An Evidence-Based Guideline Improves Outcomes for Patients with Hemophagocytic Lymphohistiocytosis and Macrophage Activation Syndrome

Abstract

Objective:

To compare clinical outcomes in children with hemophagocytic lymphohistiocytosis (HLH) and macrophage activation syndrome (MAS) who were managed before and after implementation of an evidence-based guideline (EBG).

Methods:

A management algorithm for HLH/MAS was developed at our institution based on literature review, expert opinion, and consensus building across multiple pediatric subspecialties. An electronic medical record (EMR) search retrospectively identified hospitalized patients with HLH/MAS in the pre-EBG (10/15/15–12/4/17) and post-EBG (1/1/18–1/21/20) time periods. Predetermined outcome metrics were evaluated in the two cohorts.

Results:

After the EBG launch, 57 children were identified by house staff as potential HLH/MAS patients, and rheumatology was consulted for management. Ultimately, 17 patients were diagnosed with HLH/MAS by the treating team. Of these, 59% met HLH 2004 Criteria, and 94% met 2016 MAS Classification Criteria for Systemic Juvenile Idiopathic Arthritis (sJIA). There was a statistically significant reduction in mortality from 50% before implementation of the EBG to 6% in the post-EBG cohort (p=0.02). There was a significant improvement in time to 50% reduction in CRP level in the post-EBG vs. the pre-EBG cohorts (log-rank p<0.01). There were trends towards faster time to HLH/MAS diagnosis, faster initiation of immunosuppressive therapy, shorter length of hospital stay, and more rapid normalization of HLH/MAS-related biomarkers in the post-EBG patients.

Conclusions:

While the observed improvements may be partially attributed to advances in treatment of HLH/MAS that have accumulated over time, this analysis also suggests that a multidisciplinary treatment pathway for HLH/MAS contributed meaningfully to favorable patient outcomes.

Introduction

Hemophagocytic lymphohistiocytosis (HLH) and macrophage activation syndrome (MAS) are related disorders defined by hyperinflammation. HLH and MAS are characterized by T cell and macrophage activation, which lead to exuberant release of pro-inflammatory cytokines^1–6. HLH occurs in individuals with gene defects that impair cytotoxicity (familial HLH, FHL); but it can also develop in highly inflammatory states, such as infection and malignancy (secondary HLH)^7–14. MAS is a form of secondary HLH that complicates rheumatologic disorders⁶. Regardless of terminology, these conditions share an underlying pathology defined by excessive inflammation^3,4,15.

There are multiple obstacles that make the management of HLH/MAS challenging. HLH/MAS typically evolves in the setting of an infection, malignancy, or autoimmune condition where some degree of inflammation is expected. The telltale signs of hyperinflammation can be incorrectly attributed to the triggering illness, resulting in a delayed or missed HLH/MAS diagnosis^16–18. The multi-organ involvement of HLH/MAS often necessitates multi-disciplinary care, which may be difficult to coordinate^4,19. While HLH/MAS was traditionally treated with chemotherapy, there are now multiple anti-cytokine agents available that have demonstrated promise with less associated toxicity^20–26. This plethora of options, coupled with the lack of studies directly comparing chemotherapy- vs. non-chemotherapy-based protocols, can generate uncertainty and delays in selecting first-line therapies. These complexities are compounded by the rapid pace of disease escalation which is characteristic of HLH/MAS. Any delay in the institution of immunosuppression can result in significant morbidity for the patient.

To address the challenges intrinsic to managing HLH/MAS, we implemented an evidence-based guideline (EBG) through consensus-building across multiple pediatric subspecialties²⁷. The objectives of this effort were to facilitate the early diagnosis and the rapid initiation of immunomodulation in HLH/MAS while reducing practice variability. Herein, we compare quality of care metrics in patients with HLH/MAS before and after the launch of the EBG and show that use of this guideline was associated with improved clinical outcomes.

