Fungemia and Other Fungal Infections Associated with Use of Saccharomyces boulardii Probiotic Supplements

Use of these supplements should be considered carefully for patients whose gastrointestinal tract integrity might be compromised.

P robiotics are live microorganisms intended to provide health benefi ts when consumed (1). Typically, the endpoint in randomized controlled trials of probiotics has been the prevention of diarrhea or faster alleviation of diarrhea symptoms (2). Regarding their safety, serious adverse effects have been rare in probiotic studies (3). However, the adverse effects have not been fully reported (4). In 1 trial in which a multispecies probiotic preparation was given to patients who had severe acute pancreatitis, the mortality rate was higher in the probiotic arm (5). Nevertheless, the use of probiotics is common. According to the 2012 National Health Interview Survey in the United States, 1.6% of adults had used prebiotics or probiotics in the preceding 30 days (6).
Saccharomyces cerevisiae var. boulardii is a yeast that is used as a probiotic. In hospitals in the United States, the use of S. cerevisiae var. boulardii has been common, especially among elderly patients (7). This strain is diffi cult to distinguish microbiologically from S. cerevisiae because they have >99% genomic relatedness (8). Thus, in everyday clinical laboratory work, the S. cerevisiae var. boulardii strain is identifi ed as either Saccharomyces sp. or S. cerevisiae. A review from 2005 considered S. cerevisiae var. boulardii to be the etiologic agent of Saccharomyces fungemia if the patient received treatment with a probiotic containing S. cerevisiae var. boulardii or if a molecular typing method confi rmed the identifi cation of this yeast (9). The authors found 37 cases in the literature. We found an additional 14 reports, including 22 cases of Saccharomyces fungemia with the same diagnostic method published after this review (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23). Thus, before our study, 59 cases of fungemia with a link to the use of the probiotic had been published. All of these cases have been either individual cases or small cases series (<7 cases) without any systematic approach to quantify the problem.
Furthermore, besides fungemia, there are few reports on other clinically relevant microbiological fi ndings for this yeast (i.e., in abscesses, ascites fl uid, or the pleural cavity). The meta-analysis we mention listed 20 cases of fi ndings other than fungemia (9). These fi ndings are useful because they might also lead to a change in antimicrobial treatment.
We conducted a retrospective registry study (case series) at 5 university hospitals in Finland to evaluate the use of the S. cerevisiae var. boulardii probiotic in patients who had Saccharomyces fungemia or another clinical culture fi nding for this yeast. To evaluate the association between probiotic use and subsequent fi ndings, we compared use of S. cerevisiae var. boulardii for patients who had a Saccharomyces infection with use of S. cerevisiae var. boulardii for patients who had an infection caused by another etiologic agent, such as bacteria or Candida sp.

Background
Finland has 5 university hospitals that are secondary referral centers of their catchment areas and tertiary referral centers for other hospital districts. Their combined catchment areas cover more than half population of Finland (3.29 million of 5.6 million persons). All university hospitals use the same register (SAIregistry; Neotide Ltd, https://www.neotide.fi ), in which the local clinical microbial laboratory data are collected. These data cover all blood culture data and most of all other clinical microbial culture data of the catchment area of the university hospital.

Patient Data
At least 1 infectious diseases specialist in every university hospital collected the clinical data from the medical records for all blood culture-positive cases found in the register that were identifi ed as Saccharomyces sp. or S. cerevisiae. The use of the S. cerevisiae var. boulardii probiotic was defi ned as use at the time of the positive culture or in the preceding 7 days. Data were collected on use during the preceding 3 months. If the medical record was not available, the case-patient was classifi ed as not using a probiotic. The Quick SOFA score and the defi nition of septic shock were based on the Sepsis-3 defi nitions (24). The McCabe score was determined as reported by McCabe and Jackson (25). Data collected for casepatients who had nonblood cultures were age, sex, Because of widespread use of probiotics, their safety must be guaranteed. We assessed use of Saccharomyces boulardii probiotic yeast from medical records for patients who had Saccharomyces fungemia or other clinical Saccharomyces culture fi ndings. We evaluated all Saccharomyces sp. fi ndings at 5 university hospitals in Finland during 2009-2018. We found 46 patients who had Saccharomyces fungemia; at least 20 (43%) were using S. boulardii probiotic. Compared with a control group that had bacteremia or candidemia, the odds ratio for use of an S. boulardii probiotic was 14 (95% CI 4-44). Of 1,153 nonblood culture fi ndings, the history for 125 patients was checked; at least 24 (19%) were using the probiotic (odds ratio 10, 95% CI 3-32). This study adds to published fungemia cases linked to use of S. boulardii probiotic and sheds light on the scale of nonblood Saccharomyces culture fi ndings that are also linked to use of this probiotic. malignancy, digestive tract disease, use of probiotics, use of antifungal medication at the time of the culture, and possible change of medication resulting from finding of Saccharomyces sp. The most recent 25 cases of nonblood culture findings in each hospital district were evaluated (excluding case-patients who had positive fecal samples). Isolates were obtained from routine laboratory bacterial and fungal cultures. The anatomic site of the culture was collected from the local hospital microbial registry (SAI) for all culture-positive cases. Abdominal sites were those in which culture was taken from, for example, ascites fluid, a biliary drainage catheter, or abscess drainage fluid, but not from skin or wound secretions. Oral and respiratory tract samples were from sinus drainage, bronchial lavage cultures, and pleural drainage. Other sites included samples from perianal abscesses, mediastinum, and urine.

