Mitochondrial Heteroplasmy Interpreter
Mitochondrial Heteroplasmy Interpreter
A heteroplasmy percentage means almost nothing without the variant it belongs to and the tissue it was measured in. Blood under-represents the burden for several common variants and falls with age; the threshold effect differs from one variant to the next; and for some variants no validated threshold exists at all, which this page says rather than supplying one.
Heteroplasmy, by variant and by tissue
Variant + tissue + level → readingA 42-year-old woman with sensorineural deafness and diabetes, in whom m.3243A>G was found in blood at a level of 12%
The two pieces of arithmetic on this page, and their limits
threshold for biochemical deficiency ≈ 60% for a large deletion, up to above 90% for some single-nucleotide variants
- 0.977
- the compound annual factor by which leucocyte m.3243A>G heteroplasmy declines — about 2.3% a year. Other series put the decline at 1.4 to 2.3% a year, and one longitudinal cohort measured it as 0.7 (± 0.4) percentage points a year
- age + 12
- the exponent as published. It normalises to a fixed reference point rather than to birth, which is why the adjustment raises even a sample taken at age 0 by about a third
- what it produces
- a number that can exceed 100%. A measured 40% at age 55 adjusts to about 190%. That is a property of extrapolating a log-linear fit, not a finding about the patient, and it is the reason the page flags it rather than printing it
- the threshold effect
- mtDNA variants are functionally recessive: a cell tolerates a proportion of mutant genomes with no detectable biochemical consequence, and the deficiency appears once that proportion is exceeded. The proportion is variant-specific, tissue-specific and cell-specific, which is why a whole-tissue percentage is a weak predictor even when a threshold is known
- what no formula gives
- the level in the tissue you did not sample, and the level a child will inherit. The mitochondrial genetic bottleneck makes transmission a lottery whose distribution is wide and whose mean is not the mother’s level
Worked example
A 42-year-old woman with sensorineural deafness and diabetes, in whom m.3243A>G was found in blood at a level of 12%
The variant is m.3243A>G and the tissue is blood, so the tissue-specific reading applies
12% is above the near-undetectable range, so the variant is established and the question is what the level means
The published adjustment: 0.977 raised to the power of (42 + 12) = 0.97754 = 0.2846
Age-adjusted blood level = 12 ÷ 0.2846 = 42.2% — below 100, so the adjustment is still inside the range where it means something
That figure is for comparing this woman with another carrier measured at a different age. It is not a prediction, and no validated heteroplasmy threshold exists for this variant
If the question were whether the variant is present at all — screening a maternal relative, for instance — blood would be the wrong tissue and urinary epithelial cells the right one
Change only the age, to 62. The divisor becomes 0.97774 = 0.1787, the adjusted level becomes 67.1%, and the same 12% now reads very differently. Nothing about the woman changed; the adjustment is doing all the work, which is exactly why the tissue and the age belong on the report next to the number
The same person, three tissues
| Tissue | Typical m.3243A>G level in one reported cohort | What it is good for |
|---|---|---|
| Blood — leucocytes | 14% | Poorest for detection; best single predictor of disease burden once age-adjusted |
| Buccal cells or saliva | 36% | A usable non-invasive sample when urine is not available |
| Urinary epithelial cells | 51% | The best non-invasive tissue for establishing that the variant is present |
| Skeletal muscle | Highest and most stable | The reference tissue, and the one taken when histochemistry and enzymology are wanted too |
Thresholds that are published, and thresholds that are not
| Variant | Published threshold | Tissue behaviour |
|---|---|---|
| m.8993T>G | Below 60% usually asymptomatic or mild; 70 to 90% NARP; above 90% Leigh syndrome spectrum | Relatively uniform across tissues, which makes blood representative and prenatal testing more tractable |
| m.8993T>C | Above 90% in generally all symptomatic individuals — but some asymptomatic carriers are also above 90% | As for m.8993T>G |
| Single large-scale deletion | Biochemical threshold around 60% at cell level, but focal accumulation makes a whole-tissue figure a weak guide | Often undetectable in blood except in Pearson syndrome; muscle or urine required |
| m.3243A>G | None validated. Leucocyte level correlates only weakly with severity over time | Lowest in blood and falling with age; highest in muscle and urine |
| m.8344A>G | None validated. Near-homoplasmic levels have produced Leigh syndrome rather than MERRF | As for mtDNA disease generally |
| m.1555A>G | Not applicable — essentially always homoplasmic | The actionable finding is the aminoglycoside contraindication, not a level |
Prenatal and preimplantation figures, quoted as such
| Setting | Published figure |
|---|---|
| Prenatal diagnosis | A fetal heteroplasmy below 30% is generally treated as low risk and above 60% as high risk |
| Preimplantation genetic testing | A cut-off of 18% “was safe for most mtDNA mutations, although m.3243A>G was an exception where a lower level (<15%) may be needed” |
| What neither can do | Predict the level in tissues that were not sampled, or the level after birth. Chorionic villus and amniocyte levels are measurements of those tissues |
Why the percentage on the report is the least informative part of it
Mitochondrial DNA is present in hundreds to thousands of copies per cell, and a pathogenic variant is usually present in some of them and not others. That proportion is the heteroplasmy. Two features of it defeat the natural reading of a percentage on a report, and both are properties of the biology rather than of the assay.
