Mitochondrial DNA Copy Number and Multiple Sclerosis Progression: Insights from Bidirectional Mendelian Randomization

A recent study in Molecular Neurobiology examines an important and unresolved question in multiple sclerosis research: is mitochondrial DNA copy number (mtDNA-CN) a driver of disease progression, or is it instead a downstream consequence of worsening disease? Using a bidirectional two-sample Mendelian randomization design, the authors argue for the latter interpretation. Their central finding is that genetically predicted mtDNA-CN does not appear to causally influence multiple sclerosis progression, whereas genetically predicted progression severity in multiple sclerosis is associated with a reduction in mtDNA-CN. This result reframes mtDNA-CN less as an upstream causal trigger and more as a biologically informative readout of disease-related mitochondrial stress.

Why this question matters
Multiple sclerosis (MS) is a chronic inflammatory and neurodegenerative disease of the central nervous system, marked by demyelination, axonal injury, and, over time, progressive neurological disability. Although inflammation has long occupied the center of MS pathophysiology, mitochondrial dysfunction has emerged as a serious mechanistic candidate in progressive disease. The rationale is compelling: neurons and axons are extraordinarily energy-dependent, and mitochondrial failure can amplify oxidative stress, impair ATP production, destabilize ion homeostasis, and accelerate neurodegeneration. The article situates mtDNA-CN within this framework, noting that mitochondrial DNA is essential for oxidative phosphorylation and broader mitochondrial homeostasis, and that fluctuations in mtDNA-CN have been observed across neurodegenerative disorders, including MS.

The challenge, however, is interpretation. Prior observational studies linking mtDNA-CN with MS severity have produced inconsistent results. Some studies reported inverse associations between mtDNA-CN and disease duration or disability, while others found weak or non-significant relationships. Because such studies are vulnerable to confounding and reverse causation, they cannot easily tell us whether altered mtDNA-CN contributes to progression or merely reflects damage that is already underway. That uncertainty is precisely what this study set out to address. The core idea of the study
The authors used bidirectional two-sample Mendelian randomization (MR), a method that leverages inherited genetic variants as instrumental variables for an exposure. In principle, because these variants are assigned at conception, they are less vulnerable than observational measurements to confounding by environment, treatment, disease duration, or other downstream processes. The paper explicitly frames its analysis around the three standard MR assumptions: the selected variants must be associated with the exposure, independent of major confounders, and linked to the outcome only through the exposure pathway. The conceptual workflow is summarized in the study’s design figure and assumption schematic on pages 3 and 4.

This was a bidirectional analysis, which is especially important here. The investigators first tested whether mtDNA-CN causally affects MS progression, and then reversed the direction to test whether MS progression causally affects mtDNA-CN. That design allows a stronger interrogation of directionality than a one-way MR study would provide. Data sources and analytical framework
For mtDNA-CN, the study drew on genome-wide association summary statistics from the UK Biobank, focusing on 383,476 individuals of European ancestry. For MS progression, the authors used summary statistics from the International Multiple Sclerosis Genetics Consortium, based on 12,584 European-ancestry MS cases. MS severity was indexed using age-related MS severity scores derived from the Expanded Disability Status Scale, an attempt to normalize disability relative to age and thereby better capture progression biology.

Instrument selection was handled differently for the two traits. For mtDNA-CN, the investigators used the conventional genome-wide significance threshold of P < 5 × 10⁻⁸. For MS progression, they relaxed the threshold to P < 5 × 10⁻⁵ to secure a workable number of instruments, acknowledging that this introduces greater risk of weak instrument bias and pleiotropy. To address that risk, they filtered out weak instruments using F-statistics, applied stringent linkage disequilibrium clumping, harmonized alleles, removed problematic palindromic SNPs where necessary, and supplemented inverse variance weighted analysis with a large battery of sensitivity methods. These included MR-Egger, weighted median, mode-based approaches, RAPS, MR-cML, dIVW, MR-Lasso, MR-PRESSO, RadialMR, leave-one-out analyses, and outlier diagnostics such as Cook’s distance and standardized residuals.

