|Home | About | Journals | Submit | Contact Us | Français|
Botulism developed in a patient following surgical repair of an open radial fracture. Symptoms resolved after treatment with antitoxin and antibiotics, and hardware excision was deferred. Subsequent osteomyelitis necessitated hardware exchange, and wound cultures grew Clostridium argentinense. This case highlights the management of botulism associated with orthopedic hardware.
A 34-year-old man without significant past medical history presented with double vision. Three weeks prior to admission he suffered a compound fracture of his right radius and ulna during a soccer game. He was admitted to the orthopedic service for open reduction and internal fixation with a midulnar and radial plate (see Fig. Fig.1).1). During the procedure, grass and dirt were noted in the surgical bed. He was treated perioperatively with cefazolin and gentamicin, was discharged, and completed 10 days of cephalexin treatment. Eighteen days following surgical repair he noted both double vision and difficulty opening both eyes. He presented three days later after worsening of symptoms and the development of slurred speech and difficulty swallowing.
On admission, he denied fevers or chills, wound breakdown, paresthesias, or any other symptoms and asserted adherence to his postoperative antibiotics. The patient had no significant past medical history, took no other medications, and denied recreational drug use. His speech was dysphonic, and he had bilateral ptosis with notable diplopia. There was severe upper and lower bifacial weakness, and only trace extraocular movement was seen. His pupils were large and minimally reactive to light or accommodation. There was notable tongue and pharyngeal weakness. His right arm showed a healing surgical wound on the volar surface of his forearm. The results of the rest of his examination were unremarkable, as were his laboratory results.
A nerve conduction study and electromyogram procedure were conducted, and the interpretation documented “low baseline amplitudes, especially in proximal musculature, with significant facilitation”; the interpreting neurologist noted that in the appropriate clinical context, the features were consistent with wound botulism. The patient was promptly administered a dose of botulinum antitoxin obtained from the CDC (7,500 units of antitoxin A and 5,500 units of antitoxin type B in 100 ml of normal saline solution over 1 h). Infectious disease consultants recommended commencing high-dosage intravenous penicillin G treatment and planning for surgical debridement and removal of the implanted hardware. After receiving botulinum antitoxin, his clinical symptoms improved overnight, and surgical excision was deferred. A blood sample obtained prior to the administration of botulinum antitoxin was negative for Clostridium botulinum toxin when tested by the CDC after trypsinization by use of a mouse bioassay. His symptoms improved, and he was discharged to complete 6 weeks of oral metronidazole and clindamycin treatment.
Radiographs obtained 6 weeks after discharge were indicative of possible osteomyelitis. He returned to the operating room for irrigation and debridement with a one-stage hardware exchange. Operative cultures grew Enterobacter cloacae, Enterobacter taylorae, and clostridial species; no botulinum toxin was identified upon testing the subculture of the clostridial species after trypsinization by a mouse bioassay. For determination of the species represented by the Clostridium isolate, genomic DNA was extracted from the anaerobic subculture by following the manufacturer's instructions using a Qiagen DNA mini kit (Qiagen, Valencia, CA) and was subsequently amplified at the 16S rRNA gene and sequenced using both original primers (8). The consensus sequence was built using Sequencher (Gene Codes Corp., Ann Arbor, MI), and a BLASTN search demonstrated 96% homology with both C. botulinum type G (GenBank accession no. M59087) and C. subterminale isolate DSM 758 (GenBank accession no. AF241843). He was treated initially with intravenous pencillin G, oral ciprofloxacin, and oral metronidazole; a rash developed, and the antibiotics used for treatment were changed to intravenous piperacillin-tazobactam and oral metronidazole. After 6 weeks of antibiotic treatment, his treatment regimen was changed to suppressive sulfamethoxazole-trimethoprim (800 to 160 mg administered twice a day) and amoxicillin (1,000 mg administered twice a day). Eight weeks later, he was admitted for a final revision with insertion of a vascularized free fibula graft at a large defect in the ulna. The wound appeared to be clean, and test results from cultures obtained at that time were negative.
His suppressive antibiotics were stopped 9 months after the initial diagnosis, and he remained disease-free 10 months later, at almost 2 years after the initial injury.
Wound botulism rarely complicates infected wounds but requires prompt recognition to prevent adverse outcomes. Aside from epidemic cases traced to the subcutaneous or intramuscular injection of black-tar heroin (16), most cases are attributable to traumatic injuries, infected surgical wounds, or sinusitis (due to intranasal cocaine use). Clinically, wound botulism produces a syndrome of motor neuron weakness similar to that produced by food-borne and infant botulism; all syndromes result from the blockade of acetylcholine release from presynaptic peripheral motor neurons due to the activity of botulinum toxin. The presence of this botulinum toxin (as detected by the mouse bioassay in tests of serum or wound tissue) most readily supports the diagnosis of wound botulism; among 33 cases reported to the CDC from 1943 to 1985, toxin was detected in only 25 cases, though the detection rate improved to over 90% during the most recent epidemic of black-tar heroin-associated cases (2). Currently, seven distinct toxin types (A, B, C1, D, E, F, and G) are recognized, with wound botulism principally mediated by types A and B.
In all syndromes, the definitive therapy is the administration of antitoxin. Early administration has been associated with improved outcomes, though advanced cases may require aggressive supportive care, including intubation. Equine antitoxin is available in two formulations: a bivalent formulation (serotypes A and B) that is employed in all noninfant cases of presumed wound botulism, and a monovalent formulation (serotype E) used in certain cases at risk for type E botulism, such as those attributable to the ingestion of contaminated fish products. In the United States, the antitoxins are dispensed by the CDC from its regional quarantine stations upon a request from state and local health departments or, if the request is made after normal daily operating hours, by telephoning the CDC Director's Emergency Operations Center directly at (770) 488-7100.
