MUMultiple speciesCweakT1hypothesis only
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 from.
Executive summary
An active vaccination against interleukin 5, one of the body's own signalling molecules, has been tested against sweet itch in horses. In the placebo-controlled trial in 34 Icelandic horses, 47 per cent of the vaccinated animals halved their lesion score, against 13 per cent on placebo. The frequently quoted figure of nearly nine horses in ten comes instead from the second season, which no longer had a placebo group. Skin lesions were measured, not itch.
Key points
Sweet itch is an allergic reaction to components in the saliva of biting midges of the genus Culicoides. A review from 2023 records two points that frame everything else: there is still no curative treatment, and the care of affected horses remains lifelong and costly. The frequency named there, up to 60 per cent, is a maximum from particularly exposed regions and not a general figure.
Across a broad population the picture differs. A survey of 3,409 Belgian warmblood horses found a frequency of 10 per cent over the whole lifetime and of 6.2 per cent at the time of the survey, with the heritability of susceptibility estimated as high. The diagnosis came from the keepers themselves, however, without veterinary confirmation: a questionnaire frequency is not the same as one recorded in the clinic.
No, in an observational study the timing of the first encounter with the midge weighed more heavily. Four groups from the same environment were followed for at least nine years: among the parent animals first exposed as adults, 62.5 per cent fell ill, against 21.4 per cent of the full siblings exposed while young and none of the animals exposed from birth.
Sensitisation itself also takes time. In a cohort of 16 Icelandic horses imported into North America, no animal showed signs in the first summer; in the second summer there were 9 of 16. Judging a treatment over a single season may therefore mean measuring the natural course of sensitisation as well.
In allergic skin inflammation in the horse, eosinophil granulocytes accumulate in the skin. Their growth and their survival depend strongly on one signalling molecule, interleukin 5. The vaccine that was tested is therefore not directed against a pathogen but against this signal produced by the body itself: the horse forms antibodies against its own signalling molecule and thereby reduces the number of these cells.
Technically, a piece of the equine interleukin is attached to a virus-like particle of cucumber mosaic virus, which additionally carries an element of tetanus toxoid. No adjuvant is added. This construction is the real trick, because the immune system does not normally attack the body's own substances.
Through the carrier. A laboratory paper from 2020 showed that the T cells of vaccinated horses react to the virus-like particle and not to the interleukin itself. Antibody levels are reversible, and the interleukin present in the body is not on its own enough to restart antibody production. It is precisely on this that the authors base their assessment of tolerability.
Every report about this vaccine rests on a single efficacy trial, published in 2018 in an allergy journal. 34 Icelandic horses with sweet itch were allocated to two groups, 19 received the vaccine and 15 a dummy preparation. An independent review from 2023 describes the procedure as randomised, double-blind and placebo-controlled, with monthly vaccination over three months and a booster two months later.
The result was clear but incomplete. 47 per cent of the vaccinated horses halved their lesion score and 21 per cent improved by three quarters. On the dummy preparation the figures were 13 per cent and 0 per cent. 17 of the 19 vaccinated horses formed measurable antibodies against their own interleukin at all, two showed no immunological response, and the clinical improvement depended on a minimum level of those antibodies.
The authors describe their result as the first successful immunotherapy against a chronic disease of the horse and as a step towards a comparable vaccine for humans. Cure is nowhere mentioned. The trial ran in a single breed, over a single season, with a lesion score as its yardstick, and the company developing the product took part in it.
The most widely passed on figure does not come from this trial but from the follow-up paper on the second season. 30 Icelandic horses were analysed there: 13 animals that had received the dummy preparation the year before and were now vaccinated for the first time, and 17 animals in their second vaccination year, which received a single booster in March.
In the first group 69 per cent halved their lesion score, in the second 88 per cent. An improvement by three quarters was reached by 31 and 47 per cent respectively. Out of those 88 per cent, retelling has produced “nearly nine horses in ten”.
| Analysis | Halving of the score | Improvement by three quarters | Concurrent control group |
|---|---|---|---|
| First season, vaccinated (19 horses) | 47 per cent | 21 per cent | yes, 15 horses on the dummy preparation |
| First season, dummy preparation (15 horses) | 13 per cent | 0 per cent | not applicable |
| Second season, vaccinated for the first time (13 horses) | 69 per cent | 31 per cent | no |
| Second season, second vaccination year (17 horses) | 88 per cent | 47 per cent | no |
What matters is what this second season lacked. Every horse received the vaccine, and there was no longer a concurrent placebo group. The most quoted figure therefore rests on no controlled comparison at all, and only the first year was blinded. The authors accordingly stay cautious: an annual vaccination could be a long-term solution.
