Evaluating the causal effect of mitochondrial dysfunction on Alzheimer's disease and Parkinson's disease using polygenic risk scores and Mendelian randomization

Aadrita Chatterjee; Brian Alvarez; Rakshya U Sharma; Caroline Jonson; Heather M Wilkins; Judy Pa; Russell H Swerdlow; Alison Goate; Kristine Yaffe; Shea J Andrews; for the Alzheimer's Disease Genetics Consortium and the Alzheimer's Disease Neuroimaging Initiative

doi:10.1002/alz.71469

. 2026 May 20;22(5):e71469. doi: 10.1002/alz.71469

Evaluating the causal effect of mitochondrial dysfunction on Alzheimer's disease and Parkinson's disease using polygenic risk scores and Mendelian randomization

Aadrita Chatterjee ¹, Brian Alvarez ², Rakshya U Sharma ¹, Caroline Jonson ^3,^4,⁵, Heather M Wilkins ^6,^7,⁸, Judy Pa ^9,¹⁰, Russell H Swerdlow ^6,^7,⁸, Alison Goate ², Kristine Yaffe ^1,^5,¹¹, Shea J Andrews ^1,^✉; for the Alzheimer's Disease Genetics Consortium and the Alzheimer's Disease Neuroimaging Initiative

PMCID: PMC13239222 PMID: 42163007

Abstract

INTRODUCTION

Mitochondrial DNA copy number (mtDNAcn), a measure of mitochondrial genomes per nucleated cell, has an unclear causal relationship with Alzheimer's disease (AD) and Parkinson's disease (PD). We integrated genetic correlation, polygenic risk scores (PRSs), and Mendelian randomization (MR) to assess whether mtDNAcn influences the risk of AD and PD, and evaluate how study‐specific factors in mtDNAcn genome‐wide association studies (GWASs) distort these causal estimates.

METHODS

Using GWASs of four mtDNAcn measures, AD, AD/dementia, and PD, we evaluated genetic correlations, generated ancestry‐normalized PRS in the Alzheimer's Disease Genetics Consortium (N = 27,383), and applied MR methods including latent heritable confounder‐MR (LHC‐MR).

RESULTS

Across the four mtDNAcn GWASs, only one was consistently associated with AD/dementia and PD, with genetic correlations and PRSs showing negative correlations and MR indicating that higher mtDNAcn reduced AD/dementia and PD risk.

DISCUSSION

Higher blood‐based mtDNAcn was causally associated with reduced risk of AD/dementia and PD, with limited evidence to suggest a bidirectional effect.

Keywords: Alzheimer's disease, genetic correlations, Mendelian randomization, mitochondrial DNA copy number, Parkinson's disease, polygenic risk scores

Highlights

Genetically predicted higher mitochondrial DNA copy number (mtDNAcn) is causally associated with lower Alzheimer's disease (AD) and Parkinson's disease (PD) risk.
AD genetic liability is causally associated with higher mtDNAcn, possibly as a compensatory response.
mtDNAcn is a potential early biomarker of mitochondrial dysfunction in AD/PD.

1. BACKGROUND

Mitochondria are membrane‐bound organelles that play a central role in essential cellular processes, including managing oxidative stress, beta‐oxidation of fatty acids for energy production, and facilitating programmed cell death. ¹ These processes work together to maintain mitochondrial integrity and ensure efficient cellular function by removing or replacing damaged or dysfunctional mitochondria. ¹ An important marker of mitochondrial health is mitochondrial DNA copy number (mtDNAcn), which refers to the number of mitochondrial genomes present per nucleated cell. ² This measure has been widely used as a proxy for mitochondrial function and is implicated in various physiological and pathological conditions, including aging, metabolic disorders, and neurodegenerative diseases. ² The current gold standard for measuring mtDNAcn is quantitative polymerase chain reaction (qPCR); however, mtDNAcn can also be estimated using microarray probe intensities and whole‐genome sequencing (WGS). ³

Mitochondrial dysfunction plays a prominent role in two of the most common neurodegenerative diseases—Alzheimer's disease (AD) and Parkinson's disease (PD). ⁴ , ⁵ Higher mtDNAcn has predominantly been associated with reduced risk of AD and PD, better cognitive function, and lower tau pathology. ⁶ , ⁷ , ⁸ Confounders, such as cell type heterogeneity, and reverse causation, complicate the interpretation of associations between mtDNAcn and neurodegenerative disorders. Different cell types contain varying levels of mitochondrial DNA; for instance, platelets, due to their high energy demands, contain significantly more mitochondrial DNA than other cells but lack nuclear DNA. ² Additionally, elevated mtDNAcn may reflect compensatory upregulation—an adaptive response to impaired mitochondrial function—raising the question of whether mtDNAcn is a cause or consequence of disease. ²

Mendelian randomization (MR) studies use genetic instruments to determine the causal effect of an exposure on an outcome, and they are less susceptible to confounding and reverse causation. MR studies examining the causal effect of mtDNAcn on AD and PD have reported conflicting findings. While some two‐sample MR analyses suggest that higher blood mtDNAcn is causally associated with reduced AD risk, other studies have found no strong evidence of a causal effect on either AD or PD. ⁵ , ⁷ , ⁹ , ¹⁰ Furthermore, these analyses often do not account for unmeasured confounding and typically rely on a limited set of genetic variants.

In this study, we integrate genetic correlation, polygenic risk score (PRS), and multiple MR frameworks to assess whether genetically predicted blood mtDNAcn influences the risk of AD and PD, and to evaluate how study‐specific factors in mtDNAcn genome‐wide association studies (GWASs) may distort these causal estimates. Genetic correlations test whether there is shared genetic architecture between traits due to pleiotropy. ¹¹ PRS tests whether genetic predisposition to one particular trait, such as mtDNAcn, is associated with a second trait. ¹² However, while genetic correlations and PRS establishes correlation, they do not infer causality.

To assess the causal direction of the effect of mtDNAcn on AD/PD, we can use MR, applying state‐of‐the‐art approaches such as latent heritable confounder MR (LHC‐MR) to account for unmeasured confounding and obtain unbiased causal estimates. ¹³ , ¹⁴ Furthermore, because not all mtDNAcn GWASs adjust for platelet count—a major biological confounder—we also implement multivariable MR (MVMR) to estimate both direct and indirect effects of each exposure on the outcome. ¹⁵ Finally, by leveraging multiple studies that quantify mtDNAcn using different methods, we can assess the robustness of our findings and determine whether the observed causal effects reflect true biological relationships or are influenced by study‐specific measurement biases. Together, these approaches allow us to test both association and causation while addressing key methodological limitations in prior work.

2. METHODS

We used a combination of analytical approaches to investigate whether mtDNAcn plays a causal role in AD and PD. Genetic correlations and PRSs were used to determine whether mtDNAcn has a shared genetic architecture with AD and PD, establishing a basis for causal inference. Forward direction MR analyses were used to evaluate whether genetically determined mtDNAcn is causally associated with AD or PD. Last, we applied MVMR to adjust for platelets and LHC‐MR to provide more robust estimates and confirm univariate MR analyses, including accounting for sample overlap between exposure and outcome datasets. An overview of the study design is provided in Figure S1 in supporting information.

2.1. Data sources

We used GWAS summary statistics for each exposure and outcome dataset (Table 1 and Table S1 in supporting information). The first GWAS used for AD involved clinically diagnosed AD cases (AD; n = 94,437, n _loci = 25 late‐onset AD risk loci). ¹⁶ The second GWAS used clinical case–control status and family history of dementia to perform a GWAS‐by‐proxy (GWAX; n = 788,989, n _loci = 72) of AD and related dementias (AD/dementia). ¹⁷ The GWAS for PD (n ≈ 1.4 million, n _loci = 90) used 37.7K cases, 18.6K UK Biobank proxy cases (having a first‐degree relative with PD), and 1.4M controls. ¹⁸

TABLE 1.

Description of GWAS datasets used in this study.

Study	Trait	Cohort/ consortium	N	# of loci	Ancestry	Covariates	mtDNAcn measurement
Kunkle et al. ¹⁶	Late‐onset Alzheimer's disease	IGAP	94,437	25	European	Age, sex, PCs	N/A
Bellenguez et al. ¹⁷	AD/dementia	EADB consortium and the UK Biobank	487,511	72	European	Age, sex, PCs, APOE	N/A
Astle et al. 2016 ²³	Platelets	UK Biobank and INTERVAL	173,480	2706	European	Age, sex, BMI, smoking, alcohol consumption	N/A
Chong et al. ²¹	Mitochondrial copy number	UK Biobank	395,781	71	European	Age, sex, 20 genetic PCs, blood cell composition	Microarray
Hägg et al. ²⁰	Mitochondrial copy number	UK Biobank	295,150	50	European	Age, sex, white cell counts, BMI, genotyping batch, total autosomal DNA abundance	Microarray
Longchamps et al. 2022 ¹⁹	Mitochondrial copy number	(CHARGE) consortium/UK Biobank	470,579	97	European	Age, sex and cell counts	qPCR, microarray, whole genome sequencing
Gupta et al. ²²	Mitochondrial copy number	UK Biobank, All of Us and 1000G	163,372	46	European, African American, Hispanic, Asian, Middle Eastern	Age, sex, blood cell composition traits including platelets, time of day, month of year and fasting duration	Whole genome sequencing
Nalls et al. ¹⁸	Parkinson's disease	International Parkinson's Disease Genomics Consortium, 23andMe, UK Biobank	1.4 million	90	European	Age, sex, PCs	N/A

Open in a new tab

Abbreviations: 1000G, 100 Genomes Project; AD, Alzheimer's disease; APOE, apolipoprotein E; BMI, body mass index; CHARGE, Cohorts for Heart and Aging Research in Genomic Epidemiology; EADB, European Alzheimer & Dementia Biobank; GWAS, genome‐wide association study; IGAP, International Genomics of Alzheimer's Project; mtDNAcn, mitochondrial DNA copy number; PC, principal component; qPCR, quantitative polymerase chain reaction.

