Prediction of duodenal ulcers in children with Helicobacter pylori infection Authors: Siyuan Zheng, Chuchu Zheng, Shijie Cai, Xiaoyu Ge, Yan Chen, Weijie Ou, Yijun Lin Published: 2025 ABSTRACT AbstractBackgroundHelicobacter pylori (HP) infection is a major cause of duodenal ulcers (DUs) in children. DUs can lead to gastrointestinal bleeding and even intestinal perforation, both of which may be life-threatening. This study aimed to establish a predictive model for identifying DUs in children with HP infection, in order to assist primary care pediatricians with early diagnosis and intervention.MethodsA total of 456 children with HP infection admitted to Fujian Children’s Hospital and Fujian Maternal and Child Health Hospital were retrospectively included. The training group comprised 307 patients from Fujian Children’s Hospital, while the validation group consisted of 149 patients from Fujian Maternal and Child Health Hospital. The patients in the training group were categorized into the HP infection group or HP infection with DU group. Firstly, least absolute shrinkage and selection operator regression analysis was conducted, and the selected variables were subjected to logistic regression analysis to identify the predictors of DUs in children with HP infection. Finally, a nomogram was established for the model.ResultsSex, melena, hemoglobin level, and uric acid level were included in the final prediction model. The area under curve values for the internal training and external validation groups were 0.740 and 0.728, respectively. The Hosmer-Lemeshow tests and calibration curves indicated that the degree of fit and consistency of the model were satisfactory. These decision curve analyses indicated that the model has certain clinical application value.ConclusionsIf male children with melena, decreased hemoglobin levels and high uric acid levels are clinically encountered, DUs require high vigilance, and the children should be promptly referred to a facility with pediatric endoscopic capabilities for further diagnosis and treatment.Keywords: Children, Helicobacter pylori (HP), duodenal ulcer (DU), nomogram, prediction model Highlight box Key findings • A predictive model that included male sex, melena, hemoglobin levels and uric acid levels for duodenal ulcers (DUs) in children with Helicobacter pylori (HP) infection for primary care physicians was established. What is known and what is new? • It is known that HP infection is a major cause of DUs in children. DUs can cause gastrointestinal bleeding in children and even gastrointestinal perforation, which can be life-threatening. • A predictive model that included basic and widely accessible indices for DUs in children with HP infection for primary care physicians was established. What is the implication, and what should change now? • The model serves as a practical, non-invasive tool to help physicians stratify risk and identify high-risk children who need further investigation. Primary care pediatricians should now maintain a high level of vigilance for DUs when male children with HP infection present with a combination of melena, decreased hemoglobin levels, and elevated uric acid levels. These patients should be promptly referred to a facility with pediatric endoscopic capabilities for definitive diagnosis and treatment. Introduction In recent years, the incidence of Helicobacter pylori (HP) infection in children has significantly decreased in several affluent European nations, and it is even less than 10% in some countries. Conversely, HP infection rates remain high in many developing countries (1). A large-scale inquiry conducted in China in 2023 revealed that the incidence of HP infection among children and adolescents was 20.55% (2), indicating that this nation continues to face the persistent challenge of elevated HP infection rates in pediatric populations. Many children who undergo endoscopic examination for gastrointestinal bleeding, anemia, or abdominal pain are frequently found to have HP infection. HP infection may be linked to duodenal ulcers (DUs), which represent a severe complication in pediatric patients that has the potential to lead to gastrointestinal perforation and even mortality if not addressed promptly. Therefore, timely detection and treatment of DUs are crucial. Endoscopy remains the most definitive method for diagnosing DUs (3). Pediatric guidelines recommend gastroscopy for children with clinical