European Journal of Clinical Microbiology and Infectious Diseases

, Volume 24, Issue 1, pp 31–40

Pathogenic relevance of Lactobacillus: a retrospective review of over 200 cases

Authors

  • J. P. Cannon
    • Edward Hines, Jr. VA HospitalPharmacy Service
  • T. A. Lee
    • Midwest Center for Health Services and Policy ResearchEdward Hines, Jr. VA Hospital
    • Center for Healthcare Studies, Division of General Internal Medicine, Department of MedicineNorthwestern University Feinberg School of Medicine
    • Center for Pharmacoeconomic Research, College of Pharmacy, 833 South Wood Street (Mail Code 886)University of Illinois at Chicago
  • J. T. Bolanos
    • Department of Medicine, Section of Infectious DiseasesUniversity of Illinois at Chicago
    • Department of Pharmacy Practice, College of Pharmacy, 833 South Wood Street (Mail Code 886)University of Illinois at Chicago
Article

DOI: 10.1007/s10096-004-1253-y

Cite this article as:
Cannon, J.P., Lee, T.A., Bolanos, J.T. et al. Eur J Clin Microbiol Infect Dis (2005) 24: 31. doi:10.1007/s10096-004-1253-y

Abstract

Given that Lactobacillus has been reported to be the causative pathogen in many types of infection despite debate regarding the organism’s clinical significance, a literature review was conducted to investigate the treatments and outcomes of Lactobacillus infections reported to date. In this article, the characteristics of over 200 reported cases of Lactobacillus-associated infections are summarized. Lactobacillus was found to be frequently associated with endocarditis and bacteremia. Lactobacillus was also associated with a variety of other infections including, but not limited to, peritonitis, abscesses, and meningitis. The species casei and rhamnosus were the most common. The isolates tended to be most sensitive to erythromycin and clindamycin and most resistant to vancomycin. The species that was most sensitive to vancomycin was acidophilus. The overall mortality rate was nearly 30%. There was a significant association between mortality and polymicrobial infection (P=0.004). In the subset of patients with bacteremia, increased mortality was associated with inadequate treatment (P=0.001) and polymicrobial bacteremia (P=0.044).

Introduction

Lactobacillus, a gram-positive rod-shaped bacterium [1], is a common inhabitant of the human mouth, gastrointestinal tract, and female genital tract [26]. Some species of Lactobacillus are utilized as probiotic bacteria, either in lyophilized form or as a fermented food product, and have been shown to be effective in the treatment of infantile and adult diarrhea, antibiotic-associated diarrhea, and candidal vaginitis [7, 8].

The clinical significance of Lactobacillus isolated from normally sterile sites is a subject of ongoing debate. Some believe this organism should never be dismissed as a contaminant [9]. Others comment that “not all positive blood cultures for lactobacilli are clinically significant, and lactobacilli are occasional contaminants” [10]. Debate also surrounds the safety of Lactobacillus used as a probiotic and its link to infections, specifically in immunocompromised patients [7, 1115]. Despite these opposing views, Lactobacillus has been implicated in various types of infections. In the review presented here, we analyzed the characteristics of over 200 cases of Lactobacillus infection associated with bacteremia, endocarditis, and localized infection.

Materials and methods

Case reports of Lactobacillus-associated infections reported between 1950 and 1 July 2003 were identified through a Medline search. The search term “Lactobacillus” was used and the search was limited to the English language. Additional cases were identified from the references of the case reports. An attempt was made to obtain the original publication of each case. In situations in which the original publication was either unavailable via interlibrary loan or written in a non-English language, information involving the case was based on information in the article(s) that referenced the case report.

The following data elements were extracted from each of the cases and cases were included if they contained three or more of the elements: patient age, patient gender, patient comorbidity, type of Lactobacillus infection, source of Lactobacillus isolate, species of Lactobacillus recovered, antimicrobial sensitivity of the Lactobacillus isolate, concomitant organisms identified, treatment regimen and duration, and overall mortality. Adequacy of antibiotic therapy was determined for cases in which both antibiotic treatment and in vitro sensitivity data were available. Therapy was considered adequate if antibiotic therapy was chosen in accordance with the in vitro sensitivity data. Microbiologic methods used to identify the Lactobacillus organism in each of the cases were not evaluated. Additionally, the distinction between community-acquired infection and nosocomial infection was not investigated for each case.

