Mitochondrial donation and preimplantation testing for mtDNA disease

The authors themselves plan to keep measuring the maternal variant in the children; the design adds the sampling schedule, the tissues, the comparison groups and the assay floor they left unspecified, so that 'stable' has a testable meaning.

1. Has anyone done this since?

Status

partly addressed

Works since

• Mitochondrial Donation in a Reproductive Care Pathway for mtDNA Disease (McFarland et al. 2025, New England Journal of Medicine, doi:10.1056/NEJMoa2503658)
  The companion paper reports blood and urine heteroplasmy at birth for all 8 children (5%/9%, 16%/20% and 12%/13% in the three detectable cases), one child falling from 5% in blood at birth to undetectable at 18 months, a table with 1, 5 and 10 month columns, and states that long-term assessment of heteroplasmy is under way; it does not yet give a time series for every child or any tissue beyond blood and urine.
• First pilot study of maternal spindle transfer for the treatment of repeated in vitro fertilization failures in couples with idiopathic infertility (Costa-Borges et al. 2023, Fertility and Sterility, doi:10.1016/j.fertnstert.2023.02.008)
  Not a citing work but the closest human time series: one of six spindle-transfer children went from 0.8% maternal mtDNA at the blastocyst stage to 30 to 60% at birth, and a 2025 statement by the first author places the rise between the blastocyst stage and about 15 weeks of gestation; it shows the shift can be large and prenatal, which the Newcastle series could not observe because UK rules forbid assaying treatment embryos.

Searched

All 52 titles in the citing-works file; Europe PMC full text of PMC7617939 (McFarland) for 'heteroplasmy', 'follow-up', 'urine', 'buccal'; Europe PMC search 'mitochondrial donation children follow-up heteroplasmy tissues over time' and '"mitochondrial donation" OR "mitochondrial replacement" AND children AND follow-up AND heteroplasmy 2023-2026'; web search 'Newcastle mitochondrial donation pronuclear transfer children follow-up heteroplasmy 2026'.

Still open

No published time series beyond 18 months for one child, no tissue other than blood and urine in the pronuclear-transfer children, and no comparison against the known age-related decline of the same variants in naturally conceived carriers.

Registry query

0 of 49 records in the open node mention any of: pronuclear transfer, heteroplasmy, reversion, longitudinal, children, mtdna disease. The open node holds biology and chemistry so far, so this is the expected answer; a private node of a clinical group would be asked the same question.

2. A proposed experiment

Question

Does maternal-variant heteroplasmy in children born after pronuclear transfer rise, fall or hold across blood, urinary epithelium and buccal cells from birth to age 5?

Hypothesis

In children born after pronuclear transfer, blood heteroplasmy of the maternal variant does not rise by more than assay repeatability in any sampled tissue between birth and age 5.

Attempt type

observational-study

Design

A prospective longitudinal cohort of every child born after mitochondrial donation in the Newcastle programme (8 so far, plus those from the ongoing pregnancy and future births), with two comparison groups sampled in the same laboratory with the same assay: children born after PGT in the same programme who had detectable heteroplasmy at birth (at least the 7% child), and naturally conceived children of heteroplasmic carriers of the same variants from the NHS Highly Specialised Service cohort. Samples at birth (blood spot, cord blood, placenta), then blood, urinary epithelial cells and buccal mucosa at 6, 12, 18, 24 months and yearly to age 5. Readout: maternal-variant heteroplasmy by the same high-throughput sequencing assay the paper used (reported limit 0.06 to 0.17%), run in duplicate, plus pyrosequencing where above 3%. Primary analysis: per-child slope of heteroplasmy against age in each tissue, with a mixed model across children; the pre-specified alarm is any child whose slope is positive and exceeds the duplicate-measurement repeatability. Secondary: whether the spindle-transfer pattern (rise before birth, Costa-Borges 2023) is ever seen after pronuclear transfer, using cord blood and placenta against the birth blood spot.

Controls

• reference-case Naturally conceived children of carriers of the same variants, sampled on the same schedule and assay.
  Rules out: That a fall in blood heteroplasmy in the donation children is specific to donation rather than the ordinary age-related decline seen in carriers (about 2.3% per year for m.3243A>G in blood).
  Vocabulary note: A comparison cohort rather than a true reference case: naturally conceived carrier children, whose blood decline is already measured, give the expected trajectory the method must recover.
• designated A dilution series of patient and donor mtDNA at known ratios (for example 0.1, 0.5, 1, 3, 10, 30%) run with every batch.
  Rules out: That an apparent rise or fall is assay drift or loss of linearity at low heteroplasmy rather than a change in the child.
• instrument-noise-floor Duplicate extractions and library preparations from the same sample at every time point, giving the repeatability of the measurement.
  Rules out: That a change smaller than the duplicate difference is read as biological.
• within-subject Each child's own birth value in each tissue is the baseline; tissues are compared within the same child at the same visit.
  Rules out: Between-child differences in starting heteroplasmy and in variant being mistaken for change over time.