Methods

The methods used to develop the HLH/MAS EBG have been described previously²⁷. The EBG was launched on December 4, 2017. The pre-EBG time period was defined as October 15, 2015 to December 4, 2017. The post-EBG time period extended from January 1, 2018 to January 21, 2020. Study subjects were identified retrospectively through an electronic medical record (EMR) search algorithm that captured all patients admitted to the hospital with a consult note from rheumatology and/or oncology, temperature ≥38.2°C, and ferritin level ≥500 ng/mL. An attending rheumatologist reviewed the chart of each patient. Patients were included in the analysis if the treating team diagnosed the child with new-onset HLH/MAS. For patients with multiple admissions for HLH/MAS, the index hospitalization with the first rheumatology/oncology consult note was included. Clinical characteristics were gathered and entered into a REDCap database. Quality improvement (QI) metrics were pre-specified during development of the EBG and were calculated and recorded for each patient²⁷ (Table1).

Table 1.

Pre-Specified Quality Metrics to Evaluate the HLH/MAS EBG

Recognition & Treatment of HLH/MAS

1. Time from admission to HLH/MAS diagnosis (days)

2. Time from admission to initiation of HLH/MAS-directed therapy (days)

Duration of Illness

1. Fever duration (days)

2. Length of hospital stay (days)

3. Hospital readmission within 60 days (Y/N)

Severity of Illness

1. Need for higher level of care (Y/N)

2. Mortality during admission (Y/N)

3. Mortality at a later time (Y/N)

Normalization of HLH/MAS Laboratory Parameters

1. Time to decrease in ferritin by 50% (days)

2. Ferritin decrease by 50% during admission (Y/N)

3. Time to decrease in CRP by 50% (days)

4. Time to CRP <1 mg/dL (days)

5. Normalization of platelet count during the admission (Y/N)

6. Normalization of liver function tests during the admission (Y/N)

HLH, hemophagocytic lymphohistiocytosis; MAS, macrophage activation syndrome; EBG, evidence-based guideline; Y, yes; N, no; CRP, C-reactive protein.

The process measures included time to HLH/MAS diagnosis/treatment, and the primary outcomes measures included duration of illness, severity of illness, and time to normalization of laboratory parameters. We also tracked the number of rheumatology/oncology consults to measure resource utilization. A complete treatment response was defined as full control of HLH/MAS and hospital discharge without the need to add further immunomodulatory treatments. The study was approved by the Institutional Review Board at Boston Children’s Hospital (IRBP00020692).

Continuous variables were compared using Mann-Whitney U tests. Fisher’s exact tests were used to compare categorical variables. Kaplan-Meier survival curves were used to estimate the time from admission to HLH/MAS diagnosis, time from admission to initiation of HLH/MAS-directed therapy, time to 50% reduction in ferritin value from the peak ferritin level during admission, and time to 50% reduction in C-reactive protein (CRP) value from the peak CRP level during admission. For the CRP and ferritin survival analyses, patients were included if they had abnormal values for these parameters that were trended over time. Censoring occurred at the time of hospital discharge or death if 50% reduction in ferritin or CRP levels was not observed. Two patients had prolonged hospital admissions for indications other than HLH/MAS and were excluded from the analysis for length of hospital stay, time to HLH/MAS diagnosis/treatment, and duration of fever. Log-rank tests were used to compare differences in survival distributions between the pre- and post-EBG groups. GraphPad Prism version 8.0 was used for the statistical analyses.

Results

Implementation of an EBG for the management of HLH/MAS.

As described previously, we developed a collaborative approach to the diagnosis and treatment of patients hospitalized for HLH/MAS at our institution²⁷. Briefly, a multidisciplinary work group used nominal group technique to achieve consensus and create a clinical algorithm for HLH/MAS. Entry criteria (fever, ferritin ≥500 ng/mL) were developed to alert the house staff when to consider HLH/MAS in a hospitalized patient. The rheumatology consult team was identified as the “gate-keeper” and assumed the responsibility of coordinating the diagnostic evaluation for HLH/MAS. First-line immunomodulatory treatments were recommended in the EBG based on the acuity of illness and risk for infection. To facilitate implementation of the EBG, campaign materials were developed, educational sessions were held with house staff and sub-specialty providers. The EBG was launched in December 2017 and updated in January 2021 (Supplementary File 1).

