Critical care outcomes of hemophagocytic lymphohistiocytosis in children: a single centre retrospective observational study

Abstract

Background

Hemophagocytic lymphohistiocytosis (HLH) is a challenging and potentially life-threatening condition characterized by an uncontrolled inflammatory response resulting from hyperactivation of T lymphocytes and macrophages, leading to excessive production of inflammatory cytokines. HLH can lead to multiple organ failure and death in a significant proportion of cases. It is challenging to differentiate HLH from sepsis. This study aims to describe the clinical characteristics, risk factors, and outcomes of a cohort of children admitted to a pediatric intensive care unit (PICU) with primary and secondary hemophagocytic lymphohistiocytosis (pHLH and sHLH).

Methods

Retrospective observational analysis of all children with HLH admitted to the PICU between 2006 and 2019.

Results

Patients were evaluated at two time points: upon PICU admission (T1) and at the time of the highest ferritin level (T2). A total of 48 patients were included: 5 with pHLH, 37 with sHLH, and 6 with macrophage activation syndrome (MAS) in the context of systemic juvenile idiopathic arthritis (sJIA). The overall mortality rate in the PICU was 33.3%. Mortality rates within each group were: 60% for pHLH, 35.1% for sHLH, and nobody for MAS. The Pediatric Index of Mortality (PIM2) was higher among non-survivors. Splenomegaly and leukopenia at PICU admission were more common in non-survivors compared to survivors, as were the incidence of multiple organ failure and acute renal failure during the PICU stay. Platelet counts were significantly lower in patients who died, but only at T2. Ferritin levels remained consistently elevated in non-survivors throughout their PICU stay.

Conclusions

In this cohort, mortality among children admitted to the PICU with HLH was high, particularly among those with primary or onco-hematological forms. Conversely, patients with MAS experienced more relapses but no mortality was observed in this sample. Early presence of splenomegaly, low platelet count, elevated ferritin levels, and PIM2 score at PICU admission appear to be indicators of poor prognosis. These findings may help to early identify patients at highest risk for complications and mortality.

Background

Hemophagocytic lymphohistiocytosis (HLH) is a challenging and potentially life-threatening condition characterized by an uncontrolled inflammatory response resulting from hyperactivation of T lymphocytes and macrophages, leading to excessive production of inflammatory cytokines. The main clinical features include fever, lymphadenopathy, hepatosplenomegaly, coagulopathy, and dysfunction of the renal, hepatic, and central nervous systems. Refractory HLH to medical therapy may progress to multiple organ failure (MOF) and death in a significant proportion of cases [1].

Typical laboratory findings encompass cytopenia affecting one or more cell lines (anemia, leukopenia, and thrombocytopenia), elevated transaminases, lactate dehydrogenase (LDH), ferritin, D-dimers, and decreased fibrinogen levels [2]. HLH can be classified as primary (pHLH), also known as familial HLH, or secondary (sHLH), also referred to as reactive HLH. pHLH results from genetic defects involved in the cytotoxic pathway, whereas sHLH can be triggered by various conditions such as infections, rheumatological diseases, and malignancies [3–4]. However, in a notable percentage of cases, no specific trigger can be identified [5].

Infectious triggers are predominantly herpesviruses, especially Epstein-Barr virus (EBV) and cytomegalovirus (CMV), as well as influenza viruses (notably H1N1) and Leishmania. Rheumatological diseases are also common triggers of sHLH, particularly systemic juvenile idiopathic arthritis (sJIA), systemic lupus erythematosus, Kawasaki disease, and systemic vasculitis. These forms are often referred to as macrophage activation syndrome (MAS) [6–10]. Other predisposing conditions include malignancies (most frequently lymphoma and leukemia), immunological disorders, and metabolic diseases. In 15–40% of cases, no identifiable trigger for sHLH can be determined [11].

