Abstract
Importance
The association between the microtubule-associated protein tau (MAPT) H1 haplotype and the risk of progressive supranuclear palsy (PSP) has been well documented. However, the specific H1 subhaplotypes that drive the association have not been evaluated in large studies, nor have they been studied in relation to neuropathologic severity of disease.
Objective
To comprehensively evaluate the associations of MAPT haplotypes with the risk of PSP and the severity of tau pathology using a large series of neuropathologically confirmed PSP cases.
Design, Setting, and Participants
A case-control study was used to investigate the associations between MAPT haplotypes and the risk of PSP, and a case series was conducted for examination of associations of MAPT haplotypes with the severity of tau pathology. All 802 neuropathologically confirmed PSP cases were obtained from a neurodegenerative disorders brain bank between January 1, 1998, and December 31, 2013, and 1312 clinical controls were obtained from the neurology department of the Mayo Clinic. Statistical analysis was performed from February 17 to December 12, 2018.
Main Outcomes and Measures
Presence of PSP in case-control analysis and semiquantitative tau pathology scores for neurofibrillary tangles, neuropil threads, tufted astrocytes, and oligodendroglial coiled bodies in PSP cases.
Results
For 802 patients with PSP (376 women and 426 men), the median age at death was 75 years (range, 52-98 years). For 1312 controls (701 women and 611 men), the median age at blood collection was 69 years (range, 45-92 years). After adjustment for multiple testing, known associations with risk of PSP were observed for the H2 and H1c haplotypes. Novel associations with PSP were observed for 3 H1 subhaplotypes, including H1d (odds ratio, 1.86; 95% CI, 1.43-2.42; P = 2 × 10−6), H1g (odds ratio, 3.64; 95% CI, 2.04-6.50; P = 2 × 10−6), and H1o (odds ratio, 2.60; 95% CI, 1.63-4.16; P = 2 × 10−5). Although not significant after multiple testing adjustment, 3 of these PSP risk haplotypes (H2, H1c, and H1d) were also nominally associated with measures of severity of tau pathology in PSP cases. Nominally significant associations with severity of tau pathology were also noted for the H1e and H1q haplotypes.
Conclusions and Relevance
This study has identified novel associations with risk of PSP for 3 MAPT H1 subhaplotypes. In addition, potential weaker associations between several haplotypes (including several PSP risk haplotypes) and severity of tau pathology were observed. These findings expand the current understanding of the role of MAPT haplotypic variation in susceptibility to and neuropathologic severity of PSP.
This case-control study investigates the associations of MAPT haplotypes with the risk of progressive supranuclear palsy and the severity of tau pathology using a large series of neuropathologically confirmed progressive supranuclear palsy cases.
Key Points
Question
Are MAPT haplotypes other than H2 and H1c associated with the risk of progressive supranuclear palsy, and are any MAPT haplotypes associated with the severity of tau pathology?
Findings
In this case-control study of 802 patients with progressive supranuclear palsy and 1312 control patients, in addition to observing the known associations with progressive supranuclear palsy susceptibility for the H2 and H1c haplotypes, novel associations with the risk of progressive supranuclear palsy were identified for 3 additional H1 subhaplotypes (H1d, H1g, and H1o). Potential associations between several haplotypes (including several of those associated with the risk of progressive supranuclear palsy) and severity of tau pathology were observed.
Meaning
Multiple association signals at the chromosome 17 MAPT locus are implicated in the possible cause of tauopathy.
Introduction
Progressive supranuclear palsy (PSP) is a rare, atypical Parkinsonian disorder that is clinically characterized by vertical supranuclear gaze palsy, falls, akinesia, and cognitive dysfunction.1 Neuropathologically, PSP is classified as a tauopathy due to the presence of four-repeat tau aggregates in neurons and glia in the form of neurofibrillary tangles (NFT), neuropil threads (NT), tufted astrocytes (TA), and oligodendroglial coiled bodies (CB).2 As the clinical presentation of PSP can be similar to that of other neurodegenerative disorders (such as corticobasal syndrome or multiple system atrophy), clinical diagnosis can be challenging, and neuropathologic examination is required to make a definitive diagnosis.3
Although PSP is generally considered to be a sporadic disease, rare familial forms have been reported, and mutations in the microtubule-associated protein tau (MAPT) gene (GenBank 4137) have been observed with an autosomal dominant pattern of disease inheritance.4 During the past decade, the role of genetics in PSP has become clearer, with the MAPT gene being the major common genetic risk factor.5 The MAPT gene is located on chromosome 17 in a region of extended linkage disequilibrium that is characterized by 2 main haplotypes: H1 (common) and H2 (rarer to almost absent depending on racial/ethnic background).6 The common MAPT H1 haplotype has been associated with a substantially increased risk of PSP by numerous groups.7,8,9,10,11,12 This association is further supported by findings of the only large genome-wide association study (GWAS) of PSP that has been performed to date, in which the MAPT H1 haplotype was by far the strongest susceptibility locus.13
Despite the well-documented association between the H1 haplotype and risk of PSP, there is still much to be understood regarding the role of the MAPT variation in PSP disease risk and progression. Studies of MAPT and PSP risk have focused primarily on only the H1 and H1c (which is also associated with an increased risk of PSP) haplotypes, as well as the rs242557 variant that partially defines the H1c haplotype; however, the H1 haplotype can be further divided into more than 20 common (frequency, ≥1%) subhaplotypes,11 several of which have been previously associated with other neurodegenerative disorders such as dementia with Lewy bodies,14 Parkinson disease with dementia,15 and multiple system atrophy.16 The only study to date that has examined associations of all common MAPT haplotypes with risk of PSP did not include a large number of patients with PSP.11 Furthermore, although several studies have examined the potential role of the H1 haplotype in modifying the disease presentation and severity of PSP, sample sizes have not been large, and none have evaluated specific H1 subhaplotypes.17,18,19,20,21 In the present study, we used a large series of neuropathologically confirmed patients with PSP and evaluated associations of MAPT haplotypes with disease risk and neuropathologically assessed measures of severity of tau pathology.
