Longitudinal basophil activation testing for assessing response to low-dose initiation, slow-escalation peanut oral immunotherapy in patients with high Ara h 2-sIgE

Introduction

Peanut allergy has risen markedly worldwide over the past decade, affecting approximately 2%–3% of children in Western countries (1). Unlike milk or egg allergy, which often resolves in childhood (2), peanut allergy is frequently lifelong, with only 10%–20% of affected children achieving natural resolution (3, 4). Consequently, strict dietary avoidance remains the standard of care but is associated with frequent accidental exposures and reduced quality of life (QoL).

Allergen immunotherapy (AIT) provides an active approach to desensitization and is increasingly adopted in clinical practice globally (4). The 2020 FDA approval of Palforzia® (formerly AR101) marked an important milestone in standardizing peanut oral immunotherapy (pOIT) (4–6). Meta-analyses of clinical trials demonstrate desensitization (DS) in roughly two-thirds of patients (7) and sustained unresponsiveness (SU) in selected populations, particularly younger children (8, 9). Nevertheless, SU is not equivalent to cure (3), and protection often wanes over time without continued ingestion with approximately 50% loss of tolerance every three months (10).

While achieving SU or long-term tolerance remains a central goal in efficacy assessment (11), for many families and patients, protection against accidental exposures is their primary clinical objective. In real-world settings, most unintended peanut exposures involve small amounts of peanut, typically well below 100 mg, approximately the population ED50 (eliciting dose). A reactivity threshold of ≥300 mg is estimated to protect against >95% of accidental exposures and represents a clinically meaningful benchmark (12). Achieving and maintaining this threshold can substantially improve QoL, even when ongoing peanut ingestion remains necessary (13).

Currently, no reliable biomarker exists to predict or monitor adverse reaction (AR) risk or guide patient selection for AIT. Baseline allergen-specific IgE (sIgE) levels >50 kU/L are often associated with increased risk of anaphylaxis and OIT failure (14–16), although findings are inconsistent across studies. Ara h 2 is the most informative serologic marker of peanut allergy severity (17), yet its value for predicting OIT outcome remains unknown. For high-risk patients (e.g., sIgE >50 kU/L), modified approache—including slow dose increment, low maintenance dosing or sublingual immunotherapy (SLIT)—have been proposed to enhance safety while preserving efficacy (14, 18).

Despite advances in understanding the immunological mechanisms of OIT (19) and protocol optimization, validated biomarkers for monitoring desensitization and guiding individualized therapy remain limited. Double-blind, placebo-controlled food challenges (DBPCFC), while the gold standard, are resource-intensive and carry procedural risk. Candidate biomarkers include humoral, cellular, and metabolic measures. A hallmark of successful AIT is the induction of blocking IgG antibodies, particularly IgG1 and IgG4, which may increase 10 to 100-fold from baseline (20). However, the relationship between IgG4 induction and clinical outcomes remains inconsistent (21). Conventional serologic markers (sIgE, sIgG4) reflect humoral responses but do not directly capture effector-cell reactivity. In contrast, the basophil activation test (BAT), which quantifies allergen-induced CD63 expression via FcεRI signaling, provides a functional readout of effector-cell responsiveness and has shown promise in assessing OIT outcomes (22, 23).

In this retrospective, real-world study, we evaluated whether peanut-allergic patients with high Ara h 2–specific IgE could safely achieve clinically meaningful desensitization using a low-dose initiation, slow-escalation OIT protocol. We further examined longitudinal changes in BAT and sIgE/sIgG4 ratios as dynamic candidate biomarkers of immune modulation and treatment response.

Methods

Study design and population

This retrospective, real-world study was conducted at a single allergy clinic center between September 2021 and June 2025. Thirty-two patients aged 4–18 years with physician-diagnosed peanut allergy who underwent standard-of-care peanut OIT (pOIT) were included in this study (patient flow diagram seen in Supplementary Figure S1). Inclusion criteria comprised a history of clinical reactivity, a positive skin prick test (SPT; wheal ≥3 mm), peanut-specific IgE (sIgE) level >0.35 kU/L, and at least two basophil activation test (BAT) assessments performed ≥6 months apart. Patients with uncontrolled asthma or concurrent milk allergy, which could confound BAT results (24), were excluded. All eligible patients meeting the prespecified inclusion criteria during the study period were included; no a priori sample-size calculation was performed. Patients with incomplete OIT treatment were excluded from this biomarker analytic cohort study.