Sensitivity

The repeatability of the sequencing assay at the relevant range sets what counts as a change; the paper reports detection of 0.06 to 0.17% but not a repeatability figure, so the minimal detectable per-year change cannot be stated from the record. The one published comparator rate is the blood decline of about 2.3% per year for m.3243A>G in carriers (Grady et al. 2018), which the design should be able to recover in the comparison cohort. With 8 children no between-group test is powered; the design is per-child and descriptive, and the honest statement is that it can detect a rise in a single child larger than the duplicate repeatability, which a pilot of repeat measures must first establish.

Would count as

Supported → positive · opposed → refuted
Inconclusive if: The dilution series fails linearity below 1%, or duplicate repeatability is wider than the changes seen, or fewer than half of the children give samples at three or more visits, so that no slope can be fitted.

Known before

• Blood m.3243A>G heteroplasmy in carriers declines by about 2.3% per year; urine and buccal values are higher and correlate with blood. (Grady et al. 2018, EMBO Molecular Medicine, doi:10.15252/emmm.201708262)
  So: A fall in blood heteroplasmy in a donation child is the expected direction, so stability must be judged against this decline and in more than one tissue.
• Heteroplasmy declines with age in leucocytes, urinary epithelial cells and buccal mucosa in 82 m.3243A>G carriers, and urinary epithelial cells are the preferred non-invasive sample. (de Laat et al. 2012, Journal of Inherited Metabolic Disease, doi:10.1007/s10545-012-9465-2)
  So: Fixes the three non-invasive tissues the schedule should use and warns that a single tissue cannot answer 'across tissue types'.
• A boy born after spindle transfer had neonatal mutation load of 2.36 to 9.23% across tested tissues from a 5.7% blastocyst. (Zhang et al. 2017, Reproductive BioMedicine Online, doi:10.1016/j.rbmo.2017.01.013)
  So: Tissue-to-tissue spread of several percentage points is already documented at birth, so the design needs the within-child tissue comparison, not just blood.
• In rhesus macaques made by spindle transfer, donor, maternal and paternal mtDNA were followed in all organs into adulthood and the next generation; growth, health and fertility were unremarkable. (Ma et al. 2021, Human Reproduction, doi:10.1093/humrep/deaa308)
  So: The only longitudinal multi-tissue series after mitochondrial replacement is in primates; it gives the template and the expectation that most tissues stay low.
• A stem cell line from a pronuclear-transfer blastocyst with 4% carryover showed a progressive increase in maternal heteroplasmy; carryover was under 2% in 79% of blastocysts after optimisation. (Hyslop et al. 2016, Nature, doi:10.1038/nature18303)
  So: The group's own preclinical result is the reason the question exists: amplification from a few percent is possible and the schedule has to be dense enough early to catch it.
• Low-level carryover after nuclear transfer in human oocytes can either vanish or drift and revert to the maternal genotype in stem cell lines, independently of mitochondrial function. (Yamada et al. 2016, Cell Stem Cell, doi:10.1016/j.stem.2016.04.001)
  So: Reversion is stochastic in vitro, so the cohort should not expect a single direction and each child must be read on its own.

Searched: Europe PMC: 'm.3243A>G heteroplasmy blood age decline' (title-restricted, 2000-2020), 'mitochondrial replacement reversion heteroplasmy embryonic stem cells replicative advantage', 'mitochondrial donation children follow-up heteroplasmy tissues over time'; abstracts fetched by DOI for Grady 2018, de Laat 2012, Zhang 2017, Ma 2021, Hyslop 2016, Yamada 2016, Paull 2013; web search for a 2026 Newcastle follow-up found none.

Feasibility

The Newcastle Fertility Centre and the NHS Highly Specialised Service for Rare Mitochondrial Disorders, which already hold the children, the carrier cohort and the assay; the Monash mitoHOPE programme once it has births. · 5 years to reach the age-5 endpoint; an interim at 24 months. · low, because sampling is non-invasive and the sequencing assay already runs in the same laboratory
Blockers: Consent and retention of families over 5 years; Children with undetectable heteroplasmy at birth give no slope to fit; Ethics approval to sample healthy comparison children of carriers

Strongest objection

Blood, urine and buccal cells are all derived from lineages that may not be where reversion happens; if the rise occurs in post-mitotic tissue such as muscle or brain, a stable blood series would be false reassurance, and muscle biopsy of healthy children is not ethical. The specialist would also note that the Costa-Borges case shifted before birth, so a cohort starting at birth may be sampling after the event.

Not derivable from what is published

• The assay's repeatability at 0.1 to 5% heteroplasmy, which sets the detectable change
• The number of children needed to say anything beyond per-child description
• Whether the ongoing pregnancy and later births will be enrolled

3. Send it to a lab

Send to a lab

This design is a study in people. It cannot be sent to a laboratory service; it needs a trial network, ethics approval and a sponsor.
This is a longitudinal cohort of children; no bench job exists, although the sequencing of each visit's samples could be contracted once samples are collected.

