Calm in the midst of cytokine storm: a collaborative approach to the diagnosis and treatment of hemophagocytic lymphohistiocytosis and macrophage activation syndrome

To develop a collaborative approach to HLH/MAS management, we enlisted the support of the Evidence-Based Guideline (EBG) program, which is administered through the Department of Pediatrics (DOP) Quality Program at Boston Children’s Hospital (BCH) (Fig. 1). EBGs are clinical algorithms constructed from available evidence in the literature and the expert opinion of locally involved clinicians. The goal is to achieve consensus on how to manage a given condition and thereby reduce practice variability and improve patient care. Data support through the DOP Quality Program allows the outcomes of an EBG to be tracked; therefore, additional iterations of the EBG can be developed for further improvement. The EBG approach has been used to create guidelines for more than 60 clinical topics/conditions at BCH [33].

We identified the key subspecialties involved in the management of HLH/MAS and sought a representative from each discipline (Additional file 1: Table S1). Local experts in HLH/MAS were asked to participate by personal invitation. In addition, division chiefs of involved specialties were asked to recommend potential members. Emails were sent to clinicians and trainees in rheumatology, hematology, and oncology specialties particularly involved in HLH/MAS management, to encourage participation.

HLH/MAS EBG meetings were convened monthly from 3/2016 to 12/2017 (Additional file 2: Table S2). The leads of the workgroup (M.M.H. & L.A.H) served as moderators and identified key questions to address at each meeting. Each session had a pre-set agenda with identified speakers who typically had literature review assignments. At the meetings, the speakers presented a summary of the reviewed literature, which was followed by discussion to build consensus. Based on the nominal group technique, greater than 80% consensus agreement on each point was required in order to pass. At the end of each meeting, areas of agreement and difference were summarized, and action items were identified for the next meeting.

Prior to the EBG, inpatient consults for HLH/MAS were often channeled through different specialties, most often hematology, oncology, or rheumatology, in a somewhat arbitrary manner. The result was often discrepant diagnostic and treatment recommendations for patients with similar clinical presentations. If multiple specialties were involved in the care of the same patient, clinical recommendations often conflicted, resulting in delayed diagnosis and initiation of treatment. The workgroup sought to eliminate this variability and develop a standard pathway for diagnosis and treatment. The overarching goals of the HLH/MAS EBG workgroup included: 1) Increase communication and collaboration across specialties involved in the care of these patients; 2) Develop an HLH/MAS clinical response team; and 3) Develop a diagnostic and treatment algorithm for HLH/MAS.

In order to implement the EBG, an HLH/MAS clinical response team was created. As a model, the EBG workgroup selected the Kawasaki disease (KD) program, a successful multidisciplinary, clinical team composed of cardiologists and rheumatologists at BCH. As with HLH/MAS, rapid diagnosis and treatment of KD are essential to prevent morbidity and mortality. The EBG workgroup identified key aspects of the KD program that promote success. 1) One subspecialty (rheumatology) is identified as the “gate-keeper” and directs clinical care. 2) Communication is facilitated by an email distribution list. 3) Consensus is achieved before recommendations are given to the house staff. 4) An order set within the electronic medical record (EMR) facilitates diagnostic testing and medication administration.

At BCH, rheumatology has been increasingly engaged in the management of HLH/MAS. Accordingly, the rheumatology consult service was selected to be the “gate-keeper” of the HLH/MAS EBG. Rheumatology serves as the initial point of contact for the house staff to engage when there is concern for HLH/MAS. Rheumatology determines whether the patient should enter the EBG for further diagnostic testing and treatment. An HLH/MAS email distribution list facilitates communication and includes all rheumatology trainees and attendings, as well as representatives from subspecialties involved in the care of HLH/MAS patients (Additional file 1: Table S1). An EMR order set was designed so that diagnostic studies and medications suggested by the HLH/MAS clinical team could be ordered easily by the house staff. Direct links to requisition forms for send-out laboratory studies recommended in the EBG were also included.

Entry criteria were identified to help house staff recognize patients with potential HLH/MAS who may be eligible for the EBG. The workgroup reviewed the HLH 2004 diagnostic criteria [5], proposed MAS classification criteria in patients with sJIA [14, 34], and publications on the sensitivity and specificity of elevated ferritin in the diagnosis of HLH/MAS [34‐39]. The goal was to capture all inpatients who should undergo a diagnostic evaluation for possible HLH/MAS; hence, entry criteria were designed to be broad (Table 1). Ultimately, fever and ferritin ≥500 ng/mL were selected based on the HLH 2004 diagnostic criteria [5].