Participants were stratified a priori by baseline Ara h 2-sIgE using a threshold of 40 kU/L, selected based on published clinical data identifying approximately 40 kU/L as a level associated with a high likelihood of clinically relevant peanut allergy (25). This threshold was used to define a subgroup with marked sensitization and was not intended as an OIT-response or therapeutic cutoff. Sixteen patients with Ara h 2-sIgE ≥40 kU/L formed the “high sIgE” subgroup, while the remaining 16 with lower values constituted the “lower sIgE” comparison subgroup. All participants followed the same OIT protocol. De-identified data were analyzed under approval from the Salus Institutional Review Board (IRB).

Peanut formulation and OIT protocol

Consistent with most real-world practice, OIT was performed using food-based formulations rather than a standardized pharmaceutical product (3). In this study, peanut protein doses were prepared from in-house mixed nut butter (MNB) and peanut butter (PB), or commercially whole peanuts, peanut butter-containing candy (e.g., Reese's Pieces, M&M's), PB2 powder, with standardized peanut protein content (Supplementary Table S1). Before reaching an SCD of 300 mg, patients received peanut doses prepared from in-house MNB or PB to ensure accurate dose administration. After attaining an SCD >300 mg, patients gradually transitioned to whole peanuts or commercially available peanut-containing products.

To minimize risk, oral food challenges (OFCs) were omitted for most patients. The protocol employed a very low initial dose, slow dose escalation, and a prolonged up-dosing phase lasting 1–2 years (Supplementary Table S2). Briefly, the planned schedule comprised 18 dose levels beginning at 0.1 mg and progressing to 364.4 mg by Visit 18, thereby crossing the clinically meaningful 300 mg threshold (Figure 1A), a threshold estimated to protect against >95% of accidental exposures and considered a clinically meaningful level of “real-world” protection (4, 12), even in patients who remain allergic (9). Dose escalations were typically performed every 14–28 days during clinical visits, provided prior doses were well tolerated or associated only with mild symptoms. The maintenance dose was set at the highest individually tolerated dose, with continued daily ingestion to sustain desensitization. Given that long-term clinical benefit from AIT may require >3 years of treatment (21), participants continue daily intake beyond 36 months; however, data beyond this time point were not included in the present analysis. Real-world adherence, including missed doses, and dosing-related reactions, was systematically documented.

Clinical assessments

The clinical effectiveness of pOIT was evaluated with desensitization. Successfully consumed dose (SCD) was defined as the total daily peanut-protein dose that was successfully consumed. Effective desensitization was defined as achieving an SCD of ≥300 mg peanut protein. The primary clinical endpoint is attainment of a successfully consumed daily peanut-protein dose (SCD) ≥300 mg during OIT. Safety was assessed by documenting the frequency and severity of dose-limiting symptoms during OIT. Symptom severity was graded according to World Allergy Organization (WAO) and Consortium for Food Allergy Research (CoFAR) Criteria, Version 3.0 (26, 27); only moderate-to-severe reactions were considered dose-limiting. Mild symptoms could prompt temporary withholding of escalation or continuation at the same dose according to clinical judgment but were not classified as protocol-defined dose-limiting events (Supplementary: Methods). Anaphylaxis was defined per the 2020 WAO updated grading system.

Basophil activation test (BAT)

BAT was modified from a flow cytometry-based protocol (28). Briefly, within 24 h of blood collection, 200 µL of heparinized whole blood was incubated for 20 min at 37 °C with PBS or KLH (negative controls), anti-IgE antibody or fMLP (positive controls), or five serial concentrations of peanut protein (10,000, 1,000, 100, 10 and 1 ng/mL). Cells were subsequently stained with fluorescent antibodies against CD63 and CD123/CD193, incubated for 30 min at 2–8° C, lysed (BD lysing solution), washed with PBS, and acquired on an Attune NxT flow cytometer (Thermo Fisher). Data were analyzed with FCS Express 7 (de novo Software). Basophils were gated as CD123+/CD193+ with low side scatter. A minimum of 500 basophils per sample was analyzed, and CD63 expression was quantified. Gating was standardized using PBS controls (2 ± 0.25% CD63+). Basophil activation ≥5% CD63⁺ cells was considered positive.

