Infection-Related Death among Persons with Refractory Juvenile Idiopathic Arthritis

Bacterial sepsis led to multiorgan failure in persons receiving immunosuppressive and antiinflammatory drugs.

and after recent exposure to anti-tumor necrosis factor-α biological DMARDs, and 2 during hematopoietic stem cell transplantation procedure. Reporting all cases of severe infections and especially deaths in these children is of paramount importance for accurate surveillance. J uvenile idiopathic arthritis (JIA), a group of clinically heterogonous conditions with arthritis of unknown origin beginning before 16 years of age and persisting for >6 weeks, is the most common childhood chronic rheumatic disorder (1). The most severe forms are those with systemic (So-JIA) or polyarticular (poly-JIA) onset, progressing to polyarticular disease. Despite the success of conventional and new biological disease-modifying antirheumatic drugs (DMARDs), a substantial percentage (>30%) of patients will have ongoing active disease into adulthood that includes sequelae from chronic inflammation and considerable morbidity from joint damage, osteoporosis, growth retardation, psychosocial morbidity, and reduced quality of life and education or employment (2).
Rapidly evolving guidelines (3) include the windowof-opportunity concept, where biological DMARDs are used as tailored therapy, depending on the disease category, and in naive patients not previously treated with conventional DMARDs (e.g., corticosteroids) (2). In the United Kingdom, treatment guidelines are regulated by the National Institute for Clinical Excellence, which specifies indications for tumor necrosis factor-α (TNF-α), interleukin (IL) 6, T-cell activation, and IL-1 blocking agents (https:// www.nice.org.uk/guidance/ta373; https://www.nice.org. uk/advice/esnm36). However, no good evidence exists to guide clinicians when they confront failure of the initial biological DMARD (4), and switching to a second (or third) (5) or combining >2 biological DMARDs (6) raises concern about risks from severe infections and development of malignancy and new autoimmune disorders (7,8). For this small group of patients with refractory JIA, hematopoietic stem cell transplantation (HSCT) might be the only treatment option (9).
We report details of 4 patients referred for HSCT because of refractory JIA who died of overwhelming infection from central venous catheter (CVC)-related bacterial sepsis (3 patients) or disseminated viral infection (1 patient). Two died while receiving high-dose systemic corticosteroids and methotrexate and after recent exposure to anti-TNF-α agents, before they underwent stem cell collection, and were not started on conditioning chemotherapy at the time of death; 2 died during HSCT (10,11).

Case Reports
Parents provided informed consent for data and tissue collection. Consent was obtained when the child started treatment with new anti-TNF-α agents (British Society for Paediatric and Adolescent Rheumatology database) or at assessment for HSCT (Newcastle upon Tyne Hospitals National Health Service [NHS] Foundation Trust, Newcastle upon Tyne, UK).  (Table 2). On June 1 (one week after new CVC insertion), she was admitted with fever (38°C), macular erythematous rash, vomiting, swelling and pain of several joints, and cough (Table 2). Chest examination and radiographic findings were normal, and she was treated empirically with systemic antimicrobial drugs (teicoplanin and meropenem) for 1 week (15) and a 3-day course of IVMP for presumed MAS. After transient improvement during the next few days, on June 6 the child again became unwell, with fever (38.5°C), rash, hepatomegaly, and joint pain ( Table 2). Empirical treatment with systemic antimicrobial drugs (teicoplanin and ceftriaxone) was restarted with another 3-day course of IVMP, but she remained febrile (38.5°C) and became restless with abdominal pain and vomiting. She was transferred to the pediatric intensive care unit (PICU), where surgical reasons for acute abdomen pain were excluded. Ciprofloxacin and ambisome were added to the treatment regimen, but the child's condition rapidly progressed into multiorgan failure (Table 2), and she died on June 10. Multiple peripheral blood and CVC cultures taken during this period remained sterile. The family did not agree to autopsy. Culture from the CVC tip removed postmortem grew coagulase-negative Staphylococcus.

