HOHorseBmoderate
In 392 Quarter Horses, 57 per cent of the healthy horses carried a P variant: what the two assessments of P2, P3 and P4 actually found.
Executive summary
Two assessments have set the commercially offered P2, P3 and P4 variants against the diagnosis made under the microscope. In neither was any variant linked to the disease. Among 392 Quarter Horses, 57 per cent of the healthy horses carried at least one of them, and 61 per cent of the affected horses. The authors advise explicitly against using these genotypes for breeding, pre-purchase examination or diagnosis. PSSM1, with its GYS1 variant, is a separate matter.
Key points
Polysaccharide storage myopathy describes a muscle disease in which abnormal storage material gathers inside the muscle fibre, material that the enzyme amylase does not fully break down. In 2008, a study of 831 horses from 36 breeds, all with a diagnosis made on a muscle section, identified a variant in the gene for glycogen synthase. This GYS1 variant was found in 17 breeds. It was present in 87 per cent of the marked cases among draught breeds and in 72 per cent of Quarter Horse related breeds, but in only 18 per cent of Warmbloods.
That left a residual group: horses with an abnormal muscle sample and without the GYS1 variant. The paper of the time offered two possible explanations, either a false positive reading of the section or a second, still unknown storage disease. The term PSSM2 grew out of that residual group. From the outset it has denoted no proven genetic defect, but whatever remains under the microscope once PSSM1 has been ruled out.
It means that a muscle sample showed storage material and that the GYS1 variant was absent. A 2025 review states explicitly that the form described in the Quarter Horse is established through muscle biopsy precisely because its genetic basis is unknown. To this day no confirmed genotype stands behind the label.
PSSM1 offers the opposite picture. A 2025 review describes the disease as an autosomal dominantly inherited storage disorder that occurs in more than twenty breeds and is established either through a gene test or through muscle biopsy. A prevalence estimate in selected North American and European breeds found the predisposition in 0.5 to 62.4 per cent of the horses examined, depending on breed, and demonstrated it in eleven breeds. In several other breeds examined, among them the Thoroughbred, the Connemara, the Icelandic horse and the Hanoverian, it was not found; the authors note that a failure to detect is not proof of absence.
How strong a real association looks is shown by a study of a Swiss Warmblood family. One affected stallion and 71 of his descendants were tested for the GYS1 variant. Within the family, 39 per cent showed signs of exercise related muscle breakdown, 51 per cent carried the variant, and carriers were at roughly seven times the risk. There too, not every carrier showed signs, and the paper points out that other causes of the same picture exist in Warmbloods.
A second example makes visible the distance a variant has to travel. For an immune mediated muscle inflammation in the Quarter Horse, the genomes of 36 affected horses and 54 comparison horses matched for age and breed were first searched for signals, after which the finding was confirmed in a further 35 affected and a further 22 comparison horses. Finally it was checked whether the variant occurs in 175 horses from 21 unrelated breeds. It did not occur there.
The first purpose built assessment appeared in 2021. It compared 54 Warmblood comparison horses with 68 Warmbloods in which PSSM2 or a myofibrillar myopathy had been established under the microscope, together with 30 Arabians in each group and 205 further genotypes from public databases. Not one of the genomic sites examined was linked to the diagnosis made on the section, neither singly nor as the sum of affected sites or of deviating alleles. Among the horses with a microscopic diagnosis, fewer than a third in each case carried the variant sold for it.
| Variant | Warmblood, healthy / affected | Arabian, healthy / affected | Quarter Horse, healthy / affected |
|---|---|---|---|
| P2 | 8 % / 15 % | 12 % / 17 % | 24 % / 21 % |
| P3a | 5 % / 6 % | 2 % / 2 % | 7 % / 12 % |
| P3b | 5 % / 6 % | 2 % / 2 % | 7 % / 11 % |
| P4 | 9 % / 13 % | 7 % / 12 % | 6 % / 8 % |
The second assessment followed in 2023, in 392 Quarter Horses: 229 healthy comparison horses and 163 horses with the biopsy diagnosis PSSM2 and without the GYS1 variant. At least one P variant was carried by 57 per cent of the healthy horses and by 61 per cent of the affected ones. On a scale where 1 stands for perfect separation and 0.5 for mere guessing, the variants reached 0.67. The authors worked out what use of the test would mean: 57 per cent of the healthy horses would be wrongly classed as diseased, and 40 per cent of the horses with microscopic evidence would be missed. P3a and P3b also occurred almost always together and therefore do not supply two independent pieces of information.
