FormalPara Key Summary Points

Why carry out this study?

Adult-onset Still's disease (AOSD) is an inflammatory disorder with unknown causes.

It is treated with corticosteroids, immunosuppressive agents, and biological therapies.

However, the potential risk of developing macrophage activation syndrome (MAS) with the use of biological therapies in AOSD patients has not been well understood.

What was learned from this study?

Our research revealed a higher incidence of MAS in patients with AOSD treated with tocilizumab (TCZ) compared to those treated with anakinra (ANA).

Moreover, we observed that elevated inflammatory markers at the time of TCZ initiation might increase the risk of developing MAS.

Introduction

Adult-onset Still’s disease (AOSD) is a rare systemic inflammatory disorder characterized by recurrent fever, rash, arthritis, and multi-organ involvement [1, 2]. The etiology of AOSD remains unclear, and its pathogenesis is believed to involve the unrestrained stimulation of innate immune cells and excessive production of diverse proinflammatory cytokines such as interleukin (IL)-1, IL-6, IL-18, and tumor necrosis factor (TNF)-α [3]. A range of treatments are available for AOSD, including corticosteroids, immunosuppressive agents, and biological agents [4,5,6]. However, in some patients, the disease remains refractory, and they may develop macrophage activation syndrome (MAS), which is a severe complication that can lead to multi-organ dysfunction, hemophagocytosis, and ultimately, high mortality rates [7]. Notably, cases of MAS have also been reported in patients receiving biological treatments [8,9,10]. Thus, evaluating the risk of MAS is necessary.

A recent meta-analysis indicated that IL-1 and IL-6 receptor inhibitors are the most effective biological therapeutic agents for AOSD [5]. Specifically, it has been proposed that IL-1 receptor inhibitors, such as anakinra (ANA), may be more effective in treating the systemic phenotype, whereas IL-6 receptor inhibitors, such as tocilizumab (TCZ), may show better responses in managing the chronic articular form [11]. Conversely, changes in the cytokine balance may trigger MAS in response to biological agents. Numerous cases of MAS during treatment with TCZ in patients with systemic juvenile idiopathic arthritis (sJIA), an analogous disease, have been reported [12]. As a result, MAS is currently recognized as a severe complication of IL-6 inhibitor therapy in patients with sJIA [13]. Although a systematic review of the efficacy and adverse reactions of biological agents in AOSD has been published [14, 15], to the best of our knowledge, no meta-analysis has been conducted on the incidence of MAS in patients treated with biological therapies. Therefore, it is unclear whether the type of biological agent affects the incidence of MAS. Furthermore, a recent study on the characteristics of MAS in patients with AOSD after TCZ administration has been reported [16], but it is limited to a small number of cases from a single institution.

In this study, we conducted a meta-analysis to investigate the incidence of MAS in patients treated with TCZ, a humanized anti-IL-6 receptor monoclonal antibody, and ANA, a widely used anti-IL-1 receptor antibody. Furthermore, to provide more detailed information on the clinical characteristics of patients who developed MAS while receiving TCZ, which has been approved in Japan for the treatment of refractory AOSD, we present a series of our cases.

Methods

Systematic Review and Meta-analysis

This study adhered to the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) protocol (Supplementary Fig. 1) and was registered with the Clinical Trial Registry of the University Hospital Medical Information Network (Japan) as UMIN000049243 in Japan [17]. Given the nature of the study, the need for institutional review board approval and informed consent was waived.

We conducted a systematic search of the following databases on October 17, 2022: PubMed, Web of Science, Cochrane Central Register of Controlled Trials, and Embase. The search formulas are presented in Supplementary text 1. The search was conducted using relevant keywords, such as “Still’s disease”, “biologics”, “tocilizumab”, and “anakinra” to identify the most pertinent articles. To ensure that all relevant studies and recent reviews were included, we also manually reviewed the bibliographies of the included studies. KT-M and AM independently screened the candidate articles by reviewing the title and abstract after uploading the citation list to Endnote X9 software (Thomson Reuters, Philadelphia, PA, USA) and Rayyan (https://www.rayyan.ai/). Following independent screening, articles and abstracts still considered candidates by at least one investigator were subjected to further screening based on the final full-text review. Any discrepancies between the two investigators were resolved before making a final decision.

