https://orcid.org/0000-0001-7080-2293
TO THE EDITOR:
Immune thrombocytopenia (ITP) is a rare autoimmune bleeding disorder in children with an incidence of 5 per 100 000 children per year.1 Children typically present with thrombocytopenia and variable bleeding symptoms, rarely including severe bleeding, such as intracranial hemorrhage.2,3 Chronic ITP (cITP), with persistence >12 months, occurs in 20% to 30% of patients and is associated with decreased quality of life and an increased need for second-line treatments.1,4-6 Identifying patients at risk for cITP can help providers recognize those who require early investigation for secondary causes of ITP and those in whom earlier interventions may help alter the disease course and overall outcomes. Older age is a well-documented risk factor for cITP. The lack of an identifiable trigger or mucosal bleeding, higher platelet count, and positive direct antiglobulin test at diagnosis have also been associated with the development of cITP.7-9 However, the definitive predictors of chronicity at the time of ITP diagnosis remain largely unknown.
Prior analysis of baseline immunoglobulin levels in a smaller institutional cohort of children with ITP indicated their potential for predicting chronicity but without definitive clinically applicable conclusions. Thus, we analyzed retrospective data obtained from children with ITP treated at 2 tertiary care referral centers to investigate the utility of immunoglobulin levels obtained at the time of diagnosis in predicting the development of cITP.
A retrospective review of pediatric patients (aged 6 months to 21 years) diagnosed with ITP by a hematologist at 2 tertiary care children’s hospitals in the United States between 2015 and 2022 was conducted under an institutional review board–approved protocol. Included patients had immunoglobulin levels obtained within 3 months of ITP diagnosis and before any treatment and those lost to follow-up or without at least 1 year of follow-up were excluded. Of the total patients in this selected cohort (N = 624), 45 were excluded from the final regression analysis because of missing values for at least 1 type of immunoglobulin (n = 579). Patients were determined to have spontaneously resolved ITP if they achieved a platelet count >150 × 109/L on 2 consecutive occasions, or 1 occasion in the presence of normalized mean platelet volume, absent of therapy, before 12 months from ITP diagnosis. Patients were considered to have no therapy before achieving spontaneous resolution based on treatment-specific definitions of minimum time since receiving therapy, as follows: 6 weeks since IV immunoglobulin or Rh immune globulin, 3 weeks since corticosteroids, 1 week since immunomodulators, 2 weeks since thrombopoietin receptor agonists, or achieving B-cell repopulation after rituximab therapy. Immunoglobulins were recorded as continuous variables and classified for the purposes of description by age-related reference values.
The 2 groups (spontaneously resolved and cITP) were compared using Pearson χ2, Fisher exact, and Wilcoxon rank sum test as appropriate. A multivariable logistic regression model was built to determine the independent associations between immunoglobulin levels and the development of cITP after adjusting for age and sex. R (version 4.2.1) was used for analysis. A P value <.05 was considered to be statistically significant.
Among the 624 pediatric patients with ITP who met the inclusion criteria for the study, 415 patients (66.5%) had spontaneously resolved ITP before 12 months, whereas 209 patients (33.5%) developed cITP. Male sex was more frequent in the spontaneously resolved vs cITP groups (54% vs 44%; P = .013). The median age and platelet count at ITP diagnosis were lower in the spontaneously resolved group than in the chronic group (4.0 vs 9.5 years [P < .001]; 6 × 109/L vs 12 × 109/L [P < .001]). Both low and high immunoglobulin G (IgG) levels for age were more prevalent in cITP, with a higher median IgG level in the cITP group. Low IgA levels were more prevalent in cITP although the median IgA level was higher in cITP, which may be attributable to the older age of the cITP group. Both low and high IgM levels were more prevalent in cITP, with the median IgM levels being similar in both groups (Figure 1).
Age and immunoglobulin levels for individual subjects in spontaneously resolving and chronic ITP cohorts. Age (A), IgG (B), IgA (C), and IgM (D) levels at the time of ITP diagnosis in spontaneously resolving vs chronic ITP.
Age and immunoglobulin levels for individual subjects in spontaneously resolving and chronic ITP cohorts. Age (A), IgG (B), IgA (C), and IgM (D) levels at the time of ITP diagnosis in spontaneously resolving vs chronic ITP.