The first is the threshold effect. mtDNA variants are functionally recessive: a cell tolerates a substantial burden of mutant genomes with no detectable biochemical consequence, and the deficiency appears only once a proportion is exceeded. Reviews put that proportion at about 60% for single large-scale deletions and at over 90% for some single-nucleotide variants — a range so wide that a level meaning nothing for one variant is well past the line for another. Below the threshold the relationship between level and disease is close to flat; above it, it is steep. So a heteroplasmy figure is not a dose, and treating a rise from 30 to 40% as though it were 33% worse is a misreading of the shape of the curve.
The second is that the proportion differs between tissues in the same person, sometimes by several fold. Dividing cells lose some variants over time while post-mitotic tissues retain them, so leucocytes generally give the lowest figure of any tissue sampled. One cohort of m.3243A>G carriers reported 14% in leucocytes, 36% in buccal saliva and 51% in urothelium in the same people. A report that says “heteroplasmy 12%” without naming the tissue has not told you very much, and two reports on the same patient that appear to contradict each other usually do not.
m.3243A>G makes both points sharply and is worth understanding in its own right, because it is the commonest pathogenic mtDNA variant. Leucocyte levels fall at a compound rate of roughly 2.3% a year, so the same person measured at 20 and at 60 gives markedly different numbers, and an older asymptomatic maternal relative is exactly the person in whom a blood test is likely to miss it. A published adjustment exists to put samples taken at different ages on a common footing — dividing the measured level by 0.977 raised to the power of the age plus twelve — and it is worth applying, with one caution this page enforces: at older ages the divisor is small enough that a moderate measured level extrapolates past 100%, which is a property of a log-linear fit rather than a finding about the patient.
Here the received wisdom needs correcting. It is usually said that blood is simply the wrong tissue for m.3243A>G and urine or muscle the right one. The first half is true for DETECTION: urinary epithelial cells hold much higher levels and are the non-invasive tissue of choice for establishing that the variant is there. But the largest cohort to measure all three found that, once age-adjusted, blood was “the most highly correlated mutation measure for disease burden and progression”, explaining about 27% of the variance in disease burden on its own and about 40% together with muscle mtDNA copy number. The two questions have different answers: use urine to find the variant, and age-adjusted blood to estimate how much disease to expect. And note what 40% of the variance means — most of the difference between two carriers with the same level is still unexplained, which is why no validated heteroplasmy threshold predicts phenotype for this variant and why this page prints none.
One thing that follows from none of the above, and matters more than any of it. Whatever a woman’s own level is, in whatever tissue, it does not predict her child’s. The mitochondrial genetic bottleneck during oogenesis means the level transmitted can be far higher or far lower than hers, and the spread is wide. Reproductive counselling follows from the variant being present, not from how much of it there is, and it belongs in a specialist mitochondrial service. Related pages: the ACMG variant classification interpreter for classifying the variant itself — note that the nuclear-gene framework needs mitochondrial-specific modification — and the NGS variant confirmation interpreter, since mitochondrial variants are named in the AMP and NSGC report as a special case for orthogonal confirmation, and a low-level heteroplasmy near the assay’s limit of detection is exactly the call that needs it.
Frequently asked questions
What heteroplasmy level causes symptoms?
It depends entirely on the variant, and for several common variants no validated threshold exists. Reviews give the general range as about 60% of mutant genomes for single large-scale mtDNA deletions up to over 90% for some single-nucleotide variants. Where variant-specific figures are published they are specific: for m.8993T>G, below 60% is usually asymptomatic or mild, 70 to 90% is associated with NARP and above 90% with Leigh syndrome spectrum; for m.8993T>C, symptomatic individuals generally exceed 90%. For m.3243A>G and m.8344A>G there is no validated threshold, and this page prints none rather than borrowing one from another variant.
Why is blood the wrong tissue for mitochondrial testing?
Because dividing cells lose some pathogenic mtDNA variants over time while post-mitotic tissues retain them, so leucocytes usually give the lowest heteroplasmy of any tissue — and for m.3243A>G the level falls at a compound rate of roughly 2.3% a year, so the older the person the more likely blood is to miss it. A single large-scale deletion is frequently undetectable in blood altogether outside Pearson syndrome. Urinary epithelial cells are the usual non-invasive alternative and skeletal muscle the reference tissue. The exception worth knowing is that for estimating how much disease an m.3243A>G carrier has, age-adjusted blood outperformed both urine and muscle in the largest cohort to compare them — so blood is the worst tissue for finding the variant and the best for quantifying its consequences.
How is age-adjusted blood heteroplasmy calculated for m.3243A>G?
Divide the measured blood heteroplasmy by 0.977 raised to the power of the age in years plus twelve. A 12% level at age 42 becomes 12 divided by 0.977 to the power of 54, which is 12 ÷ 0.2846 = 42.2%. The purpose is to make levels measured at different ages comparable, because leucocyte heteroplasmy declines at a compound rate of about 2.3% a year. Two limits: it applies to m.3243A>G in blood and has no established role for other variants or tissues, and at older ages it can produce a figure above 100%, which is an artefact of extrapolating a log-linear fit and should not be reported as a heteroplasmy.
