Sensitivity of Mouse Bioassay in Clinical Wound Botulism

Transcription

1 MAJOR ARTICLE Sensitivity of Mouse Bioassay in Clinical Wound Botulism Charlotte Wheeler, 1 Gregory Inami, 2 Janet Mohle-Boetani, 1 and Duc Vugia 1 1 Infectious Diseases Branch and 2 Microbial Diseases Laboratory, Division of Communicable Disease Control, Center for Infectious Diseases, California Department of Public Health, Richmond (See the editorial commentary by Sobel on pages ) Background. California has an ongoing epidemic of wound botulism (WB) among injection drug users (IDUs). We retrospectively studied a cohort of patients with WB and determined the sensitivity of the mouse bioassay the gold standard laboratory test for confirmation of botulism in verifying WB. Methods. We defined a clinical case of WB as an acute, bilateral, descending, flaccid paralysis starting with 1 cranial nerve palsies in an IDU with no other explainable diagnosis. We calculated the sensitivity of the mouse bioassay as the proportion of clinical WB cases that had positive serum toxin test results by mouse bioassay. We compared serum toxin positive with serum toxin negative patients. Results. Of 73 patients with WB, 50 tested serum toxin positive, yielding a sensitivity of 68%. Serum toxin positive patients did not differ significantly from serum toxin negative patients with respect to demographic characteristics or injection drug use practices or in days from patient symptom onset to collection of specimens for testing. Patients did not differ significantly by clinical characteristics, except that serum toxin positive patients were more likely than serum toxin negative patients to have required mechanical ventilation during their hospital courses (74% vs. 43%; P p.01). Conclusions. In this study, the mouse bioassay failed to detect botulinum toxin in the serum samples of nearly one-third of IDUs with characteristic WB. Such patients should be considered to have probable WB. Physicians should be aware of the test s limitations and base their final diagnosis of suspected WB on clinical criteria when the mouse bioassay produces negative results. Wound botulism (WB) is a neuroparalytic disorder caused by the in situ elaboration of toxin by Clostridium botulinum in infected tissue. WB was first reported in 1951 [1] and remained an exceedingly rare condition for nearly 40 years, occurring nearly exclusively in persons with contaminated traumatic or surgical wounds [2, 3]. An upsurge of cases among injection drug users (IDUs), which began in the 1990s, has changed the epidemiology of WB. Currently, the typical presentation for WB is cranial nerve palsies in an IDU with abscesses from injecting heroin subcutaneously [2, 4]. The epidemic is centered in the western United States, and Received 10 October 2008; accepted 12 January 2009; electronically published 12 May Reprints or correspondence: Dr. Charlotte Wheeler, Infectious Diseases Branch, Div. of Communicable Disease Control, Center for Infectious Diseases, California Dept. of Public Health, 850 Marina Bay Pkwy., Bldg. P, Richmond, CA (charlotte.wheeler@cdph.ca.gov). Clinical Infectious Diseases 2009; 48: This article is in the public domain, and no copyright is claimed /2009/ DOI: / California has consistently reported approximately three-quarters of WB cases in the country [5]. Despite its unique presentation, botulism remains challenging to diagnose because of its relative rarity among patients presenting to emergency departments. Guillain-Barré syndrome and myasthenia gravis can also cause cranial nerve palsies and are more common illnesses than botulism. Because treatment is urgent and because laboratory testing can take as long as 4 days to yield results, the initial diagnosis of botulism must be based on clinical criteria alone. Laboratory testing, which involves the administration of patient sera with and without monovalent antitoxin to pairs of mice, serves to confirm the botulism diagnosis and to define the type of circulating toxin (e.g., A or B). The Centers for Disease Control and Prevention base their estimates of the burden of WB in the United States on laboratory-confirmed illness only [6]. Physicians may also base clinical decisions, such as whether to pursue further diagnostic evaluation or the designation of the patient s final diagnosis, on laboratory confir- Mouse Bioassay Sensitivity in Wound Botulism CID 2009:48 (15 June) 1669

