ABSTRACT
Infant botulism is an infectious intestinal toxemia that results from colonization of the infant large bowel by Clostridium botulinum (or rarely, by neurotoxigenic Clostridium baratii or Clostridium butyricum), with subsequent intraintestinal production and absorption of botulinum neurotoxin that then produces flaccid paralysis. The disease is often initially misdiagnosed as suspected sepsis or meningitis, diagnoses that require prompt empirical antimicrobial therapy. Antibiotics may also be needed to treat infectious complications of infant botulism, such as pneumonia or urinary tract infection. Clinical evidence suggests (see case report below) that broad-spectrum antibiotics that are eliminated by biliary excretion may cause progression of the patient’s paralysis by lysing C. botulinum vegetative cells in the large bowel lumen, thereby increasing the amount of botulinum neurotoxin available for absorption. The purpose of this antimicrobial susceptibility study was to identify an antimicrobial agent with little or no activity against C. botulinum that could be used to treat infant botulism patients initially diagnosed with suspected sepsis or meningitis, or who acquired secondary infections, without lysing C. botulinum. Testing of 12 antimicrobial agents indicated that almost all California infant botulism patient isolates are susceptible to most clinically utilized antibiotics and are also susceptible to newer antibiotics not previously tested against large numbers of C. botulinum patient isolates. No antibiotic with little or no activity against C. botulinum was identified. These findings reinforce the importance of promptly treating infant botulism patients with human botulism immune globulin (BIG-IV [BabyBIG]).
INTRODUCTION
Infant botulism (IB) is an acute symmetric descending flaccid paralysis that occurs in infants 12 months of age or less that results from botulinum neurotoxin (BoNT) produced by Clostridium botulinum (and rarely Clostridium butyricum or Clostridium baratii). Illness develops after spores of these organisms are swallowed and reach the infant’s large intestine, where they germinate, multiply, and release BoNT into the intestinal lumen. Intraluminal toxin is absorbed via lymphatics into the circulation and transported to the neuromuscular junction, where it binds presynaptically, blocks acetylcholine release, and produces flaccid paralysis (1–3).
Although more than 40 years have elapsed since the initial recognition of infant botulism, “suspected sepsis” remains the most common admission diagnosis of patients who are later confirmed by laboratory studies to have infant botulism (1, 2). Infants initially diagnosed with possible sepsis or suspected meningitis require prompt empirical treatment with parenteral broad-spectrum antibiotics, some of which reach the lumen of the colon via biliary excretion. Additionally, patients with laboratory-confirmed IB may require antimicrobial therapy for community-acquired or nosocomial infections. A common concern in clinical management is that lysis of C. botulinum in the large intestine by such antibiotics may worsen the patient’s paralysis by increasing the amount of BoNT available for absorption (4, 5).
Hence, identification of broad-spectrum antibiotics that have limited or absent bactericidal activity against C. botulinum would provide clinicians with better therapeutic choices. Two antimicrobial susceptibility studies of C. botulinum were reported more than 35 years ago (6, 7). Fifteen years ago, the antimicrobial susceptibilities of six neurotoxigenic Clostridium butyricum type E strains obtained from six patients with intestinal toxemia botulism were reported (4).
This study sought to determine whether California C. botulinum infant botulism patient isolates (i) were sensitive to ceftriaxone (because it is commonly used), (ii) were resistant to other cephalosporins or fluoroquinolones, and (iii) displayed the same susceptibility patterns seen with previously tested antimicrobials (6, 7). The intent was to identify an antimicrobial agent that could be used to treat secondary infections in IB patients that would not lyse C. botulinum vegetative cells temporarily residing in the large intestine, the same question that Fenicia et al. investigated with neurotoxigenic C. butyricum type E (4).
We report here the susceptibility of 260 strains of C. botulinum and a single strain of neurotoxigenic C. baratii isolated from California infant botulism patients, as well as three additional strains of neurotoxigenic C. baratii that were not isolated from California IB patients, to a variety of antibiotics.
