Apolipoprotein E genotype and rate of decline in probable Alzheimer's disease.

Background: The risk of Alzheimer's disease (AD) appears to increase, and the age at onset to decrease, with the number of \g=elunate\4 alleles. If this relationship is due to increased rate of pathophysiological change, the presence of \g=elunate\4 would be expected to influence progression of disease, predicting a more rapid decline with increasing number of \g=elunate\4 alleles. Objective: To determine if the frequency of the \g=elunate\4 allele of the apolipoprotein E (ApoE) gene affects the rate of clinical progression in AD.

for age at onset did not affect the results.
Conclusions: The ApoE genotype does not strongly in- fluence the rate of decline in AD, implying that \ g=el unat e\ 4 might predispose to the development of the disease without ac- celerating its pathogenesis or progression.The effects of \ g=el unat e\ 4on both age at onset and rate of decline need to be further investigated.(Arch Neurol.1996;53:345-350)   The recent discovery of an overrepresentation of the e4 alíele of the apolipopro¬ tein E (ApoE) gene in Alz¬ heimer's disease (AD)1"3 has opened a whole new area of investigation and new hope for the discovery of at least one cause of AD.The gene coding for ApoE, a lipid transport protein, is lo¬ cated on chromosome 19 and has three al¬ íeles-e2, e4, and e3-the last being by far the most frequent.4Almost 50% of spo¬ radic cases of AD3 and more than 65% of late-onset familial cases1 carry the e4 al¬ íele, compared with approximately 20% in the normal population.1•5"7The strength of this association, replicated by many groups,813 makes ApoE e4 the major ge- netic susceptibility factor thus far identi¬ fied for AD. 14 Furthermore, the risk of AD appears to increase, and the age at onset to decrease, with the number of e4 alíeles carried by an individual.5 •11 It has been postulated that ApoE e4 plays a role in the pathogenesis of AD ei¬ ther by influencing the rate or amount of amyloid deposition in the brain1•2 or by promoting neurofibrillary tangle forma¬ tion. 15Regardless of mechanism, ApoE al¬ íele e4 affects sporadic and late-onset fa- Sec Subjects and Methods on next page From  Stroke-Alzheimer's Disease and Related Disorders Asso¬ ciation for probable or possible AD.25 Medical history, neu¬ ropsychological evaluation, neurological and psychiatric ex¬ aminations, and appropriate laboratory studies constituted the initial evaluation for entry into the ADRC.Exclusion criteria were current or past history of major mental ill¬ ness, current alcohol or drug abuse, or central nervous sys¬ tem disorder (eg, stroke, epilepsy, severe traumatic brain injury, or Parkinson's disease).Extrapyramidal signs in¬ sufficient to diagnose Parkinson's disease or marked be¬ havioral disturbance did not exclude entry into the study.Procedures were fully explained to the patients and their guardians, and informed consent was obtained before en¬ rollment.At each semiannual visit to the ADRC, interim histories were taken and neurologic examination and neu¬ ropsychological evaluations were repeated.This informa¬ tion was used to monitor each patient's disease course and to determine whether clinical diagnoses should be recon¬ sidered.
Patients were eligible for the present study if they (1) were given the diagnosis of probable AD at study entry and at each subsequent semiannual evaluation, (2) met the cri¬ teria of the Consortium to Establish a Registry for Alzhei¬ mer's Disease (CERAD)26 for definite AD (if an autopsy was performed), and (3) had DNA available from fresh blood samples or frozen brain tissue.A total of 101 patients met the above criteria.Thirty-two percent were male and 90% were white.The average education of the subjects was 12.9 years, and 55% of the patients had a family history of de¬ mentia.The mean age at onset was 65.6 years, with an av¬ erage illness duration of 4.1 years and a mean Mini- Mental State Examination (MMSE)27 score of f 4.7 at entry.
For analysis of the influence of ApoE genotype on rate of decline, the subjects had to meet the additional criterion of having an MMSE score of at least 10 at study entry.Sev¬ enty-eight patients (35% male and 88% white, with an average education of 13 years) met this additional crite¬ rion; 53% of these 78 patients had a family history of de¬ mentia.The mean age at onset was 66.1 years, with 3.6 years of illness and a mean MMSE score of 17.0 at entry.

