BILINGUALISM DELAYS EXPRESSION OF ALZHEIMER’S CLINICAL SYNDROME

Abstract

Objective

To evaluate the effects of bilingualism on the emergence of Alzheimer’s clinical syndrome.

Background

Studies have proposed an increase of cognitive and neural reserve from management and control of two languages, with a consequent delayed expression of dementia.

Methods

In a clinic with a large immigrant population, we identified 253 patients with Probable Alzheimer’s disease (AD) with intermediate or high evidence of AD pathophysiological process. These patients were reviewed for demographic variables, native language (L1) other than English, ages of onset and presentation, Mini-Mental State Examination (MMSE), digit spans, word fluencies, naming, and memory.

Results

Among these patients, 74 (29.2%) were bilinguals with various L1s (Farsi, Spanish, Chinese, Tagalog, Arabic, others). When compared to the 179 monolingual AD patients, those who were bilingual had significant delays in ages of onset and presentation of approximately four years (p=0.003). These delays persisted despite bilinguals having worse MMSE scores on presentation. There were no significant group differences on other variables except that for worse naming in English among bilinguals vs. monolinguals. Caregiver/informants reported that 66 (89.2%) of the 74 bilingual AD patients had gradually regressed to predominant use of their L1.

Conclusions

In line with published reports worldwide, we found that bilingualism delays the expression of Alzheimer’s clinical syndrome. We also found frequent reversion to the first learned language. These findings suggest that, among bilinguals, the availability of an L1 “back-up” either facilitates compensation or masks emergence of the early symptoms of dementia.

INTRODUCTION

Alzheimer’s disease (AD), the most common dementia, is an age-related disorder that has reached epidemic proportions with the aging of populations. Despite the increasing prevalence of AD and the burden of this disorder to patients, families, and society, there has been a lack of success in developing disease-modifying therapies. Consequently, there is increasing emphasis on prevention or identifying modifiable risk factors that can prevent, or at least retard, the development of this disorder [1].

The frequent use of two or more languages could be one of the most important modifiable risk factors for delaying the expression of Alzheimer’s clinical syndrome [2]. Much of the world is bilingual, and economically advanced societies are experiencing an influx of immigrant populations with different native languages (L1s). Elderly bilinguals, or those who can use two languages in everyday life and are at risk for AD, can clarify whether there is a “bilingual advantage” in the onset of dementia symptoms. Some, but not all, studies suggest that bilingualism promotes cognitive and neural resilience and delays the onset of the symptoms of dementia [3–7]. If learning or knowing a second language could postpone dementia symptoms even for a few years, this could mean major benefits, not only for quality of life in later years but also for decreasing related healthcare costs [8].

The literature on the relationship between bilingualism and AD remains controversial as bilingualism is complex and the diagnosis of AD can be imprecise. Variations in bilingualism include age of acquisition, proficiency in the second language (L2), and differences in contextual use. Moreover, AD is not equivalent to the typical amnestic-predominant clinical syndrome of dementia. Accordingly, from a university referral program, we evaluated all our sequential bilinguals (L2 acquisition after age 5 and historically able to use both L1 and L2 in everyday life with L2 primarily outside of home) and diagnosed with biomarker-supported Alzheimer’s clinical syndrome uncomplicated by other potential neuropathology. The bilingual patients from our program varied in their original language, but all had English as their second language. We hypothesized that, compared to monolinguals with AD, bilinguals with AD have a significant delay in onset of symptoms of dementia and suggest potential mechanisms for this effect.

METHODS

The participants in this study presented for evaluation, over a 15-year period, to a university clinical program located in a U.S. urban area with a large immigrant population. The program is also known as a center for early-onset dementias and a high proportion of our referrals are for early-onset AD. All patients had the insidious onset and progression of cognitive changes and underwent a comprehensive evaluation, laboratory assessment, and magnetic resonance imaging (MRI); many also had fluorodeoxy-glucose positron emission tomography (FDG-PET) of the brain. A total of 253 patients met National Institute on Aging-Alzheimer’s Association (NIA-AA) Criteria for “Probable AD with evidence of AD pathophysiological process”, either Intermediate (primarily FDG-PET) or High (primarily low cerebrospinal fluid [CSF] amyloid-β levels) [9,10]. NIA-AA criteria were re-assessed retrospectively for this study. Patients tended to have either FDG-PET or CSF studies (all CSF studies included total tau and phospho-tau). Only three had amyloid PET imaging and none had tau PET (hence, we did not use AT(N) research criteria) [11]. Patients with evidence of vascular changes on MRI or other potential neuropathology were excluded from consideration. All patients had “moderate dementia” based on “Extensive functional impact on daily life with impairment in basic activities. No longer independent and requires frequent assistance with daily life activities” [11]. The participants for this study were part of an institutional review board approval for this research.