Severe forms of HLH may necessitate admission to the PICU and the use of advanced organ support [11–12]. In this context, accurate differential diagnosis from acute conditions such as sepsis—which requires entirely different treatment strategies—is essential to improve patient outcomes [12]. Therefore, the primary objective of this study is to describe the clinical characteristics, risk factors—including the need for extracorporeal therapies—laboratory parameters, and prognostic outcomes (such as duration of mechanical ventilation, length of PICU stay, and survival) of patients admitted to the PICU with HLH. Additionally, the study aims to identify early predictors of clinical outcomes.

Methods

We conducted a retrospective review of all patients with primary and secondary HLH who required PICU admission from 2006 to 2019. The diagnosis of HLH was based on the HLH-2004 diagnostic guidelines [13] and the MAS-2016 classification criteria [14]. As the ACR 2016 classification criteria [14] are used only to identify patients with MAS in the context of sJIA, we have separates MAS-sJIA from MAS secondary to other rheumatic conditions. We have used the HLH-2004 classification criteria [13] to diagnose the other forms of HLH.

The study received approval from the Institutional Review Board of Children Hospital Bambino Gesù. Informed consent was waived due to the retrospective nature of the study.

Clinical features, laboratory parameters, and supportive therapies were extracted from electronic medical records. Patients were evaluated at two time points: at PICU admission (T1) and at the time of the highest ferritin level (T2) during their PICU stay.

The evidence supports that the peak and post-treatment ferritin levels, as well as their dynamic changes, are strongly associated with prognosis in hemophagocytic lymphohistiocytosis, making ferritin a valuable independent prognostic marker [15–24].

The risk of mortality was assessed using the Pediatric Index of Mortality (PIM2) score at PICU admission (T1). Organ dysfunction was evaluated using the hepato-biliary dysfunction and diffuse intravascular coagulation (DIC) scores [1, 25].

The primary objective was to describe the clinical and laboratory parameters related to HLH, as well as prognostic outcomes, comparing survivors and non-survivors after PICU admission. Mortality was defined as death occurring during the PICU stay.

We considered the following risk factors for mortality in our population: invasive mechanical ventilation (MV), extracorporeal organ support (including extracorporeal membrane oxygenation, ECMO), and extracorporeal blood purification techniques (such as Continuous Renal Replacement Therapy, CRRT; hemoadsorption; and plasma exchange). Additionally, we evaluated complications as potential risk factors for mortality, including disseminated intravascular coagulation, hepatobiliary dysfunction, and central nervous system involvement during PICU stay.

Descriptive statistics were employed in this study. The distribution of the variables under investigation was assessed using the Shapiro-Wilk test, along with measures of skewness and kurtosis. Since the primary variables did not follow a Gaussian distribution, continuous data were summarized using medians and interquartile ranges (first and third quartiles) and compared with the Mann-Whitney U test. Categorical variables were expressed as absolute and percentage frequencies, with group comparisons performed using the Chi-square test or Fisher’s exact test, as appropriate. A p-value of less than 0.05 (two-tailed) was considered statistically significant. Data were stored in an Excel database, and all statistical analyses were conducted using Stata software version 15.1 (StataCorp, College Station, TX).

Results

48 patients were included in this study. Baseline clinical characteristics are summarized in Table 1. The sample comprised 22 females (45.8%) and 26 males (54.2%).

Table 1.

Characteristics of the 48 patients stratified by diagnosis (pHLH, MAS in sJIA and sHLH by other triggers)

