Interpreting results and understanding limitations

Advanced

Why PGT-mt is a quantitative ranking method

In PGT-M, the goal is an embryo without the pathogenic genotype (or with a single allele, depending on the mode of inheritance). In PGT-mt, quantitative heteroplasmy values are compared between embryos of the same cycle. The decision is made relative to the available embryos and the chosen cut-off. A difference of a few percentage points affects the choice only when the measured level lies close to the cut-off; for extreme values (0–5% or a high load) such a correction does not change the decision.

The biopsy and later distribution across tissues

Trophectoderm, chorionic villi, amniocytes, cord and peripheral blood, urinary and buccal epithelium represent different cell populations (Stewart & Chinnery, 2015; Burr & Chinnery, 2024). Agreement is therefore judged not by matching numbers but by whether the clinically relevant range is preserved. For m.8993T>G, transfer of embryos with very low or undetectable heteroplasmy was followed by the birth of children with low or undetectable variant levels (Steffann et al., 2006; Spath et al., 2021). After TE biopsy and transfer of a blastocyst in an m.3243A>G carrier, the variant was not detected in the child’s tested tissues (Heindryckx et al., 2014). After transfer of a blastocyst with 1.8% m.3250T>C, the variant was not detected in cord blood at birth or in blood and buccal epithelium at two years (Hu et al., 2026). After transfer of two embryos with heteroplasmy below 18% for the m.14487T>C mutation, the newborn’s cord blood had an mtDNA heteroplasmy level of approximately 5%. Because two embryos were transferred, it is not possible to determine which one gave rise to the child (Sallevelt et al., 2017b). When embryos with heteroplasmy of about 50–60% (m.8344A>G, m.3697G>A) were transferred, comparable levels were found in the foetus or child (Steffann et al., 2014; Ji et al., 2021). For m.3243A>G, reported blood heteroplasmy levels in one child ranged from less than 10% to 46–47% across analyses performed at different ages and using different methods (Treff et al., 2012; Mitalipov et al., 2014). Neurological and multisystem abnormalities were reported in the child; a link to the variant was not established (Mitalipov et al., 2014).

In the Newcastle programme, PGT resulted in 18 live births; the variant level in blood was known for 10 of the children and ranged from undetectable to 7% (Hyslop et al., 2025). In eight children after single-embryo transfer, the variant was not detected in either the test blastomere or blood (pyrosequencing threshold: 3%) (Hyslop et al., 2025).

The 18% and 30% reference points: origin and scope

• 18% — the result of a systematic review of mtDNA heteroplasmy levels in muscle tissue for 159 point variants, assuming an a priori probability of being affected of 0.477; it is a probability boundary across variants, and for m.3243A>G the chance of remaining unaffected is significantly lower (Hellebrekers et al., 2012). The reference point was used to estimate which women might benefit from mitochondrial donation (Pickett et al., 2019) and to select embryos in individual cases (Hu et al., 2026).
• 15–30% — the literature range of the symptomatic threshold for m.3243A>G, used in modelling estimates (Ji et al., 2023).
• 30% — the operational transfer criterion of the Newcastle programme with single day-3 blastomere biopsy — the only one of these reference points with a published clinical outcome for a series (Hyslop et al., 2025).

A cut-off used for embryo selection serves a different purpose from a tissue-specific biochemical or clinical threshold and may be more conservative: one part is measured at one time point, whereas the load in all tissues over a lifetime is what matters clinically. None of these numbers defines a safe threshold for an individual embryo.

Aneuploidy in the joint interpretation of PGT-mt and PGT-A

Even at a younger reproductive age in a carrier of a pathogenic mtDNA variant, some embryos with low heteroplasmy may be aneuploid or mosaic: in a 33-year-old carrier of m.3250T>C, all six blastocysts were below 18%, but three were aneuploid or mosaic (Hu et al., 2026). The scenario calculation in the PGT-mt × PGT-A model section assumes that heteroplasmy and ploidy are independent. This assumption is unproven: one hypothesis is that reduced mitochondrial function of the oocyte, combined with age-related metabolic changes, increases the risk of chromosome missegregation (Mihalas et al., 2024). Paired heteroplasmy–PGT-A data for each blastocyst have been published only for individual cycles (Spath et al., 2021; Ji et al., 2021; Hu et al., 2026; Tofilo et al., 2026).

Insufficient data and publication bias

• Embryos in published datasets are not independent: many come from one patient in one cycle, and mtDNA variants are unevenly represented.
• Embryo–child comparisons include different tissues, time points, methods and laboratories for the same child; there are more measurements than independent clinical observations.
• Successful outcomes are published more often; there is no systematic correction for this bias. The pooled proportion of low-heteroplasmy embryos therefore cannot be interpreted as an individual risk.

Unsuccessful cycles and complete accounting of embryos

Cycles without a suitable embryo, pregnancy losses and results for non-transferred embryos, immature and unfertilised oocytes and aneuploid blastocysts make it possible to trace the distribution of a variant through oogenesis and early development. Without them, it is impossible to move from describing individual cases to quantitative risk estimation. A useful minimum description of a cycle includes the carrier’s age and tissue levels, the numbers of oocytes and embryos, the biopsy day, the PGT-A result and the heteroplasmy of each embryo, the method (WGA, region, linearity, limit of detection) and the outcome with tissue, age and method (Mavraki et al., 2023). How such data can be shared with the Atlas is described in Contribute data.

This material does not replace a consultation with a clinical geneticist and does not provide an individual prognosis.