Case report.A previously healthy 3.5-month-old female infant with a 2-week history of constipation was admitted to a South Carolina hospital, presenting with marked fatigue and a urinary tract infection. Her urine culture grew an Enterobacter sp., and a 7-day course of intravenous cefotaxime was started at admission. By hospital day 3 (HD 3), she had developed ptosis and general weakness. A stool specimen tested at the National Botulism Reference Laboratory, Centers for Disease Control and Prevention (Atlanta, GA) contained BoNT/A and C. botulinum type A, thereby establishing the diagnosis of infant botulism. The patient was not treated with human botulism immune globulin (BIG-IV [BabyBIG]) because at the time of her illness, the product was only available if administered within 7 days of hospital admission (2, 3). On HD 17, the patient was treated with a single dose of intravenous ceftriaxone for suspected pneumonia caused by a Staphylococcus species isolated from a tracheal aspirate. Within 24 h, she developed complete paralysis that required intubation and ventilatory support for the subsequent 27 weeks (i.e., more than half a year). In the months following her treatment with ceftriaxone, she experienced recurrent urinary tract infections, respiratory syncytial virus pneumonitis, and several failed attempts at extubation. However, with continuous meticulous supportive care alone (i.e., no tracheostomy and no gastrostomy), she recovered completely and was discharged to home after a total hospital stay of 9.5 months.
Ceftriaxone is excreted by the kidney into the urine and by the liver into the biliary tree and small intestinal lumen (8). Approximately 33 to 67% of ceftriaxone is eliminated in the bile, and the remainder is eliminated in the urine (8). This case provided the impetus to identify an antibiotic(s) that could be used to treat nosocomial or community-acquired infections in infant botulism patients without risking possible lysis of intestinal C. botulinum with concomitant release of intracellular BoNT.
RESULTS
The percentages of C. botulinum type A (n = 156), C. botulinum type B (n = 89), and C. botulinum type Ba (n = 12) strains inhibited across a range of MICs by the following 12 antibiotics are shown in Tables 1 to 3: ampicillin (AMP), penicillin (PEN), amoxicillin-clavulanic acid (AMC) (2/1), cefuroxime (CXM), cefotaxime (CTX), ceftriaxone (CRO), cefepime (FEP), moxifloxacin (MXF), gentamicin (GEN), trimethoprim-sulfamethoxazole (SXT) (1/19), nalidixic acid (NAL), and vancomycin (VAN). The MICs for C. botulinum type Bf (n = 2) and C. baratii type F (n = 4) strains were also determined.
Percentage of C. botulinum type A strains inhibited across a range of MICs of 12 antimicrobial agents
Percentage of C. botulinum type B strains inhibited across a range of MICs of 12 antimicrobial agents
Percentage of C. botulinum type Ba strains inhibited across a range of MICs of 12 antimicrobial agents
Penicillin (with one exception) demonstrated the greatest inhibitory activity of the antibiotics tested. Ninety-nine percent of all C. botulinum type A and B strains were inhibited by ≤0.50 μg/ml PEN. Similar results were achieved with ≤1.0 μg/ml AMP and ≤1.0/0.5 μg/ml AMC. Most C. botulinum type Ba strains and both C. botulinum type Bf strains were inhibited by ≤0.50 μg/ml PEN and AMP and by ≤2.0 μg/ml AMC.
Unexpectedly, a single strain of C. botulinum type A demonstrated high-level resistance to both PEN and AMP, each having an MIC of 32 μg/ml. However, this strain was susceptible to AMC, a beta-lactam–beta-lactamase inhibitor combination, with MICs of 0.75/0.375 μg/ml (9). This penicillin-resistant type A strain was found to produce beta-lactamase using the chromogenic nitrocefin disk test (Nitrocef; Hardy Diagnostics, Santa Maria, CA).