Two children have far more maternal mtDNA than their carryover scores predict, and the authors offer two mechanisms; this asks for an experiment on research embryos that can tell the two mechanisms apart.

1. Has anyone done this since?

Status

nothing found

Works since

• The bottleneck for maternal transmission of mtDNA is linked to purifying selection by autophagy (Kremer et al. 2025, Science Advances, doi:10.1126/sciadv.aea4660)
  Mouse models show bottleneck size modulates purifying selection and inter-offspring variance; it is about germline transmission, not post-transfer carryover, and does not test either mechanism in human embryos.
• Revisiting the promise and pitfalls of mitochondrial replacement therapies (Costa-Borges and Wells 2026, Human Reproduction, doi:10.1093/humrep/deag020)
  An opinion piece stating that recent clinical data raise questions about reliability; no new experiment.
• Common Principles Underlie Mitochondrial DNA Heteroplasmy Dynamics in the Germline and Soma (Ryall, Chinnery and van den Ameele 2026, Annual Review of Genomics and Human Genetics, doi:10.1146/annurev-genom-120324-032239)
  A review naming the candidate processes (relaxed replication, vegetative segregation, preferential replication, mitophagy) and the single-cell tools to separate them; no data on donation embryos.

Searched

All 52 citing-works titles, abstracts fetched for the three above; Europe PMC 'mitochondrial replacement reversion heteroplasmy embryonic stem cells replicative advantage', '"mitochondrial replacement" AND (reversion OR "replicative advantage" OR "D-loop") AND haplotype', 'single-cell heteroplasmy human blastocyst inner cell mass trophectoderm mitochondrial replacement carryover segregation'; web search 'Costa-Borges spindle transfer pilot child heteroplasmy reversion'.

Still open

Nobody has measured, in human donation embryos, whether a rise in maternal mtDNA appears as a uniform increase across cells (replication advantage) or as a few cells with high load (segregation), nor whether it tracks the donor and patient D-loop genotypes.

Registry query

0 of 49 records in the open node mention any of: pronuclear transfer, carryover, reversion, d-loop, single-cell heteroplasmy, inner cell mass. The open node holds biology and chemistry so far, so this is the expected answer; a private node of a clinical group would be asked the same question.

2. A proposed experiment

Question

In PNT embryos, is a rise in maternal mtDNA uniform across cells and predicted by the donor-patient D-loop pairing, or confined to a few cells with unchanged mean?

Hypothesis

Where a PNT blastocyst exceeds 3.9% maternal heteroplasmy, cell-to-cell variance is no higher than in low cases and the excess is predicted by the patient-donor D-loop CSBII pair.

Attempt type

wet-lab

Design

Use PNT embryos that are not suitable for treatment and are donated to research under the existing HFEA research licence R0152, as the paper already did for 62 embryos. Arm A (haplotype): whole-mtDNA sequence the patient and donor for every embryo in the existing set and all new ones; classify each pair by D-loop conserved sequence box II genotype and by total sequence divergence; regress embryo heteroplasmy on pair class with the carryover score as covariate. Arm B (segregation): culture research PNT embryos to day 5 or 6, dissociate each blastocyst into single cells, separate inner cell mass from trophectoderm by immunosurgery or by marker-sorted single-cell picking, and measure maternal heteroplasmy per cell by the group's high-throughput assay. Readouts: per-embryo mean and between-cell variance, inner cell mass versus trophectoderm mean. Replication advantage predicts a higher mean with low variance and a dependence on D-loop class; unequal segregation predicts a few high cells with an unchanged mean and no haplotype dependence. Compare against PGT research embryos carrying natural heteroplasmy, where segregation is reported to be even.

Controls

• reference-case The Hyslop 2016 stem cell line from a 4% carryover PNT blastocyst that rose progressively, and the paper's own 62 embryos with carryover scores, as cases with a known answer the assay must reproduce.
  Rules out: That the assay or the scoring cannot detect a rise that the group has already documented.
• within-subject Inner cell mass and trophectoderm cells from the same blastocyst, and single cells against the whole-embryo mean.
  Rules out: That between-embryo differences in carryover are read as lineage or cell-level segregation.
• designated Dilution series of patient and donor mtDNA at known ratios run alongside single-cell libraries, plus pooled-cell samples reconstituted from the same embryo.
  Rules out: That single-cell amplification noise at low copy number is mistaken for cell-to-cell heteroplasmy variance.
• background PGT research embryos from heteroplasmic carriers, with no manipulation, dissociated and sequenced the same way.
  Rules out: That whatever variance is found is a general property of human blastocysts rather than a consequence of pronuclear transfer.

Sensitivity

The paper gives the whole-embryo distribution (mean 1.18%, s.d. 0.90%, range undetectable to 3.9%, n=62), so a rise to 12 to 16% at whole-embryo level is more than ten standard deviations away and would be unmistakable; what cannot be set from the record is the single-cell measurement variance at 1 to 5% heteroplasmy, which Rubens et al. 2025 found to be substantial in 1 of 8 embryos for single trophectoderm cells. A pilot of 5 embryos with 20 cells each is needed to fix that floor before the variance test is powered. For Arm A, the number of embryos needed to detect a D-loop class effect is not derivable: the paper gives no haplotype data and the effect size in Kang et al. 2016 is from stem cell lines, not embryos.