Dose–response curves were generated by plotting the percentage of CD63⁺ basophils against the peanut protein concentrations. Given the heterogeneous and often bell-shaped nature of BAT curves (24), the area under the curve (AUC) was used as the primary integrated measure. BAT AUC was calculated by trapezoidal integration of %CD63+ basophils across the five log10-transformed peanut concentrations. The maximum dose response (MDR) was also recorded for clinical comparison.

BAT was validated and is in clinical use as a laboratory-developed test (LDT) conducted at our CLIA-certified, CAP-accredited Virant Laboratory. The core analytical validation data are summarized in the Supplementary Methods (Supplementary Tables S3–S5). BAT assays were performed, analyzed, and initially reported by one investigator who did not have access to immunoglobulin results or clinical outcome data at the time of interpretation, and the BAT results were subsequently audited by a second investigator under the same blinding conditions. Clinical treatment and outcome data were collected separately by another investigator without prior knowledge of laboratory results. The laboratory and clinical datasets were subsequently integrated for statistical analysis by another author.

Total IgE, sIgE and sIgG4 measurements

Serum levels of total IgE, peanut- specific and two component-specific IgE (Ara h 2, Ara h 6), and corresponding sIgG4 were measured at baseline, during up-dosing phase, and during maintenance. Assays were performed using the ImmunoCAP platform (Phadia Immunoassay Analyzers; Thermo Fisher Scientific Inc., Waltham, MA) at a CLIA-certified commercial laboratory. For graphical presentation and descriptive summaries, values exceeding the upper limit of quantification (100 kU/L for sIgE, 30 mg/L for sIgG4) were displayed at the assay upper limit. For inferential longitudinal analyses, these observations were treated as right-censored.

Statistical analysis

Continuous variables are presented as mean ± standard error of the mean (SEM). BAT AUC and SCD were analyzed using longitudinal mixed-effects models with participant-specific random intercepts and assessment period as a categorical fixed effect. Skewed continuous outcomes were analyzed on the natural-log scale. Peanut-, Ara h 2-, and Ara h 6-specific IgE and peanut-specific IgG4 were analyzed using right-censored log-normal mixed-effects models with participant-specific random intercepts. No imputation was performed for missing biomarker measurements. Mixed-effects models used all available observations for each participant. Sample size at each assessment period is reported in the corresponding tables and figures. For each biomarker, the five follow-up-vs.-baseline contrasts were adjusted for multiplicity using the Holm procedure.

Cross-sectional exploratory associations were evaluated using Pearson's correlation analysis, as specified in individual figure legends. For analyses involving measurements that exceeded the assay upper limit, correlation results based on assay-limit values were considered descriptive.

Categorical variables were compared using Fisher's exact test. Receiver operating characteristic (ROC) analysis was used to assess discrimination of continuous candidate biomarkers. The identified thresholds were evaluated as exploratory candidate thresholds rather than validated diagnostic cutoffs in this study. Diagnostic-performance estimates were reported with confidence intervals where applicable.

An exploratory time-lagged analysis was performed to evaluate whether biomarker measurements at an earlier time point were associated with subsequent clinical desensitization. Candidate biomarker measurements were compared with the SCD achieved during a subsequent follow-up period.

All tests were two-tailed, and a p value < 0.05 was considered statistically significant, * p < 0.05, ** p < 0.01, *** p < 0.001. The mixed-effects, censored mixed-effects, Holm contrasts, and bootstrap CIs were performed in STATA 18.0, while GraphPad Prism 10 and SRplot (29) were used for BAT AUC calculations, Pearson correlation analysis, Fisher's exact test, ROC analyses.

Results

High desensitization rates and favorable safety

Of 49 patients who initiated clinical peanut OIT at IAA, 9 had incomplete OIT follow-up, including 5 patients who were lost to follow-up and counted as nonattainment, and were excluded from biomarker analytic cohort. Among 40 patients with adequate OIT follow-up, 8 had fewer than two BAT assessments performed ≥6 months apart and also excluded from the biomarker study (Figure 1A). Therefore, 32 patients aged 4–18 years met all screening criteria and were included in this longitudinal biomarker-analysis cohort study, with 16 having high baseline Ara h 2-sIgE (≥40 kU/L). The cohort was male-predominated in both high (75%) and low (81%) sIgE subgroups (Table 1). While mean total IgE level was higher in the high-sIgE subgroup, this difference was not statistically significant. The descriptive mean baseline Ara h 2-sIgE was 80.3 kU/L, with observed values ranging from 42.1 to ≥100 kU/L. The majority of patients with high Ara h 2-sIgE also had elevated peanut sIgE (15/16, Table 1). Ara h 2-sIgE and Ara h 6-sIgE showed strong and medium descriptive correlations with peanut sIgE, respectively (Supplementary Figure S2). Because several values exceeded the assay upper limit (100 kU/L), these correlations were interpreted descriptively. The true mean sIgE levels in high-sIgE group are likely underestimated.