The inclusion criteria were as follows: (1) a published case–control, cohort, or randomized clinical trial; (2) patients were clinically diagnosed with AOSD; (3) the experimental group was treated with ANA or TCZ, with or without a control treatment arm; (4) the study demonstrated the safety of biological treatment; and (5) availability of the full-text paper. The exclusion criteria were as follows: (1) systematic reviews or meta-analyses; (2) case reports or case series with less than five cases; (3) republished literature, unless it included new findings related to adverse events listed in the inclusion criteria; (4) studies with no or insufficient safety results at the time of the literature search; and (5) studies published in languages other than English. The diagnosis of AOSD was defined based on the Yamaguchi criteria [18].

Two authors (SA and AM) independently extracted information from the full text using an Excel data extraction sheet. The extracted information included study design, literature details (including first author and year of publication), study site, inclusion criteria, treatment protocol, and the number of MAS cases after biological administration. The frequency of MAS incidence following biological treatment was pooled using a random-effects meta-analysis by the generic inverse variance method (Review Manager ver 5.4, Cochrane Collaboration, London, UK). The standard error was calculated using Agrestia’s method [19]. The difference between the outcomes of TCZ and ANA was assessed by expressing P values based on random-effects analysis in Review Manager, with a P value less than 0.05 considered significant. Meta-regression was performed using “metafor” package of R software. The Newcastle–Ottawa Scale was used to assess the risk of bias in each cohort study [20]. The quality of the cohort study was independently assessed by two reviewers (SA, AM), and any disagreements were resolved through discussion or by involving a third reviewer (KT-M). Publication bias was assessed using funnel plots.

Patients and Design of the Yokohama City University Institutional Database

We searched for cases that met the Yamaguchi criteria from among patients who visited the Yokohama City University Hospital between January 2000 and December 2022. We collected data on patients who received TCZ, and those who met the diagnostic criteria proposed by Ravelli et al. [21] after TCZ administration were defined as having MAS. This study was approved by the Yokohama City University Ethics Committee (B210200064). The study was conducted according to the Declaration of Helsinki. Clinical symptoms and blood test data were retrieved from the patients' medical records. We compared the blood test results and clinical symptoms during flares in cases of MAS development under TCZ with those who did not develop MAS. Statistical analysis of the blood test data was performed with the Mann–Whitney U test using the GraphPad Prism software (La Jolla, CA, USA). Clinical symptoms were subjected to a Fisher’s exact test in SPSS (version 26). In this study, receiver operating characteristic (ROC) curves along with area under the curve (AUC) values were calculated to determine the optimal cutoff data that differentiate cases with MAS from cases without MAS. Finally, odds ratios (ORs) with 95% confidence intervals (CIs) were determined for each variable. P values less than 0.05 after adjustment of multiple comparisons were considered significant.

Ethical Approval

This meta-analysis is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors. Concerning the single-center analysis, the protocol and patient’s informed consent documentation were reviewed and approved by the Yokohama City University Ethics Committee (B210200064). The study was conducted according to the Declaration of Helsinki.

Results

Incidence of MAS in Patients with AOSD During ANA or TCZ Treatment

Of the 455 prospective articles, we identified 11 eligible retrospective cohort studies that presented the efficacy of biologics (ANA or TCZ) in treating patients with AOSD and reported the incidence of MAS during treatment (Fig. 1). The diagnosis of MAS was made by a clinician, and the diagnostic criteria used in some studies were unclear. The detailed characteristics of the patients are presented in Table 1. Five studies reported the incidence of MAS following treatment with ANA [22,23,24,25,26], and six studies reported the incidence following treatment with TCZ [8, 27,28,29,30,31]. The cohort studies were conducted in France (n = 4), Japan (n = 2), Italy (n = 1), China (n = 1), Taiwan (n = 1), Turkey (n = 1), and Israel (n = 1). The meta-analysis comprised 345 patients with AOSD who received ANA treatment and 94 patients with AOSD who received TCZ treatment. The pooled incidence of MAS following biological treatment was 1.50% (95% CI 0–3.36) for ANA and 14.01% (95% CI 4.51–23.51) for TCZ, respectively. The difference in incidence between the two treatments was statistically significant (P = 0.01). These findings are summarized in Fig. 2. According to the meta-regression using data for follow-up duration, the durations were not associated with the risk of MAS caused by ANA (Supplementary Fig. 2A, P = 0.75) and TCZ (Supplementary Fig. 2B, P = 0.35). The overall risk of bias for most studies evaluated was low (Supplementary Table 1).