Identification of patients managed by the EBG.

An EMR search algorithm was developed to systematically identify patients who entered the HLH/MAS EBG. In the post-EBG period, 86 inpatients with a rheumatology or oncology consult note, fever ≥38.2°C and ferritin ≥500 ng/mL were flagged as potential HLH/MAS cases. Upon further chart review, 57 entered the EBG, and rheumatology was consulted for an evaluation for HLH/MAS (Figure1 TableS1). After assessment, 30 patients underwent a diagnostic evaluation for HLH/MAS, and 17 were ultimately considered to have HLH/MAS by the treating providers. The same EMR search was used to find 81 potential patients with HLH/MAS in the pre-EBG period (Figure1, TableS1). After chart review, 30 individuals were found to have undergone a diagnostic evaluation for HLH/MAS, and 10 received a diagnosis of HLH/MAS.

Figure 1. Inclusion & Exclusion of Patients Identified in the EMR Search Algorithm.

EMR, electronic medical record; EBG, evidence-based guideline; rheum, rheumatology; once, oncology; HLH, hemophagocytic lymphohistiocytosis; MAS, macrophage activation syndrome

Clinical characteristics of patients with HLH/MAS.

The most common clinical manifestations of HLH/MAS were persistent fevers, rash, and hepatosplenomegaly (Table2). The median ferritin level was 12,188 ng/dL and 3,082 ng/dL in pre-EBG and post-EBG groups, respectively. C-reactive protein (CRP) and alanine transaminase (ALT) values were higher while the fibrinogen level was lower in the pre-EBG group. In the pre-EBG patients, 3/10 met HLH 2004 Diagnostic Criteria and 7/10 fulfilled the 2016 Classification Criteria for MAS Complicating sJIA^21,28. Specialized studies required to fulfill the HLH 2004 Criteria were not sent in full on many pre-EBG patients, potentially explaining the low number of pre-EBG patients who qualified for the HLH diagnosis. NK cell function, bone marrow biopsy, and genetic sequencing were performed in 4/10, 3/10, and 5/10 pre-EBG patients, respectively. For children in the post-EBG group, 10/17 and 16/17 met the HLH 2004 and 2016 MAS Classification Criteria, respectively. Nearly one-quarter of patients in the pre- and post-EBG cohorts had a pre-existing diagnosis of systemic juvenile idiopathic arthritis (sJIA). Infection was identified in a majority of cases (~80%) as the trigger for HLH/MAS. Two of the 17 children in the post-EBG group had an underlying malignancy as the cause of HLH/MAS. During the diagnostic evaluation, almost 25% of patients in the post-EBG cohort were found to have a genetic cause of HLH/MAS with bi-allelic, pathogenic variants uncovered in PRF1 (n=1), UNC13D (n=1), STAT2 (n=1), and COG4 (n=1) while a single child with a heterozygous and disease-causing variant in NLRC4 was identified in the pre-EBG cohort.

Table 2.

