Introduction
Pediatric autoimmune hemolytic anemia (AIHA) is a rare disease caused by complex immune dysregulation leading to autoantibody-mediated erythrocyte destruction, which is associated with significant morbidity.1 AIHA is heterogeneous, classified based on the thermal reactivity of the autoantibody, including warm AIHA, cold agglutinin syndrome, and paroxysmal cold hemoglobinuria, and can be further subdivided into primary (idiopathic) or secondary to an underlying disease.2, 3, 4 Presenting symptoms range from mild signs of hemolytic anemia including fatigue and pallor to life-threatening hemolysis with consequent renal failure and cardiovascular collapse.4,5 Evidence guiding the diagnostic evaluation and management of pediatric AIHA is largely based on small observational studies or extrapolated from adult data.2,6
Prior observational studies have demonstrated a high treatment burden in children with warm AIHA, with reports of 25% to 35% of patients requiring multiple therapies and remaining on long-term immunosuppression.5,7 The mortality in children with isolated AIHA reported in the literature ranges from 3% to 4%.5,7 For children with Evans syndrome (ES), with ≥2 concomitant or sequential immune-mediated cytopenias, reported mortality is higher at 10% to 30%.2,5, 6, 7, 8, 9, 10, 11 Pediatric ES is often associated with inborn errors of immunity (IEI).8, 9, 10, 11, 12 For those with isolated AIHA, the rate of underlying immune disorders is not known. A high proportion of cases may be secondary to autoimmune or immunologic disease, but diagnostic evaluation is inconsistent.5,8 Immunophenotyping studies, including quantitative immunoglobulin levels and T-cell phenotyping, may be helpful indicators of IEI in children with immune cytopenias, but little is known about clinical factors that predict the likelihood of an underlying immunologic diagnosis among children with AIHA.12,13
First-line treatment of warm AIHA includes corticosteroids, although reported pediatric remission rates with steroid treatment are lower than for adults.2,5 Rituximab, other immunosuppressants, and splenectomy are used as second- and third-line therapies.6,8,14, 15, 16 Steroid-sparing agents, such as sirolimus and mycophenolate mofetil, have been used successfully in patients with AIHA after solid organ transplantation and in patients with refractory cytopenias due to ES,17, 18, 19, 20 and cyclosporine has also been reported to successfully treat refractory immune cytopenias in patients with underlying immunodeficiencies.21 There are limited data to compare the efficacy of these agents or predict clinical response.
Although previous studies have informed our understanding of AIHA, little is known about factors that influence the risk of recurrence, secondary diagnoses, efficacy of steroid-sparing agents, and mortality in pediatric AIHA. To address these questions, the ITP Consortium of North America (ICON) conducted an observational cohort study of pediatric AIHA with the aim of providing a detailed report of the current diagnostic evaluation and findings, as well as the morbidity, mortality, and outcomes, to inform management and treatment options.
Methods
Study design
A multicenter observational cohort study of pediatric patients with AIHA was conducted at ICON sites in the United States and Canada after local institutional review board/research ethics board approval. A waiver of consent was received from each center.
Participants
Eligibility included patients aged 3 months to 22 years with a diagnosis of AIHA who received a hematology consultation between January 2011 and December 2020. Patients with all subtypes of AIHA were included: warm AIHA (presence of a warm-binding antibody), cold agglutinin syndrome (presence of cold agglutinin), paroxysmal cold hemoglobinuria (presence of Donath-Landsteiner antibody), mixed AIHA (presence of both warm- and cold-binding antibodies), and direct antiglobulin test–negative AIHA if diagnosed as AIHA by the hematologist. Patients with ES who had multilineage immune cytopenias including AIHA (AIHA and immune thrombocytopenia [ITP] and/or autoimmune neutropenia) were also included.
Patients who had additional diagnoses such as autoimmunity of another organ system or known IEI were included. Those with AIHA after solid organ transplant were included. Those with a history of hematopoietic stem cell transplant before onset of AIHA were excluded, whereas those who underwent hematopoietic stem cell transplant after AIHA onset were included.
Data collection
Deidentified demographic, laboratory, medical history, and treatment data were collected and entered in a centralized, secure, web-based electronic database (REDCap) hosted at Boston Children’s Hospital. Each participating center had access only to the records of their respective patients, whereas the central site had access to data from all sites. Definitions of disease, treatment response, AIHA course, and remission are included in supplemental Tables 1-3.
Steroid-sparing therapy was defined as any immune-modulating therapy that could decrease the need for corticosteroid treatment.
Statistical analysis
Demographic features, clinical features, diagnosis characteristics, and laboratory data for the first AIHA episode, treatment of the first or subsequent AIHA episodes, and outcomes were summarized using frequencies and percentages for binary factors and medians and ranges for continuous factors with the total number of patients with data available for the characteristic. Univariate logistic regression models identified factors associated with the number of episodes of AIHA (chronic/multiple AIHA episodes vs single episode) and those associated with an IEI or autoimmune disorder. Odds ratios (ORs) were reported with a 95% confidence interval.
Kaplan-Meier curves of overall survival were generated. A log rank test was used to test for differences in overall survival between specific subgroups. Continuous variables were summarized as mean (standard deviation) or median (range) and compared with a 2-sided t test.
