Death by antibody

Abstract

A 42-year-old woman with a background of psoriatic arthritis presented with a 7-day medical history of fevers of unknown source; she had recently undergone elective shoulder arthroscopy, and her medications included anti-interleukin 17A (anti-IL-17A) drug, secukinumab.

She went on to develop sepsis-induced cardiomyopathy, requiring veno-arterial extracorporeal membrane oxygenation (ECMO), from which she was successfully weaned after 12 days. However, she then went on to develop a candidaemia, with new intra-abdominal collections found incidentally on CT; despite appropriate anti-fungal therapy and attempts at drainage, she passed away.

Both anti-IL-17A treatment and ECMO have been shown to be significant independent risk factors for Candida infection. The use of monoclonal antibody therapy in the management of autoimmune disease, and the use of ECMO in the intensive care setting are each becoming increasingly widespread. Fungal infection should be screened for early in this critically unwell group of patients, and treatment initiated as indicated.

Background

We present a case report which brings together emerging risk factors for infection, both of which are likely to become increasingly significant in clinical practice. We discuss a patient with a persistent candidaemia who had two significant risk factors for fungal infection. First, she had been undergoing immune-modulatory therapy in the form of an interleukin-17A (IL-17A) monoclonal antibody, which may play an increasing role in the management of a wide range of autoimmune and inflammatory conditions. Second, having initially developed sepsis-induced cardiomyopathy, she received extracorporeal membrane oxygenation, an invasive aspect of intensive care medicine that is developing a role in a growing number of critically unwell patients, but which again is associated with complications, of which infectious diseases (ID) play a significant role.

This combination of risk factors has not previously been described in literature, therefore we present this case in order to highlight the challenges that the case presented and open discussion around the themes.

Case presentation

We present the case of a 42-year-old woman who presented with a 7-day medical history of fevers. She had a background medical history of psoriatic arthritis, for which she was being treated with the novel IL-17A inhibitor secukinumab for 2 months prior to presentation; this had been commenced by her dermatology team following unsuccessful treatment with tumour necrosis factor-alpha (TNF-α) blocker infliximab. She had also recently undergone an elective shoulder arthroscopy.

She was admitted to her local hospital and extensively investigated. She gave no history of infective symptoms other than fever and malaise, no recent travel or known contact with any illness. Blood tests showed a predominant neutrophilia (24.1×10⁹/L) with C reactive protein of 320. No evidence of the source of infection was found on clinical examination, which included senior orthopaedic review of her shoulder to rule out septic arthritis, given her recent arthroscopy. Chest and shoulder radiographs, urine culture and blood cultures were all negative. She was commenced on 1.2 grams of co-amoxiclav three times a day, as per her local hospital protocol for a sepsis of unknown origin.

By day 3, she remained febrile with temperatures greater than 39°C, and was haemodynamically deteriorating with tachycardia and hypotension, so was transferred to her local intensive care unit (ICU). No source of infection was identified on whole body CT imaging. Due to ongoing sepsis, she had developed significant acute kidney injury (creatinine was 415 μmoles/L) and required inotropic support with norepinephrine. She had developed persistent culture-negative diarrhoea and thrombocytopaenia. Multiorgan involvement, fever, hypotension and erythroderma (chronic secondary to psoriasis), meant that despite negative cultures she met four out of five of the diagnostic criteria for toxic shock syndrome, therefore her antibiotics had been switched to clindamycin and ceftriaxone and she had received a stat dose of gentamicin (3 milligrams per kilogram of bodyweight).

On day 7 of her admission, she developed a new right bundle branch block on an ECG. A subsequent echocardiogram showed a left ventricular ejection fraction of 26% with global hypokinesis; there were no valvular abnormalities or vegetations. She was transferred to Barts Heart Centre for further cardiology input and ongoing supportive management.

By the 10th day of admission, the patient was deteriorating, with increasing inotropic requirement (0.4 μg/kg/min of norepinephrine), pulmonary oedema and rising blood lactate levels (persistently greater than 3 millimoles per litre) with a metabolic acidosis. Repeated transthoracic echocardiogram showed significant biventricular systolic dysfunction, with a left ventricular ejection fraction of 6%, as shown in figure 1; her cardiac index measured with a pulmonary artery catheter was 1.6–1.8 L/min/m² and her mixed venous oxygen saturation was 40%–50%.

Figure 1.

Transthoracic echocardiogram showing diffusely mildly increased left ventricular wall thickness with severe global hypokinesia. Left ventricular ejection fraction was 6% measured by Simpson’s biplane method. The right ventricular cavity size and wall thickness are normal with severely impaired right ventricular systolic function.

Due to this further deterioration, the decision was made to intubate the patient and initiate veno-arterial extracorporeal membrane oxygenation (VA ECMO). This was complicated by a short cardiac arrest following induction and therefore emergency femoral ECMO cannulation was undertaken on ICU rather than in theatre. The arterial access was subsequently revised to a right subclavian artery cannulation in theatre, due to absent arterial flow to the distal right leg. Furthermore, the right femoral artery cannulation site became infected with Pseudomonas aeruginosa, and wound necrosis required repeated debridement; antibiotic therapy was therefore escalated to meropenem and linezolid to cover skin and soft tissue infection. Transoesophageal echocardiogram was used during ECMO cannulation, and images again showed global hypokinesia but no evidence of infective endocarditis.

