Abstract: Hospitalized patients with cirrhosis have a high rate of mortality. In the report by Bajaj et al., the negative impact of fungal infections (FI) on outcomes in a large US cohort of hospitalized cirrhotics is highlighted. Risk factors for FI are identified. Increasing awareness of FI along with the application of new diagnostic tools in species identification will provide the opportunity to improve patient outcomes.
Patients with cirrhosis requiring hospitalization are at a high risk for mortality and their care in hospital is more expensive than many common conditions such as pneumonia, congestive heart disease, and cerebrovascular disease (1). Through the work of the North American Consortium for the Study of End-Stage Liver Disease (NACSLED), we have a much clearer understanding of the factors associated with in-hospital mortality for cirrhotics, in particular the negative impact of infections (2, 3). In this month’s journal, Bajaj et al. (4), on behalf of NACSELD, sought to define the risk factors for and impact of fungal infections (FI) in a large, multi-center, cohort of hospitalized patients with cirrhosis. Their contribution is timely, adding valuable information to an area where data are limited.
From a cohort of 2,743 cirrhotic inpatients, of which 1,052 were diagnosed with infection, Bajaj et al. (4) reported that 134 (4.9%) had an FI. Most patients with FI (77.6%) had both fungal and bacterial infections, while only 30 patients were diagnosed with FI alone. All FI were nosocomial, with a median time to diagnosis of 5 days after admission. A urinary source for the FI was found in 43.5% of patients. Invasive FI were considered to be peritonitis, fungemia, urinary and lung infections. The majority of infections (77.6%) were caused by Candida species. By multivariable analysis, predictors of developing a FI included diabetes mellitus, acute kidney injury (AKI), admission to the intensive care unit (ICU), and infection on admission. Male gender was found to be protective. Patients with FI were more likely to have been on prophylaxis for spontaneous bacterial peritonitis and be sicker, as evidenced by higher model for end-stage liver disease (MELD) scores. These patients had longer hospitalizations, were more likely to require transfer to an ICU, and develop more organ failure. Overall 30-day survival was worse with a case fatality rate of 30%. Fungemia and fungal peritonitis had 30-day crude mortality rates of 64% and 57%, respectively. On multi-variable analysis, age, AKI, admission MELD, acute on chronic liver failure, ICU admission, and the ordinal infection variable were negatively associated with survival.
In the NACSELD study, FI were defined according to the Infectious Disease Society of America guidelines (5, 6, 7). All FI were treated as invasive infections. However, it is unclear if all patients actually had invasive infections. Fifteen patients were reportedly diagnosed with fungal lung infections. Candida species are often isolated from the respiratory tract of intubated patients and most frequently represent colonization. Candida pneumonia is extremely rare and typically occurs in markedly immunocompromised persons via hematogenous spread of the organism to the lungs. Cross-sectional imaging of the chest often reveals multiple pulmonary nodules, not lobar infiltrates. In order to make a firm diagnosis of Candida pneumonia, a tissue biopsy is required to confirm the presence of histopathologic evidence of invasive disease (7). It is also unclear if all study patients diagnosed with a fungal urinary tract infection (UTI) actually had a UTI or if the Candida represented colonization. There was no documentation of the source of the urinary specimens (e.g., clean catch vs. in and out catheterization vs. indwelling bladder catheters). Although patients with cystitis are symptomatic and have abnormal urinalyses, patients with indwelling bladder catheters may have abnormal urinalyses (pyuria) and elevated colony count in the absence of a FI. Furthermore, multiple studies have shown that the presence of candiduria does not frequently increase the risk of developing candidemia (8, 9). The presence of candiduria was associated with an increased risk of mortality, but death was not due to Candida infection and treatment of Candida did not improve outcomes (8, 10). Despite these observations concerning the diagnosis of invasive disease, Bajaj et al. clearly documented adverse patient outcomes in cirrhotic patients with FI. Their data support other studies which found that airway colonization with Candida was associated with worse clinical outcomes and higher rates of death, while the presence of candiduria may reflect the severity of underlying illness (7).
Given the poor rates of survival associated with FI, the clinician must have a high index of suspicion for a FI in critically ill patients and utilize the appropriate diagnostic tools in evaluation and management. Culture allows the laboratory to recover organism(s) from the specimen, isolate, and grow a pure culture from which susceptibility testing can be performed. Although culture remains the gold standard for confirming the diagnosis, alternative techniques are emerging to improve turnaround time and ensure that patients receive the appropriate antifungal therapy. Matrix-assisted laser desorption ionization-time-of-flight mass spectrometry is a methodology that is playing an increasing role in species diagnosis, but it is dependent on a pure culture and the availability of databases that are used to compare the identification generated in the laboratory (11). PCR methods for identifying yeasts are currently being used in a number of blood culture and respiratory panels but are not available for filamentous fungi. PCR is very sensitive and specific but only a limited number of species can be detected (11). Rapid molecular methods such as magnetic resonance are very specific and have good sensitivity direct from the blood draw (without the need for blood cultures), but only a very limited set of Candida can be detected using this technology, which is very expensive (www.t2biosystems.com/t2sepsissolution/t2candidapanel).
Bajaj et al. (4) found that the rate of FI did not vary significantly based on etiology of liver disease. Despite drawing data from a large patient cohort, only 4% of patients had autoimmune hepatitis and were on corticosteroids peri-admission. While 39% of patients with FIs had alcoholic cirrhosis or hepatitis C and alcoholic cirrhosis, the authors did not specify how many of these patients had received corticosteroids prior to their hospitalization. Hmoud et al. (12) found in a meta-analysis that FI were more common in patients with severe alcoholic hepatitis who were treated with prednisolone. As a consequence, we think it would be prudent to consider the increased likelihood of a FI when evaluating patients with liver disease who manifest signs of sepsis that have been treated with corticosteroids.
Antibiotic therapy for bacterial infections appears to influence the risk of subsequently developing a FI. Schwender et al. (13) found that admission antibiotics were the biggest risk factor for the development of intra-abdominal FI after reviewing a series of 479 admitted to their hospital with acute pancreatitis. Changes in the fungal diversity of gut bacteria have been reported in outpatients with cirrhosis who were treated with antibiotics (14). Therefore, the finding that most FI in the NACSLED cohort were preceded by an initial bacterial infection highlights the importance of hospital-based antibiotic stewardship, which is now an expectation of the Joint Commission (www.jointcommission.org/assets/1/6/New_Antimicrobial_Stewardship_Standard.pdf).
Up to the current study, much of the literature concerning FI in liver disease related to their impact on post-liver transplantation (LT) outcomes. In a recent review of antibiotic prophylaxis in cirrhosis, Fernandez et al. (15) reported that invasive post-transplant FI were associated with a case fatality rate of 60%. Targeted vs. universal antifungal prophylaxis of LT recipients has been studied (16, 17). Bajaj et al. (4) identified a population of hospitalized patients with cirrhosis that were at increased risk of developing FI. Targeted surveillance and initiation of antifungal prophylaxis in cirrhotics are both strategies that will require additional study to determine their clinical efficacy and safety, as well as cost-effectiveness.
Defining the appropriate management of hospitalized patient with cirrhosis requires appreciation of the risks of specific complications. The ongoing analysis provided by NACSLED of their large patient cohort, as exemplified by this month’s paper, serves the community by highlighting the impact of FI. Reducing the mortality risk associated with FI will likely require the application of new diagnostic tools to hasten the time to diagnosis along with appropriately targeted prophylaxis therapy in at-risk patient groups.