Table 1. Demographic and clinical characteristics of two patient groups undergoing pOIT.

BAT, basophil activation test; AUC, area under the curve; SCD, successfully consumed dose daily; OIT, peanut oral immunotherapy.

*statistically significant difference.

Compared with the commercial Palforzia® protocol, a conservative, slow-escalation regimen was used (Supplementary Table S2), starting at an extremely low dose (0.1 mg peanut protein) and progressing through a prolonged up-dosing phase (≥18 clinic visits) to reach a target of ≥300 mg peanut protein (Figure 1A). Most scheduled doses were completed during the initial 18 visits, although a few missing doses occurred (Figure 1B), followed by individualized dosing (continuing escalation or maintenance) for up to 36 months (maximum 35 visits). The highest successfully consumed dose (SCD), assessed at about 6-month intervals, increased steadily through 24 months and remained stable for 25–36 months (Figure 1C). Treatment duration did not differ significantly between subgroups; however, the study was not powered to establish differences in escalation time. The planned interval between dose-escalation visits was 14–28 days during the first 18 visits; however, the observed intervals were generally longer than planned (Figure 1D) and longer than the 14-day interval recommended for Palforzia® (6), reflecting a very-low-dose initiation with extended-interval, prolonged up-dosing protocol. All 32 participants completed the first 18 visits within 12–24 months. The speed of desensitization varied; the fastest patient reached 300 mg in 14 visits. By 24 months, effective desensitization –-defined as a SCD ≥300 mg— was achieved in 68.8% (11/16) of the high sIgE subgroup and 62.5% (10/16) of the lower sIgE subgroup (relative risk 1.10, 95% CI 0.67–1.82) (Table 1). Although observed desensitization proportions were numerically similar between the high- and lower-Ara h 2-sIgE subgroups, due to the wide confidence intervals, the study was not powered to establish therapeutic equivalence across sensitization strata. Most successfully desensitized patients continued dose escalation to further raise their tolerated daily dose; nine exceeded 1,000 mg, with the highest reaching 7,000 mg (Table 1). Notably, six of eight patients who completed OIT but were not included in this biomarker study attained SCD ≥300 mg. Therefore, among all 49 patients who initiated OIT, 27 (55.1%) had documented attainment of SCD ≥300 mg. Among all 40 evaluable patients with adequate OIT follow-up to ascertain dose progression, this rate was 67.5% (27 participants), which aligns with prior reports for conventional OIT (62.0%–67.2%) (7, 30) and low-dose OIT (74.2%) (14).

During pOIT, some patients experienced mild-to-moderate dosing-related adverse events, including mouth itch, transient abdominal pain, nausea and taste aversion (not systematically analyzed). Anaphylaxis occurred in one patient per subgroup (Table 1). No other protocol-defined severe adverse events, hospitalizations, or treatment discontinuations due to adverse reactions were reported and no cases of eosinophilic esophagitis were diagnosed.

Overall, no statistically significant differences between subgroups were detected in the observed desensitization or severe adverse-event outcomes under this protocol; however, the estimates were imprecise because of the small sample size. Because the subsequent biomarker analyses were designed to characterize longitudinal associations across the treated cohort, and no significant time-by-sensitization-group interaction was detected for SCD trajectories, the subgroups were pooled for exploratory biomarker analyses.

Effective peanut OIT leads to longitudinal suppression of basophil activity

Given the heterogeneous dose-response patterns characteristic of BAT (31) (Supplementary Figure S3), the area under the curve (AUC) was used as the primary metric for basophil activity, as it showed close descriptive concordance with maximum dose response (MDR) (Supplementary Figure S4). To assess whether higher peanut sIgE levels are associated with increased basophil reactivity, baseline peanut sIgE levels were compared with available baseline BAT AUC in 25 patients (Baseline BAT was unavailable in 7 participants because their first BAT measurement occurred after OIT initiation). No significant correlation was observed (p = 0.42) (Supplementary Figure S5A), and baseline basophil activity did not statistically differ between two groups (Table 1). Moreover, baseline BAT AUC was not associated with desensitization outcome (p = 0.76) (Supplementary Figure S5B), and no predictive cutoff could be identified (Supplementary Figure S5C).