indications (4), especially for those who also have a concomitant HP infection. However, most primary health care institutions in the Asia-Pacific region are unable to conduct endoscopic procedures on children (5). There are several reasons: Firstly, the endoscopists at primary health care institutions often lack sufficient experience in performing such examinations on pediatric patients. Secondly, pediatric endoscopic procedures also typically necessitate general anesthesia. Since most primary health care institutions do not perform pediatric surgery, anesthesiologists may be inadequately trained in managing the specific risks associated with anesthesia, such as determining appropriate anesthetic agent dosage for children (6). Third, guardians express concerns regarding complications arising from both the endoscopic procedure and general anesthesia. In addition, some school-aged children need referral to a higher-level hospital far from their hometown for gastroscopy, which disrupts school functioning and creates a financial burden for the family, ultimately leading to refusal of the procedure. Because of these different factors, gastroscopy in children is often difficult to perform in primary health care institutions. The aim of this study was to characterize the clinical manifestations of children with HP infection at two tertiary pediatric hospitals and to develop and validate prediction model of active (A-stage) or healing (H-stage) DUs. Children with a high likelihood of DUs should be promptly referred by primary pediatricians to institutions with pediatric endoscopy capability. We present this article in accordance with the TRIPOD reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-557/rc). Methods Participants This retrospective study included children who were admitted to Fujian Children’s Hospital and Fujian Maternal and Child Health Hospital for their first gastroscopy between January 1, 2018 and December 31, 2023. The inclusion criteria were as follows: (I) diagnosis of HP infection that was confirmed by satisfying any one of the following two criteria (7): primary diagnosis of HP infection by invasive gastric biopsy-oriented methods, such as the following: (i) positive bacterial culture or (ii) histopathological manifestations of HP gastritis, as evaluated by the updated Sydney classification, need to be supported by no less than an additional positive test, such as a rapid urease test or available molecular-based assays; and (II) age between 1 and 14 years old. The exclusion criteria were as follows: (I) abdominal pain primarily attributable to nongastrointestinal diseases; (II) severe multiorgan failure; (III) DUs resulting from other gastrointestinal conditions, including, but not limited to Henoch-Schoenlein purpura, hormone therapy, nonsteroidal antiinflammatory drugs (NSAIDs) use, inflammatory bowel disease, eosinophilic gastroenteritis, lymphoma, Behcet’s disease, trauma-induced stress responses, chemotherapy medications, and hereditary metabolic disorders; (IV) children with gastric ulcers, small intestine ulcers, and colonic ulcers; (V) digestive symptoms stemming from infections caused by other types of pathogens such as cytomegalovirus or Epstein-Barr virus; (VI) children with DUs at the scar stage (S-stage); and (VII) incomplete patient data. Following the application of these criteria, 456 patients were incorporated into the final analysis. Children from Fujian Children’s Hospital composed the training group, whereas the validation group included children from Fujian Maternal and Child Health Hospital. Children with A-stage or H-stage DUs were placed in the HP + DU group, whereas children without DUs were placed in the HP group. This study was approved by Ethics Committee of Fujian Children’s Hospital (approval No. 2022ETKLRD10033). The other institution was informed and agreed with the study. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Informed consent was waived by the ethics committee. Data collection The subsequent information from the electronic medical record system were extracted upon admission as potential predictors. To ensure the model applicability for primary care physicians, basic and widely accessible indicators were collected. The baseline demographics included sex, age, body mass index (BMI), family history of HP infection, family history of digestive diseases (excluding HP infection), history of digestive system diseases, and season of hospital admission for patients were extracted. The clinical characteristics included loss of appetite, vomiting, constipation, weight loss, heartburn, chest pain, melena, diarrhea, mesenteric lymph node size, platelet count (PLT), neutrophil count (NE), lymphocyte count (LYM), leukocyte count (WBC), neutrophil/lymphocyte ratio (NLR), platelet/lymphocyte ratio (PLR), hemoglobin (Hb), red cell distribution width (RDW), C-reactive protein (CRP), gastrin-17 (G-17), triglycerides (TG), total cholesterol (TC), uric acid (UA), serum creatinine (Scr), blood urea nitrogen (BUN), amylases (AMY), lipase (LPS), and the erythrocyte sedimentation rate (ESR). Statistical analysis Statistical analyses were performed with SPSS 26.0 (IBM Corp., Armonk, NY, USA) and R software (version 4.1.3; R Foundation for Statistical Computing, Vienna, Austria). Enumeration variables are presented as frequencies (cases, %), and contrasts among different groups were carried out via the Chi-squared test. With respect to variables that passed the normality test, measurement data are presented as the means ± standard deviations, and intergroup comparisons were performed via t-tests. Variables not normally distributed are represented by the median and interquartile range (IQR), with the Mann-Whitney U test employed for intergroup comparisons. These initial univariable analyses were conducted for descriptive purposes to characterize the study cohort. To prevent overfitting and potential collinearity among the measured indicators of the same child, least absolute shrinkage and selection operator (LASSO) regression analysis was applied in the training group to identify nonzero regression coefficients as relevant factors, followed by logistic regression to develop the relevant prediction model, which was grounded in which a nomogram was drawn. The validation and evaluation of the model mainly relied on receiver operating characteristic (ROC) curves. The Hosmer-Lemeshow test was used to evaluate the goodness of fit, and the calibration curve was used to examine the consistency between the predicted and actual values. The efficacy of our model in clinical practice was evaluated via decision curve analysis (DCA). P<0.05 was considered statistically significant. Results Flow chart of the study process Children who underwent gastroscopy and were admitted to Fujian Children’s Hospital or Fujian Maternal and Child Health Hospital between January 1, 2018 and December 31, 2023 were included. Following the elimination of 1,942 ineligible patients, 456 patients were incorporated into the final analysis. Among them, 307 patients from Fujian Children’s Hospital composed the internal training group, and 149 patients from Fujian Maternal and Child Health Hospital composed the external validation group. The flowchart illustrating the study process is depicted in Figure 1. Flow chart of the study process. DCA, decision curve analysis; HP, Helicobacter pylori infection group; HP + DU, Helicobacter pylori infection with duodenal ulcer group; LASSO, least absolute shrinkage and selection operator; ROC, receiver operating characteristic. Comparison of baseline demographics between the internal training and external validation groups According to the analysis results, the percentage of patients with DUs in the training group (34.9%) was lower than that in the validation group (63.1%) (P<0.001). The median age of the participants in the training group was 9 years, and the range was 7 to 11 years. This age was significantly greater than that in the validation group, which median age was 8 years, and the range was 5 to 10 years (P<0.001). The prevalence of children with a history of digestive system diseases in the training group (11.1%) was lower than that in the validation group (22.8%) (P<0.001). There were no marked differences in sex (P=0.09), BMI (P=0.26), family history of HP infection (P=0.53), or family history of digestive diseases (P=0.27) between the two groups (Table 1). Table 1Comparison of baseline demographics between internal training group and external validation groupCharacteristicsTraining group (n=307)Validation group (n=149)Z/χ2PCombined with DU107 (34.9)94 (63.1)32.441<0.001Sex2.9250.09 Male187 (60.9)103 (69.1) Female120 (39.1)46 (30.9)Age, years9 (7, 11)8 (5, 10)−5.062<0.001Age-based stratification, years27.944<0.001 >1341 (13.4)8 (5.4) 10–12106 (34.5)36 (24.2) 7–9107 (34.9)48 (32.2) <753 (17.3)57 (38.3)BMI, kg/m216.09 (14.36, 19.11)15.75 (14.11, 18.90)−1.1340.26Family history of HP infection274 (89.3)130 (87.2)0.3980.53Family history of digestive diseases128 (41.7)54 (36.2)1.2430.27History of digestive system diseases34 (11.1)34 (22.8)10.904<0.001Data are presented as n (%) or median (IQR). BMI, body mass index; DU, duodenal ulcer; HP, Helicobacter pylori; IQR, interquartile range. Comparison of baseline demographics between internal training group and external validation group Data are presented as n (%) or median (IQR). BMI, body mass index; DU, duodenal ulcer; HP, Helicobacter pylori; IQR, interquartile range. Comparison of baseline demographics and clinical characteristics between the HP and HP + DU groups in the training group Among the 307 children included in the training group, 200 were included in the HP group, whereas 107 were included in the HP + DU group. Sex (P<0.001), age (P=0.002), age-based stratification (P=0.02), BMI (P=0.006), melena (P<0.001), PLT (P=0.01), PLR (P=0.02), Hb (P=0.008), RDW (P=0.009), TC (P=0.04), UA (P=0.01) and Scr (P=0.03) between the HP and the HP + DU groups are significantly differed. The other indices did not significantly differ between the groups (P>0.05). The detailed results are presented in Table 2. Table 2Comparison of baseline demographics and clinical characteristics between HP group and HP + DU group in the internal training groupCharacteristicsHP group (n=200)HP + DU group (n=107)Z/χ2PSex21.350<0.001 Male103 (51.5)84 (78.5) Female97 (48.5)23 (21.5)Age, years9 (7, 11)10 (8, 12)−3.1110.002Age-based stratification, years9.5730.02 >1323 (11.5)18 (16.8) 10–1261 (30.5)45 (42.1) 7–974 (37.0)33 (30.8) <742 (21.0)11 (10.3)BMI, kg/m215.76 (14.10, 18.62)16.46 (15.15, 19.70)−2.7680.006Family history of HP infection180 (90.0)94 (87.9)0.3360.56Family history of digestive diseases83 (41.5)45 (42.1)0.0090.93History of digestive system diseases24 (12.0)10 (9.3)0.4990.48Loss of appetite60 (30.0)31 (29.0)0.0350.85Season of hospital admission6.9240.07 Spring43 (21.5)36 (33.6) Summer72 (36.0)38 (35.5) Autumn50 (25.0)17 (15.9) Winter35 (17.5)16 (15.0)Vomiting47 (23.5)28 (26.2)0.2690.60Constipation37 (18.5)24 (22.4)0.6760.41Melena10 (5.0)23 (21.5)19.770<0.001Diarrhea26 (13.0)12 (11.2)0.2050.65Weight loss18 (9.0)9 (8.4)0.0300.86Heartburn13 (6.5)3 (2.8)1.9280.17Chest pain11 (5.5)1 (0.9)2.7480.10The size of the mesenteric lymph node size, mm1.40 (1.18, 1.70)1.50 (1.15, 1.75)−1.0390.30PLT, ×109/L297.00 (249.75, 354.25)316.00 (270.50, 387.50)−2.5750.01NE, ×109/L3.62 (2.78, 5.12)3.75 (2.82, 5.89)−0.9710.33LYM, ×109/L2.87 (2.37, 3.51)2.86 (2.34, 3.38)−0.0920.93WBC, ×109/L7.22 (5.79, 9.19)7.60 (6.11, 9.03)−0.8120.42NLR1.25 (0.92, 1.80)1.36 (1.02, 1.97)−1.1160.27PLR100.38 (81.87, 128.50)120.60 (85.41, 147.26)−2.2830.02Hb, g/L129.00 (122.00, 136.00)124.00 (102.50, 135.00)−2.6400.008RDW, %40.90 (39.20, 42.50)42.00 (39.60, 44.60)−2.6270.009C-reactive protein, mg/L2.93 (1.64, 4.93)2.92 (1.84, 4.46)−0.0890.93G-17, pmol/L20.09 (12.56, 35.01)17.00 (9.17, 35.30)−1.0630.29TG, mmol/L0.90 (0.60, 1.33)0.80 (0.60, 1.10)−1.8190.07TC, mmol/L4.00 (3.60, 4.50)3.80 (3.35, 4.40)−2.0190.04UA, μmol/L280.50 (239.75, 326.50)301.00 (251.00, 370.00)−2.4860.01Scr, μmol/L40.00 (34.00, 46.00)42.00 (36.50, 48.50)−2.1490.03BUN, mmol/L4.40 (3.70, 5.10)4.50 (3.70, 5.50)−1.2370.22AMY, U/L70.00 (53.75, 93.10)70.00 (52.50, 91.00)−0.2460.81LPS25.00 (18.00, 36.00)23.00 (16.00, 33.00)−1.3890.17ESR, mm/h9.75 (6.00, 18.33)9.00 (5.00, 15.50)−0.9980.32Data are presented as n (%) or median (IQR). AMY, amylases; BMI, body mass index; BUN, blood urea nitrogen; DU, duodenal ulcer; ESR, erythrocyte sedimentation rate; G-17, gastrins-17; Hb, hemoglobins; HP, Helicobacter pylori; IQR, interquartile range; LPS, lipase; LYM, lymphocyte count; NE, neutrophils count; NLR, neutrophil/lymphocyte ratio; PLR, platelet/lymphocyte ratio; PLT, platelet count; RDW, red cell distribution width; Scr, serum creatinine; TC, total cholesterol; TG, triglycerides; UA, uric acid; WBC, leukocyte count. Comparison of baseline demographics and clinical characteristics between HP group and HP + DU group in the internal training group Data are presented as n (%) or median (IQR). AMY, amylases; BMI, body mass index; BUN, blood urea nitrogen; DU, duodenal ulcer; ESR, erythrocyte sedimentation rate; G-17, gastrins-17; Hb, hemoglobins; HP, Helicobacter pylori; IQR, interquartile range; LPS, lipase; LYM, lymphocyte count; NE, neutrophils