Kazım Okan Dolu

Background and Aim: Familial Mediterranean Fever (FMF) is an autoinflammatory disorder caused by mutations in the MEFV gene, leading to dysregulated inflammatory responses. Persistent immune activation in FMF may increase susceptibility to allergic diseases, such as asthma, allergic rhinitis (AR), and atopic dermatitis (AD). This study aimed to compare the prevalence of allergic diseases and aeroallergen sensitization in children with FMF and age-matched healthy controls.Materials and Methods: This prospective cohort study included 183 children with FMF who were in clinical remission and receiving regular colchicine therapy and 105 age-matched healthy controls. The FMF group consisted of 112 males (61.2%) and 71 females (38.8%). All participants were evaluated by pediatric allergy specialists. Diagnoses of asthma, AR, and AD were made according to current clinical guidelines. Aeroallergen sensitization was assessed using skin prick testing serum-specific IgE measurements. Demographic characteristics, eosinophil counts, total IgE levels, and use of anti-allergic medications were recorded.Results: The prevalence of asthma (41.8% vs. 25.0%; p = 0.002) and AR (53.0% vs. 35.0%; p<0.001) were more common in the FMF group, whereas AD prevalence did not differ. Aeroallergen sensitization rates (43.0% vs. 42.0%; p = 0.815), eosinophil counts, and total IgE levels were similar.Conclusion: Children with FMF have an increased prevalence of asthma and AR. Similar sensitization rates despite higher clinical allergic disease suggest that chronic inflammation and immune dysregulation beyond IgE-mediated mechanisms may contribute to this association.

Donor human milk is the recommended alternative when a mother's own milk is unavailable, especially for high-risk infants. Türkiye has no operational human milk bank, and the attitudes of actively breastfeeding mothers are poorly characterised. To describe the knowledge, attitudes and concerns of actively breastfeeding mothers regarding human milk banks and to identify factors associated with their willingness to recommend one. In this single-centre, cross-sectional study, 310 actively breastfeeding mothers with infants aged 0-6 months were recruited by consecutive sampling at a training and research hospital (March-May 2025). Data were collected using an interviewer-administered, 32-item questionnaire that was content-validated and piloted (not a formally validated scale). Associations were assessed with chi-square tests and binary logistic regression reporting adjusted odds ratios (ORs) with 95% confidence intervals (CIs). Only 37.1% of mothers were willing to recommend a human milk bank, 40.3% were unwilling and 22.6% undecided. Educational level was the only independent predictor of willingness to recommend (adjusted OR 1.59, 95% CI 1.15-2.20, p = 0.005); knowledge and employment were not significant after adjustment. Knowledge of milk banking was higher among mothers who were not employed (44.5% vs 14.5%, p < 0.001). Religious and moral concerns were prominent among those opposing a milk bank (21.9%, and the distribution of concerns differed significantly according to mothers' views on establishment (p < 0.001). Willingness to recommend was low but broadly consistent with international figures. Higher education was associated with greater willingness yet appears necessary rather than sufficient, as many reservations are religious or cultural. Expanding milk banking in Türkiye will require culturally sensitive engagement alongside public-awareness activities.

Objective: This study aimed to systematically evaluate pediatric patients referred with suspected penicillin allergy and determine the rate of confirmed penicillin allergy.Materials and Methods: A retrospective study was conducted on 57 patients who presented to the Pediatric Allergy and Immunology Department of Istanbul Kanuni Sultan Süleyman Training and Research Hospital with suspected penicillin allergy between August 2023 and August 2025. All patients underwent specific IgE testing for amoxicillin, ampicillin, penicillin G, and penicillin V. Skin testing was performed in patients with parenteral drug reactions. Oral drug provocation testing was conducted in patients with negative specific IgE and skin tests.Results: The mean age was 7.4±4.9 years (median: 6 years). The most common suspected drug was amoxicillin-clavulanate (77.2%). Urticaria/angioedema (61.4%) was the most frequent reaction type. Specific IgE positivity was detected in only 3 patients (5.3%). Drug provocation testing was performed in 53 patients, with 51 patients (96.2%) tolerating the drug without complications and 2 patients (3.8%) showing positive reactions. Penicillin allergy was confirmed in 6 patients (10.5%), whereas it was excluded (delabeled) in 51 patients (89.5%).Conclusion: Our study demonstrated that 89.5% of penicillin allergy labels were incorrect. The low sensitivity of specific IgE testing was confirmed. Direct drug provocation testing is safe and effective in low-to-moderate-risk patients.