Control Group
To evaluate the practice of probiotic use in the hospital ward in which the patient who had a Saccharomyces finding was given treatment, a control group was obtained from the same SAI register. For every Saccharomyces fungemia case-patient, 2 blood culture-positive patients (1 chronologically closest before and 1 after) from the same ward as the case were selected. For every clinical culture sample (other than blood), there was 1 chronologically closest positive culture sample from the same ward as the case-patient who served as a control. Data collected for the controls were date, ward, microbe, age, sex, malignancy, digestive tract disease, and S. cerevisiae var. boulardii probiotic use at the time of the positive culture or in the previous 3 months.

Statistical Analysis
We used SPSS version 22.0 software (IBM Corp., https://www.ibm.com) for statistical analyses. The study was centrally approved by the Ethics Committee of Tampere University Hospital, Tampere, Finland. The requirement for informed consent was waived.

Blood Cultures
There were 46 patients with a positive blood culture for Saccharomyces in the 5 hospitals during between January 2009-December 2018. The median age of case-patients was 68 (range 30-93) years and a male predominance (63%). The most common underlying condition was a digestive tract disease (59%). There was a medical record confirming the use of the S. cerevisiae var. boulardii probiotic on the day of the blood culture or during the preceding 7 days for 20 case-patients (43%). Medical records were not available for 10 case-patients (22%), and these were classified as nonusers.
Of the 20 case-patients, 17 were using S. cerevisiae var. boulardii probiotic on the day of the blood culture and 3 case-patients had already stopped using it (2 patients on the day before and 1 patient 5 days earlier). Five additional case-patients had used the probiotic in the preceding 3 months, of whom 1 patient had already stopped using it 26 days earlier. For 4 case-patients, the time when the use of the probiotic was terminated could not be determined. Most case-patients (16/20, 80%) received the S. cerevisiae var. boulardii probiotic in the hospital, 3 casepatients in some other facility, and 1 case-patient was using it at home. All S. cerevisiae var. boulardii probiotics found in the medical records were from the same strain (Precosa; Biocodex Ltd., https://www. biocodex.com). We provide characteristics, underlying diseases, and severity of the disease for these patients (Table 1, https://wwwnc.cdc.gov/EID/ article/27/8/21-0018-T1.htm; Table 2).
Antimicrobial drugs were commonly used by the patients (72%) during 4 weeks preceding the fungemia. Antifungal treatment was commenced or changed because of Saccharomyces fungemia for 23 patients (50%). For an additional 8 patients (17%), the culture result came after the patient had died. Casefatality rates by day 7 were 22% (10 patients) and by day 28 were 37% (17 patients). Of patients who died by day 28, 6 patients had an ultimately fatal disease (McCabe score 2) and 5 patients had a rapidly fatal disease (McCabe score 3).

Nonblood Cultures
There were 1,153 nonblood Saccharomyces culture findings (fecal samples excluded). There was considerable variation between hospital districts in numbers of the microbial cultures and anatomic sites from which cultures were obtained (Table 3). We evaluated use of probiotics for 125 case-patients. Medical records were not available for 6 of them. Use of S. cerevisiae var. boulardii probiotic was confirmed for 24 case-patients (19%). This finding was divided by the anatomic site as follows: 17 (21%) of 82 from the abdominal region, 4 (13%) of 30 from the oral or respiratory tract, and 3 (23%) of 13 from other sites. Antifungal medication was already in use at the time of culture for 38% (47/125, the medical record was not available for 1 case-patient) of the case-patients. This finding led to a modification of the antifungal medication in for 35% (44/125, medical records not available for 2 case-patients) of the case-patients.

Controls
The controls for the fungemia case-patients (n = 76) were mostly bacteremic (n = 65), but there were 5 case-patients infected with Candida sp. Medical records were not available for 6 control case-patients. Median age for this group was 70 years (vs. 68 years for case-patients), 70% were males (versus 63% for the case-patients), 47% had digestive tract disease (vs. 59% of the case-patients), and 17% had a malignancy (vs. 13% of the case-patients) (data were available for 64 case-patients). Use of S. cerevisiae var. boulardii probiotic by the Saccharomyces fungemia group was 43% compared with 5% (4/76) for the control group (odds ratio [OR] 14, 95% CI 4-44).
Microbes detected for controls who had nonblood cultures (n = 123) were also mostly bacteria (n = 97), but Candida sp. or other yeasts (n = 51; with or without a concomitant bacterial finding) were more common than in blood cultures. Median age for this group was 65 years (vs. 64 years for case-patients), 44% were males (vs. 59% for case-patients), 70% had digestive tract disease (vs. 69% of case-patients), and 28% had a malignancy (vs. 38% of case-patients) (data on underlying diseases were available for 100 controls). Use of S. cerevisiae var. boulardii probiotic was 19% (24/125) in the Saccharomyces culture-positive group compared with 2% (3/123) for the control group (OR 10, 95% CI 3-32).