From their intake presentation, we abstracted demographic and examination information and documentation of the presence of bilingualism. Among all patients, we abstracted ages of onset and presentation, sex, handedness, years of education, and Mini-Mental State Examination (MMSE) scores. A Neurobehavioral Status Examination (NBSE) available in most patients and included digit span forward and backward, numbers of “F” words/minute and category (animals)/minute, the 15-item mini-Boston Naming Test (mBNT), and a 10-item (5 registration trials) verbal learning test with 15-minute delayed recall and 20-item recognition [12]. If bilingualism was present, we further abstracted available information on language use in the first five years of life, the later acquisition of English, immigrant status, the historical ability to use both languages on a daily basis, whether English was the predominant language at work or outside the home, and any change in current language use.

Statistical Analysis

The demographic and neurobehavioral characteristics of the two dementia groups were compared using t-tests for continuous variables and Fisher’s exact test or χ² for sex and handedness. For both age of onset and age or presentation, we performed logistic regressions with groups as dependent variable and significant or near significant non-language variables as predictors. These were followed with analysis of covariance (ANCOVA) controlling for dementia severity as measured by MMSE. The statistical analysis was performed using SPSS (version 23; SPSS, Chicago, IL, USA).

RESULTS

Of the 253 patients with Probable AD, 74 (29.2%) were sequential bilinguals who were previously fluent in English, and, of these bilingual patients, 66 (88.2%) were reported as having regressed to predominant or exclusive use of L1 both inside and outside the home. All the bilingual participants were immigrants, and they varied in their original language. The L1s in our immigrant population were Farsi (22), Spanish (21), Mandarin Chinese (5), Tagalog (4), Arabic (4), Korean (3), Hebrew (3), Hindi (3), French (2), Russian (2), and one each Bengali, Croatian, Greek, Hausa, and Hungarian.

The bilingual AD patients were older in age of onset (t=−2.99; df 251, p=0.003) and presentation (t=−3.03; df 251, p=0.003). They also had lower MMSE scores consistent with more advanced disease (t=3.38; df 251; p=0.001) (See Table 1). Although there was quite a difference in numbers of males vs. females between the two groups, this did not reach statistical significance (p=0.052 Fisher’s Exact Test or χ²=3.63, df 1, p=0.057). There were not significant group differences on the remaining neurobehavioral variables except for worse confrontational naming (mBNT) among the bilingual Probable AD patients (t=1.98; df 215, p=0.049).

TABLE 1:

Characteristics of Probable AD Patients

	Bilingual n=74	Monolingual n=179
Ages OnsetA	62.4 (10.5)	58.0 (10.7)
Age of PresentationA	66.0 (10.5)	62.1 (8.8)
Male/Female	46/28	85/92
Education (years)	14.8 (4.8)	14.5 (4.2)
Handedness R/L	70/4	171/17
MMSE scoresA,B	18.1 (8.1)	21.6 (6.3)
Neurobehavior Status ExaminationC
Digit Span: Forward	5.22 (1.11)	5.37 (1.36)
Digit Span: Backward	2.87 (1.63)	2.84 (1.18)
“F” Word Fluency	7.21 (6.12)	7.67 (5.57)
Category Fluency	9.22 (4.55)	10.26 (6.07)
mBNTA,D	8.55 (4.08)	9.96 (4.83)
Delayed verbal recall	1.74 (2.08)	2.06 (2.42)
Memory recognition	14.95 (2.98)	15.44 (2.85)

^AStatistically significant group differences

^BMMSE=Mini-Mental State Examination

^CAvailable on 58 bilinguals and 159 monolinguals

^DmBNT=mini-Boston Naming Test.