Methods
Study Patients
A total of 802 PSP cases were obtained from the Mayo Clinic brain bank for neurodegenerative disorders between January 1, 1998, and December 31, 2013, and were neuropathologically assessed for severity of tau pathology; 1312 clinical controls were also included in this study. Progressive supranuclear palsy cases were examined by a single neuropathologist (D.W.D.) using standardized histopathologic methods and phospho-tau immunochemistry. All cases met neuropathologic criteria for PSP,22 and more than 85% (707 of 801 with clinical diagnosis information available) of patients had a clinical diagnosis of PSP. Controls were free from neurologic disease and were seen at the Mayo Clinic in Jacksonville, Florida (n = 881) or Rochester, Minnesota (n = 431). The controls were collected in part through the Mayo Clinic Florida Morris K. Udall Parkinson’s Disease Research Center of Excellence, Alzheimer’s Disease Research Center, and Mayo Clinic Study of Aging. All patients were non-Hispanic, white, and unrelated. Characteristics of patients with PSP and controls are shown in Table 1. This study was approved by the Mayo Clinic Institutional Review Board. All patients or legal next of kin provided written informed consent.
Table 1. Characteristics of Patients With PSP and Controls.
| Characteristic | Patients With PSP (N = 802) | Controls (N = 1312) |
|---|---|---|
| Age, median (range), ya | 75 (52-98) | 69 (45-92) |
| Sex, No. (%) | ||
| Female | 376 (46.9) | 701 (53.4) |
| Male | 426 (53.1) | 611 (46.6) |
| Braak stage, No. (%) | ||
| 0 | 117 (14.6) | NA |
| I | 133 (16.6) | NA |
| II | 235 (29.3) | NA |
| III | 247 (30.8) | NA |
| IV | 52 (6.5) | NA |
| V | 12 (1.5) | NA |
| VI | 6 (0.8) | NA |
| Thal phase, No. (%) | ||
| 0 | 356 (44.4) | NA |
| 1 | 133 (16.6) | NA |
| 2 | 54 (6.7) | NA |
| 3 | 195 (24.3) | NA |
| 4 | 43 (5.4) | NA |
| 5 | 21 (2.6) | NA |
| CB overall tau pathology score, median (range)b | 1.5 (0.3-2.4) | NA |
| NFT overall tau pathology score, median (range)b | 2.2 (0.8-2.9) | NA |
| TA overall tau pathology score, median (range)b | 1.0 (0.1-2.1) | NA |
| NT overall tau pathology score, median (range)b | 2.2 (0.3-2.9) | NA |
Abbreviations: CB, coiled bodies; NA, not applicable; NFT, neurofibrillary tangles; NT, neuropil threads; PSP, progressive supranuclear palsy; TA, tufted astrocytes.
Age at death is given for patients with PSP; age at blood collection is given for controls.
One overall score was created for each separate measure of tau pathology (NFT, CB, TA, and NT) by calculating the mean of the semiquantitative measures (0, 1, 2, or 3) for each patient with PSP across all neuroanatomical regions. For patients who did not have information available regarding the given measure of tau pathology for a given neuroanatomical region, these scores were imputed by using the mean of the values of the patients with PSP who did have this information available for the purposes of calculating the overall scores. Any patient with missing data for greater than 50% of the neuroanatomical regions for a given measure of tau pathology was not included in any analysis involving the overall score for that measure. Overall tau pathology scores were unavailable for CB (1 patient), TA (31 patients), and NT (1 patient).
Genetic Analysis
Using standard protocols, total genomic DNA was extracted from peripheral blood lymphocytes in controls and from brain tissue of patients with PSP.23 Six MAPT variants (rs1467967, rs242557, rs3785883, rs2471738, rs8070723 [the H2 haplotype–tagging variant], and rs7521) were genotyped to assess the most common MAPT haplotypes, as previously described.11,24 The MAPT variants were genotyped using TaqMan single-nucleotide polymorphism genotyping assays on an ABI 7900HT Fast Real-Time polymerase chain reaction system (Applied Bio-systems) according to manufacturer instructions (primer sequences available on request). Genotypes were called using TaqMan Genotyper Software, version 1.3 (Applied Bio-systems). Genotype call rates were 100% for each variant. There was no evidence of a deviation from Hardy-Weinberg equilibrium in controls (all P > .01 after Bonferroni correction). The 21 different MAPT haplotypes that were observed in 1% or more of individuals in any of the association analyses that were performed in our study are displayed in Table 2. Genotype frequencies for each variant are reported in eTable 1 in the Supplement.
