Every LabelCheck every label, every time.

← Cross-reactivity guides

Oral allergy syndrome (birch pollen and PR-10 proteins)

By Symon, Every Label ·

If raw apple, hazelnut, or carrot makes your mouth and lips itch but the cooked version does not, the protein behind it is almost always one family called PR-10. PR-10 proteins look, to the immune system, a great deal like the major birch-pollen allergen (a protein called Bet v 1). So a person sensitized to birch pollen can react to the PR-10 protein in certain raw foods. That reaction is oral allergy syndrome, also called pollen-food syndrome: itching or tingling in the mouth, lips, and throat, usually soon after eating. This page is built around the protein, because the protein is what explains both why the reaction is usually local and why, in a minority of cases, it is not.

One fact drives the whole pattern: PR-10 proteins are fragile. They break down with heat and they break down in the stomach. That is why the raw food often itches and the cooked or baked version often does not, and it is why these reactions are usually confined to the mouth. But “usually local” is not the same as “always safe to eat,” and this page does not tell you any of these foods is safe for you. A minority of PR-10 reactions go beyond the mouth, the soy case in particular has documented systemic reactions, and the same foods can also carry a completely different, heat-stable protein that is a genuine red flag. Which of those is true for you is decided by component testing and your allergist, not by a rule on a page.

This is the mechanism hub. It explains PR-10 first, then groups the foods by how strongly they actually travel with a birch allergy, then sends you to each food’s own page and to your allergist. Where a claim is a verified cross-reactivity fact, it is drawn from the project’s conservative cross-reactivity floor. The reassurance most people come here looking for, that it is “just oral allergy syndrome” and fine to eat, is exactly the claim this site holds back until a clinician signs it off, so you will not find it stated as fact here.

The molecular why: PR-10, a birch-pollen protein wearing a fruit disguise

PR-10 is the spine of this whole story, so it goes first.

PR-10 stands for “pathogenesis-related protein, family 10,” a small defense protein that many plants make. The prototype, and the reason this cluster exists, is Bet v 1, the major allergen in birch pollen. The PR-10 proteins in certain fruits, nuts, vegetables, and legumes are close enough in shape to Bet v 1 that an immune system trained on birch pollen recognizes them too. So the food reaction is not really a food allergy that started with the food. It usually starts in the airway, with hay-fever-type sensitization to birch pollen, and the food reaction is the pollen allergy reaching the plate. That is why this is called pollen-food syndrome, and why it clusters in people with spring tree-pollen hay fever.

Two properties of PR-10 set the entire pattern below:

  • It is heat-labile and digestion-labile. PR-10 falls apart with cooking, baking, or canning, and it is broken down by stomach acid and enzymes. So the raw food can provoke a reaction while the cooked version often does not, and the reaction tends to stay in the mouth and throat, where the raw protein first makes contact, rather than being absorbed whole and acting on the whole body. This fragility is the reason the pattern is usually local. It is a mechanism, not a guarantee, and it is not a clearance to eat the food.
  • It is one shared shape, so a birch allergy reaches many foods at once. Because all of these foods carry a Bet v 1 lookalike, a single birch sensitization can light up apple, hazelnut, soy, peanut, stone fruit, carrot, celery, and more on a test at the same time. A positive test across many of these foods is the PR-10 fingerprint, and it is common. It tells you the immune system has noticed the shared shape. It does not, by itself, tell you which of those foods will actually cause a reaction, or how severe.

There is a crucial second layer. Several of these same foods also carry a different, heat-stable protein (a storage protein, or a lipid transfer protein) that is the actual driver of serious, whole-body reactions. Hazelnut, soy, and peanut each carry both kinds: a fragile PR-10 component that fits the usually-local picture, and a tough storage component that does not. That is why a bare “I tested positive to peanut” or “to soy” cannot be read as either reassuring or alarming on its own. Telling the two apart is what component testing does, and it is the whole point of the section on serious markers below.

The member foods, grouped by how strongly they travel with a birch allergy

This is the map of the cluster: the foods whose PR-10 protein cross-reacts with birch, sorted by how reliably a birch allergy reaches them. A reminder before the list: the verified cross-reactivity floor for this cluster has not yet cleared any of these individual food edges, so each one below is described as the documented PR-10 pattern, not as a settled per-food rule, and each one routes to component testing and your allergist rather than to a yes-or-no answer here.

These are the foods most reliably reported in birch oral allergy syndrome.