The identified cases were organized into three mutually exclusive categories of infection: (i) bacteremia; (ii) endocarditis; and (iii) localized infections. Each case was assigned to one of the categories and comparisons were made between the three types of infection.

Statistical analysis

The characteristics of patients in each of the three infection categories were determined by calculating averages for continuous variables and calculating frequencies and percentages for categorical variables. When comparing the three groups, continuous variables were evaluated with an ANOVA and categorical variables were evaluated using a contingency table approach with the chi-square and Fisher’s exact tests. Statistical tests were used to look for associations or differences between the site of the infection and various characteristics, rather than testing for pair-wise differences between infection sites. Specifically, we compared the average age of patients in the three groups of infection types; we also determined if differences existed among the groups in terms of patient gender, species of Lactobacillus, presence of a polymicrobial infection, antibiotic sensitivity, and treatment regimen. Finally, we evaluated the association between mortality and several factors including the type of infection, adequacy of treatment, presence of a polymicrobial infection, and species. Additional analyses looking at the outcome of mortality were conducted in patients with bacteremia since this group had the largest sample size of the three infection types, which allowed for additional statistical comparisons.

Results

We reviewed a total of 92 manuscripts in their entirety. From these manuscripts, 241 cases of Lactobacillus infection associated with bacteremia (129 cases), endocarditis (73 cases), and localized infection (39 cases) were identified [26, 10, 16101]. The characteristics of the patients and their infections are listed in Table 1. Overall, the majority of the patients were male (53.1%). The average age was 50.1 years with a range of 1 day–90 years.
Table 1

Patient characteristics and species isolated overall and by type of infection

Characteristic

All cases (n=241)

Bacteremia (n=129)

Endocarditis (n=73)

Localized (n=39)

P value

Age (years)

  Mean

50.1

53.5

43.2

49.3

0.007b

  Median

52

57

43.5

52

  Range

<1–90

<1–90

6–80

<1–79

Sex, n (%)

  Male

128 (53.1)

73 (56.6)

35 (48.0)

20 (51.3)

0.892c

  Female

86 (35.7)

51 (39.5)

21 (28.8)

14 (35.9)

  Unknown

27 (11.2)

5 (3.9)

17 (23.3)

5 (12.8)

Speciesa, n (%)

  L. acidophilus

14 (10.0)

5 (9.4)

6 (9.4)

3 (13.0)

0.101d

  L. brevis

1 (0.7)

1 (1.9)

  L. casei

50 (35.7)

15 (28.3)

26 (40.6)

9 (39.1)

  L. confuses

3 (2.1)

1 (1.9)

2 (8.7)

  L. curvantus

3 (2.1)

2 (3.8)

1 (1.6)

  L. delbrueckii

1 (0.7)

1 (1.9)

  L. fermentum

4 (2.9)

1 (1.9)

1 (1.6)

2 (8.7)

  L. gasseri

1 (0.7)

1 (1.9)

  L. jensenii

3 (2.1)

1 (1.9)

2 (3.1)

  L. lactis

1 (0.7)

1 (4.4)

  L. leichmannii

2 (1.4)

2 (3.8)

  L. paracasei

10 (7.1)

4 (7.6)

5 (7.8)

1 (4.4)

  L. plantarum

14 (10.0)

2 (3.8)

11 (17.2)

1 (4.4)

  L. rhamnosus

32 (22.9)

17 (32.1)

11 (17.2)

4 (17.4)

  L. salvarius

1 (0.7)

1 (1.6)

Polymicrobial, n (%)

  Yes

69 (28.6)

50 (38.8)

3 (4.1)

16 (41.0)