Table 2. MAPT Haplotypes That Were Observed in 1% or More of Individuals in Any of the Association Analyses.
| Haplotype | MAPT Variant | |||||
|---|---|---|---|---|---|---|
| rs1467967 | rs242557 | rs3785883 | rs2471738 | rs8070723 | rs7521 | |
| H1b | G | G | G | C | A | A |
| H1c | A | A | G | T | A | G |
| H1d | A | A | G | C | A | A |
| H1e | A | G | G | C | A | A |
| H1f | G | G | A | C | A | A |
| H1g | G | A | A | C | A | A |
| H1h | A | G | A | C | A | A |
| H1i | G | A | G | C | A | A |
| H1k | A | A | A | C | A | G |
| H1l | A | G | A | C | A | G |
| H1m | G | A | G | C | A | G |
| H1o | A | A | A | C | A | A |
| H1p | G | G | G | T | A | G |
| H1q | A | A | G | T | A | A |
| H1r | A | G | G | T | A | G |
| H1u | A | A | G | C | A | G |
| H1v | G | G | A | T | A | G |
| H1x | G | A | A | T | A | G |
| H1y | A | A | A | T | A | G |
| H1z | G | A | G | T | A | G |
| H2 | A | G | G | C | G | G |
Neuropathologic Assessment
Semiquantitative measures of tau pathology in patients with PSP were assessed by a single neuropathologist (D.W.D.) on a severity scale of 0 to 3 (0 indicates none; 1, mild; 2, moderate; and 3, severe). All sections were processed identically with phospho-tau monoclonal antibody (CP13, from Peter Davies, PhD, Feinstein Institute, Long Island, New York), and immunochemistry was performed using a DAKO Autostainer. Four different measures of tau pathology (NFT, CB, TA, and NT) were scored in between 17 and 20 different neuroanatomical regions that are vulnerable to PSP (eTable 2 in the Supplement).25
For the primary analysis, 1 overall score for each of the 4 different measures of tau pathology (NFT, CB, TA, and NT) was created by calculating the mean of the semiquantitative scores (0-3) for each patient with PSP across all neuroanatomical regions; a higher overall score for a given measure of tau pathology indicates more severe tau pathology (Table 1). In calculating the overall scores for tau pathology for patients who did not have information available in a given region for a given measure of tau pathology, scores were imputed by using the mean of the values of the patients who did have this information available. Any patients who had missing data for more than 50% of neuroanatomical regions for a given measure of tau pathology were not included in analysis involving the overall score for that measure. Overall tau pathology scores were unavailable for CB in 1 patient, TA in 31 patients, and NT in 1 patient.
All PSP cases were assessed for Alzheimer-type pathology with thioflavin S fluorescent microscopy. Based on the density and distribution of plaques and tangles, Braak NFT stage26 and Thal amyloid phase27 were assessed for each case as previously described.28
Statistical Analysis
Statistical analysis was performed from February 17 to December 12, 2018. All statistical tests were 2-sided. Associations between 6 variant MAPT haplotypes and risk of PSP were examined using score tests for association under a logistic regression framework,29 in which tests were adjusted for age and sex. Odds ratios (ORs) and 95% CIs were estimated. In only the patients with PSP, associations between MAPT haplotypes and the 4 overall tau pathology scores (CB, NFT, TA, and NT) were evaluated using score tests of association under a linear regression framework29 with adjustment for age at death, sex, Braak stage, and Thal phase. Regression coefficients (β) and 95% CIs were estimated and are interpreted as the change in the mean overall tau pathology score corresponding to each additional copy of the given haplotype. Haplotypes were examined under an additive model (ie, the effect of each additional copy of the given haplotype) in all analyses.
In secondary analysis, MAPT haplotypes that displayed at least a nominally significant (P < .05) association with a given overall tau pathology score were further investigated by evaluating associations between the given haplotype and the severity scores in the individual neuroanatomical regions using score tests for association under a proportional odds logistic regression framework29 in which adjustment was made for age at death, sex, Braak stage, and Thal phase. Haplotypes occurring in less than 1% of individuals in a given association analysis were excluded from that analysis.
A Bonferroni correction for multiple testing was used in the primary analysis examining associations of MAPT haplotypes with risk of PSP and with the 4 overall tau pathology scores. Specifically, tests of association were performed for between 16 and 20 haplotypes depending on the outcome measure, after which P values between P ≤ .0031 (0.05/16) and P ≤ .0025 (0.05/20) were considered as statistically significant after Bonferroni correction. The necessary sample sizes to have 80% power to detect various degrees of association between haplotypes and outcomes at the aforementioned significance levels are shown in eTable 3 in the Supplement (associations with PSP) and eTable 4 in the Supplement (associations with tau pathology scores). Statistical analyses were performed using R statistical software, version 3.2.3 (R Foundation for Statistical Computing).
Results
Five different MAPT haplotypes were significantly associated with risk of PSP (Table 3, P ≤ .0025 considered significant). In addition to observing the previously reported associations with PSP risk for the H2 (OR, 0.16; 95% CI, 0.12-0.21; P = 7 × 10−49) and H1c (OR, 2.15; 95% CI, 1.76-2.62; P = 2 × 10−14) haplotypes, novel significant associations with PSP were observed for 3 other H1 subhaplotypes, including H1d (OR, 1.86; 95% CI, 1.43-2.42; P = 2 × 10−6), H1g (OR, 3.64; 95% CI, 2.04-6.50; P = 2 × 10−6), and H1o (OR, 2.60; 95% CI, 1.63-4.16; P = 2 × 10−5). The rare H1z haplotype was also associated with a higher risk of PSP, but this finding was not significant (OR, 3.06; 95% CI, 1.34-6.99; P = .004). Although the H1 haplotype is strongly associated with an increased risk of PSP, several H1 subhaplotypes (H1l and H1p) were associated with a nominally significant decreased risk of PSP (H1l: OR, 0.55; 95% CI, 0.33-0.90; P = .02; H1p: OR, 0.36; 95% CI, 0.13-0.95; P = .03) (Table 3).