Age-adjusted multivariable logistic regression analysis demonstrated the independent predictive value of immunoglobulins for the development of cITP (Table 1). Each 100 mg/dL increase in IgG level was associated with a 14% increase in the odds of developing cITP (P = .002); whereas each 100 mg/dL increase in IgA or IgM levels was associated with decreased odds of developing cITP (33% [P = .038] and 58% [P < .001] decrease, respectively). The multivariable model demonstrated a C-index of 0.74 and Nagelkerke R2 of 0.22, indicating that immunoglobulin values, age, and sex can help identify patients at diagnosis who are at higher risk for developing cITP. We also found on univariable analysis that patients with a known secondary cause of ITP at the time of diagnosis had an increased likelihood of developing cITP (P < .001).
Outcomes of a multivariable linear regression model to predict development of cITP
| Odds ratio (95% confidence interval) | P value | |
|---|---|---|
| Age at ITP diagnosis | 1.18 (1.12-1.23) | <.001 |
| IgG value (mg/dL) at ITP diagnosis | 1.14 (1.05-1.24) | .002 |
| IgA value (mg/dL) at ITP diagnosis | 0.67 (0.46-0.98) | .038 |
| IgM value (mg/dL) at ITP diagnosis | 0.42 (0.27-0.65) | <.001 |
| Male sex | 0.70 (0.47-1.04) | .073 |
| Odds ratio (95% confidence interval) | P value | |
|---|---|---|
| Age at ITP diagnosis | 1.18 (1.12-1.23) | <.001 |
| IgG value (mg/dL) at ITP diagnosis | 1.14 (1.05-1.24) | .002 |
| IgA value (mg/dL) at ITP diagnosis | 0.67 (0.46-0.98) | .038 |
| IgM value (mg/dL) at ITP diagnosis | 0.42 (0.27-0.65) | <.001 |
| Male sex | 0.70 (0.47-1.04) | .073 |
Given that the course of ITP is unpredictable, with many children having spontaneously resolving disease within months but others going on to have a chronic course requiring second-line agents and a greater impact on health-related quality of life, finding predictors of disease chronicity at the time of diagnosis is key. We investigated the predictive power of immunoglobulin levels obtained at diagnosis and found that immunoglobulins add predictive power independent of and when controlling for age and sex. In this way, immunoglobulins could be part of a risk prediction model to determine the future risk of cITP early in a patient’s disease course.
Hypergammaglobulinemia has been associated with inflammatory and autoimmune diseases.10 Systemic lupus erythematosus can be associated with hypergammaglobulinemia, and in a study of 500 pediatric patients, noniatrogenic IgG levels >2000 mg/dL had predictive value for autoimmune/autoinflammatory disease, with systemic lupus erythematosus being the most common diagnosis.11,12 Viral infections may also be a cause.13 In clinical practice, immunoglobulins are often obtained later in the disease course if patients do not respond to treatments as expected or after developing cITP, to identify potential hypogammaglobulinemia and immunodeficiency. However, our study focused on immunoglobulins at the time of diagnosis, which could lead to earlier identification of patients but also provide data beyond screening for immunodeficiency.
The limitations of this study include those inherent to the retrospective approach, such as missing information and lack of adequate follow-up. Additional immunologic studies (ie, vaccine titers and lymphocyte subsets) were not available for most patients and could add to the data interpretation. Data were obtained from 2 tertiary referral centers, which may bias the trends toward higher rates of both chronic and secondary ITP. Future directions include prospective and longitudinal analyses of immunoglobulin levels in children with ITP, along with supplementary immunologic markers. A risk prediction model incorporating additional factors at the time of ITP diagnosis could strengthen these findings and validate them in an external cohort.
In conclusion, pretreatment IgG, IgA, and IgM levels obtained in patients with newly diagnosed ITP are predictive of cITP development, even after adjusting for age. Aligned with prior studies, age at diagnosis is also predictive of the development of cITP. Better defining the predictive value of clinical data available at the time of ITP diagnosis will allow hematologists to provide improved prognostication and individualized management much earlier in a patient’s disease course, ultimately improving the quality of care for children with ITP.
Acknowledgment: K.H. acknowledges the support of the Sala Elbaum Pediatric Research Scholars Program.
Contribution: K.H., D.M., J.C., S.E.K., C.O., T.O.K., and A.B.G. were involved in data collection; K.H., M.Z., and A.B.G. were involved in data analysis and figure creation; K.H. wrote the manuscript with input from all authors; and A.B.G. was involved in the project concept, design, and oversight.
Conflict-of-interest disclosure: D.M. is supported in part by the Division of Clinical Research at National Institute of Allergy and Infectious Diseases/National Institutes of Health. A.B.G. has research funding from Novartis. The remaining authors declare no competing financial interests.
Correspondence: Amanda B. Grimes, Department of Pediatrics, Baylor College of Medicine, 6701 Fannin St, Houston, TX 77030; email: abgrimes@bcm.edu.