PROCEDURES
Clinical and cognitive data collected during semiannual vis¬ its were used to define illness progression.The following mea¬ sures had sufficient and necessary longitudinal data for assessing rate of decline: MMSE, Spatial Delayed Recogni¬ tion Span,28 Boston Naming Test,25 Category Fluency Test (cities, animals, and foods),30 and Physical Capacity sub- scale of the Psychogeriatric Dependency Rating Scale (PGDRS).31We did not consider several other measures.Lon¬ gitudinal data on the Block Design subtest of the Wechsler Adult Intelligence Scale-Revised32 were not available be¬ cause the test was given only at baseline and not adminis¬ tered to patients with a MMSE score under 15.Two other scales from the PGDRS also were not included in assessing the rate of illness progression.The Behavior Disorder scale was not used because scores do not increase monotonically across the illness; eg, behavioral disturbance would de¬ crease with neuroleptic therapy and at end-stage of the ill- milial AD in such a way that it leads to earlier symptom expression.If this relationship is due to an increased rate of pathophysiological change, ApoE genotype may also influence the rate of clinical progression in AD.A few studies have been published on rate of decline in AD as a function of ApoE genotype, and these have yielded discordant results.Frisoni et al16 reported slower cogni¬ tive decline with increasing doses of e4 in a sample of subjects with late-onset sporadic AD (aged 70 years or older).Progression was estimated by retrospective de¬ termination of both age at onset of symptoms and as¬ sumed baseline cognitive performance.Besides the po¬ tential limitations of retrospective determinations, Frisoni and colleagues' study was highly selective as to the age range of their subjects and excluded subjects with late- onset familial AD, a group in which the effect of e4 has been demonstrated.Studying autopsy-confirmed cases of AD, DeKosky et al17 reported improved survival in e4- positive patients with AD, but no effect was found on neu¬ ropsychological measures.These findings might sug¬ gest a survival bias rather than a pathogenetic mechanism to account for part of the association between e4 and prevalent AD.Another study of autopsy-confirmed cases of AD failed to find a correlation between genotype and rate of clinical progression, although the presence of e4 was found to correlate strongly with amyloid burden.18 Faster disease progression in e4 carriers has been re- ported in three other studies, although short follow-up periods19•20 or small sample size21 make it difficult to in¬ terpret the results.When progression was analyzed as measured by survival, two studies found longer survival in e4 carriers,22,23 while the e2 genotype was associated with increased risk of disease and mortality in earlyonset AD. 23 The lack of conclusive findings is probably due, in part, to the inherent methodological difficulties of longitudinal studies of rate of decline. 24e sought first to determine ApoE allelic distribu¬ tion in subjects from the Johns Hopkins Alzheimer's Dis¬ ease Research Center (ADRC), Baltimore, Md, and then to test the hypothesis that the presence of ApoE e4 af¬ fects the rate of clinical decline in AD.To this end, we used random-effects regression models to analyze the clini¬ cal and neuropsychological decline of our cohort of pa¬ tients with AD, who were prospectively followed up for approximately 10 years.