  • Apple, and the stone fruits (cherry, peach, apricot, plum) and pear. These are the textbook PR-10 foods. Raw apple and raw stone fruit causing an itchy mouth in a birch-allergic person is the most recognized form of this syndrome. Apple’s PR-10 is Mal d 1; peach’s is Pru p 1; cherry’s is Pru av 1; pear’s is Pyr c 1. The reaction is usually local and usually to the raw fruit, because the PR-10 protein is fragile.
  • Hazelnut. Hazelnut is both a classic PR-10 food (its PR-10 is Cor a 1) and a nut that carries serious storage proteins, so it sits in two stories at once. The PR-10 side is the birch oral allergy syndrome side. The storage-protein side is covered in the serious-markers section below, and it is the one that matters most for safety.
  • Kiwi and carrot. Kiwi (Act d 8) and carrot (Dau c 1) carry PR-10 proteins and are part of the same birch-driven oral allergy pattern.

What to do with that: a birch-allergic person who reacts to raw apple very commonly tests positive across this whole group, and may or may not react to each one. Whether any specific food in this group is safe for you is a component-testing and allergist question, not a rule. This page does not clear any of them.

Soy and peanut both carry a PR-10 protein (soy’s is Gly m 4, peanut’s is Ara h 8) homologous to birch Bet v 1, so a birch-allergic person can have a PR-10-type reaction to them. Both also carry heat-stable storage proteins that are the primary driver of serious peanut and soy allergy, which is the separate and more dangerous story in the serious-markers section.

Soy carries the cluster’s most important caution. The usual PR-10 picture is a local, oral reaction, but birch-sensitized people have documented systemic reactions to large doses of lightly-processed soy, classically soy milk, soy protein drinks, and edamame, through the PR-10 component Gly m 4. The fragility rule still applies (heavily processed and cooked soy is generally lower risk because the protein is degraded), but a big dose of barely-processed soy protein is the known exception where a “just oral” food turns whole-body. This is exactly why the cluster cannot be summed up as harmless, and why soy in particular is a conversation to have with your allergist, not a food to retry at home on the strength of “it is only oral allergy syndrome.”

What to do with that: a positive peanut or soy test in a birch-allergic person could be the mild PR-10 marker or the serious storage marker, and the two are managed completely differently. That distinction is made by component testing, below, with your allergist. This page does not clear soy or peanut for you, and it specifically does not tell you the soy is safe.

Carrot, celery, and the other root and umbellifer vegetables: watch celery in particular

Carrot and celery are part of the birch and mugwort pollen-food pattern through their PR-10 proteins (carrot Dau c 1, celery Api g 1). For most of this vegetable group the reported pattern is the usual local one.

Celery is the exception worth stating plainly. In pollen-sensitized people, especially those sensitized to mugwort as well as birch, celery is known to cause systemic reactions, including anaphylaxis, driven by heat-stable proteins (a lipid transfer protein and a defensin) rather than by the fragile PR-10. So celery does not follow the “cooked is fine” pattern reliably, and it should be treated as a possible systemic reactor, not folded into a mild-mouth-only group. Confirm with your allergist before treating celery as low-risk. The deeper celery and spice mechanism lives on the mugwort, celery, and spice page and the LTP syndrome page rather than here.

The serious markers: the heat-stable proteins that are the real red flags

This is the most important section on the page, because it is where the “usually mild” picture has its hard limit.

Several of the foods above carry a second, completely different protein that does not follow the fragile-PR-10 rule. These are the storage proteins (called 2S albumins, 7S vicilins, and 11S legumins) and the lipid transfer proteins (LTPs). Unlike PR-10, these proteins are heat-stable and survive digestion, so cooking and baking do not defuse them, and a reaction driven by them can be whole-body rather than confined to the mouth.

For the tree nuts in this cluster, this is verified against the cross-reactivity floor: a positive test to a tree-nut storage protein (for example cashew Ana o 1 and Ana o 2, walnut Jug r 2 and Jug r 4, pistachio Pis v 2, Pis v 3 and Pis v 5, pecan Car i 2 and Car i 4) is a red flag for whole-body reactions, not a reassuring low-risk result; these are heat-stable and survive digestion, so roasting, baking, or cooking does not make the nut safe. For hazelnut specifically, the storage proteins Cor a 9 and Cor a 14 are associated with systemic reactions, and hazelnut and walnut commonly cause dual allergy through their shared 2S albumins and legumins. The same logic applies to soy and peanut: their storage proteins (soy Gly m 5, Gly m 6 and Gly m 8; peanut Ara h 1, Ara h 2, Ara h 3 and Ara h 6) are the primary drivers of serious soy and peanut allergy, a separate and more dangerous picture than the birch PR-10 one. For the most concentrated and higher-risk soy forms, such as textured or hydrolyzed soy protein, see where soy hides.