Would count as

Supported → positive · opposed → negative-no-association
Inconclusive if: No research PNT embryo in the series exceeds 3.9%, so there is no high case to dissect, or single-cell dropout makes per-cell heteroplasmy unreadable below 1%.

Known before

• After spindle transfer, some embryonic stem cell lines reverted to the maternal haplotype; a D-loop conserved sequence box II polymorphism was proposed as the cause of preferential replication, and donor matching was suggested. (Kang et al. 2016, Nature, doi:10.1038/nature20592)
  So: Gives Arm A its specific genotype to test, and the prediction that reversion is pair-dependent.
• Reversion after replacement of mitochondrial DNA was re-examined by the Newcastle group in a Brief Communications Arising on Kang 2016; the paper cites it alongside Kang for a replicative advantage of patient mtDNA. (Hudson, Takeda and Herbert 2019, Nature, doi:10.1038/s41586-019-1623-3)
  So: The two groups disagree on what drives reversion, so the design must pre-register which genotype classification it will use; the text was not accessible (no abstract on Europe PMC, Nature page paywalled), so its exact claim is taken from the paper's citation context.
• Haplogroup matching of donor and patient would delay treatment by more than 4 years for rare haplogroups and intra-haplogroup divergence is often as large as between random pairs. (Takeda et al. 2023, EMBO Reports, doi:10.15252/embr.202154540)
  So: Rules out haplogroup as the right unit; the design should test the specific D-loop genotype and sequence divergence, not haplogroup.
• In primate embryos, mtDNA variants partition between blastomeres during preimplantation development and fetuses and stem cells shift toward homoplasmy. (Lee et al. 2012, Cell Reports, doi:10.1016/j.celrep.2012.03.011)
  So: Direct evidence that segregation can happen before implantation, which is the mechanism Arm B is built to see.
• In human embryos carrying pathogenic variants, heteroplasmy segregates evenly among blastomeres and between inner cell mass and trophectoderm, but single trophectoderm cells vary substantially in 1 of 8 embryos. (Rubens et al. 2025, Human Reproduction, doi:10.1093/humrep/deaf050)
  So: Provides the natural-heteroplasmy baseline for the background control and the single-cell noise figure the pilot must beat.
• Human primordial germ cells carry about 5 mtDNA molecules per mitochondrion with selection against non-synonymous, tRNA and D-loop variants. (Floros et al. 2018, Nature Cell Biology, doi:10.1038/s41556-017-0017-8)
  So: A selection mechanism acting on D-loop variants exists in human development, supporting the haplotype arm as biologically plausible.

Searched: Europe PMC searches as listed under followUp plus 'TITLE:("mitochondrial replacement" OR "mitochondrial donation" OR "nuclear transfer") AND (haplogroup OR haplotype) AND (matching OR compatibility OR selection)'; abstracts fetched for Kang 2016, Yamada 2016, Lee 2012, Takeda 2023, Floros 2018, Rubens 2025, Craven 2010; Hudson 2019 could not be read.

Feasibility

The Newcastle Biosciences Institute group under its existing HFEA research licence, or the Monash mitoHOPE laboratory once its research licence covers pronuclear transfer; any embryo-research centre with donated PNT embryos and single-cell mtDNA sequencing. · 12 to 24 months, paced by the accrual of research PNT embryos · moderate, because single-cell mtDNA libraries on dozens of embryos are costly but no new clinical procedure is needed
Blockers: Research PNT embryos accrue only as fast as clinical cycles produce unsuitable embryos; UK rules forbid assaying treatment embryos, so the high cases that matter clinically are never in the research set; Immunosurgery or sorting of inner cell mass from human blastocysts needs a licence amendment if not already covered

Strongest objection

The embryos that reach research are the arrested or poor-quality ones; the two clinical cases were good blastocysts that implanted, and the paper's own data show arrested embryos never exceed 3.9%, so the research set may contain no instance of the phenomenon. The reversion may also happen after implantation, as the Costa-Borges case suggests, in which case no preimplantation experiment can see it.

Not derivable from what is published

• Single-cell heteroplasmy measurement variance at 1 to 5%
• The number of embryos per D-loop class needed to detect a pairing effect
• Whether the existing 62 embryos still have DNA available for whole-mtDNA sequencing

3. Send it to a lab

The authors suggest, without data, that moving the nuclear genome before fertilisation would avoid the low fertilisation they saw in carrier eggs; the design turns that suggestion into a paired preclinical comparison with the comparator, the confounder (egg vitrification) and the carryover readout they left open.

