| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a Andrus Gerontology Center, University of Southern California, Los Angeles
b Department of Psychology, University of Southern California, Los Angeles
c Department of Medical Epidemiology, The Karolinska Institute, Stockholm, Sweden
d Department of Psychology, University of California, Riverside
e Institute of Gerontology, Jönköping, Sweden
Margaret Gatz, Department of Psychology, University of Southern California, 3620 McClintock Ave., Los Angeles, CA 90089-1061 E-mail: gatz{at}rcf.usc.edu.
Decision Editor: Margie Lachman, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
ALTHOUGH normal aging is associated with decline in some areas of cognition (Salthouse 1999
), older adults who later develop dementia seem to demonstrate preclinical cognitive changes that can be discriminated from normal aging (e.g., Elias et al. 2000; Jacobs et al. 1995; Johansson and Zarit 1997; La Rue and Jarvik 1987; Nielsen, Lolk, Andersen, Andersen, and Kragh-Sorensen 1999; Storandt, Botwinick, Danzinger, Berg, and Hughes 1984). Although researchers agree that some signs of cognitive decline predict the onset of dementia long before its clinical onset (Fabrigoule et al. 1998; Flicker, Ferris, and Reisberg 1991; La Rue and Jarvik 1987; Linn et al. 1995; Nielsen et al. 1999), there is no clear evidence for which cognitive measures best identify those who will develop dementia. Preclinical signs of dementia have most often been associated with tests of verbal learning and delayed recall (Fox, Warrington, Agnew, and Rossor 1998; Fabrigoule et al. 1998; Linn et al. 1995; Masur, Sliwinski, Lipton, Blau, and Crystal 1994; Tierney et al. 1996), category verbal fluency (Dartigues et al. 1997; Fabrigoule et al. 1998; Masur et al. 1994; Monsch et al. 1992; Storandt et al. 1984), perceptual speed (Fabrigoule et al. 1998; Masur et al. 1994; Storandt, et al. 1984), and attention or mental control (La Rue and Jarvik 1986; Nielsen et al. 1999; Storandt et al. 1984; Tierney et al. 1996). Still, effect sizes tend to be small and findings inconsistent across studies even when identical measures are used.
Studies by Jorm 1986, Fabrigoule and colleagues 1998, and Amieva, Rouch-Leroyer, Fabrigoule, and Dartigues 2000 represent attempts to search for a pattern that could guide future research efforts and help clinicians distinguish early signs of dementia from normal aging. They suggest that the preclinical phase of dementia may be associated with cognitive deficits in controlled, but not automatic, processing (Shiffrin and Schneider 1977). Controlled information processes rely on intact attentional resources and involve voluntary retrieval and integration of information. Automatic processes generally involve well-learned, spontaneous responses and require relatively less cognitive effort. This distinction may explain, for example, why some tests of verbal skills such as category verbal fluency and word finding seem to decline prior to dementia onset (e.g., Dartigues et al. 1997; Jacobs et al. 1995; Monsch et al. 1992; Storandt et al. 1984), whereas more automated verbal skills such as reading remain fairly intact (Fox et al. 1998; Monsch et al. 1992).
Over the past two decades, early and accurate detection of preclinical cognitive changes became of interest in dementia research (Masur et al. 1994). Early detection of dementia has also assumed greater medical value with the emergence of new preventive and treatment strategies. For example, use of antioxidants Vitamin C or E (Morris et al. 1998) and other oxidation inhibitors (Munch et al. 1998) may lower the risk of Alzheimer's disease in healthy older adults. Early detection of developing dementia and consequent early implementation of treatment may slow down the neurodegenerative process and, in turn, delay dementia onset or slow down its progression.
Detection of early signs of dementia with cognitive tests has often been thwarted by the confounding effect of demographic variables such as age and education (e.g., Jacobs et al. 1995; Slooter et al. 1998). In addition, genetic factors may affect cognitive performance (Brandt et al. 1993; McClearn et al. 1997; Pedersen, Plomin, Nesselroade, and McClearn 1992; Swan, La Rue, Carmelli, Reed, and Fabsitz 1992). For example, Swan and colleagues 1992 suggested that low scores on Digit Symbol may reflect lower inherent intellectual level rather than signs of dementia. Matching older adults on genetic and demographic variables may reduce the possibility that these confounding factors, rather than incipient dementia, are responsible for observed cognitive deficits.