Characteristics	pHLH	sHLH		p value
		MAS in sJIA	sHLH from other triggers	pHLH vs. sHLH	MAS vs. sHLH
	(N = 5)	(N = 6)	(N = 37)
Gender, Females 1	2 (40)	4 (66,7)	16 (43,2)	1,0	0,39
Age at onset (years) 2	1,0 (0,6 − 1,8)	8,5 (1,8–12,3)	5,6 (1,2–11,6)	0,068	0,57
Age at PICU admission (years) 2	1,8 (1,2–3,7)	8,5 (1,8–13,3)	5,5 (2,6–11,5)	0,095	0,56
Trigger 1				-	-
Onco-Hematological	-	0	2 (5,4)	-	-
Infectious	-	0	21 (56,8)	-	-
Rheumatological	-	6 (100)	8 (21,6)	-	-
Undefined	-	0	6 (16,2)	-	-
Perforin expression reduction NK activity reduction MAS criteria1	1 (20) 3 (60) -	0 1 (16,7) 6 (100)	2 (5,4) 4 (10,8) -	- - -	- - -
Criteria HLH-20041	5 (100)	-	30 (81,1)	-	-

¹Number (%) –Fisher’s exact test; ²Median (1st-3rd quartile) - Mann-Whitney test

The functional tests (perforin expression, cytotoxic activity) were performed in all patients with HLH and the analysis of familial HLH genes was done in those with defects of functional tests or in those highly suggestive of familial HLH.

All patients diagnosed with pHLH underwent genetic testing that revealed mutations; specifically: IFN-gamma-mediated autoinflammatory disease with the STXBP2 mutation (Familial HLH type 5), Griscelli syndrome, and, in two patients, homozygous mutation in the PRF1 gene.

Among the patients with sHLH, 16 did not undergo genetic testing, 22 tested negative for pHLH or had heterozygous variants for other unrelated conditions, 1 was diagnosed with Kabuki syndrome, 2 had mutations in the NLRC4 gene, and 1 had a mutation in the RAB3-GEP/MADD gene.

Patients with secondary hemophagocytic lymphohistiocytosis (sHLH) were stratified according to their predisposing conditions. In two patients (4.7%), the predisposing condition was a hematological disease—specifically, primary combined immunodeficiency and Hodgkin’s lymphoma.

In 21 patients (48.8%), the predisposing factor was an infection, including one case of Leishmania, viral agents in 12 patients (Rhinovirus, H1N1, EBV, HHV6, VZV, Bocavirus), and bacterial etiology in 8 patients.

The Authors would like to emphasize that rhinovirus is considered an uncommon trigger of HLH. In this single patient, rhinovirus was the only infectious trigger identified. However, it is possible that this infection was incidental rather than causative in the pathogenesis of HLH; in that case, the patient should instead be classified among the HLH cases with an unknown trigger. As no evidence was available to conclusively support either hypothesis, the patient was included in the infectious-trigger group to ensure consistency in data collection and classification.

Fourteen patients (32.6%) had a rheumatological disease as the predisposing factor, with specific diagnoses including six cases of systemic juvenile idiopathic arthritis (sJIA), two with NLRC4-gain of function, one with systemic lupus erythematosus, one with Kawasaki syndrome, one with an interferonopathy (spondyloenchondrodysplasia, SPENCD), one with Rasmussen syndrome, one with antiphospholipid antibody syndrome, and one with ANCA-positive vasculitis. In 56.8% of cases, HLH was triggered by an infectious event (e.g., EBV, CMV, H1N1, sepsis).

The different forms of HLH showed varying ages at disease onset: primary HLH (pHLH) at approximately 1 year, macrophage activation syndrome (MAS) in sJIA at around 8.5 years, and sHLH at about 5.6 years.

All patients with pHLH presented with fever, splenomegaly, and hemophagocytosis observed in bone marrow aspirates. Conversely, in sHLH, fever was the most common clinical symptom (94.6%), followed by splenomegaly in 56.8% of cases and in 33.3% of MAS cases. Hemophagocytosis in bone marrow was observed in 76% of sHLH cases and 60% of MAS cases (see Table 2).

Table 2.

