Microsatellite Instability (MSI) Interpreter
Microsatellite Instability (MSI) Interpreter
Read an MSI-PCR result and a mismatch repair immunohistochemistry panel together — the five Bethesda loci, the two obligate heterodimers, the reflex testing MLH1 loss demands, and what to do when the two methods disagree.
MSI and mismatch repair
PCR loci + MMR IHC → interpretationMLH1 and PMS2 both lost on immunohistochemistry, three of five loci unstable by PCR, no reflex testing yet
The two methods, and what each one measures
≥ 2 of 5 unstable = MSI-H · 1 of 5 = MSI-L · 0 of 5 = MSS
MMR IHC — four proteins, two obligate heterodimers: MLH1–PMS2 and MSH2–MSH6
NGS — the same question asked of tens to thousands of microsatellite loci, scored as the fraction unstable
- the Bethesda panel
- the original five-locus reference panel: two mononucleotide repeats, BAT-25 and BAT-26, and three dinucleotide repeats, D2S123, D5S346 and D17S250. Two or more unstable markers defines MSI-high. It has largely been displaced by mononucleotide-only pentaplex panels — BAT-25, BAT-26, NR-21, NR-24 and NR-27 — whose markers are quasi-monomorphic, so a matched normal sample is not always needed
- the heterodimers
- MLH1 pairs with PMS2, MSH2 pairs with MSH6, and MLH1 and MSH2 are the dominant partners. Inactivating a dominant partner destabilises its junior partner, so both disappear together; inactivating a junior partner leaves the dominant one in place. That asymmetry is what makes four stains more informative than two
- why MLH1 loss is different
- THE POINT OF THE PAGE: every other loss pattern points at a germline gene, but MLH1 loss usually does not. About 10 to 15% of colorectal cancers are mismatch repair deficient through somatic methylation of the MLH1 promoter, which is sporadic. BRAF V600E and MLH1 promoter methylation are the tests that separate the two, and neither is optional before a referral decision
- discordance
- the two methods disagree in a small but real minority of cases — 1.7% in one 516-case series, and published rates from under 1% to over 20% depending on the cancer type and the panel. A discordant pair is a finding, not a technical failure to be explained away, and the recommendation from the data is to run both methods where the answer matters
- the NGS version
- sequencing-based callers interrogate far more than five loci — from a few dozen up to several thousand, with more than 11,000 usable microsatellites identifiable from whole-exome data — and score the fraction that are unstable rather than counting to five. Performance plateaus surprisingly early: one widely used tool works with as few as 50 sites and its accuracy plateaus at the top 20
Worked example
MLH1 and PMS2 both lost on immunohistochemistry, three of five loci unstable by PCR, no reflex testing yet
The PCR result is unambiguous: three of five unstable markers is MSI-high, since two or more defines it
The immunohistochemistry agrees, and adds what PCR cannot — which protein has gone. MLH1 is the dominant partner of the MLH1–PMS2 heterodimer, so its loss destabilises PMS2 and both stains go negative together
That pair is concordant, so no discordance rule fires and the interpretation turns on what happens next
And this is the one pattern that cannot be referred as it stands. About 10 to 15% of colorectal cancers are mismatch repair deficient because the MLH1 promoter has been somatically methylated — sporadic, not inherited, and indistinguishable from Lynch syndrome on the slide
Reflex testing separates them. Detecting BRAF V600E would argue strongly for the sporadic route; 69% of MLH1-loss cases carry it, and it is essentially absent in germline MLH1 carriers. Detecting MLH1 promoter methylation would do the same; 86.5% of MLH1-loss cases are methylated
Neither finding, and the result becomes a referral: BRAF wild type with no methylation leaves an inherited cause unexcluded. Note the order matters practically — BRAF alone, used as the sole discriminator, generates about 2.3 times as many genetics referrals as methylation testing does, because a negative BRAF result predicts methylation poorly
The five Bethesda loci and what the count means
| Detail | |
|---|---|
| Mononucleotide markers | BAT-25 and BAT-26 |
| Dinucleotide markers | D2S123, D5S346 and D17S250 |
| 0 of 5 unstable | Microsatellite stable (MSS) |
| 1 of 5 unstable | MSI-low (MSI-L) — conventionally grouped with MSS |
| 2 or more of 5 unstable | MSI-high (MSI-H) |
| If more than five markers are tested | 30 to 40% or more of markers unstable defines MSI-high |
| The panel most laboratories now run | A mononucleotide pentaplex — BAT-25, BAT-26, NR-21, NR-24, NR-27 — whose quasi-monomorphic markers reduce the need for a matched normal sample |
The four proteins, the two pairs, and what each loss pattern means
| IHC pattern | Why the pattern looks like that | What it implies | Next step |
|---|---|---|---|
| MLH1 and PMS2 both lost | MLH1 is the dominant partner; losing it destabilises PMS2 | Sporadic MLH1 promoter methylation in most cases; germline MLH1 in a minority | Reflex BRAF V600E and/or MLH1 promoter methylation before any referral decision |