Patient 4
Patient 4 was a 13-year-old girl whose rheumatoid factor + poly-JIA was diagnosed at 4 years of age. She had progressive and debilitating polyarthritis refractory to conventional (corticosteroids, methotrexate) and biological (infliximab, etanercept, anakinra, rituximab) DMRADSs (JADAS-10 score over the years [24][25][26][27][28][29][30]. She underwent HSCT in 2009. α-hemolytic Streptococcus grew from CVC culture taken during a febrile episode after receipt of anti-T-cell globulin, and she was treated empirically with meropenem and teicoplanin; unusually, she rapidly progressed into multiorgan failure requiring ventilatory, inotropic, and renal support in the PICU. Because results of initial liver function tests, including clotting, were normal, and C-reactive protein (CRP) response was adequate, the impression was of bacterial (or fungal) septicemia and renal failure. After transient improvement, she completed conditioning and HSCT and, despite renal failure, maintained stable neutrophil engraftment but remained platelet dependent. Bone marrow biopsy was hypocellular and showed some evidence of macrophage activation. Subsequently, and in parallel with acute pancreatitis, encephalopathy, and progressive enteral and liver failure, the girl manifested prolonged hyperinflammatory response (CRP 100-170 mg/L [reference 0-5 mg/L]; fibrinogen 6-10 g/L [reference 1.5-4.0 g/L]; raised neutrophil count >20 × 10 9 cells/L) despite broad-spectrum antimicrobial and antifungal therapy. Multiple cultures and viral PCRs from different sites (blood, CVC, and other line tips; bone marrow and intestine biopsy; cerebrospinal fluid; maxillary sinus washing) remained negative. She died on day 43 after HSCT; autopsy confirmed multiorgan failure with severe secondary pancreatitis (11).

Treatment with Combined DMARDs, Including Biologicals
At death, patients 1 and 2 had active disease treated with high-dose systemic corticosteroids and methotrexate; because they previously had been exposed to long-term and multiple DMARDs, including anti-TNF-α biologicals (Table 1). Patients 3 and 4 underwent autologous Tcell-depleted HSCT after a severely immunosuppressive conditioning regimen. Clinical observation of increased risk from severe infections in children with JIA, often requiring treatment in a hospital (17), recently was confirmed in a study of a large JIA cohort (18). The increased risk for concurrent immunosuppressive therapy, in addition to the underlying diseaserelated immune dysfunction (8,17), was supported by this study, in which high-dose systemic corticosteroids, but not methotrexate and/or anti-TNF-α agents, substantially increased susceptibility to severe infections (18). Although simultaneous use of different biological DMARDs is not common (5,6), a recently published study highlighted the exposure to multiple and often combined immunosuppressive drugs, such as corticosteroids, methotrexate, cyclosporine, cyclophosphamide, and a variety of biological antiinflammatory DMARDs, including TNF-α, IL-1, and IL-6 blocking agents, as a major risk factor for severe infections and the unusually high rate of death for a selected group of children with refractory So-JIA, often complicated by MAS, and associated with pulmonary hypertension, interstitial lung disease, and alveolar proteinosis (19). It is well recognized that severe immunosuppression targeting B-and/or T-lymphocyte functions (e.g., B-cell-depleting agents, such as rituximab, T-cell activation blocking agent abatacept, anti-CD52 monoclonal antibody alemtuzumab, HSCT procedure) is often complicated with infections caused by a wide spectrum of pathogens, such as pyogenic bacteria, viruses, and fungi (7,(9)(10)(11). In contrast, blocking specific inflammatory pathways (e.g., IL-1, IL-6, TNF-α) mimics some of the very rare primary immunodeficiencies of the innate immune system (20), with susceptibility to a relatively narrow range of pathogens (21,22). Infections with encapsulated pyogenic bacteria have been reported in patients with deficiencies of IL-6 function (23), whereas the essential role of TNF-α in defense against intracellular pathogens (24) was highlighted by the initially observed increased risk for mycobacterial infections associated with anti-TNF-α biologicals (7,12), prompting the introduction of effective screening and prevention measures (24). Although inconsistently reported (25), unusually prolonged, severe, and life-threatening infections with common bacterial pathogens are being seen in children with JIA treated with combined DMARDs, including anti-TNF-α agents (26,27). As observed in these patients (23,28), because of blocking of the inflammatory response mediated by TNF-α, IL-1, or IL-6 cytokines, these patients might not have high fever and raised CRP, a fact of which clinicians should be aware (20)(21)(22)(23)(24)29). Regardless of active disease and combined immunosuppressive therapy, patients 1 and 2 (i.e., those not undergoing HSCT) had preserved inflammatory responses (Table 2) and appropriate routine immunologic testing (data not shown).

