Henoch–Schönlein purpura (HSP), also known as immunoglobulin A (IgA) vasculitis, is the most popular leukocytoclastic vasculitis, affecting the small vessels of the skin, joints, gastrointestinal (GI) tract, and kidneys in children. The main clinical features are non-thrombocytopenic purpura, GI manifestations, arthritis, and renal involvement. Most HSP cases have a sound outcome, with renal involvement being the most key prognostic factor influencing morbidity and mortality (1). This is especially true for severe kidney disease (SKD), as severe renal involvement is related to chronic renal impairment, nephritic or nephrotic syndrome, renal insufficiency, and renal failure. Therefore, the predictors for severe kidney disease have been a focus, and the risk factors at the time of diagnosis may aid in disease management. However, a standard and effective method to identify the risk factors for renal involvement is still lacking. Many studies focus on the prognostic factors of developing renal involvement in HSP, without consistent results (2–7). Some studies are not cohort studies, and some studies have a limited sample size. In addition, some studies targeting the Chinese population were restricted by a small sample size, the neglect of socioeconomic factors, and other initial changes of this disease, like HSP triggers and central nervous system (CNS) involvement (8). Considering these limitations of previous studies, the children with HSP were prospectively assessed to better identify the demographic findings and risk factors for SKD, thereby facilitating the better management of the disease. This prospective study aimed to explore risk factors of SKD within 3, 6, 12, and 24 months after the HSP diagnosis by analyzing the follow-up medical records within the same aforementioned timeframe.

The nomogram shows an easy and practical graphical tool to predict the risk assessments of a disease or death of patients and provide references for treatment, follow-up, and prognosis. At present, the nomogram has not been applied to the prediction of SKD.

Therefore, our goal was to build a prognostic model of 6-, 12-, and 24-month SKD outcomes in children with HSP, combining socioeconomic factors, clinical features, and treatments of this disease based on a large dataset gathered from high-volume institutions. So we analyzed two large prospective cohorts (development and validation cohorts) of patients with HSP to assess the performance of a nomogram to predict SKD status. The predictive value was determined according to discrimination, calibration, and clinical utility in separate internal and external validation cohorts.

This research was approved by the Medical Ethical Committee of the West China Second University Hospital of Sichuan University for Children and followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) statement on reporting predictive models. As a result, we show an easy and practical graphic tool that maybe expand the clinical use of the nomogram in HSP.

We developed and validated, for the first time, clinical feature-based nomograms for the individualized prediction of SKD in patients with HSP. The nomogram incorporates four items of clinical features: serious skin rash or digestive tract purpura, severe GI manifestations, recurrent symptoms, and renal involvement. Model performance, as reflected by the C-statistic, was 0.751 (95% CI, 0.734–0.769) in the training dataset and 0.714 (95% CI, 0.678–0.750) in the validation dataset. It demonstrates that the relatively simple nomogram provides a good prediction of outcomes in prospective study-based data.

Our cohort is representative of pediatric HSP patients. The SKD prevalence of 14.98% in the present cohort of HSP patients in children is in line with previously reported rates of 15% in studies with at least 2 years of follow-up (15, 16). In this study, 85.19% of SKD occurred during the 3 months of follow-up, whereas 97.22% occurred during the 6 months of follow-up, which is in line with previously published data. A previous study also conducted 6 to 12 months’ follow-ups (17). KDIGO (Kidney Disease: Improving Global Outcomes) guideline suggests that urinary monitoring is necessary for 6 months (18). Our follow-up data are supportive of the suggestion that a 6-month follow-up is a more prudent method for disease evaluation and amelioration (19).