| MSH2 and MSH6 both lost | MSH2 is the dominant partner; losing it destabilises MSH6 | Germline MSH2, or a 3′ EPCAM deletion silencing the MSH2 promoter. No common sporadic cause | Refer. Make sure EPCAM is covered as well as MSH2 |
| PMS2 lost, MLH1 retained | PMS2 is the junior partner; its loss does not affect MLH1 | Germline PMS2 | Refer, and tell the laboratory — PMS2 analysis is complicated by neighbouring pseudogenes |
| MSH6 lost, MSH2 retained | MSH6 is the junior partner; its loss does not affect MSH2 | Germline MSH6 | Refer. Expect a weaker PCR instability signal: MSH6 loss destabilises mononucleotide repeats preferentially |
| All four retained | No pair has been destabilised | Mismatch repair proficient — good evidence, not proof, since a missense variant can leave a detectable but dead protein | Consider PCR or NGS-based MSI if the clinical suspicion is real |
| Equivocal or patchy | Usually technical: internal controls, fixation, a thin or treated block | Nothing, until it is repeated | Repeat on a better block; run the PCR, which fails differently |
Reflex testing after MLH1 and PMS2 loss
| Finding | Frequency among cases with MLH1 loss | What it means |
|---|---|---|
| BRAF V600E present | 69% (95% CI 60.1–76.8%) | Sporadic. BRAF V600E is not found alongside a germline MLH1 variant, so it argues strongly against Lynch syndrome |
| MLH1 promoter methylation present | 86.5% (95% CI 79–91.7%) | Sporadic silencing. Covers more cases than BRAF does, which is why it is the more complete discriminator |
| BRAF absent | — | Weak evidence. BRAF has an excellent positive predictive value for MLH1 methylation and a poor negative one, so a wild-type result does not establish an inherited cause |
| Neither present | — | Refer. An inherited cause has not been excluded; germline MLH1 testing, and where that is negative, somatic testing for two hits in the tumour |
| Using BRAF alone as the sole test | — | Generates about 2.3 times as many genetics referrals as methylation testing alone. Testing BRAF first and reserving methylation for the wild-type cases is the efficient sequence |
Two tests, one question, and the case where they part company
Microsatellite instability is what happens downstream of mismatch repair failure. Short repeated sequences slip during replication, the repair machinery normally corrects the slippage, and when it cannot, the repeats accumulate length changes that are visible as new alleles on electrophoresis. The reference definition is the Bethesda panel: five loci, two mononucleotide and three dinucleotide, with two or more unstable defining MSI-high, one MSI-low, and none stable. In practice most laboratories now run a mononucleotide-only pentaplex whose markers are quasi-monomorphic, which removes the need for a matched normal sample in most cases, and sequencing-based callers interrogate anything from a few dozen to several thousand loci and report the fraction unstable rather than counting to five. All three are called MSI testing and they are not the same assay.
Immunohistochemistry asks the question from the other end: is the protein there? Four stains cover two obligate heterodimers, MLH1 with PMS2 and MSH2 with MSH6, and the crucial structural fact is that the pairing is asymmetric. MLH1 and MSH2 are the dominant partners: inactivate one and its junior partner is destabilised and disappears too. Inactivate a junior partner — PMS2 or MSH6 — and the dominant partner stays put, because it has other binding options. So loss comes in four informative patterns, and each points at a different gene. That is the advantage immunohistochemistry has over PCR: instability tells you that repair has failed, staining tells you which protein failed, and it is the protein that decides which gene is sequenced and which relatives are offered testing.
One pattern breaks the rule, and it is the commonest one. Loss of MLH1 and PMS2 usually does not mean Lynch syndrome. Around 10 to 15% of colorectal cancers are mismatch repair deficient because the MLH1 promoter has been somatically methylated — an acquired, sporadic event that produces a slide identical to the inherited disease. Two reflex tests separate them, and they are not equivalent. BRAF V600E is found in 69% of cases with MLH1 loss and essentially never alongside a germline MLH1 variant, so finding it rules the sporadic route in firmly; not finding it rules very little out, because its negative predictive value for methylation is poor. MLH1 promoter methylation is present in 86.5% of MLH1-loss cases and is the more complete discriminator. Using BRAF as the sole test generates roughly 2.3 times as many genetics referrals as methylation testing would, which is why testing BRAF first and reserving methylation for the wild-type cases is the sequence that gets recommended. An MLH1-loss result released without either test has not answered the question it appears to answer.