At BCH, ferritin is typically obtained as part of the fever of unknown origin evaluation and is often readily available. The workgroup leveraged i2b2, a centralized repository of de-identified clinical data from BCH, to review the number of inpatients within the preceding year with a ferritin ≥500 ng/mL. Twenty-seven patients were identified, a number that was agreed to be reasonably handled by the HLH/MAS EBG.

In addition to fever and ferritin levels, other clinical findings were highlighted to help house staff consider a diagnosis of HLH/MAS: a history of a rheumatologic/hematologic/immunologic disease that predisposes to HLH/MAS, Epstein-Barr virus (EBV) infection, neurologic symptoms, hepatosplenomegaly, coagulopathy, and transaminitis.

Once a patient with potential HLH/MAS is identified, the rheumatology team is consulted and determines whether the patient should enter the EBG and undergo a diagnostic evaluation (Fig. 2, Table 2). While the EBG provides recommendations, the diagnostic assessment is at the discretion of the rheumatology consult team.

Based on the HLH diagnostic criteria [5] and the ACR/PRINTO 2016 MAS classification criteria [14], laboratory evaluation includes assessment for cytopenias, transaminitis, coagulopathy, and elevated triglycerides (Table 2). Functional and genetic studies commonly used in the diagnosis of FLH are not typically ordered in patients with MAS. Review of the literature by the EBG workgroup demonstrated that mutations in FHL genes and decreased NK cell function are increasingly documented in patients with presumed secondary HLH due to infection and MAS associated with sJIA [21‐25]. sJIA patients with MAS are more likely to have variants in FHL-associated genes than sJIA patients without MAS, indicating a potentially more severe disease course that may require more aggressive and sustained therapies [23, 40]. Accordingly, the EBG workgroup agreed that assessment of cytolytic activity with NK cell function and/or CD107a mobilization as well as genetic testing should be considered broadly (Table 2). While genetic testing can take over a month to complete, flow cytometric evaluation of perforin, granzyme, SLAM-associated protein (SAP), and XIAP expression is often available within days and should be considered in patients with high concern for FHL. While bone marrow biopsies are commonly performed in patients with presumed FHL, they are less frequently obtained in children with secondary HLH due to infection or MAS associated with rheumatic disease; therefore, this procedure was not categorically recommended in the EBG.

Cytokine storm is a central feature of HLH/MAS with a particularly important role for IFNγ and IL-18. The EBG workgroup reviewed literature on the importance of these cytokines as supported by data in mouse models and humans [2‐4, 12, 26‐29, 31, 41‐44]. While IFNγ is difficult to measure in the peripheral blood, CXCL9, an IFNγ-inducible cytokine, is easily measured and reflects IFNγ levels. CXCL9 levels are high in HLH and can differentiate patients with active sJIA and MAS [3, 13, 31, 32]. IL-18 is increased in HLH but more so in MAS, where chronic IL-18 elevations are thought to be a risk factor for MAS [12, 28, 29]. Like CXCL9, there is evidence that IL-18 can be used to discriminate between active sJIA and sJIA with MAS [12, 29]. Both CXCL9 and IL-18 are available for clinical testing and are likely biomarkers of HLH/MAS; therefore, these tests were included in the HLH/MAS EBG.

Dedicated EBG workgroup meetings were held with representatives from neuroimmunology (L.B., M.G.) and infectious diseases (G.P., N.S.) to select criteria for involving these subspecialties (Fig. 2). Rarely, children present with HLH/MAS as the first manifestation of underlying malignancy [45]. The EBG workgroup reviewed characteristics of malignancy-associated HLH/MAS in children [45]. In conjunction with input from representatives of oncology (B.A.D.) and hematology (K.W.), criteria for considering additional oncologic evaluation were identified (Fig. 2), reflecting the most common malignancies known to present as HLH/MAS, T and B cell lymphomas.

