It is well-known that patients with acute myeloid leukemia (AML) who receive induction and reinduction chemotherapy and any patients undergoing hematopoietic stem-cell transplant (HSCT) are at high risk for infections and prolonged hospital stays. Intensive chemotherapy regimens used in these settings cause patients to be neutropenic for prolonged durations. Despite preventive measures, patients often become febrile and some have severe, life-threatening infections. Currently, the standard for preventing infection-related complications in this patient population is the use of prophylactic antibiotics, antifungals, and antivirals. In addition, some providers use probiotics with the goal of maintaining or restoring the normal intestinal flora damaged by chemotherapy.
The use of probiotic prophylaxis in susceptible patients has been widely thought of as a simple, inexpensive solution to some infectious problems.1 However, studies indicate mixed results on the effectiveness of probiotics in preventing infection.1-9 Probiotics have proved effective in the treatment of many diarrheal illnesses, such as antibiotic-associated diarrhea and Clostridium difficile infection.2,3 One study has also demonstrated that probiotics prevent candidemia and candiduria in the pediatric intensive care unit (ICU) setting.4
Alternatively, studies and case reports have shown possible adverse effects associated with probiotic use, including bacterial sepsis, fungal sepsis, and probiotic sepsis.1 Risk factors for these complications include a compromised immune system, the use of a central venous catheter, and an impaired intestinal epithelial barrier—all of which are frequently present in patients with AML and patients undergoing HSCT who are receiving chemotherapy.1
A wide variety of demographic and clinical factors, including age, ethnicity, the strain of probiotic used, and the patient’s normal flora, may also affect the effects demonstrated in these studies.2-9 Probiotic effectiveness has not been studied in patients with AML or patients undergoing HSCT,1,5 and neither efficacy nor safety can be easily generalized to these patient populations because of the complexity of their disease states and the severity of their immunosuppression.
Issues related to the use of probiotics in these populations include the presence of prolonged neutropenia and mucositis, which can impair the epithelial barrier in the gastrointestinal tract. The latter 2 are risk factors for probiotic adverse effects, because they increase patients’ risks for infection-related complications, as well as increase the risk for probiotic-related sepsis.
The purpose of our study was to determine the effectiveness of prophylactic probiotics in this immunocompromised population of patients with AML and patients undergoing transplant.
This retrospective, single-center study was approved by the local Institutional Review Board. Adult patients admitted to a community teaching hospital in Indianapolis, IN, with a diagnosis of AML who received induction or reinduction chemotherapy between January 2008 and January 2015, and patients who were undergoing HSCT during that period, were included in this study. Patients who started probiotics more than 7 days after chemotherapy initiation were excluded from the study to avoid inclusion of patients who received probiotics as an adjunct to infection-related treatment rather than as prophylaxis.
Demographic and clinical characteristics were collected from the electronic health records of all eligible patients, and included age, sex, height, weight, diagnosis, length of stay, and chemotherapy regimen. In addition, probiotic drug, probiotic start day in relation to chemotherapy, prophylactic antibiotic use, antibiotic use and duration, receipt of gastric acid–suppressing drugs, and use of a granulocyte colony-stimulating factor (G-CSF) during the hospital stay were also identified.
During the study, filgrastim was the standard G-CSF used for patients receiving select regimens (ie, fludarabine phosphate, cytarabine, and filgrastim, with or without idarubicin) for AML in the relapsed setting, as well as for those undergoing HSCT.
The standard for the institution was not to use growth factors routinely during induction therapy for AML unless the patient had severely prolonged neutropenia and a life-threatening infection that could benefit from quicker recovery of the neutrophil count. The event parameters collected for the study included documented mucositis, febrile neutropenia, time to first fever, documented infection, type of infection, C difficile infection, and 30-day readmission for an infectious issue.
The primary outcome measure was the incidence of febrile neutropenia, defined as a temperature ≥100.5°F and an absolute neutrophil count ≤500/mm3. Secondary outcomes were the incidence of C difficile infection (determined by a positive stool culture), time to first fever (described as the number of days from start of chemotherapy to first temperature ≥100.5°F), time to C difficile infection, incidence of documented infection (ie, positive culture results), and 30-day readmission for an infectious issue.