The patient remained on VA ECMO for 12 days, before reduced inotropic requirements and improved cardio-respiratory function permitted successful weaning and de-cannulation. Left ventricular systolic function had recovered to an ejection fraction of 60%, seen in figure 2.

Figure 2.

Transthoracic echocardiogram in M-mode on the parasternal long axis showing left ventricular ejection fraction of 60%.

Five sets of blood cultures taken from her vascath, central line and peripherally, on day 25 flagged positive for a Candida glabrata; these were the first positive blood cultures since admission. Our patient was commenced on amphotericin B (ambisome) but this was soon switched to anidulafungin, 100 milligrams once a day, due to her deteriorating liver function. We know from the Mycology Reference Lab report that the C. glabrata was fully sensitive, so the antifungal choice was appropriate, and following a course of treatment the blood cultures remained negative for 16 days, however the patient remained septic.

On day 68, C. glabrata was again isolated from a peripheral blood culture. This prompted further investigation for a possible hidden source, as she had been treated with appropriate antifungals but was again candidaemic. A repeat CT abdomen to identify the source of this candidaemia demonstrated significant peri-pancreatic, peri-gastric and peri-duodenal collections, as shown in figure 3. These collections were not present on the initial CT scans and it is unlikely that they were the cause of her initial presentation, given persistently negative blood cultures at this time.

Figure 3.

(A) Axial slice of CT abdomen showing a large submucosal gastric collection measuring up to 10.1 cm in antero-posterior depth and 13.2 cm in cranio-caudal extent. (B) Axial slice of CT abdomen showing further large peri-pancreatic collection abutting the head and uncinate process and measuring up to 6.1 cm in antero-posterior diameter and 8.1 cm in cranio-caudal dimension.

Multiple radiologically-guided percutaneous drains were inserted into the collections over several weeks, and culture of the peri-gastric collection also showed C. glabrata. However, despite appropriate anti-fungal treatment and intervention, complete resolution was never achieved.

Differential diagnosis

The initial presentation of our patient was with sepsis of unknown origin despite extensive investigation. Toxic shock syndrome was considered, as the patient was immunosuppressed, had psoriasis and met four out of five diagnostic criteria, however causative organisms were never isolated. Septic arthritis was also considered, given recent arthoroscopy in an immunosuppressed patient, however examination and imaging excluded this as the cause.

She subsequently became haemodynamically unstable, due to sepsis-induced cardiomyopathy with associated biventricular dysfunction requiring VA ECMO.

Following recovery from this initial insult and decannulation from VA ECMO, the patient developed a new candidaemia and sepsis secondary to uncontrollable intra-abdominal fungal collections. A risk factor for this was secukinumab.

Outcome and follow-up

Thirteen weeks following her initial presentation, the patient further deteriorated with an upper gastrointestinal bleed, rising lactate and increasing inotropic requirements. CT abdomen demonstrated widespread pneumatosis coli and portal venous gas consistent with bowel ischaemia, with necrosis leading to the gastrointestinal bleeding. She was taken to theatre and underwent extensive resection of large and small bowel, however owing to multiorgan failure (fulminant liver failure, dependence on renal replacement for 2 months, and high-dose multi-inotrope dependence), the decision was made to withdraw active care and she sadly passed away.

Discussion

Secukinumab is a human monoclonal antibody that binds to IL-17A and inhibits its function. It is currently licensed in the treatment of psoriatic arthritis, ankylosing spondylitis and in moderate to severe plaque psoriasis in which there has been an inadequate response to anti-TNF-α.

IL-17 is a pro-inflammatory cytokine produced by T cells in response to a pathogen, which recruits monocytes and neutrophils to the site of infection. It therefore plays a key role in fighting infection caused by extracellular bacteria and fungi (in particular Candida species), and in regulating gut microbes. IL-17A deficiency, either in mice studies or in humans with hyper-IgE syndrome, leads to susceptibility to Candida infections.¹

IL-17 has also been shown to be involved in the pathophysiology of several autoimmune and inflammatory diseases, such as multiple sclerosis, psoriasis, rheumatoid arthritis, asthma and chronic obstructive pulmonary disease. Novel immune-modulatory therapy such as anti-IL-17 medications have proven to be a significant breakthrough in the management of these conditions. Further to these, early evidence suggests anti-IL-17A can also decrease MRI lesion activity in multiple sclerosis,² and may be effective in the management of systemic lupus erythematosus, rheumatoid arthritis, airway disease and inflammatory bowel disease^{1 3}; the role of anti-IL-17A drugs may therefore expand as the evidence for their use grows.