RESULTS
Of the 101 patients with AD who were genotyped, 14 (13.9%) were homozygous for the ApoE e4 alíele, 56 (55.4%) were heterozygous (e3/e4 or e2/e4), and 31 (30.7%) had no e4 alíeles.The frequencies of the alíeles are presented in Table I.The baseline characteristics of the patients by number of e4 alíeles are listed in ness.The Orientation scale of the PGDRS was also not used because many patients were placed in nursing homes dur¬ ing the course of the study.The decision for institutional care often is based on families' social circumstances as much as on level of dementia.All the measures described above were used to compare baseline scores of groups of patients de¬ fined by ApoE genotype.GENETIC TESTING Genomic DNA was extracted from white blood cells in liv¬ ing patients and from frozen or fixed brain tissue in au¬ topsy cases.The ApoE genotyping was conducted accord¬ ing to the method of Hixson and Vernier. 33Briefly, the DNA region of interest was amplified by polymerase chain reac¬ tion with primers F4 and F6 and then digested with the Hha I restriction enzyme.Digestion products were separated on 3% agarose gel (NuSieve, FMC Bioproducts, Rockland, Me), stained with ethidium bromide, and visualized and photo¬ graphed under UV illumination.The pattern of migration of the digestion fragments is indicative of ApoE genotype.STATISTICAL ANALYSIS Random-effects regression models34 were used to predict the change in clinical and cognitive measures associated with the presence of ApoE e4 over time.Each regression equa¬ tion included a random intercept term for the individual's performance on the outcome variable at baseline, a covari- ate (ApoE genotype), visit number, and a term represent¬ ing the interaction between the visit number and the covar- iate.A significant interaction term indicates that ApoE genotype significantly predicted the change in the outcome measure over time.The random-effects regression models were fit using SAS (Statistical Analysis System).35Separate equations were computed to assess the influence of ApoE genotype on rate of decline on each outcome variable.
Because our statistical model assumes linear change, data on each measure were considered from study entry until a "floor" was achieved.For the MMSE, scores were no longer considered after the first score of 5.For the Physi¬ cal Capacity subscale of the PGDRS, scores were no longer considered after the first score of 32 or more (maximum score is 36, indicating complete physical disability and to¬ tal sensory impairment).For the Category Fluency Test, scores were no longer considered after the first total score of 2 (indicating zero items generated for one or two cat¬ egories).For the Delayed Spatial Recognition Span, scores were no longer considered when the average of three tri¬ als was less than 1.There were no apparent floor con¬ straints for the Boston Naming Test.Thus, the data considered likely capture a phase of the disease during which decline is linear, even though de¬ cline across the entire illness may be nonlinear.36Further¬ more, a previous investigation of a subset of ADRC pa¬ tients found that a linear model of decline provides the best fit to their longitudinal data on the MMSE. 37ach subject had on average 6.3 observations for the MMSE (range, 1 to 14), 6.4 for the Delayed Spatial Recog¬ nition Span (range, 1 to 14), 6.1 for the Category Fluency Test (range, 1 to 15), 7.1 for the Boston Naming Test (range, 2 to 15), and 11.6 for the PGDRS Physical Capacity sub- scale (range, 2 to 18).
Finally, ApoE genotype was added to the multivari¬ ate model of predictors of decline in MMSE score in our previous study.24This was done to assess the predictive strength of ApoE genotype in relation to other predictor variables.
Table 2.At study entry, the three subgroups (€4/e4, e3/e4 or e2/e4, and no e4) were similar in all respects except for estimated age at onset of symptoms.Age at onset of symptoms was greatest among the group heterozygous for €4 (mean age, 67.2 years), followed by the group ho- mozygous for e4 (mean age, 65.4 years) and the e4- negative group (mean age, 62.4 years).It is of note, how¬ ever, that the mean age at onset was relatively early in all three patient groups.
Because the presence of ApoE e4 has been associ¬ ated with late-onset familial and sporadic AD, but not early-onset familial AD, we also evaluated the character¬ istics of the three subgroups, excluding patients with dis¬ ease onset before the age of 60 years5 to eliminate po¬ tential early-onset familial cases.In these subjects, the age at onset was greatest among the group homozygous for e4 (70.5 years), followed by the group heterozygous for e4 (70.1 years) and the e4-negative group (68.6 years).Seventy-eight of the 101 genotyped patients met cri¬ teria for the analysis on rate of decline.To determine whether these subjects differed from the rest of the co¬ hort, we compared baseline characteristics.Since one cri¬ terion for this study was an MMSE score of 10 or more, these subjects had significantly higher MMSE scores at entry and had been ill for a shorter time (Table 3), sug¬ gesting that individuals who were eliminated from the analysis because of their low MMSE scores were similar except that more time had elapsed since the onset of the disease.Thus, they did not have a more rapid rate of de¬ cline.
Table 4 lists the rates of decline by ApoE geno¬ type as estimated by random-effects regression models.Patients with one ApoE e4 alíele declined more rapidly on the MMSE and the Category Fluency Test than those with no e4 alíeles.Patients homozygous for e4 declined more rapidly on the Physical Capacity subscale of the PGDRS.Because the groups differed in terms of age at onset, and because earlier age at onset has been found to predict more rapid illness progression, we also per¬ formed random-effects regressions covarying for age at onset.The findings remained significant.
Presence of ApoE e4 was added to the final model of predictors of rate of MMSE decline derived in a pre¬ vious study of this cohort through the use of backward stepwise multiple regression.24Variables in this model were handedness, education, family history of demen¬ tia, and performance on the Boston Naming Test, the Gol- lin Incomplete Figures Test, Block  The equation was similar in the present sample, but not completely replicated.The ApoE variable was not sig¬ nificant when considered simultaneously with the vari¬ ables previously found to predict rate of decline.Be¬ cause a family history of dementia and ApoE genotype  fP<. 05uld be related, we also examined the effects of e4 after eliminating family history from the model.Again, the pres¬ ence of ApoE e4 did not predict rate of decline.The ApoE genotype did not emerge as a significant factor in this model when entered as a dichotomous variable (ie, e4 positive vs e4 negative) or a trichotomous variable (ie, no, one, or two e4 alíeles).