Logistic regression was performed for group (bilingual vs. monolingual Probable AD) with non-language variables that were significant or near significant (ages of onset or presentation, MMSE, and sex). The results showed that increased ages of onset and presentation and worse severity of dementia per MMSE scores predicted membership in the bilingual Probable AD group. Subsequent ANCOVAs controlling for MMSE scores, did not have overall significant models but bilingual group membership remained significant for ages of onset and presentation (See Table 3). Notably, there were no group*gender interaction effects.

TABLE 3:

Analysis of Covariance

Dependent Variable: Age at Onset with MMSE score as covariate
Source	Sum of Squares	df	Mean Square	F	Sig.
Corrected Model	951.856B	4	237.964	2.109	.081
Intercept	76820.636	1	76820.636	680.864	.000
MMSE score	11.365	1	11.365	.101	.751
GroupA	887.208	1	887.208	7.863	.006
Gender	52.808	1	52.808	.468	.495
Group*Gender	.062	1	.062	.001	.981
Error	23581.098	209	112.828
Total	760232.750	214
Corrected Total	24532.954	213
Dependent Variable: Age at Presentation with MMSE as covariate
Source	Sum of Squares	df	Mean Square	F	Sig.
Corrected Model	763.430B	4	190.857	2.324	.058
Intercept	90376.846	1	90376.846	1100.475	.000
MMSE score	3.270	1	3.270	.040	.842
GroupA	686.759	1	686.759	8.362	.004
Gender	1.905	1	1.905	.023	.879
Group*Gender	5.595	1	5.595	.068	.794
Error	17410.561	212	82.125
Total	866768.000	217
Corrected Total	18173.991	216

^AGroup: Bilingual vs. Monolingual Probable AD

^BR²=0.039 (Adjusted R²=0.02)

^AGroup: Bilingual vs. Monolingual Probable AD

^BR²=0.042 (Adjusted R²=0.024)

DISCUSSION

We found evidence to support a delay in the onset of dementia symptoms among well-characterized patients with Alzheimer’s clinical syndrome who were bilingual in different languages. Like other reports in the literature, the bilingual Probable AD patients had the onset and presentation of their dementia symptoms about 4 years later than comparable monolingual patients with AD despite worse dementia severity (i.e., MMSE scores) on presentation. The more advanced MMSE scores suggested that bilingualism compensates for cognitive decline as well as for the neuropathology of AD in everyday life so that disease is more advanced on reported onset and presentation.

In 2007, Bialystok and colleagues reported that bilingual patients with probable AD were diagnosed about four years later than monolinguals with probable AD [13]. Since then, investigators and most retrospective studies from around the world have suggested that bilingualism delays clinical symptoms of dementia by 4–5 years [3–7,13–40] (See Table 4). In addition, pervasive multilingualism could be a factor in the low prevalence of cognitive complaints and dementia among older adults in Luxembourg [41], and, in the Lothian Birth Cohort (Scotland), those who learned a second language manifested better cognition in late life [42]. Others report that the more languages spoken, the better the cognitive performance in the aged [43].

TABLE 4:

Studies Supportive of a Delay in Dementia Onset

Bialystok E, et al. 2007 [13]	Canada; 184 Patients with dementia, 51% bilingual which showed dementia symptoms 4–5 years later than monolinguals.
Craik FI, et al.2010 [14]	Canada: follow-up of the above study; 211 new patients with probable AD, bilinguals (n=102) were diagnosed 4.3 years later + symptom onset 5.1 years later than monolinguals (n=109).
Zheng Y, et al. 2018 [15]	China; Cantonese-Mandarin bilinguals was older at AD onset, and older at first clinic visit than monolinguals.
Alladi S, et al. 2016, 2017 [17,18]	India; bilinguals(n=390) developed clinical dementia 4.5 years later, delay in post-stroke cognitive impairment and bvFTD (but not other forms of FTLD or PSP) than monolinguals (n=257).
Woumans E, et al. 2015 [19]	Belgium; 134 patients with probable AD, bilinguals (n=69) had first symptoms (4.6 years) and diagnosis (4.8 years) later after controlling for demographics.