A systematic review in 2017 found a small, dose-dependent, increase in the rates of oral candidiasis in patients treated with anti-IL-17A drugs⁴; however, all of these infections were mild to moderate and either self-resolved or were responsive to standard therapy. They did not report any severe or systemic infection, however they raised this as a possibility, as neutropaenia had been widely reported as an adverse effect of IL-17 inhibitors. They proposed screening and treatment algorithms for Candida infections in patients to be treated with anti-IL-17A drugs to minimise both exacerbation of pre-existing conditions and the risk imposed by immunosuppression; screening may include tests for tuberculosis, hepatitis and HIV, and advice regarding planning for live vaccinations and major surgery. The screening algorithm suggested by Saunte et al in their 2017 review in the British Journal of Dermatology is shown in figure 4.

Figure 4.

Saunte et al’s proposed screening algorithm for Candida infections in patients treated with anti-interleukin (IL)-17 drug.⁴

In the context of a patient presenting with biventricular dysfunction in whom myocarditis was a plausible diagnosis, review of the literature reports no significant cardiovascular adverse effects related to anti-IL-17A treatment. Conversely, treatment with anti-IL-17 in mice with viral myocarditis actually ameliorated myocardial inflammation.⁵ Similarly, the T helper cells which produce IL-17 were found to be critical in the onset of experimental autoimmune myocarditis, therefore inhibition in fact represents a potential target for prevention of myocardial inflammation.⁶

ECMO itself is widely known to be a significant risk factor for infection, given its invasive nature, in a population of critically unwell patients. The Extracorporeal Life Support Organisation (ELSO) set up a task force in 2008 in response to growing concerns around the prevention, diagnosis and treatment of infections related to the use of ECMO. This task force was comprised of both experts in ECMO, and also in ID, both in adult paediatric populations. They sought to review the ELSO database, relevant literature and assorted local guidelines, in order to provide evidence-based recommendations on the recognition and management of ECMO-related infections. Their analysis and recommendations were subsequently published in the ELSO ‘Red Book’ in 2012.

Their analysis of data from 1998 to 2008 found an overall incidence of infection of 11.7%, with the highest rates found in the adult population, and lower in the paediatric and neonatal groups (30.6, 20.8 and 10.1 infections per 1000 ECMO days). VA ECMO was the mode of support associated with the highest risk of infection, and also that prevalence of infection increased with duration of ECMO support over 14 days.⁷ The organisms that were cultured most commonly, and therefore which the group recommend to be covered with empiric antimicrobial therapy are coagulase negative Staphylococcus (incidence of 15.9%), Candida species (12.7%) and P. aeruginosa (10.5%).⁸

There were caveats to this data, in that the culture source (blood/urine/sputum) was not included in the database, neither was the date and other comorbid factors such as open chests, however they felt that the pattern of infection, the respective incidences and mortality data would nevertheless prove useful in guiding prophylaxis and treatment in patients developing sepsis. Pluim et al analysed the Extracorporeal Life Support Organisation’s (ELSO) clinical database of patients placed on ECMO between 1997 and 2009; they found that fungal infection either preceding or during ECMO was associated with a statistically significant higher risk of mortality in all age groups.⁹

In fact, due to the higher than expected incidence of Candida infection and the associated high mortality rates, the task force went on to state: ‘…it is the strong recommendation of the task force ID experts, that clinicians raise their index of suspicion for yeast in significantly ill patients suspected to have sepsis on ECMO and lower their threshold for antifungal treatment in these patients.’ ⁸ This surprisingly high incidence of Candida infection has been consistent with other research groups; a 2012 observational study in an Australian centre found 24 cases of bloodstream infections within their group of 139 patients who had undergone ECMO cannulation; Candida was the most common cause of bloodstream infection, in nine out of the 24 cases.¹⁰

The task force investigated the use of prophylactic antibiotic therapy during ECMO; they found that practice is very varied but that there is no robust evidence for the use of prophylactic antibiotics during ECMO runs, and therefore recommended that they should be avoided beyond the surgical cannulation procedure. However, this recommendation does not extend to central VA ECMO, wherein an open chest significantly predisposes to mediastinitis¹¹ and therefore they suggest that antibiotic therapy is guided locally by the surgeons and ECMO team.⁸ Significantly, the task force acknowledges the unpredicted high incidence of fungal infections in ECMO patients and the significant mortality of this group; they therefore recommend ‘cautious, but aggressive’ empirical antifungal prophylaxis in patients who are assess as high risk of infection. These include patients who have had prolonged poly-antibiotic therapy, those with a long period with an open chest, and also those who are significantly immunocompromised.⁸

Learning points.

• Anti-interleukin-17 medications increase susceptibility to both bacterial and fungal infections.

• Screening and treatment algorithms for fungal infections in this group of susceptible patients exist.

• Extra-corporeal membrane oxygenation is a significant risk factor for systemic infections, including Candida species.

• Extracorporeal Life Support Organisation recommend ‘cautious but aggressive’ use of antifungal prophylaxis in patients deemed to be at particularly high risk of infection, including immunocompromised patients.