To evaluate basophil activation dynamics during pOIT, each patient underwent at least two BAT measurements spaced ≥6 months apart. In contrast to some prior reports (23), our protocol induced a transient increase in basophil activity—measured by peanut-induced %CD63 basophils (AUC and MDR%)—during months 2–6 of the build-up phase, followed by a progressive and significant decline (Table 2, Figure 2A). In mixed-effects analysis accounting for within-participant repeated measurements and multiplicity, BAT activity showed a significant overall change over time (Table 3, p = 7.65 × 10⁻⁷), with statistically significant suppression relative to baseline emerging at 19–24 months (Holm-adjusted p = 0.037) and becoming more pronounced at 25–36 months (Holm-adjusted p = 0.00028). A similar trajectory was observed in the high Ara h 2-sIgE subgroup (Figure 2B, Table 3). It has to be mentioned that one patient had a pre-OIT BAT with an anti-IgE-negative/fMLP-positive pattern consistent with a non-releaser phenotype. This measurement was excluded from all BAT-specific analyses and was not imputed. Subsequent BAT assessments from the same participant obtained during OIT demonstrated anti-IgE-positive/fMLP-positive responses and were therefore included in the analyses.

Table 2. Descriptive longitudinal changes of biomarkers in response to peanut OIT.

SCD: successfully consumed total daily peanut-protein dose; values shown are the maximum SCD recorded within each assessment period.

For descriptive summaries in Table 2 and graphical presentation, sIgE and sIgG4 values exceeding the upper limits of quantification (100 kU/L and 30 mg/L, respectively) were displayed at the assay limits. For inferential longitudinal analyses, these observations were treated as right-censored.

The values from the lab test reports were directly used to calculate the sIgE/sIgG4 [(kU/L)/(mg/L)] ratio.

Table 3. Estimated marginal means from longitudinal mixed-effects models by assessment period.

Estimates were obtained from mixed-effects models with assessment period as a categorical fixed effect and participant-specific random intercepts. Values are back-transformed estimates with 95% confidence intervals. They are not expected to equal the arithmetic means reported descriptively in Table 2. sIgE and sIgG4 were analyzed using right-censored log-normal mixed-effects models. The sIgE/sIgG4 ratio was calculated from reported laboratory values and was analyzed as an exploratory secondary outcome. Global time effects were significant (p < 0.001) for all listed outcomes; follow-up-vs.-baseline contrasts were adjusted using the Holm procedure. The unit for sIgE is kU/L and for sIgG4 is mg/L.

Means.

95% confidence intervals.

BAT serial peanut stimulation revealed a stepwise reduction in CD63⁺ basophils over time. Significant suppression was first observed at 13–18 months at low peanut concentration (100 ng/mL), followed by intermediate (1,000 ng/mL) at 19–24 months, and high concentration (10,000 ng/mL) at 25–36 months (Figures 2D–F). These findings indicate a progressive increase in the basophil activation threshold during pOIT, aligned with the rise in SCD (Figure 1C). Notably, even at 25–36 months—when effective desensitization is achieved—basophils remain responsive but require higher allergen doses for activation.

Despite an overall downward trend following the initial increase, substantial inter-patient variability was observed in both baseline BAT levels and longitudinal response patterns (Figure 2C). BAT AUC at 19–24 months did not linearly correlate with treatment effectiveness. Among participants with evaluable BAT measurements, although 11/25 (44.0%) had a baseline BAT AUC <100 before OIT, BAT AUC ≥100 was observed during the up-dosing phases in 13/15 (86.7%) patients at 2–6 months and 14/17 (82.4%) patients at 7–12 months (Figure 2A, Table 2). When the last available BAT AUC during months 13–36 was compared with the participant's maximum SCD achieved during the same period, a BAT AUC <100 was significantly associated with effective desensitization (SCD ≥300 mg) (Figure 2G) in an exploratory patient-level analysis, with a sensitivity of 89.5% (95% CI, 66.9–98.7%) and specificity of 60.0% (95% CI, 26.2–87.8%). These findings suggest that longitudinal BAT suppression may reflect acquisition of clinical desensitization; however, the same-cohort threshold is an exploratory study-derived threshold. Its performance was not internally validated and is likely optimistic. The low specificity (only 60%) indicated that an AUC <100 alone is insufficient to distinguish patients who achieved effective desensitization. Consistent with this limitation, continuous ROC analysis yielded an AUC of 0.784 (95% CI, 0.536–0.983), with the wide confidence interval reflecting limited precision. The observed cutoff should be considered a hypothesis-generating candidate threshold and requires validation in larger independent prospective studies.