Categorical variables were summarized as counts or percentages and compared with a 2-sided χ2 test for all expected cells counts ≥5, and the Fisher exact test was used when the expected cell count was <5. P values <.05 were considered statistically significant.
Results
Clinical characteristics
This study included 399 patients with AIHA seen for a hematology consultation between January 2011 and December 2020 at 15 ICON sites (Table 1). The number of patients included from each institution ranged from 2 to 65, with a median of 20 patients included per site. The median age at diagnosis was 7 years (range, 0-22; interquartile range [IQR], 3-13; n = 396), with a median follow-up of 4.9 years (range, 0-19.4; IQR, 1.6-8.1; n = 398).
Approximately half were female (180/397 [45%]). The median age at AIHA diagnosis was significantly younger in males than females (5 vs 9 years; P < .001; supplemental Figure 1).Table 1.Clinical characteristics of overall cohort with AIHA (N = 399)n∗Median (range; IQR)/n (%)Age at diagnosis, y3967.0 (0-22.0; 3-13)Length of follow-up, y3984.9 (0-19.4; 1.6-8.1)Female sex397180 (45)Race223 American Indian/Alaskan3 (1) Asian29 (9) Black/African American42 (13) White184 (59) Hawaiian/Pacific Islander1 (0.3) Multiple Race7 (2) Other46 (15)Ethnicity, Hispanic32879 (24)First- or second-degree relative with a diagnosis of immune cytopenia, immunodeficiency, or autoimmunity29678 (26) Immune cytopenia22 (28) Immunodeficiency10 (13) Autoimmunity60 (77)ES385142 (37)Solid organ transplant†39026 (7)Type of AIHA399 Warm AIHA257 (64) Paroxysmal cold hemoglobinuria24 (6) Cold agglutinin syndrome47 (12) Mixed/other AIHA33 (8) Unspecified AIHA38 (10)Type of AIHA course373 Single acute episode that remitted217 (58) Single chronic episode that remitted48 (13) ≥2 episodes of AIHA that remitted55 (15) Chronic and ongoing AIHA53 (14)∗Number with available data.†Types of solid organ transplants included cardiac (n = 7), liver (n = 12), kidney (n = 5), multivisceral (n = 3), and others (n = 3).
Most patients had warm AIHA (257/399 [64%]). Of 399 patients, cold agglutinin syndrome was present in 47 (12%), paroxysmal cold hemoglobinuria in 24 (6%), mixed AIHA in 33 (8%), and unspecified type of AIHA in 38 (10%). There was a family history of immune cytopenia, immunodeficiency, or autoimmunity in a first or second degree relative in 78 of 296 patients (26%; Table 1).
Clinical presentation and laboratory findings
Symptoms, physical examination findings, and laboratory findings at initial presentation are in supplemental Table 4. The median nadir hemoglobin was 5.3 g/dL (range, 1.7-14.5; IQR, 4.0-7.0). Nadir hemoglobin and highest reticulocyte count did not vary by type of AIHA; highest lactate dehydrogenase and indirect bilirubin levels were higher in those with paroxysmal cold hemoglobinuria and cold agglutinin than those with warm AIHA (supplemental Table 4).
A peripheral blood film was examined in 241 of 294 patients (82%), with 49% (119/241) reporting spherocytes, seen across all types of AIHA. Of 397 patients, a direct antiglobulin test was positive in 353 (88.9%), negative in 22 (6%), and unknown or not done in 22 (6%). In those with a positive direct antiglobulin test, 264 (75%) were immunoglobulin G (IgG) positive, 193 (56%) were C3 positive, and 131 (39%) were both IgG and C3 positive.
An enhanced direct antiglobulin test was performed in 24 of 338 patients (7%), 10 of whom were negative or had unknown results for both IgG and C3 on initial direct antiglobulin testing. Enhanced direct antiglobulin testing was positive in 16 of 24 patients (66.7%) tested. Seventy-six patients (22%) had a bone marrow evaluation.
Functional immune assays were performed in 305 of 399 (76%), with abnormal test results in 249 of 305 (82%; Table 2; supplemental Tables 5 and 6). Genetic testing to evaluate for an IEI was performed in 109 of 348 (31%), including a gene panel (69/109 [63%]) or exome sequencing (27/109 [25%]). Pathogenic findings consistent with an IEI were identified in 34 of 109 (31%) tested (Table 3).Table 2.Functional immune and antibody testing in patients with autoimmune hemolytic anemia (N = 399)Frequency tested, n (%)Frequency of abnormal results, n/N (%)Any immune assay305 (76)249/305 (82)Immunoglobulin levels∗245 (61)150/245 (61)Lymphocyte subsets177 (44)144/177 (81)ALPS flow cytometry testing97 (24)53/97 (55)Vaccine serology†88 (22)23/88 (26)Anti-nuclear antibody test183 (46)63/183 (34)Other rheumatologic antibody testing‡125 (31)30/125 (24)General antibody testing§202 (51)87/202 (43)Anti-phospholipid antibodies76 (19)28/76 (37)Complement testing||145 (36)85/145 (59)ALPS, autoimmune lymphoproliferative syndrome.∗Immunoglobulins included IgG, IgA, IgE, IgM, and IgG subclasses.†Vaccine serology included tetanus and pneumococcal titers.‡Anti-ds DNA, anti-Smith, Anti-RNP, Anti-SSA or SSB, Anti-Scl 70, and Anti-Jo 1.§ANA, anti-ds DNA, anti-Smith, Anti-RNP, Anti-SSA or SSB, Anti-Scl 70, Anti-Jo 1, lupus anticoagulant, and antiphospholipid antibodies.||Complement testing included C3, C4, and CH50.Table 3.Genetic testing in patients with autoimmune hemolytic anemia (N = 399)n/No. with known status (%)Genetic testing109/348 (31) Gene panel69/109 (63) WES27/109 (25) Other31/109 (28)Genetic test results Pathogenic34/109 (31) VUS41/109 (38) Negative32/109 (29)VUS, variant of uncertain significance; WES, whole exome sequencing.