Clinical Characteristics of Patients with HLH/MAS

	Pre-EBG N=10	Post-EBG N=17
Age (mean ± SD)	10.0 ± 9.2	9.5 ± 7.7
Sex (#, % female)	(8, 80%)	(8, 47%)
Clinical manifestations (#, % patients)
Persistent Fever	(7, 70%)	(17, 100%)
Rash	(5, 50%)	(5, 29%)
HSM	(5, 50%)	(5, 29%)
Coagulopathy	(5, 50%)	(3, 18%)
Neurologic Involvement	(2, 20%)	(4, 24%)
Pre-existing Rheum Diagnosis (#, % patients)
sJIA	(2, 20%)	(4, 24%)
SLE	(1, 10%)	(1, 6%)
Autoinflammatory♦	(1, 10%)	(0, 0%)
KD	(0, 0%)	(0, 0%)
Other□	(0, 0%)	(1, 6%)
HLH/MAS Trigger (#, % patients)
Infection	(8, 80%)	(14, 82%)
Autoimmune/Autoinflammatory flare	(2, 20%)	(0, 0%)
Malignancy	(0, 0%)	(2, 12%)
Other◊	(4, 40%)	(2, 12%)
Cytopenias* (#, % patients)	(4, 40%)	(12, 71%)
Highest ferritin level, ng/dLΔ○ (median, range)	12,188, 6,741–100,000	3,082, 526–36,040
Highest CRP level, mg/dLΔ (mean ± SD)	16.3 ± 12.0	6.0 ± 5.8
Highest ALT level, units/LΔ (mean ± SD)	409 ± 549	226 ± 255
Lowest fibrinogen level, mg/dLΔ (mean ± SD)	192 ± 93	211 ± 168
Abnormal elevation in sIL2R level (#, % patients)	(7, 70%)	(10, 59%)
Abnormal NK cell function (#, % patients)	(2, 20%)	(3, 18%)
Hemophagocytosis (#, % patients)	(2, 20%)	(4, 24%)
HLH 2004 Criteria3 (#, % patients)	(3, 30%)	(10, 59%)
2016 MAS Classification Criteria2,Ϯ (#, % patients)	(7, 70%)	(16, 94%)
Identified Genetic Diagnosis (#, % patients)
FHL▲	(0, 0%)	(2, 12%)
PID▼	(0, 0%)	(1, 6%)
Other●	(1, 20%)	(1, 6%)

^♦In the pre-EBG group, there was one patient with an undifferentiated autoinflammatory disorder.

^□In the post-EBG group, there was a patient with ANCA vasculitis.

^◊Other triggers for HLH/MAS included decreased immunosuppression for the underlying rheumatologic disorder, n=1; renal failure, n=1; a history of atypical HUS, n=1; and cystic fibrosis s/p lung transplant, n=1 in the pre-EBG group and decreased immunosuppression for the underlying rheumatologic disorder, n=1; an unknown pre-existing primary immunodeficiency, n=1 in the post-EBG group.

^*Cytopenias affecting ≥2 of 3 lineages in the peripheral blood with Hgb <9 gm/dL, plts <100×10³/mL, Neutrophils <1×10³/mL

^ΔThe highest or lowest indicated laboratory value recorded during the hospital admission.

^○2 patients in the pre-EBG group had a ferritin level >100,000 ng/dL, which was considered a value of 100,000 to calculate the median ferritin level.

^ϮPatient fulfilled the 2016 MAS Classification Criteria except for sJIA diagnosis.

^▲In the post-EBG group, biallelic mutations in PRF1 (n=1) and UNC13D (n=1) were found.

^▼In the post-EBG group, a patient with pathogenic and homozygous variants in STAT2 was found.

^●In the pre-EBG group, other genetic diagnosis uncovered during the diagnostic evaluation included a pathogenic and heterozygous variant in NLRC4, n=1. In the post-EBG group, one patient had compound heterozygous mutations in COG4 (congenital glycosylation defect type 2).

HLH, hemophagocytic lymphohistiocytosis; MAS, macrophage activation syndrome; EBG, evidence-based guideline; HSM, hepatosplenomegaly; sJIA, systemic juvenile idiopathic arthritis; SLE, systemic lupus erythematosus; KD, Kawasaki disease; CRP, C-reactive protein; ALT, alanine aminotransferase; sIL2R, soluble IL-2 receptor; NK, natural killer; FHL, familial HLH; PID, primary immunodeficiency.

HLH- and MAS-directed treatment in the pre- and post-EBG cohorts.