RESEARCH IN CONTEXT

Systematic review: Mitochondrial dysfunction, measured by mitochondrial DNA copy number (mtDNAcn), has been linked to Alzheimer's disease (AD) and Parkinson's disease (PD). However, Mendelian randomization (MR) studies on this relationship have shown inconsistent results, have not applied advanced MR methods that address prior limitations, or examined study‐specific biases.
Interpretation: Using genetic correlations, polygenic scores, and MR, we triangulated evidence across complementary methods. We found that results varied depending on the dataset (e.g., clinically diagnosed AD vs. family history of AD) and study design factors such as mtDNAcn measurement techniques. Despite these biases, higher mtDNAcn was consistently associated with a lower risk of AD and PD, supporting a mitochondrial mechanism in both diseases.
Future directions: Our findings highlight mtDNAcn as a potential biomarker for AD/PD, emphasizing the importance of measurement methods. Future research is needed to explore the biological pathways underlying this relationship.

Blood‐derived mtDNAcn came from four different GWASs. Longchamps et al. (n = 470,579, n _loci = 97) quantified mtDNAcn through WGS, qPCR, and microarray probe intensities. ¹⁹ Hägg et al. (n = 295,150, n _loci = 50) estimated mtDNAcn by the weighted intensities of genotyping probes. ²⁰ Chong et al. (n = 395,781, n _loci = 71) measured mtDNAcn by using a microarray‐based method called automatic mitochondrial copy number (AutoMitoC). ²¹ This was unique in the following ways: (1) replacement of autosomal signal normalization of common variants with globally rare variants, (2) assessment of the association of corrected probe signal intensities with off‐target genome intensities to detect cross‐hybridizing probes, and (3) mitochondrial signal is ascertained via principal component analysis, rather than by median signal intensity. Finally, Gupta et al. (n = 163,372, n _loci = 46) quantified mtDNAcn using WGS. ²²

To account for platelet abundance in MVMR, we used a GWAS for platelets (n = 173,480, n _loci = 2706). ²³

All datasets included age and sex as covariates and were standardized using the gwas2vcf pipeline. ²⁴ Table S1 details the study design, eligibility, and quality control measures.

2.2. Genetic correlations

We formatted and munged GWAS using Linkage Disequilibrium Score Regression (LDSC) version 1.0 and estimated genetic correlations between each trait using GeNetic cOVariance Analyzer (GNOVA). ²⁵ , ²⁶ GNOVA is a principled framework that leverages GWAS summary statistics to assess annotation‐stratified genetic covariance between traits and generally provides higher statistical power and more precise genetic covariance/genetic correlation estimates than LDSC, particularly when single nucleotide polymorphism (SNP) heritability is modest, while also correcting for sample overlap. ²⁶ Genetic correlations allow us to see the shared genetic basis of our traits of interest and can result from horizontal or vertical pleiotropy. They can be quantified as strong (rg > 0.6), medium (rg 0.2–0.6), or weak (rg < 0.2). ¹¹ To correct for multiple testing, false discovery rate (FDR) adjustment was applied to genetic correlations.

2.3. PRSs

PRSs estimate an individual's genetic liability to a trait or disease by aggregating the effects of multiple SNPs, each weighted by its effect size derived from a GWAS. ¹² We constructed mtDNAcn PRS using data from the Alzheimer's Disease Genetics Consortium (ADGC), one of the largest multi‐ancestry AD datasets. Genotyping and imputation procedures for the contributing cohorts have been described previously. ²⁷ For this study, we applied genotype quality control to TOPMed‐imputed datasets using GenoTools. ²⁸ Variants were filtered to exclude those with a call rate < 95% and minor allele frequency < 1%. Samples were evaluated for relatedness (kinship coefficient > 0.0884) and excessive heterozygosity. Related individuals and potential duplicate samples (Identity‐by‐state [IBS] > 0.354) were identified and removed, leaving one representative sample per pair.

mtDNAcn summary statistics were filtered to HapMap3 SNPs, and score files were generated using PRS‐CS‐auto (v1.1.0) to automatically learn the global shrinkage parameter (ϕ) using a fully Bayesian approach. ²⁹ PRSs were calculated and applied to the ADGC using PGSC‐calc (v2.0.1) for ancestry normalization. ³⁰ Genotypes were lifted over to the GRCh38 reference genome and matched to the ADGC dataset using our custom score file. ³⁰ We used the Human Genome Diversity Project + 1000 Genomes Project reference panel for ancestry inference and excluded the apolipoprotein E (APOE) ε4 region (19:44408822‐19:45408822, build 38) due to confounding. ³⁰ To determine the association of PRSs with AD status in ADGC, we performed logistic regression adjusting for age, sex, APOE ε4 status, and population stratification (PC 1–4).

To validate that our PRS models predicted mtDNAcn, we evaluated the association of each mtDNAcn PRS with mtDNAcn in participants contributing to ADGC from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database (Table 2 and Table S2 in supporting information). ADNI was launched in 2003 as a public–private partnership, led by Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test whether serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early AD. Demographic, clinical, and genomic data have been previously described. ³¹ Blood‐derived mtDNAcn was estimated from WGS as previously described. ³² Linear regression models adjusting for age, sex, APOE ε4 genotypes, diagnosis, and population stratification (PC 1–4) were used to examine the association of mtDNAcn PRS with mtDNAcn.

TABLE 2.

Demographic characteristics of ADGC.

	Alzheimer's disease	Cognitively unimpaired
	N = 17,005 ^a	N = 10,378 ^a
Age	75.9 ± 8.3	73.2 ± 15.8
Sex
Female	10,496 (62%)	6112 (59%)
Male	6509 (38%)	4266 (41%)
Ancestry
AFR	1228 (7.2%)	629 (6.1%)
AMR	1191 (7.0%)	674 (6.5%)
EAS	878 (5.2%)	920 (8.9%)
EUR	13,708 (81%)	8155 (79%)
APOE ε4 carrier
Non‐carrier	12,539 (74%)	4331 (42%)
Carrier	4466 (26%)	6047 (58%)

Open in a new tab

Abbreviations: ADGC, Alzheimer's Disease Genetics Consortium; AFR, African; AMR, Admixed American; EAS, East Asian; EUR, European; APOE, apolipoprotein E; SD, standard deviation.

^{^a}

Mean ± SD; n (%).

2.4. Mendelian randomization

2.4.1. Genetic instrument selection and harmonization of data

To select genetic instruments and harmonize the data, we applied clumping (r ² = 0.001, 10Mb window, EUR reference population) to keep independent genome‐wide significant SNPs (p < 5e‐8). We used LDlinkR version 5 (EUR reference, r ² > 0.8) to find linkage disequilibrium (LD) proxies for instrumental variables absent in the outcome GWAS dataset. ³³ The exposure and outcome datasets were harmonized so that SNP effect estimates aligned to the same effect allele, and we used allele frequency information to infer palindromic SNPs. ³⁴ , ³⁵ The APOE region (19:44912079–19:45912079, build 37) was excluded due to its pleiotropic effects and variants in this locus have very large effects on the outcome (AD risk) and can therefore disproportionately influence MR estimates and outlier detection, independent of their association with the exposure. ³⁶ , ³⁷

2.4.2. Statistical analysis

MR uses genetic variants as instrumental variables to infer causal relationships between an exposure–outcome pair. Because genetic variants are randomly allocated at conception, they are free from confounding and serve as genetic instruments for causal inference. ¹⁵ There are three assumptions in an MR analysis: (1) the genetic variants are associated with the exposure, (2) there are no confounders between the genetic variant and outcome, and (3) all the effects of the genetic variant on the outcome are through the exposure. ¹⁵

Univariate MR was conducted to quantify the causal effect of mtDNAcn on AD and PD using a fixed‐effects inverse‐variance weighted (IVW) approach as the primary analysis method. This method provides a single causal estimate by weighting variant–exposure and variant–outcome associations based on the inverse of their variances, under the assumption that all instruments are valid and there is no horizontal pleiotropy. ³⁷

2.4.3. Diagnostics

We used the Cochran Q test to examine heterogeneity and MR Egger regression intercept to detect horizontal pleiotropy. Diagnostics were used to evaluate whether the causal estimates are robust to violations of the MR underlying assumptions. ¹⁵ Additionally, we used F statistics to test the strength of the instruments, whereby higher F statistics (> 10) indicate stronger instruments and reduce the likelihood of weak instrument bias. ¹⁵ Outliers were identified and excluded from the MR analysis using Radial MR version 1.0. ³⁸

2.4.4. Sensitivity analyses

Sensitivity analyses were used to evaluate the robustness of each exposure–outcome analysis under different assumptions, including random‐effects IVW, MR Egger, weighted median (WME), and weighted mode based estimator (WMBE). ³⁹ , ⁴⁰ , ⁴¹ We defined a robust causal effect as a significant IVW analysis (p < 0.05) after outlier removal and when there was no evidence of heterogeneity (p > 0.05) or pleiotropy (p > 0.05). When there was evidence of heterogeneity or pleiotropy, at least one of the sensitivity analyses also had to be significant (p < 0.05) and have the same direction of effect for a robust causal association. Finally, Steiger directionality tests were conducted across all exposure–outcome pairs to assess whether genetic instruments explained more variance in mtDNAcn than in AD, AD/dementia, and PD. ¹⁵

2.4.5. MVMR

MVMR evaluates the direct causal effect of an exposure on an outcome in the presence of other exposures. SNPs are first clumped (r ² = 0.001, 10Mb clumping window, EUR reference) and independent genome‐wide significant SNPs (p < 5e‐8) were extracted from each exposure dataset. ¹⁵ For any SNPs not present in either of the exposure datasets, LD proxies are identified and combined. Next, exposure SNPs were extracted from the outcome GWAS, and for SNPs not present in the outcome GWAS, proxy SNPs were identified. TwoSampleMR version 0.5.11 was used to harmonize the mtDNAcn and AD/PD datasets. Finally, MVMR was performed using the MVMR package version 0.3 and MendelianRandomization package version 0.9.0. ⁴² , ⁴³

2.4.6. LHC‐MR

Addressing exchangeability and horizontal pleiotropy is challenging in MR as many SNPs display pleiotropic effects, some of which may be mediated through confounders in the exposure–outcome relationship. LHC‐MR addresses these issues by introducing a latent heritable confounder. This unmeasured confounder is a latent factor whose effects are estimated from the pleiotropic effects of instrumental variables (IVs) in the exposure–outcome pairs being analyzed. ¹⁴ Including this LHC in the model adjusts for confounding, providing an unbiased beta estimate. Other benefits to LHC‐MR is that it (1) mitigates sample overlap effects by not having to remove overlapping segments which reduces statistical power, (2) enables us to estimate bidirectional effects of the outcome on the exposure, and (3) boosts statistical power by using the whole genome to estimate causal effects, rather than just genome‐wide significant SNPs. ¹⁴ To correct for multiple testing across the various exposure–outcome pairs analyzed, we applied FDR correction to the LHC‐MR p values.