Patients with IEI or other autoimmunity had a higher proportion of abnormal immunoglobulin levels, general antibody testing, antinuclear antibody testing, other rheumatologic antibody testing, antiphospholipid antibodies, and complement testing than those without (supplemental Table 7). Vaccine titers and lymphocyte subset results were not different in patients with an IEI or other autoimmunity compared to those without. An immunologist was consulted in 136 of 342 patients (40%).
Infectious evaluations including viral polymerase chain reaction, serologies, and/or blood cultures were performed in 261 of 367 patients (71%). Testing was consistent with an acute infection at the time of AIHA diagnosis in 122 of 261 (47%; supplemental Table 8), including most commonly Epstein Barr Virus (34%), enterovirus (33%), and mycoplasma pneumoniae (32%). Of those with positive infectious testing, 37 of 118 (31%) were found to have an IEI or other autoimmunity.
Other diagnoses
IEI
Of the 379 patients with known data, 68 (18%) were found to have an IEI, with the most frequent diagnoses being common variable immunodeficiency (29/68 [43%]), autoimmune lymphoproliferative syndrome (18/68 [26%]), and 22q11.2 deletion syndrome (DiGeorge syndrome; 9/68 [13%]; Table 4; Figure 1). Patients with a diagnosed IEI were more likely to be older at the first AIHA episode (median, 8 years [range, 1-21; IQR, 5-13] vs 6 years [range, 0-22; IQR, 2-13], respectively; P = .021) and have warm AIHA (P = .001; supplemental Table 9).Table 4.Comorbid diagnoses in a cohort with AIHA (N = 399)Warm AIHA (n = 257), n/N (%)Cold agglutinin (n = 47), n/N (%)PCH (n = 24), n/N (%)Mixed/other AIHA (n = 33), n/N (%)Unspecified AIHA (n = 38), n/N (%)Overall (N = 399), n/N (%)IEI55/243 (23)0/45 (0)0/24 (0)6/32 (19)7/35 (20)68/379 (18) Common variable immunodeficiency26/55 (47)001/6 (17)2/7 (28)29/68 (43) ALPS11/55 (20)004/6 (67)3/7 (43)18/68 (26) 22q11 deletion syndrome7/55 (13)001/6 (17)1/7 (14)9/68 (13) CTLA4 haploinsufficiency6/55 (11)00006/68 (9) STAT3 gain of function disease5/55 (9)00005/68 (7) LRBA deficiency1/55 (2)00001/68 (1) Other∗10/55 (18)0001/7 (14)11/68 (16)Autoimmunity of other organ systems68/253 (27)3/46 (7)0/23 (0)10/33 (30)5/37 (14)86/392 (22) Systemic lupus erythematosus23/68 (34)005/10 (50)2/5 (40)30/86 (35) Neurologic disease17/68 (25)002/10 (20)1/5 (20)21/86 (24) Antiphospholipid antibodies7/68 (10)1/3 (33)02/10 (20)1/5 (20)10/86 (12) Hashimoto thyroiditis6/68 (9)1/3 (33)0007/86 (8) Inflammatory bowel disease9/68 (13)00009/86 (10) Sjogren syndrome1/68 (15)1/3 (33)0002/86 (2) Mixed connective tissue disorder1/68 (15)001/10 (10)02/86 (2) Other†21/68 (31)1/3 (33)02/10 (20)2/5 (40)26/86 (20)Malignancy‡8/254 (31)2/47 (42)01/33 (30)011/396 (3)ALPS, autoimmune lymphoproliferative syndrome; PCH, paroxysmal cold hemoglobinuria.∗Other IEI included Kabuki syndrome (n = 1), severe combined immunodeficiency (n = 1), T-cell lymphopenia (n = 3), humoral immune deficiency (n = 1), RAG2 deficiency (n = 1), IgA deficiency (n = 2), bare lymphocyte syndrome (n = 1), and unspecified immunodeficiency with associated HIV (n = 1)†Other autoimmunity of other organ systems (some individual patients had >1 type of other autoimmunity) included granulomatous-lymphocyte interstitial lung disease (n = 3), immune hepatitis (n = 1), giant cell hepatitis (n = 3), autoimmune encephalitis (n = 1), multisystem inflammatory syndrome vs atypical Kawasaki disease (n = 1), poststreptococcal glomerulonephritis (n = 1), Addison disease (n = 1), immune complex–mediated glomerulonephritis (n = 1), type 1 diabetes mellitus (n = 1), primary sclerosing cholangitis vs autoimmune hepatitis (n = 1), and Kawasaki disease (n = 1).‡Types of malignancy included bilateral Wilms tumor (n = 1), Hodgkin lymphoma (n = 4), low-grade B-cell lymphoma (n = 1), posttransplant lymphoproliferative disorder (n = 3), rhabdomyosarcoma (n = 1), and seminoma (n = 1).Figure 1.Frequency of primary AIHA and AIHA associated with ES, IEI, and other autoimmunity (N = 361).