Eight patients in the pre-EBG period and 15 in the post-EBG cohort received immunomodulatory therapy for HLH/MAS. Of note, 4 children with HLH/MAS who were not treated with immunosuppression experienced a clinical improvement without intervention (n=3) or after treatment of the underlying condition that triggered the HLH/MAS (n=1) (see supplemental Table2 for more information). The management recommendations outlined in the EBG were followed in 16/17 patients. The single deviation occurred in a child with immune dysregulation who was managed by oncology without rheumatology’s input.

The EBG provides a list of immunomodulatory medications that can be used alone or in combination as first-line treatment of HLH/MAS, including anakinra, intravenous immunoglobulin (IVIG), cyclosporine, tacrolimus, and glucocorticoids. After implementation of the EBG, IVIG (n=6), IL-1 blockade (n=6), and glucocorticoids (n=5) were the most commonly selected medications for the initial treatment of HLH/MAS (Figure2). IL-1 blockade and IVIG were used at higher frequencies in the post-EBG group. The increased use of IVIG was notable (10% of pre-EBG vs. 35% of post-EBG patients). IVIG monotherapy was selected as first-line treatment in patients who were either non-critically ill (4/6 patients treated with IVIG) or in patients where there was a high degree of concern for an invasive infection (4/6). More patients in the pre-EBG cohort received the HLH 2004 protocol (20% of pre-EBG vs. 6% in the post-EBG groups). Tocilizumab was not included in the EBG recommendations and it was prescribed for one pre-EBG and no post-EBG patients. Twenty-five percent of pre-EBG compared to 67% of post-EBG patients had a complete response to first-line therapy, although this difference was not statistically significant (p=0.09, Fisher’s exact test).

Figure 2. First-line Immunomodulatory Treatment for HLH and MAS.

The bar graph depicts the proportion of patients in the pre- and post-EBG cohorts treated with the given medications.

EBG, evidence-based guideline; HLH, hemophagocytic lymphohistiocytosis; MAS, macrophage activation syndrome; IVIG, intravenous immunoglobulin

Clinical Outcomes after Implementation of an EBG for HLH/MAS.

A goal of the EBG was to decrease the time to HLH/MAS diagnosis and treatment. The median time from hospital admission to HLH/MAS diagnosis was 4 days (IQR 31.3 days) in the pre-EBG period compared to 2 days (IQR 3.5 days) in the post-EBG era, although this difference did not achieve statistical significance (Figure3A). Similarly, there was a non-significant trend towards faster initiation of treatment in the post-EBG group with a median time from admission to the initiation of HLH/MAS-directed therapy of 5 days (IQR 40.0 days) in the pre-EBG cohort compared to 2 days (IQR 5.0 days) after EBG launch (Figure3B).

Figure 3. Clinical Outcomes in the Pre- and Post-EBG Cohorts.

Kaplan-Meier estimates of the cumulative probability of A) remaining without a diagnosis of HLH/MAS over days since hospital admission stratified by pre- (n=8) and post-EBG (n=17) cohorts, or B) remaining without HLH/MAS directed immunomodulatory therapy over days since hospital admission, stratified by pre- (n=7) and post-EBG (n=15) cohorts. C) Median number of febrile days with interquartile range (IQR) in the pre- and post-EBG patients. D) Median length of hospital admission with IQR in the pre- and post-EBG patients. E) Cumulative probability of not yet achieving 50% reduction in ferritin over days since peak ferritin level, by pre- (n=9) and post-EBG (n=16) cohorts. F) Cumulative probability of not yet achieving 50% reduction in CRP over days since peak CRP level, by pre- (n=7) and post-EBG (n=13) cohorts. The p-values represent log-rank tests comparing survival curves.