Conclusions based on findings from twin studies may increase our confidence in accurate detection of predictors of dementia among cognitive tests. Twin comparisons enable researchers to control for genetic and environmental factors, including those affecting cognitive performance. In addition, twins are completely matched for age and gender (except for opposite fraternal twins), and tend to be similar in education (Lichtenstein, Pedersen, and McClearn 1992).
Controlling for genetic factors affecting cognitive performance can be particularly important in dementia research. Twin researchers suggest that heritability for cognitive performance is high throughout adulthood (Bouchard, Lykken, McGue, Segal, and Tellegen 1990; Finkel, Pedersen, and McGue 1995; Finkel, Pedersen, Plomin, and McClearn 1997; Patrick 2000; Pedersen et al. 1992; Plomin, Pedersen, Lichtenstein, and McClearn 1994), although heritability estimates seem to decrease in very old cohorts (Finkel et al. 1995,Finkel et al. 1997; McClearn et al. 1997). Studies of specific cognitive abilities suggest heritabilities around 65% in older adult samples; for example, for tests of verbal skills (McClearn et al. 1997; Pedersen et al. 1992; Plomin et al. 1994; Swan et al. 1999), perceptual speed (McClearn et al. 1997; Pedersen et al. 1992), spatial skills (Pedersen et al. 1992), and memory (Finkel and McGue 1993; McClearn et al. 1997), and Mini-Mental State Examination (MMSE) scores (Swan et al. 1990). High heritability of cognitive abilities means that monozygotic twin pairs are more similar than dizygotic twin pairs. However, dizygotic twins also have substantial intrapair correlations. Thus, within both monozygotic and dizygotic twins, the two members of the pair can be expected to show similar cognitive performance. If differences in cognition become noticeable in one member of the pair, these differences may signal pathological change.
In the present study, we investigated differences in cognitive performance in twin pairs who were initially nondemented, but in which one of the twins went on to develop dementia. We tested the hypothesis that the twin who would later become demented would score lower than the twin partner who continued to be nondemented on cognitive tests that predominantly involve controlled processing and require attention and unassisted retrieval of information. On the other hand, twin partners would score similarly on tests of automatic information processing for which performance is automated and/or retrieval cues are provided.
In addition, we compared the cognitive scores of the nondemented twin partners with the scores of matched individuals randomly selected from a pool of nondemented twin pairs. We know that concordance for dementia is quite high. For example, in the Bergem, Engedal, and Kringlen 1997 study of twins identified through the Norwegian Twin Register, probandwise concordance rates for all dementias combined was 70% for monozygotic twin pairs and 40% for dizygotic twin pairs. Concordance rates among twin pairs identified through the Swedish Twin Registry were 50% for monozygotic twin pairs and 30% for dizygotic twin pairs (Gatz et al. 1997). Therefore the nondemented twin partner of a demented twin can be regarded as being at elevated risk for dementia. This second comparison evaluated whether twins at higher risk for dementia differed from a comparison group of twins who were not at higher risk.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
) or members of the OCTO-Twin Study (McClearn et al. 1997). Both studies represent defined subsamples of the population-based Swedish Twin Registry (Cederlof and Lorich 1978).
The Study of Dementia in Swedish Twins identified cases from the Swedish Adoption/Twin Study of Aging (SATSA), a longitudinal study of personality, health, and aging among same-gender twins (Pedersen et al. 1991). SATSA includes all pairs from the twin registry who indicated having been reared apart and a matched sample who were reared together. SATSA data collection included in-person cognitive assessments on a 3-year rolling schedule. For purposes of the dementia study, all twins identified for the SATSA sample born in 1935 or previously were included, if one or both members of the pair were alive in 1987, whether or not they had responded to SATSA data collection efforts (N = 1,978). Baseline screening took place in 1987 and 1988, with the final 20% completed by 1991, and incident cases were identified at each additional data collection.