We found that severe GI manifestations are a significant predictor for HSP nephritis in our Cox regression analysis, regardless of renal involvement symptom (test for protein and/or blood in urine) for patients at hospital admission. It implied that patients with severe GI manifestations have a higher risk of developing SKD, even though they may have normal urine test results. A recent study in China found that newly enrolled patients with GI were more likely to develop renal involvement during follow-up (20). Another study using machine learning prediction found similar results (21). Thus, patients with normal urine rest require 6 months of follow-up (22). Pediatric patients with severe GI symptoms, such as abdominal pain with GI bleeding and colic, require special attention in monitoring nephritis-related symptoms (2, 6, 23).

We found that the severe skin rash/digestive tract purpura and recurrent symptoms were strongly predictive of SKD. The risk factors for developing nephropathy in the course of HSP, according to a meta-analysis carried out by Chan et al., included persistent purpura and the occurrence of relapses (23). It was consistent with another study that reported that persistent skin rash for more than 1 month was a crucial risk factor for HSP patients to develop nephropathy and recurrence (24). Another group argued that patients presenting with a recurrent episode are supposed to receive renal screening again regardless of previous normal findings and followed up to ensure complete remission (25). Jiang et al. and Cao el al. found that relapse was an independent risk factor of HSP nephritis (HSPN) or renal damage (20, 21). Thus, early-onset HSP patients with severe skin symptoms and recurrence should receive special care.

We found that 34.66% of patients with renal involvement developed SKD in a 2-year follow-up. In Cox regression analysis, renal involvement at hospital admission is a risk factor for SKD. Other studies found that the severity of renal involvement at HSP onset determines the severity of nephritis and long-term HSP prognosis (15, 26). Thus, patients with renal involvement, especially those with protein in the urine, require close monitoring of urine since disease onset (3).

In our study, 97.11% of hospitalized cases underwent hormone treatment. However, we did not find a significant correlation between hormone treatment with disease prognosis in our Cox regression analysis. Obviously, clinical symptoms have more predictive power in prognosis. Other studies and reviews also mentioned that there is no evidence supporting a preventive role of hormone treatment during HSP onset in nephritis progression (4, 17). A systemic evaluation implicated that steroids could not provide protection against complications and should not be used as a preventive measure (19).

Here, age at onset was not a predictor of poor prognosis in Cox multivariate analysis, even if our previous and others’ studies showed that older children (≥6 years old) are more likely to have the more severe disease at presentation (5–7, 20, 27–30).

However, the present study has several limitations. First, the accuracy of our nomogram was determined through both internal and external validation, but the nomogram was developed and externally validated based on two prospective cohorts from one medical institution; therefore, potential bias was inevitable. A multicenter trial would allow further evaluation of the accuracy and applicability of this novel nomogram in a wider population of patients with HSP. Second, there was attrition in follow-up between 1 and 24 months, which might due to the difficulty of extensive follow-up for patients with overall good outcomes to be involved in extensive follow-up; thus, this might cause a bias toward good prognosis in the patients lost to follow-up. However, our dropout rate was similar to that of other studies. Thus, we established a reliable predictive model, providing vital results to interpret the long-term outcome in this patient population.

In summary, our nomogram study reveals four clinical features as risk factors for SKD: serious skin rash or digestive tract purpura, severe GI manifestations, recurrent symptoms, and renal involvement. Specific clinical interventions to these clinical features may provide a positive influence in preventing SKD in HSP patients (23). However, due to the limitations in this study described above, caution should be taken in the interpretation of our results.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by West China Second University Hospital. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

KW and L-QD contributed to the study design. S-LJ, LL, YC, and XP supervised the data collection. Data analysis was performed by KW. KW and X-MS wrote the main manuscript text. L-QD and L-NQ reviewed and edited the manuscript. All authors contributed to the article and approved the submitted version.

This research was supported by funding from The Central Government Funds of Guiding Local Scientific and Technological Development for Sichuan Province of China (No. 2021ZYD0105), the Sichuan Province Key Research and Development Program (No. 22ZDZX0014) and Fundamental Research Funds for the Central Universities (SCU2021E4251).

We would like to thank Wen Deng for proofreading our manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

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.