There are several other supportive studies, as well as those that question the relationship of bilingualism and dementia. Among patients with mild cognitive impairment (MCI), a frequent prelude of dementia, one report found that bilinguals were several years older at age of symptom onset [24]. Others found that multilinguals had a lower risk for amnestic MCI and higher temporal tissue density [25–29]. Another study found bilingualism associated with later age-of-diagnosis only among those with a low (<11 years) educational level [30]. Much disagreement appears to stem from reports of a lack of effect of bilingualism on incidence of dementia or AD [4,5,35–37,44,45]. However, assessing for an effect on new incident cases of dementia is distinct from retrospectively assessing for an effect on age of onset or delay in clinical dementia.

In this study, the Probable AD patients showed a “bilingual advantage” in the expression of Alzheimer’s clinical syndrome. Bilingualism could increases cognitive reserve through the constant management of two simultaneously active and competing languages [13,41,46–58]. Better reserve, with better compensation or masking of worsening cognition, could explain the worse dementia severity per MMSE scores on presentation for bilinguals. Furthermore, increased cognitive and neural reserve from bilingualism may be beneficial when there is cognitive decline from neuropathology such as that from AD [13,17,48,50,51]. In the presence of greater cognitive reserve and optimal compensatory use of brain resources, there may be a need for more advanced AD pathology in order to express dementia symptoms [49]. A few studies show that bilinguals with dementia may have more advanced temporal neuropathology than monolinguals with dementia [7,14,29,53,59–62]. Computerized tomography scans of bilingual and monolingual AD patients who were matched for level of cognitive performance showed that the bilingual patients had more disease (greater mesotemporal atrophy-radial width of the temporal horn and the temporal horn ratio) than the monolingual patients [59]. Others report similar or lower tissue density in medial temporal areas in multilingual patients with AD versus monolingual patients with AD despite comparable impairment [29]. Consistent with these structural changes, cerebral hypometabolism was more extensive among AD bilinguals vs. AD monolinguals, indicating more advanced disease [61].

As with other research on bilingualism in dementia [63], this study has potential limitations. One of the most important factors is the social diversity of language use including adaptations to interactional contexts [64]. Bilingualism’s effects may depend on whether they are in a home or familial vs. formal or work context, along with social situation and psychological distancing effects [42,65–67]. The bilinguals in this report used English outside the home and at work, but often their native language at home. Another important issue is the objective determination of L2 proficiency without relying on self-reports [68,69]. In this study there was collateral verification of “routine daily use” proficiency among the bilinguals. Bilingualism in dementia may also be affected by factors such as immigration status, as well as acculturation, education, specific L1 language groups, and other variables [16,31,42,63,65–73]. All of the bilingual participants in this study were immigrants. There were more males among the bilinguals, possibly because of immigrant status (but gender differences did not meet significance). Finally, many studies fail to fully characterize the dementia as consistent with Alzheimer’s clinical syndrome, as is done in this study.

In conclusion, bilingualism appears to delay presentation with Alzheimer’s clinical syndrome, possibly from compensation or masking of cognitive decline. Learning a second language, even in older age, may be an important mechanism for promoting compensatory reserve and for delaying a functional decline from early dementia [74,75]. Even subsequent regression to L1 with advancing dementia may suggest ongoing compensation from better retained semantic access in L1 [76]. The findings of this study can lead to more research and investigation of the use of two languages as an important delaying factor for dementia symptom onset.

Table 2:

Logistic Regressions

Dependent Variable: Group (Bilingual Probable AD vs. Monolingual Probable AD)
	B	S.E.	Wald	df	Sig.	Exp(B)
Age of Onset	.047	.017	7.543	1	.006	1.048
MMSE score	−.072	.022	10.516	1	.001	.930
Sex	.572	.331	2.990	1	.084	1.771
Constant	−2.676	1.120	5.704	1	.017	.069
	B	S.E.	Wald	df	Sig.	Exp(B)
Age at Presentation	.048	.018	7.551	1	.006	1.049
MMSE score	−.071	.022	10.207	1	.001	.931
Sex	.571	.329	3.004	1	.083	1.769
Constant	−2.954	1.214	5.922	1	.015	.052

Nagelkerke R² 0.14