<0.001d

  No

172 (71.4)

79 (61.2)

70 (95.9)

23 (59.0)

aData for cases in which the species was reported: all cases, n=140; bacteremia, n=53; endocarditis, n=64; localized, n=23

bP value calculated using ANOVA comparing means of the three types of infection; where H0 is, the average age is the same in all three infection sites

cP value calculated using the chi-square test comparing the proportion in each category across the three types of infection; where H0 is, there is no association between gender and infection site (unknowns excluded)

dP value calculated using Fisher’s exact test comparing the proportion in each category across the three types of infection; where H0 is, no relationship exists between infection site and species or infection site and the presence of polymicrobial infection

The Lactobacillus isolate was identified to the species level in 140 of the cases. The breakdown by species is also reported in Table 1, both overall and for each type of infection (i.e., bacteremia, endocarditis, and localized infection). The type of species was similar for each type of infection (P=0.101). The most common species were casei, rhamnosus, and plantarum. The infection was polymicrobial in 28.6% of the cases, and a greater proportion of the bacteremia and localized infection cases were polymicrobial than the cases of endocarditis (P<0.001).

Antimicrobial sensitivity data is listed in Table 2. Overall, Lactobacillus tended to be the most sensitive to erythromycin and clindamycin (94.3% and 90.9% sensitive, respectively). There were differences in antimicrobial sensitivity according to the site of infection, with bacteremic infections being less sensitive to ciprofloxacin (P=0.010), while endocarditis infections were less sensitive to gentamicin (P=0.002). Resistance to vancomycin was high with only 22.5% of the isolates sensitive. The isolates sensitive to vancomycin were either acidophilus or not speciated. Of the eight acidophilus isolates tested for vancomycin sensitivity, 75% were sensitive.
Table 2

Antibiotic sensitivity of Lactobacillus overall and by type of infection

Antibiotic agent

Percent sensitive (no. of cases)a

All cases

Bacteremia

Endocarditis

Localized

P valueb

Ampicillin

63.6% (n=44)

60.0% (n=25)

84.6% (n=13)

33.3% (n=6)

0.060

Cefazolin

47.8% (n=23)

50.0% (n=14)

75.0% (n=4)

20.0% (n=5)

0.309

Ciprofloxacin

64.3% (n=28)

35.7% (n=14)

100.0% (n=4)

90.0% (n=10)

0.010

Clindamycin

90.9% (n=55)

94.1% (n=34)

80.0% (n=10)

90.9% (n=11)

0.268

Erythromycin

94.3% (n=53)

96.7% (n=30)

90.9% (n=11)

91.7% (n=12)

0.400

Gentamicin

70.0% (n=40)

94.4% (n=18)

36.4% (n=11)

63.6% (n=11)

0.002

Penicillin

54.9% (n=71)

54.3% (n=35)

65.0% (n=20)

43.8% (n=16)

0.184

Vancomycin

22.5% (n=80)

26.7% (n=45)

26.7% (n=15)

10.0% (n=2)

0.333

aPercentages indicate the percent of cases in which the isolate was sensitive to the antibiotic tested; numbers represent the total number of cases in which the isolate was tested against the specific antibiotic

bFisher’s exact test comparing proportions across the three types of infection; where H0 is, there is no association between the site of the infection and sensitivity to the antibiotic

Patients were treated with a variety of antibiotic regimens for various durations (Table 3). The most common regimens included a penicillin antibiotic as monotherapy (n=35), a combination of a penicillin antibiotic with an aminoglycoside (n=20), and a cephalosporin as monotherapy (n=16). Based on in vitro sensitivity data, antibiotic therapy was deemed adequate in 69 of 93 (74%) cases. The average duration of antibiotic therapy was 25 days, with a median of 14 days.
Table 3

Treatment regimens overall and by type of infection

Treatment

All cases (n=241)

Bacteremia (n=129)

Endocarditis (n=73)

Localized (n=39)

Monotherapy, n (%)

81 (33.6)

54 (41.9)

15 (20.5)

