Delivering on the Antimicrobial Resistance (AMR) Agenda Is Not Possible Without Improving

Delivering on the Antimicrobial Resistance (AMR) agenda is not possible without improving fungal diagnostic capabilities

David W. Denning1,2, David. S. Perlin2,3, Eavan G. Muldoon1, Arnaldo Lopes Colombo2,4, Arunaloke Chakrabarti2,5, Malcolm D. Richardson2.6and Tania C. Sorrell2,7

1 The National Aspergillosis Centre, University Hospital of South Manchester, The University of Manchester, Manchester Academic Health Science Centre, Manchester, UK

2 Global Action Fund for Fungal Infections, Geneva, Switzerland;

3 Public Health Research Institute, Rutgers Biomedical and Health Sciences, Newark, NJ, USA;

4 Division of Infectious Diseases, Universidade Federal de São Paulo, São Paulo, Brazil.

5Center of Advance Research in Medical Mycology, Postgraduate Institute of Medical Education & Research, Chandigarh 160012, India

6 Mycology Reference Centre, Manchester; Manchester Academic Health Science Centre; University Hospital of South Manchester, Manchester and International Society for Human and Animal Mycology

7Marie Bashir Institute for Infectious Diseases & Biosecurity, University of Sydney, Australia.

*Correspondence:

Professor David Denning

National Aspergillosis Centre, Education and Research Centre, University Hospital of South Manchester, Southmoor Road, Manchester M23 9LT

Tel.: +44161 291 5811

E-mail:

One line summary: Widespread use of accurate and sensitive fungal diagnostic testing will reduce unnecessary antibacterial and antifungal therapy

Key words: Aspergillus, Candida, Pneumocystis, Histoplasma, Cryptococcus, resistance

Word count: 2196

ABSTRACT

Antimicrobial resistance (AMR) is a major public health concern largely arising from the excess use of antibacterial and antifungal drugs. Here we argue that the lack of routine fungal disease diagnostic testingexacerbates the problem of antimicrobial empiricism, both antibiotic and antifungal. We cite 4 common clinical situations that are illustrative. 1. Accurate diagnosis of fungal sepsis in the hospital and the Intensive Care Unit to reduce the use of broad-spectrum antibacterial therapy administered inappropriately to patients with invasive candidiasis. 2. Failure to diagnose smear-negative pulmonary ‘tuberculosis’ as chronic pulmonary aspergillosis. 3. Mis-diagnosing ‘fungal asthma’ that results in unnecessary courses of antibiotics instead of antifungals and missing life-threatening invasive aspergillosis in COPD patients. 4. Both over- and under-treatment of Pneumocystis jirovecii pneumonia in HIV positive patients. All communities should have access to non-culture fungal diagnostics, which will have a substantial benefit for clinical outcome, antimicrobial stewardship and AMR control.

Antimicrobial resistance (AMR) is a significant public health concern, and a major threat to modern medicine (1). It is estimated that in the United States antibiotic resistant infections are associated with 23,000 deaths per year (2), and that excess healthcare costs are in the region of US$20-25 billion (3). MinimizingAMR has been the focus of accelerating efforts with multi-pronged approaches tailored to individual countries and healthcare settings. Even if the difficult task of developing new antimicrobials is successful, current efforts aimed at reducing the development of resistance will need to be maintained to protectthese novel compounds.

A central tenet of controlling AMR is antibiotic stewardship, which seeks to limit inappropriate antibiotic usage either by avoiding unnecessary prescribing or discontinuing antibiotic therapy immediately it is not required. Within the context of stewardship programs, inadequate attention has been paid to fungal infection as the cause of antibacterial treatment failure. Furthermorethe importance of accurate and timely diagnosis of fungal infection in defeating AMR has been starkly absent from policydiscussions (4). Accurate diagnosis or exclusion of fungal infection impacts substantially on antimicrobial usage and our ability to limit bacterial AMR.

Generally,physicians prescribe most antimicrobials empirically as veryfew infections are microbiologically diagnosed in real time. Further, most infections are never microbiologically documented, which matters little if the patient improves and infection resolves, but leads to additional empirical antimicrobial use if the patient deteriorates (5). Fungal infections are often undiagnosed and untreated, and they are a frequent fatal complication(superinfection) of numerous diseases (cancer, liver, respiratory and/or renal failure, sepsis, AIDS etc) that contribute to inappropriate antibiotic usage. This must stop.

We provide four specific clinical examples where we argue for greater application of existing fungal diagnostics andimproved overall fungal diagnostic capability in line with the recently issued ’95-95 by 2025’ Roadmap from the Global Action Fund for Fungal Infections (6).