1. Has anyone done this since?

Status

nothing found

Works since

• mitoHOPE Program progress and 2026 pre-clinical research announcement (Monash University mitoHOPE programme pages, 2025 to 2026, https://www.monash.edu/medicine/mitohope/progress)
  The Australian programme has started maternal spindle transfer training on donated eggs under a pre-clinical licence, is seeking a second licence for pronuclear transfer, and expects first pilot-trial participants in the second half of 2027; it plans both techniques but has published no comparison and no results.
• From disease prevention to fertility treatment: rethinking the use of mitochondrial donation for oocyte-related infertility in the light of safety and efficacy (Pasturino et al. 2026, Human Reproduction, doi:10.1093/humrep/deag027)
  An ethics and policy paper on MST and PNT for infertility; it notes the scarcity of clinical data and does not compare the two techniques on fertilisation.

Searched

All 52 citing-works titles; Europe PMC 'TITLE:"maternal spindle transfer" 2017-2026', 'TITLE:"spindle transfer" AND (heteroplasmy OR carryover) 2019-2026', '"pronuclear transfer" AND "spindle transfer" AND human AND fertilization AND oocytes 2013-2026'; web search 'maternal spindle transfer mitochondrial disease live births heteroplasmy children 2025 2026' and 'mitoHOPE Monash mitochondrial donation Australia pronuclear transfer spindle transfer clinical trial'.

Still open

No human study has compared spindle transfer with pronuclear transfer on eggs from the same women, and no spindle-transfer series exists in mtDNA-disease carriers beyond single cases.

Registry query

0 of 49 records in the open node mention any of: spindle transfer, pronuclear transfer, fertilisation rate, oocyte vitrification, mtdna carryover, human oocyte. The open node holds biology and chemistry so far, so this is the expected answer; a private node of a clinical group would be asked the same question.

2. A proposed experiment

Question

Do carrier eggs reconstituted by spindle transfer fertilise normally at the donor-egg rate, where sibling eggs through pronuclear transfer fertilise at the paper's 48.5%?

Hypothesis

Among sibling eggs from one carrier, normal (2PN) fertilisation after spindle transfer exceeds that after the paper's pronuclear-transfer pathway, with maternal mtDNA carryover no higher.

Attempt type

wet-lab
Vocabulary note: A preclinical paired comparison on research-donated human eggs; it is bench work on human gametes under licence, not a trial, so wet-lab is used although the arms are allocated like a trial.

Design

Eggs donated for research by carriers of the variants that fertilised worst in the paper (the paper says fertilisation varied by variant but gives the breakdown only in a supplementary figure) and by unaffected donors, under a research licence. Each carrier's mature eggs are split, by alternation after warming or retrieval, between three arms: spindle transfer into an enucleated donor metaphase II egg followed by ICSI (intracytoplasmic sperm injection); the paper's pronuclear-transfer pathway (ICSI, then transfer at 8 to 13 hours); and unmanipulated ICSI. Because clinical PNT eggs are vitrified while donor eggs are fresh, each carrier arm is run with both vitrified and fresh eggs where numbers allow, so the vitrification effect is separated from the technique effect. Readouts: proportion with exactly 2 pronuclei at 16 to 18 hours (primary), abnormal pronuclear number, blastocyst rate on days 5 and 6, blastocyst quality grade, aneuploidy by trophectoderm biopsy, and maternal mtDNA carryover by high-throughput sequencing. Analysis: fertilisation compared within carrier by a paired test across arms, stratified by vitrified versus fresh.

Controls

• active-comparator Sibling eggs taken through the paper's pronuclear-transfer protocol in the same laboratory.
  Rules out: That any improvement is due to the laboratory or the batch rather than the technique.
  Vocabulary note: The pronuclear-transfer arm is the comparator the issue names, and its known rate (48.5% normal fertilisation, 46.7% day-6 blastocysts) is the benchmark that shows the system is behaving as in the clinic.
• no-treatment Unmanipulated sibling carrier eggs injected by ICSI, fresh and vitrified.
  Rules out: That the carrier eggs' low fertilisation is caused by manipulation at all rather than by the variant or by vitrification.
• reference-case Donor eggs with ICSI only, which fertilised at 63.8% in the paper.
  Rules out: That the donor cytoplasm itself is the limiting factor in a given batch.
• designated Spindle transfer between two donor eggs (reciprocal transfer as in Tachibana 2013) run in parallel.
  Rules out: That the spindle-transfer procedure, which in 2013 gave 52% abnormal fertilisation, is itself failing in this laboratory's hands.

Sensitivity

The paper supplies the two rates that define the difference worth detecting: 48.5% normal fertilisation after the PNT pathway against 63.8% for donor eggs and 66.5% for PGT patient eggs. The number of eggs per arm needed to detect a rise from 48.5% to the mid-60s at a stated power follows from a standard two-proportion calculation that this design does not perform; the researcher must choose the power and the pairing structure. Carryover sensitivity is set by the sequencing assay the paper used (reported detection 0.06 to 0.17%). Tachibana 2013 gives the hazard rate for spindle transfer in human eggs: 52% abnormal fertilisation, so the design must be able to see abnormal pronuclear number as well as the 2PN rate.