Background/Objectives: We aimed to evaluate the etiologic distribution of markedly elevated IgE (≥2000 IU/mL) and the association of parasitic infections with longitudinal IgE dynamics and National Institutes of Health Hyper-IgE Syndrome (NIH-HIES) score trajectories in pediatric patients. Methods: We retrospectively enrolled 127 pediatric patients with IgE ≥2000 IU/mL and ≥1 year of longitudinal follow-up at a tertiary allergy clinic (2019–2024). The primary outcome was the IgE reduction percentage difference between parasitic and non-parasitic groups. Results: Atopic sensitization was identified in 81.0% of tested patients; genetic evaluation was performed in 40 patients (31.5%), with confirmed IEI in one patient (0.8% of the cohort; 2.5% of those evaluated). Parasitic infections were present in 24.4% of patients (n = 31; intestinal parasites 15.7%, scabies 8.7%, hydatid cyst 1.6%; two patients had concurrent intestinal parasitosis and scabies). Median IgE reduction was 63.8% vs. 27.0% in parasitic versus non-parasitic groups (p < 0.001), persisting after multivariable adjustment (p < 0.001). Parasitic infection independently predicted IgE normalization below 2000 IU/mL (OR = 8.26; 95% CI: 2.76–24.68; p < 0.001). All three NIH-HIES inflammatory components (IgE, eosinophilia, eczema) regressed more often in the parasitic group (all p ≤ 0.01); no patient reached the ≥40-point HIES threshold. Conclusions: Parasitic infections produce a clinical phenotype overlapping with hyper-IgE syndrome, constituting a reversible phenocopy of primary immune dysregulation. In populations with substantial parasitic prevalence, parasitological evaluation may be a useful consideration in children with markedly elevated IgE and indeterminate clinical scores; however, this approach should complement rather than replace comprehensive clinical assessment.

I read the article by Xu et al. with great interest. Identifying reliable in vitro markers to reduce the burden of oral food challenges (OFC) is a priority in pediatric allergy. However, I have identified several fundamental statistical concerns that, in my view, may affect the validity of the primary conclusions.The Independence Problem (Pseudoreplication)The most striking issue is the mismatch between the number of subjects (n=83) and the number of challenges (n=209). It appears that each challenge was treated as an independent data point. In clinical research, multiple observations from the same child are inherently correlated. By failing to account for this clustering, the standard errors are likely underestimated, and the p-values may be artificially small. Without employing Generalized Estimating Equations (GEE) or mixedeffects models to adjust for within-subject correlation, the reported diagnostic associations remain statistically questionable.The authors conclude that the sIgE/tIgE ratio is "better" than sIgE alone for the total sample and egg white groups. However, their own ROC analysis-as documented in the results and Figure 1-states that there was no significant difference between the AUCs of the two methods. If the difference between two diagnostic markers is not statistically significant, one cannot definitively claim the superiority of one over the other. This creates a contradiction between the data presented and the final inference.In Table 3, the authors include both sIgE and the sIgE/tIgE ratio within the same multifactorial model. Since the ratio is a direct mathematical derivative of sIgE, these variables are highly collinear. Such multicollinearity can make the regression model unstable and may explain why sIgE loses its significance only when the ratio is introduced into the model.I believe these methodological issues, particularly the lack of adjustment for clustered data and the interpretation of non-significant AUC differences, warrant further clarification. A corrected analysis is essential to ensure that the proposed cut-off of 10.42 can be safely and accurately implemented in clinical practice.KOD: Conceptualization, formal analysis, writing -original draft, writing -review and editing.

I read with great interest the article by Yorulmaz et al

Yang et al. developed a machine learning-based nomogram to predict airway hyperresponsiveness (AHR) among patients with suspected asthma using baseline pulmonary function parameters (1). While the clinical aim is relevant, I identified several internal inconsistencies in statistical reporting that currently limit interpretability and reproducibility of the model and warrant formal correction.Table 2 reports the Model C coefficient for MMEF75-25%pred as -0.31 with a 95% confidence interval (CI) of (-0.06, -0.006) (1). By statistical definition, a 95% CI must contain its point estimate; here the reported interval does not include the coefficient. This appears to be a reporting error (e.g., transcription of the coefficient and/or CI) and directly affects the nomogram because coefficient weights determine individual risk calculations. I request confirmation and correction of the exact coefficient and its 95% CI for this predictor, alongside any downstream updates required for the nomogram and Figure 4.The Results section reports an AUC of 0.790 for the training cohort while also presenting a 95% CI of (0.724-0.760) (1). This interval does not include the stated AUC value, and it is presented adjacent to the 'optimal cut-off' value (0.354), creating ambiguity as to whether the interval pertains to AUC or another quantity. Similar ambiguity appears for the validation cohort (AUC = 0.756; CI: 0.600-0.814) (1). I ask the authors to provide the correct AUC estimates and their corresponding 95% CIs for both cohorts, and to clearly label which CI belongs to which metric. Figure 5 should also be updated to reflect corrected values.The Methods describe random splitting into training/validation groups and the use of multiple imputation by chained equations (MICE) for missing values, with predictors including 'all variables in the analysis' (1). However, the manuscript does not explicitly state whether imputation was performed strictly within the training set (and then applied to the validation set) or on the combined dataset prior to splitting. Because pre-split imputation can introduce information leakage that compromises the independence of internal validation, I request explicit clarification of the exact sequence (i.e., split → impute within training only → apply to validation, or an alternative approach). If a different sequence was used, the authors should discuss potential implications for the reported validation performance (AUC = 0.756).Given the statistical inconsistencies documented above-particularly the coefficient-CI mismatch and the AUC-CI reporting ambiguity-I respectfully request that the journal and authors issue a formal Correction/Addendum to: (i) correct the Table 2 coefficient and CI for MMEF75-25%pred in Model C, (ii) provide accurate AUC values and correctly labeled 95% CIs for both cohorts in the text and Figure 5, and (iii) clarify imputation timing relative to data splitting. These corrections are essential to ensure accurate interpretation and clinical implementation of the proposed model. To facilitate reproducibility, I also encourage sharing analysis code and, where feasible, de-identified data via a public repository.KOD: Conceptualization, formal analysis, writing -original draft, writing -review and editing.