Introduction

It is estimated that more than 25 million people worldwide have Alzheimer’s disease (AD), with these numbers expected to triple by the year 2050 [1]. Familial early-onset AD is well recognized as an entity [2] but accounts for only about 2–3% of AD cases. FH is also a significant risk factor for, and predictor of late-onset AD [2]–[4] and studies suggests a 2 to 4-fold greater risk for AD in such individuals with a first-degree relative who has developed late-onset AD. Common gene polymorphisms (e.g. the ε4 variant of the APOE gene) account for about 50% of the heritability of late-onset AD [2] and despite recent genetic findings of new candidate genes, there is still a significant unexplained heritability. If one assumes that the average person with AD has 3 living first-degree relatives (1 sibling, 2 children), then there would be some 75 million worldwide with a positive FH of AD. Subjects with a positive FH often participate in research studies as controls, thus making it particularly important to understand how FH affects biomarker phenotype.

While previous multisite research has shown associations between AD biomarkers and specific genetic variations (such as APOE, APP, PSEN1, and PSEN2) [5]–[11], in most research studies and in routine practice, clinicians usually rely on a simple “yes/no” of self-reported FH status. Prior studies of the effects of FH status on biomarkers have shown effects on PET glucose metabolism, hippocampal volumes and amyloid markers [11]–[18]. If true, these findings have great significance since subjects with a positive FH are routinely enrolled in diagnostic and prognostic biomarker studies as “normal controls” and their inclusion might affect the accuracy of biomarker cut-points established for discriminating AD from controls.

The Alzheimer’s Disease Neuroimaging Initiative (ADNI) is considered one of the more successful biomarker research studies and has been widely analyzed and reported [19]–[22]. Its data forms the basis, in part, for newly proposed research diagnostic criteria for preclinical AD as well as MCI due to Alzheimer’s [23]–[24]. A positive family history was not an exclusion for controls or MCI in ADNI nor its subsequent studies ADNI-GO and ADNI-2. Most ADNI subjects were recruited from the community using advertisements and referrals in a manner similar to that used for most therapeutic and biomarker trials in many disorders.

The aims of our study were to systematically assess how subjects with a FH of AD differed from those without a FH in demographics, cognition, as well as neuronal and pathologic biomarker profiles, at the time of enrollment into a national biomarker study. Another aim was to examine whether the effect of FH was solely due to ApoE4 status. We also examined whether the effects of FH status on biomarkers differed across the spectrum of cognitively normal (CN) to MCI to mild AD to test the timing of such effects in relation to development of cognitive symptoms and clinical dementia.

Materials and Methods

Statistical Analysis

We first ran unadjusted t-tests and chi-square analyses comparing demographics and biomarker data (CSF Aβ42, t-tau, ptau181p, and hippocampal volume, HV) between FH+ and FH− subjects within each diagnostic group. Biomarker data were log-transformed to normalize their distributions but results were essentially similar before and after log transformation. Table 1 shows demographics and Figures 1–2 show unadjusted biomarker data. We then ran a multivariate linear model, with age, baseline MMSE, gender, and ApoE4 as predictors to examine effect of FH on tau/Aβ42 and HV within each diagnostic group above and beyond the effects of ApoE4 (Tables 2 and 3). We also examined the effect of FH status on intracranial volume (ICV) in each diagnostic group using multivariate linear models and also adjusted for ICV in models examining the effect of FH on HV. With chi-square, we tested the hypothesis that a positive FH status would be associated with a greater proportion of CN and MCI subjects with a CSF pathological signature of AD (CSF tau/Aβ42>0.39). Finally, ROC curves were generated to examine whether the specificity and sensitivity of CSF Tau/Aβ42 ratio for distinguishing AD from FH− CN differed from that for FH+ CN. We also ran alternative models adjusting for education and without adjusting for MMSE.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. CSF Biomarkers and Hippocampal Volume in CN, MCI, and AD by FH status.

a–d. Comparative boxplots of biomarker distribution and hippocampal volume in (mm3) as a function of diagnosis (on x-axis) and family history (present = +, absent = −): (a)TAU/Aβ42 (b)Aβ42 (c)TAU. The y-axis of each plot is on a logarithmic scale to transform the biomarker or hippocampal volume distribution to approximately Gaussian after adjusting for differences due to diagnosis and family history. P-values give the significance of a test of equality between the mean levels of biomarker or hippocampal volume (on the logarithmic scale) in FH+ and FH− groups, within each diagnosis category, using a two-tailed two sample t-test. In (a), the dashed horizontal line at 0.39 denotes the standard threshold value for diagnosis of AD.

https://doi.org/10.1371/journal.pone.0060747.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Intracranial Volumes by FH status. Figure shows that ICV (intracranial volume) did not significantly vary by FH status within diagnostic groups.