EBG, evidence-based guideline; HLH, hemophagocytic lymphohistiocytosis; MAS, macrophage activation syndrome; HR, hazard ratio; CI, confidence interval; CRP, C-reactive protein

To evaluate the duration of illness before and after implementation of the EBG, we measured fever duration, length of hospital admission, and need for readmission to the hospital after discharge. The median number of febrile days was 4 in the pre-EBG group compared to 5 in the post-EBG group, while the length of hospital stay was 28 vs. 11 days in these two groups, respectively (Figure3C-D). There was wide variability in number of days with fever and length of admission in both pre- and post-EBG patients, thereby limiting comparisons. During the pre-EBG period, 1 patient was re-admitted for HLH/MAS within 60 days of the discharge date. After initiation of the EBG, 5/17 patients in the post-EBG cohort were readmitted for active HLH/MAS.

Laboratory biomarkers of HLH/MAS were assessed in both cohorts. During hospital admission, liver function tests (LFTs) normalized in 20% of pre-EBG and 29% of post-EBG patients. Thrombocytopenia resolved during the hospital stay in 50% compared to 65% of children in the pre- vs. post-EBG groups. There was a trend towards faster decline in ferritin values in children enrolled in the EBG (Figure 3E). There was a statistically significant improvement in time to 50% decrease in CRP level (log-rank p<0.01) (Figure3F). The median time to a CRP of less than 1 mg/dL could not be estimated given the large proportion of patients who did not achieve this event during hospital admission.

A greater proportion of patients in the pre-EBG group were directly admitted to the intensive care unit (6/10, 60%) compared to the post-EBG cohort. Of the 6 pre-EBG patients initially managed on the pediatric ward, 2 were transferred to the ICU at a later time. In the post-EBG group, 12/17 (71%) of patients were admitted to the pediatric ward. Of these 12 post-EBG patients, 6 were transferred to the intensive care unit (ICU).

There was a statistically significant difference in mortality between the pre- and post-EBG cohorts. Half of the pre-EBG patients (5/10, 50%) died during hospital admission while there was only one fatality in the post-EBG group (1/17, 6%) (p=0.02, Fisher’s exact test) (Table3). Patients in the pre-EBG group were 8.5 times more likely to die from HLH/MAS during admission than individuals in the post-EBG cohort (RR 8.5, 95% CI 1.6–50.9). During a subsequent hospital admission, an additional patient (n=1) in the post-EBG group died from sJIA-related interstitial lung disease.

Table 3.

Mortality in the Pre and Post HLH/MAS EBG Cohorts

	Pre-EBG N=10	Post-EBG N=17	RR, 95% CI
Mortality During Admission (#, %)	5, 50%	1, 6%	8.5, 1.6–50.9
Mortality at a Later Time (#, %)	0, 0%	1, 6%	0, 0.0–6.0

HLH, hemophagocytic lymphohistiocytosis; MAS, macrophage activation syndrome; EBG, evidence-based guideline; RR, risk ratio; CI, confidence interval

Discussion

Children with HLH/MAS typically present with life-threatening hyperinflammation and multi-organ dysfunction, requiring coordinated subspecialty care. To facilitate the recognition of HLH/MAS and improve clinical outcomes, a pathway for the management of HLH/MAS was developed at our institution. As part of the EBG, predetermined quality improvement (QI) measures were instituted to track outcomes. Compared to the pre-EBG era, children treated for HLH/MAS in the post-EBG period had a significantly faster improvement in CRP level and reduction in mortality. In addition, there was a non-statistically significant trend towards earlier diagnosis, faster initiation of HLH/MAS-directed therapies, and shorter length of hospital stay in the post-EBG group. Hyperferritinemia, thrombocytopenia, and transaminitis also resolved in a greater proportion of children in the post-EBG cohort. These findings suggest that implementation of a guideline created through consensus building across multiple pediatric sub-specialties improved outcomes for children with HLH/MAS.