Table 3. Association Between MAPT Haplotypes and Risk of PSPa.
| Haplotype | Haplotype Frequency, % | OR (95% CI) | P Value | |
|---|---|---|---|---|
| Patients With PSP (N = 802) | Controls (N = 1312) | |||
| H1b | 17.2 | 16.0 | 1.06 (0.87-1.29) | .57 |
| H1c | 19.8 | 11.3 | 2.15 (1.76-2.62) | 2 × 10−14 |
| H1d | 11.1 | 7.1 | 1.86 (1.43-2.42) | 2 × 10−6 |
| H1e | 8.7 | 9.0 | 0.99 (0.76-1.28) | .92 |
| H1f | 0.0 | 1.2 | 0.60 (0.21-1.74) | .29 |
| H1g | 3.7 | 1.1 | 3.64 (2.04-6.50) | 2 × 10−6 |
| H1h | 5.3 | 4.1 | 1.19 (0.82-1.71) | .36 |
| H1i | 5.0 | 4.4 | 1.56 (1.08-2.27) | .02 |
| H1l | 1.6 | 3.0 | 0.55 (0.33-0.90) | .02 |
| H1m | 2.2 | 2.9 | 0.77 (0.48-1.24) | .27 |
| H1o | 4.4 | 2.3 | 2.60 (1.63-4.16) | 2 × 10−5 |
| H1p | 0.3 | 1.5 | 0.36 (0.13-0.95) | .03 |
| H1q | 1.2 | 1.0 | 1.81 (0.89-3.68) | .10 |
| H1r | 1.5 | 1.1 | 1.15 (0.56-2.35) | .70 |
| H1u | 3.2 | 2.4 | 1.45 (0.89-2.35) | .13 |
| H1v | 0.8 | 1.2 | 0.44 (0.19-1.06) | .06 |
| H1x | 1.2 | 1.3 | 1.11 (0.53-2.03) | .78 |
| H1y | 0.8 | 1.6 | 1.03 (0.48-2.22) | .93 |
| H1z | 1.8 | 0.8 | 3.06 (1.34-6.99) | .004 |
| H2 | 4.7 | 22.7 | 0.16 (0.12-0.21) | 7 × 10−49 |
Abbreviations: OR, odds ratio; PSP, progressive supranuclear palsy.
The ORs (95% CIs) and P values result from score tests of association (under a logistic regression framework) that were adjusted for age and sex. The ORs correspond to each additional copy of the given haplotype. P ≤ .0025 was considered statistically significant after applying a Bonferroni correction for the 20 tests of association that were performed.
Table 4 displays associations of MAPT haplotypes with CB, NFT, TA, and NT overall tau pathology scores in patients with PSP. Although no haplotypes were significantly associated with overall tau pathology scores after correcting for multiple testing, a number of nominally significant associations were identified. Specifically, compared with other haplotypes, H2 was correlated with less severe tau pathology scores for CB (β, –0.10; 95% CI, –0.18 to –0.02; P = .01), NFT (β, –0.11; 95% CI, –0.18 to –0.03; P = .005), and NT (β, –0.10; 95% CI, –0.18 to –0.01; P = .03), while carriers of H1c had a more severe tau pathology for CB (β, 0.05; 95% CI, 0.01-0.09; P = .03) and TA (β, 0.05; 95% CI, 0.01-0.10; P = .02). In addition, the H1d haplotype, which was associated with greater risk of PSP, was also associated with worse tau pathology scores for TA (β, 0.07; 95% CI, 0.01-0.13; P = .03). Furthermore, H1e was associated with more severe tau pathology for NFT (β, 0.08; 95% CI, 0.02-0.14; P = .008), and H1q was associated with less severe pathology for both NFT (β, –0.17; 95% CI, –0.32 to –0.02; P = .02) and NT (β, –0.24; 95% CI, –0.41 to –0.06; P = .005).
Table 4. Associations of MAPT Haplotypes With CB, NFT, TA, and NT Overall Tau Pathology Scoresa.
| Haplotype | Haplotype Frequency in Patients With PSP, % | Association With CB Overall Tau Pathology Score | Association With NFT Overall Tau Pathology Score | Association With TA Overall Tau Pathology Score | Association With NT Overall Tau Pathology Score | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Regression Coefficient (95% CI) | P Value | Regression Coefficient (95% CI) | P Value | Regression Coefficient (95% CI) | P Value | Regression Coefficient (95% CI) | P Value | |||||
| H1b | 17.2 | −0.04 (−0.09 to 0.00) | .07 | 0.00 (−0.05 to 0.04) | .97 | −0.05 (−0.10 to 0.00) | .06 | 0.00 (−0.05 to 0.06) | .86 | |||
| H1c | 19.8 | 0.05 (0.01 to 0.09) | .03 | −0.01 (−0.05 to 0.03) | .71 | 0.05 (0.01 to 0.10) | .02 | 0.00 (−0.04 to 0.05) | .88 | |||
| H1d | 11.1 | 0.02 (−0.04 to 0.08) | .54 | 0.03 (−0.03 to 0.08) | .36 | 0.07 (0.01 to 0.13) | .03 | 0.05 (−0.02 to 0.11) | .17 | |||
| H1e | 8.7 | 0.00 (−0.06 to 0.07) | .95 | 0.08 (0.02 to 0.14) | .008 | −0.06 (−0.12 to 0.01) | .10 | 0.06 (−0.01 to 0.13) | .10 | |||
| H1g | 3.7 | 0.06 (−0.04 to 0.16) | .26 | 0.08 (−0.01 to 0.17) | .11 | 0.08 (−0.02 to 0.19) | .13 | 0.03 (−0.08 to 0.14) | .64 | |||
| H1h | 5.3 | 0.01 (−0.07 to 0.10) | .76 | 0.01 (−0.06 to 0.09) | .73 | −0.02 (−0.11 to 0.06) | .61 | −0.02 (−0.11 to 0.07) | .71 | |||
| H1i | 5.0 | 0.01 (−0.08 to 0.10) | .86 | 0.00 (−0.08 to 0.09) | .99 | 0.05 (−0.04 to 0.15) | .28 | −0.04 (−0.14 to 0.06) | .44 | |||
| H1kb | 1.0 | NA | NA | NA | NA | 0.08 (−0.15 to 0.30) | .54 | NA | NA | |||
| H1l | 1.6 | −0.10 (−0.25 to 0.06) | .22 | 0.06 (−0.08 to 0.20) | .44 | −0.02 (−0.18 to 0.14) | .80 | 0.03 (−0.13 to 0.20) | .70 | |||
| H1m | 2.2 | 0.08 (−0.05 to 0.21) | .24 | 0.08 (−0.04 to 0.20) | .19 | 0.04 (−0.09 to 0.18) | .57 | 0.04 (−0.10 to 0.17) | .63 | |||
| H1o | 4.4 | 0.01 (−0.09 to 0.11) | .82 | −0.04 (−0.13 to 0.05) | .41 | −0.04 (−0.14 to 0.06) | .46 | −0.01 (−0.11 to 0.10) | .88 | |||
| H1q | 1.2 | −0.02 (−0.18 to 0.14) | .84 | −0.17 (−0.32 to −0.02) | .02 | −0.15 (−0.32 to 0.01) | .07 | −0.24 (−0.41 to −0.06) | .005 | |||