Discussion
We report 20 cases of Saccharomyces fungemia in patients who used S. cerevisiae var. boulardii probiotic. These cases have increased the number of cases reported in the literature by 34%.
We evaluated 125 of 1,153 patients who had a nonblood Saccharomyces culture finding and confirmed use of S. cerevisiae var. boulardii probiotic by 19% of these case-patients. To our knowledge, the magnitude of findings other than fungemia has not been reported. Although some of these nonblood findings might represent colonization, positive Saccharomyces cultures led to a change in antimicrobial drugs for 44 (35%) patients who had evaluated cases.
We also evaluated use of S. cerevisiae var. boulardii probiotic for patients who had a Saccharomyces culture  finding and compared it with that of a control group who had different positive blood and nonblood cultures and were in the same ward around the same time. The Saccharomyces fungemia patients had an OR of 14 and nonblood culture-positive patients had an OR of 10 for use of this probiotic compared with respective controls. Moreover, case-patient 7 (Table  1) is an example of a patient in whom probiotic use unequivocally caused the fungemia. The patient had had a percutaneous endoscopic gastrostomy feeding tube inserted 2 days before the fungemia, had septic shock, and then died. The probiotic was administered at least once through the tube, and the tip of the tube was unintentionally displaced from its ventricular position, leading to an accidental peritoneal application of the probiotic. Saccharomyces fungemias occurred most often in patients who had diseases of the gastrointestinal tract (59%). This finding is nearly identical to the amount reported in a meta-analysis (58%) (9). Furthermore, there are reports of Saccharomyces fungemias in patients not given pretreatment with a S. cerevisiae var. boulardii probiotic that have been believed to have been derived from contaminated central venous catheters (26)(27)(28).
Bacteremias and fungemias have not been encountered in clinical trials with probiotics in general. There were probiotic studies conducted with susceptible patients who did not have blood culture findings and who had hepatic encephalopathy (29). However, serious concurrent conditions have usually been an exclusion criterion; thus, the safety profile remains unclear. Furthermore, there are other reported safety issues with probiotics, such as contamination of a probiotic supplement with a pathogenic microbe or possible transfer of an antimicrobial drug resistance gene from a probiotic microbe to pathogenic microbes (30)(31)(32).
Regarding the benefits of probiotics, is there evidence showing that adults should use S. cerevisiae var. boulardii probiotic in conjunction with antimicrobial drugs to prevent Clostridioides difficile infections (CDIs) that cause diarrhea? A meta-analysis during 2017 combined S. cerevisiae var. boulardii probiotic studies conducted in adult populations to prevent CDIs (33). There were 5 studies. All studies had a low number of CDIs (15 cases of CDI in control groups) and all had nonsigficant results (pooled risk ratio 0.63, 95% CI 0.29-1.37).
Currently, several companies sell S. cerevisiae var. boulardii yeast, but total consumption of this yeast in Finland is not known. Thus, we are not able to relate our study results to its use. However, nationwide consumption of probiotics does not reflect the risk for fungemia or nonblood culture findings, or use of this probiotic by susceptible patients in hospitals. Cautious use of S. cerevisiae var. boulardii probiotic in gastrointestinal surgery wards would probably be one of the most effective ways to decrease these culture findings.
Moreover, the benefits for the indication for which the probiotic is used need to be established. There are 2 recent US guidelines on administration of probiotics in general for primary prevention of CDI. The first guideline states that there are insufficient data to recommend the administration (34), and the second guideline states that in certain circumstances certain probiotics could be used, but the quality of evidence is low (35). The first limitation of this study is that we were not able to obtain microbiological evidence that the fungal infections were caused by the S. cerevisiae var. boulardii probiotic strain and not by another strain. Furthermore, the retrospective design using medical records can lead to a bias in reporting. Only confirmed use of probiotics was reported in this study, and casepatients whose medical records were not available were defined as not using a probiotic. Thus, the percentage of probiotic users was the minimum estimate in all groups. All medications given to patients in the wards were documented in medical records of patients, but patients might have used these medications before they were admitted to the hospital. Moreover, most patients who had fungemia were given bacterial antimicrobial drugs, which could have decreased the routine of taking blood cultures. Recall bias (e.g., the S. cerevisiae var. boulardii probiotic would have been mentioned in the medical charts more often in case-patients than in control-patients because of Saccharomyces culture finding) was not a limitation in this study. For all but 1 case-patient, the probiotic was recorded in the charts before the culture result was complete.
S. cerevisiae var. boulardii probiotics are not recommended for patients who have indwelling catheters, are immunocompromised, or are critically ill. Our results indicate that use of S. cerevisiae var. boulardii probiotics should also be considered carefully for patients whose gastrointestinal tract integrity might be compromised. Furthermore, more data are needed to elucidate the health benefits of S. cerevisiae var. boulardii probiotics in general .