Categorical data were compared between groups using Fisher’s exact test or Pearson’s chi-square test, as appropriate. Continuous variables between groups were compared using the t-test for parametric continuous data and the Mann-Whitney U test for nonparametric continuous data. Binary logistic regression was completed to determine the odds ratio (OR) of the significant predictors of febrile neutropenia (ie, G-CSF use, probiotic use, mucositis, and duration of neutropenia) or C difficile infection (ie, probiotic use, mucositis, duration of neutropenia, antibiotic duration, and proton pump inhibitor use). Time to first fever was analyzed using the Kaplan-Meier survival analysis. These analyses were performed using SPSS Statistics Software (Version 22.0; IBM, Armonk, NY). A P value <.05 was considered significant.
A total of 197 patients were identified; however, 22 patients were excluded because probiotics were initiated >7 days after starting chemotherapy. Therefore, 175 patients were included in the analysis. Of these patients, 29 received probiotics and 146 did not. Baseline characteristics are provided in Table 1.
The only significant difference between the 2 groups was in hospital length of stay, which was longer in the probiotic group (33 days vs 26 days; P = .018).
Of the 29 patients who received probiotics, 27 (93%) received Lactobacillus acidophilus and 2 (7%) received Saccharomyce s boulardii as probiotics (Table 2).
Dosing for L acidophilus ranged from 1 to 2 tablets taken 2 to 3 times daily, and S boulardii was administered as 1 capsule once or twice daily. Gastric acid suppressants were heavily used in the probiotic and nonprobiotic groups (86% vs 75%, respectively; P = .203; Table 1).
Febrile neutropenia was common in both groups, affecting 79% of the probiotic group and 71% of the nonprobiotic group (P = .337). There was a significant difference between the probiotic and nonprobiotic groups in documented infections (48% vs 29%, respectively; P = .04), specifically bloodstream infections (45% vs 21%, respectively; P = .007). The incidence of other infections (ie, urinary tract and C difficile) was similar between the probiotic and nonprobiotic groups (Table 3).
Overall, 30-day readmissions occurred less frequently in patients receiving probiotics during hospitalization than in those not receiving probiotics (28% vs 44%, respectively; P = .104), as did 30-day readmissions for an infectious issue (0% vs 7%, respectively; P = .147); however, these differences were nonsignificant in both cases (Table 3). Time to C difficile was not significantly longer in the probiotic group than in the nonprobiotic group (16 days vs 6.44 days, respectively; P = .120; Table 3).
A logistic regression model, including probiotic use, presence of mucositis, use of G-CSF, and duration of neutropenia, demonstrated that mucositis (OR, 0.435; 95% confidence interval [CI], 0.203-0.933; P = .033) and duration of neutropenia (OR, 1.043; 95% CI, 1.01-1.077; P = .011) were both significant predictors of febrile neutropenia (Table 4).
Similarly, mucositis (OR, 6.426; 95% CI, 1.105-37.359; P = .038), neutropenic duration (OR, 1.129; 95% CI, 1.009-1.263; P = .034), and antibiotic duration (OR, 0.91; 95% CI, 0.841-0.985; P = .019) were significant predictors of C difficile infection (Table 5).
The median time to fever via Kaplan-Meier survival analysis was 10 days in those receiving a probiotic, and 12 days in those not receiving a probiotic (P = .244; Figure).
As a result of the increasing use of probiotics in the medical and the consumer settings, it is important to ensure their safety and efficacy. Although research on probiotics use in various settings is limited, previous findings cannot be generalized because of the heterogeneity of those results. Determining the patient population for which probiotics are useful is imperative to help provide the best quality of care, as well as prevent any harm.
Probiotic effectiveness has been widely studied in patients at risk for antibiotic-associated diarrhea and C difficile infection.2 A meta-analysis of randomized controlled trials spanning from 1946 to 2012 determined that probiotics used concurrently with antibiotics reduced antibiotic-associated diarrhea by approximately 40% and C difficile infection by approximately 60%.2
Our current study of the use of probiotics in a highly immunosuppressed population, however, did not show the same level of improvement as shown in these studies. The reasons for the disparity between our findings and previously published data could be related to the immunosuppressed nature of the population we studied and the effects of chemotherapy on the gastrointestinal system. In addition, our results show an increased length of hospital stay in the probiotic group, which could increase patients’ risk for C difficile infection.