| H1r | 1.5 | −0.08 (−0.24 to 0.08) | .33 | −0.01 (−0.16 to 0.14) | .93 | −0.11 (−0.28 to 0.05) | .18 | −0.04 (−0.22 to 0.13) | .65 | |||
| H1u | 3.2 | −0.07 (−0.18 to 0.05) | .25 | −0.03 (−0.14 to 0.07) | .55 | −0.05 (−0.17 to 0.07) | .45 | −0.06 (−0.19 to 0.06) | .33 | |||
| H1x | 1.2 | 0.08 (−0.05 to 0.21) | .46 | 0.08 (−0.04 to 0.20) | .64 | 0.04 (−0.09 to 0.18) | .55 | 0.04 (−0.10 to 0.17) | .98 | |||
| H1z | 1.8 | 0.11 (−0.05 to 0.27) | .22 | −0.11 (−0.26 to 0.03) | .15 | 0.06 (−0.11 to 0.23) | .50 | −0.08 (−0.26 to 0.09) | .40 | |||
| H2 | 4.7 | −0.10 (−0.18 to −0.02) | .01 | −0.11 (−0.18 to −0.03) | .005 | −0.02 (−0.10 to 0.07) | .67 | −0.10 (−0.18 to −0.01) | .03 | |||
Abbreviations: CB, coiled bodies; NA, not applicable; NFT, neurofibrillary tangles; NT, neuropil threads; PSP, progressive supranuclear palsy; TA, tufted astrocytes.
The regression coefficients (95% CIs) and P values result from score tests of association (under a linear regression framework) that were adjusted for age at death, sex, Braak stage, and Thal phase. Regression coefficients are interpreted as the change in the mean overall tau pathology score corresponding to each additional copy of the given haplotype. Overall tau pathology scores were unavailable for CB (1 patient), TA (31 patients), and NT (1 patient); these patients were not included in haplotype analysis involving the given overall tau pathology score. P ≤ .0031 (analysis involving CB, NFT, and NT overall tau pathology scores; 16 tests each) or P ≤ .0029 (analysis involving TA overall tau pathology score; 17 tests) were considered statistically significant after applying Bonferroni correction.
Indicates a haplotype that was observed at an overall frequency of less than 1% in a particular analysis and was therefore excluded from that analysis.
We examined these nominally significant findings in further detail by assessing associations between the aforementioned haplotypes and the measures of tau pathology in each separate neuroanatomical region (eTable 5 in the Supplement). A number of associations were observed, the strongest of which occurred for H1c and CB (inferior olive and medullary tegmentum), H1c and TA (basal nucleus), H1e and NFT (midbrain tectum), H1q and NT (caudate putamen), H2 and NFT (oculomotor complex), and H2 and NT (medullary tegmentum) (eTable 6 in the Supplement).
Discussion
The MAPT H1 haplotype is the strongest genetic risk factor for PSP that has been identified to date.13 However, other than an increased risk of PSP that is associated with the H1c subhaplotype (which has also been associated with increased MAPT expression30), the specific H1 subhaplotypes that drive the H1-associated risk of PSP have not been well defined. Furthermore, a comprehensive evaluation of how specific H1 subhaplotypes may affect pathologic severity of PSP has not been performed. In this study, we have identified novel associations with the risk of PSP for 3 H1 subhaplotypes (H1d, H1g, and H1o), with an additional subhaplotype (H1z) displaying a suggestive association. In addition, we noted potential weaker associations between several haplotypes and severity of tau pathology. These associations included the H1, H1c, and H1d disease risk haplotypes, suggesting that MAPT haplotypic variation may alter both the risk of PSP and the severity of disease.
The association between MAPT H2 and risk of PSP that we observed is consistent with that observed in previous studies7,8,9,10,11,12 and the PSP GWAS.13 In the GWAS, a significant association between MAPT rs242557 and PSP was identified that was independent of H1 (rs8070723). MAPT rs242557 partially tags the H1c haplotype and, like H1c, has been shown to influence MAPT expression.10 The GWAS authors highlight the fact that the rs242557 association (OR, 1.96) accounts for only part of the total risk that is associated with H1 (OR, 5.46); our study sheds light on this finding. Specifically, although all of the significant PSP risk H1 subhaplotypes in our study are associated with the rs242557 risk allele (A), the H1m subhaplotype that is associated with the rs242557 risk allele was observed less frequently in PSP cases than in controls, while, conversely, 3 subhaplotypes (H1b, H1h, and H1r) associated with the rs242557 protective allele were observed more commonly in patients with PSP. Thus, the MAPT rs242557 variant is not consistently associated with a greater risk of PSP depending on the haplotype. Our results indicate that the increased risk of PSP associated with the H1 haplotype is driven primarily by the aforementioned 4 to 5 specific H1 subhaplotypes and that haplotypic structure is best taken into account to most precisely understand how MAPT variation alters the risk of PSP.