Would count as

Supported → positive · opposed → negative-no-effect
Inconclusive if: The reciprocal donor-donor spindle-transfer control fertilises below the Tachibana 2013 rate, showing the technique was not working in this laboratory, or carrier egg numbers per arm fall below the pre-set minimum because eggs from the worst-fertilising variants are scarce.

Known before

• Spindle transfer in 65 research human oocytes fertilised at 73%, like controls, but 52% of spindle-transfer zygotes had an abnormal number of pronuclei; normally fertilised ones reached blastocyst at 62%. (Tachibana et al. 2013, Nature, doi:10.1038/nature11647)
  So: The authors' premise that fertilisation in wild-type cytoplasm is cleaner is not obviously true: spindle transfer has its own fertilisation failure mode that the design must count.
• Nuclear genome transfer between unfertilised oocytes did not reduce blastocyst development provided premature activation was avoided; carryover fell below 1% and became undetectable in stem cell lines. (Paull et al. 2013, Nature, doi:10.1038/nature11800)
  So: Shows that the timing of spindle transfer relative to activation controls both development and carryover, so the protocol must be fixed before the comparison.
• Spindle transfer in eggs from carriers of common mtDNA syndromes produced embryos with over 99% donor mtDNA; some stem cell lines later reverted to the maternal haplotype. (Kang et al. 2016, Nature, doi:10.1038/nature20592)
  So: The only spindle-transfer data on carrier eggs; it reports replacement efficiency but not a fertilisation comparison against pronuclear transfer.
• A live birth after spindle transfer for m.8993T>G, with 5.7% heteroplasmy in the blastocyst and 2.36 to 9.23% in neonatal tissues. (Zhang et al. 2017, Reproductive BioMedicine Online, doi:10.1016/j.rbmo.2017.01.013)
  So: Carryover after clinical spindle transfer can be higher than the under-2% the Newcastle PNT protocol achieves, so carryover must be a co-primary readout, not an afterthought.
• In 28 spindle-transfer cycles for idiopathic infertility, 6 children were born; 5 had over 99% donor mtDNA and 1 rose from 0.8% at blastocyst to 30 to 60% at birth. (Costa-Borges et al. 2023, Fertility and Sterility, doi:10.1016/j.fertnstert.2023.02.008)
  So: The largest human spindle-transfer series; it shows the technique is deliverable at scale and that reversion is not specific to pronuclear transfer.
• In 35,654 sibling donor oocytes, IVF with vitrified oocytes was less efficient than with fresh oocytes, though efficacy per transfer was comparable. (Cornet-Bartolomé et al. 2020, Human Reproduction, doi:10.1093/humrep/deaa178)
  So: Vitrification of carrier eggs, which the PNT pathway requires, is a plausible cause of the low fertilisation on its own; the design stratifies on it for that reason.
• Earlier second polar body transfer with physical removal of residual mitochondria gave near-normal blastocyst rates and further reduced carryover in mice and human oocytes. (Li et al. 2023, MedComm, doi:10.1002/mco2.217)
  So: A third technique exists that also fertilises in donor cytoplasm; a researcher may ask why it is not an arm.

Searched: Europe PMC searches listed under followUp plus 'oocyte vitrification fertilization rate ICSI fresh versus vitrified sibling oocytes' and '"second polar body transfer" AND mitochondrial AND human'; abstracts fetched by DOI for Tachibana 2013, Paull 2013, Kang 2016, Zhang 2017, Costa-Borges 2023, Cornet-Bartolomé 2020, Li 2023, Craven 2010.

Feasibility

A centre holding both a research licence for nuclear transfer in human eggs and access to carrier egg donors: Newcastle, the Monash mitoHOPE laboratory (now training on spindle transfer), Oregon Health and Science University, or Embryotools with a Greek partner clinic. · 18 to 36 months, paced by carrier egg donation · high, because every data point is a human egg from a rare donor and two micromanipulation teams are needed
Blockers: Carrier eggs for research are scarce and compete with treatment; Spindle-transfer competence takes a training period (Monash is in that period now); UK HFEA research licence scope for spindle transfer would need confirming

Strongest objection

The paper's own data argue the low fertilisation tracks the variant, not the cytoplasm the sperm meets: fertilisation is a sperm-egg event in the carrier's egg either way, and in spindle transfer the carrier's spindle still sits in donor cytoplasm only after a manipulation that Tachibana showed doubles abnormal pronuclear number. A specialist would say the better first test is cheaper: compare fresh against vitrified carrier eggs with plain ICSI, which the paper hints at and which needs no nuclear transfer at all.

Not derivable from what is published

• The per-variant fertilisation rates (in Fig. S3C-D, not in the text)
• How many carrier eggs per arm the licence and donors can supply
• The spindle-transfer protocol (timing relative to activation, fusion method) to be fixed before comparison

3. Send it to a lab

Send to a lab

No part of this design can be run as a laboratory job.
The work is micromanipulation of human eggs by a licensed embryologist and cannot be quoted by a lab-automation service; only the downstream carryover sequencing of embryo lysates could be contracted.