Our case highlights several key microbiologic and clinical issues germane to the management of wound botulism. Surgically, the necessity of debridement in wound botulism is unclear, and we can identify no prior reports of wound botulism associated with exogenous material in which a cure was effected with medical therapy alone in the setting of retained foreign material. Though debridement is often recommended, our patient's radial instability precluded immediate and total hardware removal. Typically, orthopedic hardware infections achieve the highest cure rates by means of total excision and one- or two-stage replacement. In this case, the absence of local, radiographic, or systemic signs of purulent infection allowed deferral of the planned surgical revision in the hope that bone union could be achieved after both antitoxin administration to mitigate neurologic symptoms and antibiotic therapy to eradicate organisms limited to the soft tissue. Though this case demonstrates that clinical botulism may be cured despite the retention of infected hardware, it also shows that the diagnosis of wound botulism may indicate polymicrobial wound contamination even in the absence of other compelling clinical or laboratory indications of infection. This possibility should impact both antibiotic therapy and surgical planning.
Medically, the role of antibiotic therapy of wound botulism is not clearly defined. Treatment recommendations emphasize antitoxin administration, and though penicillin is often recommended, its efficacy is unproven. Previous case reports describe successful adjunctive therapy with penicillin, metronidazole, and macrolides, among others, though usually in the setting of abscess drainage or the removal of infected material. According to the results of in vitro testing, C. botulinum isolates are reliably susceptible to tetracycline, metronidazole, and chloramphenicol and usually susceptible to penicillin, vancomycin, and clindamycin (14). In our case, though the combination of metronidazole and clindamycin may have contributed to the prevention of neurologic relapse, their inability to eradicate clostridia from the hardware and their lack of activity against Enterobacter bacteria allowed osteomyelitis to develop.
Notably, motor weakness was alleviated by the combination of a single dose of antitoxin and ongoing therapy with metronidazole and clindamycin; debridement and culture 6 weeks later revealed viable Clostridium bacteria in the surgical bed, which we presume were present at the time of clinical botulism, though we cannot know definitively. Several possibilities may account for the successful resolution of symptoms despite the failure to eradicate clostridia. It is possible that an additional classically toxigenic clostridial species was also present at the time of clinical symptoms and was either eradicated by medical therapy prior to subsequent wound culture or simply undetected by subsequent wound culture. Alternatively, if the isolated Clostridium species had been responsible for toxin production, antibody levels may have remained sufficient to prevent toxin activity during convalescence: although the half-life of antibodies in antitoxin is only 5 to 8 days, the amount of neutralizing antibody in a licensed antitoxin far exceeds that necessary to neutralize toxin (6). Additionally, the passive immunity conferred by antitoxin could have been augmented by the acquisition of active immunity. The immunogenicity of botulinum toxin has been previously demonstrated by the development of neutralizing antibodies both in animals (15) and in patients receiving repeated therapeutic toxin type A for dystonia (7). Our case differs in that, instead of intermittent exposure to therapeutic toxin injections, the patient could have been exposed to continuous ongoing toxin production; no data describe the acquisition of immunity in this scenario. Finally, the resolution of symptoms may have been due to decreased toxin production resulting from the disruption of bacterial protein synthesis by the administration of clindamycin. Though clindamycin therapy is efficacious in treatment of other toxin-mediated diseases (11) and in downregulating toxin production in other clostridial species (12), its utility in preventing botulinum toxin production is unknown.
The importance of the isolated Clostridium species is unclear: ours is the first reported case of wound botulism in which Clostridium argentinense may have been causative. C. argentinense is genetically distinct from other clostridial species but phenotypically diverse, subsuming both the toxinogenic C. botulinum type G and the nontoxinogenic strains previously identified as C. subterminale and C. hastiforme (13). The significant homology of the 16S rRNA gene sequences of C. botulinum type G and C. subterminale prevents us from definitively identifying this isolate and implicating it as the causative agent of clinical botulism, but several aspects make C. argentinense an unusual suspect in this case of wound botulism. C. botulinum type G has been definitively isolated only from soil samples in Argentina (5) and Switzerland (10). It was identified at autopsy in a series of five unexpected human deaths (9), but its presence was of unknown clinical significance, and it has not previously been definitively implicated in cases of human botulism. Additionally, though it does produce a type G neurotoxin that has resulted in botulism in experimentally treated monkeys (3), the toxin should not have been neutralized by antitoxin to types A and B; though toxin types B and G share a high degree (58%) of amino acid sequence homology (1), it is not believed that antitoxins possess significant cross-reactivity between serotypes. Furthermore, C. subterminale can possess the gene encoding toxin type B (4), but neither it nor the other strains comprising C. argentinense typically produce toxin. Though it is possible that the Clostridium species we isolated was culpable with respect to toxin production, this ambiguity serves to underscore the possibility that a second, unidentified clostridial species producing toxin A or B was present at the time of clinical symptoms.
This case demonstrates that, despite initial hardware retention, a one-stage hardware exchange, and bone graft implantation, mitigation of clinical botulism is possible. Our case, in which neither traditionally recognized forms of botulinum toxin nor C. botulinum itself was identified, also underscores the absolute necessity of clinical recognition in the diagnosis of botulism.
We thank the Duke University Neurology Service for assistance in caring for the patient, Keith Simmon and Cathy Petti for sequence identification, and Peggy Althaus and L. Barth Reller for microbiologic assistance.
No funding support was received for this work.
S.M.T., C.R.W., T.C.D., D.S.R., and G.M.C. have no conflicts of interest to report.
Published ahead of print on 9 December 2009.