Peanut OIT slowly reduces peanut and Ara h 2/6-sIgE levels

Longitudinal changes of sIgE against peanut and its major components (Ara h 2 and Ara h 6) were evaluated. Consistent with prior reports (23), peanut sIgE levels demonstrated a biphasic pattern, with an early increase during months 2–6 (100% of patients exceeding 50 kU/L for peanut sIgE), followed by a gradual decline at later time points (Table 2, Figures 3A–D). However, the sIgE/total IgE ratio remained generally stable over time, with no significant overall longitudinal time effect (Figure 3B). Mixed-effects analysis confirmed a significant overall effect of time for peanut sIgE, with an early increase followed by a later decline (Table 3). Notably, the rate of decline was slower than that reported in other pOIT protocols (15), and significant reductions from baseline were observed only after 25–36 months (Holm-adjusted p = 0.0063). This delay may be partly attributable to censoring of values >100 kU/L in real-world laboratory measurements. Notably, at 19–24 months, a substantial proportion of patients retained elevated Ara h 2-sIgE despite achieving clinical desensitization (Figure 3C bottom), indicating that serologic IgE decline lagged behind functional and clinical improvement. Peanut sIgE levels at 19–36 months did not correlate with SCD (data not shown).

Peanut- and Ara h 2/6-specific sIgG4 increase early and significantly during pOIT

In contrast to sIgE, sIgG4 levels against peanut and its components increased early and significantly (Table 2 and Figure 4A). The right-censored log-normal mixed-effects model showed a significant increase already at 2–6 months, ratio ≈ 1.96, adjusted p = 0.0095, with progressively larger increases thereafter (Table 3). Notably, at the cohort level, mean peanut sIgG4 increased over time while mean BAT AUC declined, demonstrating opposing longitudinal trajectories (Figure 5). Peanut-specific sIgG4 at 13–24 months did not differ significantly according to baseline BAT category (Supplementary Figure S6).

Allergen-specific IgG4 functions as a blocking antibody by competing with IgE for allergen binding during OIT. Accordingly, the sIgE/sIgG4 ratio may provide a more integrated marker of treatment response. As shown in Figure 4B, this ratio decreased progressively during therapy, with minimum values <50 observed in 13 of 18 participants (72%) at 13–18 months. Although the lowest post-treatment peanut sIgE/sIgG4 ratio did not show a linear correlation with desensitization thresholds (n = 31, R² = 0.1066, p = 0.07), it demonstrated apparent discriminatory performance in ROC analysis (AUC = 0.836, 95% CI 0.695–0.976). An exploratory dichotomous cutoff of 50 yielded a positive predictive value (PPV) of 80.9% and a negative predictive value (NPV) of 60.0% for effective desensitization (SCD ≥300 mg) (Figure 4C). Given the small cohort, its diagnostic performance is likely optimistic. The ratio <50 threshold was exploratory and was evaluated in the same cohort in which it was identified; therefore, it requires independent validation. For clinical applicability, raw laboratory values (kU/L for sIgE, mg/L for sIgG4) were used directly to calculate this ratio without unit conversion (Table 2).

Finally, we performed an exploratory time-lagged analysis to assess whether biomarker measurements at 13–18 months were associated with subsequent successful clinical desensitization (SCD ≥300 mg) at 19–24 months. These intervals were selected based on the longitudinal trajectories observed in our cohort: BAT suppression first became evident during 13–18 months, the sIgE/sIgG4 ratio had substantially declined by this period, and SCD increased substantially during 19–24 months, crossing the 300-mg clinical threshold at the cohort level. For BAT AUC (n = 17), the apparent AUC was 0.788, but the estimate was imprecise, with a wide bootstrap 95% CI 0.54–1.00. The BAT AUC <100 threshold did not significantly discriminate subsequent SCD ≥300 mg (Fisher's exact test, p = 0.102). For the sIgE/sIgG4 ratio (n = 18), the lagged AUC was 0.569 with a bootstrap 95% CI 0.28–0.84, and a ratio <50 did not significantly discriminate subsequent SCD ≥300 mg. Overall, these exploratory time-lagged analyses did not provide sufficiently precise evidence to establish prospective predictive utility.