COMMENT
In AD, as in many dementia syndromes, clinical variabil¬ ity is the rule rather than the exception.Variations in age at onset, earliest symptoms, pattern of cognitive and func¬ tional impairment, presence of psychiatric and neuro¬ logic complications, and rate of decline are pro¬ nounced.Many researchers in the field have therefore suggested that AD is not a single disease but a clinical syndrome caused by many, possibly interacting, dis¬ eases with distinct pathogenetic mechanisms resulting in a relatively uniform histopathologic appearance.38•39Undoubtedly, the findings on ApoE in AD have sug¬ gested new perspectives in the search for causative fac¬ tors and pathogenetic mechanisms.Even the role of other putative causative factors (eg, head injury,40"43 family his¬ tory of Down's syndrome,44 maternal age at birth,45  jP<.05 when compared with rest of cohort.<.07 when compared with rest of cohort.|P<.07 when compared with no ApoE e4. posure to toxins and metals,4748 lack of education,49 52 and age53) may need to be revisited in light of the ApoE geno¬ type findings.
Several hypotheses on the mechanism of €4 as an etiologic agent in AD have been posed,1215 but none is completely satisfactory at this point.If we believe that the clinical picture of AD is the result of a gradual accu¬ mulation of pathologic changes in the central nervous system (whatever these might be), we could hypoth¬ esize that a gene might continue operating over time, pro¬ ducing more and more of these changes and leading to increasing severity of disease.This would be particu¬ larly true for a genetic factor such as e4, which may have its greatest influence in long-lived individuals.If e4 causes earlier age at onset of AD by means of faster accumula¬ tion of histopathologic changes that lead to clinical symp¬ toms, it might also result in a more rapid rate of decline.In addition, it could be hypothesized that this rate would be even faster in individuals homozygous for e4, consis¬ tent with the gene-dose effect and earlier age at onset.
Alternatively, ApoE could be influencing age at onset in AD by virtue of a mechanism other than progressive ac¬ cumulation of lesions.For example, ApoE may interact at a given point in time with other etiologic factors (eg, head trauma, other genes, and environmental factors), triggering a cascade of neurodegenerative effects that sub¬ sequently continue independently of ApoE genotype.
Our study represents one of the largest investiga¬ tions of ApoE and rate of decline in AD.The size of the sample, together with the extended (10-year) follow-up of patients, is unique.In this study, ApoE genotype did not relate significantly to clinical characteristics of pa¬ tients at baseline, and only very small trends for more severe disease could be identified from the MMSE and the Category Fluency Test.The scores of the patients with one ApoE e4 alíele declined more rapidly on the MMSE and the Category Fluency Test than did those of the pa¬ tients with no e4 alíeles.On the other hand, this effect was not found in patients with two "doses" of the impli¬ cated gene.In these patients, rate of change in physical capacity was more rapid than in subjects without an e4 alíele, but was not significantly different in the hetero¬ zygous individuals.Although these findings are statisti¬ cally significant, the differences between the groups in rate of decline were not consistent or large and there was no gene-dose effect.Furthermore, ApoE €4 was not a sig¬ nificant factor when included in a multivariate model with other variables previously found to predict rate of de¬ cline in this cohort.Thus, other variables, such as fam¬ ily history of dementia or pattern of cognitive impair¬ ment, appear to account for more of the variance in the rate of decline in MMSE scores in AD than does ApoE genotype.
The results of this study are far from excluding the possibility of an effect of e4 on rate of decline.In a dis¬ ease often spanning decades, a 10-year follow-up period might not represent a sufficiently long interval to detect a small effect that is dispersed over the course of time or perhaps evident at the earliest stages of the disease.It is possible, therefore, that e4 has a small but consistent ef¬ fect that our period of study did not allow us to detect.In support of this hypothesis, our measures of global cog¬ nitive and functional impairment, probably most sensi¬ tive to overall decline,54 did show a small but significant relationship to ApoE genotype.
Although we could not demonstrate a gene-dose ef¬ fect, there were only a small number of e4 homozygous individuals in our sample, limiting our statistical power.Study of a much larger sample of homozygous patients would be necessary to arrive at a more definitive con¬ clusion.
Study of the baseline characteristics of our patients did not confirm earlier age at onset with increasing doses of e4.In fact, we found a trend in the opposite direction.
In particular, the group of patients heterozygous for e4 had an older age at onset, while the youngest age at on¬ set was seen in patients with no e4 alíeles.This trend re¬ mained after patients with symptom onset before the age of 60 years were eliminated.The cutoff point of 60 years of age was chosen so that we could compare our results with those of the seminal study5 in which gene-dose ef¬ fects of e4 on age at onset were first demonstrated.We note, however, that the finding of an effect of e4 on age at onset has not been universally replicated.Indeed, the results of at least two other studies, those of Duara et al55 and, most notably, Corder et al (1995),n fail to confirm this finding.
Several methodological limitations of our study are acknowledged.First, our sample comes from a referral- based population through geriatric neurology and neu¬ ropsychiatry clinics.It was mainly composed of whites with a higher-than-average education who also tend to have more dedicated spouses or caregivers than do pa¬ tients in population-based studies.In addition, while many studies use the date of diagnosis of dementia to define the onset of disease, we used the time of onset of symp¬ toms, as established retrospectively with highly reliable methods,56 as our disease-onset date.This way to deter¬ mine age at onset might also contribute to the younger ages of our subjects compared with the subjects of other studies.
Except for mean age at onset, our sample appeared to be typical of AD cohorts, including the frequency of the ApoE e4 alíele.Our cohort did not include more cases with a family history of dementia than other studies; there¬ fore, it is unlikely that earlier age at onset is the result of "contamination" with a high number of early-onset fa¬ milial AD cases, in which other genetic causes may be implicated.
In summary, the effect of e4 on both age at onset and rate of decline needs further investigation.If repli¬ cated, the overall finding of minimal effect on clinical pro¬ gression might suggest that the presence of e4 affects de¬ velopment of AD by mechanisms other than the progressive accumulation of histopathologic lesions.Sig¬ nificant methodological difficulties in the longitudinal study of AD complicate the evaluation of rate of decline.