12 (30.8)

  pcn

35

24

10

1

  ceph

16

10

2

4

  carba

3

2

0

1

  clinda

8

5

0

3

  quin

3

3

0

0

  ag

2

2

0

0

  vanco

5

4

0

1

  macro

2

1

1

0

  sulfa

3

1

2

0

  tetracycline

2

1

0

1

  chloramphenicol

1

1

0

0

  lincomycin

1

0

0

1

Dual therapy, n (%)

45 (18.7)

25 (19.4)

15 (20.5)

5 (12.8)

  pcn + ag

20

2

15

3

  ceph + ag

1

1

0

0

  carba + ag

1

1

0

0

  vanco + ag

2

2

0

0

  macro + ag

1

1

0

0

  clinda + ag

3

3

0

0

  pcn + vanco

1

1

0

0

  ceph + vanco

6

6

0

0

  ceph + quin

1

1

0

0

  carba + macro

4

3

0

1

  carba + clinda

1

1

0

0

  pcn + pcn

1

1

0

0

  quin + macro

1

1

0

0

  vanco + quin

1

1

0

0

  rif + ag

1

0

0

1

Triple therapy, n (%)

6 (2.5)

2 (1.5)

4 (5.5)

0

Multiple coursesa, n (%)

54 (22.4)

27 (20.9)

19 (26.0)

8 (20.5)

No therapy, n (%)

10 (4.1)

6 (4.7)

3 (4.0)

1 (2.6)

Unknown therapy, n (%)

45 (18.7)

15 (11.6)

17 (23.0)

13 (33.3)

Adequacy of therapy assessedb

(n=93)

(n=46)

(n=27)

(n=20)

  Adequate, n (%)

69 (74)

35 (76)

19 (70)

15 (75)

  Not adequate, n (%)

24 (26)

11 (24)

8 (30)

5 (25)

Duration of therapy (days)

(n=136)

(n=79)

(n=43)

(n=14)

  Mean

25

14

49

18

  Median

14

10

42

14

  Range

1–160

1–111

10–160

3–63

ag, aminoglycoside; carba, carbapenem; ceph, cephalosporin; clinda, clindamycin; macro, macrolide; pcn, penicillin; quin, fluoroquinolone; rif, rifampin; vanco, vancomycin

aPatients who received sequential antibiotic therapy with various classes of antibiotics

bAdequacy assessed according to in vitro sensitivity results

In cases for which patient outcome was reported, the overall mortality rate was 29.1%. In the overall analysis, mortality was not associated with the type of infection (P=0.418) (Table 4). Of those patients who were adequately treated, 13.2% died, while of those who were inadequately treated, 31.8% died. This difference between adequate and inadequate treatment approached, but did not attain, statistical significance (P=0.053). There was, however, a significant association between mortality and the presence of a polymicrobial infection. A total of 43.5% of patients with a polymicrobial infection died versus 23.0% of patients without a polymicrobial infection (P=0.004).
Table 4

Outcome overall and by site of infection in relation to various factors

Associated factor

Outcome

Survived

Died

P valuea

Type of infection, n (%)

  Bacteremia

77 (67.5)

37 (32.5)

0.418

  Endocarditis

47 (77.1)

14 (22.9)

 

  Localized

25 (71.4)

10 (28.6)

 

Adequately treatedb, n (%)

  No

15 (68.2)

7 (31.8)

0.053

  Yes

59 (86.8)

9 (13.2)

 

Treatment received, n (%)

  No

4 (40.0)

6 (60.0)

0.066

  Yes

145 (72.5)

55 (27.5)

 

Polymicrobial infection, n (%)

  No

114 (77.0)

34 (23.0)

0.004

  Yes

35 (56.5)

27 (43.5)

 

Speciesc, n (%)

  L. acidophilus

10 (71.4)

4 (28.6)

0.416

  L. casei

31 (67.4)

15 (32.6)

 

  L. plantarum

9 (69.2)

4 (30.8)

 

  L. rhamnosus

19 (86.4)