Autoimmunity of other organ systems
Autoimmunity of other organ systems was present in 86 of 392 patients (22%) in this cohort, including systemic lupus erythematosus (30/86 [35%]), immune neurologic disease (21/86 [24%]), and antiphospholipid antibodies (10/86 [12%]; Table 4; Figure 1). Females were more likely to have other autoimmune disorders than males (OR, 2.78; P < .001; Figure 2). For every 1-year increase in age, the risk of having an autoimmune disorder increased by 12% (P < .001).
Patients with other autoimmune disorders were more likely to have warm AIHA (OR, 2.47; P = .002), abnormal immune testing results (OR, 3.42; P = .007), and a family history of any immune cytopenia, immunodeficiency, or systemic autoimmunity (OR, 2.10; P = .015) than those without other autoimmune disorders (supplemental Table 10).Figure 2.Forest plot of univariate logistic regression model of factors associated with autoimmunity of other organ systems vs no other autoimmunity in pediatric patients with AIHA (N = 392).
ES
ES was common (142/385 [37%]) including ITP (131/142 [92%]) and/or immune neutropenia (56/132 [42%]; supplemental Table 11; Figure 1). Patients with ES (n = 142) were more likely than those without ES (n = 243) to be older during the first AIHA episode (median age, 9 years [range, 0-21; IQR, 4-13] vs 5 years [range, 0-22; IQR, [2-12.5]; P = .011), have warm AIHA (84% vs 53%; P < .001), receive steroid-sparing treatment in the first AIHA episode (36% vs 19%; P < .001), and have genetic testing for an IEI (55% vs 20%; P < .001; supplemental Table 11).
Solid organ transplantation
AIHA was associated with a history of solid organ transplant in 26 of 399 patients (7%). Types of transplants included liver (n = 12), cardiac (n = 7), kidney (n = 5), multivisceral (n = 3), and other (n = 3).
Malignancy
Malignancy was present in 11 of 396 patients (3%), including 8 patients with lymphoma or posttransplant lymphoproliferative disorder (Table 5). Malignancy was diagnosed before AIHA in 10 patients and after the diagnosis of AIHA in 1 patient. Of the 11 patients with a malignancy, 3 had an identified IEI.
Mortality in patients with malignancy was 36% (4/11).
Treatments
For the treatment of the first episode of AIHA, the median number of medications was 2 (n = 338; range, 0-17; IQR, 1-3; Table 4; supplemental Figure 2). Of those with warm AIHA, steroid-sparing therapy was prescribed to 79 patients (31%) for their first episode of AIHA, and 41 of 79 (52%) received these treatments within the first month of diagnosis (Table 5). The median number of steroid-sparing medications prescribed per patient for the first episode of warm AIHA was 0 (n = 257; range, 0-6; IQR, 0-1).Table 5.Treatment outcomes for the first episode of autoimmune hemolytic anemia overall and by type of autoimmune hemolytic anemia (N = 399)Warm AIHA (n = 257)Cold agglutinin syndrome (n = 47)Paroxysmal cold hemoglobinuria (n = 24)Mixed/other AIHA (n = 33)Unspecified AIHA (n = 38)Overall (N = 399)nmedian (range)nmedian (range)nmedian (range)nmedian (range)nmedian (range)nmedian (range)No. of treatments per patient2572 (0-17)470 (0-4)242 (0-3)333 (0-16)382 (0-8)3992 (0-17)No. of steroid-sparing treatments per patient2570 (0-6)470 (0-1)240 (0-0)330 (0-6)380 (0-4)3990 (0-6)n (%)n (%)n (%)n (%)n (%)n (%)Proportion of patients treated∗ Corticosteroids218 (85)18 (38)18 (75)31 (94)30 (79)315 (79) IVIG83 (32)1 (2)0 (0)10 (30)10 (26)104 (26) Rituximab60 (23)0 (0)0 (0)10 (30)2 (5)72 (18) Mycophenolate21 (8)0 (0)0 (0)5 (15)4 (11)30 (8) Sirolimus17 (7)1 (2)0 (0)3 (9)1 (3)22 (6) Observation6 (2)5 (11)3 (12)2 (6)2 (5)18 (4)Patients treated with steroid-sparing therapy79 (31)1 (2)0 (0)14 (42)7 (18)101 (25)n/No. with steroid-sparing therapy (%)n/No. with steroid-sparing therapy (%)n/No. with steroid-sparing therapy (%)n/No. with steroid-sparing therapy (%)n/No. with steroid-sparing therapy (%)n/No. with steroid-sparing therapy (%)Patients treated with steroid-sparing therapy within first month from diagnosis41/79 (52)1/1 (100)0 (0)6/14 (43)7/7 (100)55/100 (55)nmedian (range)nmedian (range)nmedian (range)nmedian (range)nmedian (range)nmedian (range)Median duration of AIHA at first dose, mo Corticosteroids2150.07 (−0.26-8.96)170.07 (0.03-5.12)180.05 (0.03-0.43)300.08 (0.03-3.64)270.07 (0.03-24.26)3070.07 (−0.26-24.26) IVIG780.13 (−0.26-60.40)10.07 (0.07-0.07)0NA100.12 (0.03-2.71)80.17 (−0.13-1.26)970.13 (−0.26-60.40) Rituximab590.93 (0.03-59.27)0NA0NA101.72 (0.13-6.64)20.30 (0.03-0.56)710.93 (0.03-59.27) Mycophenolate202.26 (−29.92-53.88)0NA0NA54.96 (0.03-19.70)40.12 (0.10-0.53)291.42 (−29.92-53.88) Sirolimus171.65 (0.03-69.65)10.60 (0.60-0.60)0NA30.26 (0.13-30.94)10.30 (0.30-0.30)220.89 (0.03-69.65)n (%)n (%)n (%)n (%)n (%)n (%)Of patients treated with corticosteroids, response to corticosteroids CR129 (59)13 (72)6 (33)17 (55)21 (70)186 (59) PR83 (38)5 (28)10 (56)11 (35)7 (23)116 (37) No response6 (3)0 (0)2 (11)3 (10)2 (7)13 (4) Unknown000000Of patients treated with IVIG, response to IVIG CR12 (14)0 (0)0 (0)2 (20)4 (40)18 (17) PR52 (63)1 (100)0 (0)4 (40)3 (30)60 (58) No response19 (23)0 (0)0 (0)4 (40)3 (30)26 (25) Unknown000000Of patients treated with rituximab, response to rituximab CR22 (37)0 (0)0 (0)5 (50)1 (50)28 (39) PR31 (52)0 (0)0 (0)3 (30)1 (50)35 (49) No response7 (12)0 (0)0 (0)2 (20)0 (0)9 (13) Unknown000000Of patients treated with mycophenolate, response to mycophenolate CR16 (80)0 (0)0 (0)4 (80)3 (75)23 (79) PR4 (20)0 (0)0 (0)0 (0)1 (25)5 (17) No response0 (0)0 (0)0 (0)1 (20)0 (0)1 (3) Unknown100001Of patients treated with sirolimus, response to sirolimus CR8 (47)1 (100)0 (0)3 (100)1 (100)13 (59) PR7 (41)0 (0)0 (0)0 (0)0 (0)7 (32) No response2 (12)0 (0)0 (0)0 (0)0 (0)2 (9) Unknown000000∗As a proportion of the total number of patients.
Other treatments included no medication (n = 55), 6-mercaptopurine (n = 11), azathioprine (n = 5), cyclosporine (n = 4), splenectomy (n = 4), erythropoietin (n = 3), plasmapheresis (n = 3), hematopoietic stem cell transplant (n = 2), abatacept (n = 2), bortezomib (n = 2), and hydroxychloroquine (n = 2).
Corticosteroids were the most common treatment prescribed for the first episode of warm AIHA (218/257), followed by IV immunoglobulin (n = 83), rituximab (n = 60), mycophenolate (n = 21), and sirolimus (n = 17). The median starting dose was 4 mg/kg per day for IV methylprednisolone, and 2 mg/kg per day for oral prednisone. Of 218 patients who received corticosteroids for the first episode of warm AIHA, 129 (59%) had a complete response, 83 (38%) had a partial response, and 6 (3%) had no response.
Complete response rates to steroid-sparing therapy used in the first episode of AIHA varied, including 80% with mycophenolate, 47% with sirolimus, and 37% with rituximab. Most patients treated with steroid-sparing therapies were prescribed 1 medication (rituximab [53/60], mycophenolate [17/21], and sirolimus [8/17]). For those treated with >1 steroid-sparing therapy, patients most often were initiated on rituximab, and then mycophenolate or sirolimus was added.