Discussion

In this retrospective real-world cohort, clinically meaningful increases in ingested peanut dose were observed during a low-dose initiation, slow-escalation pOIT protocol in children with high Ara h 2-sIgE, a population traditionally considered high risk. Observed desensitization proportions and severe adverse-event frequencies did not differ significantly between sensitization subgroups, although the study was not designed or powered to establish equivalence or comparative safety. The percentage of patients who had adequate OIT follow-up and achieved SCD ≥300 mg is consistent with published OIT outcomes. Baseline BAT and sIgE were not associated with clinical responses. In contrast, progressive reductions in basophil activation and shifts in the sIgE/sIgG4 balance consistently paralleled successful dose escalation during treatment. Specifically, longitudinal decreases in BAT AUC (<100) and in the sIgE/sIgG4 ratio (<50) were significantly associated with effective desensitization (SCD ≥300 mg) in exploratory analyses. These findings suggest that longitudinal immune trajectories may provide more informative measures of treatment-associated immune changes than baseline biomarker measurements.

High baseline sIgE has historically been linked to increased adverse reactions, slower desensitization, and higher dropout rates, leading to its use as an exclusion criterion in earlier guidelines (e.g., EAACI 2017) (32). The 40 kU/L Ara h 2-sIgE threshold was selected to characterize marked sensitization based on published clinical data (25) and should not be considered a universal OIT-response threshold. Because Ara h 2 distributions and diagnostic performance may vary across populations and geographic settings, the clinical applicability of this threshold may differ. Our findings therefore primarily support the feasibility of clinically meaningful desensitization under this conservative protocol in patients with high Ara h 2-sIgE, rather than equivalence between sensitization strata. The study also cannot determine whether this protocol reduced risk relative to other OIT approaches. The low-dose initiation, slow escalation, and individualized maintenance approach produced outcomes consistent with prior reports using conventional OIT approaches (7, 9). These findings support the practical value of prioritizing tolerability and gradual dose escalation, although whether such an approach improves long-term adherence or treatment persistence requires prospective evaluation. Real-world adherence and treatment persistence are important determinants of the effectiveness of peanut OIT. Recent real-world experience with food-based OIT has similarly highlighted treatment discontinuation and loss to follow-up as important challenges (33). In our cohort, the observed high desensitization rate should be interpreted in the context of the biomarker-analytic cohort, which required completion of OIT and at least two BAT assessments and may therefore have preferentially excluded patients who discontinued treatment early. Although missed doses and dosing-related reactions were documented during treatment, long-term adherence after the observation period was not systematically assessed.

Peanut OIT induces characteristic immunologic shifts. Allergen-specific IgE typically rises early and then declines gradually during maintenance, but elevated IgE may persist in some patients (19, 23). In our cohort, IgE decline lagged behind clinical and functional improvement, with 45% of participants retaining Ara h 2-sIgE ≥40 kU/L at 19–24 months despite achieving desensitization. In contrast, allergen-specific IgG4 increased early, with increase in peanut and Ara h 6-sIgE evident as early as 2–6 months. Given the ability of IgG4 to competitively inhibit IgE-mediated activation (34), the sIgE/sIgG4 ratio provides an integrated measure of the balance between allergen-specific IgE and IgG4. We observed a progressive decline in this ratio during OIT, and an exploratory ratio <50 was associated with effective desensitization, supporting its potential utility as a practical adjunctive measure of treatment-associated immune change.