In a study similar to ours, pathologically diagnosed patients with PSP and controls were genotyped for the same 6 variants to define MAPT haplotypes in a case-control series from the United Kingdom (83 PSP cases and 169 controls) and the United States (238 PSP cases and 131 controls)11; H2 and H1c were the only haplotypes that were significantly associated with PSP. However, power to detect an association would have been low in the relatively small series. Regardless of statistical significance, the authors observed higher frequencies of the H1c and H1g subhaplotypes in both series, while results were discordant between the 2 series for H1d and H1o. The inconsistently varying direction of association for H1d and H1o between our study and the previous study is likely owing to imprecise haplotype frequency estimates in the smaller UK and US series in the latter study. This finding underscores the need for large studies when examining MAPT haplotypes, as many of the H1 subhaplotypes are relatively rare.
H1 subhaplotypes have previously been assessed for association with other neurologic diseases. Several of the PSP risk haplotypes that we identified have previously been associated with Alzheimer disease (H1c),30,31 corticobasal degeneration (H1c),11 Parkinson disease (H1d and H1o),32 and dementia with Lewy bodies (H1g).14 The identification of multiple novel PSP risk H1 subhaplotypes in our study further highlights the heterogeneity of the role of MAPT variation in susceptibility to different neurodegenerative disorders.
We also observed associations between several MAPT haplotypes and severity of tau pathology. Specifically, when considering overall tau pathology scores, nominally significant associations were noted for 3 of the PSP risk haplotypes (H2, H1c, and H1d), with additional associations identified for H1e and H1q. These findings did not survive correction for multiple testing and, correspondingly, are much weaker compared with the strong disease risk associations; with tau pathology scored as 0 (none), 1 (mild), 2 (moderate), and 3 (severe), haplotypes that displayed nominally significant associations had fairly modest mean increases or reductions in tau pathology scores between 0.05 and 0.24 for each additional copy of the given haplotype. Nonetheless, the direction of association with tau pathology scores (higher or lower) was in agreement with that of disease risk (risk or protection) for the H2, H1c, and H1d haplotypes. The H2 haplotype was most robustly correlated with severity of tau pathology, displaying associations with lower scores for CB, NFT, and NT but not TA. Given the associations between these 3 haplotypes and the risk of PSP, an additional role in modifying the severity of disease seems biologically plausible. As the H1e and H1q subhaplotypes were not associated with risk of PSP, their associations with tau burden are less certain. All these associations appear to have been driven by findings in a subset of neuroanatomical regions, where in some cases associations with haplotypes were fairly strong. The mechanism by which MAPT haplotypes alter severity of tau pathology will be an important topic for future study.
Several previous studies have attempted to evaluate the role of the MAPT haplotypes in influencing clinical and pathologic severity of PSP, most of which have focused solely on H1 or the MAPT rs242557 variant. In a previous investigation involving a subset of approximately 50% of the PSP cases included in our study, H2 was not associated with overall severity of NFT, CB, TA, or NT, whereas rs242557 was associated with more severe CB and TA20; neuroanatomical region–specific analyses were not performed. Our larger study enabled the detection of stronger and nominally significant associations with H2, and our findings suggest that the associations that were previously observed for rs242557 may have been driven by the specific H1c and H1d haplotypes. In clinical studies, although sample sizes have mostly been small, the H1 haplotype has been inconsistently linked with an earlier age at PSP onset17,18,19 but has not been associated with symptom severity,18 disease duration,18 or survival.17,18 Although larger clinical studies are needed, given that the nominally significant associations between MAPT haplotypes and tau burden that we observed were relatively modest, the lack of a dramatic association with clinical presentation would not be surprising. It is also important to consider the biological effect of our observed haplotype associations, as it is likely that the alternate haplotypes have an effect on MAPT isoform expression that may be tissue specific. It will be crucial to examine this expression in appropriate brain tissue from carriers of both common and rare haplotypes in both patients with PSP and unaffected individuals in future studies.
Limitations
Although the strengths of our study (definitive diagnosis of PSP and relatively large sample size) are important to highlight, several limitations should be considered. Although we identified novel associations between several H1 subhaplotypes and risk of PSP, our study did not include a replication series; therefore, validation of these findings (particularly those involving rare H1 subhaplotypes) will be important. The controls were clinically rather than neuropathologically assessed, and although unlikely, genetic mosaicism could occur between tissue types. In addition, the sample size of our study, although large for a study on PSP, is still somewhat limited for a genetic association study. Therefore, the lack of power to detect mild to moderate associations and corresponding possibility of a type II error (ie, false-negative finding) is important to consider, particularly for rare haplotypes and disease risk analysis.