Accurate diagnosis offungal sepsis in hospital and Intensive Care Unit (ICU)

Hospitalized patients, especially those in the ICU, are often inappropriately placed on broad-spectrum antibiotics because fungal diseases involving Candida spp are not diagnosed. Bloodstream infection and invasive candidiasis are substantially more common than appreciated. Multiple factors are likely to be responsible, including unrestrained antibiotic use, indwelling devices, increasing populations of immunocompromised patients, increased renal support and others. Bloodstream infections with Candida spp. have a population incidence of 1.2-26/100,000 in multiple studies, with the highest rates in middle- and high-income countries, notably the USA(7,8). Most (93%) of episodes are healthcare-associated (~80% as inpatients) and 7% are community acquired (9). In Brazil, with a population of 194 million, Candida bloodstream infections are seen in 14.9/100,000 population, or 29,000 people each year (10), based on prospective data obtained 10 years ago from 11 medical centres (11). A recent prospective study from 27 ICUs in India found a mean incidence of 6.51 cases of ICU-acquired candidemiaper 1,000 ICU admissions with a 35-75% mortality(12). An estimated 14.3 million patients are admitted to ICUs in India each year. Undoubtedly this high rate underestimates the problem, as blood culture is only ~40% sensitive for invasive candidiasis (including intra-abdominal candidiasis), with fluconazoleand echinocandins substantially reducing the yield from blood culture (13-15).It is likely therefore that the actual burden and death rate in ICU exceeds 200,000 patients and ~100,000 deaths. If, as is found in other countries, Candida bloodstream infection (and presumably invasive candidiasis) outside ICUs is at least twice ascommon as in ICUs, over 600,000 patients are estimated to have invasive candidiasis in India each year, with ~300,000 deaths, assuming patients are treated, and many more if not.

In addition, a high prevalence of candidemia has been reported in children and neonates in India and Latin America. Central nervous system involvement is quite common in premature infants leading to a high rate of neurological sequelae(16).

While invasive candidiasis is strongly associated with prior bacterial infection and anti-bacterial therapy, inappropriate escalation and combination antibacterial therapy will typically have been administered to patients with invasive candidiasis. In one large study of 444 patients with Candida bloodstream infections, 81% were exposed to multiple antibacterial drugs, either concomitantly or sequentially (17) and in the ICU study from India described above, 95% were already on antibiotics, usually two or more (12). Early therapy of Candida bloodstream infection greatly improves patient outcomes as shown in several observational studies, and even better if ‘correct’ therapy is given immediately (18).

Once Candida sepsis is confirmed, antibacterial agents can usually be stopped and, if ruled out, empirical antifungal therapy can be stopped. Inflammation without infection requires no antimicrobial therapy. Three well-validated diagnostic tools, two configured for ruling out a diagnosis of invasive candidiasis, are now available - the 1,3 beta-D-glucan assay(19),the Candida albicansgerm tube antibody test (CAGTA) (serum samples) (20) and a newly FDA approved nonculture-based molecular assay which is substantially more sensitive than blood culture and useful for making the diagnosis Candida (EDTA blood)(21). In patients without invasive candidiasis, antimicrobial stewardship programsbased on such diagnostics have successfully curtailed the use of antifungal therapy in the ICU without worsening patient outcomes(18,22). The economics depend on the cost of diagnostic reagents and testing, antifungal costs and incidence of infection(23). Overuse of antifungal agents is very costly and can promote antifungal resistance, may carry toxicity and the potential for numerous detrimental drug interactions (18). Modeling will be required to determine the magnitude of the impact on antibacterial prescribing.What is not in dispute is that patient outcomes for those with fungal infection will improve. Antimicrobial stewardship programs will be even more effective if these diagnostic tools, with their rapid turnaround times, are readily available(24).Widespread implementation of rapid non-culture Candidadiagnostics will greatly improve prescribing practices for the complex hospital patient.

Mis-diagnosis of smear-negative pulmonary ‘tuberculosis’ as tuberculosis

Smear-negative pulmonary tuberculosis (TB) is aproblematic area for clinicians and policy makers. Post-tuberculous sequelae are common, poorly studied and may be mistaken for active, recurrent TB (25), and itis apparent that an important under-recognised issue for these patients is chronic pulmonary aspergillosis (CPA), which can mimic TB presentation. Among 544 patients previously treated for TB, with a residual cavity, in the UK, 34% developed precipitating antibodies to Aspergillus fumigatus of whom 63% developed aspergilloma within 2 years (a late stage of chronic pulmonary aspergillosis) and 42% of these patients developed hemoptysis (26). Prospective studies of CPA (after treatment for TB) are otherwise lacking, but conservatively an assumed rate of ~10% is likely, with ~1.2 million individuals affected (26).