In the pediatric ICU population, a retrospective, comparative study of probiotics administered concomitantly with antibiotics demonstrated a reduction in candidemia infection by approximately 70% and candiduria infection by approximately 50% (common variants of Candida infection) in those receiving probiotics.4 Candida infection was not an outcome of our study because of the infrequency of candidemia and candiduria infections, likely caused by the use of antifungal prophylaxis; however, 3 of the 175 patients had candidemia, and none of them received probiotics.
Alternatively, in the adult ICU setting, the use of probiotics has also been shown to be ineffective.9 In that study, probiotics did not reduce the duration of mechanical ventilation, length of hospital stay, or mortality in this population.9 This demonstrates, once again, why studies of probiotics cannot be generalized to other patient populations.
Other patient populations, such as patients with ventilator-associated pneumonia, have shown mixed results in studies using probiotics. One meta-analysis indicated that probiotics had no effect on ventilator-associated pneumonia, length of hospital stay, or mortality.7 However, this analysis was limited by great heterogeneity of the included studies. A more homogeneous meta-analysis came to the same conclusions, showing that probiotics had no benefit in decreasing the duration of mechanical ventilation, diarrhea, or incidence of urinary tract infection.8 Contrary to these results, the previously discussed meta-analysis proposed that probiotics were beneficial in preventing ICU-acquired pneumonia, decreasing the odds by 40%,9 suggesting that probiotic effectiveness may vary even in the same patient population.
Our study demonstrated that documented infection was almost twice more likely to occur in patients using prophylactic probiotics than in those not taking probiotics. Bloodstream infection was increased more than any other infection, occurring in more than twice as many patients receiving probiotics than in those not receiving probiotics.
It is important to note that none of the patients receiving probiotics were infected with L acidophilus; however, 1 patient had Saccharomyce s cerevisiae fungemia. The patient was a 71-year-old woman with relapsed AML who received reinduction chemotherapy with cytarabine and clofarabine. She received S boulardii as a probiotic 3 times daily starting 3 days after hospital admission and until discharge 33 days later. This patient’s S cerevisiae–positive blood culture was obtained 6 days before discharge. She was discharged to hospice care because of the presence of persistent AML.
Although S boulardii and S cerevisiae were once believed to be 2 distinct strains, it is now known that S boulardii is a member of the S cerevisiae strain.10,11 Moreover, S cerevisiae fungemia has occurred in patients receiving S boulardii probiotics.12
It is not known why patients receiving probiotics in our study had a higher risk for infection, because baseline characteristics and risk factors for infections were similar between the 2 groups. One disparity that could potentially explain the increased risk for infection and bloodstream infections was the prolonged length of hospital stay in those receiving probiotics compared with those not receiving probiotics; however, in our study, the duration of neutropenia was similar between the 2 groups (16 days vs 17.5 days, respectively).
Although not a significant difference, patients in the group without probiotics in our study were approximately 50% more likely to be readmitted within 30 days for an infectious issue, and had a delayed time to onset of C difficile infection, signifying that probiotics may have some effects in this patient population.
Probiotic administration after hospital discharge was not captured in this data set and could account for this decrease in readmission among patients taking probiotics. As for the nonsignificant delay in C difficile infection, this could be caused by the effects of the probiotics; however, there was a low incidence of C difficile in this study.
Although not one of the main focuses of our study, we found that significant predictors for febrile neutropenia were mucositis (OR, 0.435; 95% CI, 0.203-0.933; P = .033) and neutropenic days (OR, 1.043; 95% CI, 1.01-1.077; P = .011). As expected, patients who had neutropenia for a longer duration were more likely to become febrile. Conversely, mucositis may be a protective factor. Although this is contrary to what is known about mucositis, it could be that patients without mucositis were at a higher risk for infection than those without mucositis for reasons other than those captured in our study.
Conversely, mucositis was shown to increase the risk for C difficile infection (OR, 6.426; 95% CI, 1.105-37.359; P = .038) in our study. This risk has been documented in previous studies, including one study in the autologous HSCT setting, demonstrating that mucositis was the strongest predictor for C difficile infection.13
Another compelling finding in our study was the effect of antibiotic duration (OR, 0.91; 95% CI, 0.841-0.985; P = .019) on the incidence of C difficile infection. The protective effect could have been caused by the utilization pattern of prophylactic fluoroquinolones. Patients receiving shorter durations of antibiotic treatment would have longer exposure to fluoroquinolones, a known risk factor for C difficile infection. However, patients were treated with broad-spectrum antibiotics, including fourth-generation cephalosporins and antipseudomonal penicillins, which also increase the risk for
C difficile infection.