So the same food can carry both kinds of protein. Hazelnut carries a fragile PR-10 (Cor a 1, the birch oral allergy syndrome marker) and tough storage proteins (Cor a 9 and Cor a 14, the systemic markers). Soy carries Gly m 4 (PR-10, plus the soy-milk systemic exception) and Gly m 5, 6, and 8 (storage, primary systemic). Peanut carries Ara h 8 (PR-10) and Ara h 1, 2, 3, and 6 (storage, primary systemic). This is why a single positive result to “peanut” or “soy” or “hazelnut” cannot be read as either reassuring or alarming on its own. Which protein your antibodies are aimed at is the question that matters, and it is the question component testing answers.

The full storage-protein story lives on the seed and tree-nut storage-protein page, and the LTP story on the lipid transfer protein page. They are the serious siblings of this syndrome.

What is NOT this syndrome: the boundaries worth drawing

The PR-10 / birch cluster is broad, but it has edges, and drawing them keeps the picture honest.

A positive PR-10 test is not the same as a serious food allergy, and a positive storage-protein or LTP test is not oral allergy syndrome. This is the central distinction, repeated here because it is the one that keeps people safe. A result driven by the fragile PR-10 component fits the usually-local picture; a result driven by a heat-stable storage protein or LTP is a red flag for whole-body reactions and is not reassuring (verified against the cross-reactivity floor). The two can coexist in the same food. Neither one, on its own, tells you a food is safe to eat.

Profilin is a different birch-driven pan-allergen, and it usually means something else. Birch pollen also drives sensitization to a separate shared protein called profilin, which lights up on tests across many pollens and plant foods. Profilin positivity is most often serological noise: a positive test without a consistent real-world reaction. It is a different protein with a different meaning from PR-10, and it is covered separately. A broad positive plant-food panel in a birch-allergic person is frequently profilin, not a clinically meaningful PR-10 reaction, which is one more reason a raw panel needs component testing to interpret.

This is not a clearance for the cooked food. The heat-lability of PR-10 explains why the cooked form is often better tolerated, but “often” is not “always,” celery and the soy-milk Gly m 4 case break the rule, and the cleared verification floor for this site has not signed off any “safe to eat” claim for these foods. So this page does not give you one. Whether you can eat a specific food, raw or cooked, is a component-testing and supervised-challenge question for your allergist.

Where studies disagree

Two genuinely unsettled areas are worth seeing as disagreements rather than settled facts.

Co-sensitization versus real reactions, the whole reason this cluster needs component testing. A birch-allergic person very commonly tests positive across a long list of PR-10 foods, but a real-world reaction follows much less often, and when it does it is usually local. The test and the clinical reaction are measuring two different things, and the gap between them is wide for this cluster specifically, because the shared Bet v 1 shape produces so many positive results. This gap is exactly why a broad positive panel is not an instruction to avoid every food on it, and equally why it is not permission to eat them, and why component testing plus history, not the panel alone, is what decides.

How often a PR-10 reaction goes systemic, and in whom. The textbook teaching is that birch oral allergy syndrome is local and mild, but the literature documents systemic reactions, the soy Gly m 4 large-dose case being the clearest, and the true rate and the risk factors (dose, how processed the food is, the specific food, co-sensitization to other proteins) are not settled. The practical consequence is conservative: the usually-local pattern is real, but it is not reliable enough to act on without a clinician, which is why this page describes the mechanism and holds the reassurance.

Allergy shots for birch in someone with these food reactions. Active birch immunotherapy (allergy shots or drops), especially during pollen season, is documented to lower the reaction threshold to Bet v 1-homolog foods for some people, so a food that was tolerated can become more reactive during treatment. Whether and how much this happens is variable. The practical step is to tell your allergist about your pollen-food reactions before and during birch immunotherapy. This is a conversation to have, not a reason to refuse a treatment your allergist recommends.

Testing and confirmation

Cross-reactivity questions in this cluster are answered the same way as the core diagnosis: component-resolved testing, and where needed a supervised challenge.

A standard whole-extract blood test or skin prick to a food tells you the immune system has noticed it, but for this cluster that is often just the shared birch shape showing up, and it cannot tell a fragile PR-10 result from a heat-stable storage-protein or LTP result. Component testing breaks the result down to the actual proteins. For a birch-allergic person, the useful split is PR-10 (for example Cor a 1 for hazelnut, Gly m 4 for soy, Ara h 8 for peanut, the usually-local markers) versus the storage proteins and LTPs (Cor a 9 and Cor a 14, Gly m 5, 6 and 8, Ara h 1, 2, 3 and 6, the serious markers). That distinction is what decides whether a positive food test reflects the mild pattern or the dangerous one, and it changes the management completely. The soy components in particular are worth testing, because Gly m 4 is under-detected on whole-soy extract testing.

Where component testing and history still disagree, the supervised oral food challenge is the test that settles whether a specific food can actually be eaten, raw or cooked. It is done with your allergist, never at home, and it is the only thing that turns “related on a test” into “safe to eat.” Nothing on this page is that clearance.