The yardstick in all these trials is a points system for skin lesions: scaling, hair loss, rubbing marks, thickening, distributed across body regions. Only in 2026 was such a system tested comprehensively for sweet itch, in 44 privately kept horses in Switzerland and by six trained assessors. Specialists call it the first broadly validated score for this disease. Conversely, that means the older trials, the vaccine trial among them, worked with scales that had not undergone this testing.
The same paper contains a second, rarely quoted finding. The trained assessors agreed very closely with one another and with themselves. The keepers' visual judgement, by contrast, matched the tested clinical score only moderately, although the keepers agreed strongly among themselves. Broad consensus among keepers therefore does not replace a clinical measurement.
Not necessarily. In a randomised trial of an omega-3 cream, the skin lesions on the treated half of the body improved over the first four weeks compared with the untreated half, while the itch values did not change. After four weeks both sides were treated, so the later comparison was no longer controlled.
A randomised crossover trial of an essential oil spray in 20 horses shows how far perception and tissue can diverge. The values recorded by the veterinary surgeons did improve there against the dummy preparation, itch included. The keepers, however, reported complete disappearance of the complaints in 85 per cent of the horses, while of four tissue samples examined only one was entirely unremarkable.
The cleanest evidence for this comes from a double-blind, randomised trial of desensitisation with commercially available extracts. 20 horses, ten against ten, one year of follow-up, a check every four months: at no point did the clinical values differ. Both groups nevertheless improved to a similar degree, which the specialists attribute to the repellent that both groups received weekly.
That a control arm improves therefore has solid reasons: an accompanying measure that everyone receives, a variable midge year, a sensitisation that is not yet complete, and the increased attention a study horse receives in any case.
Two papers from other species show how strongly the frame of a study shifts the judgement. An analysis of 64 randomised trials in atopic dermatitis in humans found the response to the dummy treatment higher the longer the study ran, the more accompanying treatments were permitted and the fewer study arms there were; absence of blinding raised the response particularly at the level of skin lesions. In ageing dogs, the animals in the placebo arm were rated as improved by their keepers over six months, while dogs that were merely observed were rated as worse; the tests carried out on site changed in neither group.
A separate paper deals with tolerability. It followed 34 vaccinated Icelandic horses, spread across the first to the fifth vaccination year, together with affected horses on the dummy preparation and healthy unvaccinated animals. Kidney and liver values as well as the blood count were observed over time. The T cells of the vaccinated horses reacted strongly to the carrier and not at all to the equine interleukin itself.
The numbers behind the long-term observation are small, however. Beyond the third season, six horses were followed, four in the fourth and two in the fifth vaccination year. Surveillance after market launch is not mentioned in the papers analysed here, nor is an authorisation status or a price. This literature says nothing about them.
A laboratory paper from 2024 adds a point about timing. The number of eosinophil cells falls in the first vaccination year already, yet those that remain initially keep an inflammatory appearance. Only after two years does the population shift to a resting state of the kind healthy horses show. Cellular normalisation therefore takes two seasons, and this paper too comes from the developer's group.
The same working group has tested a second vaccine, directed against interleukin 31, the signalling molecule of itch. In a controlled trial in 18 horses the clinical values fell against the untreated previous season and against the horses on the dummy preparation. Improvement in this type of protocol is therefore not tied to interleukin 5.
A completely different approach reaches proportions that are at least as high. With desensitisation using nine genetically produced midge allergens, 67 per cent of the treated horses against 25 per cent of the sham-treated horses halved their score in the first year, and 89 against 14 per cent in the second. Behind these proportions stand single-digit numbers: six of nine, two of eight, eight of nine, one of seven. Allocation was moreover not random but balanced by stable, origin, age and sex; the keepers and the assessing clinic were blinded.
For placing all this in context, that means only one thing at first: several very different interventions produce similar-looking success rates in small groups. A direct comparison between them has never been carried out, and until recently no single tested scoring system was even available for the purpose.
Blockade of interleukin 5 has long been tested in humans on a large scale, though in a different way. A Cochrane review brings together 17 randomised studies with around 7,600 participants with severe eosinophilic asthma: the rate of worsening episodes falls roughly by half. The gains in quality of life and lung function, by contrast, mostly stay below the threshold a patient would notice at all.
These studies use ready-made antibodies, given at regular intervals, and not an active vaccination that lets the body produce them itself. A vaccine of this construction has not yet been tested in humans. What exists is a perspective article from the same group, published in 2018, proposing such a vaccine against asthma: a proposal, not a programme.
A paper from 2025 additionally showed that vaccination against interleukin 5 lowers allergen-specific class E antibodies, in mouse and horse, and dampens the signalling molecules in diseased skin. The only clinically measured endpoint in that paper, swelling of the ear, was recorded in the mouse alone. The preventive use hinted at there is expressly framed as a hypothesis and has never been tested.