Description of symptoms and organ damage of the 48 patients stratified by diagnosis (pHLH, MAS in sJIA and sHLH by other triggers)

Symptoms and organ damage	pHLH	sHLH		p value
		MAS in sJIA	sHLH from other triggers	pHLH vs. sHLH	MAS vs. sHLH
	(N = 5)	(N = 6)	(N = 37)
Fever	5 (100)	5 (83,3)	35 (94,6)	1,0	0,37
Splenomegaly	5 (100)	2 (33,3)	21 (56,8)	0,14	0,39
Hemophagocytosis (Bone marrow aspirate or biopsy)	4/4 (100)	3/5 (60,0)	19/25 (76,0)	0,55	0,64
Central Nervous System involvement	2 (40,0)	2 (33,3)	24 (64,9)	0,35	0,19
Multi-organ failure	2 (40,0)	1 (16,7)	14 (37,8)	1,0	0,40
Acute kidney injury	2 (40,0)	1 (16,7)	17 (46,0)	1,0	0,38
HBD at PICU admission	0 (0,0)	3 (50,0)	19 (51,4)	0,053	1,0
HBD at T2	2 (40,0)	1 (16,7)	26 (70,3)	0,31	0,021

Legend: HBD= hepato-biliary dysfunction. Number (%) –Fisher’s exact test

The median time from hospital admission to PICU admission was 3 days (interquartile range [IQR]: 0–19 days). Among patients with pHLH, the median was 0 days (IQR: 0–8 days); among patients with MAS, it was 2 days (IQR: 0–9 days); and among patients with sHLH, it was 3 days (IQR: 0–28 days).

In this patient population, clinical deterioration represented the primary reason for PICU admission, and the diagnosis of an HLH episode was established either immediately before or shortly after PICU admission, with a fairly homogeneous distribution across all patient groups.Relapse of disease, defined as a new HLH episode during PICU stay, occurred in 20% of pHLH patients, 66.7% of MAS patients with sJIA, and 18.9% of sHLH patients from other predisposing conditions. The difference between MAS and sHLH relapse rates was statistically significant (p = 0.029; see Table 3). Table 3 also details the Pediatric Index of Mortality 2 (PIM2) scores at T1, length of PICU stay, and mortality rates.

Table 3.

Outcomes of the 48 patients stratified by diagnosis (pHLH, MAS in sJIA and sHLH by other triggers)

Outcomes	pHLH	sHLH		p value
		MAS in sJIA	sHLH from other triggers	pHLH vs. sHLH	MAS vs. sHLH
	(N = 5)	(N = 6)	(N = 37)
PIM2 at PICU admission2	19,6 (6,9–59) n = 4	2,1 (1,1–13,9) n = 3	8,2 (4,6–31,7) n = 14	0,57	0,36
Relapse of disease 1	1 (20,0)	4 (66,7)	7 (18,9)	1,0	0,029
PICU stay, days2	24 (4–45)	19 (9–21)	19 (11–35)	0,88	0,61
PICU stay > 28 days1	2 (40)	1 (16,7)	16 (43,2)	1,0	0,38
Death in PICU1	3 (60)	0 (0)	13 (35,1)	0,275	0,097

¹Number (%) –Fisher’s exact test; ²Median (1st-3rd quartile) - Mann-Whitney test

Laboratory parameters at admission (T1) were compared across the 48 patients stratified by diagnosis (pHLH, MAS in sJIA, and sHLH from other triggers) (see Table 4). These included white blood cell count, neutrophils, platelets, ferritin, AST, ALT, LDH, triglycerides, fibrinogen, and D-dimers. Notably, white blood cell counts differed between survivors and non-survivors at PICU admission [5.34 × 10³/µL vs. 9.32 × 10³/µL; p = 0.076], as did neutrophil counts [0.9 × 10³/µL vs. 7.12 × 10³/µL; p = 0.012] (see Table 5).

Table 4.

Comparison of laboratory test values and medications of 48 patients stratified by diagnosis (pHLH, MAS in sJIA and sHLH by other triggers)

Table 5.