Collectively, cephalosporin antibiotics demonstrated higher MICs than did PEN or AMP. CRO was the most active cephalosporin against all C. botulinum strains and against C. baratii type F strains, followed in descending order by CTX, CXM, and FEP. More than half of all C. botulinum type A and B strains were inhibited by an MIC of ≤2.0 μg/ml of CRO, CTX, or CXM. FEP was the least active of the cephalosporins, with just 50% of type A and type B strains being inhibited at concentrations between 2 and 4 μg/ml. In general, all bivalent C. botulinum type Ba and Bf and C. baratii type F strains were more resistant to the cephalosporins than were C. botulinum type A and B strains. More than half of the type Ba strains were inhibited only by >8 μg/ml FEP. This evident increase in FEP MIC for the 12 type Ba strains may reflect in part the issue that relatively fewer bivalent Ba than type A and B strains were tested (Table 3). The two C. botulinum type Bf bivalent strains were inhibited by 3.0 and 12.0 μg/ml FEP, respectively, and 2 of 4 C. baratii type F strains were completely resistant to FEP at ≥32 μg/ml (low-range Etest strip). The other 3 cephalosporins evaluated inhibited the 2 C. botulinum type Bf and the 4 C. baratii type F strains at an MIC of ≤6.0 μg/ml.
MXF was the only fluoroquinolone evaluated in this study. MXF was active against all strains of C. botulinum and C. baratii type F tested. Of the C. botulinum type A and type B strains, 87, 97, 99, and 100% were inhibited at 0.5, 1, 2, and 4 μg/ml, respectively. The MIC at which more than 50% of C. botulinum type A, B, and Ba strains were inhibited by MXF was ≤0.50 μg/ml. All strains of C. baratii type F were inhibited by ≤0.50 μg/ml MXF. The 2 strains of C. botulinum type Bf had MXF MICs of 1.0 and 4.0 μg/ml.
The MIC of SXT for 76% of C. botulinum type A and 85% of C. botulinum type B strains was ≤2/38 μg/ml when tested on blood Mueller-Hinton II (BMH) agar (Tables 1, 2, and 4). Twenty-two C. botulinum type A strains and 12 C. botulinum type B strains had SXT MICs of ≤0.008/0.152 μg/ml when grown on BMH but had SXT MICs of ≥32/608 μg/ml when grown on both formulations of WCA. One type A strain had an MIC of ≥32/608 μg/ml when grown on WCA supplemented with 5% lysed horse blood (LHB) but had an MIC of 3/57 μg/ml when grown on WCA supplemented with thymidine phosphorylase (TP).
Percentage of California C. botulinum strains inhibited across a range of MICs of SXT compared to percent inhibition reported by Swenson et al.
No differences in antibiotic activity were found between strains from different geographical regions of California (data not shown).
DISCUSSION
At clinical presentation, patients with infant botulism often appear to have sepsis or meningitis and hence are treated empirically with broad-spectrum and combination antimicrobial therapy. Recommended empirical triple-antibiotic therapy includes VAN and AMP plus CTX, CRO, or FEP (10, 11). Patients with laboratory-confirmed IB may also require antimicrobial therapy for secondary or community-acquired pulmonary or urinary tract infections, and such therapy often includes CTX or the combinations of CRO and GEN or AMP and GEN, respectively (10, 11). CRO, which is widely used in the United States for pediatric patients, has a broad antimicrobial spectrum. However, CRO did not exist in 1980 when the first extensive in vitro antibiotic susceptibility study of C. botulinum was done (6). Clinical observations suggest that CRO, which is excreted by the biliary system into the small intestine (8), may lyse C. botulinum vegetative cells with the release of BoNT and exacerbation of illness (4).
California C. botulinum patient isolates were found to be highly susceptible to PEN. Nearly all strains, both types A and B, were inhibited by less than 0.50 μg/ml PEN. However, a single strain of C. botulinum type A demonstrated substantial resistance to both AMP and PEN and was found to produce beta-lactamase. Mazuet et al. reported a similar strain also isolated from an IB patient in France (12). Additionally, C. botulinum was sensitive to the 4 cephalosporins tested, some of which have not been previously tested against large numbers of C. botulinum strains. CRO had the greatest activity of the cephalosporins tested; all C. botulinum type A strains were inhibited at ≤8 µg/ml, and all C. botulinum type B strains were inhibited at ≤4 µg/ml. These findings suggest that IB patients treated with a cephalosporin, especially CRO, may be at increased risk of developing more extensive and sustained paralysis (4).