The OCTO-Twin Study enrolled all twin pairs aged 80 and older if both members of the pair were alive during the first wave of data collection in 1991–1994 (N = 702). Three waves of data collection including in-person cognitive assessment took place at 2-year intervals. Cases of dementia were identified at each of the waves.
Screening Procedures
Case ascertainment in the Study of Dementia in Swedish Twins used a two-stage process. Participants were screened for dementia using either the MMSE (Folstein, Folstein, and McHugh 1975) or telephone screening protocol (Gatz et al. 1995). Those identified by screening as suspected cases of dementia were evaluated by an assessment team employing a nurse, a psychologist, and a physician. The protocol parallels Consortium to Establish a Registry for Alzheimer's Disease procedures for physical and neurological evaluations, laboratory tests, neuropsychological testing, and neuroimaging (Morris et al. 1989). Findings were presented at a consensus diagnosis conference, attended by the clinicians and chaired by a psychologist who had not met the twin; diagnoses were assigned following Diagnostic and Statistical Manual of Mental Disorder (3rd ed., rev.) criteria for dementia (American Psychiatric Association 1987), National Institute of Neurological and Communicative Disorders and Stroke (NINCDS)/Alzheimer's Disease and Related Disorders Association (ADRDA) criteria for probable and possible Alzheimer's disease (McKahn et al. 1984) and—once available—National Institute of Neurological Disorders and Stroke–Association Internationale pour la Recherche et l'Enseignement en Neurosciences (NINDS-AIREN) criteria for vascular dementia (Roman et al. 1993). Twin partners of probands were given an identical clinical workup. If the twin partners were deceased, their diagnostic assessment included informant interviews and review of medical records, including death certificates. Cases and twin partners were followed longitudinally every 18 months, with postmortem neuropathological examination for any who died. Autopsy confirmation of diagnosis was available for 25 cases.
Case ascertainment from the OCTO-Twin sample entailed a review of MMSE scores and cognitive performance information collected from all twins, whether or not they were demented. The battery paralleled that used in the Study of Dementia in Swedish Twins. For those suspected of dementia, the same informant protocol was followed, medical records were reviewed, and a consensus diagnosis assigned by a physician and psychologist. The same psychologist chaired the diagnostic conferences for both samples.
As part of the informed consent procedure participants (or their proxy) were asked what information they wished to receive. If requested, the physician provided a report of the clinical evaluation to the participant's health clinic with any comments about findings deserving follow-up. Participants were also informed that, in the case of significant finding requiring immediate attention, they and their primary physician would be notified.
In all, 33 twin pairs were identified in which both twins were initially diagnosed as nondemented but the pair became discordant for dementia during longitudinal follow-up. The nondemented twin partner was required to remain nondemented and alive for a minimum of 3 years counting from the year of dementia onset in the preclinical twin. Out of the 33 nondemented twin partners, two became demented 5 and 6 years, respectively, from dementia onset of the preclinical twin. In both cases, the nondemented twin partner was tested 7 years prior to dementia onset.
One control was matched to each of the 33 pairs by forming a pool of all potential matches based on age (± 3 years), gender, and zygosity. Then we randomly chose one control using a table of random numbers. The procedure also assured that no twin pairs had both members in the control sample, assuring independence of the data. To be included in the pool, both twin partners had to (a) have participated in at least two waves of testing, (b) stay nondemented and alive for at least 3 years after age of onset of the matched proband, and (c) match a discordant twin pair in gender, zygosity, and age at cognitive testing conducted by SATSA or OCTO-Twin. The first of three waves of cognitive testing was used for controls taken from OCTO-Twin. Controls from SATSA were included in the pool if their chronological age at one of the first two waves of cognitive testing corresponded to the age at testing of a discordant pair.
