Background

Healthcare-associated infections (HAIs) due to fungal organisms have been increasing in incidence, particularly among immunocompromised patient populations, and have been recognized by the World Health Organization as a global health concern.¹ Yeasts such as Candida species and various molds such as Aspergillus species and Mucorales have caused numerous outbreaks in healthcare settings, especially related to construction, inadequate air filtration, or contaminated medical products such as linens.^{Reference Glowicz, Benowitz and Arduino2–Reference Hartnett, Jackson and Perkins5} Construction or renovation activities may account for up to 50% of healthcare-associated Aspergillus outbreaks.^{Reference Kanamori, Rutala, Sickbert-Bennett and Weber6} Identification of fungal HAIs may therefore provide an important indicator of a systemic issue requiring mitigation.^{Reference Kanamori, Rutala, Sickbert-Bennett and Weber6}

Outbreaks of invasive fungal infections (IFIs) are quite challenging to investigate given complex diagnostic criteria, protracted hospitalizations among at-risk patients requiring extensive record review, and unclear incubation periods.^{Reference Hartnett, Jackson and Perkins5} Because of these challenges, consensus definitions, such as those developed by the Infectious Diseases Group of the European Organization for Research and Treatment of Cancer and the Mycoses Study Group (EORTC/MSG) have been used to standardize epidemiologic studies of IFIs.^{Reference Peter Donnelly, Chen and Kauffman7} These definitions incorporate multiple domains, such as host factors, clinical features, and laboratory evidence to define a suspect case of IFI as “proven,” “probable,” and “possible.” In addition, the Centers for Disease Control and Prevention (CDC) has developed tools to assist with investigating possible outbreaks of IFIs,⁸ including the Guide to Investigating Suspected Outbreaks of Mucormycosis in Healthcare.^{Reference Hartnett, Jackson and Perkins5}

Surveillance for IFIs by healthcare infection prevention (IP) teams requires systematic case identification to detect temporal trends and outbreaks. This typically relies on review of positive microbiology cultures for fungal organisms by the IP team. Reliance on fungal cultures can be problematic as they may require invasive diagnostic procedures for testing, have low diagnostic sensitivity, and are limited by long turn-around times.^{Reference Huygens, Schauwvlieghe and Wlazlo9,Reference Lestrade, Bentvelsen and Schauwvlieghe10} . Some institutions include results from serum antigen markers and histopathology reports in their surveillance of fungal HAIs in addition to fungal cultures.^{Reference Patterson, Zidouh, Miniter, Andriole and Patterson11} However, widespread use of serum antigen biomarkers such as Aspergillus galactomannan or 1,3-beta-D-glucan (BDG) in effect results in many patients with IFI labeled as “probable” and not “proven.”

Upon identification of a possible fungal HAI, additional detailed review by the IP team is required to collect information to investigate the case, such as admission dates, sites of infection, bed movement, patient risk factors, construction activities, and common procedures, among others.⁸ Patients at risk for fungal infections may have prolonged hospitalizations, making assessment of potential exposures and commonalities labor-intensive requiring lengthy medical record review. IDSA has defined a case of aspergillosis as likely to be healthcare-associated if it occurs more than seven days after admission,^{Reference Patterson, Thompson and Denning12} and this cutoff has been widely used for fungal surveillance.^{Reference Patwari, Cutrell, Bhaskaran, Trevino and Sreeramoju13} However, incubation periods for many IFIs are unclear and a seven-day cutoff is arbitrary. Additionally, symptoms of fungal infections are generally nonspecific and the diagnosis is often delayed, therefore determining date of onset (and thus attributing the infection as “healthcare-associated”) may not be possible. As identification of an outbreak of IFI (particularly due to mold) will prompt extensive investigation including environmental assessments, this presents a challenge to IP teams.^{Reference Hartnett, Jackson and Perkins5} IFIs related to construction may involve multiple pathogens over a prolonged period and can therefore be particularly difficult to detect. These investigations typically require extensive work by a multidisciplinary team led by an infection preventionist and involving members of facilities, environmental services, clinical microbiology, and public health experts.