The SNPs used as instrumental variables, along with their harmonized effect estimates, are provided in Tables S3–S14 in supporting information.

3. RESULTS

3.1. mtDNAcn is negatively correlated with AD and PD

We found strong positive genetic correlations between the four mtDNAcn measures and strong positive genetic correlations between both AD datasets (Figure 1). After FDR correction, the Longchamps mtDNAcn GWAS remained significantly negatively correlated with AD (rg = −0.082, FDR p = 1.25 × 10⁻²) and PD (rg = −0.078, FDR p = 1.25 × 10⁻²). The Gupta mtDNAcn GWAS showed a significant negative genetic correlation with AD (rg = −0.096, FDR p = 9.79 × 10⁻³). The Chong mtDNAcn GWAS demonstrated a significant negative correlation with AD/dementia (rg = −0.097, FDR p = 1.11 × 10⁻²) and AD/dementia and PD showed a significant negative genetic correlation (rg = −0.096, FDR p = 6.78 × 10⁻³).

Genetic correlations of mtDNAcn, AD, AD/dementia, and PD. The figure displays the lower triangular portion of the genetic correlation matrix. Box color indicates the direction and magnitude of the genetic correlation (red = positive, blue = negative). Box size reflects statistical significance, with larger tiles indicating significant associations (p ≤ 0.05). Asterisks denote significance thresholds: p ≤ 0.05 (^*), p ≤ 0.01 (^**), p ≤ 0.001 (^***). AD, Alzheimer's disease; GWAS, genome‐wide association study; mtDNAcn, mitochondrial DNA copy number; PD, Parkinson's disease.

3.2. mtDNAcn‐PRS is not associated with AD

In ADNI, all mtDNAcn PRS models were significantly associated with blood‐derived mtDNAcn, with higher genetically predicted mtDNAcn corresponding to higher measured blood‐derived mtDNAcn. The Hägg mtDNAcn PRS had the strongest association with mtDNAcn (Table S15 in supporting information).

None of the PRSs derived from the mtDNAcn GWASs showed a statistically significant association with AD risk in ADGC (Figure 2). However, the Longchamps mtDNAcn PRS showed a trend toward significance (odds ratio [OR; 95% confidence interval (CI)] = 0.97, [0.95–1.00], p = 0.059).

PRS analysis of mtDNAcn and AD. Forest plot shows the PRS odds ratio of four different mtDNAcn measures and AD. Black indicates non‐significant results. AD, Alzheimer's disease; CI, confidence interval; mtDNAcn, mitochondrial DNA copy number; PRS, polygenic risk score.

3.3. Univariable and MVMR showed no causal effect of mtDNAcn on AD or PD

Univariate MR results show no significant causal effect of mtDNAcn onto AD, AD/dementia, or PD across all four mtDNAcn measures (Figure S2 in supporting information). The number of IVs used for each mtDNAcn exposure–outcome pair in the univariable MR analyses ranged from 14 to 89 and is reported in Table S16 in supporting information. While instrument strength was strong in all analyses (F statistics > 10), significant heterogeneity was observed in nearly all exposure–outcome pairs, with the exception of Gupta et al. onto PD (Cochran Q value = 33.7, p = 0.33) and Hägg et al. onto AD (Cochran Q value = 62.2, p = 0.06; Table S16). No pleiotropy was detected in any analysis based on the MR‐Egger intercept (Table S16). When adjusting for platelets in Hägg using MVMR, we similarly observed no significant causal effect of mtDNAcn on AD or PD (Table S17 and Figure S3 in supporting information). Heterogeneity remained significant, but instruments were strong across all models (Table S17). Steiger directionality tests indicated that 11 of 12 exposure–outcome pairs had 100% of variants explaining more variance in mtDNAcn than in disease outcomes, with the remaining pair showing > 95% concordance. There was no evidence of reverse causation.

3.4. LHC‐MR showed a causal effect of mtDNAcn on AD/dementia and PD, with limited evidence to support a causal association in the reverse direction

LHC‐MR results showed that higher genetically predicted mtDNAcn from all four mtDNAcn datasets was significantly associated with lower risk of AD/dementia and PD (Figure 3 and Table S18 in supporting information) after FDR correction. In contrast, higher genetically predicted mtDNAcn from the Longchamps dataset was significantly associated with a higher risk of AD (OR [95% CI] = 1.21, [1.04–1.42], p = 0.01), while the other mtDNAcn datasets yielded non‐significant associations with AD. There was significant confounding in all LHC‐MR analyses except for Chong onto AD/dementia (p = 0.47) and PD (p = 0.72). Notably, the confounder effect estimates were positive across all AD/dementia and PD analyses, whereas the corresponding causal effect estimates were negative (Table S18).

LHC‐MR results for mtDNAcn onto AD and PD. Forest plots show the MR odds ratio for four different mtDNAcn measures, AD, AD/dementia, and PD. p values represent false discovery rate–adjusted significance. Red indicates significant results. AD, Alzheimer's disease; CI, confidence interval; MR, Mendelian randomization; mtDNAcn, mitochondrial DNA copy number; PD, Parkinson's disease.

In the reverse direction, genetically liability for AD was causally associated with increased mtDNAcn in Longchamps (β [95% CI] = 0.17, [0.08, 0.26], p = 0.002) and Gupta (β [95% CI] = 0.19, [0.07–0.31], p = 0.01; Table S19 and Figure S4 in supporting information). Furthermore, there was significant confounding for AD onto Longchamps (p = 2.91e‐25).

4. DISCUSSION

This study evaluated the association between genetically predicted mtDNAcn and AD/PD risk across four mtDNAcn GWAS datasets (Longchamps, Hägg, Gupta, Chong). While most analyses yielded consistent results across these datasets, minor discrepancies likely reflect differences in study design, including sample size, platelet contamination, and the method used to measure mtDNAcn. Despite these variations, our findings suggest that different mtDNAcn measurement techniques capture a shared underlying biology. To assess the robustness of these associations, we applied multiple MR methods to evaluate the causal relationships between mtDNAcn measures and both neurodegenerative diseases. The key finding was that higher blood‐based mtDNAcn was causally associated with reduced risk of AD/dementia and PD.

Genetic correlations showed that the Longchamps mtDNAcn GWAS was negatively correlated with AD, AD/dementia, and PD; Gupta mtDNAcn GWAS with AD and AD/dementia; and Chong mtDNAcn GWAS with AD/dementia. Although none of the PRS associations of mtDNAcn onto AD reached statistical significance, the PRS based on the Longchamps mtDNAcn GWAS showed a trend toward significance. These findings highlight small dataset‐specific differences in the strength of association between mtDNAcn genetic risk and AD, which may be explained by how mtDNAcn was quantified. Longchamps was the largest of the four studies, identifying 97 associated loci and estimating mtDNAcn using three different methods: qPCR (1.8% of samples), WGS (0.3%), and microarray (97.9%). ¹⁹ Although previous research has shown that mtDNAcn measured by WGS has a stronger correlation with known mtDNA correlates (age, biological sex, white blood cell count, Duffy locus genotype, incident cardiovascular disease), combining all three approaches likely provides the most accurate estimate. ³ Thus, the results from the Longchamps GWAS are likely the most reliable. On the other hand, Hägg et al. did not adjust for platelet counts in their analysis and therefore can be considered the least reliable. ²⁰ Although adjusting for platelets did not change the significance of the MVMR results in our study, prior research has shown that approximately one third of the genetic associations identified by Hägg et al. lose statistical significance when platelet counts are included as a covariate. ²¹

The univariate MR results did not support a causal association between mtDNAcn and AD or PD; however, there was significant heterogeneity in most exposure–outcome pairs, indicating significant violations of the underlying assumptions of MR. In particular, sample overlap between the AD/dementia GWAS and mtDNAcn GWASs—both of which include individuals from UK Biobank—may introduce bias due to the winner's curse and increased risk of type I errors. ⁴⁴

LHC‐MR accounts for sample overlap and incorporates a latent, unmeasured confounder—an environmental or genetic factor that influences both the exposure and outcome—and estimates both its presence and effect. ¹⁴ Such factors in practice could reflect shared genetic influences on mitochondrial biogenesis/turnover, immune and inflammatory function, blood cell–type composition, or metabolic mechanisms. ⁴⁵ , ⁴⁶ LHC‐MR results showed that genetically higher mtDNAcn is causally associated with a lower risk of AD/dementia. Additionally, LHC‐MR results suggest that mtDNAcn is associated with a lower risk of PD. These results are consistent with prior studies that have found that damaged mtDNA and mtDNA mutations can lead to PD‐like pathology. ⁴⁷ , ⁴⁸ Comparing confounder estimates alongside causal estimates helps reveal how confounding may bias causal estimates in univariable MR. In the LHC‐MR analysis, heritable confounders showed positive effects, while the causal estimates for mtDNAcn on AD/dementia and PD were negative. These opposing directions may explain why standard MR methods were non‐significant.