CVC-Related Bacterial Infections
Three patients died of CVC-related bacterial sepsis with α-hemolytic Streptococcus and coagulase-negative Staphylococcus, common and fully sensitive organisms, despite timely administered appropriate antimicrobial therapy (15). Long-term CVCs are essential for patients requiring frequent blood tests and intravenous treatments. Unfortunately, CVC-related bloodstream infection is a well-recognized and potentially severe complication. Coagulase-negative Staphylococcus species are the most common pathogens causing CVC-related infections. Guidelines recommend treatment with 10-14 days of systemic antimicrobial drugs and antibiotic locks, but routine CVC removal is not recommended because most patients have a benign course and rarely develop sepsis or poor outcome (15). S. aureus, enteric gram-negative bacilli and Candida are less frequent but potentially more severe pathogens. Coagulase-negative Staphylococcus species (S. epidermidis in particular) were the most common (>50%) pathogens identified from 146 episodes of bacteremia in 64 children with primary immunodeficiencies undergoing HSCT in Great North Children's Hospital, whereas Enterococcus species, gram-negative organisms, and Candida were isolated only in few cases each (30). Most (80%) episodes were successfully treated with appropriate systemic and antibiotic locks; CVC was removed in 12 patients, and the only death resulted from overwhelming C. albicans infections despite CVC removal (30). Contrary to that study, severe and life-threatening CVC-related sepsis caused by S. epidermidis has been reported in a significant percentage of children with systemic vasculitis treated with infliximab and combined immunosuppressive and/or antiinflammatory therapies (31). Two deaths from CVC-related sepsis resulting from coagulasenegative Staphylococcus and combined Escherichia coli and Candida infection were reported from a cohort of children with inflammatory bowel disease treated with adalimumab in combination with other immunosuppressive medications (32). This high risk for severe CVC-related infections associated with active disease that requires multiple immunomodulatory therapies was also reported from a group of children with pediatric rheumatic disease-related complications admitted to PICU, with substantially high rate of death (50%), of which 44% resulted from multiorgan failure (33).