Recently, plasma cell-free DNA (cfDNA) metagenomic next-generating sequencing assays have been developed as a method to detect IFIs.^{Reference Huygens, Schauwvlieghe and Wlazlo9} At present, Karius Inc is the only commercial laboratory to offer such a test. The Karius Test can detect more than 1000 pathogens from a plasma specimen.^{Reference Hong, Blauwkamp, Kertesz, Bercovici, Truong and Banaei14} cfDNA holds the promise of detecting various pathogens from a single plasma sample in a pathogen-agnostic manner and the potential to avoid additional invasive diagnostic procedures. In contrast to various fungal antigens such as the serum galactomannan immunoassay or BDG cfDNA is typically used solely for diagnosis of suspected infection and not for routine biomarker screening given the high cost ($2,000) and lack of data on this approach.

At UCLA Health, our IP team did not previously review cfDNA results as part of our IFI surveillance. cfDNA tests at our institution are ordered at the discretion of consulting infectious disease physicians to assist with diagnosis for select cases, such as culture-negative endocarditis. However, following a cluster of cases of apparent hospital-onset invasive Rhizomucor cases in highly immunocompromised patients diagnosed solely by cfDNA and therefore missed by routine clinical surveillance, our IP team began to explore how to incorporate cfDNA results into our IFI surveillance.

Benefits of cfDNA testing

cfDNA has been shown to have moderate sensitivity but high specificity for IFI.^{Reference Huygens, Schauwvlieghe and Wlazlo9} Fungal results of cfDNA tests, particularly mold results, have been shown to have relatively higher positive clinical impacts.^{Reference Dong, Saleh, Kaur and Yang15} Studies have suggested improved diagnosis for several conditions such as pneumonia in immunocompromised patients, however there may be variable performance in different patient populations. While prophylactic or empiric antifungal therapy may decrease the sensitivity of the galactomannan immunoassay, it does not appear to impact the sensitivity of cfDNA.^{Reference Marr, Laverdiere, Gugel and Leisenring16,Reference Huygens, Schauwvlieghe and Wlazlo17} In the PICKUP study, addition of cfDNA to conventional diagnostic testing for pneumonia increased the diagnostic yield by 12%.^{Reference Bergin, Chemaly and Dadwal18} cfDNA may have the most clinical impact in detection of fastidious organisms not easily grown in cultures.^{Reference Dong, Saleh, Kaur and Yang15}

By incorporating cfDNA results into a facility’s existing epidemiological IFI surveillance, detection of hospital-onset IFI could be improved. Earlier identification of IFI outbreaks may prompt additional investigation that would uncover underlying shortcomings in construction, linen management, or building design, resulting in mitigation of these issues.

Challenges of cfDNA testing

Use of cfDNA for surveillance of healthcare-associated IFIs, while promising, presents several challenges:

1. 1. Variable availability and diagnostic stewardship, potentially resulting in delayed cfDNA test ordering, artificially shifting community-onset cases to healthcare-associated.

2. 2. Data transmission and lack of integration into electronic health records requiring manual review.

3. 3. Need for clinical interpretation to determine the significance of the results.

4. 4. Inability to compare with environmental or clinical cultures, such as by phenotypic drug susceptibility or whole genome sequencing.