A side finding of the 2021 paper places the variants in time. All the P variants examined could be detected in genetic material from domestic horses between 400 and 5,500 years old, and P2 additionally in the Przewalski's horse. A variant that has sat in the population for millennia, and that also occurs in a separate wild lineage, is hard to read as a rare disease defect. That does not rule out a role, but it does not make one likely.
Healthy horses carry a great many striking variants. An analysis of 605 fully sequenced horses from 48 breeds found around 33 million variants. Per horse, an average of 730 of them were computationally predicted to be harmful. The same paper notes that the causal variant is known for fewer than 20 per cent of the suspected inherited diseases of the horse. Anyone who searches the genome of a sick horse for a rare, protein altering variant will therefore almost always find one, without anything having been explained by it.
This is precisely the point that was worked through for myofibrillar myopathy. One study compared eight Warmbloods carrying that diagnosis with eight Warmbloods without it, and drew in genomic data from more than 200 further horses. In 16 candidate genes, 426 variants were found, 26 of them altering the protein sequence. None of them was linked to the clinical picture, not even the two variants sold commercially as a test for this disease.
That binding rules of assessment were missing until recently is no accident. In 2024 an international working group developed and validated the first criteria for judging, in animals, whether a variant causes disease. Its reasoning is blunt: breeding decisions based on invalid DNA tests can damage the health of a population in the long term and, in the individual animal, lead to the wrong treatment or to a decision that ends its life.
Yes. The laboratory measures reliably whether a variant is present, and none of the assessments doubts that. The open question is a different one: whether that variant has anything to do with the disease. A measurement can be exact and its message still empty, if healthy comparison horses carry the same variant at a similar rate.
A 2025 review lists five validated gene tests for muscle diseases of the horse: hyperkalaemic periodic paralysis, malignant hyperthermia, glycogen branching enzyme deficiency, PSSM1 and myosin heavy chain myopathy. The same paper states explicitly that gene tests in veterinary medicine are not regulated and that every newly offered test in the horse must be carefully evaluated and confirmed as valid before it is used.
An uncomfortable observation follows: that a test is offered, advertised and paid for says nothing in itself about whether it has ever been examined in an independent sample. For the P2, P3 and P4 variants that examination now exists, and it came out negative.
That PSSM2 in the Quarter Horse is a disease in its own right was shown by the same research group in 2023. It compared 64 horses with PSSM2, 30 with PSSM1 and 185 comparison horses. Muscle glycogen content averaged 129 millimoles per kilogram in the PSSM2 horses, 175 in PSSM1 and 80 in the comparison horses. The value for amylase resistant storage material lay, for PSSM2, between that of the PSSM1 horses and that of the comparison horses, which had none at all.
The cause nevertheless remained open. The coding sections of twelve genes that trigger glycogen storage diseases in other species were compared between seven PSSM2 horses and five comparison horses. Of 29 variants with a predicted strong or moderate effect, none occurred exclusively in the affected horses. Descent, by contrast, argues for inheritance: 17 of 30 PSSM2 horses with an available pedigree traced back within four generations to one of three stallions.
Myofibrillar myopathy now counts as a disease of its own and no longer as a storage disease. A study of ten affected Warmbloods and eight comparison horses found a markedly higher muscle damage score and accumulations of the fibre protein desmin in the wrong place, with unremarkable glycogen content. The same accumulations were found in one mare line across three generations. A 2025 review describes two forms: in the endurance Arabian, muscle breakdown at the end of the ride with raised muscle enzymes; in the Warmblood, pain behaviour, loss of performance, shifting lameness and normal muscle enzymes.
