiko and Eva Braak may be compared with the prevalence curves of dementia that have been published. We used the data proposed by Katzman and Kawas [25], and relied on the logistic regression to facilitate the comparison with the pathological data (Fig. 3). The prevalence of dementia in the younger age groups coincide fairly well with the prevalence values of stage V–VI and was clearly below
Hypothetical Progression of the Stages
We may now summarize the main conclusions of the analysis and suggest a sequence of events by taking into account the peaks of prevalence and the mean duration of the stages: around the age of 40, the first neurofibrillary changes appear in the entorhinal region; their incidence increases, reaching a maximum at age 47—by that age, half the population is affected. Then, around the age of 62, amyloid precipitates in the basal neocortex; the maximal incidence of stage A is observed at age 72. By
How Valid Were the Initial Hypotheses?
The “random death” hypothesis cannot be easily accepted. It is, indeed, probable that the mortality rate is higher for cases with lesions than for those who are spared. It is also probable than intellectually deficient patients are more often dying in hospitals, where autopsies can be performed, than at home. This bias, which seems at least plausible, leads to an overestimation of the lesion prevalence rate. It underlines the necessity of undertaking longitudinal study, including postmortem
Discussion of the Consequences of the Model
This discussion has perhaps convinced the reader of the limits of the model. It may also have led the reader to believe that some consequences of these hardly questionable observations will remain valid even if the hypotheses are not fully verified, most noticeably the relation between lesions and aging, the long duration and early onset of the tangle pathology, the dissociation and time-lag between neurofibrillary and amyloid changes. We would like to discuss in more detail several of these
Aging and Disease
The disease, as described in this model, appears so systematically and closely associated with age than one may reasonably ask if the two processes can be distinguished. The extrapolations of the curves over 100 years (an age that obviously has to be taken as the vanishing point in a perspective) converge to a disease prevalence of 1, because the slopes of the estimated prevalence curves of the lesions are positive and steep when the usual life span stops their lines. There is, to our
Relation Between Amyloid and Tangles
The relationship between neurofibrillary changes and amyloid appears asymmetrical in this model. Tangles are more frequent and precocious than most neuropathologists would have probably thought. Moreover they anticipate the amyloid pathology for 27 years. More than half (54%) of the cases with tangles are devoid of Aβ lesions. This is not what was expected in the light of the recent findings concerning APP [8]and presenilin mutation [32], trisomy 21 [31]and the ApoE4 genotype 3, 33all inducing
When Does AD Start? What Is the Significance of Stage I–II?
The high prevalence of stage I–II, even in the young, and its extraordinary long duration had certainly not been fully appreciated until now, although the presence of lesions in young people has already been reported in the literature [34]. Because stage I–II is so common, how can it be so serious? It could be normal aging, and remain stable. This possibility is supported by the long interval between stages I–II and III–IV (around 40 years). The main argument presented by Heiko and Eva Braak
Relationship of the Lesions with Dementia
Finally, we would like to examine the differences between dementia prevalence and the prevalence of the lesions, although many confounding variables may obscure the relation between the lesions and the clinical symptoms, most noticeably the presence of Lewy bodies [21]and vascular lesions [30]. It is usually admitted that the cognitive decline is more significantly correlated with the neurofibrillary tangles than with the Aβ deposits 1, 6, 10, 11, 13, 15, 16, 28. The lesions are thought to be
Conclusions
The length of this article should at least convince the reader of the deep interest these findings have raised. Whatever criticisms, doubts or discussion that arouse, the large amount of data brought by Heiko and Eva Braak will have to be assimilated by the scientific community. They cannot be simply dismissed. We believe that they add highly significant information to our understanding of brain aging and Alzheimer disease.
References (35)
- et al.
Apolipoprotein E allele epsilon 4 is linked to increased deposition of amyloid beta-peptide (A-beta) in cases with or without Alzheimer’s disease
Neurosci. Lett.
(1994) - et al.
Image analysis of β-amyloid load in Alzheimer’s disease and relation to dementia severity
Lancet
(1995) - et al.
Large amounts of neocortical βA4 deposits without Alzheimer changes in a nondemented case
Neurosci. Lett.
(1990) - et al.
βA4 deposits are constant in the brain of the oldest oldan immunocytochemical study of 20 French centenarians
Neurobiol. Aging
(1993) - et al.
Modeling the relation between neurofibrillary tangles and intellectual status
Neurobiol. Aging
(1997) - et al.
Does amyloid precede paired helical filaments in the senile plaque? A study of 15 cases with graded intellectual status in aging and Alzheimer disease
Neurosci. Lett.
(1988) An ‘anatomical cascade hypothesis’ for Alzheimer’s disease
Trends Neurosci.
(1992)- et al.
The pattern of acquisition of plaques and tangles in the brains of patients under 50 years of age with Down’s syndrome
J. Neurol. Sci.
(1989) - et al.
Neurofibrillary tangles but not senile plaques parallel duration and severity of Alzheimer’s disease
Ann Neurol.
(1992) - et al.
Neurofibrillary predominant form of Alzheimer’s diseasea rare subtype in very old subjects
Acta Neuropathol (Berl.).
(1994)