This extensive sensitivity framework is one of the strengths of the paper. The authors were clearly aware that claims about causality in MR studies depend heavily on instrument validity and robustness to pleiotropy. What the study found
1. No evidence that mtDNA-CN causes MS progression
In the forward analysis, the authors began with 6,694 genome-wide significant mtDNA-CN-associated SNPs, which were reduced through clumping and harmonization to a smaller instrument set and then further refined after outlier detection. The final forward MR model used 43 instrumental SNPs. Across the primary inverse variance weighted analysis, there was no significant association between mtDNA-CN and MS progression (P = 0.487). Other MR approaches yielded similarly non-significant results. The forest plot on page 5 shows this clearly: effect estimates hover around zero with confidence intervals crossing the null across methods.

Just as importantly, the diagnostics did not suggest major technical artifacts. Heterogeneity was negligible, with Cochran’s Q and Rucker’s Q both non-significant and I² equal to 0%. The MR-Egger intercept was also non-significant, arguing against detectable horizontal pleiotropy. The authors further report consistency across leave-one-out and scatter/funnel/forest inspections. Taken together, the forward direction appears negative not because the model was obviously broken, but because the signal was simply not there.

2. Evidence that MS progression reduces mtDNA-CN
The reverse analysis is where the paper becomes especially interesting. Starting from 356 SNPs associated with MS progression at the chosen significance threshold, the authors ultimately retained 76 SNPs after filtering and outlier handling. In the primary inverse variance weighted model, genetically predicted MS progression was associated with a decrease in mtDNA-CN, with β = −0.010, 95% CI −0.019 to −0.001, P = 0.036. Several alternative MR methods supported the same general direction and reached nominal statistical significance, including penalized IVW, robust IVW, penalized robust IVW, RAPS, MR-cML, dIVW, and MR-Lasso. The reverse-direction forest plot on page 7 and the comparative scatter plot on page 8 both show a consistent negative slope across methods.

Again, the supporting diagnostics were reassuring. The analysis showed no meaningful heterogeneity, no evidence of horizontal pleiotropy by MR-Egger intercept, and a correct direction according to the MR Steiger test. While some secondary MR methods did not individually reach significance, the direction of effect was broadly aligned, which strengthens the authors’ interpretation that worsening MS is likely upstream of lower mtDNA-CN.

How should these findings be interpreted biologically?
The discussion section offers a mechanistic model that is plausible and well aligned with current views of progressive MS biology. Mitochondrial DNA is especially vulnerable to oxidative damage because it lacks histones and has less efficient repair systems than nuclear DNA. The mitochondrial D-loop, which regulates replication, is particularly mutation-prone, and the authors cite evidence that MS patients harbor more mutations in this region. In that context, chronic inflammatory stress, excitotoxic injury, and impaired oxidative phosphorylation in MS could plausibly drive mtDNA damage and eventually reduce mtDNA-CN.

The paper sketches a disease cascade in which chronic inflammation and progressive demyelination increase neuronal energy demand, including heightened Na⁺/K⁺-ATPase activity. Activated immune cells and injured neurons generate reactive oxygen species and glutamate-related excitotoxic stress, while inflammatory mediators such as TNF-α further impair oxidative phosphorylation. Over time, these pressures produce mitochondrial dysfunction, mtDNA alterations, reduced ATP generation, ionic imbalance, calcium overload, and ultimately axonal degeneration and neuronal death. In that model, reduced mtDNA-CN is not the initial spark but rather part of the mitochondrial collapse that accompanies ongoing progression.

This is a subtle but important distinction. The study does not suggest that mtDNA-CN is irrelevant. On the contrary, it may be highly informative. But it may function more as a state marker of biological deterioration than as an inherited causal determinant of progression.

A biomarker, but perhaps not a master switch
One of the most valuable implications of the study is its repositioning of mtDNA-CN as a candidate biomarker. The authors note that several observational studies have reported lower mtDNA-CN with longer disease duration or greater disability, and they argue that their MR findings are broadly consistent with those reports. They further suggest that mtDNA-CN could complement established biomarkers such as neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP). Conceptually, that makes sense: mtDNA-CN may capture mitochondrial health, NfL may reflect neuroaxonal injury, and GFAP may index astroglial activation. A composite biomarker strategy could therefore provide a richer view of disease activity and progression than any single marker alone.