Table 1 presents demographic characteristics of the sample. The age at testing for both the preclinical and nondemented twin partners ranged from 60 to 88 years (M age = 78.1 years for preclinical twins and 78.2 years for nondemented twin partners). Matched controls were tested between ages 61 and 87 (M age = 78.1 years). The average age of onset in the preclinical twins was 80.1 years (SD = 6.6 years); the mean time difference between test administration and clinical dementia onset was 2.0 years (SD = .87 years). The preclinical twins were tested between 1 and 4 years before dementia onset. Fifty-eight percent of the preclinical twins were later diagnosed with Alzheimer's disease; 18%, with vascular dementia; 6%, with mixed dementia; 3%, with secondary dementia (e.g., Parkinson's disease or hydrocephalus); and 15%, with other dementias.
|
Seventy percent of the twin pairs were female and 30% were male; 36% were monozygotic and 64% were dizygotic. Matched controls were selected such that gender and zygosity corresponded precisely to that of the twin pairs who later became discordant for dementia. Because the sample included dizygotic twins, the results reflect only partial control of heritable factors.
Measures
Tests of verbal ability included a Swedish version of the Wechsler Adult Intelligence Scale (WAIS) Information subtest (Jonsson and Molander 1964) and Synonyms, a forced-choice vocabulary test from the Swedish Dureman—Sälde Battery (Dureman, Kebbon, and Osterberg 1971). Visuospatial–reasoning measures included Koh's Block Design (Dureman et al. 1971), similar to the WAIS Block Design subtest, and Figure Logic (Dureman et al. 1971), a matrix test in which the correct design must be chosen to complete a series. Perceptual speed was assessed using Figure Identification (Dureman et al. 1971), a timed pattern-matching test, and Symbol Digit Test, a task in which participants verbally indicate which digits correspond to a row of symbols, similar to the Digit Symbol test from the WAIS. Tests of memory and attention included Digit Span Forward and Digit Span Backward and Thurstone's Picture Memory. Digit Span Forward and Digit Span Backward were scored separately as the highest number of digits repeated correctly (Jonsson and Molander 1964). Thurstone's Picture Memory is a test of recognition memory of drawings of common items such as a truck and a table (Dureman et al. 1971). All tests were taken from standard batteries, with reliability coefficients ranging from .82 for Thurstone's Picture Memory to .96 for Figure Identification (Pedersen et al. 1992).
A brief mental status screening test was included: the MMSE (Folstein et al. 1975) with an extension of the three-item recall task. The participants who were unable to recall all three items were given a recognition test for any missed items. The recognition test entailed providing the correct item and two distractors. Correct responses were each scored 0.5 points. The addition of the recognition component was prompted by the same modification used by the Alzheimer's Disease Research Center at University of Southern California (S. H. Zarit, personal communication, November 9, 1992).
Analyses
Initially, we examined differences in baseline cognitive performance between those twins who later became demented and their nondemented twin partners using paired sample t tests. We also calculated Cohen 1977 d statistic using the formula for estimating effect sizes in studies with correlated designs (Dunlap, Cortina, Vaslow, and Burke 1996). To corroborate our findings with a more stringent statistical method, we conducted conditional logistic regression analyses for case-control studies (Guo et al. 2000), controlling for age and education differences. For these analyses, we used the SAS Program Version 8 (SAS Institute 1999). We estimated adjusted odds ratios to determine whether low cognitive scores were associated with being the twin in a preclinical phase of dementia, independent of any within-pair differences in age at testing or education. Finally, we calculated correlation coefficients to examine whether the magnitude of score differences within twin pairs was associated with mean age at testing, education (using the mean level of education for the pair), or zygosity of the pair. These correlation coefficients gave some indication of whether variations in age at testing, education, and zygosity might confound the results.
In the second set of analyses, we compared the cognitive test scores of the nondemented twin partners and the matched controls. Because of the matching procedure, paired t tests were used and effect sizes were tested with Cohen's d statistic for correlated designs. This comparison enabled us to examine whether having a twin partner in the preclinical phase of dementia was associated with lower cognitive scores in the nondemented twin partner. As in the first set of analyses, adjusted odds ratios and correlations with age at testing, education, and zygosity were calculated.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
Odds ratios adjusted for age and education yielded the same results as paired-sample t tests reported above (see Table 2 ) with the exception of Thurstone's Picture Memory, for which the difference between preclinical twins and their nondemented twin partners was reduced to only a statistical trend (p = .070).