MXF was included in this study as a representative of the fluoroquinolones, an antibiotic class that has not previously been tested against large numbers of C. botulinum strains. Although MXF is generally contraindicated for treating pediatric patients because of possible cartilage damage and possible prolongation of the QT interval, it can be used under certain circumstances for treating serious infections if no other effective treatments are available (13). Like the cephalosporins, all C. botulinum and neurotoxigenic C. baratii type F strains tested were sensitive to MXF.
No differences in susceptibilities between C. botulinum types A and B strains were observed. The antimicrobial susceptibilities of bivalent C. botulinum type Ba and Bf strains were comparable to those of the monovalent type A and B strains (Tables 1 to 3). Also, no differences in susceptibilities were found between strains from different geographical regions of California. Furthermore, the MICs of most antibiotics previously studied, including those for GEN, NAL, and VAN, were similar to those in our findings (Tables 1 to 3). Like Dezfulian and Dowell (6) and Swenson et al. (7), we found that GEN was relatively inactive against CA strains of C. botulinum and C. baratii. However, the possible clinical appeal of GEN use is tempered by its synergism with BoNT in potentiating neuromuscular blockade (14).
A notable result of our study is the apparent discordant MIC data for SXT compared to those previously reported by Dezfulian and Dowell (6) and Swenson et al. (7). In their 1980 study of C. botulinum antibiotic susceptibility, Swenson et al. reported that most type A and B strains were resistant to the trimethoprim-sulfamethoxazole (SXT) combination (7). Dezfulian and Dowell reported similar findings (6). Unexpectedly, we found the opposite. In our study, the percentage of C. botulinum type A and B strains inhibited by SXT over a range of MICs indicated that these strains were mostly susceptible to this antimicrobial. In accordance with standard interpretive guidelines and the Etest manufacturer’s instructions, the MIC of SXT was interpreted at 80% inhibition when diffuse growth was present (15). Distinct zones of growth inhibition were evident for many of the strains inhibited by a low SXT MIC, thereby precluding the possibility of misinterpretation due to diffuse or sparse growth.
Swenson et al conducted their antimicrobial susceptibility study using the agar dilution method and Wilkins-Chalgren agar (WCA) (Table 4). To prevent inhibition of SXT by thymidine, they supplemented WCA with 5% lysed horse blood, which is a natural source of thymidine phosphorylase, the enzyme that converts thymidine to thymine (5). It is possible that the lysed horse blood added by Swenson et al. (7) did not provide enough thymidine phosphorylase in the media. The residual thymidine may have then inhibited the SXT, thereby allowing C. botulinum to grow and give the appearance of resistance to SXT. In contrast, in our susceptibility studies, we used a commercially prepared BMH agar known to contain inconsequential amounts of sulfonamide inhibitors, the specifics of which are proprietary (16).
Our initial MIC data found that most California C. botulinum strains were inhibited by ≤2/38 μg/ml SXT when grown on BMH (Tables 1 to 4). However, a subset of strains was found to be resistant to 32/608 μg/ml SXT when grown on WCA. All media passed quality control testing for sulfonamide inhibition (15). Monthly quality control testing of media in our laboratory using a standardized suspension of C. botulinum results in smaller colonies on botulinum selective medium than the colonies grown on nonselective egg yolk agar. This is also true for wild-type strains of C. botulinum. These findings further suggest that C. botulinum is partially inhibited by SXT. It is unknown if the Etest method used to determine the MIC of SXT in our study played a role or if other factors, such as the homogeneity of the source of the California strains, is contributory. SXT was the only drug we tested in which our results did not correlate with those previously reported (6, 7). Nonetheless, these discordant observations may warrant further investigation.