(
Dr Troncoso), The Johns Hopkins University School of Medicine, and the Department of Biostatistics, The Johns Hopkins University School of Public Health (Dr Brookmeyer), SUBJECTS AND METHODS SUBJECTS Between November 1984 and March 1987, the ADRC en¬ rolled for longitudinal study 209 patients who met the cri¬ teria of the National Institute of Neurological Disorders and Design, the Benton Visual Retention Test, and the Responsive Naming Test.

Table 2 .
1. Apolipoprotein E Allelic Frequency for 101 Patients With Alzheimer's Disease Baseline Characteristics of 101 Patients * Values other than percentages are expressed as mean (SD).

Table 3 .
Baseline Characteristics of All 131 Patients From the Initial ADRC Cohort Not Included in the Analysis of Rate of Decline and the Subset of 78 Patients Meeting All Inclusion Criteria (MMSE Score at Entry of 10 or More), for Rate of Decline Analysis* Values other than percentages are expressed as mean (SD).ADRC indicates Alzheimer's Disease Research Center: MMSE, Mini-Mental State Examination. *

Table 4 .
Rate of Decline on Neuropsychological The rate of decline is expressed in points per year.MMSE indicates Mini-Mental State Examination; PGDRS, Psychogeriatric Dependency Rating Scale.+P<.05 when compared with no ApoE e4. *