Original analysis
| Approach | Best design found | Horses in the trial | What it was compared with | What this design does not support |
|---|---|---|---|---|
| Vaccination against interleukin 5, first season | Randomised, double-blind, against a dummy preparation | 34 (19 against 15) | Concurrent placebo group in the same season | That the improvement lasts beyond that one season |
| Vaccination against interleukin 5, second season | Half-crossover follow-up without a control arm | 30 (13 vaccinated for the first time, 17 with a booster) | The same horses' own previous season | That the 88 per cent is due to the vaccination |
| Vaccination against interleukin 31 | Controlled trial, randomisation not reported | 18 | Previous season and horses on a dummy preparation | That the effect depends on one particular signalling molecule |
| Desensitisation with whole-body extracts | Randomised, double-blind, against a dummy preparation | 20 (10 against 10) | Concurrent placebo group over one year | That the improvement in both groups came from the extract |
| Desensitisation with recombinant midge allergens | Double-blind, allocation not random but balanced | 17 (9 against 8, then 9 against 7) | Concurrent sham treatment over two seasons | That the groups were comparable in unknown features too |
| Essential oil spray | Randomised crossover trial, double-blind, 28 days per phase | 20 | Dummy preparation in the other phase | That the 85 per cent reported by keepers is reflected in the tissue |
| Cream with omega-3 fatty acids | Randomised half-body comparison, single-blind | 28 enrolled, 21 completed | The untreated half of the body, day 0 to day 28 | That the itch improves along with it |
This question cannot be answered from the published evidence, because the scientific papers contain no information on authorisation, availability or price. What is documented is the following: a vaccine against interleukin 5, one of the body's own signalling molecules, was tested in 2018 in a randomised, placebo-controlled trial in 34 Icelandic horses, with partial improvement in just under half of the vaccinated animals. Whether and in what form it may be used in Switzerland is for the attending veterinary practice to clarify, not for an article. This text describes what was measured and makes no recommendation.
The figure exists, but it comes from a season without a control group. In the second vaccination season, 88 per cent of 17 horses that received a single booster halved their lesion score. In that season every participating horse was given the vaccine, and there was neither a concurrent placebo group nor blinding. The more robust value comes from the first year: 47 per cent of the vaccinated horses against 13 per cent of the sham-treated ones. Both figures are correctly reported, they simply do not carry equally far.
No, and the studies do not claim so either. What was measured is a partial improvement in skin lesions, not removal of the allergy. A review from 2023 states expressly that there is no curative treatment for sweet itch and that care remains lifelong. In the second season the horses received a booster, and antibody levels recede if no further vaccination follows. Specialists therefore speak of a possible long-term treatment, not of a cure.
Because almost every trial counts skin lesions rather than rubbing behaviour. That is not a detail: in a randomised trial of an omega-3 cream, the appearance of the skin improved on the treated half of the body, while the itch values did not change. Only in 2026 was a scoring system for this disease tested broadly, in 44 privately kept horses in Switzerland. The same paper found that the keepers' visual judgement matches the clinical score only moderately, although keepers agree very strongly among themselves.
That is common and well documented. In a randomised trial of desensitisation, treated and sham-treated horses improved to a similar degree, which the specialists attribute to the repellent that both groups received weekly. Added to this are midge activity, which varies from year to year, and the fact that sensitisation takes time: in a cohort of 16 imported Icelandic horses none fell ill in the first summer and nine did in the second. A before-and-after comparison over a single season therefore says little.
The available data do not argue against tolerability, but they are thin. A separate paper followed 34 vaccinated Icelandic horses across up to five vaccination years and observed kidney and liver values as well as the blood count. The T cells reacted to the carrier and not to the body's own signalling molecule, and antibody levels recede. Beyond the third season, however, only six horses were followed, the endpoints were laboratory values rather than clinical events, and surveillance after market launch is nowhere mentioned.
Studied yes, equally well supported no. Desensitisation with nine genetically produced midge allergens reached 89 per cent against 14 per cent on sham treatment in the second year in 17 horses, though without random allocation and with single-digit groups. An older desensitisation with commercially available whole-body extracts showed no difference from sham treatment in a randomised trial. An essential oil spray improved the values recorded by veterinary surgeons in a crossover trial in 20 horses. A direct comparison between these approaches is missing.
Terms defined in this text
Comparative Health
Of 107 untreated sarcoids, 1.9 per cent regressed: what links bovine and human papillomaviruses, and where the comparison ends.
MUMultiple speciesBmoderateT1hypothesis only Open access
Equine Health
Pastern dermatitis is not a diagnosis but a skin reaction pattern: mites and an altered skin flora are found consistently, skin fungi are not.
HOHorseBmoderate Open access
Equine Health
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.
HOHorseBmoderate Open access
Once a month: new syntheses, checked studies, and what has changed in the state of the evidence.