The paper also raises an important issue about tissue specificity. mtDNA-CN varies dramatically across tissues, and measurements in blood may not perfectly reflect what is happening in the central nervous system. Blood-based mtDNA-CN may track systemic oxidative stress and immune activation, whereas CSF or brain tissue measures may more directly reflect compartmentalized CNS pathology. This matters for future translational work: if mtDNA-CN is to be developed as a progression biomarker, the biological meaning of the sampled tissue will need to be carefully defined.

Strengths of the study
Several features make this paper notable. First, it addresses a biologically important question with a design capable of probing directionality, rather than simple association. Second, it uses very large GWAS resources for both mtDNA-CN and MS progression. Third, it applies a broad suite of MR sensitivity analyses and pleiotropy diagnostics rather than relying on a single estimator. Fourth, the reported absence of heterogeneity and detectable horizontal pleiotropy increases confidence that the main results are not trivially explained by unstable instruments or obvious off-target genetic effects.

Limitations that should temper interpretation
The authors are appropriately cautious. Their results are derived primarily from populations of European ancestry, so generalizability to other ancestries remains uncertain. The GWAS-based nature of the analysis also means that residual issues such as population structure and relatedness cannot be entirely dismissed. Most importantly, MR can strengthen causal inference but does not by itself reveal the precise molecular pathway connecting disease progression to lower mtDNA-CN. Mechanistic work remains necessary.

Another limitation is the relaxed SNP inclusion threshold used for MS progression instruments (P < 5 × 10⁻⁵ rather than genome-wide significance). The authors justify this choice as a practical step to increase the number of instruments, but it does heighten concern about weak instruments and pleiotropy. They attempted to mitigate that through F-statistic filtering and multiple robust MR methods, yet the issue does not disappear entirely. Finally, the reverse effect size is statistically significant but modest, which means overstatement should be avoided. This is an informative signal, not a definitive blueprint of mitochondrial pathology in MS.

Broader implications for MS research and therapy
The study fits into a broader movement in MS research that increasingly recognizes progressive disease as not merely an inflammatory extension of relapsing disease, but as a syndrome deeply shaped by bioenergetic failure, compartmentalized inflammation, and CNS resilience. In that context, mitochondria are attractive not only as biomarkers but also as therapeutic targets. The authors suggest that interventions aimed at improving mitochondrial function or reducing oxidative stress could complement standard immunomodulatory therapies, particularly in progressive MS. That proposal remains hypothetical here, but it is biologically coherent.

What this paper adds is a sharpened causal framing: if disease progression drives mitochondrial genomic depletion, then therapies that slow progression may also preserve mitochondrial integrity, and therapies that directly protect mitochondria may still be useful even if mtDNA-CN itself is not an upstream inherited cause of progression. In other words, mtDNA-CN may still matter clinically even if it is not the initial causal lever.

Final perspective
This study advances the field by disentangling two possibilities that observational studies could not cleanly separate. The evidence presented argues against a causal role for inherited variation in mtDNA-CN as a driver of MS progression, while supporting the idea that progressive MS biology contributes to declining mtDNA-CN. That makes mtDNA-CN a promising marker of mitochondrial stress and disease state, rather than a proven initiating mechanism. It is a meaningful distinction, and one with implications for biomarker development, mechanistic research, and therapeutic design.

In formal terms, the paper does not close the case on mitochondrial dysfunction in MS; it refines it. Mitochondria remain central to the story, but perhaps less as the hidden author of progression than as one of its most sensitive witnesses.

Disclaimer: This blog post is based on the provided research article and is intended for informational purposes only. It is not intended to provide medical advice. Please consult with a healthcare professional for any health concerns.

References:
Sabaie, H., Taghavi Rad, A., Shabestari, M. et al. Mitochondrial DNA Copy Number as a Hidden Player in the Progression of Multiple Sclerosis: A Bidirectional Two-Sample Mendelian Randomization Study. Mol Neurobiol 62, 11643–11653 (2025). https://doi.org/10.1007/s12035-025-04980-9