3 (13.6)

 

aP values are for tests of association between variables listed in the rows and mortality; where the H0 is, there is no association between the characteristics in the rows and mortality

bAdequacy assessed according to in vitro sensitivity results

cRestricted to the four most prevalent species

Lactobacillus bacteremia

One-hundred and twenty-nine cases of bacteremia were identified [46, 10, 1650]. The characteristics of those patients and their infections are listed in Table 1. L. rhamnosus and L. casei were the most common species isolated. Other organisms in addition to a Lactobacillus sp. were isolated from the blood in 50 (38.8%) of the cases. The most common concomitant organisms isolated from the blood were Candida spp. (n=13 cases), streptococci (n=13 cases), and enterococci (n=9 cases). Patients had various underlying conditions. The most common included cancer (n=31), diabetes (n=21), broad-spectrum antibiotic therapy (n=19), transplantation (n=18), and abscesses (n=15). The most common type of cancer was leukemia (n=8). Types of transplant included liver (n=9), renal (n=3), lung (n=3), heart (n=2), and bone marrow (n=1).

Decreased sensitivity (≤50% of isolates sensitive) to cefazolin, ciprofloxacin, and vancomycin was noted for these cases (Table 2). The treatment regimens were reported for 108 cases (Table 3). In 54 of the cases, patients were treated with monotherapy, typically a beta-lactam antibiotic (n=36). In 24 of the cases, patients were treated with dual therapy, typically a cephalosporin antibiotic plus vancomycin (n=6) or a beta-lactam antibiotic plus an aminoglycoside (n=4). Six patients were not treated with any antibiotic therapy. The average duration of therapy was 14 days. In the 46 cases in which adequacy of therapy could be assessed, 76% of the patients were treated appropriately.

Outcomes were reported for 114 of the cases. In 67.5% of the cases, the patient recovered. Among these patients, increased mortality was associated with inadequate treatment and polymicrobial infection. None of the patients who were treated adequately died, but four of the ten inadequately treated patients died during follow-up (P=0.001). A total of 43.5% of the patients with a polymicrobial infection died, whereas 25.0% of the patients without a polymicrobial infection died (P=0.044). Other factors such as receipt of treatment, species, and underlying condition were not associated with any differences in mortality.

Lactobacillus endocarditis

The demographics of the 73 cases of endocarditis are listed in Table 1 [24, 10, 1618, 5081]. Of these patients, 63% had an underlying structural heart disease and 12% reported a previous episode of endocarditis. A dental procedure or a dental condition was a possible predisposing cause in 47% of the cases. In three of the cases “heavy dairy consumption” was noted as a possible risk factor. In 26% of the cases, systemic emboli were documented. The most common species identified was casei, followed by rhamnosus and plantarum.

The isolates were most sensitive to ciprofloxacin, erythromycin, and ampicillin (Table 2). Decreased sensitivity (≤50%) to gentamicin and vancomycin was found. The most common treatment regimens included a combination of a penicillin antibiotic and an aminoglycoside (n=15) and a penicillin antibiotic as monotherapy (n=10) (Table 3). Of the 27 cases in which adequacy of therapy could be assessed, therapy was determined to be adequate in 70%. Average duration of therapy was 49 days, with a median of 42 days.

Outcomes included a valve replacement procedure in 14 cases and a relapse of endocarditis in seven cases. Overall mortality was nearly 23%. Differences in mortality were not statistically significantly between patients who were treated adequately versus those who were not (P=0.99).

Localized Lactobacillus infections

There were 39 reported cases of localized Lactobacillus infection [19, 21, 16, 44, 81102]. The patient characteristics are listed in Table 1. Of the 39 patients, 15 had a Lactobacillus infection related to a pulmonary infection, eight had an abscess, four had peritonitis, two had chorioamnionitis, and two had an intra-abdominal infection. There was one case each of endophthalmitis, esophageal infection, erysipeloid infection, throat infection, meningitis, wound infection, vascular graft infection, and fistula. Common underlying factors included diabetes (n=7), renal failure (n=5), cancer (n=4), and neutropenia (n=4). One patient who presented with a liver abscess reported heavy consumption of dairy drinks containing L. rhamnosus GG in the months prior to admission.