An analysis of 3,615 submitted muscle samples shows where the difficulty lies. In Warmbloods with PSSM2, muscle breakdown was reported in 27 per cent of cases, in Warmbloods with PSSM1 in 75 per cent. Conversely, it was above all the gait that was abnormal in the PSSM2 horses, in 66 per cent of cases. The decisive comparison figure stands in the same paper: in Warmbloods without PSSM the gait was likewise abnormal in 53 per cent. Muscle enzymes in blood and glycogen content lay within the reference range in the PSSM2 Warmbloods.
A second paper measured again. Between 36 Warmbloods with PSSM2 and 23 Warmbloods without recognisable muscle disease, mean muscle glycogen content did not differ. Only the upper end of the distribution remained striking: horses with the highest values more often showed a loss of performance. The owner survey in the same project gave an onset of signs around the sixth year of life and, in more than half, a loss of performance and resistance to collection. Eighty per cent perceived an improvement on a change of feed, while 53 per cent of the horses still failed to progress as expected.
A 2025 paper attempted to reorder the exercise associated muscle diseases computationally. From 109 muscle samples, two broad clinical pictures emerged, one of them with visible muscle pain, reluctance to move, abnormal gait and neurological signs. A blinded examination of sections in 69 horses could not, however, subdivide these pictures any further, and typical features of other equine myopathies were found in neither subtype.
According to a 2025 review, muscle biopsy is often indispensable for a definitive diagnosis of neuromuscular disease. It is not, however, an automatic reading: assessing it requires experienced pathologists who bring the picture together with breed, age, history and clinical findings. The sample is taken either surgically as an open biopsy or percutaneously with a needle. Research laboratories prefer fresh and then frozen tissue, while diagnostic laboratories work with formalin fixed tissue and special stains.
The standard itself has also shifted. In a study of Arabians with exercise related muscle breakdown, 15 of 25 archive samples had previously been classed as PSSM2. On reassessment, central cell nuclei were found in 12 of 13 prospectively examined horses against 4 of 14 comparison horses, and desmin accumulations in 12 of 13 against 3 of 14. The authors concluded that glycogen gathering in disordered muscle fibrils had created the false impression of a storage disease.
Two limitations follow, and they apply to the gene test as well. A test cannot be more reliable than the standard against which it is measured. And that standard is itself a stand-in: it describes a change in tissue, not what matters to the horse. Whether the severity of the microscopic change corresponds to what a horse experiences in pain and loss of performance has not so far been shown.
Original analysis
| Test | How the variant was found | Published independent check | Outcome of the check | What a positive result establishes |
|---|---|---|---|---|
| PSSM1, GYS1 variant | Search in 831 horses from 36 breeds diagnosed on a muscle section, followed by a prevalence estimate in selected breeds | Yes, among others in a Swiss Warmblood family of 72 horses | Carriers were at roughly seven times the risk of exercise related muscle breakdown | Increased susceptibility; carriers without any sign do occur |
| Myosin heavy chain myopathy, MYH1 variant | Search in the genomes of 36 affected horses and 54 matched comparison horses | Yes, in a further 35 affected horses, a further 22 comparison horses and 175 horses from 21 other breeds | Association confirmed; in the 175 horses of other breeds the variant was entirely absent | Increased susceptibility to an immune mediated muscle inflammation |
| PSSM2 panel, P2 variant | Offered commercially; no peer reviewed first publication could be found in this search | Yes, twice: Warmbloods and Arabians in 2021, Quarter Horses in 2023 | No link to the diagnosis on the muscle section; variant also present in equine genetic material between 400 and 5,500 years old and in the Przewalski's horse | Nothing about the presence of a muscle disease |
| PSSM2 panel, P3a and P3b variants | Offered commercially; no peer reviewed first publication could be found in this search | Yes, the same two assessments | No link; the two variants occurred almost always together and therefore do not count twice | Nothing about the presence of a muscle disease |
| PSSM2 panel, P4 variant | Offered commercially; no peer reviewed first publication could be found in this search | Yes, the same two assessments | No link to the diagnosis on the muscle section | Nothing about the presence of a muscle disease |
| Myofibrillar myopathy test, two coding variants | Offered commercially; no peer reviewed first publication could be found in this search | Yes, in eight affected and eight unaffected Warmbloods across 16 candidate genes | None of the 426 variants found was linked to the clinical picture | Nothing about the presence of a myofibrillar myopathy |
That does not follow from the result alone. In the study of 392 Quarter Horses, 57 per cent of the healthy comparison horses carried at least one P variant and 61 per cent of the affected ones: the variant is almost as common in healthy horses. In Warmbloods and Arabians, too, no link could be shown between the variants and the diagnosis made on the muscle section. On these data a positive result barely shifts the probability of a muscle disease. Whether your horse has a muscle problem is decided by the signs it shows and by the veterinary examination, not by this genotype.