Culture of Mycobacterium tuberculosis from smear-negative patients is slow and may be falsely negative. The use of new, highly sensitive,DNA detection assays (e.g. Xpert MTB/RIF) directly on respiratory specimenshas transformed the rapidity of detecting positive results, but there remains a considerable proportion of unwell, smear negative, PCR-negative patients. Some of these are patients who have ‘relapsed’ following anti-tuberculous therapyhave CPA. Those with ‘smear-negative TB’ and HIV infection, have a higher mortality than smear positive patients (27), probably because many do not have TB at all. It is increasingly being recognized that many of these patients are chronically infected with Aspergillus spp. resulting in CPA that is largely undiagnosed and untreated.

Weight loss, worsening cough, chest pains, dyspnea and fatigueare common to TB and CPA, and the radiological abnormalities are similar (Figure 1). In studies from the UK, Brazil, Korea, Iran and India, the frequency of elevated Aspergillusserum antibody after TB varied upwards from 20%(28,29. Aspergillus antibody detection is the key diagnostic test for CPA, and performs well (96-97% sensitivity, 92-98% specificity (30,31), compared with controls. Given that CPA is a common sequel to TBand has a 5-year mortality of 75-80%, it needs to be sought actively using antibody testing(32). Sera from almost all patients with ‘recurrent TB’ in Iran were Aspergillus antibody positive(33). Pneumocystis jirovecii DNA was identified in the sputum ofup to 7% of patients with smear-negative TB in Brazil and 6.7% in Uganda, providing an alternative, potentially treatable diagnosis, and there are others such as histoplasmosis (Figure 2) and coccidioidomycosis (5).

Empirical anti-tuberculous therapy is unnecessary in patients with either chronic pulmonary aspergillosis or other fungal infections, as it isineffectiveand exposes the patient to potential toxicity. Yet, it remains the default approach for most complicated TB patients. When such treatment fails, as it usually does, there is a risk that patients are assumed to have multidrug resistant (MDR) TB, and the inappropriate substitution of second and third line anti-tuberculous agents adds to potential toxicities and healthcare costs. The size of the problem is substantial with the WHO reporting 2,755,870 patients with smear negative TB in 2013 (34). Aspergillus antibody detection should be widely available and integrated into TB control programs with CPA as post-TB sequelae emphasised.All symptomatically recurrent TB cases should be screened for Aspergillus antibody.

Fungal exacerbations of asthma and chronic obstructive airways disease (COPD)

It is common practice in the community to treat asthma and exacerbations of COPD with antibiotics and corticosteroids, although guidelines caution against unnecessary use, especially in COPD patients (35). Most patients respond, even if the exacerbation is virus-induced. Patients with moderate and severe COPD are frequently admitted to hospital and have an in-hospital mortality of 6%. The majority patients (>80%) are treated with antibiotics (36), although multiple guidelines advise against this in the absence of purulent sputum and pulmonary infiltrates. It is now well established that Aspergillus airways colonization (or infection) is strongly associated with exacerbations (37). Two studies have addressed the frequency of invasive aspergillosis in these patients - with a 1.3% rate in Spain (38) and a 3.9% rate in southern China (39), based on sputum cultures revealing Aspergillus spp (which likely under-estimates the problem as Aspergillus culture is insensitive). The respective mortality rates were 65% and 43%. Most are not treated for invasive aspergillosis but are treated unnecessarily with antibiotics. The scale of the problem is large: In China, 87 per 10,000 people over 40 years of age (an estimated 11,858,000) COPD patients were admitted to hospital in 2012 (40,41), ~462,000 may have developed invasive aspergillosis and ~200,000 may have died. Presumably other countries with high COPD rates such as Hungary, Ireland, and New Zealand face a similar situation.

Asthmatics with frequent exacerbations and/or poorly controlled disease may take several antibiotic courses a year and some are managed with long-term antibiotics.Recently, awareness and understanding of ‘fungal asthma’ has increased and a timely diagnosis and the use of antifungal therapy could substantially reduce antibiotic prescriptions.Long-term antifungal therapy is efficacious in 60-80% of patients with fungal asthma, notably allergic bronchopulmonary aspergillosis (ABPA) and severe asthma with fungal sensitisation (SAFS) (42). Both antibiotic and corticosteroid use fall with successful antifungal therapy and may reducehospitalization.

Diagnosis of fungal asthma relies ontotal and fungal specific IgE testing, or skin prick testing which is simple to do, yet not often done. Rapid antigen detection with a lateral flow device might accelerate diagnosis, although no data are published on utility for sputum, or on COPD patients. Few studies have addressed the role of fungi in precipitating exacerbations of COPD, although culture, antigen and Aspergillus IgG antibody all contribute to the diagnosis of chronic and invasive aspergillosis.