Changes in mtDNAcn may influence AD/dementia and PD risk through multiple, overlapping mechanisms. In PD, dopaminergic neurons of the substantia nigra are particularly vulnerable due to their high energetic demands and dependence on mitochondrial respiration, with studies linking complex I dysfunction, mtDNA damage, and impaired mitochondrial quality control to neuronal degeneration. ⁴⁹ Similarly, in AD, deficiencies across multiple respiratory chain complexes, disrupted mitochondrial dynamics, impaired mitophagy, and increased oxidative stress have been observed in both sporadic and familial AD. ⁴⁹ Because mtDNA encodes essential subunits of the oxidative phosphorylation system, alterations in mtDNAcn may influence mitochondrial respiratory capacity, reactive oxygen species production, and cellular energy homeostasis. ⁴⁹ Experimental models of mtDNA depletion further demonstrate that mitochondrial dysfunction can induce widespread transcriptional and signaling changes affecting amyloid processing, synaptic function, immune responses, and proteostasis pathways implicated in neurodegeneration. ⁵⁰ Changes in mtDNAcn may therefore contribute to bioenergetic deficits, linking mitochondrial dysfunction to downstream neurodegenerative processes.

Higher mtDNAcn in the Longchamps dataset was found to be causally associated with increased AD risk but a decreased risk for AD/dementia. These opposing findings may be explained by sample composition, differences in the outcome definition, and diagnostic heterogeneity in both AD/dementia and AD GWASs. The AD/dementia GWAS has a larger effective sample size and encompasses multiple dementia subtypes, each with distinct underlying risk factors and biological mechanisms. ² , ⁵¹ Furthermore, AD/dementia case definitions may be influenced by survival and reporting bias, as many cases are based on self‐reported parental diagnoses. ⁵² However, for the AD GWAS, there may be misclassification in clinical AD, with one study finding that 30% of controls were found to be in preclinical stages of AD even at typical cohort ages. ⁵³ Such misclassification can inflate reverse associations, potentially contributing to the reverse causality in our results. Finally, while sample overlap could influence causal estimates, we applied LHC‐MR, which is robust to sample overlap, making this an unlikely explanation for the GWAS‐specific differences observed.

When looking at the bidirectional effects, higher genetic risk of AD was associated with increased mtDNAcn in both the Longchamps and Gupta datasets. This may reflect compensatory upregulation of mtDNAcn in response to mitochondrial impairment, in which cells increase mtDNAcn through stress‐induced mitochondrial biogenesis and/or mtDNA replication to offset mitochondrial dysfunction, oxidative stress, or accumulated damage. ² In both AD and PD, processes including impaired electron transport chain activity, increased reactive oxygen species production, and disrupted mitochondrial dynamics lead to a decline in mitochondrial efficiency. ² In response, cells may initially increase mitochondrial biogenesis and mtDNAcn to maintain energy production and homeostasis. ⁵⁴ However, as mitochondrial damage accumulates, increased mtDNAcn might not be able to compensate, leading to neurodegeneration. ⁵⁴

There are several limitations in our analyses. First, all GWAS datasets analyzed were predominantly of European ancestry, limiting the generalizability of our findings to other populations. Second, although the Longchamps GWAS included three methods for measuring mtDNAcn, > 98% of samples were measured using microarray; therefore, it is difficult to draw conclusions about the specific impact of WGS or qPCR. Third, while independent validation is important, currently available external cohorts such as FinnGen have limited power for AD, PD, and dementia outcomes due to smaller effective sample sizes and reduced instrument strength. ⁵⁵ Large future GWAS releases may allow for such validation. ⁵⁶ Fourth, LHC‐MR assumes linear relationships between the exposure and outcome. While non‐linear effects between mtDNAcn and AD or PD risk are biologically plausible, future studies using individual‐level data will be required to investigate potential non‐linear effects. Last, while our PRS is based on blood‐derived mtDNAcn, it may not fully capture mtDNAcn in brain tissue where AD manifests, and baseline blood‐based mtDNAcn levels may not reflect the brain's adaptive capacity to compensate for age‐related mitochondrial decline.

Strengths of our study include the use of multiple mtDNAcn and AD GWAS that were conducted using different study designs. Different mtDNAcn GWASs rely on distinct participant samples and analytical approaches, while the AD GWASs rely on different definitions of cases versus controls based on underlying AD pathology. These differences can affect the estimated associations between SNPs and the trait and introduce variability that can impact MR causal effect estimates. ⁵⁷ Another strength is a bidirectional MR approach to assess the directionality of effects and the application of both PRS and MR to distinguish association from causation.

In conclusion, our study demonstrates a causal effect of mtDNAcn on AD and PD using multiple complementary analytical approaches and different mtDNAcn measures. Additional research is needed to extend these findings to other AD endophenotypes to better understand the biological pathways underlying these findings. Nevertheless, these findings highlight mtDNAcn as an important biomarker of mitochondrial dysfunction in the early stages of AD and PD.

CONFLICT OF INTEREST STATEMENT

Aadrita Chatterjee, Shea J. Andrews, Russell H. Swerdlow, and Heather M. Wilkings declare no conflicts of interest. Caroline Jonson's participation in this project was part of a competitive contract awarded to DataTecnica LLC by the National Institutes of Health to support open science research. Alison M. Goate serves on the SRB for Genentech and Muna Therapeutics. She has consulted for Merck and Arbor.bio. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All participants provided informed consent.

Supporting information

Supporting Information

ALZ-22-e71469-s004.docx^{(583.1KB, docx)}

Supporting Information

ALZ-22-e71469-s005.xlsx^{(180.7KB, xlsx)}

Supporting Information

ALZ-22-e71469-s001.docx^{(35.6KB, docx)}

Supporting Information

ALZ-22-e71469-s002.docx^{(42.6KB, docx)}

Supporting Information

ALZ-22-e71469-s003.pdf^{(2MB, pdf)}

ACKNOWLEDGMENTS

ADNI

Data used in preparation of this article were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.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.usc.edu/wp‐content/uploads/how_to_apply/ADNI_Acknowledgement_List.pdf.

Data collection and sharing for this project was funded by the Alzheimer's Disease Neuroimaging Initiative (ADNI; National Institutes of Health Grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH‐12‐2‐0012). ADNI is funded by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: AbbVie; Alzheimer's Association; Alzheimer's Drug Discovery Foundation; Araclon Biotech; BioClinica, Inc.; Biogen; Bristol‐Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann‐La Roche Ltd and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & Development, LLC; Johnson & Johnson Pharmaceutical Research & Development LLC; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, LLC; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics. The Canadian Institutes of Health Research is providing funds to support ADNI clinical sites in Canada. Private‐sector contributions are facilitated by the Foundation for the National Institutes of Health (www.fnih.org). The grantee organization is the Northern California Institute for Research and Education, and the study is coordinated by the Alzheimer's Therapeutic Research Institute at the University of Southern California. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of Southern California.

ADGC

ADGC, The National Institutes of Health, National Institute on Aging (NIH‐NIA) supported this work through the following grants: ADGC, U01 AG032984, RC2 AG036528. Samples from the National Cell Repository for Alzheimer's Disease (NCRAD), which receives government support under a cooperative agreement grant (U24 AG21886) awarded by the National Institute on Aging (NIA), were used in this study. We thank contributors who collected samples used in this study, as well as patients and their families, whose help and participation made this work possible. Data for this study were prepared, archived, and distributed by the National Institute on Aging Alzheimer's Disease Data Storage Site (NIAGADS) at the University of Pennsylvania (U24‐AG041689); GCAD, U54 AG052427.

ACT

Adult Changes in Thought—Kaiser Permanente Washington, U19AG06656.

ADC

Alzheimer's Disease Centers/NACC National Alzheimer's Coordinating Center—The NACC database is funded by NIA/NIH Grant U24 AG072122. NACC data are contributed by the NIA‐funded ADRCs: P30 AG062429 (PI James Brewer, MD, PhD), P30 AG066468 (PI Oscar Lopez, MD), P30 AG062421 (PI Bradley Hyman, MD, PhD), P30 AG066509 (PI Thomas Grabowski, MD), P30 AG066514 (PI Mary Sano, PhD), P30 AG066530 (PI Helena Chui, MD), P30 AG066507 (PI Marilyn Albert, PhD), P30 AG066444 (PI David Holtzman, MD), P30 AG066518 (PI Lisa Silbert, MD, MCR), P30 AG066512 (PI Thomas Wisniewski, MD), P30 AG066462 (PI Scott Small, MD), P30 AG072979 (PI David Wolk, MD), P30 AG072972 (PI Charles DeCarli, MD), P30 AG072976 (PI Andrew Saykin, PsyD), P30 AG072975 (PI Julie A. Schneider, MD, MS), P30 AG072978 (PI Ann McKee, MD), P30 AG072977 (PI Robert Vassar, PhD), P30 AG066519 (PI Frank LaFerla, PhD), P30 AG062677 (PI Ronald Petersen, MD, PhD), P30 AG079280 (PI Jessica Langbaum, PhD), P30 AG062422 (PI Gil Rabinovici, MD), P30 AG066511 (PI Allan Levey, MD, PhD), P30 AG072946 (PI Linda Van Eldik, PhD), P30 AG062715 (PI Sanjay Asthana, MD, FRCP), P30 AG072973 (PI Russell Swerdlow, MD), P30 AG066506 (PI Glenn Smith, PhD, ABPP), P30 AG066508 (PI Stephen Strittmatter, MD, PhD), P30 AG066515 (PI Victor Henderson, MD, MS), P30 AG072947 (PI Suzanne Craft, PhD), P30 AG072931 (PI Henry Paulson, MD, PhD), P30 AG066546 (PI Sudha Seshadri, MD), P30 AG086401 (PI Erik Roberson, MD, PhD), P30 AG086404 (PI Gary Rosenberg, MD), P20 AG068082 (PI Angela Jefferson, PhD), P30 AG072958 (PI Heather Whitson, MD), P30 AG072959 (PI James Leverenz, MD).