Development of a therapeutic pathway for HLH/MAS patients was a major goal for the EBG workgroup. The intent of the treatment algorithm is to provide an agreed upon framework for therapeutic decision-making to reduce practice variability and improve clinical outcomes. Decisions surrounding the start of therapy in HLH/MAS patients are often the most difficult and can be associated with disagreements amongst subspecialists that delay much needed intervention. It is for this reason that the HLH/MAS workgroup focused its recommendations on treatment initiation for the hospitalized patient. The guidelines are meant as suggestions, and clinicians retain their autonomy to deviate from the EBG.

The EBG workgroup recommended that FHL patients identified during the diagnostic workup should in most cases receive HLH directed therapy under oncology. In addition, patients with HLH/MAS in the setting of a known malignancy or secondary to chemotherapy were excluded from the treatment algorithm because of insufficient guidance in the pediatric literature and the limited knowledge about safety of biologic medications in this population.

Based on the opinion of the participating clinicians, initiation of treatment should be considered when there is progressive clinical deterioration, worsening laboratory parameters of HLH/MAS, and/or persistent fevers. The workgroup members agreed that treatment should be started as soon as HLH/MAS pathophysiology is recognized and often before the patient meets the full criteria for HLH/MAS. Thus, absolute clinical and laboratory parameters for treatment were not defined. Instead, the trend in the clinical profile is indicative of the need for treatment (Fig. 2), which is in keeping with previously published expert consensus on the dynamics of laboratory values in MAS [46].

Since the EBG recommends treatment at the first sign of HLH/MAS, many patients will not meet diagnostic criteria for HLH or classification criteria for MAS. The use of chemotherapy with agents such as etoposide is not recommended until patients meet diagnostic specifications for HLH. Alternately, medications typically used by rheumatology for MAS are a viable option early in the disease course because they are associated with less toxicity. The EBG workgroup reviewed the literature that demonstrated efficacy for medications such as cyclosporine and anakinra in some patients who meet criteria for not only MAS but also secondary HLH [18, 47‐49]. Since these medications have a safer side effect profile than chemotherapy, they were selected as first-line agents (Table 3). An inadequate response to the suggested therapies should immediately trigger re-involvement of oncology to consider more aggressive therapeutic protocols (Fig. 2).

The workgroup agreed to present a list of recommended medications that could be used alone or in combination for HLH/MAS, depending on the preference of the treating physician. Medication selection and dosing are stratified by the severity of illness and risk of infection, particularly bacterial or fungal infections. The actual assessment of disease severity and infectious risk is made by the treating physician. In moderately ill patients with concern for an underlying infection, less immunosuppressive agents such as IVIG and anakinra are generally preferred. In critically ill patients with uncontrolled cytokine storm (typically patients requiring ICU level care), greater immunosuppression is often warranted despite the increased risk. Drug dose, route, and frequency were based on the pharmacy formulary at BCH and the experience of the involved clinicians.

The EBG workgroup reviewed a number of medications for inclusion in the treatment algorithm. Investigational drugs including NI-0501/anti-IFNγ monoclonal antibody and recombinant human IL-18 binding protein were discussed but ultimately were not included in the EBG because they were not FDA approved at the time of the workgroup meetings and thus were difficult to obtain outside of a clinical trial [31, 44]. Literature on the use of tocilizumab for HLH/MAS was reviewed by the EBG workgroup, including the impact of IL-6 on NK cell function [24, 25], rates of MAS in sJIA patients treated with tocilizumab [9, 50, 51], use of tocilizumab to treat MAS in sJIA and Adult-onset Still’s disease (AOSD) [52, 53], and efficacy of tocilizumab in cytokine release syndrome secondary to T cell engaging oncologic therapies [54, 55]. The reviewed phase III clinical studies demonstrated that MAS can occur in patients with sJIA while on tocilizumab, even with adequate sJIA disease control [9]. In addition, tocilizumab alters markers of MAS/HLH such as C-reactive protein (CRP) and ferritin, making diagnosis difficult [9, 51, 56]. Articles that supported the use of tocilizumab in MAS mostly included patients with sJIA, AOSD, or oncology patients receiving therapies such as blinatumomab and chimeric antigen receptor (CAR) T cells. There was little evidence in the literature for the use of tocilizumab in secondary HLH due to infection. Accordingly, tocilizumab was not recommended as a first-line agent in our treatment algorithm.

Before the era of biologic agents, IVIG, cyclosporine, and glucocorticoids were used regularly to treat rheumatologic-associated MAS [15‐17]. Presently, these medications are still used alone or in conjunction with other biologic agents for MAS and have been shown to be effective in HLH [5, 47, 49, 57]. Accordingly, IVIG, cyclosporine, and glucocorticoids were included in the treatment algorithm. Tacrolimus targets the same pathway as cyclosporine and is occasionally prescribed by EBG workgroup clinicians for MAS, particularly in patients who may have difficulty tolerating cyclosporine. As such, tacrolimus was included in the treatment algorithm.

An entire HLH/MAS workgroup meeting was dedicated to discussing anakinra in the treatment of HLH/MAS. Several manuscripts were reviewed by the workgroup that described the successful use of anakinra to treat MAS in sJIA patients, particularly with doses greater than 1-2 mg/kg [9, 18, 19]. In some of these reports, sJIA patients who met the full criteria for secondary HLH still responded to anakinra [18]. In addition, some data supported the efficacy for anakinra in secondary HLH due to infection [48, 58]. Notably, a re-analysis of the original phase III randomized, placebo-controlled study of anakinra for severe sepsis showed that patients with features of MAS (hepatobiliary dysfunction and severe coagulopathy) had improved survival [58]. The HLH/MAS workgroup leaders (L.A.H, M.M.H.) analyzed additional literature in support of anakinra therapy in secondary HLH and MAS [20, 47, 49, 57, 59]. It was unanimously agreed to include anakinra in the treatment algorithm. In the phase III sepsis study, high dose anakinra was used safely; therefore, high dose anakinra was recommended in the treatment algorithm for critically ill patients, even those with suspected infections [58]. There were little data in the literature to support the use of other IL-1-inhibiting drugs in HLH/MAS [9].

The EBG was presented to major stakeholders at departmental meetings: rheumatology, immunology, hematology, department of medicine hospitalists, and pediatric interns/residents (Additional file 1: Table S1). The presentations were structured to encourage feedback, which was then incorporated into the EBG. In addition, pharmacy reviewed the suggested medications and dosages to ensure harmonization at the institution level. The created HLH/MAS email distribution list and EMR order set also facilitated implementation of the EBG. In order to increase awareness of the new HLH/MAS EBG, laminated copies were distributed to house staff, email announcements were sent, and the EBG was posted on the DOP intranet site.

Since our aim in implementing this EBG was to improve the quality of care provided to this patient population, we defined several quality metrics in concert with the DOP QI team. First, we sought to define our target population and to obtain baseline data regarding these patients prior to EBG implementation. We thus developed a search algorithm for the EMR that allowed us to identify patients who met entry criteria of the EBG and would potentially benefit from its use. The following search criteria were used: ferritin ≥500 ng/mL, fever ≥38.2C, presence of a rheumatology or oncology inpatient clinical note, and clinical note including the term “HLH” or “MAS”. Using these criteria, we evaluated the outcomes of a historical control cohort by pooling the medical records of patients treated within the 2 years prior to EBG implementation. To validate our search strategy, the charts were manually reviewed by rheumatologists who confirmed the presence or absence of HLH/MAS. During the defined time period, a total of 75 patient charts were selected based on the above parameters. After review, 23 patients were confirmed to have the diagnosis of MAS/HLH based on the expert opinion of the reviewing rheumatologist. Prior to EBG implementation, HLH/MAS patients were cared for by different subspecialties: oncology (n = 1), rheumatology and oncology (n = 2), rheumatology and hematology (n = 8), and rheumatology alone (n = 10). Treatments for patients diagnosed with HLH/MAS varied and included systemic steroids (n = 13), anakinra (n = 11), HLH 2004 protocol (n = 3), IVIG (n = 4), cyclosporine (n = 3), tacrolimus (n = 2), infliximab (n = 1), and tofacitinib (n = 1). These findings underscore the wide range of therapies used for HLH/MAS at our institution historically. The goal of the EBG is reduce this treatment variation and improve outcomes. In order to monitor efficacy of the EBG in achieving these goals, we selected quality measures to be assessed after EBG implementation: length of stay, readmission, time to diagnosis, time to HLH/MAS directed treatment, duration of fever, time to decrease in laboratory parameters of disease activity (ferritin, CRP), need for higher level of care, and death (Table 4).