In conclusion, our study was unable to identify an antimicrobial agent to which C. botulinum and neurotoxigenic C. baratii strains were consistently resistant that could then be safely used to treat concomitant bacterial infections in patients with infant botulism. We sought to identify such an antibiotic to avoid the possibility that intraluminal lysis of C. botulinum vegetative cells in the colon might increase subsequent botulinum toxemia. Patients treated with BIG-IV before starting antimicrobial therapy avoid this risk because of the large toxin-neutralizing capacity and long half-life of BIG-IV (2). Treatment with BIG-IV before administration of antibiotics (or as soon as possible thereafter) enables physicians to select the optimal antibiotic regimen for the patient without concern for susceptibility of C. botulinum to the antibiotics prescribed.
MATERIALS AND METHODS
Strains.A total of 260 botulinum neurotoxin-producing strains of Clostridium isolated from California patients hospitalized with infant botulism was studied, consisting mainly of C. botulinum type A (n = 156) and proteolytic C. botulinum type B (n = 89) strains. Proteolytic C. botulinum type Ba bivalent strains (those that produce both BoNT/A and BoNT/B) (n = 12), proteolytic C. botulinum type Bf bivalent strains (n = 2), and a neurotoxigenic Clostridium baratii type F strain (n = 1), all isolated from California IB patients, were also tested. Three additional neurotoxigenic C. baratii type F strains isolated from 2 non-California infant botulism patients and one CA adult patient were also studied. All isolates came from the Infant Botulism Treatment and Prevention Program (IBTPP) culture collection.
Strain selection criteria.To ensure statewide geographic coverage of isolates tested, strains from all counties in California identified as the county of residence at the onset of illness were tested (n = 42 [CA has 58 counties]). All strains from each county (n = 15) that had 4 or fewer infant botulism patients were included. For those counties (n = 24) that had more than 4 but fewer than 60 patients, 30% of the patient strains were tested. For those counties (n = 3) with 60 or more patients, 20% of the patient strains were tested. All C. botulinum type Ba and Bf strains and C. baratii type F strains in the collection at the time of the study were tested. In the more densely populated counties of California, both C. botulinum type A and type B strains cause infant botulism. The number of toxin type A and toxin type B strains selected for testing from each county having more than 4 cases was determined by the percentage of cases caused by each toxin type. For those counties in which >4 infant botulism cases occurred, the oldest case isolate and the most recent case isolate were selected for testing, with the remaining isolates chosen to be evenly distributed by year of isolation.
Susceptibility testing.Etest strips (AB Biodisk, Solna, Sweden) were used to determine the MICs of the following 12 antimicrobial agents: amoxicillin-clavulanic acid (AMC) (2/1), ampicillin (AMP), cefepime (FEP), cefotaxime (CTX), ceftriaxone (CRO), cefuroxime (CXM), gentamicin (GEN), moxifloxacin (MXF), nalidixic acid (NAL), penicillin (PEN), trimethoprim-sulfamethoxazole (SXT) (1/19), and vancomycin (VAN).
Susceptibility testing was performed on 150-mm-diameter blood Mueller-Hinton II (BMH) agar plates supplemented with 5% sheep blood (Hardy Diagnostics, Santa Maria, CA). This medium was chosen for its ability to support the luxurious growth of C. botulinum and for its low thymidine concentration, a necessary condition for accurately testing bacterial susceptibility to SXT (15). Large quantities of BMH and Etest strips (each spanning just two lot productions) were purchased to avoid possible fluctuations in MIC values as a consequence of lot-to-lot product variation. The BMH agar and Etest strips underwent in-house quality control testing to verify performance characteristics that included testing for possible presence of sulfonamide inhibitors in the BMH agar (15).