count; NLR, neutrophil/lymphocyte ratio; PLR, platelet/lymphocyte ratio; PLT, platelet count; RDW, red cell distribution width; Scr, serum creatinine; TC, total cholesterol; TG, triglycerides; UA, uric acid; WBC, leukocyte count. LASSO regression for the selection of predictors for DU in children with HP infection in the training group This study incorporated all 34 potential influencing variables into the LASSO regression analysis. The 10-fold cross-validation technique was used to identify the optimal lambda parameter. The parameter 1se was chosen as the best fit for the final model (Figure 2A,2B). Drawing on this parameter, the number of variables exhibiting nonzero regression coefficients was statistically assessed. The LASSO regression results indicated that five factors (i.e., sex, age, melena, Hb level and UA level) significantly predicted DUs in children with HP infection. LASSO regression screening process for predictors. (A) The process of selecting the most suitable λ in the LASSO regression model. The left dashed line denoted the optimal lambda value (lambda.min) based on the evaluation metric. The right dashed line indicated the lambda value (lambda.1se) for the model where the evaluation metric was within one standard error of the optimal value. (B) LASSO regression screening variable coefficient curve. Each line in the graph corresponded to the coefficient variation of a particular variable as it changed with the lambda coefficient (after taking the logarithm) of the LASSO penalty term. LASSO, least absolute shrinkage and selection operator. Logistic regression to build DUs prediction models for children with HP infection in the training group Taking these five predictors as independent variables, a logistic regression prediction model was constructed. Four independent predictors of DUs in children with HP were identified, including: being of the male sex (P<0.001), melena (P=0.047), Hb level (P=0.002) and UA level (P=0.001) (Table 3). Table 3Logistic regression analysis for DU in children with HP infectionCharacteristicsStandard errorWaldPOR95% CISex0.30214.503<0.0010.3170.175, 0.573Melena0.4962.9200.0472.3321.013, 6.161Hb0.00910.0210.0020.9710.953, 0.989UA0.00210.8980.0011.0061.002, 1.010Constant2.0860.1780.670.415–CI, confidence interval; DU, duodenal ulcer; Hb, hemoglobin; HP, Helicobacter pylori; OR, odds ratio; UA, uric acid. Logistic regression analysis for DU in children with HP infection CI, confidence interval; DU, duodenal ulcer; Hb, hemoglobin; HP, Helicobacter pylori; OR, odds ratio; UA, uric acid. Nomogram for DUs in children with HP infection in the training group Sex, melena, and Hb and UA levels were used as predictors, and the presence of DUs was used as an outcome variable in this model. A nomogram for DUs in children with HP infection was drawn (Figure 3). Nomogram for DUs in children with HP infection. On the left side of the figure, variable names were listed, while on the right side, each variable was represented by its fractional coordinate conversion. The length of the line segment reflected the magnitude of each variable’s contribution to the outcome event. “Points” represented scores ranging from 0 to 100. This indicated the individual scores corresponding to each variable in different groupings within the model. The “Total points” scale corresponded to the aggregate score, which was the sum of individual scores assigned to all variables. “Linear predictor” denoted the linear predictor value. “Risk” represented the probability of the outcome occurring, converted from the score. In the variable “Sex”, 1 represents male and 2 represents female. In the variable “Melena”, 1 represents the presence of melena and 0 represents the absence of melena. DU, duodenal ulcer; HP, Helicobacter pylori. Validation of the DU prediction model in children with HP infection The area under curves (AUCs) for the internal training and external validation groups were 0.740 [95% confidence interval (CI): 0.679–0.800; Figure 4A] and 0.728 (95% CI: 0.647–0.808; Figure 4B), respectively, indicating that the model exhibited robust discriminative ability and predictive efficiency. Validation of the model for DUs in children with HP infection. (A) The ROC curve for the training group. (B) The ROC curve for the validation group. (C) The calibration curve of the training group. (D) The calibration curve of the validation group. The “Apparent” curve represented the prediction curve, the “Bias corrected” curve represented the calibration curve, and the “Ideal line” represented the ideal curve. (E) The DCA curve of the training group. (F) The DCA curve of the validation group. The “purple curve” represented the change in the model’s net benefit as the threshold for high-risk probability varies. The area between the “no treatment line” (green curve) and the “full treatment line” (red curve) in the model curve reflected the clinical utility of the model. AUC, area under curve; CI, confidence interval; DCA, decision curve analysis; DU, duodenal ulcer; FPR, false positive rate; HP, Helicobacter pylori; ROC, receiver operating characteristic; TPR, true positive rate. The Hosmer-Lemeshow tests revealed that for the training group, χ2=11.241 (P=0.19), and for the validation group, χ2=6.841 (P=0.55). This finding indicated that the model yielded a strong fit. The calibration curve further confirmed a high degree of consistency between the predicted risk and actual risk, indicating that the prediction model strongly aligned with the actual probability of the event occurring (Figure 4C,4D). To evaluate the clinical therapeutic outcome of the model, we plotted DCA curves to further investigate the clinical benefit of the nomograms for DUs in children with HP infection. In Figure 4E,4F are the DCA results of the prediction model applied to the training and validation groups. According to the DCA curves, the two groups showed wide applicability of the high-risk threshold, which indicated that the model had significantly clinical applications. Discussion Our study collected data from two tertiary pediatric hospitals and developed a model for predicting DUs in children with HP infection that included male sex, melena, Hb levels and UA levels. To ensure the model applicability for primary care physicians, basic and widely accessible indicators were chosen, and endoscopy and pathological examination ensured the accuracy of the diagnoses for HP infection and DU. Since the number of cases with other types of ulcers at our hospital was small, including them in the analysis could have resulted in heterogeneity, potentially affecting the relevance of predictive indicators and the performance of the model. Therefore, these cases were excluded from our study. Children with DUs in the S-stage were also excluded from enrollment. In our opinion, most S-stage ulcers do not require specific ulcer-related treatment, nor do they necessitate urgent referral from primary healthcare facilities to hospitals with endoscopic capabilities. Moreover, S-stage ulcers usually no longer have active inflammation, and their clinical symptoms are often mild or difficult to distinguish from those in children without ulcers. Notably, the detection rate of DU in the training group in our study was 34.9%, much higher than that reported in other regions (5.1–19%) (8-10). This may be attributed to the higher proportion of males included and the relatively more severe cases in our cohort. The prediction model was thoroughly assessed and validated, showed strong external validation and discriminatory capacity. The results of this study provide valuable insights for primary health care facilities. Specifically, it was revealed that, among children with HP infection, the prevalence of DUs in male children (78.5%) was significantly greater than that in female children (21.5%). These findings align with those of a prior one conducted in Taiwan, in which the male-to-female ratio among children diagnosed with DUs was 2.6:1, and the sex ratio of HP-positive children with DUs reached 3.7:1 (11). The research conducted in the United States of America examining the prevalence of peptic ulcers among HP-positive children also showed that female sex is a protective factor against DUs (12). It was indicated that males exhibit greater basal gastric acid secretion; gastric acid erosion of mucosal tissue can lead to the development of DU, which could explain the greater incidence of DU in male children (13). Furthermore, in some studies, a lower occurrence of DUs in female children was attributed to the effects of estrogen. Estrogen is shown to enhance HCO3- production in the duodenal mucosa, thereby strengthening its function as a protective barrier (14). However, the precise mechanisms underpinning this effect remain unclear. Authors of some studies have suggested that endogenous estrogen can increase the functional activity of