Analysis of laboratory values and medications of 48 patients stratified by survival in PICU

Platelet count, on the other hand, was similar at PICU admission (T1) but showed a significant difference at T2 [32,000 /µL vs. 72,000 /µL (p = 0.022)]. All patients received some form of mechanical ventilation; in 80% of cases, support was provided via invasive mechanical ventilation. Long-term mechanical ventilation (greater than 14 days) was necessary in 50% of pHLH patients, 60% of MAS patients, and 42.4% of sHLH patients.

Regarding extracorporeal organ support, ECMO was used only in the sHLH group: one patient with MAS (support duration of 108 days) and nine cases among sHLH patients from other predisposing conditions (median duration of 7 days). Continuous renal replacement therapy was employed in 60% of pHLH cases, 50% of MAS cases, and 59.5% of sHLH cases, with median durations of 22, 21, and 16 days, respectively. Plasma exchange was used exclusively in the secondary HLH group: in 2 out of 6 MAS cases (33.3%) and in 14 out of 37 other sHLH cases (37.8%). Hemoadsorption, in combination with CRRT, was performed in one pHLH case and four sHLH cases, with a median duration of 72 h.

Long-term mechanical ventilation was observed in 37.5% of non-survivors compared to 40.6% of survivors. Additionally, ECMO support was provided to 25% of non-survivors versus 18.8% of survivors; the median duration of ECMO support was 7 days among non-survivors and 7.5 days among survivors.

Table 6 reports the median duration of extracorporeal supports, distinguishing between survivors and non-survivors. A statistically significant difference was found in the duration of CRRT (p = 0.021) and the number of Hemoadsorption sessions (p = 0.037) between these groups.

Table 6.

Comparison of extracorporeal supports and outcomes of 48 patients stratified by survival

Extra-corporeal supports and outcomes	Non-survivors (N = 16)	Survivors (N = 32)	p value
Mechanical ventilation 1	16 (100)	26 (81,3)	0,159F
Duration (days) 2	13 (8–23)	13 (6–29)	0,969
ECMO1	4 (25)	6 (18,8)	0,712F
Duration (days) 2	7,5 (2–16)	7 (6–54)	0,476
CRRT1	13 (81,3%)	15 (46,9)	0,021C
Duration (days) 2	20 (10–30)	16 (10–31)	0,821
Plasma exchange1	5 (31,3)	11 (34,4)	1,0C
Hemoadsorption 1	4 (25)	1 (6,3)	0,037F
PIM2 upon entering PICU2	25 (10,1–64,8)	4,6 (2,1–13,6)	0,010
Relapse of disease 1	1 (6,3)	11 (34,4)	0,040C
Days in PICU2	27 (16–37)	19 (9–36)	0,308
PICU stay > 28 days 1	8 (50)	11 (34,4)	0,307C

¹Number (%) - Chi-square test C/ Fisher exact test F; ²Median (1st-3rd quartile) - Mann-Whitney test

Mortality was higher among primary HLH (60%) and onco-hematological cases (100%). Secondary HLH cases showed a mortality rate of 35.1%, while MAS cases had no observed mortality (0%) in this study sample. MAS also exhibited a higher relapse rate (66.7%) compared to pHLH (20%) and sHLH (18.9%).

Among pharmacological treatments (see Tables 4 and 5), corticosteroids were administered in 100% of patients with pHLH, 100% of patients with MAS, and 95% of patients with sHLH. Intravenous immunoglobulins (IVIG) were used in 33% of MAS cases and in 27% of sHLH cases. Cyclosporine A was administered to a total of 24 patients, corresponding to 40% of pHLH cases, 50% of MAS cases, and 51.4% of sHLH cases. Etoposide was used in 8 patients, all with secondary forms of HLH, corresponding to 50% of MAS cases and 14% of sHLH cases. Among biologic agents, anakinra was used exclusively in secondary forms, specifically in 67% of MAS cases and 8.1% of sHLH cases. Conversely, emapalumab was administered in 100% of patients with pHLH, 33% of patients with MAS, and 14% of patients with sHLH. Cyclophosphamide was the only treatment that showed a statistically significant difference in use between non-survivors and survivors (6% vs. 28%; p = 0.038).