The number of subsequent warm AIHA episodes ranged from 0 to 9. In subsequent episodes of AIHA, steroid-sparing agents were used in 63 of 399 patients (16%; Table 6). Complete response to treatment also varied in these subsequent episodes: 65% with mycophenolate, 67% with sirolimus, and 47% with rituximab.Table 6.Treatment outcomes for subsequent episodes of autoimmune hemolytic anemia overall and by type of autoimmune hemolytic anemia (N = 399)Warm AIHA (n = 257)Cold agglutinin syndrome (n = 47)Paroxysmal cold hemoglobinuria (n = 24)Mixed/other AIHA (n = 33)Unspecified AIHA (n = 38)Overall (N = 399)nmedian (range)nmedian (range)nmedian (range)nmedian (range)nmedian (range)nmedian (range)No. of all subsequent episodes per patient2220 (0-9)400 (0-1)220 (0-1)300 (0-8)290 (0-3)3430 (0-9)n (%)n (%)n (%)n (%)n (%)n (%)Patients treated with steroid-sparing therapy56 (22)1 (2)0 (0)5 (15)1 (3)63 (16)Proportion of patients treated∗ Corticosteroids72 (28)1 (2)1 (4)6 (18)2 (5)82 (21) IVIG19 (7)0 (0)0 (0)3 (9)1 (3)23 (6) Rituximab36 (14)0 (0)0 (0)3 (9)0 (0)39 (10) Mycophenolate20 (8)1 (2)0 (0)2 (6)1 (3)24 (6) Sirolimus14 (5)0 (0)0 (0)1 (3)0 (0)15 (4)Of patients treated with corticosteroids, response to corticosteroids CR41 (57)0 (0)1 (100)5 (83)2 (100)49 (60) PR25 (35)1 (100)0 (0)1 (17)0 (0)27 (33) No response6 (8)0 (0)0 (0)0 (0)0 (0)6 (7) Unknown000000Of patients treated with IVIG, response to IVIG CR4 (22)0 (0)0 (0)1 (33)0 (0)5 (23) PR9 (50)0 (0)0 (0)2 (67)1 (100)12 (55) No response5 (28)0 (0)0 (0)0 (0)0 (0)5 (23) Unknown100001Of patients treated with rituximab, response to rituximab CR16 (46)0 (0)0 (0)2 (67)0 (0)18 (47) PR13 (37)0 (0)0 (0)1 (33)0 (0)14 (37) No response6 (17)0 (0)0 (0)0 (0)0 (0)6 (16) Unknown100001Of patients treated with mycophenolate, response to mycophenolate CR13 (68)0 (0)0 (0)1 (50)1 (100)15 (65) PR5 (26)1 (100)0 (0)1 (50)0 (0)7 (30) No response1 (5)0 (0)0 (0)0 (0)0 (0)1 (4) Unknown100001Of patients treated with sirolimus, response to sirolimus CR9 (64)0 (0)0 (0)1 (100)0 (0)10 (67) PR2 (14)0 (0)0 (0)0 (0)0 (0)2 (13) No response3 (21)0 (0)0 (0)0 (0)0 (0)3 (20) Unknown000000∗As a proportion of the total number of patients.
Other treatments included 6-mercaptopurine (n = 2), hydroxychloroquine, ocrelizumab, tocilizumab, tofacitinib, cyclophosphamide, tacrolimus, and hematopoietic stem cell transplant.
Clinical course
Most patients with IEI, autoimmunity of other organ systems, and ES were diagnosed after AIHA diagnosis. In this cohort, 43 of 53 (81%) were diagnosed with IEI after AIHA diagnosis, 44 of 58 (76%) were diagnosed with autoimmunity of other organ systems after AIHA diagnosis, and 109 of 136 (80%) were diagnosed with ES after AIHA diagnosis.
The majority of patients had a single episode of AIHA that remitted (217/373 [58%]; Table 1). Chronic (>6 months duration) and recurrent courses of AIHA were not uncommon, including 2 episodes of AIHA that remitted (29/373 [8%]), multiple relapsing episodes (26/373 [7%]), a chronic course that remitted (48/373 [13%]), or a chronic ongoing course (53/373 [14%]). The median total number of AIHA episodes was 1 (range, 1-10).
The majority of patients with cold antibodies had a single episode of AIHA that resolved (cold agglutinin syndrome [43/45 (96%)] and paroxysmal cold hemoglobinuria [22/23 (96%)]; supplemental Table 12). Of the 156 patients who had chronic/relapsing AIHA, the majority had warm AIHA (82%) or mixed/other AIHA (11%). Patients with chronic/relapsing AIHA were more likely to have had a genetic evaluation for an IEI (50% vs 20%; P < .001) and received treatment with a steroid-sparing agent in the first episode (35% vs 18%; P < .001).
Those with chronic/relapsing AIHA did not have a significantly different age at presentation of the first AIHA episode or in the rate of IEI diagnosis.
Patients with chronic/relapsing AIHA were at higher risk of having ES than those with a single episode of AIHA (OR, 4.85; P < .001; Figure 3; supplemental Table 13). Patients with chronic/relapsing episodes of AIHA were more likely than those with a single episode to be female (OR, 1.55; P = .038) and have abnormal immune testing results (OR, 2.29; P = .012), an autoimmune disorder (OR, 3.29; P < .001), and a family history of any immune cytopenia, immunodeficiency, or systemic autoimmunity (OR, 1.99; P = .013).Figure 3.Forest plot of univariate logistic regression models of factors associated with chronic/relapsing episodes of AIHA vs a single episode (N = 373).
Outcomes
With a median length of follow-up of 4.9 years (range, 0-19.4; IQR, 1.6-8.1; n = 398; Table 1; supplemental Figure 3), of 257 patients with warm AIHA, 72 (28%) had active disease on treatment, 7 (3%) had active disease and were being observed, and 175 (69%) were in remission (supplemental Table 14). At the time of last follow-up, 126 of 392 patients (32%) had a negative direct antiglobulin test.
Mortality was 2.5% (10/399) and was significantly higher in those with a chronic/relapsing AIHA course (4% vs 1%; P = .038). Of the 10 patients who died (supplemental Figures 4-8), 5 had ES, 2 had a solid organ transplant, 2 had a systemic autoimmune disorder, 2 had an identified IEI, and 4 had a malignancy. For these 10 patients, the causes of death included complications after post-AIHA hematopoietic stem cell transplant (n = 3), complications after solid organ transplant (n = 1), malignancy (n = 2), and infection (n = 4).
Discussion
Pediatric AIHA is a rare and heterogeneous disease with limited data available regarding disease course and outcomes. This multicenter North American collaborative effort from 15 centers describes the clinical features, current management strategies, and outcomes in this rare disease population, identifying significant unmet medical needs in terms of both diagnostic evaluation and treatment. Even among these institutions with expertise in immune hematology, patients were inconsistently evaluated with comprehensive immune testing, highlighting the need for pediatric-specific guidance to standardize evaluations.