cystic fibrosis transmembrane conductance regulator (CFTR) and SLC26A6 within the duodenal tissue (15). In other studies, it was indicated that, at the physiological level, estrogen can augment HCO3- secretion in this tissue through the activation of ERα- and PI3K-dependent pathways in response to PGE2 stimulation (16). Therefore, for primary physicians should be especially vigilant of DUs in male children with HP infection. In our study, the percentage of patients in the HP + DU group who had melena was 21.5%, which was markedly greater than that in the HP group (5%). In the HP group, the occurrence of melena may be related to diet, iron supplementation, or bismuth-containing medications use. Besides this, melena is a common symptom of upper gastrointestinal bleeding (UGIB) (17). Moreover, the characteristics of melena can indicate both the rate and severity of UGIB (18). In cases of UGIB, red blood cells within the gastrointestinal tract undergo degradation, and hydrogen sulfide reacts with Hb released from damaged cells to produce ferrous sulfide. Following an interaction with the intestinal microbiota and digestive enzymes, this substance ultimately transforms into a black, viscous tar-like material that results in the appearance of melena in affected children (19). DUs are the most prevalent cause of UGIB in pediatric patients. A previous multicenter study revealed that, among the 194 patients with UGIB included in the analysis, 75 (39%) exhibited melena. This rate was second only to that of children who presented with hematemesis. This multicenter study identified ulcers as the most common etiology for UGIB, accounting for 30% of all cases subjected to endoscopic evaluation (20). Another study investigating the etiology of melena in pediatric patients indicated that DUs were the predominant etiology in pediatric populations (21). Each of these prior studies revealed that DUs were highly correlated with melena, which aligns with the conclusions drawn in the current study. Therefore, when children with HP infection present with melena, DUs should be considered, and endoscopy should be performed promptly after ruling out dietary and medication factors. We found that Hb levels were lower in the HP + DU group than in the other groups, indicating that Hb levels are a significant predictor of DU. A prior study on anemia secondary to UGIB showed that up to 89.4% of patients with UGIB had peptic ulcers (22), indicating that anemia is significantly associated with peptic ulcers (23,24). In addition, as we mentioned above when discussing the relationship between melena and DUs, children with DUs are prone to UGIB, which can eventually lead to a decrease in hemoglobin. Moreover, 60% of patients might contract iron deficiency anemia subsequent to acute nonvariceal UGIB in cases in which portal hypertension was absent (22). Furthermore, Children with DUs often have upper abdominal pain and nausea, leading to reduced appetite and food intake, which can result in anemia (11). Chronic bleeding, impaired iron absorption, and ongoing inflammation caused by DUs also contribute to anemia (25). For children with HP infection who have decreased Hb levels, primary pediatricians should make prompt referrals to institutions equipped with pediatric endoscopy capability for further diagnosis and treatment. UA is the ultimate product of purine metabolism, with approximately one-third being excreted via the intestinal tract. Among all intestinal juices, duodenal juice has the highest UA concentration, which is positively correlated with serum UA levels (26). In an animal trial, Wang reported that, when the intestine was exposed to oxidative stress, the body activated the Nrf2 pathway to increase UA synthesis and secretion to counteract oxidative stress and protect intestinal epithelial cells. Therefore, elevated serum UA levels may serve as a self-protective mechanism for intestinal epithelial cells during the occurrence of DUs. However, a previous study revealed that high concentrations of UA could also exacerbate oxidative damage (27). One of the proposed mechanisms underlying injury associated with UA involves the fact that UA can activate an inflammatory response mediated by NALP (also referred to as NLRP). In research conducted by John K Crane and colleagues, it was indicated that UA crystals in the gastrointestinal tract exhibited chemotactic and secretory effects on immune cells, and it potentially