The members (each food’s own page):

The neighboring hubs:

  • Seed and tree-nut storage proteins: the serious-marker syndrome (the heat-stable proteins that are the red flags)
  • Lipid transfer protein (LTP) syndrome: the other heat-stable serious-marker cluster, and the celery story
  • Profilin pan-allergen: the other birch-driven shared protein, the one that is usually serological noise (page forthcoming)
  • Mugwort, celery, and spice (pollen-food): where the celery and spice mechanism lives in full (page forthcoming)
  • Tree nuts and legumes: the family pages (the category, the label-scan and FALCPA detail)

Frequently asked questions

My mouth itches when I eat raw apple but not cooked apple. What is that?

That pattern is the signature of oral allergy syndrome, also called pollen-food syndrome. The protein in raw apple (called PR-10, or Mal d 1) looks like the major birch-pollen protein, so a birch-allergic immune system reacts to it; the protein is fragile and breaks down with cooking, which is why the cooked apple often does not itch. This page explains the mechanism, but it does not tell you the apple is safe to eat, because how a given person’s reactions behave varies and a minority go beyond the mouth. Confirm what is safe for you with component testing and your allergist.

Is oral allergy syndrome dangerous, or is it always mild?

It is usually local, an itchy mouth, lips, or throat, but “usually” is not “always.” A minority of reactions go beyond the mouth, soy in particular has documented systemic reactions to large doses of lightly-processed soy such as soy milk, and celery can be systemic in pollen-sensitized people. Because of that, this site does not publish a blanket “it is just oral allergy syndrome and safe” reassurance. Whether your reactions are mild and stay that way is a question for your allergist, not a rule you can read off a page.

I tested positive to peanut and soy but I only have hay fever. Does that mean I am allergic?

Maybe not in the way the test makes it look. In a birch-allergic person, a positive peanut or soy test is often the mild PR-10 component (Ara h 8 for peanut, Gly m 4 for soy) that travels with the pollen allergy, rather than the serious storage proteins that drive primary peanut and soy allergy. A positive storage-protein test is the related red-flag rule, verified against the cross-reactivity floor. Component testing tells the two apart, and that distinction changes everything about how it is managed. Take the result to your allergist for the breakdown rather than reading it as a yes or no.

If the protein is destroyed by cooking, can I just eat these foods cooked?

Cooking does often help, because the PR-10 protein is heat-labile, but this is a mechanism, not a guarantee, and this site does not turn it into a green light to eat the cooked food. Two known exceptions break the rule: a large dose of lightly-processed soy can cause a systemic reaction through Gly m 4, and celery can be systemic through heat-stable proteins that cooking does not destroy. Some of these foods also carry heat-stable storage proteins that are a genuine red flag, verified against the cross-reactivity floor. Whether you can eat a specific food cooked is a supervised question for your allergist.

Why did my whole plant-food panel light up when I only react to a couple of things?

Because birch sensitization drives broad cross-reactivity to shared proteins (PR-10 and profilin) that turn up in many plant foods, so the panel shows lots of positives that often do not match real reactions. Profilin in particular is usually serological noise, a positive test without a consistent reaction. This is why a broad positive panel in a birch-allergic person is read with component testing and your history, not taken at face value, and why it is neither a reason to avoid everything that lit up nor permission to eat it.

Should I worry about birch allergy shots affecting my food reactions?

Tell your allergist. Active birch immunotherapy, especially during pollen season, can lower the reaction threshold to birch-related foods for some people, so a food that was tolerated may become more reactive during treatment. How much this happens varies. It is a reason to keep your allergist informed about your pollen-food reactions, not a reason to refuse a treatment they recommend.

References and medical review

This page is pending independent medical review. The PR-10 membership map and the syndrome-level mechanism (the Bet v 1 homology, the heat- and digestion-lability, the usually-local pattern, the Gly m 4 systemic exception, and the birch-immunotherapy threshold note) resolve to the project’s cross-reactivity database, which is not yet signed off by a medical reviewer, so this page is not published until a named reviewer and date are in place. The verified serious-marker claims (a positive tree-nut storage-protein test is a heat-stable red flag rather than a reassurance; hazelnut Cor a 9 and Cor a 14 are associated with systemic reactions, and hazelnut and walnut commonly cause dual allergy through their shared 2S albumins and legumins; concentrated soy protein such as textured or hydrolyzed soy is the higher-risk soy form) resolve to the project’s conservative cross-reactivity floor, each carrying its own tier-1 source there. The “usually mild, mouth-only, safe cooked” reassurance is deliberately held back from publication until a clinician signs it off.

← Cross-reactivity guides