Basophil activation provides a functional measure of allergen-induced effector-cell responsiveness. Although prior studies have reported associations between baseline BAT responsiveness and OIT outcomes (22, 23), and that participants with low baseline basophil responsiveness are more likely to achieve treatment success, baseline BAT was not associated with desensitization in our cohort. Longitudinal BAT assessment instead revealed a biphasic pattern, with transient activation during early up-dosing (months 2–6) followed by progressive suppression during extended dosing and maintenance. This trajectory paralleled the gradual increase in tolerated dose and the longitudinal changes observed in humoral biomarkers. Importantly, subsequent progressive reduction in basophil activation, including BAT AUC <100, accompanied by an increased activation threshold, was associated with effective desensitization in an exploratory patient-level analysis. However, the <100 threshold was derived and evaluated in the same cohort and should therefore be considered hypothesis-generating and requires validation in an independent cohort. Longitudinal BAT may thus provide a complementary functional measure of treatment-associated immune modulation, although its clinical utility remains to be established prospectively.

Collectively, these findings highlight fundamental limitations of relying on static baseline biomarkers alone. Rather than demonstrating prospective predictive value, our results indicate that longitudinal changes in BAT and the sIgE/sIgG4 ratio were associated with increasing tolerated dose during pOIT. These dynamic measures may therefore provide adjunctive information about treatment-associated immune changes and complement clinical assessment during dose escalation and maintenance.

These observations have potential clinical implications. If prospectively validated against standardized oral food challenges and in independent cohorts, longitudinal BAT and sIgE/sIgG4 trajectories may provide adjunctive information during clinical assessment of pOIT. Early increases in sIgE and BAT activity were observed in some participants despite subsequent increases in tolerated dose, indicating that early biomarker changes should not be interpreted in isolation. Continued longitudinal assessment may provide a more informative context for interpreting these dynamic immune changes during treatment and for counseling patients and families.

Limitations

This study has several limitations. Its single-center, retrospective design and modest sample size may limit generalizability and statistical power for subgroup analyses. It lacked an untreated, placebo, or alternative OIT-protocol comparator; therefore, the contribution of the low-dose, slow-escalation strategy itself cannot be isolated from the effects of OIT more generally. Eligibility required at least two BAT assessments separated by ≥6 months, which may have preferentially excluded patients who discontinued treatment early and could introduce selection bias. Biomarker availability varied across assessment periods. No imputation was performed, and mixed-effects analyses used all available observations. Clinical desensitization was defined using the successfully consumed dose during routine care rather than a standardized blinded oral food challenge; therefore, the candidate biomarkers were not validated against the diagnostic gold standard.

The BAT assay was developed by study authors, and although laboratory and clinical data generation incorporated internal blinding and separation of responsibilities, no independent external reanalysis was performed. AUC measurement uncertainty remains unquantified. Dichotomous cutoff values (AUC <100, sIgE/sIgG4 ratio <50) are exploratory and require validation in large independent cohorts. In addition, patient-reported outcomes, including quality-of-life measures, were not systematically collected and represent an important area for future studies. Long-term adherence was not systematically assessed, and sustained unresponsiveness after treatment discontinuation was not evaluated. Finally, the exclusion of children under 4 years leaves unanswered questions regarding biomarker utility in the population most likely to benefit from disease-modifying intervention. Prospective studies are needed to determine whether longitudinal biomarker assessment provides clinically useful information and whether this OIT approach improves safety, adherence, or treatment outcomes relative to alternative protocols.

The calendar-time assay-control analysis also has limitations. Available longitudinal control data covered January 2024 through July 2026 and therefore did not include the earlier study period from September 2021 through December 2023. In addition, the linear trend analysis was designed to detect an overall monotonic temporal trend and may not capture nonlinear variation in control responses. Finally, anti-IgE and fMLP positive controls assess basophil responsiveness and general assay performance but do not directly detect lot-specific changes in the peanut stimulation reagent.

Conclusions

In this retrospective cohort, clinically meaningful increases in ingested peanut dose were observed in participants with both high and lower baseline Ara h 2-sIgE, although equivalence between sensitization subgroups was not established. Longitudinal suppression of basophil activation and reductions in the sIgE/sIgG4 ratio were associated with higher SCD, whereas baseline biomarkers and exploratory time-lagged analyses did not establish prospective predictive utility. These findings support further prospective evaluation of BAT and antibody trajectories as adjunctive biomarkers during pOIT, with validation against standardized oral food challenges and in independent cohorts required before clinical implementation.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/falgy.2026.1902025/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Péter Csonka, Tampereen yliopisto—Kaupin Kampus, Finland

Reviewed by: Benedetta Pessina, Azienda Ospedaliera Universitaria Meyer IRCCS—Firenze, Italy

Adnan Al Ali, United Arab Emirates University, United Arab Emirates