Correlation coefficients indicated no association between either mean age at testing or zygosity with intrapair cognitive score differences for any of the measures. Correlation coefficients ranged from -.41 to .29 for age and intrapair differences in cognitive performance, and from -.21 to .29 for zygosity and cognitive performance. Mean education for the pair was associated with the difference score on the Thurstone's Picture Memory test, with higher education related to greater differences in cognitive test scores, r = .46, p < .05. No other correlation coefficients for education and cognitive score differences were significant, ranging from -.12 to .33.
Nondemented Twin Partners Versus Matched Controls
Matched controls performed better than nondemented twin partners on all cognitive tests, with some of the differences being significant (see Table 3 ).
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
; Johansson and Zarit 1997; Nielsen et al. 1999; Rubin et al. 1998; Storandt et al. 1984; Tierney et al. 1996). Therefore, our finding that Thurstone's Picture Memory test predicts dementia may be expected in spite of some opposing evidence. Some researchers argue that primary memory (Elias et al. 2000; Linn et al. 1995) or cued recall (Grober, Lipton, Hall, and Crystal 2000) may be the components of memory that remain relatively intact during the preclinical phase of dementia. Thurstone's Picture Memory test carries both a primary memory (i.e., only a brief retention of information is required) and a cued recall (i.e., cues in the form of possible answers are presented to the respondent) component. On the other hand, the score difference between preclinical and nondemented twins on Thurstone's Picture Memory test was not significant when we accounted for differences in age at testing and education level in the analyses. In addition, nondemented twin partners did not perform significantly better on the other tasks assessing primary memory—Digit Span Forward and Digit Span Backward.
Our finding that preclinical deficits may exist in perceptual speed and visuospatial–reasoning skills is supported by previous research. Preclinical deficits in perceptual speed are well documented (Amieva et al. 2000; Fabrigoule et al. 1998; Masur et al. 1994; Storandt et al. 1984). Several studies also reported that visuospatial–reasoning skills (Elias et al. 2000; Fabrigoule et al. 1998; Fox, Warrington, Agnew, and Rossor 1998; Jacobs et al. 1995) may deteriorate prior to dementia onset. Although our results are based on measures that are similar but not identical to the tests used in other studies, our findings still seem to substantiate the existence of multiple cognitive deficits in the preclinical phase of dementia.
Fabrigoule and colleagues 1998 and Amieva and colleagues 2000 suggested that the sensitivity of a cognitive test to preclinical deterioration may not depend on a specific cognitive area involved in a task. Instead, it may depend on attentional and executive demands of each task regardless of the specific cognitive area involved. Moscovitch and Umilta 1990 suggested that tasks requiring selective attention involve voluntary, "executive" processes and are prone to decline in dementia. On the other hand, tasks requiring a specific or well-learned response involve mandatory, modular, or autonomous processes and tend to be more resistant to neurodegeneration. Similarly, Jorm 1986, in his application of Shiffrin and Schneider 1977 theory of controlled and automatic processing to pathological cognitive changes in older adults, proposed that controlled but not automatic processes predict dementia onset.
The controlled–automatic processing theoretical framework may shed some light on the seeming divergence of our results. For example, although both Block Design and Figure Logic assess visuospatial–reasoning skills, they may require different cognitive processing. In Figure Logic, participants are given several options from which to choose; therefore, their response is cued and hence more automatic. However, more centralized, controlled cognitive processes may be involved in Block Design for which accurate performance requires initiation of active retrieval plans (Jorm 1986) and controlled cognitive manipulation. Our results indicate that cognitive performance may vary depending on what underlying processes are involved in the performance of a task.
Tests of perceptual speed may be especially useful in detecting early signs of dementia. Moscovitch 1992 identified perceptual speed as an important aspect of memory retrieval on more complex tasks. Fabrigoule and colleagues 1998 found that perceptual speed contributed to performance on tests that best predicted dementia because of its controlled processing component. Our finding that both measures of perceptual speed—Symbol Digit and Figure Identification—predict dementia complements the conclusions of Fabrigoule and colleagues 1998 and the hypothesis proposed by Jorm 1986 that deterioration of controlled cognitive processes is the first sign of dementia.
