Neurol Genet. 2026 Apr;12(2):
e200365
Laura Bermejo-Guerrero,
Juan Luis Restrepo-Vera,
Paloma Martin-Jimenez,
Maria Navarro-Riquelme,
Rocío Garrido-Moraga,
Luz Edith Ochoa,
Adrián González-Quintana,
Aurelio Hernandez-Lain,
Jessica Castillo-Villalba,
Germán Morís,
Nuria Muelas,
Elena García-Arumí,
Victoria González,
Elena Martínez-Sáez,
Ana Vesperinas,
Laura Llansó,
Raul Juntas-Morales,
Carmen Paradas,
Solange Kapetanovic,
Alberto Blázquez,
Ramon Martí,
Joaquín Arenas,
Miguel A Martín,
Cristina Domínguez-González.
Background and Objectives: POLG-related disorders exhibit marked phenotypic heterogeneity and frequent clinical overlap, often leading to delayed diagnosis. A precise delineation of their clinical spectrum, natural history, and the identification of reliable biomarkers is essential to improve diagnostic accuracy and guide therapeutic development.
Methods: We analyzed a cohort of 34 patients with confirmed pathogenic POLG variants, assessing clinical phenotypes, molecular findings, and biomarkers (plasma growth differentiation factor-15 [GDF15] in 16, plasma neurofilament light chain [NF-L] in 14, and mitochondrial DNA [mtDNA] copy number in muscle in 16).
Results: Thirty four patients (0.6-71 years) from 33 families were included. Juvenile/adult onset (12-40 years) was the most common presentation (62%). The predominant phenotypic categories were ataxia-neuropathy spectrum ([ANS], 44%), autosomal recessive PEO-plus (arPEO-plus, 26%), and autosomal dominant PEO-plus ([adPEO-plus], 15%), with frequent phenotypic overlap. Recessive inheritance accounted for 74% of cases, with the most common variants being p.([Thr251Ile; Pro587Leu]) paired on 1 allele, p.(Ala467Thr), and p.(Trp748Ser). Dominant variants were associated with milder, primarily myopathic phenotypes. The most common dominant variant was p.(Tyr955Cys). No clear genotype-phenotype correlations were identified among recessive variants. Compared with previously reported cohorts, our patients exhibited a lower prevalence of seizures, hepatopathy, and stroke-like episodes. GDF15 was elevated in 87.5% of patients, with a mean level of 3,315 pg/mL (±1,559.79), showing no significant differences between myopathic and ANS phenotypes, supporting its role as a general biomarker of mitochondrial dysfunction. NF-L was elevated in 78.6% of tested individuals but did not correlate with phenotype or clinical severity (as per Newcastle Mitochondrial Disease Adult Scale score).On average, muscle mtDNA copy number in patients was 76% of that observed in controls, with no differences by phenotype or inheritance pattern. All but 1 patient exhibited multiple mtDNA deletions, likely representing the primary mechanism of oxidative phosphorylation dysfunction rather than mtDNA depletion.
Discussion: POLG-related disorders demonstrate extensive clinical variability with no consistent genotype-phenotype correlation. GDF15 and NF-L may serve as useful, though nonspecific, biomarkers of mitochondrial and neuroaxonal dysfunction, respectively. Prospective studies incorporating advanced molecular profiling are essential to establish reliable outcome measures and inform future therapeutic strategies.