ADNI

Alzheimer's Disease Neuroimaging Initiative—Data collection and sharing for the Alzheimer's Disease Neuroimaging Initiative (ADNI) is funded by the National Institute on Aging (National Institutes of Health Grant U19 AG024904). The grantee organization is the Northern California Institute for Research and Education. In the past, ADNI has also received funding from the National Institute of Biomedical Imaging and Bioengineering, the Canadian Institutes of Health Research, and private sector contributions through the Foundation for the National Institutes of Health (FNIH) including generous contributions from the following: AbbVie; Alzheimer's Association; Alzheimer's Drug Discovery Foundation; Araclon Biotech; BioClinica, Inc.; Biogen; Bristol‐Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann‐La Roche Ltd and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & Development, LLC; Johnson & Johnson Pharmaceutical Research &Development LLC; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, LLC; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics.

BIOCARD

Biomarkers of Cognitive Decline Among Normal Individuals—Johns Hopkins University, U19AG033655.

CHAP

Chicago Health and Aging Project—Rush University, R01 AG11101.

EAS

Einstein Aging Study—Einstein College of Medicine, P01 AG03949.

EFIGA

Estudio Familiar de Influencia Genetica en Alzheimer—Columbia University, RF1AG015473.

GenerAAtions

Genetic & Environmental Risk Factors for AD Among African Americans—Johns Hopkins BIOCARD R01AG020688.

The GSK dataset is from the Multi‐Site Collaborative Study for Genotype‐Phenotype Associations in Alzheimer's disease (Genetic Alzheimer's Disease Associations – GenADA), funded by GlaxoSmithKline.

Indianapolis

Indianapolis African Americans/Ibadan Study of Aging—Indianapolis‐Ibadan Dementia Project, R01AG009956.

We are very grateful to the members of Japanese Genetic Study Consortium of Alzheimer's Disease (JGSCAD) for the collection of blood samples. The members were listed in previous JGSCAD publications [Kuwano R, 2006; Miyashita A,2007].

Japanese Alzheimer's Disease Neuroimaging Initiative (J‐ADNI) was supported by the following funding sources: the Translational Research Promotion Project from the New Energy and Industrial Technology Development Organization of Japan; Research on Dementia, Health Labor Sciences Research Grant; the Life Science Database Integration Project of Japan Science and Technology Agency; the Research Association of Biotechnology (Astellas Pharma Inc., Bristol‐Myers Squibb, Daiichi‐Sankyo, Eisai, Eli Lilly and Company, Merck‐Banyu, Mitsubishi Tanabe Pharma, Pfizer Inc., Shionogi & Co., Ltd., Sumitomo Dainippon, and Takeda Pharmaceutical Company), Japan; and a grant from an anonymous foundation. The investigators within J‐ADNI contributed to the design and implementation of J‐ADNI and/or provided data but did not participate in the analysis or the writing of this report. A complete listing of J‐ADNI investigators can be found at https://humandbs.biosciencedbc.jp/en/hum0043‐j‐adni‐authors.

MAYO

Mayo Clinic – Mayo Clinic, R01 AG032990, U01 AG046139, R01 NS080820, RF1 AG051504, P50 AG016574.

MIRAGE

Multi‐Institutional Research in Alzheimer's Genetic Epidemiology—Boston University, R01‐AG048927, U01‐AG082655.

NIA‐LOAD/NCRAD

NIA‐LOAD/National Centralized Repository for Alzheimer's Disease/National Centralized Repository for Alzheimer's Disease (NCRAD)—Columbia University, U24AG056270, P30AG072976.

OHSU

Oregon Health and Science University—Oregon Health and Science University, P30 AG008017, R01 AG026916.

PRADI

Puerto Rican Alzheimer's Disease Initiative – University of Miami, RF1AG054074.

REAAADI

Research in African‐American Alzheimer's Disease Initiative—Wakeforest University & University of Miami, AG052410.

RMAYO

Rochester Mayo Clinic—Mayo Clinic, P50 AG016574; P30 AG062677.

ROS/MAP

Religious Orders Study/Memory and Aging Project – Rush University, P30AG10161 (ROS), R01AG15819 / R01AG17917 (MAP).

TARCC

Texas Alzheimer's Research and Care Consortium—The Texas Alzheimer's Research and Care Consortium (TARCC) is funded by the state of Texas through the Texas Council on Alzheimer's Disease and Related Disorders.

TGEN

Translational Genomics Research Institute—TGEN—AG041232, Genotyping of the TGEN2 cohort was supported by Kronos Science. The TGen series was also funded by NIA grant AG041232 to AJM and MJH, The Banner Alzheimer's Foundation.

UM/CWRU/MSSM: Subset A‐C and MTV/MTC

University of Miami/Case Western Reserve University/Mount Sinai School of Medicine—University of Miami, Case Western Reserve University, Mount Sinai School of Medicine, Texas Alzheimer's Research and Care Consortium, R01 AG027944, R01 AG028786, R01 AG019085.

UPITT

University of Pittsburgh, P30 AG066468, R01 AG030653, AG064877.

WASHU

Washington University—Washington University, P50 AG05681, P01 AG03991, P01 AG026276.

WHICAP

Washington Heights/Inwood Columbia Aging Project – Columbia University, RF1AG054023.

Aadrita Chatterjee is supported by R35AG071916. Shea J. Andrews is supported by the Alzheimer's Association (AARF‐20‐675804). Caroline Jonson is supported (in part) by the Intramural Research Program of the National Institutes of Health (NIH), project number ZO1 AG000534, as well as the National Institute of Neurological Disorders and Stroke (NINDS). The content of this publication is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. This research was supported (in part) by the Intramural Research Program of the National Institutes of Health (NIH). The contributions of the NIH author(s) are considered Works of the United States Government. The findings and conclusions presented in this paper are those of the author(s) and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services. Heather M. Wilkins is supported by the Margaret “Peg” McLaughlin and Lydia A. Walker Opportunity Fund, the University of Kansas Alzheimer's Disease Center P30AG072973, R01AG078186, and by the Alzheimer's Association 23AARG‐1023294. Russell H. Swerdlow and Heather M. Wilkins are supported by AG072973.

Chatterjee A, Alvarez B, Sharma RU, et al. Evaluating the causal effect of mitochondrial dysfunction on Alzheimer's disease and Parkinson's disease using polygenic risk scores and Mendelian randomization. Alzheimer's Dement. 2026;22:e71469. 10.1002/alz.71469

Aadrita Chatterjee and Brian Alvarezs are co‐first authors.

For the Alzheimer's Disease Genetics Consortium and the Alzheimer's Disease Neuroimaging Initiative, see ACKNOWLEDGMENTS section.

DATA AVAILABILITY STATEMENT

All statistical analyses were conducted using R (v4.3.0). The analysis code is publicly available at: https://github.com/AndrewsLabUCSF/Brian‐Estimating‐Causality‐mtDNAcn‐on‐Alzheimer/tree/aadrita_mtdnacn. The original summary statistics are available at the following websites: https://dss.niagads.org/datasets/ng00075/, https://www.ebi.ac.uk/gwas/publications/35379992, https://pmc.ncbi.nlm.nih.gov/articles/PMC8422160/#S15, https://www.ebi.ac.uk/gwas/publications/27863252, https://www.ebi.ac.uk/gwas/publications/35023831, https://link.springer.com/article/10.1007/s00439‐020‐02249‐w#data‐availability, https://pmc.ncbi.nlm.nih.gov/articles/PMC8758627/#notes2, https://www.ebi.ac.uk/gwas/publications/37587338