A key component of the HLH/MAS EBG included efforts to increase awareness of HLH/MAS in house staff and consultant services through educational materials, clinical conferences, and electronic orders sets. These efforts appeared to have had an impact on consults for HLH/MAS. In the 25.6 months prior to establishment of the guideline, 34 patients were referred to a subspecialist for consideration of a diagnosis of HLH/MAS (Figure1). In the 24.7 months after the EBG became active, consults for HLH/MAS were requested for 57 children (Figure1). At HLH/MAS diagnosis, children in the post-EBG group tended to have lower markers of disease activity compared to pre-EBG patients, including lower ferritin, ALT, and CRP levels along with higher fibrinogen values (Table 2). This would suggest that patients with HLH/MAS were recognized faster and directed to subspecialty care earlier in the disease course after establishment of the EBG. Indeed, the median time from hospital admission to HLH/MAS diagnosis was 2 days in the post-EBG group compared to 4 days in the pre-EBG group. Once recognized, children with HLH/MAS were quickly treated with immunomodulatory therapy, often on the same day as diagnosis. The median time from admission to the initiation of HLH/MAS-directed therapy was 2 days in the post-EBG patients. We suspect a significant driver in the trend towards faster diagnosis and treatment of HLH/MAS after implementation of the EBG was a change in referral pattern that resulted in quicker engagement with experts in hyperinflammation along with clinical decision support in the form of the EBG.

A second major goal of the EBG was to reduce variability in the management of HLH/MAS. Prior to the initiation of the EBG, multiple services could be consulted for a patient with suspected HLH/MAS. The EBG established rheumatology as the “gate-keeper” to engage for the initial management of HLHL/MAS. These recommendations were followed as 16/17 patients diagnosed with HLH/MAS in the post-EBG period had an initial consult with rheumatology. For treatment, the EBG provided a list of medications that could be used alone or in combination for first-line immunomodulatory therapy in HLH/MAS. Of note, patients with familial HLH (FHL) were excluded from these treatment recommendations and directed to oncology for treatment. Patients treated for HLH/MAS in the post-EBG cohort received first-line immunomodulatory treatments as per guideline recommendations (glucocorticoids (n=5), IL-1 blockade (n=6), and IVIG (n=6)) (Figure2). Interestingly, IVIG use increased most dramatically. Treatment with IVIG for MAS has been reported in the literature, but it is not thought to be as effective as glucocorticoids and anakinra.29,30 IVIG was included in the EBG to provide a therapeutic option for patients with moderate levels of illness where immunosuppression might not be warranted due to concern for a serious underlying infection. Indeed, IVIG was given to such patients in the post-EBG group, and its increased use may reflect lower disease severity in patients diagnosed earlier in the disease course. Tocilizumab was not recommended in the EBG given limited evidence supporting the efficacy of the medication for this indication⁶. In keeping with the guidelines, no HLH/MAS patients received tocilizumab in the post-EBG cohort. A smaller proportion of patients received etoposide-based therapy in the post-(1/17, 6%) vs. pre-EBG (2/10, 20%) groups. The single child treated with the HLH 2004 protocol in the post-EBG group had homozygous and pathogenic mutations in PRF1. By contrast, the patients treated with etoposide in the pre-EBG group included a child with autoinflammation with infantile enterocolitis due to a heterozygous variant in NLRC4 and another patient with an unknown inflammatory disorder. In total, these findings indicate that the recommendations outlined in the HLH/MAS EBG were largely followed in clinical practice.