Discussion

Strengths of the manuscript include the focus on a life-threatening hyperinflammatory condition in a high-acuity setting and the inclusion of primary and secondary HLH across a range of underlying aetiologies, reflecting the full spectrum of hyperinflammation. Furthermore, this is the largest documented case series of paediatric cases of HLH, specifically in the ICU setting.

Limitations of this study include the retrospective design and the single-centre setting. HLH is inherently a rare condition and does not always require intensive care; therefore, results are limited to small subgroups of patients. Furthermore, this study was designed for a PICU-only population. All these considerations may constitute potential confounding factors and limit the generalizability of the results; however, at the same time, they delineate the operational framework within which the study was intended to be conducted.

The long inclusion period (2006–2019) encompasses substantial changes in HLH/MAS management. Although treatment variability and temporal changes are beyond the scope of the present manuscript, we nevertheless provided a descriptive overview of the treatments administered across the different patient groups under investigation. These results were included in Tables 4 and 5 together with hematologic parameters, allowing readers to easily compare the findings. In our cohort, the distribution between males and females (22 vs. 26, respectively) was similar. We found that primary HLH forms tend to have an earlier onset in childhood, and among the secondary forms, infectious predisposition was the most common (48.8%).

The shorter time from hospital admission to PICU admission observed in patients with HLH is likely attributable both to the more rapidly progressive acute clinical presentation and to increased clinical awareness among physicians; this consideration may also apply to patients with MAS. In contrast, in patients with sHLH, the diagnosis is less frequently anticipated, and the clinical picture may require more time to become fully manifest.

A previous study [26] reported an overall survival rate of 45%, with increased mortality associated with HLH triggered by infection or malignancy. However, differences in survival did not correlate with genetic profile or the extent of therapy. The same study [26] discusses how the uncertain performance of the HLH-2004 criteria and the consequences of missed or incorrect diagnoses have led to widely variable approaches to diagnosis and treatment [4, 27]. These findings are consistent with our sample, where overall mortality in the PICU was high (33.3%), particularly in the primary HLH group (60%) and the onco-hematological group (100%), but no mortality was observed among the MAS group (0%).

In our cohort, patients with the worst prognosis exhibited features such as splenomegaly at onset, lower platelet counts, and higher ferritin levels at PICU admission. Extracorporeal support therapies were used uniformly across all patients, but for longer durations in the most severe cases, which correlated with higher mortality.

The higher mortality observed in primary HLH and sHLH related to onco-hematological disease may be linked to underlying genetic conditions that predispose these patients to early childhood risks, as well as to compromised general health and comorbidities in the latter group. Conversely, the more frequent relapses seen in the MAS group, with no associated mortality, likely reflect the classic pattern of recurrent exacerbations of rheumatological disorders. Early diagnosis and targeted therapies in patients with known diagnoses may also contribute to this trend.

In our population, pHLH always presented with a full symptom complex. Fever was the most common clinical symptom, even in secondary forms. Splenomegaly and bone marrow hemophagocytosis were frequent but not universal. According to the literature [28], splenomegaly occurs in about 79% of HLH cases (survivors 84.85%, non-survivors 66.67%), likely due to immune activation, especially of lymphocytes and macrophages. In our study, splenomegaly at onset was more common in patients who died compared to those who survived (overall 58.3%, survivors 53.1%, non-survivors 68.8%). This suggests that more severe immune activation may lead to a more severe clinical presentation, with splenomegaly potentially serving as a negative prognostic factor.

The incidence of multi-organ failure (MOF), central nervous system impairments, and renal dysfunction was higher among non-survivors, though these differences did not reach statistical significance.