REFERENCES

1. Castellani CA, Longchamps RJ, Sun J, Guallar E, Arking DE. Thinking outside the nucleus: mitochondrial DNA copy number in health and disease. Mitochondrion. 2020;53:214‐223. doi: 10.1016/j.mito.2020.06.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
2. Picard M. Blood mitochondrial DNA copy number: what are we counting? Mitochondrion. 2021;60:1‐11. doi: 10.1016/j.mito.2021.06.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
3. Longchamps RJ, Castellani CA, Yang SY, et al. Evaluation of mitochondrial DNA copy number estimation techniques. PLOS ONE. 2020;15(1):e0228166. doi: 10.1371/journal.pone.0228166 [DOI] [PMC free article] [PubMed] [Google Scholar]
4. Baloyannis SJ. Mitochondrial alterations in Alzheimer's disease. J Alzheimer's Dis. 2006;9(2):119‐126. doi: 10.3233/jad-2006-9204 [DOI] [PubMed] [Google Scholar]
5. Wang L, Han J, Fearnley LG, et al. Peripheral immune cell abundance differences link blood mitochondrial DNA copy number and Parkinson's disease. NPJ Parkinson's Dis. 2024;10:219. doi: 10.1038/s41531-024-00831-x [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Harerimana NV, Paliwali D, Romero‐Molina C, et al. The role of mitochondrial genome abundance in Alzheimer's disease. Alzheimers Dement. 2023;19(5):2069‐2083. doi: 10.1002/alz.12812 [DOI] [PMC free article] [PubMed] [Google Scholar]
7. Qin P, Chen X, Ma P, et al. Mitochondrial DNA copy number and Alzheimer's disease and Parkinson disease. Mitochondrion. 2025;83:102032. doi: 10.1016/j.mito.2025.102032 [DOI] [PubMed] [Google Scholar]
8. Risi B, Imarisio A, Cuconato G, Padovani A, Valente EM, Filosto M. Mitochondrial DNA (mtDNA) as fluid biomarker in neurodegenerative disorders: a systematic review. Eur J Neurol. 2025;32(1):e70014. doi: 10.1111/ene.70014 [DOI] [PMC free article] [PubMed] [Google Scholar]
9. Lei S, Liu Y. Identifying blood mitochondrial DNA copy number as a biomarker for development of neurodegenerative diseases: evidence from Mendelian randomization analysis. Neuroscience. 2025;573:421‐429. doi: 10.1016/j.neuroscience.2025.04.003 [DOI] [PubMed] [Google Scholar]
10. Xia H, Wang Zi‐H, Wang X‐B, et al. Investigating the genetic association of mitochondrial DNA copy number with neurodegenerative diseases. BMC Neurol. 2025;25:160. doi: 10.1186/s12883-025-04176-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
11. van Rheenen W, Peyrot WJ, Schork AJ, Lee SH, Wray NR. Genetic correlations of polygenic disease traits: from theory to practice. Nat Rev Genet. 2019;20(10):10. doi: 10.1038/s41576-019-0137-z [DOI] [PubMed] [Google Scholar]
12. Choi SW, Mak TSH, O'Reilly PF. Tutorial: a guide to performing polygenic risk score analyses. Nat Protoc. 2020;15(9):2759‐2772. doi: 10.1038/s41596-020-0353-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
13. Davies NM, Holmes MV, Davey Smith G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ. 2018;362:k601. doi: 10.1136/bmj.k601 [DOI] [PMC free article] [PubMed] [Google Scholar]
14. Darrous L, Mounier N, Kutalik Z. Simultaneous estimation of bi‐directional causal effects and heritable confounding from GWAS summary statistics. Nat Commun. 2021;12(1):7274. doi: 10.1038/s41467-021-26970-w [DOI] [PMC free article] [PubMed] [Google Scholar]
15. Sanderson E, Glymour MM, Holmes MV, et al. Mendelian randomization. Nat Rev Methods Primers. 2022;2(1):1. doi: 10.1038/s43586-021-00092-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
16. Kunkle BW, Grenier‐Boley B, Sims R, et al. Genetic meta‐analysis of diagnosed Alzheimer's disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat Genet. 2019;51(3):3. doi: 10.1038/s41588-019-0358-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
17. Bellenguez Cé, Küçükali F, Jansen IE, et al. New insights into the genetic etiology of Alzheimer's disease and related dementias. Nat Genet. 2022;54(4):412‐436. doi: 10.1038/s41588-022-01024-z [DOI] [PMC free article] [PubMed] [Google Scholar]
18. Nalls MA, Blauwendraat C, Vallerga CL, et al. Identification of novel risk loci, causal insights, and heritable risk for Parkinson's disease: a meta‐genome wide association study. Lancet Neurol. 2019;18(12):1091‐1102. doi: 10.1016/S1474-4422(19)30320-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Longchamps RJ, Yang SY, Castellani CA, et al. Genome‐wide analysis of mitochondrial DNA copy number reveals loci implicated in nucleotide metabolism, platelet activation, and megakaryocyte proliferation. Hum Genet. 2022;141(1):127‐146. doi: 10.1007/s00439-021-02394-w [DOI] [PMC free article] [PubMed] [Google Scholar]
20. Hägg S, Jylhävä J, Wang Y, Czene K, Grassmann F. Deciphering the genetic and epidemiological landscape of mitochondrial DNA abundance. Hum Genet. 2021;140(6):849‐861. doi: 10.1007/s00439-020-02249-w [DOI] [PMC free article] [PubMed] [Google Scholar]
21. Chong M, Mohammadi‐Shemirani P, Perrot N, et al. GWAS and ExWAS of blood mitochondrial DNA copy number identifies 71 loci and highlights a potential causal role in dementia. ELife. 2022;11:e70382. doi: 10.7554/eLife.70382 [DOI] [PMC free article] [PubMed] [Google Scholar]
22. Gupta R, Kanai M, Durham TJ, et al. Nuclear genetic control of mtDNA copy number and heteroplasmy in humans. Nature. 2023;620(7975):839‐848. doi: 10.1038/s41586-023-06426-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
23. Astle WJ, Elding H, Jiang T, et al. The allelic landscape of human blood cell trait variation and links to common complex disease. Cell. 2016;167(5):1415‐1429.e19. doi: 10.1016/j.cell.2016.10.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
24. Lyon MS, Andrews SJ, Elsworth B, Gaunt TR, Hemani G, Marcora E. The variant call format provides efficient and robust storage of GWAS summary statistics. Genome Biol. 2021;22(1):32. doi: 10.1186/s13059-020-02248-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
25. Bulik‐Sullivan BK, Loh Po‐Ru, Finucane HK, et al. LD Score Regression Distinguishes Confounding from Polygenicity in Genome‐Wide Association Studies. Nat Genet. 2015;47(3):291‐295. doi: 10.1038/ng.3211 [DOI] [PMC free article] [PubMed] [Google Scholar]
26. Lu Q, Li B, Ou D, et al. A Powerful Approach to Estimating Annotation‐Stratified Genetic Covariance via GWAS Summary Statistics: Am J Hum Genet. Accessed March 14, 2025;101(6):939‐964. https://www.cell.com/ajhg/fulltext/S0002‐9297(17)30453‐6 [DOI] [PMC free article] [PubMed] [Google Scholar]
27. Rajabli F, Benchek P, Tosto G, et al. Multi‐ancestry genome‐wide meta‐analysis of 56,241 individuals identifies known and novel cross‐population and ancestry‐specific associations as novel risk loci for Alzheimer's disease. Genome Biol. 2025;26(1):210. doi: 10.1186/s13059-025-03564-z [DOI] [PMC free article] [PubMed] [Google Scholar]
28. Vitale D, Koretsky MJ, Kuznetsov N, et al, GenoTools: an open‐source Python package for efficient genotype data quality control and analysis G3: Genes, Genomes, Genet. 2024;15:jkae268. doi: 10.1093/g3journal/jkae268 [DOI] [PMC free article] [PubMed] [Google Scholar]
29. Ge T, Chen CY, Ni Y, Feng YCA, Smoller JW. Polygenic prediction via Bayesian regression and continuous shrinkage priors. Nat Commun. 2019;10(1):1776. doi: 10.1038/s41467-019-09718-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
30. Lambert SA, Wingfield B, Gibson JT, et al. Enhancing the Polygenic Score Catalog with tools for score calculation and ancestry normalization. Nat Genet. 2024;56(10):1989‐1994. doi: 10.1038/s41588-024-01937-x [DOI] [PMC free article] [PubMed] [Google Scholar]
31. Petersen RC, Aisen PS, Beckett LA, et al. Alzheimer's Disease Neuroimaging Initiative (ADNI). Neurology. 2010;74(3):201‐209. doi: 10.1212/WNL.0b013e3181cb3e25 [DOI] [PMC free article] [PubMed] [Google Scholar]
32. Tong T, Zhu C, Farrell JJ, et al. Blood‐derived mitochondrial DNA copy number is associated with Alzheimer disease, Alzheimer‐related biomarkers and serum metabolites. Alzheimers Res Ther. 2024;16(1):234. doi: 10.1186/s13195-024-01601-w [DOI] [PMC free article] [PubMed] [Google Scholar]
33. Myers TA, Chanock SJ, Machiela MJ. LDlinkR: an R Package for Rapidly Calculating Linkage Disequilibrium Statistics in Diverse Populations. Front Genetics. 2020;11: 157. doi: 10.3389/fgene.2020.00157 [DOI] [PMC free article] [PubMed] [Google Scholar]
34. Hemani G, Zheng J, Elsworth B, et al. The MR‐Base platform supports systematic causal inference across the human phenome. ELife. 2018;7:e34408. doi: 10.7554/eLife.34408 [DOI] [PMC free article] [PubMed] [Google Scholar]
35. Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLOS Genet. 2017;13(11):e1007081. doi: 10.1371/journal.pgen.1007081 [DOI] [PMC free article] [PubMed] [Google Scholar]
36. Mandl MM, Boulesteix AL, Burgess S, Zuber V. Outlier Detection in Mendelian Randomization. Stat Med. 2025;44(15‐17):e70143. doi: 10.1002/sim.70143 [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Hemani G, Bowden J, Davey Smith G. Evaluating the potential role of pleiotropy in Mendelian randomization studies. Hum Mol Genet. 2018;27(R2):R195‐R208. doi: 10.1093/hmg/ddy163 [DOI] [PMC free article] [PubMed] [Google Scholar]
38. Bowden J, Spiller W, Del Greco M F, et al. Improving the visualization, interpretation and analysis of two‐sample summary data Mendelian randomization via the Radial plot and Radial regression. Int J Epidemiol. 2018;47(4):1264‐1278. doi: 10.1093/ije/dyy101 [DOI] [PMC free article] [PubMed] [Google Scholar]
39. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Intl J Epidemiol. 2015;44(2):512‐525. doi: 10.1093/ije/dyv080 [DOI] [PMC free article] [PubMed] [Google Scholar]
40. Bowden J, Davey Smith G, Haycock PC, Burgess S, Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol. 2016;40(4):304‐314. doi: 10.1002/gepi.21965 [DOI] [PMC free article] [PubMed] [Google Scholar]
41. Hartwig FP, Davey Smith G, Bowden J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol. 2017;46(6):1985‐1998. doi: 10.1093/ije/dyx102 [DOI] [PMC free article] [PubMed] [Google Scholar]
42. Sanderson E, Spiller W, Bowden J. Testing and correcting for weak and pleiotropic instruments in two‐sample multivariable Mendelian randomization. Stat Med. 2021;40(25):5434‐5452. doi: 10.1002/sim.9133 [DOI] [PMC free article] [PubMed] [Google Scholar]
43. Patel A, Ye T, Xue H. MendelianRandomization v0.9.0: updates to an R package for performing Mendelian randomization analyses using summarized data. Wellcome Open Res. 2023;8:449. doi: 10.12688/wellcomeopenres.19995.2 [DOI] [PMC free article] [PubMed] [Google Scholar]
44. Burgess S, Davies NM, Thompson SG. Bias due to participant overlap in two‐sample Mendelian randomization. Genet Epidemiol. 2016;40(7):597‐608. doi: 10.1002/gepi.21998 [DOI] [PMC free article] [PubMed] [Google Scholar]
45. Parchwani D, Singh R, Patel D. Biological and translational attributes of mitochondrial DNA copy number: laboratory perspective to clinical relevance. World J Methodol. 2025;15(3):102709. doi: 10.5662/wjm.v15.i3.102709 [DOI] [PMC free article] [PubMed] [Google Scholar]
46. Hu MM, Shu HB. Mitochondrial DNA‐triggered innate immune response: mechanisms and diseases. Cell Mol Immunol. 2023;20(12):1403‐1412. doi: 10.1038/s41423-023-01086-x [DOI] [PMC free article] [PubMed] [Google Scholar]
47. Tresse E, Marturia‐Navarro J, Sew WQiG, et al. Mitochondrial DNA damage triggers spread of Parkinson's disease‐like pathology. Mol Psychiatry. 2023;28(11):4902‐4914. doi: 10.1038/s41380-023-02251-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
48. Buneeva O, Fedchenko V, Kopylov A, Medvedev A. Mitochondrial Dysfunction in Parkinson's Disease: focus on Mitochondrial DNA. Biomedicines. 2020;8(12):591. doi: 10.3390/biomedicines8120591 [DOI] [PMC free article] [PubMed] [Google Scholar]
49. Macdonald R, Barnes K, Hastings C, Mortiboys H. Mitochondrial abnormalities in Parkinson's disease and Alzheimer's disease: can mitochondria be targeted therapeutically?. Biochem Soc Trans. 2018;46(4):891‐909. doi: 10.1042/BST20170501 [DOI] [PubMed] [Google Scholar]
50. Menta BW, Schueddig E, Ranjan A, et al. MtDNA‐depleted neuronal cell transcriptomes reveal Alzheimer's disease‐related changes. Alzheimers Dementia. 2025;21(11):e70929. doi: 10.1002/alz.70929 [DOI] [PMC free article] [PubMed] [Google Scholar]
51. Anderson EL, Davies NM, Korologou‐Linden R, Kivimäki M. Dementia prevention: the Mendelian randomisation perspective. J Neurol Neurosurg Psychiatry. 2024;95(4):384‐390. doi: 10.1136/jnnp-2023-332293 [DOI] [PubMed] [Google Scholar]
52. Wu Y, Sun Z, Zheng Q, et al. Pervasive biases in proxy GWAS based on parental history of Alzheimer's disease. bioRxiv. Preprint posted online October 17, 2023:2023.10.13.562272. doi: 10.1101/2023.10.13.562272 [DOI] [Google Scholar]
53. Escott‐Price V, Baker E, Shoai M, et al. Genetic analysis suggests high misassignment rates in clinical Alzheimer's cases and controls. Neurobiol Aging. 2019;77:178‐182. doi: 10.1016/j.neurobiolaging.2018.12.002 [DOI] [PubMed] [Google Scholar]
54. Swerdlow RH, Khan SM. The Alzheimer's Disease Mitochondrial Cascade Hypothesis: an Update. Exp Neurol. 2009;218(2):308‐315. doi: 10.1016/j.expneurol.2009.01.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
55. Kurki MI, Karjalainen J, Palta P, et al. FinnGen provides genetic insights from a well‐phenotyped isolated population. Nature. 2023;613(7944):508‐518. doi: 10.1038/s41586-022-05473-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
56. Uffelmann E, Wightman DP, Bahrami S, et al. Genomic analyses reveal new insights into Alzheimer's disease. medRxiv. Published online January 1, 2025:2025.10.10.25337470. doi: 10.1101/2025.10.10.25337470 [DOI] [Google Scholar]
57. Nguyen TN, Terry L, Guggenheim JA, Mendelian Randomization studies of Myopia: choosing the right summary statistics. medRxiv. Preprint posted online June 5, 2025:2025.06.04.25328975. doi: 10.1101/2025.06.04.25328975 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supporting Information

