We read with interest Rho et al. [1]’s article about a retrospective observational study of 40 Korean patients with maternally inherited diabetes and deafness syndrome (MIDD) due to the mutant mitochondrial DNA (mtDNA) variant m.3243A>G in mitochondrially encoded tRNA leucine (MT-TL1) with a mean heteroplasmy rate of 30% [1]. The most common comorbidity besides diabetes was hearing loss in 90% of patients, albuminuria in 60% of patients, seizures in 38% of patients, and stroke in one-third of patients [1]. There was a negative correlation between heteroplasmy rate and age at diabetes onset [1]. The work is compelling, but some points should be discussed.
The first point is that the diagnosis of MIDD cannot be maintained for all 40 patients. Since 13 patients had stroke-like episodes (SLEs) and SLEs are pathognomonic of mitochondrial encephalopathy, lactic acidosis, and stroke-like episode (MELAS) syndrome, at least these 13 patients should be reclassified as MELAS syndromes and excluded from the analysis. Since serum lactate was apparently elevated in several patients [1], we should know how many of the patients with SLE’s also suffered from lactic acidosis, a common phenotypic feature of MELAS. How many of the SLE patients also had magnetic resonance spectroscopy (MRS), and in how many of these did the MRS show a lactate peak and a reduced N-acetyl aspartate peak? In this context, it must also be clarified what the authors mean with the statement that 13 and 15 patients had SLEs [1]. Does this mean that the 15 patients with seizures also had SLEs? Did patients with SLEs have higher heteroplasmy compared to MIDD patients without SLEs?
A second point is that the reason for excluding patients with a glycosylated hemoglobin (HbA1c) replace by <6.5 from the analysis is not understandable. Patients with MIDD do not necessarily need have full-blown diabetes, but may also have prediabetes. Therefore, m.3243A>G carriers with an HbA1c between 5.7 and 6.5 should also be included in the analysis. Prediabetes could be interpreted as part of the phenotypic heterogeneity in MIDD patients.
A third point is that heteroplasmy rates were determined in blood lymphocytes, not in a tissue or organ that was presumably clinically affected [1]. Since heteroplasmy rates can vary significantly between tissues and organs, it would have been desirable to determine heteroplasmy rates not only in unaffected tissues/organs, but especially in phenotypically affected tissues. The significance of heteroplasmy rates from non-affected tissues/organs is of limited importance. This is also reflected in the fact that the heteroplasmy rates in blood were very low at an average of 30% [1]. Did any of the 40 patients have immune deficiency due to leukocyte dysfunction?
A fourth point is that for some parameters in Table 1 [1] only mean values were provided, but not the actual number of patients with an abnormal result. To assess how many actually had type 1 diabetes mellitus, we should know how many people actually had reduced C-peptide levels. In addition, we should know how many had elevated serum lactate and how many had short stature. Was the chronic renal failure that was present in all patients really due to diabetes in all included patients?
A fifth point is that no adequate explanation has been provided for the negative correlation between heteroplasmy rate and age at diabetes onset. Was this result interpreted as an artefact? One would expect that high heteroplasmy rates would be associated with increased phenotype severity. Therefore, a high heteroplasmy rate should be associated with early-onset diabetes but not with late-onset diabetes.
In summary, the excellent study has limitations, which complicate the interpretation of the results. Addressing these limitations could strengthen and reinforce the statement of the study. In a study of MIDD patients, MELAS should be excluded from the analysis. Before correlating heteroplasmy rates with phenotypic expression, the reliability of heteroplasmy rates should be checked.