The retrospective design of our study posed some limitations, including the difference in sample size between the groups. Physician interpretation and documentation of adverse events is also a concern, given the retrospective design of the study.
A baseline characteristic that might have affected the study outcomes was increased length of hospital stay in the probiotic group (33 days vs 26 days; P = .018).
In addition, prescribers might have been more inclined to use probiotics in patients at higher risk for infection. Specifically, because of the retrospective nature of our study, there could have been a propensity for prescribers to use probiotics in patients who were at higher risk for C difficile infection.
Many studies have hypothesized that probiotics could help prevent the occurrence of infection in a multitude of settings. Ideally, the use of probiotics would provide an inexpensive, readily available option that could reduce costs and potentially improve morbidity and mortality rates in the immunocompromised setting.
Although previously shown to help prevent some diarrheal illnesses,2,3 probiotics in patients with AML and patients undergoing HSCT in our study did not demonstrate beneficial effects on the rates of febrile neutropenia or C difficile infection. Patients receiving probiotics did have a higher incidence of documented infection, specifically bloodstream infections. This likely indicates that probiotics are not beneficial in patients at risk for prolonged neutropenia. Future prospective studies are needed to definitively test the efficacy of probiotics in immunosuppressed patients at high risk for infection.
Author Disclosure Statement
Mr Przybylski and Dr Reeves have no conflicts of interest to report.
1. Boyle RJ, Robins-Browne RM, Tang ML. Probiotic use in clinical practice: what are the risks? Am J Clin Nutr. 2006;83:1256-1264.
2. Pattani R, Palda VA, Hwang SW, Shah PS. Probiotics for the prevention of antibiotic-associated diarrhea and Clostridium difficile infection among hospitalized patients: systematic review and meta-analysis. Open Med. 2013;7:e56-e67.
3. Kotzampassi K, Giamarellos-Bourboulis EJ. Probiotics for infectious diseases: more drugs, less dietary supplementation. Int J Antimicrob Agents. 2012;40:288-296.
4. Kumar S, Singhi S, Chakrabarti A, et al. Probiotic use and prevalence of candidemia and candiduria in a PICU. Pediatr Crit Care Med. 2013;14:409-416.
5. Lee AC, Siao-Ping Ong ND. Food-borne bacteremic illnesses in febrile neutropenic children. Hematol Rep. 2011;3:32-33.
6. Sikorska H, Smoragiewicz W. Role of probiotics in the prevention and treatment of meticillin-resistant Staphylococcus aureus infections. Int J Antimicrob Agents. 2013;42:475-481.
7. Wang J, Liu KX, Ariani F, et al. Probiotics for preventing ventilator-associated pneumonia: a systematic review and meta-analysis of high-quality randomized controlled trials. PLoS One. 2013;8:e83934.
8. Gu WJ, Wei CY, Yin RX. Lack of efficacy of probiotics in preventing ventilator-associated pneumonia. Chest. 2012;142:859-871.
9. Barraud D, Bollaert PE, Gibot S. Impact of the administration of probiotics on mortality in critically ill adult patients: a meta-analysis of randomized controlled trials. Chest. 2013;143:646-655.
10. van der Aa Kühle A, Jespersen L. The taxonomic position of Saccharomyce s boulardii as evaluated by sequence analysis of the D1/D2 domain of 26S rDNA, the ITS1-5.8S rDNA-ITS2 region and the mitochondrial cytochrome-c oxidase II gene. Syst Appl Microbiol. 2003;26:564-571.
11. Mitterdorfer G, Mayer HK, Kneifel W, Viernstein H. Protein fingerprinting of Saccharomyces isolates with therapeutic relevance using one- and two-dimensional electrophoresis. Proteomics. 2002;2:1532-1538.
12. Herbrecht R, Nivoix Y. Saccharomyce s cerevisiae fungemia: an adverse effect of Saccharomyce s boulardii probiotic administration. Clin Infect Dis. 2005;40:1635-1637.
13. Alonso CD, Dufresne SF, Hanna DB, et al. Clostridium difficile infection after autologous stem cell transplantation: a multicenter study of epidemiology and risk factors. Biol Blood and Marrow Transplant. 2013;19:1502-1508.