ALZ-22-e71469-s004.docx^{(583.1KB, docx)}

Supporting Information

ALZ-22-e71469-s005.xlsx^{(180.7KB, xlsx)}

Supporting Information

ALZ-22-e71469-s001.docx^{(35.6KB, docx)}

Supporting Information

ALZ-22-e71469-s002.docx^{(42.6KB, docx)}

Supporting Information

ALZ-22-e71469-s003.pdf^{(2MB, pdf)}

Data Availability Statement

[alz71469-bib-0001] 1. Castellani CA, Longchamps RJ, Sun J, Guallar E, Arking DE. Thinking outside the nucleus: mitochondrial DNA copy number in health and disease. Mitochondrion. 2020;53:214‐223. doi: 10.1016/j.mito.2020.06.004 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0002] 2. Picard M. Blood mitochondrial DNA copy number: what are we counting? Mitochondrion. 2021;60:1‐11. doi: 10.1016/j.mito.2021.06.010 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0003] 3. Longchamps RJ, Castellani CA, Yang SY, et al. Evaluation of mitochondrial DNA copy number estimation techniques. PLOS ONE. 2020;15(1):e0228166. doi: 10.1371/journal.pone.0228166 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0004] 4. Baloyannis SJ. Mitochondrial alterations in Alzheimer's disease. J Alzheimer's Dis. 2006;9(2):119‐126. doi: 10.3233/jad-2006-9204 [DOI] [PubMed] [Google Scholar]

[alz71469-bib-0005] 5. Wang L, Han J, Fearnley LG, et al. Peripheral immune cell abundance differences link blood mitochondrial DNA copy number and Parkinson's disease. NPJ Parkinson's Dis. 2024;10:219. doi: 10.1038/s41531-024-00831-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0006] 6. Harerimana NV, Paliwali D, Romero‐Molina C, et al. The role of mitochondrial genome abundance in Alzheimer's disease. Alzheimers Dement. 2023;19(5):2069‐2083. doi: 10.1002/alz.12812 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0007] 7. Qin P, Chen X, Ma P, et al. Mitochondrial DNA copy number and Alzheimer's disease and Parkinson disease. Mitochondrion. 2025;83:102032. doi: 10.1016/j.mito.2025.102032 [DOI] [PubMed] [Google Scholar]

[alz71469-bib-0008] 8. Risi B, Imarisio A, Cuconato G, Padovani A, Valente EM, Filosto M. Mitochondrial DNA (mtDNA) as fluid biomarker in neurodegenerative disorders: a systematic review. Eur J Neurol. 2025;32(1):e70014. doi: 10.1111/ene.70014 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0009] 9. Lei S, Liu Y. Identifying blood mitochondrial DNA copy number as a biomarker for development of neurodegenerative diseases: evidence from Mendelian randomization analysis. Neuroscience. 2025;573:421‐429. doi: 10.1016/j.neuroscience.2025.04.003 [DOI] [PubMed] [Google Scholar]

[alz71469-bib-0010] 10. Xia H, Wang Zi‐H, Wang X‐B, et al. Investigating the genetic association of mitochondrial DNA copy number with neurodegenerative diseases. BMC Neurol. 2025;25:160. doi: 10.1186/s12883-025-04176-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0011] 11. van Rheenen W, Peyrot WJ, Schork AJ, Lee SH, Wray NR. Genetic correlations of polygenic disease traits: from theory to practice. Nat Rev Genet. 2019;20(10):10. doi: 10.1038/s41576-019-0137-z [DOI] [PubMed] [Google Scholar]