There were high rates of underlying diagnoses, including ES, other autoimmunity, and IEI, in patients who had complete evaluations. In this cohort, most children had warm AIHA, and more than one-third had a chronic/relapsing course. Patients with chronic/relapsing AIHA had the highest likelihood of underlying autoimmune or immune conditions and a higher mortality rate, underlining the need for comprehensive diagnostic evaluation, multidisciplinary management, and better treatment approaches.
This study confirms previously published reports by the CEREVANCE group establishing the high rate of IEI in children with AIHA regardless of known infection at the time of diagnosis (supplemental Table 15).5 The high frequency of ES in children with AIHA and higher mortality in those with ES are also similar to prior reports. This study adds key new information about AIHA in children, including identification of clinical laboratory tests associated with IEI and other autoimmune disorders, as well as clinical features associated with risk of a chronic or relapsing course. The frequency of treatment with steroid-sparing therapies and the efficacy of steroid-sparing therapies in the first month after diagnosis is also described, suggesting potential benefits of early treatment with these agents in some patients.
The variability in diagnostic evaluation observed in this cohort, including frequencies of immunologic studies, autoimmune evaluation, and genetic testing, reflects the heterogeneity of real-world clinical practice. It is important to note that among those who had relevant testing, there was a high rate of abnormalities and additional diagnoses. This may suggest bias in those patients who underwent systematic evaluation or could also suggest that failure to evaluate for additional diagnoses places patients at risk of incomplete understanding of their disease.
It is possible that consistent implementation of comprehensive immunologic and genetic testing may reveal additional immunologic differences associated with IEI that were not identified in this cohort.
An IEI or other autoimmunity was more likely to be identified in children diagnosed with AIHA at an older age. Furthermore, patients who had an IEI were more likely to have abnormalities in immunoglobulin levels, antinuclear antibody testing, antiphospholipid antibody testing, other rheumatologic antibody testing, and/or complement testing, suggesting that these tests may be most helpful in distinguishing patients with AIHA who have an IEI or other autoimmunity. Testing for these conditions is most likely to be useful among patients with warm or mixed AIHA, who were significantly more likely to have an IEI than those with cold subtypes of AIHA (IEI in warm AIHA, [55/68 (81%)], mixed and unspecified AIHA [19/68 (13%)], and none with paroxysmal cold hemoglobinuria or cold agglutinin syndrome).
This distinction is clinically and prognostically significant, because the presence of an IEI or other autoimmunity was correlated with chronic/relapsing course and worse outcomes, including a higher mortality rate. Identifying underlying diagnoses based on these results can lead to modification in follow-up and treatment strategies targeted to the secondary diagnosis. Although only one-third of patients had genetic testing, one-third of those tested had pathogenic findings on gene panels; findings that also affect monitoring and treatment approaches.
In this cohort, the rate of pathogenic genetic findings was comparably high in patients with warm AIHA with and without ES (32% vs 30%), which may reflect a bias in those patients who were evaluated with genetic testing. However, these high rates of AIHA due to an IEI, both in this study and in previous reports, indicate a need for wider availability of genetic testing with systematic testing in all children with warm AIHA.5,22,23
At the time of first presentation with AIHA, many patients had positive testing for an infection. These patients did not have a lower risk of an underlying IEI or other autoimmunity. Therefore, positive infectious testing should not affect decision-making regarding further evaluation for an underlying immune or autoimmune process.
These data support immunologic evaluation in all patients with warm AIHA regardless of any associated infection at the time of presentation.
Long courses of corticosteroids, often for greater than 3 to 6 months, continue to be a mainstay of treatment of warm AIHA, both initially and at relapse, despite the significant morbidity associated with high doses and prolonged courses. This is likely due to the high efficacy of steroids in the treatment of warm AIHA and the life-threatening nature of acute autoimmune hemolysis. More than half of patients were treated with steroid-sparing agents within the first month of diagnosis, with high response rates.
Although there were not standardized criteria for initiation of steroid-sparing therapy during the study period, these data suggest that there may be a role for early initiation of steroid-sparing agents in pediatric patients with warm AIHA. In this cohort, there were comparable overall response rates between sirolimus, mycophenolate, and rituximab. Patients particularly likely to require steroid-sparing therapies include those with highest risk of relapse or chronic disease: female sex, diagnosis of ES, presence of other autoimmunity or IEI, and positive family history for autoimmune or immune disease.
There was also a high rate of corticosteroid treatment in patients with paroxysmal cold hemoglobinuria and cold agglutinin syndrome despite the lack of evidence for efficacy in these settings.24,25 The high rate of steroid use in these patients may reflect diagnostic uncertainty in the initial treatment period and, given the high morbidity and mortality of untreated warm AIHA, a tendency toward treatment with corticosteroids while the type of AIHA is being discerned. It is important to discontinue corticosteroids once these cold autoantibody subtypes are diagnosed, given their lack of efficacy in this setting and associated side effects. There was a low rate of complete response to IV immunoglobulin (IVIG) in this population.