In 23 of the cases, the Lactobacillus organism was speciated. A single species could not be associated with a specific type of infection. The most common species identified was casei (Table 1). In 16 cases, other organisms in addition to Lactobacillus were recovered. These organisms included Candida spp. (n=5), Klebsiella spp. (n=4), and Enterococcus spp. (n=3).

Isolates were most sensitive to ciprofloxacin and erythromycin and most resistant to penicillin, ampicillin, cefazolin, and vancomycin (Table 2). Patients were treated with a variety of regimens, but monotherapy was administered in most cases (n=12). One patient received no antibiotic therapy. Abscess drainage was performed in conjunction with antibiotic therapy for four (50%) of the patients with an abscess. The sole patient with an infected vascular graft underwent graft removal in conjunction with antibiotic therapy. Among the 20 cases in which adequacy of therapy could be evaluated, 75% were determined to have received adequate therapy. The duration of therapy varied and ranged from at least 3 days to more than 2 months.

Outcomes were reported for 35 cases. The majority of the patients (71.4%) recovered. In regard to the specific sites of infection, 40% of the 15 patients with a pulmonary infection recovered, 40% died, and outcome was not available for 20%. The majority (88%) of the eight patients with an abscess recovered. Recovery was 100% for patients with peritonitis, chorioamnionitis, intra-abdominal infection, vascular graft infection, and fistula. Patients with an esophageal infection, meningitis, wound infection, and erysipeloid infection died. The outcome was not available for the patient with a throat infection. Differences in mortality were not statistically significant between patients who were treated adequately versus those who were not (P=0.99).

Discussion

Lactobacillus is a gram-positive rod-shaped bacterium that varies from long and slender forms to short coccobacilli, occasionally forming short chains [1]. This organism is anaerobic (facultative or strict), non-spore-forming, and lacks motility. Lactobacillus is ubiquitous in humans, inhabiting the mouth, gastrointestinal tract, and the female genital tract [26]. Specific strains of Lactobacillus have been found to be beneficial in the treatment of certain types of diarrhea and vaginosis [7].

A high overall mortality (nearly 30%) associated with Lactobacillus infection was found in our review. It bears reiteration that we looked at “overall” mortality and not “attributable” mortality. The high mortality rate could therefore be accounted for by the underlying immunosuppressive conditions of the patients, since the presence of Lactobacillus has been proposed to be a “marker of serious underlying illness and poor long-term prognosis” [10]. We were unable to confirm this statistically, however, since the underlying conditions in the cases we evaluated were too diverse for creating groups large enough to have sufficient statistical power. Although the mortality rate for the cases of Lactobacillus endocarditis (23%) in our review was consistent with the literature on the overall mortality of endocarditis [103], we found a statistically significant higher mortality rate for patients with polymicrobial infection (P=0.004) and for patients with bacteremia who were treated inadequately (P=0.001). This emphasizes the importance of weighing the clinical significance of Lactobacillus isolated from the blood before disregarding it as a contaminant. While some investigators have proposed criteria for the differentiation of true infection versus contamination [10, 29], the ability to create a uniform consensus of the meaning of Lactobacillus isolated from normally sterile sites is complicated by questions regarding identification, clinical relevance, treatment, and probiotic safety.

Lactobacillus can go unrecognized by clinical laboratories, since growth of anaerobes requires special media and extended incubation [104]. Even after recovery and isolation, Lactobacillus can be problematic to correctly identify since most commercial identification systems employed for anaerobic rod identification are insufficient for the identification of Lactobacillus [104]. In a case series of Lactobacillus bacteremia reported by Salminen et al. [12], 27% of 66 Lactobacillus isolates that were analyzed for species identification were found to be other microorganisms.

