For patients with food allergies, the use of the oral food challenge (OFC) has long been considered the gold-standard diagnostic test. However, conducting an OFC test can be difficult for both the patient and the clinical staff. Researchers at the 2024 AAAAI Annual Meeting presented data on different testing strategies that may help expand diagnostic options.

Following these presentations, featured expert Edwin Kim, MD, MS, was interviewed by Conference Reporter Associate Editor-in-Chief Mona Shah, PharmD. Dr Kim’s clinical perspectives on these findings are presented here.

Most allergy offices still use food allergy skin prick testing to evaluate patients for a suspected IgE-mediated food allergy. I think we have found that such testing can do a pretty good job. At the same time, however, we know that these are not perfect tests, they are screening tests. We do see false-positive results (ie, essentially tests suggesting that there is IgE present, but the patient is still able to tolerate the food). This has driven the need for additional follow-up testing, and, historically, the OFC has been considered the gold standard.

The conversation around the utility, requirements, and role of OFCs has been ramping up leading up to the 2024 AAAAI Annual Meeting, in part because we know that OFCs are difficult. The most obvious problem with the OFC is the risk of the person reacting during testing. Although we know that life-threatening reactions are rare, that chance is always there. Moderate to severe reactions can also happen, which can be frightening and traumatic to the patient and difficult for the clinical staff. Other considerations include the financial costs and the time commitment. Although the OFC is the most definitive test that we currently have, there are many reasons why it is not the perfect test. To try to address that, some of the steps that have been taken over the years include attempting to fine-tune our current IgE testing. In other words, getting down into the actual nitty-gritty of the foods themselves at the protein level and seeing if that can be helpful.

At the AAAAI meeting, an abstract by Yanagida et al focusing on milk allergy evaluated the utility of using casein as a specific component within milk and how useful that could be for improving diagnostics (abstract 105). And what they found is in line with what we have seen in previous studies of components. It seems like there may be an incremental benefit compared with the standard milk test. However, the question remains: Is a little additional accuracy worth the extra costs and perhaps the extra steps that are needed to do that test? There would be a financial burden to patients, as well as the requirement of sending off tests to specialty labs. My take on this is that it is an incremental improvement. It is not a gold-standard type of test, so I think that we need to continue looking for additional tests that could be helpful.

Abstract 118, also presented at the meeting, was looking at the avidity of specific IgE antibodies (sIgE). So, not necessarily the measured range of IgE, but how strongly it binds. These concepts make a lot of sense at the immunological level, but, at the same time, I think that access to these tests is going to be important. Until ideas like this are validated in larger settings, I think that they are watch-and-wait scenarios.

One concept presented by Schuler and colleagues at the AAAAI meeting that I think is important is the use of machine learning (abstract 804). Every day, we are hearing about artificial intelligence in different scenarios, and medicine and allergy should be no different. This abstract described using machine learning to predict peanut OFC outcomes. While I do think that this is an important concept, it is in the early stages. We do not have a lot of literature out there on machine learning, but I feel strongly that we have a lot of data potentially available. Is a skin test perfect? No. Is the blood IgE test perfect? No. Component testing? Not perfect. But that combination of information, plus patient-specific characteristics, could be helpful. If we had a way to combine all that information to perhaps come up with a calculation or number, it makes sense that this could get us closer as well. We often try to do this in our own brains right now, but it is a lot of information. If we have the ability through machine learning to not only model for the patient in front of us but also recognize that it is a continuous learning process, then perhaps it can continually improve as more patients are plugged into the program. That seems like an important step forward.

The closest tests that we have right now to replacing OFCs are probably the basophil activation test (BAT) and the mast cell activation test (MAT). Both cellular-based tests are trying to show the release of actual mediators that ultimately cause the reaction. Some scenarios that really can be confounding and difficult for allergists are when patients have a very high IgE to a certain food yet are able to eat it, or vice versa. Tests such as the BAT and MAT that measure the mediator release may be more functional and more accurate tests. One of the biggest challenges in terms of incorporating these tests into clinical practice is figuring out how to make them more practical for use in allergy offices all over the country. That is one of the steps being worked on now.

It is important to note that I do not think that we ever want to fully replace the OFC, but what we really want to be thinking about is how to restrict its use to the patients who really need it. In my mind, there are going to be gray areas. Based on the progress that is being made with all these other tests and machine learning, I think that we should be able to get to a point where we can more accurately diagnose food allergies. But there will always be a little bit of gray, and I think that we want to have that OFC option for these scenarios. So, I do want to be careful to say that we are looking at how to narrow down when the OFC is needed so that we are using it in those most-needed scenarios, rather than fully replacing it.