Variable availability and diagnostic stewardship

While increasing in popularity, cfDNA is not universally available at all medical centers that conduct surveillance for healthcare-associated fungal infections. cfDNA is expensive ($2,000), and therefore different medical centers may have widely divergent approaches as to when to order cfDNA, how and to whom ordering the test is restricted, and/or requirements for conventional workup prior to sending a cfDNA.^{Reference Hogan, Yang and Garner19} At many centers, cfDNA may not be ordered until the clinical team has exhausted conventional microbiologic workup including awaiting culture results. There has been active discussion in the field of clinical microbiology of whether cfDNA should be reserved as a last resort for cases for which conventional microbiological testing has failed to provide an answer, or whether it should be performed earlier and in parallel to conventional tests.^{Reference Hogan, Yang and Garner19} In a medical center with tight restrictions on cfDNA ordering, an IFI identified solely by positive cfDNA may artificially appear to occur well into a patient’s hospital stay, making distinguishing between community vs healthcare-associated IFI challenging, potentially resulting in extensive investigation for an infection that was in reality community-acquired and present on admission. Selection bias and variable determinations by clinicians of pre-test probability for IFI could also impact detection of IFI outbreaks. The dynamics of cfDNA in relation to clinical signs and symptoms and radiographic abnormalities remain unclear, which could further complicate determination of the source of acquisition of the infection. The turn-around time for cfDNA (2–4 days, including shipping) may additionally add delays that could complicate adjudication of IFIs.

Additionally, centers that use cfDNA commonly would be expected to have a higher number of IFIs detected than a center where use of the test is tightly restricted. This could result in widely varying prevalence rates of IFI, which might confound investigation and outbreak control measures.

Data transmission and lack of integration into electronic health records

cfDNA results are currently provided as an electronic report which displays the names of all identified microbial cell-free DNA organisms as well as the concentration of their DNA. In addition to obligate and opportunistic pathogens such as Aspergillus species or Coccidioides immitis/posadasii, cfDNA may also identify commensal microbes or transient microbial DNA (such as Escherichia coli, Enterococcus faecalis, and Pseudomonas aeruginosa) and DNA viruses such as cytomegalovirus and Epstein-Barr virus. Since the reports do not presently interface directly into electronic health systems, incorporation of cfDNA into IP surveillance requires manual review of reports which can be labor intensive for IP teams. This can be mitigated by summary reports sent on a standing basis (eg, quarterly), but this approach could delay identification of a mold outbreak.

Need for clinical interpretation

Interpretation of a positive cfDNA test requires clinical judgment by an experienced physician who can integrate history and physical exam findings, results of imaging studies, histopathological findings, and other microbiologic results such as cultures and serum biomarkers.^{Reference Dong, Saleh, Kaur and Yang15} Determining whether positive cfDNA represents true IFI involves considerably more review than surveillance for other organisms, such as LabID for Clostridioides difficile. A positive cfDNA report for mold such as Aspergillus spp. does not necessarily imply an IFI. cfDNA does not distinguish commensal/colonizing organisms from pathogenic organisms. In addition, high genetic similarity among mold species can often lead to inaccurate species identification due to the nature of metagenomic sequencing, which often detects a limited number of DNA fragments without sufficient fungal genome coverage to allow definitive identification.^{Reference Salem-Bango, Price, Chan, Chandrasekaran, Garner and Yang20} Occasionally, multiple closely-related mold species may be reported by cfDNA yet only one mold infection is confirmed by culture (clinical observation, unpublished findings). Therefore, caution should be taken when interpreting a stand-alone positive cfDNA without integration of the patient’s entire clinical picture.

Inability to compare with environmental or clinical cultures

An additional challenge with conducting IFI surveillance by cfDNA is that since no pathogen is recovered in clinical culture, no clinical isolate is available for comparison to samples obtained from other patients or potential suspect environmental sources such as ventilation systems or linens. Comparison of environmental and clinical isolates by phenotypic drug susceptibility testing is not possible when no phenotypic testing can be done, and molecular epidemiologic tools such as whole genome sequencing to assess relatedness of isolates cannot be utilized.^{Reference Salem-Bango, Price, Chan, Chandrasekaran, Garner and Yang20} Identification of, say, multiple cases of Aspergillus fumigatus infection solely by cfDNA could potentially be misinterpreted as an outbreak, whereas clinical isolates could be sequenced to determine relatedness.