The last point is the one most often mishandled. The two methods do not always agree, and a discordant result is a finding rather than an error to be explained away. In a series of 516 cancers the methods agreed in 98.3%, with nine discordant cases — but that concordance was 99.3% in gastrointestinal cancers and 92.3% elsewhere, and published discordance rates across tumour types and panels run from under 1% to 23.8%. The mechanisms are real: a missense variant can leave an antigenic but catalytically dead protein that stains perfectly while microsatellites destabilise; MSH6 deficiency destabilises mononucleotide repeats preferentially and reads weakly on a panel containing dinucleotides; loss can be subclonal; and low tumour content or poor fixation degrade staining without touching the PCR. The conclusion drawn by the authors of that series is the practical one: where the answer matters — a treatment indication, a referral to genetics — run both, report both, and let the clinical question rather than a preference between methods decide what happens next.
Frequently asked questions
What are the five loci of the Bethesda panel?
Two mononucleotide repeats, BAT-25 and BAT-26, and three dinucleotide repeats, D2S123, D5S346 and D17S250. Two or more unstable markers defines MSI-high, one is MSI-low and none is microsatellite stable. Most laboratories now use a mononucleotide-only pentaplex — BAT-25, BAT-26, NR-21, NR-24 and NR-27 — instead.
Why do MMR proteins disappear in pairs?
Because they work as obligate heterodimers and the pairing is asymmetric. MLH1 pairs with PMS2 and MSH2 with MSH6, with MLH1 and MSH2 the dominant partners. Losing a dominant partner destabilises the junior one so both stains go negative; losing a junior partner leaves the dominant one standing, which is why isolated PMS2 and isolated MSH6 loss are informative patterns.
Why does MLH1 loss need BRAF or methylation testing?
Because most MLH1 loss is sporadic, not inherited. Around 10 to 15% of colorectal cancers are mismatch repair deficient through somatic MLH1 promoter methylation, which looks identical on the slide. BRAF V600E is present in 69% of MLH1-loss cases and effectively excludes a germline MLH1 variant; MLH1 promoter methylation is present in 86.5% and is the more complete test.
Is BRAF V600E testing enough on its own after MLH1 loss?
It rules sporadic disease in well and rules it out badly. BRAF has an excellent positive predictive value for MLH1 promoter methylation and a poor negative one, so a wild-type BRAF result leaves many sporadic cases unexplained. Used alone it generates about 2.3 times as many genetics referrals as methylation testing; testing BRAF first and reflexing methylation only in wild-type cases is more efficient.
What should be done when IHC and MSI-PCR disagree?
Report both and treat the discordance as real. One 516-case series found 98.3% concordance overall — 99.3% in gastrointestinal cancers, 92.3% elsewhere — and published discordance runs from under 1% to 23.8% depending on tumour type and panel. Missense variants leaving a dead but stainable protein, subclonal loss, MSH6 deficiency and fixation problems all produce genuine disagreement.
How does NGS-based MSI testing differ from PCR?
It uses far more loci. Sequencing-based callers score the fraction of unstable microsatellites across tens to thousands of sites — more than 11,000 are usable from whole-exome data — rather than counting unstable markers to five. Performance plateaus early: one widely used caller works from as few as 50 sites and its accuracy plateaus at the top 20.
Related calculators
References
- Baudrin LG, Deleuze JF, How-Kit A. Molecular and computational methods for the detection of microsatellite instability in cancer. Front Oncol. 2018;8:621.
- Wu EY, Resnick M. Predictive and prognostic implications of microsatellite instability and mismatch repair deficiency in carcinomas of the gastrointestinal tract, liver, and pancreas. Precis Cancer Med. 2019;2:5168.
- Adar T, Rodgers LH, Shannon KM, et al. A tailored approach to BRAF and MLH1 methylation testing in a universal screening program for Lynch syndrome. Mod Pathol. 2017;30(3):440–447.
- Matsubayashi H, Oishi T, Sasaki K, et al. Discordance of microsatellite instability and mismatch repair immunochemistry occurs depending on the cancer type. Hum Pathol. 2023;135:54–64.
- Nádorvári ML, Lotz G, Kulka J, et al. Microsatellite instability and mismatch repair protein deficiency: equal predictive markers? Pathol Oncol Res. 2024;30:1611719.
- Jia P, Yang X, Guo L, et al. MSIsensor-pro: fast, accurate, and matched-normal-sample-free detection of microsatellite instability. Genomics Proteomics Bioinformatics. 2020;18(1):65–71.
- Idos G, Valle L. Lynch syndrome. In: Adam MP, Feldman J, Mirzaa GM, et al, eds. GeneReviews. University of Washington, Seattle — on tumour testing and the reflex algorithm.
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.