MAS versus Multiorgan Failure
Because the 3 patients we describe who died of CVC-related bacterial sepsis manifested unusually severe and rapidly progressing multiorgan failure, we considered the possibility of MAS in the differential diagnoses (14,16,33,34).
MAS is a well-recognized major risk factor for death in children with JIA admitted to PICU (33,34). A recent international study of a large cohort of So-JIA patients identified the most common clinical features as fever, organomegaly, central nervous system involvement, and hemorrhage (16). The most useful laboratory parameters were thrombocytopenia; hyperferritinemia; increased liver transaminases, lactate dehydrogenase, triglycerides, and D-dimer levels; and decreased leukocyte count, erythrocyte sedimentation rate, and fibrinogen (16).
Some features that manifested near death, such as persisting fever, falling neutrophil and platelet counts, and increased serum transaminase levels ( Table 2; patient 1 on March 1) suggest MAS in patient 1 (16). Unfortunately, not all of the laboratory parameters highlighted as essential for the clinical diagnosis of MAS were available (14) ( Table  2). However, the fact that she collapsed after CVC was accessed and that α-hemolytic Streptococcus grew from the CVC line tip, lung tissue, and pleural effusion samples after death favors infection as the cause of death. In patient 2, persisting fever, hepatomegaly, and high serum ferritin level suggested MAS, but increasing platelet and neutrophil counts, erythrocyte sedimentation rate, and fibrinogen and normal liver transaminase levels did not support MAS (14) ( Table 2; patient 2 on June 6). Although rapid deterioration to terminal multiorgan failure was associated with falling platelet count, deranged liver function, and clotting ( Table 2; patient 2 on June 10), the high neutrophil count suggested overwhelming fungal infection (which cannot be ruled out because autopsy was not performed), regardless of negative fungal cultures. In patient 4, unusually severe, progressive multiorgan failure developed after CVC-related sepsis, but normal liver function results and clotting and adequate CRP response did not support MAS. Features of macrophage activation in bone marrow biopsy were seen later in the course of progressive multiorgan failure and in parallel with unusual hyperinflammation, suggesting possible fungal infection. However, multiple cultures from different sites remained negative, and autopsy confirmed severe pancreatitis. In patient 3, disseminated adenovirus and EBV infection/reactivation during HSCT led to bone marrow aplasia. In the terminal stage of multiorgan failure with A. fumigatus infection, results of liver function and clotting tests were normal, and inflammatory markers were raised (erythrocyte sedimentation rate 80 mm/h [Westergren method; reference 1-10 mm/h]; CRP 200 mg/L [reference 0-5 mg/L]; ferritin 11,000 µg/L [reference 20-60 µg/L]).