The PGT arm biopsies one cell on day 3 and loses 42.4% of its best blastocysts to high heteroplasmy; moving to day-5 trophectoderm biopsy would be gentler on the embryo but the question is whether the biopsy still reports the embryo's load when heteroplasmy is in the 10 to 60% band where the 30% decision is made.

1. Has anyone done this since?

Status

partly addressed

Works since

• Concordant mtDNA heteroplasmy between trophectoderm and inner cell mass supports blastocyst-stage PGT-mt for m.14487T>C (Tofilo et al. 2026, Journal of Assisted Reproduction and Genetics, doi:10.1007/s10815-026-04012-y)
  One carrier, six blastocysts, bimodal heteroplasmy (11.4 to 13.5% and 73.2 to 99.6%); trophectoderm biopsy and the post-warming remainder (trophectoderm, inner cell mass or both) were closely concordant; it covers one variant and no embryo in the 15 to 70% band, and the authors still advise prenatal diagnosis.
• Successful live birth after preimplantation genetic testing for combined nuclear RNASEH2C and mitochondrial MT-TL1 variants (Hu et al. 2026, Journal of Assisted Reproduction and Genetics, doi:10.1007/s10815-026-04041-7)
  A single case of blastocyst-stage combined testing; by title it adds one more data point and does not test agreement across a range.

Searched

All 52 citing-works titles, abstract fetched for Tofilo 2026; Europe PMC '"trophectoderm" AND "inner cell mass" AND mtDNA AND heteroplasmy AND (concordance OR discrepancy)' and 'TITLE:("preimplantation genetic" AND mitochondrial) AND (trophectoderm OR blastocyst) AND heteroplasmy 2014-2026'.

Still open

No study has measured trophectoderm-to-inner-cell-mass agreement in human embryos binned across the whole heteroplasmy range with enough embryos in the 15 to 60% band to say whether the 30% decision could be wrong.

Registry query

0 of 49 records in the open node mention any of: trophectoderm biopsy, inner cell mass, heteroplasmy, pgt-mt, m.3243a>g, limits of agreement. The open node holds biology and chemistry so far, so this is the expected answer; a private node of a clinical group would be asked the same question.

2. A proposed experiment

Question

Across heteroplasmy bands from under 10% to over 60%, how far does a 5 to 10 cell trophectoderm biopsy depart from the inner cell mass, and does it ever cross 30%?

Hypothesis

In blastocysts carrying pathogenic mtDNA variants, a 5 to 10 cell trophectoderm biopsy is within 10 percentage points of the inner cell mass in 95% of embryos in every band, including 15 to 60%.

Attempt type

wet-lab
Vocabulary note: This is a benchmarking of a sampling method, so evaluation also fits; wet-lab is used because every measurement is on human embryo cells.

Design

Embryos from PGT cycles for mtDNA variants that are unsuitable for transfer (the paper's series discarded 42.4% of good blastocysts for heteroplasmy over 30%, so high-band embryos exist) and are donated to research, from several variants including m.3243A>G, m.8344A>G and MT-ATP6 variants. Each embryo already has a day-3 single-blastomere value. On day 5 or 6, take a standard 5 to 10 cell trophectoderm biopsy, then a second trophectoderm biopsy from the opposite pole, then separate the inner cell mass; measure heteroplasmy in each by the paper's pyrosequencing and high-throughput sequencing, in duplicate. Pre-define bands by day-3 value: under 10%, 10 to 30%, 30 to 60%, over 60%, and recruit until each band holds a pre-set number of embryos. Analysis: Bland-Altman limits of agreement between biopsy and inner cell mass per band; the fraction of embryos where biopsy and inner cell mass fall on opposite sides of 30%; and agreement between the two biopsies as the sampling floor. Report by variant.

Controls

• within-subject Two trophectoderm biopsies from opposite poles of the same blastocyst, and the inner cell mass from the same embryo.
  Rules out: That disagreement between biopsy and inner cell mass is just the biopsy-to-biopsy sampling noise.
• reference-case The day-3 single-blastomere value already obtained for each embryo in the clinical cycle.
  Rules out: That the blastocyst-stage measurement is inconsistent with the stage currently used for decisions.
• designated Dilution series of mutant and wild-type mtDNA at 1, 3, 10, 30, 60 and 90% run in each batch, and 5-cell and 10-cell pools of a homogeneous cell line.
  Rules out: That assay non-linearity or small-input amplification bias produces the band-dependent disagreement.
• instrument-noise-floor Duplicate library preparations from each biopsy lysate.
  Rules out: That a difference smaller than technical repeatability is reported as biological discordance.