Several twin studies imply that perceptual speed may play an important role in detecting cognitive deficits in nondemented older adults because of its relatively high dependency on age and shared genetic factors. For example, Finkel and colleagues 1997 found perceptual speed a main cross-sectional indicator of cognitive deficits in older adults. Finkel and Pedersen 2000 concluded that the relation between cognitive abilities and perceptual speed is mediated primarily by genetic predisposition. Both McClearn and colleagues 1997 and Pedersen and colleagues 1992 reported relatively high heritability ratings for perceptual speed. Age and genetic predisposition seem to account for a relatively large portion of variance on tests of perceptual speed. Therefore, on average we would expect to find minimal intrapair differences on the measures of perceptual speed. Our results, however, indicate that twins in a preclinical phase of dementia score significantly lower than their nondemented twin partners on both tests of perceptual speed. Because previous research with nondemented twins seems to preclude such a finding, we may presume that the deficits in perceptual speed found here reflect the true predictive ability of perceptual speed in the preclinical phase of dementia.
Previous research provides some support for the applicability of the controlled–automatic theoretical framework to preclinical detection of dementia. For example, measures with the most stringent requirements on attentional resources such as some tests of memory (Fox et al. 1998; Jacobs et al. 1995; Linn et al. 1995; Nielsen et al. 1999; Tierney et al. 1996) and tests of visuospatial abilities (Fabrigoule et al. 1998; Masur et al. 1994), attention or mental control (La Rue and Jarvik 1986; Nielsen et al. 1999; Storandt et al. 1984; Tierney et al. 1996), and category verbal fluency (Dartigues et al. 1997; Masur et al. 1994; Nielsen et al. 1999) were most often cited as good predictors of dementia. On the other hand, tests that involve automatic information processing such as reading (Fox et al. 1998) or letter fluency (Monsch et al. 1992) may not be sensitive to early signs of dementia.
Similarly, we found only nonsignificant deficits among preclinical twins on tests of verbal skills—Information and Synonyms. Both these tests assess well-learned verbal ability with limited demands on controlled cognitive processing and, therefore, may not be sensitive to preclinical signs of dementia.
MMSE seemed to be a good determinant of incipient dementia in preclinical twins (see Table 2 ). Because MMSE is composed of multiple tests and assesses both controlled and automatic processes, this finding is somewhat more difficult to explain within the controlled–automatic processes framework. Nevertheless, the finding indicates that MMSE may be a valuable tool in assessing early dementia.
The comparison analyses of nondemented twin partners and matched controls yielded similar results to analyses of the preclinical and nondemented twins (see Fig. 1). Paired t tests showed significant differences on five measures for preclinical twins and their twin partners. Nondemented twin partners and matched controls differed on three of these five measures plus Information. Whereas nondemented twin partners performed similarly to preclinical twins on Information, they performed worse on this test than matched controls. Block Design and Figure Identification were the only two tests on which differences were found between nondemented twins and their preclinical twin partners but not between nondemented twin partners and matched controls.
There are two ways to explain this pattern of results. First, it is possible that in spite of selective demonstration of dementia within twin pairs, accelerated change in cognitive functioning may occur in both preclinical and nondemented twin partners because of shared genetic and/or environmental factors. Second, because we were not able to assess twin pairs longitudinally, it may be that the inherently lower intellectual level in some twin pairs predetermines relatively poor performance on some cognitive tests independently of age and education. Incipient dementia in the preclinical twins may further exacerbate the low cognitive performance. In any case, similarity of findings in the two sets of analyses seems to converge in pointing to particular sorts of cognitive deficits that presage dementia.