Other factors may also make the identification of lactobacilli difficult. Microscopy reveals that Lactobacillus morphology resembles members of other genera, including Corynebacterium, Clostridium, Nocardia, and Streptococcus [29]. Additionally, since Lactobacillus is a fastidious gram-positive rod, it may be confused with diphtheroids on Gram stain and dismissed as a contaminant [56].

Swenson et al. [105] demonstrated the high resistance of lactobacilli to vancomycin. Lactobacillus isolates were tested against 32 antimicrobial agents with none of the isolates proving susceptible to vancomycin (MIC90>256 μg/ml for all isolates). In a case series of Lactobacillus bacteremia, all of the 42 isolates tested for sensitivity to vancomycin were resistant [11]. The findings in our review were inconsistent with these data. Of the 80 isolates that were tested against vancomycin, 22.5% were sensitive. Husni et al. [10] reported a similar percentage in their case series with 27% of 22 isolates found to be sensitive to vancomycin. In our review, the Lactobacillus species that were sensitive to vancomycin were either acidophilus or not speciated. A study conducted by Hamilton-Miller et al. [106] verified that the species acidophilus, in addition to delbrueckii, can be sensitive to vancomycin, while Ruoff et al. [44] found L. leichmannii to be sensitive to vancomycin. However, this begs the question of whether the non-speciated isolates sensitive to vancomycin in our review were acidophilus, delbrueckii, or leichmannii, or quite possibly other organisms misidentified as Lactobacillus.

The most common species of Lactobacillus recorded in our review were casei and rhamnosus. We found that the type of species had no bearing on the type of infection or outcome. Although many species of Lactobacillus have been isolated, some experts consider identification to the species level unnecessary since antibiotic treatment is not altered based on speciation [1]. Furthermore, identification to the species level is not only difficult to achieve, but often inaccurate [104].

Debate surrounding the clinical significance of Lactobacillus is ongoing [9, 10, 29]. Antony et al. [29] considered Lactobacillus a true pathogen rather than a contaminant when describing 12 patients with Lactobacillus bacteremia. Each of these patients had either two sets of positive blood cultures or isolation of the organism from the blood, plus another site of clinical infection. Further evidence of the pathogenic potential of lactobacilli has been provided by in vivo and in vitro studies. Early animal studies of L. casei explored the toxicity potential of this organism [107]. Several studies have demonstrated that specific Lactobacillus species (e.g. rhamnosus and paracasei subspecies paracasei) isolated from patients with endocarditis possess pathogenic traits, such as platelet aggregation, binding to fibronectin, fibrogen, and collagen, and production of enzymes enabling the breakdown of human glycoproteins and the synthesis of human fibrin clots [77, 108110]. All of these characteristics allow the organism to survive and to colonize vascular surfaces.

Husni et al. [10] did not consider all positive blood cultures of Lactobacillus to be significant. In their retrospective review of Lactobacillus blood isolates at the Cleveland Clinic Foundation, each case was categorized as a definite, probable, or possible case of Lactobacillus bacteremia. The definitions were based on the number of positive blood cultures and/or site of another positive culture, presence of polymicrobial bacteremia, and signs and symptoms of infection. Those cases that did not fit the criteria were considered to have Lactobacillus contaminants and were excluded. Interestingly, only a small percentage (11.8%) of the 51 patients identified in their case series was excluded based on their definitions.

If the decision to treat Lactobacillus is made, most experts agree that the treatment of choice should consist of high-dose intravenous penicillin (>25 million units [MU]/day) and an aminoglycoside (typically gentamicin) for synergy [1, 2, 29, 58]. This recommendation is based on in vivo data obtained in the early 1970s from treating a small number of patients with Lactobacillus endocarditis [3]. Bayer et al. [111] suggest the use of either high-dose penicillin (>25 MU/day) or ampicillin (200–250 mg/kg/day) plus an aminoglycoside for synergy. Their recommendation was based on in vivo experience in treating a small number of patients with Lactobacillus endocarditis as well as on in vitro synergy data.