Conclusions
Our findings advance the current understanding of the role of the MAPT gene in altering susceptibility to PSP by demonstrating that the increased risk of PSP that is associated with the H1 haplotype is driven by 4 to 5 specific H1 subhaplotypes. In addition, several of these PSP risk haplotypes may also have a smaller influence on the severity of tau pathology.
eTable 1. Allele and Genotype Counts and Frequencies
eTable 2. Summary of Semi-Quantitative Tau Pathology Measures for CB, NFT, TA, and NT
eTable 3. Number of PSP Patients and Controls Needed to Have 80% Power at the P < .0025 Significance Level to Detect Given Odds Ratios at Various Haplotype Frequencies in Controls
eTable 4. Number of PSP Patients Needed to Have 80% Power at the P < .0031 (Analysis of CB, NFT, and NT Overall Tau Pathology Scores) and P < .0029 (Analysis of TA Overall Tau Pathology Score) Significance Levels to Detect Given Mean Differences at Various Haplotype Frequencies
eTable 5. P Values for Association Between MAPT Haplotypes and Semi-Quantitative Tau Pathology Measures (0 = none; 1 = mild; 2 = moderate; 3 = severe) in Each Separate Neuroanatomical Region for Haplotypes That Showed a Nominally Significant (P ≤ .05) Association With Overall Tau Pathology Scores
eTable 6. Details of Strongest Associations (P ≤ .001) Between MAPT Haplotypes and Semi-Quantitative Tau Pathology Measures in Specific Neuroanatomical Regions
References
- 1.Höglinger GU, Respondek G, Stamelou M, et al. ; Movement Disorder Society-endorsed PSP Study Group . Clinical diagnosis of progressive supranuclear palsy: the Movement Disorder Society criteria. Mov Disord. 2017;32(6):853-864. doi: 10.1002/mds.26987 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Dickson DW, Ahmed Z, Algom AA, Tsuboi Y, Josephs KA. Neuropathology of variants of progressive supranuclear palsy. Curr Opin Neurol. 2010;23(4):394-400. doi: 10.1097/WCO.0b013e32833be924 [DOI] [PubMed] [Google Scholar]
- 3.Josephs KA, Dickson DW. Diagnostic accuracy of progressive supranuclear palsy in the Society for Progressive Supranuclear Palsy brain bank. Mov Disord. 2003;18(9):1018-1026. doi: 10.1002/mds.10488 [DOI] [PubMed] [Google Scholar]
- 4.Boxer AL, Yu JT, Golbe LI, Litvan I, Lang AE, Höglinger GU. Advances in progressive supranuclear palsy: new diagnostic criteria, biomarkers, and therapeutic approaches. Lancet Neurol. 2017;16(7):552-563. doi: 10.1016/S1474-4422(17)30157-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Scholz SW, Bras J. Genetics underlying atypical parkinsonism and related neurodegenerative disorders. Int J Mol Sci. 2015;16(10):24629-24655. doi: 10.3390/ijms161024629 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Stefansson H, Helgason A, Thorleifsson G, et al. . A common inversion under selection in Europeans. Nat Genet. 2005;37(2):129-137. doi: 10.1038/ng1508 [DOI] [PubMed] [Google Scholar]
- 7.Conrad C, Andreadis A, Trojanowski JQ, et al. . Genetic evidence for the involvement of tau in progressive supranuclear palsy. Ann Neurol. 1997;41(2):277-281. doi: 10.1002/ana.410410222 [DOI] [PubMed] [Google Scholar]
- 8.Baker M, Litvan I, Houlden H, et al. . Association of an extended haplotype in the tau gene with progressive supranuclear palsy. Hum Mol Genet. 1999;8(4):711-715. doi: 10.1093/hmg/8.4.711 [DOI] [PubMed] [Google Scholar]
- 9.Pastor P, Ezquerra M, Perez JC, et al. . Novel haplotypes in 17q21 are associated with progressive supranuclear palsy. Ann Neurol. 2004;56(2):249-258. doi: 10.1002/ana.20178 [DOI] [PubMed] [Google Scholar]
- 10.Rademakers R, Melquist S, Cruts M, et al. . High-density SNP haplotyping suggests altered regulation of tau gene expression in progressive supranuclear palsy. Hum Mol Genet. 2005;14(21):3281-3292. doi: 10.1093/hmg/ddi361 [DOI] [PubMed] [Google Scholar]
- 11.Pittman AM, Myers AJ, Abou-Sleiman P, et al. . Linkage disequilibrium fine mapping and haplotype association analysis of the tau gene in progressive supranuclear palsy and corticobasal degeneration. J Med Genet. 2005;42(11):837-846. doi: 10.1136/jmg.2005.031377 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Melquist S, Craig DW, Huentelman MJ, et al. . Identification of a novel risk locus for progressive supranuclear palsy by a pooled genomewide scan of 500,288 single-nucleotide polymorphisms. Am J Hum Genet. 2007;80(4):769-778. doi: 10.1086/513320 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Höglinger GU, Melhem NM, Dickson DW, et al. ; PSP Genetics Study Group . Identification of common variants influencing risk of the tauopathy progressive supranuclear palsy. Nat Genet. 2011;43(7):699-705. doi: 10.1038/ng.859 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Labbé C, Heckman MG, Lorenzo-Betancor O, et al. . MAPT haplotype H1G is associated with increased risk of dementia with Lewy bodies. Alzheimers Dement. 2016;12(12):1297-1304. doi: 10.1016/j.jalz.2016.05.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Setó-Salvia N, Clarimón J, Pagonabarraga J, et al. . Dementia risk in Parkinson disease: disentangling the role of MAPT haplotypes. Arch Neurol. 2011;68(3):359-364. doi: 10.1001/archneurol.2011.17 [DOI] [PubMed] [Google Scholar]