The p-values are from a model covarying diagnosis and FH status only but similar findings were noted in a multivariate model covarying for age, gender, and ApoE4 also. Only gender had an effect on ICV (data not shown).

https://doi.org/10.1371/journal.pone.0060747.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of ADNI Study Sample.

https://doi.org/10.1371/journal.pone.0060747.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Effect of FH on t-tau/Aβ42 ratios after covarying for ApoE4.

https://doi.org/10.1371/journal.pone.0060747.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Effect of FH on HV after covarying for ApoE4.

https://doi.org/10.1371/journal.pone.0060747.t003

Results

The overall, prevalence of FH+ in subjects who volunteered for CSF in ADNI-1 (50.4%) was slightly higher than the prevalence of FH+ among all ADNI-1 subjects with FH data (42%). The prevalence of a FH+ status did not differ significantly between diagnostic groups (p = 0.36). Within each diagnostic group, there were no differences in key demographic or cognitive baseline characteristics by FH status (Table 1).

Discussion

Three main findings emerged from our study. We found significantly higher t-tau/Aβ42 ratios and a higher prevalence of the rate of a “pathologic t-tau/Aβ42 endophenotype” in FH+ (versus FH−) CN and MCI groups. There was an additional effect of FH on these markers above and beyond that of ApoE4 in MCI subjects and the model estimated suggests that this additional effect is about half of the size of the ApoE4 effect. We found no FH effects on CSF pathologic markers in AD. We also found no FH effects on neuronal loss marker (hippocampal volume) both before and after adjusting for intracranial volume. We also found no FH effects on ICV. These data extend findings from prior studies of FH effects [5]–[17] to the national ADNI sample and to MCI subjects. Our study also found that almost half of all normal controls with FH+ would have met research criteria for preclinical AD (based on CSF) [23] at entry into ADNI but only about 20% of FH− subjects would have met such criteria. This result is also consistent with the view that a family history of AD does not reduce cognitive reserve, as there were no significant differences in cognitive test scores between FH+ and FH− groups. Rather, one can speculate that the risk of family history is probably mediated by earlier development of amyloid pathology or more rapid development of amyloid pathology with the same time of onset.

Of prior studies examining FH effects, three are particularly relevant to our analyses. In a study of 269 cognitively normal controls, Xiong et al [12] reported that FH status was linked to a decrease in CSF Aβ42, a finding that we extend by reporting a similar and even more robust change in MCI. Honea et al [11] examined the relationship between biomarkers and parental history of all dementia types in the ADNI sample, thus potentially including also FH of vascular dementia, DLB or FTD or other etiologies. In their study, the rate of FH+ subjects was different from ours and unlike our study, the effects of FH on t-tau/Aβ42 ratio and t-tau effects in MCI did not reach significance. They did report a significant FH effect on Aβ42 in MCI consistent with our finding, but in their study the FH effects in MCI were not significant after adjusting for ApoE4, whereas ours remained significant. They also found pathologic signatures of AD in a smaller percent of CN than we did. Thus, their looser definition appears to have resulted in an underestimation of the effect of FH of AD. Andrawis et al [18] found MCI subjects with positive maternal history of dementia had smaller baseline hippocampal volumes and greater 12-month atrophy rates. The effect of positive maternal history of dementia on hippocampal atrophy in MCI and AD was significant after controlling for age, ApoE4 genotype, and paternal history of dementia. Taken together, these studies along with prior studies showing potential FH effects on brain networks and glucose metabolism highlight the need to further examine FH effects on multiple biomarkers simultaneously.