Fig. 1
figure 1

Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flow diagram for the meta-analysis. The studies were selected in accordance with the PRISMA guidelines

Full size image
Table 1 Overview of the included studies
Full size table
Fig. 2
figure 2

Forest plot of the incidence rate of MAS. MAS incidence rates were compared between groups treated with ANA and TCZ. CI confidence interval, ANA anakinra, TCZ tocilizumab

Full size image

Summary of Single-Center Analysis

Of 76 patients diagnosed with AOSD based on Yamaguchi's criteria at Yokohama City University Hospital between January 2000 and December 2022, 17 consecutive patients who received TCZ were included in this study. Six patients developed MAS after TCZ administration, while 11 patients did not. Although age, sex, height, and weight at starting TCZ did not differ significantly between the two groups, the time from disease onset to TCZ induction was significantly shorter in the MAS group (P = 0.007). Furthermore, the MAS group had significantly higher white blood cell (WBC) counts, neutrophil counts, lactate dehydrogenase (LDH), C-reactive protein (CRP), and ferritin levels at TCZ induction (P = 0.014 for WBC, P = 0.015 for neutrophil counts, P = 0.001 for LDH, P = 0.007 for CRP, and P = 0.001 for ferritin) (Fig. 3). ROC analysis was conducted on laboratory data at the initiation of TCZ treatment to assess the risk of MAS development. The estimated AUC values for WBC, LDH, CRP, and ferritin were calculated (Supplementary Fig. 3), and the optimal cutoff thresholds for predicting MAS development were determined as 12.6 × 103/μl, 250 U/l, 2.0 mg/dl, and 560 ng/ml, respectively. The ORs indicated that higher levels of LDH, CRP, and ferritin were associated with the development of MAS (OR 13.3 for WBC; 95% CI 1.07–166.4, OR 22.5 for LDH, CRP, and ferritin; 95% CI 1.61–314.6). No clinical symptoms were significantly more prevalent in the MAS group. The systemic score also did not differ significantly between the two groups. The patient characteristics, symptoms, and laboratory data are presented in Table 2.

Fig. 3
figure 3

Scatter dot plots of the laboratory data of patients with and without MAS. The significantly different WBC and neutrophil counts and LDH, CRP, and ferritin levels are shown in the scatterplot dot plot. - lines on the graphs indicate median values. *P < 0.05, **P < 0.01, ***P < 0.001 by the Mann–Whitney U test. AOSD adult-onset Still’s disease, MAS macrophage activation syndrome, ANA anakinra, TCZ tocilizumab, WBC white blood cell, LDH lactate dehydrogenase, CRP C-reactive protein

Full size image
Table 2 Patient characteristics, symptoms, and laboratory data of our cases
Full size table

Discussion

The present study aimed to investigate the incidence of MAS in patients with AOSD treated with biological agents and to identify potential predictors of MAS. A meta-analysis of 11 retrospective cohort studies showed a significantly higher incidence rate of MAS under TCZ than ANA. Furthermore, analysis of the patients diagnosed at our institution revealed that certain laboratory data at TCZ induction, including LDH, CRP, and ferritin levels, could be predictive of the development of MAS. Recognizing the clinical and laboratory features of MAS is useful in clinical practice and may allow clinicians to identify patients at high risk for MAS and prevent its development.

While multiple factors are known to contribute to the development of MAS, there is a focus on the need to determine and address not only a single cause but also a combination of factors. In 2022, the ACR and the EULAR proposed the early stages of diagnosis and management of suspected hemophagocytic lymphohistiocytosis/MAS [32]. Immunotherapies, including biologics, may trigger MAS; however, MAS frequently emerges as a result of AOSD itself or due to infections. In patients with AOSD, genetic etiologies constitute a minority but noteworthy aspect necessitating attention, along with the investigation for underlying malignancy. When the development of MAS is suspected, a thorough workup should be initiated immediately to identify the underlying contributing factors (genetic causes, predisposing conditions, and acute triggers) of MAS [32].

In the studies included in the meta-analysis, the underlying cause of MAS remained indeterminate. Consequently, discerning whether MAS was triggered by the administration of ANA, TCZ, or potentially other factors, proves to be a complex endeavor. Furthermore, eight of the 12 patients who developed MAS after TCZ administration were Japanese. This may be attributed to racial differences, genetic characteristics, and the fact that TCZ is the only approved biological agent available for AOSD in Japan. In addition, there is no established evidence for biological agents other than ANA and TCZ for AOSD. For example, it has been reported that three patients with AOSD diagnosed with MAS were treated successfully by intravenous ANA and all patients experienced improvement [33]. Some reports have shown positive results for TNF inhibitors [34], while others have shown only a partial response [5, 35]. Therefore, this study analyzed the incidence of MAS after administration of ANA and TCZ, but not after treatment with other biological agents. Further data on the outcomes of biological agents other than ANA and TCZ are warranted.