- 16.Labbé C, Heckman MG, Lorenzo-Betancor O, et al. . MAPT haplotype diversity in multiple system atrophy. Parkinsonism Relat Disord. 2016;30:40-45. doi: 10.1016/j.parkreldis.2016.06.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Litvan I, Baker M, Hutton M. Tau genotype: no effect on onset, symptom severity, or survival in progressive supranuclear palsy. Neurology. 2001;57(1):138-140. doi: 10.1212/WNL.57.1.138 [DOI] [PubMed] [Google Scholar]
- 18.Baba Y, Putzke JD, Tsuboi Y, et al. . Effect of MAPT and APOE on prognosis of progressive supranuclear palsy. Neurosci Lett. 2006;405(1-2):116-119. doi: 10.1016/j.neulet.2006.06.032 [DOI] [PubMed] [Google Scholar]
- 19.Cruchaga C, Vidal-Taboada JM, Ezquerra M, et al. ; Iberian Atypical Parkinsonism Study Group Researchers . 5′-Upstream variants of CRHR1 and MAPT genes associated with age at onset in progressive supranuclear palsy and cortical basal degeneration. Neurobiol Dis. 2009;33(2):164-170. doi: 10.1016/j.nbd.2008.09.027 [DOI] [PubMed] [Google Scholar]
- 20.Allen M, Burgess JD, Ballard T, et al. . Gene expression, methylation and neuropathology correlations at progressive supranuclear palsy risk loci. Acta Neuropathol. 2016;132(2):197-211. doi: 10.1007/s00401-016-1576-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Gerstenecker A, Roberson ED, Schellenberg GD, et al. . Genetic influences on cognition in progressive supranuclear palsy. Mov Disord. 2017;32(12):1764-1771. doi: 10.1002/mds.27196 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hauw JJ, Daniel SE, Dickson D, et al. . Preliminary NINDS neuropathologic criteria for Steele-Richardson-Olszewski syndrome (progressive supranuclear palsy). Neurology. 1994;44(11):2015-2019. doi: 10.1212/WNL.44.11.2015 [DOI] [PubMed] [Google Scholar]
- 23.Labbé C, Soto-Ortolaza AI, Rayaprolu S, et al. . Investigating the role of FUS exonic variants in essential tremor. Parkinsonism Relat Disord. 2013;19(8):755-757. doi: 10.1016/j.parkreldis.2013.03.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Allen M, Kachadoorian M, Quicksall Z, et al. . Association of MAPT haplotypes with Alzheimer’s disease risk and MAPT brain gene expression levels. Alzheimers Res Ther. 2014;6(4):39. doi: 10.1186/alzrt268 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Dickson DW, Hauw J-J, Agid Y, Litvan I Progressive supranuclear palsy and corticobasal degeneration. In: Dickson DW, Weller RO, eds. Neurodegeneration: The Molecular Pathology of Dementia and Movement Disorders. 2nd ed. Hoboken, NJ: Wiley-Blackwell; 2011. [Google Scholar]
- 26.Braak H, Braak E. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol. 1991;82(4):239-259. doi: 10.1007/BF00308809 [DOI] [PubMed] [Google Scholar]
- 27.Thal DR, Rüb U, Orantes M, Braak H. Phases of A beta-deposition in the human brain and its relevance for the development of AD. Neurology. 2002;58(12):1791-1800. doi: 10.1212/WNL.58.12.1791 [DOI] [PubMed] [Google Scholar]
- 28.Murray ME, Lowe VJ, Graff-Radford NR, et al. . Clinicopathologic and 11C-Pittsburgh compound B implications of Thal amyloid phase across the Alzheimer’s disease spectrum. Brain. 2015;138(pt 5):1370-1381. doi: 10.1093/brain/awv050 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Schaid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA. Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am J Hum Genet. 2002;70(2):425-434. doi: 10.1086/338688 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Myers AJ, Kaleem M, Marlowe L, et al. . The H1c haplotype at the MAPT locus is associated with Alzheimer’s disease. Hum Mol Genet. 2005;14(16):2399-2404. doi: 10.1093/hmg/ddi241 [DOI] [PubMed] [Google Scholar]
- 31.Myers AJ, Pittman AM, Zhao AS, et al. . The MAPT H1c risk haplotype is associated with increased expression of tau and especially of 4 repeat containing transcripts. Neurobiol Dis. 2007;25(3):561-570. doi: 10.1016/j.nbd.2006.10.018 [DOI] [PubMed] [Google Scholar]
- 32.Vandrovcova J, Pittman AM, Malzer E, et al. . Association of MAPT haplotype-tagging SNPs with sporadic Parkinson’s disease. Neurobiol Aging. 2009;30(9):1477-1482. doi: 10.1016/j.neurobiolaging.2007.11.019 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Allele and Genotype Counts and Frequencies
eTable 2. Summary of Semi-Quantitative Tau Pathology Measures for CB, NFT, TA, and NT
eTable 3. Number of PSP Patients and Controls Needed to Have 80% Power at the P < .0025 Significance Level to Detect Given Odds Ratios at Various Haplotype Frequencies in Controls
eTable 4. Number of PSP Patients Needed to Have 80% Power at the P < .0031 (Analysis of CB, NFT, and NT Overall Tau Pathology Scores) and P < .0029 (Analysis of TA Overall Tau Pathology Score) Significance Levels to Detect Given Mean Differences at Various Haplotype Frequencies
eTable 5. P Values for Association Between MAPT Haplotypes and Semi-Quantitative Tau Pathology Measures (0 = none; 1 = mild; 2 = moderate; 3 = severe) in Each Separate Neuroanatomical Region for Haplotypes That Showed a Nominally Significant (P ≤ .05) Association With Overall Tau Pathology Scores
eTable 6. Details of Strongest Associations (P ≤ .001) Between MAPT Haplotypes and Semi-Quantitative Tau Pathology Measures in Specific Neuroanatomical Regions