Sensitivity

The decision threshold is 30%, so the quantity that matters is the limits of agreement in the 10 to 60% band. Published figures that bound it: Treff 2012 found trophectoderm predicted inner cell mass with r squared 0.97 in one carrier's embryos; Heindryckx 2014 found r 0.90 among trophectoderm cells; in a heteroplasmic mouse model Neupane 2014 found r squared 0.90 but a 95% prediction interval of 23.58 percentage points between trophectoderm and whole blastocyst, which would cross a 30% threshold from either side; Rubens 2025 found substantial single-cell variation in 1 of 8 embryos and set 5 to 10 cells as the minimum. The number of embryos per band needed to estimate limits of agreement to a stated precision cannot be set from these because none reports a human per-band variance; a pilot of 10 embryos per band is the first step.

Would count as

Supported → positive · opposed → negative-no-effect
Inconclusive if: The 30 to 60% band cannot be filled because clinics rarely culture such embryos to blastocyst, or the two-biopsy sampling floor is as wide as the biopsy-to-inner-cell-mass difference, so the method cannot be separated from its own noise.

Known before

• Trophectoderm mutation load predicted inner cell mass (r squared 0.97) across embryos ranging 7 to 91% for m.3243A>G; a 12% biopsy led to a boy with tissue loads from undetectable to 15%. (Treff et al. 2012, Fertility and Sterility, doi:10.1016/j.fertnstert.2012.07.1119)
  So: The strongest human concordance figure, but from one carrier, and the paper cites it as the case where biopsy and infant diverged.
• Limitations of preimplantation genetic diagnosis for mitochondrial DNA diseases, commenting on the Treff case as a clinically relevant discrepancy. (Mitalipov et al. 2014, Cell Reports, doi:10.1016/j.celrep.2014.05.004)
  So: Frames the issue as the paper does: the infant can differ from the biopsy, which this design cannot test but can bound at the embryo level.
• A mutation-free baby after trophectoderm biopsy for m.3243A>G; correlation r 0.90 among trophectoderm cells and r 0.95 between cytoplasmic fragments and trophectoderm. (Heindryckx et al. 2014, Mitochondrion, doi:10.1016/j.mito.2014.08.005)
  So: Shows within-trophectoderm agreement is high in a low-heteroplasmy case; the design needs the high bands.
• In a heteroplasmic mouse model, trophectoderm against whole blastocyst gave r squared 0.90 with a 95% prediction interval of 23.58 percentage points; polar bodies were unreliable. (Neupane et al. 2014, Human Reproduction, doi:10.1093/humrep/deu016)
  So: A prediction interval that wide at the 30% threshold is the quantitative reason the question is open; the design reports the same statistic in humans.
• In 112 embryos carrying pathogenic variants, heteroplasmy segregated evenly among day-4 blastomeres and between inner cell mass and trophectoderm, but single trophectoderm cells varied substantially in 1 of 8 control embryos, so at least 5 to 10 cells are needed. (Rubens et al. 2025, Human Reproduction, doi:10.1093/humrep/deaf050)
  So: Sets the biopsy size and gives the closest existing answer, though its single-cell work was on control embryos with polymorphisms rather than pathogenic variants.
• A single day-3 blastomere gave a false-negative at a 15% threshold in 1 of 294 blastomeres from 73 m.3243A>G embryos. (Sallevelt et al. 2017, Journal of Medical Genetics, doi:10.1136/jmedgenet-2017-104633)
  So: The error rate of the current method is 0.34%; a trophectoderm protocol must be shown to be no worse against the same kind of threshold.

Searched: Europe PMC searches as under followUp plus abstracts fetched by DOI for Treff 2012, Mitalipov 2014, Heindryckx 2014, Neupane 2014, Rubens 2025, Sallevelt 2017, Spath 2021, Monnot 2011; searched for and did not find any human study binning trophectoderm-to-inner-cell-mass agreement by heteroplasmy band.

Feasibility

A PGT centre with an mtDNA programme and research embryo donation: Newcastle, Maastricht (Sallevelt), Paris Necker (Rubens, Monnot), Ghent (Heindryckx), or the Chinese centres publishing PGT-mt cases. · 12 to 24 months to fill the middle bands · low to moderate, because the embryos are already discarded and the assays already run in these laboratories
Blockers: Clinics may not culture high-heteroplasmy day-3 embryos to blastocyst, emptying the bands that matter; Research consent for discarded PGT embryos; Inner cell mass isolation needs skill and may be incomplete, contaminating with trophectoderm

Strongest objection

Even perfect trophectoderm-to-inner-cell-mass agreement does not answer the clinical question, because the documented discrepancy is between embryo and child, and heteroplasmy shifts after implantation (Monnot 2011; the Tofilo authors still advise prenatal diagnosis). A specialist would say the design measures the part of the problem that is measurable and should be labelled as such, and that the day-3 to day-5 shift inside the same embryo may be the larger source of error.

Not derivable from what is published

• Per-band variance of biopsy-to-inner-cell-mass difference in human embryos, needed to size each band
• How many high-band embryos the Newcastle series still holds in storage and whether they are consented for research
• Whether inner cell mass isolation without trophectoderm contamination is achievable at the required purity

3. Send it to a lab