In the cases of Lactobacillus infections we reviewed, very few of the patients were treated with the suggested regimen of high-dose penicillin (>25 MU/day) or high-dose ampicillin (200–250 mg/kg/day) plus an aminoglycoside. We were not able to statistically analyze outcomes in relation to specific antibiotic treatment given the large number of various treatments that were reported. However, the overall sensitivity data from these cases suggest that ampicillin plus gentamicin may be a better empiric option than penicillin and gentamicin. The disappointing sensitivity data reported for penicillin, ampicillin, and gentamicin may be due to the ability of lactobacilli to lower the pH of their environment via lactic acid production. It has been postulated by Sussman et al. [2] that the large quantities of lactic acid produced by lactobacilli can hinder the activity of aminoglycoside antibiotics. Kim et al. [112] found that the autolytic enzyme, which is essential for the bactericidal effect of β-lactam antibiotics, is less active at a lower pH. This may account for the better activity of ciprofloxacin compared to penicillin or cefazolin. Although not commonly endorsed to treat gram-positive infections, ciprofloxacin and perhaps the newer fluoroquinolones with enhanced activity against gram-positive organisms may offer therapeutic alternatives. Our review found erythromycin and clindamycin to be the most effective agents, although most practitioners would hesitate to use these antibiotics as first-line therapy for serious infections due to their bacteriostatic nature. Though not studied in this review, newer antibiotics that specifically target gram-positive organisms, such as linezolid or daptomycin, may offer alternative treatment options.

Debate surrounding the safety of Lactobacillus as a probiotic agent also exists. Some consider probiotics to be safe [7] and lacking association with bacteremia, as reported in two case series [12, 47]. Others are not convinced, especially in light of the increasing population of immunosuppressed individuals [13, 15]. Salminen and Arvilommi [113] point out that, “Any viable microbe can be associated with bacteremia or other infections under certain conditions, especially in severely ill or immunocompromised hosts.” These concerns are warranted, as demonstrated in a recently published review of cases of Lactobacillus bacteremia in Finland in which the authors reported 11 (8%) cases in which an isolate identical to the probiotic L. rhamnosus GG was found [11]. In our review of the literature, we found a very small percentage (1.7%) of cases associated with heavy dairy consumption. Three cases were associated with endocarditis and one with a liver abscess [52, 56, 65, 90]. The case of the patient with the liver abscess highlights the probiotic safety issues, especially because the Lactobacillus isolate recovered was indistinguishable from a probiotic strain and the only underlying conditions of the patient were diabetes and hypertension.

In summary, we found that Lactobacillus can cause a variety of infections, but it is most commonly associated with bacteremia and endocarditis. Patients of all ages and both genders were affected. Cancer, diabetes, and transplantation (specifically liver) were the more common underlying conditions associated with Lactobacillus bacteremia or localized infections. Concomitant organisms were recovered from nearly a quarter of the patients. Common organisms included Candida, Enterococcus, and Klebsiella spp. Overall mortality approached one-third in the patients we reviewed and was significantly associated with polymicrobial infection.

Although the recommended treatment for Lactobacillus infections has traditionally been high-dose penicillin or ampicillin in combination with an aminoglycoside, the sensitivity data gathered in this review indicates otherwise. While empiric therapy with ampicillin and gentamicin may be a better initial choice versus penicillin and gentamicin, the importance of obtaining sensitivity data to help guide therapy is imperative since the susceptibility of Lactobacillus to the above-mentioned antibiotics was relatively low (54.9–70%). While vancomycin may be an option for the treatment of L. acidophilus, the probability of acidophilus being the causative species is low (10%). The results of our review do not support the view that duration of therapy for Lactobacillus infections associated with endocarditis, bacteremia, or other localized infections should be different from that used in the management of infection due to other pathogens. Furthermore, non-medication management of certain types of infection, such as abscesses or vascular graft infections, should be no different than if any other pathogen was involved. Finally, given the high overall mortality found in our review, it seems that a standardized approach to the treatment of Lactobacillus infections may be warranted.

Copyright information

© Springer-Verlag 2004