All isolates were subcultured from frozen stocks to culture tubes containing prereduced chopped meat glucose starch broth (CMGS; produced in-house) and incubated at 35°C for 72 h in a Bactron IV anaerobic chamber (90% N2, 5% CO2, and 5% H2 gas atmosphere; Sheldon Mfg., Cornelius, OR), in which all further experimental manipulations occurred. One hundred microliters of actively growing CMGS broth culture was streaked for isolation on 4% egg yolk agar plates (produced in-house) and incubated at 35°C for 48 h. Isolated colonies were emulsified in physiologic saline and adjusted to a turbidity equal to 0.5 (±0.05) McFarland standard using a Vitek densitometer (bioMérieux, Hazelwood, MO). BMH agar plates prereduced in the anaerobe chamber overnight were inoculated with the adjusted bacterial suspensions using standard techniques (AB Biodisk Etest technical guide 1B: Susceptibility testing of anaerobes Solna, Sweden) (9, 17). Each isolate was inoculated onto two BMH agar plates, and 6 different Etest strips were applied to each plate, thereby testing a total of 12 antibiotics per isolate. Plates were examined after approximately 24 h of incubation for a confluent lawn of growth and for distinct elliptical zones of bacterial inhibition before interpreting the MIC of each antibiotic. Less than 3% of the isolates required extended incubation (always less than 48 h) to reach visible confluent growth. The MIC values for each antibiotic were interpreted in accordance with Etest manufacturer guidelines (AB Biodisk Etest technical guide 1B: Susceptibility testing of anaerobes, Solna, Sweden). A single C. botulinum type B strain could only be cultured on Schaedler agar supplemented with 5% sheep blood and 100 units/liter thymidine phosphorylase (TP). There were no significant differences in MIC values for this single isolate compared to all other type B isolates tested on BMH agar.
Trimethoprim-sulfamethoxazole testing.Botulinum selective medium (BSM) that contains trimethoprim and sulfamethoxazole enables selective isolation of C. botulinum from human fecal flora (18). In addition, previous studies reported that C. botulinum type A and proteolytic C. botulinum type B strains exhibited resistance to trimethoprim and sulfamethoxazole (6, 7). However, initial results using BMH agar indicated that most strains in the IBTPP collection were susceptible to SXT. For this reason, we tested 36 strains (24 type A and 12 type B) that showed susceptibility to SXT when grown on BMH and compared their sensitivities to SXT when grown on Wilkins-Chalgren agar (WCA).
Two formulations of WCA were prepared in-house and compared to BMH agar for determining the MIC of SXT. Dehydrated Wilkins-Chalgren medium (Oxoid, Basingstoke, England) was rehydrated per the manufacturer’s instructions, and agar (Oxoid) was added to a final concentration of 2.5%. The first preparation of WCA was supplemented with 5% lysed horse blood (LHB) (7), and the second preparation was supplemented with 100 units/liter thymidine phosphorylase (TP), the amount present in BSM (18). Both formulations of WCA were poured in 150-mm-diameter petri dishes to a standard depth of 4 mm and were quality control tested for sulfonamide inhibition (15).
Twenty-four C. botulinum type A and 12 C. botulinum type B strains were cultured on WCA that contained 5% LHB and on WCA that contained 100 units/liter TP. Both formulations of WCA passed quality control testing to verify the absence of sulfonamide inhibition. Media that yield a zone of growth inhibition diameter of ≥20 mm for Enterococcus faecalis ATCC 29212 around a disk impregnated with SXT are considered acceptable for testing sulfonamides (15). Interestingly, in quality control testing, BMH agar demonstrated a zone of inhibition of 40 mm, while for both formulations of WCA, the zone of inhibition was only 23 mm. Additionally, the WCA that contained LHB had heavier diffuse growth around the SXT disk (but still within 80% inhibition) than did the WCA that contained 100 units/liter TP.
MIC interpretation.Interpretive standards for anaerobic Gram-positive organisms exist only for PEN, AMP, and AMC (9). There are no MIC susceptibility breakpoints for the other 9 antibiotics when tested against Gram-positive anaerobes. For this reason, testing results are reported here as the percentage of strains inhibited by each antibiotic across a range of MICs.
ACKNOWLEDGMENTS
We declare no conflicts of interest.
This study was supported by the Infant Botulism Treatment and Prevention Fund of the California Department of Public Health.
FOOTNOTES
- Received 27 July 2018.
- Returned for modification 24 August 2018.
- Accepted 22 September 2018.
- Accepted manuscript posted online 1 October 2018.
- Copyright © 2018 American Society for Microbiology.