Failure to properly diagnose and treat fungal asthma and COPD patients with Aspergillus colonisation continues to increase the inappropriate use of antibiotics (and corticosteroids) among these patients. Recognition of fungal infection and/or allergy with directed antifungal therapy would greatly reduce exacerbations, medical consultations and hospital admissions.There is an urgent need to evaluate and implement both fungal culture and non-culture diagnostics (Aspergillus IgE, Aspergillus IgG, Aspergillus antigen and PCR) for COPD and asthma exacerbations, and to diagnose fungal asthma and treat it with antifungals.

Making and excluding the diagnosis ofPneumocystis pneumonia in AIDS

Pneumocystis pneumonia (PCP) in AIDS is often diagnosed empirically based on a subacute onset of cough, breathlessness out of proportion to the radiological abnormalities and subtle, bilateral chest X-ray changes, in the context of a low CD4 cell count (Figure 3). Co-trimoxazole (trimethoprim-sulphamethoxazole, Bactrim®, Septrin®) is the most effective agent for both prevention and therapy of PCP. A low dose is effective for prophylaxis, but a 3-week course of high and potentially toxic doses is required for effective therapy. The differential diagnosis of PCP is broader in children as bacterial pneumonia is more common. If a precise diagnosis could be achieved in most cases of PCP, much inappropriate use of co-trimoxazole would be prevented.

Rates of PCP among newly hospitalized adults with advanced HIV infection are highly variable (<1% to 60%), with rising rates as gross domestic product increases (43). Without adequate diagnostics available, many patients will unnecessarily receive high dose co-trimoxazole with or without corticosteroids for 3 weeks. If the actual number of PCP cases in patients with AIDS is 400,000, then 100,000s may be given co-trimoxazole unnecessarily, with toxicity rates as high as 90%(5,44). Early detection and diagnosis of PCP can help prevent unnecessary hospitalization and cost. Currently bronchoscopy and microscopic examination of bronchoalveloar lavagefluid is the commonest definitive means of establishing the diagnosis, with a sensitivity of 75-90% depending on the microscopy technique (45). Pneumocystis jirovecii is non-culturable in routine laboratories and so molecular detection with PCR is commonly used in Europe with a sensitivity of 95-99%(46). Pneumocystis PCR performed on expectorated sputum is also effectivefor detecting P. jirovecii(47-49), yet infrequently used.In breathless children, PCR on nasopharyngeal aspirates is currently the only realistic means of establishing the diagnosis. In serum,1,3 beta-D-glucan is detectablein nearly all cases of PCP (19), and if negative,effectively rules out infection. Assuming that 25% of PCP cases are mild, immediate diagnosis and use of oral therapy will potentially avoid 100,000 admissions to hospital annually, or even more if the diagnosis is ruled out and patients are not admitted for unnecessary PCP therapy(5). Mild PCP responds well to treatment, and prevents progression to moderate or severe infection.

Provision of rapid Pneumocystisdiagnostics will enable early diagnosis, allow discontinuation of broad spectrum antibiotics if positive and discontinuation of high dose cotrimoxazole and corticosteroids if negative. Missed PCP diagnoses, obscured by concurrent bacterial infection, will be minimized.Pneumocystis PCR should be widely available for all respiratory samples serving large immunocompromised populations and 1,3 beta-D-glucan in high volume laboratories. These diagnostics will also definitively improve outcome for non-AIDS immunocompromised patients for the same reasons as in AIDS.

Other clinical scenarios

We have not addressed multiple other clinical situations in which precise fungal diagnosis could reduce inappropriate antibacterial prescribing, including cryptococcal meningitis(Figure 4), overtreatment of Candida found in the respiratory tract and urinewhich can be reduced with the use of better markers of infection, under-diagnosis of invasive aspergillosis in critical care, over-treatment with antifungals of febrile neutropenia in leukaemia, Aspergillus bronchitis in bronchiectasis, allergic, chronic and invasive fungal sinusitis and PCP in HIV-negative patients. In all these clinical settings, inappropriate antibacterialor antifungal prescribing is common, through lack of adequate fungal diagnostic testing. We have also not addressed either rapid detection of antifungal resistant fungi (ie Aspergillus terreus,Candida krusei) or the use of diagnostics to prevent super-infection with high resistance potential fungi, such as Candida glabrata or Rhizopus oryzae. Antifungal resistance is problematic in some settings, demands informed prescribing, therapeutic drug monitoring and development of new antifungal agents(50).