Regarding single-center studies, there is a previous report on the predictors of MAS incidence [16]. The report compared six and eight patients with and without MAS, respectively, and found that the MAS group had significantly lower neutrophil counts and fibrinogen levels at the start of TCZ treatment than the “without MAS” group. This finding diverges from our data and has not yet been validated. This discrepancy may be attributed to the deep attenuation of baseline IL-6 activity in the MAS group and differences in background immunosuppressive therapy. Unlike previous reports, our patients had higher ferritin concentrations and neutrophil counts. Ferritin has been shown to promote neutrophil activation and neutrophil extracellular trap (NET) formation through activation of the ferritin receptor, macrophage scavenger receptor 1; and mice treated intraperitoneally with ferritin exhibited systemic inflammation characterized by cytokine storm and hepatitis, and markers of increased neutrophil mobilization and NET formation were evident in liver tissue [36]. Although inflammatory cytokines were not measured in this study, the findings of our cases suggest that the risk of developing MAS may increase if TCZ treatment is initiated when the levels of inflammatory markers are high. We previously reported that serum heme oxygenase-1, a heme-degrading enzyme highly expressed in Mo/Mφ serves as a biomarker for AOSD [37], and that the gasdermin D N-terminal level, a pore-forming protein playing central roles in inflammasome-mediated inflammation, was also elevated in active AOSD [38]. Monocyte distribution width is a novel cytometric parameter that correlates with the cytomorphological changes that occur with massive monocyte activation, suggesting that it may be useful as an indicator of AOSD activity [39]. Although these markers were not measured in the present study, they might be used to assess the risk of MAS development. In our case, inflammatory laboratory data on WBC count, neutrophil count, and LDH, ferritin, and CRP levels indicated inflammation, but these parameters were not included in the systemic score [40]. As a result, the systemic score is not significantly different between the two groups, and is consistent with previous reports [16]. Consequently, while the systemic score may be useful as an assessment of disease activity, it has limitations as a risk assessment method for the development of MAS, and suitable assessment index and biomarkers are necessary.

While the targeted inhibition of IL-1and IL-6 in patients with AOSD is presently acknowledged as a secure and efficacious therapeutic approach for enhancing disease management, our analysis suggests that IL-6 inhibition should be administered with caution, especially in patients with elevated CRP and ferritin. The 2022 ACR/EULAR recommendations have also outlined that initial empiric immunomodulatory treatment for individuals with rapidly progressive MAS may encompass high-dose glucocorticoids, ANA, and/or intravenous immunoglobulin [32]. Empiric use of ANA is supported, although it depends on the accumulation of limited data. Nevertheless, the available evidence to optimize biologic therapy remains constrained and further research is needed including the validation of novel biological agents such as IL-18 inhibitors and JAK inhibitors.

Our study had several limitations. Firstly, the included studies used varying diagnostic criteria of MAS, which may have introduced heterogeneity into the meta-analysis. Secondly, only cohort studies could be found and incorporated into the meta-analysis, which may have introduced selection bias. Possibly, patients with severe symptoms were more frequently switched to TCZ than ANA. Thirdly, reporting bias may have been present as some reports on AOSD cases treated with ANA or TCZ did not mention the development of MAS, while others did. Fourthly, the quality of evidence was limited by the retrospective nature of the included studies. Lastly, the small sample size of our single-center analysis limits the generalizability of our findings. We have ascertained that neither malignancy nor infection was the instigating factor behind MAS in patients with AOSD who experienced MAS while undergoing treatment with TCZ within our institution. However, evaluating whether the root cause lies within AOSD itself or is a consequence of drug-induced mechanisms remains a complex challenge.

Conclusions

In conclusion, our study provides evidence that the incidence rate of MAS under TCZ treatment is significantly higher than that under ANA treatment in patients with AOSD. Our study also found that patients with AOSD with high levels of inflammatory laboratory data at TCZ induction may be at an increased risk of MAS development. Although there are multiple causal factors for the emergence of MAS in patients with AOSD, recognizing the clinical and laboratory features of MAS is critical to prompt etiological evaluations, therapeutic interventions, and initiation of monitoring prior to the development of severe complications.