Does a mother’s heteroplasmy level predict her child’s?
No. During oogenesis only a small sample of the mother’s mitochondrial genomes is passed on, a process called the genetic bottleneck, and the level in a child can be far higher or far lower than the mother’s. The distribution is wide and its shape differs between variants. That is why reproductive counselling follows from the presence of the variant rather than from the mother’s level, and why prenatal and preimplantation testing measure the fetal or embryonic level directly. Those decisions belong in a specialist mitochondrial genetics service with genetic counselling attached.
What does a negative mitochondrial DNA test in blood mean?
Much less than it appears to. For m.3243A>G the variant may be undetectable in leucocytes and detectable only in buccal mucosa, cultured fibroblasts, hair follicles, urinary sediment or, most reliably, skeletal muscle. For a single large-scale deletion, blood is negative in most patients with Kearns-Sayre syndrome or progressive external ophthalmoplegia. If clinical suspicion remains, test urinary epithelial cells or muscle. Ask also what the laboratory’s limit of detection was, because fragment analysis and restriction methods often cannot see below a few per cent while next-generation sequencing and droplet digital PCR reach much lower — so “not detected” means different things from different laboratories.
Why does the m.1555A>G result not give a meaningful percentage?
Because m.1555A>G is essentially always homoplasmic — present in effectively all mitochondrial genomes — so there is no heteroplasmy level to interpret and no threshold to apply. The finding that matters is a contraindication: carriers can suffer severe, permanent bilateral sensorineural hearing loss after a single dose of an aminoglycoside, regardless of dose, peak level or duration. Record the contraindication where it will be seen in an emergency, give the patient something they can show, and test the maternal relatives, all of whom carry it.
Related calculators
References
- Grady JP, et al. mtDNA heteroplasmy level and copy number indicate disease burden in m.3243A>G mitochondrial disease. EMBO Mol Med. 2018. doi:10.15252/emmm.201708262. Blood (n=231), urine (n=235) and skeletal muscle (n=77); “Age-adjusted blood level = (Blood heteroplasmy) / 0.977^(age + 12)”; “Blood heteroplasmy declines by ~2.3%/year”; “blood is the most highly correlated mutation measure for disease burden and progression”, R² = 0.27 alone and 0.40 with muscle mtDNA copy number.
- Mitochondrial DNA-Associated Leigh Syndrome Spectrum. In: GeneReviews. Seattle: University of Washington; NBK1173. m.8993T>G: below 60% “usually asymptomatic or have only mild pigmentary retinopathy or migraine headaches”, 70–90% NARP, above 90% “typically present with LSS”. m.8993T>C: “generally all symptomatic individuals with m.8993T>C have a heteroplasmy level greater than 90%”.
- MELAS. In: GeneReviews. Seattle: University of Washington; NBK1233. The variant “may be undetectable in mtDNA from leukocytes and may be detected only in other tissues, such as buccal mucosa, cultured skin fibroblasts, hair follicles, urinary sediment, or (most reliably) skeletal muscle”; and “correlations between the frequency of the more common clinical features and the level of mutated mtDNA in muscle, but not in leukocytes, have been observed”.
- Mitochondrial DNA disorders: from pathogenic variants to preventing transmission. PMC8490015. “mtDNA variants are functionally recessive with threshold of mutated to wild-type mtDNA of about 60% for single, large-scale mtDNA deletions up to over 90% for some single-nucleotide variants”; the steep early fall of m.3243A>G in blood; and the need to test muscle or uroepithelial cells for deletions.
- Pathogenic mitochondrial DNA 3243A>G mutation: from genetics to phenotype. Front Genet. 2022;13:951185. doi:10.3389/fgene.2022.951185. Heteroplasmy by tissue in one cohort — “buccal saliva (36%), leucocytes (14%), Urothelium (51%)”; decline in leucocytes of “around 1.4%–2.3% per year”; the Grady age adjustment; de Laat’s six-year follow-up of 151 carriers finding leucocyte levels “only weakly correlated with the severity of the disease”; fetal levels “below 30% … low risk, and those above 60% … high risk” (Steffann 2021); and a preimplantation cut-off of 18% “safe for most mtDNA mutations, although m.3243A>G was an exception where a lower level (<15%) may be needed”.
- de Laat P, et al. Clinical features and heteroplasmy in blood, urine and saliva in 34 Dutch families carrying the m.3243A>G mutation. J Inherit Metab Dis. 2012. doi:10.1007/s10545-012-9465-2.
- Prezant TR, et al. Mitochondrial ribosomal RNA mutation associated with both antibiotic-induced and non-syndromic deafness. Nat Genet. 1993. m.1555A>G and the aminoglycoside contraindication.
Medical Disclaimer: The tools and content provided here are for educational and reference purposes only. They are not intended to substitute for professional medical advice, diagnosis, or treatment. Clinical decisions should always be based on the comprehensive assessment of a qualified healthcare professional.