[alz71469-bib-0012] 12. Choi SW, Mak TSH, O'Reilly PF. Tutorial: a guide to performing polygenic risk score analyses. Nat Protoc. 2020;15(9):2759‐2772. doi: 10.1038/s41596-020-0353-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0013] 13. Davies NM, Holmes MV, Davey Smith G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ. 2018;362:k601. doi: 10.1136/bmj.k601 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0014] 14. Darrous L, Mounier N, Kutalik Z. Simultaneous estimation of bi‐directional causal effects and heritable confounding from GWAS summary statistics. Nat Commun. 2021;12(1):7274. doi: 10.1038/s41467-021-26970-w [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0015] 15. Sanderson E, Glymour MM, Holmes MV, et al. Mendelian randomization. Nat Rev Methods Primers. 2022;2(1):1. doi: 10.1038/s43586-021-00092-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0016] 16. Kunkle BW, Grenier‐Boley B, Sims R, et al. Genetic meta‐analysis of diagnosed Alzheimer's disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat Genet. 2019;51(3):3. doi: 10.1038/s41588-019-0358-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0017] 17. Bellenguez Cé, Küçükali F, Jansen IE, et al. New insights into the genetic etiology of Alzheimer's disease and related dementias. Nat Genet. 2022;54(4):412‐436. doi: 10.1038/s41588-022-01024-z [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0018] 18. Nalls MA, Blauwendraat C, Vallerga CL, et al. Identification of novel risk loci, causal insights, and heritable risk for Parkinson's disease: a meta‐genome wide association study. Lancet Neurol. 2019;18(12):1091‐1102. doi: 10.1016/S1474-4422(19)30320-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0019] 19. Longchamps RJ, Yang SY, Castellani CA, et al. Genome‐wide analysis of mitochondrial DNA copy number reveals loci implicated in nucleotide metabolism, platelet activation, and megakaryocyte proliferation. Hum Genet. 2022;141(1):127‐146. doi: 10.1007/s00439-021-02394-w [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0020] 20. Hägg S, Jylhävä J, Wang Y, Czene K, Grassmann F. Deciphering the genetic and epidemiological landscape of mitochondrial DNA abundance. Hum Genet. 2021;140(6):849‐861. doi: 10.1007/s00439-020-02249-w [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0021] 21. Chong M, Mohammadi‐Shemirani P, Perrot N, et al. GWAS and ExWAS of blood mitochondrial DNA copy number identifies 71 loci and highlights a potential causal role in dementia. ELife. 2022;11:e70382. doi: 10.7554/eLife.70382 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0022] 22. Gupta R, Kanai M, Durham TJ, et al. Nuclear genetic control of mtDNA copy number and heteroplasmy in humans. Nature. 2023;620(7975):839‐848. doi: 10.1038/s41586-023-06426-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0023] 23. Astle WJ, Elding H, Jiang T, et al. The allelic landscape of human blood cell trait variation and links to common complex disease. Cell. 2016;167(5):1415‐1429.e19. doi: 10.1016/j.cell.2016.10.042 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0024] 24. Lyon MS, Andrews SJ, Elsworth B, Gaunt TR, Hemani G, Marcora E. The variant call format provides efficient and robust storage of GWAS summary statistics. Genome Biol. 2021;22(1):32. doi: 10.1186/s13059-020-02248-0 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0025] 25. Bulik‐Sullivan BK, Loh Po‐Ru, Finucane HK, et al. LD Score Regression Distinguishes Confounding from Polygenicity in Genome‐Wide Association Studies. Nat Genet. 2015;47(3):291‐295. doi: 10.1038/ng.3211 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0026] 26. Lu Q, Li B, Ou D, et al. A Powerful Approach to Estimating Annotation‐Stratified Genetic Covariance via GWAS Summary Statistics: Am J Hum Genet. Accessed March 14, 2025;101(6):939‐964. https://www.cell.com/ajhg/fulltext/S0002‐9297(17)30453‐6 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0027] 27. Rajabli F, Benchek P, Tosto G, et al. Multi‐ancestry genome‐wide meta‐analysis of 56,241 individuals identifies known and novel cross‐population and ancestry‐specific associations as novel risk loci for Alzheimer's disease. Genome Biol. 2025;26(1):210. doi: 10.1186/s13059-025-03564-z [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0028] 28. Vitale D, Koretsky MJ, Kuznetsov N, et al, GenoTools: an open‐source Python package for efficient genotype data quality control and analysis G3: Genes, Genomes, Genet. 2024;15:jkae268. doi: 10.1093/g3journal/jkae268 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0029] 29. Ge T, Chen CY, Ni Y, Feng YCA, Smoller JW. Polygenic prediction via Bayesian regression and continuous shrinkage priors. Nat Commun. 2019;10(1):1776. doi: 10.1038/s41467-019-09718-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0030] 30. Lambert SA, Wingfield B, Gibson JT, et al. Enhancing the Polygenic Score Catalog with tools for score calculation and ancestry normalization. Nat Genet. 2024;56(10):1989‐1994. doi: 10.1038/s41588-024-01937-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0031] 31. Petersen RC, Aisen PS, Beckett LA, et al. Alzheimer's Disease Neuroimaging Initiative (ADNI). Neurology. 2010;74(3):201‐209. doi: 10.1212/WNL.0b013e3181cb3e25 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0032] 32. Tong T, Zhu C, Farrell JJ, et al. Blood‐derived mitochondrial DNA copy number is associated with Alzheimer disease, Alzheimer‐related biomarkers and serum metabolites. Alzheimers Res Ther. 2024;16(1):234. doi: 10.1186/s13195-024-01601-w [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0033] 33. Myers TA, Chanock SJ, Machiela MJ. LDlinkR: an R Package for Rapidly Calculating Linkage Disequilibrium Statistics in Diverse Populations. Front Genetics. 2020;11: 157. doi: 10.3389/fgene.2020.00157 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0034] 34. Hemani G, Zheng J, Elsworth B, et al. The MR‐Base platform supports systematic causal inference across the human phenome. ELife. 2018;7:e34408. doi: 10.7554/eLife.34408 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0035] 35. Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLOS Genet. 2017;13(11):e1007081. doi: 10.1371/journal.pgen.1007081 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0036] 36. Mandl MM, Boulesteix AL, Burgess S, Zuber V. Outlier Detection in Mendelian Randomization. Stat Med. 2025;44(15‐17):e70143. doi: 10.1002/sim.70143 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0037] 37. Hemani G, Bowden J, Davey Smith G. Evaluating the potential role of pleiotropy in Mendelian randomization studies. Hum Mol Genet. 2018;27(R2):R195‐R208. doi: 10.1093/hmg/ddy163 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0038] 38. Bowden J, Spiller W, Del Greco M F, et al. Improving the visualization, interpretation and analysis of two‐sample summary data Mendelian randomization via the Radial plot and Radial regression. Int J Epidemiol. 2018;47(4):1264‐1278. doi: 10.1093/ije/dyy101 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0039] 39. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Intl J Epidemiol. 2015;44(2):512‐525. doi: 10.1093/ije/dyv080 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0040] 40. Bowden J, Davey Smith G, Haycock PC, Burgess S, Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol. 2016;40(4):304‐314. doi: 10.1002/gepi.21965 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0041] 41. Hartwig FP, Davey Smith G, Bowden J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol. 2017;46(6):1985‐1998. doi: 10.1093/ije/dyx102 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0042] 42. Sanderson E, Spiller W, Bowden J. Testing and correcting for weak and pleiotropic instruments in two‐sample multivariable Mendelian randomization. Stat Med. 2021;40(25):5434‐5452. doi: 10.1002/sim.9133 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0043] 43. Patel A, Ye T, Xue H. MendelianRandomization v0.9.0: updates to an R package for performing Mendelian randomization analyses using summarized data. Wellcome Open Res. 2023;8:449. doi: 10.12688/wellcomeopenres.19995.2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0044] 44. Burgess S, Davies NM, Thompson SG. Bias due to participant overlap in two‐sample Mendelian randomization. Genet Epidemiol. 2016;40(7):597‐608. doi: 10.1002/gepi.21998 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0045] 45. Parchwani D, Singh R, Patel D. Biological and translational attributes of mitochondrial DNA copy number: laboratory perspective to clinical relevance. World J Methodol. 2025;15(3):102709. doi: 10.5662/wjm.v15.i3.102709 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0046] 46. Hu MM, Shu HB. Mitochondrial DNA‐triggered innate immune response: mechanisms and diseases. Cell Mol Immunol. 2023;20(12):1403‐1412. doi: 10.1038/s41423-023-01086-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0047] 47. Tresse E, Marturia‐Navarro J, Sew WQiG, et al. Mitochondrial DNA damage triggers spread of Parkinson's disease‐like pathology. Mol Psychiatry. 2023;28(11):4902‐4914. doi: 10.1038/s41380-023-02251-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0048] 48. Buneeva O, Fedchenko V, Kopylov A, Medvedev A. Mitochondrial Dysfunction in Parkinson's Disease: focus on Mitochondrial DNA. Biomedicines. 2020;8(12):591. doi: 10.3390/biomedicines8120591 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0049] 49. Macdonald R, Barnes K, Hastings C, Mortiboys H. Mitochondrial abnormalities in Parkinson's disease and Alzheimer's disease: can mitochondria be targeted therapeutically?. Biochem Soc Trans. 2018;46(4):891‐909. doi: 10.1042/BST20170501 [DOI] [PubMed] [Google Scholar]

[alz71469-bib-0050] 50. Menta BW, Schueddig E, Ranjan A, et al. MtDNA‐depleted neuronal cell transcriptomes reveal Alzheimer's disease‐related changes. Alzheimers Dementia. 2025;21(11):e70929. doi: 10.1002/alz.70929 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0051] 51. Anderson EL, Davies NM, Korologou‐Linden R, Kivimäki M. Dementia prevention: the Mendelian randomisation perspective. J Neurol Neurosurg Psychiatry. 2024;95(4):384‐390. doi: 10.1136/jnnp-2023-332293 [DOI] [PubMed] [Google Scholar]

[alz71469-bib-0052] 52. Wu Y, Sun Z, Zheng Q, et al. Pervasive biases in proxy GWAS based on parental history of Alzheimer's disease. bioRxiv. Preprint posted online October 17, 2023:2023.10.13.562272. doi: 10.1101/2023.10.13.562272 [DOI] [Google Scholar]

[alz71469-bib-0053] 53. Escott‐Price V, Baker E, Shoai M, et al. Genetic analysis suggests high misassignment rates in clinical Alzheimer's cases and controls. Neurobiol Aging. 2019;77:178‐182. doi: 10.1016/j.neurobiolaging.2018.12.002 [DOI] [PubMed] [Google Scholar]

[alz71469-bib-0054] 54. Swerdlow RH, Khan SM. The Alzheimer's Disease Mitochondrial Cascade Hypothesis: an Update. Exp Neurol. 2009;218(2):308‐315. doi: 10.1016/j.expneurol.2009.01.011 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0055] 55. Kurki MI, Karjalainen J, Palta P, et al. FinnGen provides genetic insights from a well‐phenotyped isolated population. Nature. 2023;613(7944):508‐518. doi: 10.1038/s41586-022-05473-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[alz71469-bib-0056] 56. Uffelmann E, Wightman DP, Bahrami S, et al. Genomic analyses reveal new insights into Alzheimer's disease. medRxiv. Published online January 1, 2025:2025.10.10.25337470. doi: 10.1101/2025.10.10.25337470 [DOI] [Google Scholar]

[alz71469-bib-0057] 57. Nguyen TN, Terry L, Guggenheim JA, Mendelian Randomization studies of Myopia: choosing the right summary statistics. medRxiv. Preprint posted online June 5, 2025:2025.06.04.25328975. doi: 10.1101/2025.06.04.25328975 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Evaluating the causal effect of mitochondrial dysfunction on Alzheimer's disease and Parkinson's disease using polygenic risk scores and Mendelian randomization

Aadrita Chatterjee

Brian Alvarez

Rakshya U Sharma

Caroline Jonson

Heather M Wilkins

Judy Pa

Russell H Swerdlow

Alison Goate

Kristine Yaffe

Shea J Andrews

Abstract

INTRODUCTION

METHODS

RESULTS

DISCUSSION

Highlights

1. BACKGROUND

2. METHODS

2.1. Data sources

TABLE 1.

RESEARCH IN CONTEXT

2.2. Genetic correlations

2.3. PRSs

TABLE 2.

2.4. Mendelian randomization

2.4.1. Genetic instrument selection and harmonization of data

2.4.2. Statistical analysis

2.4.3. Diagnostics

2.4.4. Sensitivity analyses

2.4.5. MVMR

2.4.6. LHC‐MR

3. RESULTS

3.1. mtDNAcn is negatively correlated with AD and PD

FIGURE 1.

3.2. mtDNAcn‐PRS is not associated with AD

FIGURE 2.

3.3. Univariable and MVMR showed no causal effect of mtDNAcn on AD or PD

3.4. LHC‐MR showed a causal effect of mtDNAcn on AD/dementia and PD, with limited evidence to support a causal association in the reverse direction

FIGURE 3.

4. DISCUSSION

CONFLICT OF INTEREST STATEMENT

CONSENT STATEMENT

Supporting information

ACKNOWLEDGMENTS

DATA AVAILABILITY STATEMENT

REFERENCES

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases