Diagnostic and management of life-threatening Adult-Onset Still Disease: a French nationwide multicenter study and systematic literature review

Background Adult-onset Still disease (AOSD) is a rare systemic inflammatory disorder. A few patients develop organ complications that can be life-threatening. Our objectives were to describe the disease course and phenotype of life-threatening AOSD, including response to therapy and long-term outcome. Methods A multicenter case series of intensive care medicine (ICU) patients with life-threatening AOSD and a systematic literature review. Results Twenty patients were included. ICU admission mostly occurred at disease onset (90%). Disease manifestations included fever (100%), sore throat (65%), skin rash (65%), and arthromyalgia (55%). Serum ferritin was markedly high (median: 29,110 ng/mL). Acute respiratory failure, shock and multiple organ failure occurred in 15 (75%), 10 (50%), and 7 (35%) cases, respectively. Hemophagocytosis was demonstrated in eight cases. Two patients died. Treatment delay was significant. All patients received corticosteroids. Response rate was 50%. As second-line, intravenous immunoglobulins were ineffective. Anakinra was highly effective. After ICU discharge, most patients required additional treatment. Literature analysis included 79 cases of AOSD with organ manifestations, which mainly included reactive hemophagocytic syndrome (42%), acute respiratory failure (34%), and cardiac complications (23%). Response rate to corticosteroids was 68%. Response rates to IVIgs, cyclosporin, and anakinra were 50%, 80%, and 100%, respectively. Conclusions AOSD should be recognized as a rare cause of sepsis mimic in patients with fever of unknown origin admitted to the ICU. The diagnosis relies on a few simple clinical clues. Early intensive treatment may be discussed. IVIgs should be abandoned. Long-term prognosis is favorable. Electronic supplementary material The online version of this article (10.1186/s13054-018-2012-2) contains supplementary material, which is available to authorized users.

A few patients develop organ complications that can be life-threatening, including reactive hemophagocytic syndrome (RHS), cardiac failure, respiratory distress, coagulopathy, severe hepatitis, and multiple organ failure (MOF) [16]. Data regarding these infrequent complications mainly arise from case reports or small series focusing on one manifestation such as RHS [17][18][19] or myocarditis [20]. Despite these patients posing important diagnostic and therapeutic dilemmas, little is known about life-threatening AOSD as a whole.
Our objectives were: (i) to describe the disease course and phenotype of life-threatening AOSD (i.e. cases with organ failure leading to ICU admission) (ii) to analyze the response to therapy and long-term outcome of these patients.
We report a multicenter case series of 20 intensive care unit (ICU) patients with life-threatening AOSD as well as a systematic literature review of organ manifestation of AOSD.

Multicenter case series Inclusion criteria
This multicenter retrospective study was performed under the auspice of the French Intensive Care Society and the French Internal Medicine Society. Inclusion criteria were: (i) admission to ICU due to AOSD-related organ failure. (ii) AOSD diagnosis fulfilling the Yamaguchi [21] and/or Fautrel [22] criteria. (ii) Exclusion of differential diagnosis, including infection, malignancy, and other systemic immune mediated disorders. (iv) Age at AOSD diagnosis > 18 years. (v) Organ failure requiring organsupporting therapeutic intervention including vasopressor agents, pericardial drainage, mechanical ventilation, renal replacement therapy (RRT) or plasmatherapy. Exclusion criteria were ICU admission without organ failure or for reasons other than AOSD.
The flow chart is in Additional file 1. This observational study was performed in accordance with the Helsinki declaration, and European and French ethics laws.

Data collection
Data were collected using a standardized form by one of the investigators (AW). Disease severity was measured using the SAPS II (Simplified Acute Physiology Score II). Organ failure definitions were adapted from SAPS II and Logistic Organ Dysfunction (LOD) system [23][24][25]. Definitions of respiratory and cardio-circulatory failures, acute kidney injury (AKI), neurologic dysfunction, and hematologic disorders are reported in Additional file 2. Response to therapy was defined as ICU discharge.

Systematic literature review
Using MEDLINE via PubMed (National Library of Medicine, Bethesda, MD, USA), we performed a computer search for English or French language publications from January 2000 to March 2014. We selected case reports from western or east Asian countries describing AOSD cases complicated by RHS or/and shock, respiratory distress, myocarditis, tamponade, hepatic failure, thrombotic microangiopathy (TMA), disseminated intravascular coagulation (DIC), MOF, AKI, and/or neurological involvement. Individual data were collected by one of the investigators (AW). Details of systematic literature review are available in Additional file 2.

Statistical analysis
Data are presented as median (min-max). Analyses were performed using SPSS software, version 21.0 (IBM Corp, Armonk, NY, USA). Continuous variables were compared using the nonparametric Mann-Whitney test. Statistical significance was defined as a two-tailed p value of < 0.05. Cumulative incidence and survival curve were created using the Kaplan-Meier method (GraphPad Prism version 5 (GraphPad Software, San Diego, CA, USA). Venn diagrams were drawn using the online tool available at http:// bioinformatics.psb.ugent.be/webtools/Venn/.

Disease course and diagnostic delay
Median time between fever onset and hospitalization was 10 days (1-61) (Fig. 2a). Median time between hospitalization and ICU admission was 4 days (0-39). Median time between fever onset and ICU admission was 14.5 days (5-61). Most patients (17/20, 85%) had received no specific treatment for AOSD upon ICU admission. Even then, treatment began more than 2 days after ICU admission in most cases (12/17, 60%) (Fig. 2b). One patient was treated after 44 days spent in the ICU. Suspected diagnoses included septic shock, pneumonia, angiocholitis, Lemierre syndrome, Rickettsiosis, Leptospirosis, drug reaction. As shown in Fig. 2c, many antibiotics were prescribed before AOSD was eventually considered.

Organ involvement
Organ involvement mainly affected the respiratory, cardiocirculatory and hematologic systems (Fig. 1a). The median SAPS II score was 33 .
Pericarditis occurred in ten patients (50%) with cardiac tamponade in three (drainage in two cases). Five patients (25%) had < 45% left ventricular ejection fraction. Their median was 40 . No patients had ventricular heart rhythm disorder. Eight patients had evidence of myocarditis (40%), which was occasionally proven by magnetic resonance imaging (MRI) or endomyocardial biopsy (one case each).

Hematologic disorders
Nineteen patients had moderate anemia. None had leucopenia. Seven patients had platelet count < 100 × 10 9 /L. Fourteen patients underwent bone marrow examination. Hemophagocytosis was seen in eight cases and correlated with thrombocytopenia and liver enzyme elevation. In contrast, hemoglobin, leukocyte, fibrinogen and ferritin levels did not predict hemophagocytosis (Fig. 3).
Treatment and outcomes Treatment efficacy in the ICU As first line, all patients received high-dose CS. Response to CS monotherapy, i.e. leaving the ICU without further treatment, was achieved in only ten cases (50%), within a median of 6 days (1-36). These patients had lower ferritin level (p = 0,017) and tended to exhibit a less severe clinical picture. An additional table shows this in more details (see Additional file 3).
Two patients (10%) died in the ICU. Both had MOF and hemophagocytosis. A 19-year-old man, with a previous history of AOSD died after ineffective CS then IVIgs treatment. Treatment had begun 17 days after first signs. A 37-year-old woman with new-onset AOSD died after 2 months in the ICU. This patient presented with hemophagocytosis uncontrolled by CS, anakinra, cyclosporine, and etoposide. She underwent diagnostic splenectomy and suffered repeated infections, notably invasive candidiasis. Autopsy disclosed hepatosplenic hemophagocytosis. Treatment had begun 32 days after fever onset.

Long-term outcome
After leaving the ICU, 11 patients with persistent disease activity required additional treatment before discharge. At 3 months, all patients were on CS and 16/18 (89%) received additional treatment, mostly anakinra (n = 10) and/or methotrexate (n = 5). Detailed long-term treatment data are reported in Table 3. At last follow-up, a single patient had received no other treatment than CS. The clinical course of AOSD was monocyclic, polycyclic and articular in ten (56%), six (33%), and two patients (11%), respectively. After a median follow-up of 3.5 years, no patient had suffered life-threatening relapse.

Discussion
Herein, we report on the first study focusing on AOSD patients requiring ICU admission due to AOSD-related organ failure. Our objectives were to describe the  Until now, most studies about AOSD have focused on the long-term risk of developing further systemic flare(s) and/or a chronic polyarthritis. Similarly, most data regarding therapeutics focuses on patients with refractory, chronic, and non-life-threatening disease [1-15, 21, 22]. The fact that AOSD patients may develop severe organ manifestations is mentioned for decades, but this subset of patients has been seldom studied [2]. The present work is the first attempt to study critically ill AOSD patients as a whole. The frequency of organ complications is difficult to estimate, but it may not be so rare. Among the last 38 patients with new-onset AOSD seen at our center (Nantes), five (13%) were diagnosed in the ICU. In a recent single-center series, 19 out of 57 patients (33%) developed one or more organ complications (with or without organ failure) [3]. In the present study, we wanted to specifically focus on life-threatening AOSD, i. e. on ICU patients with overt organ failure. This stringent inclusion criterion explains the rather small size of this multicenter nationwide series.
Both in our series and in published cases, we found that three key AOSD organ manifestations can lead to ICU admission. The most frequent is respiratory failure due to lung damage and/or pleural effusion. The second is cardiocirculatory failure (non-cardiogenic shock, myocarditis, and/or cardiac tamponade). The third complication is hematologic disorders, including RHS and/or DIC. The majority of patients exhibit a combination of organ manifestations and one-third has MOF. These complications occurred mostly during the first flare. Expectedly, patients had usual signs of AOSD. Half of them exhibited the classic triad of high-grade fever with evanescent maculopapular rash and arthromyalgia [2]. Other clues to the diagnosis, including sore throat, serositis, neutrophil leukocytosis, and hepatic cytolysis were highly prevalent. However, patients suffered significant diagnostic delay. Median time between fever onset and treatment was 3 weeks. Even once organ failure had occurred, therapeutic delay remained significant. The main  issue was the difficulty to abandon the sepsis hypothesis, as reflected by the amount of antibiotics these patients received before AOSD diagnosis. High-grade fever, sore throat, lung infiltrates, hepatic cytolysis, and neutrophilic leukocytosis clearly misguided clinicians, along with the negativity of autoimmunity screening. RHS is the most frequent AOSD complication [3,[17][18][19]. Both entities share pathophysiologic pathways and clinicobiological features [2,[17][18][19][26][27][28][29][30]. As for RHS, hyperferritinemia is a useful diagnostic clue for AOSD [2,21,22]. The sensitivity and specificity of > 1000 ng/ml hyperferritinemia are 67-80% and 36-41%, respectively, outside the ICU. In our series, all patients had > 2000 ng/mL and 17/ 20 (85%) reached 10,000 ng/ml. Two out of three patients without such a massive hyperferritinemia only suffered from cardiac tamponade. In recent AOSD series, the incidence of RHS is 15%. Its mortality rate is 10 to 15% [3,[17][18][19], which is lower than in infection and/or malignancy-related RHS [17][18][19]. No validated criteria are available for the diagnosis of AOSD-related RHS. We found that biological parameters that best correlated with bone marrow hemophagocytosis were thrombocytopenia and hepatic cytolysis. Paradoxically, other cytopenias and ferritin level had less predictive value, which emphasis the role of macrophage activation in AOSD, beyond hemophagocytosis. Recently, Bae et al. studied 109 AOSD patients including 21 with clinically defined RHS [18]. They found that low platelet count, anemia, and hepatomegaly were significant predictors of clinically defined RHS, but that leukocyte count and ferritin level were not. A scoring system has recently been proposed to estimate the probability of RHS (H-score) of any cause [26]. Missing data precluded its calculation in our patients. Interestingly, this has been found to predict survival in an Asian AOSD cohort [27].
AOSD, RHS, and sepsis share pathophysiological aspects, including the prominent role of pro-inflammatory cytokines IL-1, IL-6, and IL-18 [2,[28][29][30][31][32][33]. Interestingly, a recent study has shown that clinicobiological assessment is more reliable than cytokine profiling to distinguish AOSD from sepsis [33]. As for RHS, a careful analysis of patients' biological profile is an important step to suspect AOSD [2,32]. In the appropriate clinical setting, the combination of absolute or relative thrombocytopenia or hypofibrinogenemia, and massive hyperferritinemia should lead to consider the diagnosis of AOSD.
Our second objective was to assess the prognosis and response to therapy. Two patients of ours succumbed to AOSD (10%). Yet, mortality may be higher since some patients with fatal AOSD have certainly been missed. A significant proportion of patients experienced classical ICU complications and/or sequelae such as distal necrosis or neuropathy, but long-term prognosis was favorable.
Overall, only half the patients responded to CS. We found that these patients tend to have a less severe disease, and lower ferritin levels. IVIgs were frequently used as second-line therapy in patients who failed CS. Indeed, persisting or recurrent fever under CS in an ICU patient raises the issue of superimposed infection. In such circumstances IVIgs are an appealing therapy even if little data are available regarding their use in AOSD. Our study and literature data show that IVIgs are rarely effective, with an overall response rate of only 41%. By contrast, we found that IL-1-targeting therapy with anakinra was highly effective. In our series, five patients (25%) received anakinra in the ICU, with a response rate of 80%. One of the limitations of its use is the lack of data for patients with severe renal failure. Indeed, the only patient that did not fulfill our response criteria (i.e leaving ICU without further treatment) died after anakinra was stopped due to persistent fever, infections, and AKI requiring RRT. Twenty years ago, a phase III randomized controlled trial investigated the efficacy of anakinra in ICU patients with severe sepsis [34]. In this trial, stopped for futility, 116 patients had received anakinra. Mortality rate was not different from the placebo group. Of note, a recently published post hoc analysis suggests that anakinra could be beneficial in sepsis patients with features of RHS [35]. A significant proportion of our patients had persistent disease activity after ICU discharge. At 3 months, 16 out of 18 required additional treatment. At this time point, 11 patients had received anakinra, which was effective in ten cases (91%).
Our series have several limitations owing to its retrospective nature, limited size, and heterogeneity. However, retrospective studies are essential tools for the study of such rare diseases. They can also provide meaningful information regarding the efficacy and safety of therapeutic strategies in complex yet real-life situations.

Conclusions
AOSD should be recognized as a potential sepsis mimic in patients with fever of unknown origin admitted to ICU. In the absence of diagnostic gold standard, intensivists should be aware of the clinical and biological clue to this diagnosis. Fifty percent of patients are refractory to CS, thus early intensive treatment may be discussed on an individual patient basis [36]. Whether life-threatening AOSD is a distinct subset or merely the result of diagnostic delay remains to be determined. Further studies are needed to identify risk factors for life-threatening complications in patients with AOSD that had not yet entered the ICU.