In addition to relative lack of efficacy, IVIG also carries a risk of worsening hemolysis as a known side effect.22 Although there may be a role for IVIG in certain clinical situations, such as in patients with concomitant hypogammaglobulinemia and infection, the low rates of complete response to IVIG in this cohort suggest limited utility of IVIG for the management of AIHA alone, consistent with prior studies demonstrating that it is not a recommended standard therapy for AIHA.2,15,26
In this cohort, pediatric patients with ES had a high rate of underlying immune diagnoses, as has been previously reported.27, 28, 29, 30 Patients with ES were not more likely to have other autoimmunity than patients with AIHA alone, reflective of the high rate of other autoimmunity among all patients with AIHA. Patients with ES had a similar likelihood of complete response to steroids as patients with isolated AIHA. Most patients with ES had a combination of AIHA and ITP.
In those with ES, treatment strategies may differ based on the other cytopenia(s) present and the underlying immune process. A detailed classification system for patients with ES, for example, by phenotype or underlying IEI, may allow for an improved management approach.
The 2.5% overall mortality rate in this cohort was similar to the mortality rate reported previously.5,7 All patients who died had underlying processes including ES, preceding solid organ transplant, post-AIHA hematopoietic stem cell transplants, IEI, and/or malignancies and died of complications of treatment and/or their underlying non-AIHA disease. Underlying diagnoses were associated with chronic/relapsing AIHA and worse outcomes, highlighting the importance of comprehensive diagnostic testing. These are high-risk subpopulations of patients who require close monitoring and improved treatment strategies.
There were high rates of infections in these patients, suggesting that careful consideration should be given to antimicrobial prophylaxis for patients in these high-risk groups.
This study was limited by the availability of retrospective observational data. Available data may be biased by the cohort of patients who underwent immunologic and genetic testing, with a higher likelihood of positive results in those tested. Furthermore, because the participating ICON centers are subspecialized academic medical centers, there may be higher rates of testing and/or more availability of expansive immune testing at these locations than in other centers.
Additionally, although the follow-up period in this study was long, it was limited, such that it remains unclear what additional complications, comorbidities, or mortality may arise for these patients with childhood-onset AIHA as they age through adulthood. Given the limited length of follow-up, it is possible that there may be a higher rate of ES or relapsing AIHA observed with longer follow-up regarding the outcomes of patients with childhood-onset AIHA throughout their entire life span. It is also possible that higher rates of antinuclear antibody positivity and/or progression to lupus may be observed with longer follow-up.
Future studies could evaluate autoimmune disease progression, responses to second-line agents, and mortality in this cohort after a longer follow-up duration.
The data from this large multicenter cohort study of pediatric AIHA identify a high rate of underlying immune diagnoses, including genetic IEI diagnoses, and a high rate of up-front response to steroid-sparing therapies in children with AIHA. Those with warm and mixed AIHA are at the highest risk of a relapsing/chronic course, ES, and associated immunologic disorders, highlighting the need for comprehensive diagnostic evaluation and multidisciplinary management. These data support the need for immunologic and genetic testing in all pediatric patients with warm and mixed AIHA.
There remains significant heterogeneity in diagnostic testing and management strategies in clinical practice, underscoring the need for evidence-based guidelines for the diagnostic evaluation and treatment of children with AIHA.
Conflict-of-interest disclosure: M.S. reports advisory board fees from Novo Nordisk. S.M.B. reports consultancy fees from Agios Pharmaceuticals, Beam, Bristol Myers Squibb (BMS), Chiesi, Editas, Sobi, X4 Pharmaceuticals, and Vertex Pharmaceuticals. K.H. reports advisory board fees from Sobi; and funding from Sala Elbaum Pediatric Research Scholar.
R.J.K. reports consultancy fees from Agios Pharmaceuticals, Amgen Inc, Bayer Canada, Hoffman-La Roche, Novo Nordisk, Octapharma AG, Sanofi, and Takeda. K.A.S. reports research funding from Novartis, Sobi, and Sanofi. T.A.N. reports consulting fees from Novartis, Sanofi, and OnKure.
A.R. reports advisory board fees from Novartis, Bluebird Bio, Horizon Therapeutics, and Chiesi; and speaker bureau fees from X4 Pharmaceuticals. A.M. reports consulting fees from Metagenomi. S.N. reports consulting fees from Star Therapeutics.
S.R.-Z. reports consultancy fees from Bluebird Bio and Vertex. C.M. reports employment with Takeda Pharmaceuticals USA. W.B.L. reports data safety monitoring board participation for Medpace, Jubilant DraxImage, Merck, and Memorial Sloan Kettering Cancer Center.
J.A.R. reports research funding from Agios Pharmaceuticals, Novo Nordisk, and Bluebird Bio/Genetix Biotherapeutics. M.P.L. reports advisory board fees from Janssen, Sobi, Argenx, Platelet Disorder Support Association (PDSA), 22q Society, and Cornelia de Lange Syndrome Foundation; consulting fees from Novartis, Argenx, Agios Pharmaceuticals, Sobi, Sanofi, and Janssen; and research funding from Foundation for Women & Girls with Bleeding Disorders, PDSA, National Institutes of Health, Sysmex, Novartis, Principia, Janssen, Argenx, Dova, Octapharma, Sobi, and Sanofi. R.F.G. reports research funding from Agios Pharmaceuticals, Sobi, and Novartis; and consultancy fees from Agios Pharmaceuticals, Sanofi, and Sobi.
The remaining authors declare no competing financial interests.