PSSM1 is genetically defined, PSSM2 is not. In PSSM1 a known variant in the gene for glycogen synthase is present; it has been demonstrated in 17 breeds, is inherited in an autosomal dominant manner and, in a Swiss Warmblood family, went together with roughly seven times the risk of muscle breakdown. PSSM2, by contrast, denotes horses whose muscle sample showed storage material and in which precisely that variant is missing. It is a label by exclusion. In the Quarter Horse it was shown in 2023 that a genuine storage disease in its own right lies behind the term; its genetic cause remains unknown despite the examination of twelve candidate genes.
The authors of both assessments advise explicitly against using these genotypes for the pre-purchase examination, for breeding decisions or for diagnosing a muscle disease. The reason lies in their own figures: use of the test would wrongly class 57 per cent of the healthy horses as diseased and at the same time miss 40 per cent of the horses with microscopic evidence. A result would therefore often be misleading in both directions. This editorial team makes no recommendation for or against any test; it reports what the published assessments write about it.
Through muscle biopsy, and precisely because the genetic basis is unknown. A 2025 review states this explicitly. The same literature also describes the limits, however: assessment requires experienced pathologists who bring the picture together with breed, age, history and clinical findings, and the classification has changed over time. In a study of Arabians, 15 of 25 archive samples had first been classed as PSSM2 and were later judged differently. Biopsy is the best standard currently available, not a certain one.
Because nothing prevents it. A 2025 review states that gene tests in veterinary medicine are not regulated and that every newly offered test in the horse must be carefully evaluated and confirmed as valid before it comes into use. There is no approval authority checking this before sale. That is why an international working group developed the first criteria in 2024 for judging, in animals, whether a variant causes disease, and justified them with the harm that breeding decisions based on invalid tests can do.
No, and the separation is recent. Myofibrillar myopathy long counted as a storage disease and is today treated as a disease of its own: it is marked by disordered muscle fibrils and accumulations of the fibre protein desmin in the wrong place, with unremarkable glycogen content. Two forms are described, in the endurance Arabian with muscle breakdown at the end of the ride and raised muscle enzymes, in the Warmblood with pain behaviour, loss of performance, shifting lameness and normal muscle enzymes. PSSM2 of the Quarter Horse, by contrast, is described as a genuine storage disease. The same holds for both: the commercially offered variants did not separate affected from healthy horses.
Terms defined in this text
Comparative Health
Ten of sixteen clinically unremarkable horses stayed under 30 minutes of REM sleep: measured lying times and what sleep loss actually shows.
MUMultiple speciesBmoderateT1hypothesis only Open access
Movement & Recovery
Of 750 horses at competitions, 44 per cent left no finger under the noseband: what the taper gauge, pressure sensors and a cadaver head show.
HOHorseBmoderate Open access
Equine Health
In the placebo-controlled trial 47 per cent of vaccinated horses halved their lesion score, 13 per cent on placebo. Where the 88 per cent figure comes
MUMultiple speciesCweakT1hypothesis only Open access
Once a month: new syntheses, checked studies, and what has changed in the state of the evidence.