occurs through the activation of the NALP3 inflammasome and protein kinase C. Furthermore, UA crystals are characterized by a similar adjuvant effect to alum, suggesting that this proinflammatory UA action within the gastrointestinal system may contribute to the development of DUs (28). Presently, there is a lack of consensus regarding the mechanisms underlying UA-induced tissue damage and its protective effects, necessitating further investigation. Previous studies have shown that UA is related to ulcerative diseases. In Chinese female patients with type 2 diabetes, increased serum UA levels are strongly correlated with an increased risk of diabetic foot ulcers. Elevated UA levels may suppress nitric oxide synthase, activate the RAS, and induce endothelial dysfunction. These effects can eventually lead to ulceration (29). In our study, we identified for the first time that elevated UA levels serves as an independent predictor of DU in children with HP infection, highlighting its significance in clinical practice. However, there are several limitations in this study, such as the use of a retrospective analysis, which might lead to certain degree of bias. Second, due to the stringent exclusion criteria, a considerable number of cases were eliminated. This resulted in a relatively small sample size compared with other studies. However, this small sample size increased the precision of the research. In future studies, we plan to collect more comprehensive data on ulcer types and stages to further assess the applicability and generalizability of the model in different populations. Third, the proportion of male patients was higher in both the training and validation groups, which may be one of the reasons for the increased incidence of DU. However, our model identified male sex as a predictive factor for DU, which is consistent with previous studies, indicating that the higher proportion of males did not lead to biased results. Finally, there were baseline differences between the two hospitals from which data were collected in this study, which was due to individual variations among the patients who were admitted to different hospitals. Nevertheless, the AUC of the validation group remained high, which further underscores the general applicability of this model. Conclusions In this study, a model for predicting A-stage or H-stage DUs in children with HP infection was successfully established. Therefore, when male children have melena, decreased Hb levels, and increased UA levels, primary pediatricians should be highly vigilant for DU and promptly refer the children to institutions with pediatric endoscopic capability for further diagnosis and treatment. Supplementary The article’s supplementary files as Acknowledgments The authors express their sincere gratitude to the departmental colleagues whose steadfast support and collaborative spirit were indispensable throughout the course of this research and manuscript preparation. Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was approved by Ethics Committee of Fujian Children’s Hospital (approval No. 2022ETKLRD10033). The other institution was informed and agreed with the study. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Informed consent was waived by the ethics committee. Footnotes Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-557/rc Funding: This work was supported by the Joint Funds for the innovation of science and Technology, Fujian province (grant No. 2023Y9386); Fujian Medical University Student Innovation and Entrepreneurship Training Project (grant No. S202410392016); and Fujian Provincial Natural Science Foundation of China (grant No. 2023J011314). Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-557/coif). All authors report that this work was supported by the Joint Funds for the innovation of science and Technology, Fujian province (grant No. 2023Y9386); Fujian Medical University Student Innovation and Entrepreneurship Training Project (grant No. S202410392016); and Fujian Provincial Natural Science Foundation of China (grant No. 2023J011314). The authors have no other conflicts of interest to declare. Data Sharing Statement Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-557/dss References Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-557/prf Supplementary Materials The article’s supplementary files as Data Availability Statement Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-557/dss