The length of PICU stay was shorter in the MAS group and longer in the other groups. The hepatobiliary damage (HBD) score changed from T1 to T2, worsening in the pHLH group (which showed no alterations at T1), improving among MAS patients, and worsening in other sHLH cases—indicating a greater involvement of the hepatobiliary system in secondary non-rheumatologic forms of HLH. The Pediatric Index of Mortality 2 (PIM2) at T1 was higher in pHLH (19.6%), lower in sHLH (8.2%), and much lower in MAS (2.1%). PIM2 was also significantly higher among non-survivors (25% vs. 4.6%; p = 0.01; see Table 6). This finding suggests that PIM2 could serve as a reliable prognostic tool.

Ferritin levels at T1 were similar across pHLH, sHLH, and MAS; however, at T2, ferritin levels were higher in pHLH, lower in MAS, and much lower in other sHLH cases. Ferritin levels were approximately three times higher among non-survivors compared to survivors at T1 (p = 0.069) and about twice as high at T2 (p = 0.01), confirming that ferritin functions as a prognostic biomarker in HLH [5, 26, 27].

A retrospective study by Gregory et al. 2019 [15] examined similar outcomes in 33 children admitted to a PICU with HLH, and identified high total bilirubin, lactic acidosis at presentation, and hyperferritinemia as factors associated with worse outcomes.

Regarding white blood cell counts, we observed differences among MAS, sHLH, and pHLH, but values remained within the normal range for both survivors and non-survivors. Therefore, white blood cell count does not appear to be useful for predicting severity.

In contrast, platelet counts were lower among non-survivors at both T1 and T2, indicating that this parameter could help identify patients at higher risk of death. This trend of decreasing platelets and white blood cells may reflect a better disease status at T1 in some patients, while the uniformity of worse values at T2 across groups likely indicates clinical worsening.

LDH levels were elevated in all groups. In pHLH patients, LDH increased from three times the upper normal limit at T1 to twenty times at T2. In MAS patients, LDH remained consistently high (about 17 times the normal), while in sHLH patients, it increased from 5 to 8 times from T1 to T2. Typically, increased LDH is due to physical stress, cellular damage, or increased inflammatory activity; this could explain the increase in these concentrations at T2. However, this case study is too small to draw any conclusions, and this observation should be explored further with larger case studies.

In the comparison between pHLH and sHLH triggered by other causes (see Table 4), emapalumab was the only treatment showing a statistically significant difference in use between the two groups (100% vs. 33%; p < 0.001). Cyclophosphamide, etoposide, intravenous immunoglobulins (IVIG), and mycophenolate mofetil were not used in the pHLH group, likely because alternative therapies considered more appropriate for this patient category were selected.

In the comparison between MAS and sHLH triggered by other causes (see Table 4), only anakinra (67% vs. 8%; p = 0.038) and mycophenolate mofetil (0% vs. 11%; p = 0.044) showed statistically significant differences in use.

Cyclophosphamide was the only treatment associated with a statistically significant difference in administration between non-survivors and survivors (6% vs. 28%; p = 0.038; see Table 5).

Extracorporeal support techniques played a significant role in our patient management, especially considering that this study was conducted in the PICU setting, where all patients had severe HLH, rapidly evolving clinical conditions, or hemodynamic instability. Patients with more severe conditions required extracorporeal support more frequently and for longer durations. Often, a combination of techniques was used (e.g., invasive mechanical ventilation, CRRT, ECMO, plasma exchange, hemoadsorption).

All patients with pHLH initially received non-invasive ventilation, and most patients with pHLH, MAS, and sHLH also required invasive mechanical ventilation. Interestingly, long-term mechanical ventilation (more than 14 days) was associated with survival in our cohort. ECMO was used only in the sHLH group (a total of 10 patients, with only one patient having MAS), with a median duration of 7 days. ECMO was applied at similar rates among survivors (18.8%) and non-survivors (25%), with the same median duration (7 vs. 7.5 days). However, the mean duration was about three times longer in non-survivors (10.5 vs. 33.3 days).

CRRT was more frequently required in non-survivors (81.3%) compared to survivors (46.9%), and for longer periods (20 vs. 16 days). Plasma exchange was performed only in MAS and sHLH groups, at similar rates among survivors (34.4%) and non-survivors (31.3%). Hemoadsorption with CytoSorb was used in one case of pHLH and in four cases of sHLH, mainly among non-survivors (6.3% vs. 25%). A retrospective study by Alsohime et al. 2023 [29], regarding patient with familial HLH admitted in PICU, examined similar outcomes and identified high bilirubin level, low serum triglyceride level, bleeding, higher levels of hemodynamic and respiratory support, and positive fungal cultures. The high incidence of extracorporeal therapies and mechanical ventilation in our cohort of critically ill children admitted to the PICU reflects the complexity of these patients and the high prevalence of organ dysfunction associated with HLH as a hyperinflammatory syndrome.

Conclusions

Hemophagocytic lymphohistiocytosis (HLH) is associated with high mortality despite intensive care and maximal therapeutic support. Mortality is particularly elevated in patients with primary HLH and those with onco-hematological forms. While MAS patients were prone to relapse, no deaths occurred in this subgroup during the study period. Early presence of splenomegaly, low platelet count, and elevated ferritin levels at PICU admission are potential predictors of adverse outcomes. The PIM2 score may serve as a useful prognostic tool in HLH. Extracorporeal support therapies—including ECMO, invasive mechanical ventilation, continuous renal replacement therapy, plasma exchange, and CytoSorb—were applied consistently across patients, though for longer durations in the most severe cases, which correlated with higher mortality. Further studies may help clarify the impact of treatment evolution over time. To our knowledge, this is one of the largest pediatric cohorts of patients with both primary and secondary HLH treated in a PICU setting.

Abbreviations

CMV

Cytomegalovirus

CRRT

Continuous Renal Replacement Therapy

DIC

Diffuse intravascular coagulation

EBV

Epstein-Barr virus

ECMO

Extracorporeal membrane oxygenation

HBD

Hepatobiliary damage

HLH

Hemophagocytic lymphohistiocytosis

MAS

Macrophage activation syndrome

MOF

Multiple organ failure

MV

Mechanical ventilation

pHLH

primary hemophagocytic lymphohistiocytosis

PICU

Pediatric intensive care unit

PIM2

Pediatric Index of Mortality

sHLH

secondary hemophagocytic lymphohistiocytosis

sJIA

systemic juvenile idiopathic arthritisSPENCD: Spondyloenchondrodysplasia

Author contributions

MM, GB, CB and MDN conceptualized and designed the study, coordinate and supervised collected data and drafted the initial manuscript; FC, FDL, PM conceptualized and designed the study, coordinate and supervised collected data, and drafted the initial manuscript; DPM, collaborate to the design of the study and performed all statistical analyses, contributed to manuscript drafting, critically reviewed and revised the manuscript; MM, GB, CB, MDN, FC, FDL and PM collaborate to the design of the study, contributed to manuscript drafting, critically reviewed and revised the manuscript; CC and FDB designed the study, supervised data collection, and reviewed and revised the manuscript for important intellectual content; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Funding

All phases of this study were supported by Bambino Gesù Children’s Hospital, IRCCS, Rome, Italy. This work was supported by the Italian Ministry of Health with Current Research funds.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study received approval from the Institutional Review Board of Children Hospital Bambino Gesù. Informed consent was waived due to the retrospective nature of the study. This study was performed in line with the principles of the Declaration of Helsinki. This is an observational study. Approval was granted by the Ethics Committee of Bambino Gesù Children’s Hospital, IRCCS, Rome.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.