The mechanisms underlying the effects of FH status are not fully known but will likely vary depending on biomarker – ie genetic mechanisms underlying hippocampal volume changes are not likely to be identical to those underlying amyloid or tau processing. Prior autopsy, CSF and PET studies have linked ApoE4 to an amyloid phenotype and hippocampal changes [6]–[10] [26], and studies have documented ApoE4 effects on greater neocortical amyloid-beta deposition and/or reduced clearance [6] [27]. However, E4 does not account for all of the variance and there is interest in determining the degree of unexplained heritability not accounted for by ApoE4 as well as the genes underlying such unexplained heritability. In our ADNI sample, the effect of adjusting for ApoE4 on pathologic markers was different in different diagnostic groups - the FH effect was considerably weakened in CN, remained significant in MCI subjects, and was not significant in AD. After adjusting for ApoE4, the remaining FH effect on CSF t-tau/Aβ42 was approximately half the size of the main ApoE4 effect. Thus, our data along with others [8] [12] confirms that there are as yet unidentified genetic factors embedded in FH status that influence pathology before the onset of dementia [12].

Our sensitivity/specificity analyses also suggest that the presence of FH+ controls in an AD control group may significantly reduce the specificity of CSF pathologic biomarkers for separating AD from controls. It may be worth examining whether including FH+ controls may have reduced the accuracy of calculations in other tests, such as amyloid PET or plasma Aβ42. Likewise our data also suggests that companies planning registration studies of diagnostic biomarkers to detect AD pathology in at-risk subjects may wish to exclude FH+ controls to enhance their power for achieving the desired 80% or greater specificity.

There are also some potential limitations of this study – by design ADNI’s sample size of controls and AD was relatively smaller than the MCI group, which may have limited power to detect small effects in controls. CSF data were only collected in a subset who agreed to volunteer, a selection bias that applies to most CSF biomarker studies. FH status was determined through interviews with subjects and informants, but may have been subject to a reporter bias and lack of informative pedigree (early death of relatives due to other causes, though this problem is less likely in the US due to longer life expectancies and greater awareness of dementia). Many respondents may not be well versed enough to know the difference between a dementia and AD. Because FH in most biomarker research and practice is usually collected only by simple history, our findings are relevant. We also did not distinguish between maternal and paternal inheritance and hence our data cannot be compared with findings that maternal family history may have greater risk for metabolic changes or hippocampal atrophy [15]–[16]. Furthermore, given that there is a mitochondrial hypothesis providing an underlying biological mechanism for promoting disease on the maternal side we believe future studies should examine maternal versus paternal family history. However a recent longitudinal study of 108 middle-aged normal controls (of younger age than ADNI cohort) found that FH status predicted greater atrophy only within a posterior sub-region of the hippocampus but not in other gray matter regions, and that there was no effect of maternal versus paternal history [26]. Differences in sampling, FH ascertainment, and biomarker methods could account for some of the discrepant findings. While the means differ significantly, the overlap in CSF data boxplots between FH+ and FH− MCI groups suggests that these findings may not be as robust a biomarker as one where the boxplots do not overlap at all - unfortunately no such biomarker exists.

What do these phenotypic differences related to a positive FH in MCI mean for the subject? Other studies have linked CSF pathologic phenotypes with faster rates of future decline in CN and MCI subjects [22]. By extrapolation, this would imply that the subset of FH+ MCI and CN subjects with abnormal biomarker phenotypes would decline faster than FH− subjects. Longitudinal data from ADNI and standardization of hippocampal sub-region analyses as well as CSF soluble amyloid oligomer assays [28] may permit more definitive testing of the prognostic significance of FH differences on risk for decline.

In summary, our study, derived from a large national biomarker cohort, documents that a positive family history of AD is associated with an abnormal beta-amyloid and tau endophenotype prior to the onset of clinical AD in mildly symptomatic subjects, and that there are genetic influences embedded within FH beyond that due to ApoE4 that are most obvious in the MCI cohort. Since CSF biomarkers correlate highly with cerebral neuritic beta-amyloid and neurofibrillary tangle changes [28], we also speculate that FH status is associated with earlier onset of preclinical pathologic AD. These findings have implications for the design of future research studies, heritability of AD and personalizing testing and care of at risk subjects.

Acknowledgments

Data used in preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (adni.loni.ucla.edu). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at: http://adni.loni.ucla.edu/wpcontent/uploads/how_to_apply/ADNI_Acknowledgement_List.pdf. The ADNI was launched in 2003 by the National Institute on Aging (NIA), the National Institute of Biomedical Imaging and Bioengineering (NIBIB), the Food and Drug Administration (FDA), private pharmaceutical companies and non-profit organizations, as a $60 million, 5-year public-private partnership.