Several limitations of the present study should be addressed. First, our sample included both mono- and dizygotic twin pairs. Sample size precluded separate analyses by zygosity. Because intrapair similarity should be greater for monozygotic than for dizygotic pairs, one might predict a more pronounced preclinical effect for monozygotic than for dizygotic pairs. We did establish in our sample that zygosity of twin pairs was not significantly associated with intrapair differences on any of the cognitive measures, suggesting that there were not major differences by zygosity that were hidden by the combined sample. Nonetheless, assessing differences in dizygotic twin pairs may still attenuate our conclusions regarding control for possible genetic influences. Second, we used a relatively small sample of participants in the study. As a result, we can be less certain about our findings in terms of their general application. Both of these limitations stem partly from the fact that obtaining a sample of twins who later develop dementia is difficult. In a study similar to ours, La Rue and Jarvik 1987 assessed preclinical signs of dementia in twins but were not able to use a twin design in their data analyses because of not having sufficient numbers of complete twin pairs. Third, the present analyses combine various types of dementia. It is possible that separate analyses with different types of dementia would find distinct profiles among preclinical twins. However, we do not have sufficient pairs eligible for such comparisons. Fourth, we used cognitive measures that varied in score range. For example, Figure Logic has a restricted score range and relatively low internal consistency (Pedersen et al. 1992). Consequently, a test with a greater score range may yield significant results that are not due to its greater sensitivity to incipient dementia but to a larger spread of scores. On the other hand, these tests may be more appropriate for preclinical dementia research because they alleviate the possibility of floor or ceiling effects. Fifth, we used only a limited range of cognitive tests in this study. For example, it would be helpful to assess preclinical changes on tests of secondary memory or verbal learning. Such tests were previously found to predict dementia (e.g., Elias et al. 2000; Fox et al. 1998; Nielsen et al. 1999) but were not available prospectively in the SATSA assessment battery.
Sixth, the preclinical interval in our sample was relatively short. La Rue and Jarvik 1987 or Elias and colleagues 2000 used preclinical intervals that spanned to up to 22 years before clinical dementia onset. Assessment earlier before dementia onset may enhance the chances to discern true premorbid signs from early mild dementia, especially considering that the exact dementia onset is difficult to determine. Finally, we know that two of the nondemented twin partners of demented twins became demented themselves after 5 and 6 years, respectively. It seemed possible that the reason that nondemented twin partners scored lower than matched controls was that these two individuals depressed the mean scores for the nondemented twin-partner sample. However, excluding these two nondemented twin partners did not change the results.
In summary, our findings indicate that some cognitive tests may be more sensitive to developing dementia than others. The results seem to go along with the notion proposed by Jorm 1986 that a preclinical phase of dementia is characterized by deterioration of the central, controlled cognitive processes, whereas automatic processes remain relatively preserved. Because of small effect sizes and a small sample size in the present study, the results and implications of this study should be regarded with caution. Additional research using a larger pool of participants is needed to further evaluate the utility of the controlled–automatic processes framework in identifying preclinical dementia. Nonetheless, by assessing cognitive change in twin pairs, our study was the first to enable the distinction of premorbid signs of dementia and shared genetic or environmental factors.
|
Acknowledgments |
|---|
Received for publication November 20, 2000. Accepted for publication March 21, 2001.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  R. J. Kryscio, F. A. Schmitt, J. C. Salazar, M. S. Mendiondo, and W. R. Markesbery Risk factors for transitions from normal to mild cognitive impairment and dementia Neurology, March 28, 2006; 66(6): 828 - 832. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Gatz, A. Fiske, C. A. Reynolds, B. Johansson, L. Fratiglioni, and N. L. Pedersen Performance on Neurocognitive Tests by Co-twins to Dementia Cases Compared to Normal Control Twins J Geriatr Psychiatry Neurol, December 1, 2005; 18(4): 202 - 207. [Abstract] [PDF]
|
|
|
|
 |
|
 T. C. Rosenthal and N. Khotianov Managing Alzheimer Dementia Tomorrow J Am Board Fam Med, September 1, 2003; 16(5): 423 - 434. [Abstract] [Full Text]
|
|
|
|
 |
|
 L. Backman, S. Jones, B. J. Small, H. Aguero-Torres, and L. Fratiglioni Rate of Cognitive Decline in Preclinical Alzheimer's Disease: The Role of Comorbidity J. Gerontol. B. Psychol. Sci. Soc. Sci., July 1, 2003; 58(4): P228 - 236. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