Deaths and Reporting Deaths
Although the death rate for JIA has decreased since the 1970s, 1 of 2 recent studies referring to the period before the use of biological DMARDs reported a standardized mortality ratio of 3.4 (95% CI 2.0-5.5) for boys and 5.1 (95% CI 3.2-7.8) for girls (35). This nationwide cohort study from Scotland (1,246 children with JIA during 1981-2000) reported 39 JIA-associated deaths, for which the most common causes were the underlying disease (9 cases), circulatory complications (8 cases), and respiratory complications (6 cases) (35). In the United States, the standardized mortality ratio for the JIA pediatric rheumatology mortality database (9,604 children with JIA during 1992-2001) was lower at 1.8 (95% CI 0.66-3.92), with 6 of 19 reported deaths occurring among children with So-JIA caused by MAS and heart failure (2 each) and infection and secondary malignancy (1 each) (36).
In addition to uncontrolled disease activity and its complications, in particular amyloidosis in the past and MAS today, several recent reviews highlighted the substantial risk for death from severe infections in children with JIA who are receiving biological DMARDs (7,8). Most of these children were treated with multiple and combined classical DMARDs before or alongside biological DMARDs. Although the range of causing pathogens is broad, pyogenic bacteria and herpes viruses were the most common (7,8). An unexpectedly high death rate (68% [17/25]) in a group of children with refractory So-JIA associated with pulmonary complications was recently reported from an international group of 25 patients (19). Although disease onset ranged from the 1980s onward and most children in the cohort had received multiple immunosuppressive and antiinflammatory drugs, 19 (76%) diagnoses were made after 2000; contrary to previous reports (35,36), 68% had been exposed to biological DMARDs (19). Strikingly, the calculated death rate for these 19 patients increased by almost 50% over the figure recently reported from the US pediatric rheumatology mortality database for the period up to 2000, before the use of biological DMARDs (36).
The regional pediatric rheumatology service at Great North Children's Hospital is closely linked to the national center providing expertise in investigating and treating children with primary immunodeficiencies for northern England and a leading center for HSCT in children with severe rheumatic diseases. This database holds ≈1,250 children in whom JIA has been diagnosed and who were treated and followed during 1994-2013; each year, 50-80 new patients are referred. The calculated death rate for children in this cohort is 0.032% (4 patients died during this period). Three had So-JIA and 1 rheumatoid factor + poly-JIA, all with progressive polyarthritis and poorly controlled systemic disease for many years, including fever, rash, organomegaly, and high acute-phase reactants, with features of MAS in 2 So-JIA patients (14,16). All died during the early 2000s after long-term treatment with multiple and combined conventional and biological DMARDs: corticosteroids and methotrexate (all 4 children), cyclosporine and leflunomide (1 each); anti-TNF-α agents (3 etanercept, 4 infliximab); IL-1 blocking agent (2 anakinra); and B-celldepleting (anti-CD20) monoclonal antibody (1 rituximab). Although considered in all 4, only 2 patients underwent autologous T-cell-depleted HSCT (10,11). With the expertise in pediatric rheumatology, immunology, and infectious diseases available in centers best placed to care for these most severe and often refractory cases, fatalities are rare but do occur.
Monitoring the safety and reporting the side effects, including severe infections and deaths, of new biological DMARDs is a priority of national and international patient registries and multicenter, international collaborative research consortiums, such as the Childhood Arthritis and Rheumatology Research Alliance, Pediatric Rheumatology International Trials Organisation, Pediatric Rheumatology Collaborative Study Group, and Single Hub and Access Point of Care for Pediatric Rheumatic Diseases in Europe (37,38). However, reporting of deaths is still inconsistent. In children with JIA treated with multiple DMARDs alongside anti-TNF-α biologicals, a systematic review from 2013 reported 4 deaths, 3 of which were associated with severe infections: 2 treated with etanercept (group A Streptococcus-related purpura fulminans) and 1 with adalimumab (bacterial sepsis); for 1 treated with infliximab, infection cause was not given (27). However, Hashkes et al. (8) commented on 6 deaths, all in children treated with combined DMARDs, including anti-TNF-α agents, and associated with serious infections: 3 treated with etanercept (1 each with suspected sepsis, MAS, and tuberculosis [previously treated with infliximab]); 2 with infliximab (sepsis); and 1 with adalimumab (MAS and interstitial pneumonia). Furthermore, neither Woerner and Ritz (7) nor Swart et al. (24) referred to severe infections and infection-related deaths associated with biological DMARDs from 2 clinical trials reported in 2012 (39,40). Six deaths were reported in trials of the anti-IL-6 agent tocilizumab, including 1 each from probable streptococcal sepsis and MAS (other listed causes were traffic accident, pulmonary hypertension [2 cases], and pneumothorax) (40). Two deaths were reported in a trial of long-acting anti-IL-1 agent canakinumab, both from MAS (1 was previously treated with anakinra and tocilizumab) (39). All these patients had active disease and had been treated with a combination of multiple DMARDs before or at death; some were reported as part of the So-JIA cohort associated with pulmonary complications and high death rate (19). The most recent report from the United States for 2008-2012 highlighted 7 deaths (1 from an accident) in children with JIA treated with multiple DMARDs (methotrexate and steroids), including biologicals (4 anakinra, 1 each etanercept and infliximab), of which 3 were associated with severe infections (multiorgan failure from disseminated tuberculosis, viral illness, and sepsis) (38).

Conclusions
Three patients with refractory JIA reported here died with evidence of CVC-related bacterial sepsis caused by common pathogens (α-hemolytic Streptococcus and coagulasenegative Staphylococcus): 2 while receiving high-dose systemic corticosteroids; and methotrexate and after recent exposure to the anti-TNF-α biological DMARD infliximab; 1 during HSCT procedure. As has been previously reported, we observed unusual severity of septic shock and rapid progression to multiorgan failure despite timely and appropriate antimicrobial treatment (31)(32)(33). Patient 4 died of disseminated adenovirus and EBV infection during HSCT procedure.
Evidence from clinical trials facilitated by multicenter international collaborative research enabled introduction of biological DMARDs in the treatment of rheumatic diseases in children and unprecedented improvement in their care during the past 20 years (1)(2)(3)(4)41,42). However, severe infections are emerging as an important risk factor for death among children with JIA treated with combined and multiple conventional and new biological DMARDs (8,19,37,38). Accurately reporting all cases of severe infections and especially deaths in these children is of paramount importance, as was highlighted a decade ago (25). Although monitoring safety and reporting side effects of new biological DMARDs is in place and improving, we note marked inconsistency in the current literature (7,8,19,24,27,(38)(39)(40).