The medications recommended in the EBG, are now routinely used by rheumatologists to treat MAS in the context of autoimmune/autoinflammatory diseases^30–34. Increasingly, these medications are used to treat non-rheumatologic forms of secondary HLH^23,24,29,35. Kumar and co-authors proposed a treatment algorithm for secondary HLH in adults, recommending glucocorticoids, anakinra, IVIG, and/or cyclosporine as first-line treatments²⁴. In children, Eloseily et al. reported favorable outcomes (73% survival) in a retrospective, single-center cohort of 44 children who received anakinra for HLH/MAS²³. In this study, 36% of patients lacked an underlying rheumatologic diagnosis, suggesting that anakinra was effective and safe in secondary HLH not associated with an autoimmune condition²³. The exception was malignancy-associated HLH where the mortality rate was 100%²³. In our post-EBG cohort, 67% had a full response to first-line therapy. The survival rate was 94% during the index hospital admission, which compares favorably to historical cohorts of children with MAS/HLH where survival rates have ranged from 44–73%^23,36,37. Six children in the post-EBG group (35% of the cohort) had a rheumatologic diagnosis and there were no fatalities in this subset (Table2). Of the remaining 11 patients, 2 had FHL and were treated by oncology. The remaining 9 patients with secondary HLH fared well with the approach outlined in the EBG. The single death occurred in a child who received IV glucocorticoids as first-line treatment for HLH secondary to a new diagnosis of acute myeloid leukemia (AML), supporting the findings of Eloseily et al. that malignancy-associated HLH has a poor prognosis²³. A second child with sJIA-related lung disease died during a subsequent hospitalization. Our favorable experience in using non-chemotherapy-based protocols as first-line treatment for MAS and secondary HLH provides further support for this approach in children.

Several interesting clinical characteristics were noted in this HLH/MAS cohort. The association between malignancy and secondary HLH is well known in adults but not commonly considered in the pediatric population¹². A recent case series reported by Lehmberg et al. showed that an oncologic process was found in 8% of pediatric HLH cases, a rate much higher than previously appreciated⁷. In our post-EBG cohort, close to 12% of children were noted to have malignancy as a trigger for the HLH/MAS (Table2). These findings coupled with the high mortality rates in malignancy-associated HLH highlight the importance of considering an oncologic evaluation in children with new-onset HLH/MAS^23,38. Our EBG recommends genetic testing in most patients with HLH/MAS. A genetic diagnosis was uncovered in approximately 25% of patients in the post-EBG cohort. The high frequency of causative variants in children with HLH/MAS indicates a need for systematic genetic evaluation in this population.

Our findings should be interpreted in light of several limitations of this study. The EBG was implemented and evaluated at a single quaternary center and may not be generalizable to other institutions. A small number of patients were included in the pre- and post-EBG cohorts, reflecting the relative rarity of HLH/MAS. The small sample size limited our power to detect more modest effect sizes and did not allow for techniques to adjust for differences in the study groups, such as propensity score matching. The post-EBG patients demonstrated lower levels of HLH/MAS biomarkers, indicating that these patients were less severely ill than children in the pre-EBG group. Thus, it is possible that the improved outcomes noted after EBG implementation were related to differences in disease severity across the two groups. We believe the decreased disease severity in the post-EBG patients is attributable to earlier disease recognition because of the educational efforts launched with the EBG, but we are unable to prove this definitively. We were also unable to account for death as a competing risk, which likely resulted in an overestimate of the probability of 50% CRP or ferritin reduction in the pre-EBG cohort due to the large proportion of deaths. An improved understanding of the biology of HLH/MAS coupled with advancements in diagnostic tools and treatments have likely contributed to improved quality of care and more favorable outcomes over time. In an effort to address potential bias due to these secular trends, we restricted the pre-EBG group to the 2 years before initiation of the EBG. Finally, there may be unforeseen costs associated with implementing an intervention. In the case of this EBG, there was a greater number of rheumatology/oncology consults for HLH/MAS, and many of these patients were not ultimately diagnosed with HLH/MAS. This represents an increase in resource utilization that needs to be weighed against the benefits of the EBG. In the implementation phase of the EBG, we focused on process and outcome measures. As we accumulate more patients, we plan to add balancing metrics, including adverse events related to earlier and more aggressive treatment of HLH/MAS.

In summary, implementation of a multi-disciplinary, consensus-based guideline for the management of HLH/MAS was associated with improved clinical outcomes, including a reduction in mortality. These findings highlight the importance of a collaborative and streamlined approach to the diagnosis and treatment of HLH/MAS.