2 mation of WB. Yet, although the specificity of the bioassay is high because a positive result means having a particular botulinum toxin type, the test s sensitivity (i.e., the proportion of true cases detected by the test) in detecting WB is unknown. Descriptions of negative test results for patients presenting with characteristic features of WB have raised concerns about the test s sensitivity since 1973 [7 11]. We undertook a study to determine the sensitivity of the mouse bioassay in detecting WB and to compare serum toxin positive with serum toxin negative WB patients. We used rigorous clinical criteria for WB to define cases and drew the study cohort retrospectively from the California Department of Public Health (CDPH) surveillance data for We calculated the sensitivity of the mouse bioassay as the proportion of clinical WB cases that tested serum toxin positive by mouse bioassay. We analyzed data to determine whether demographic, risk factor, or clinical characteristics differed between serum toxin positive and serum toxin negative patients. METHODS Overview. In the United States, botulinum antitoxin is available only through the public health system. Before authorizing release of antitoxin, public health duty officers confer with attending physicians to determine whether patients histories and clinical presentations are consistent with botulism. Before administering antitoxin, clinicians collect serum specimens for botulism testing at public health laboratories. Laboratory test results are reported to clinicians a few days later, when they become available. All antitoxin releases and laboratory results are tracked by the public health system. Definitions. We defined a clinical case of WB as an acute, bilateral, descending, flaccid paralysis starting with 1 cranial nerve palsies in an IDU for whom no diagnosis other than botulism explained the illness. We defined a laboratory-confirmed case of WB as a clinical case for which the mouse bioassay serum test result was positive for botulinum toxin. Study cohort. The study cohort consisted of patients with illnesses that fit the case definition for which antitoxin was released and serum tested by the CDPH during Patients were selected from any reporting California public health jurisdiction except the jurisdiction overseen by the Los Angeles Department of Public Health, because the Los Angeles Department of Public Health performs its own antitoxin releases and serum testing. Materials. We collected botulism case report forms (BCRs) and laboratory reports for all California WB cases for which the CDPH authorized antitoxin releases during We collected hospital discharge summaries for the subset of patients with suspected WB for whose serum samples tested negative; we could not obtain discharge summaries for 4 serum toxin negative patients and excluded them from the analyses. All demographic and risk factor information was derived from BCRs. All clinical information for serum toxin positive patients was derived from BCRs. For serum toxin negative patients, clinical information was derived from BCRs; however, values for the variable required mechanical ventilation during hospitalization and the determination of death were derived from hospital discharge summaries. BCRs. For each antitoxin release, the California duty officer fills out a BCR that includes demographic, risk factor, and clinical information. The BCRs are filed with the CDPH, and copies are sent to local health jurisdictions. If a case is laboratory confirmed, the local health jurisdiction reviews the case, updates the BCR, and sends a finalized copy to the CDPH. Laboratory reports. The CDPH Microbial Diseases Laboratory (MDL) maintains records for all specimens received for botulism testing. Specimens are recorded as positive, negative, or inadequate for testing. All values related to specimens, including collection dates, serum results, and botulinum toxin types, were derived from laboratory reports. Hospital discharge summaries. Hospital discharge summaries for serum toxin negative patients were reviewed to determine whether a patient had required mechanical ventilation during hospitalization and whether the patient died. Summaries were reviewed for the presence or absence of botulism as 1 of the discharge diagnoses. Laboratory testing. All laboratory animal testing was regulated under the Institutional Animal Care and Use Committee and performed by the CDPH Microbial Diseases Laboratory. For each test, 3 pairs of mice were injected intraperitoneally with preparations of patient serum mixed with antitoxin A (pair 1), antitoxin B (pair 2), or no antitoxin (pair 3; the control pair). Injected mice were observed for up to 4 days for symptoms of botulism (wasp shape, labored abdominal breathing, weakness of the limbs, and ruffled fur) and/or death. If no pair exhibited typical symptoms or death, the test result was considered negative for C. botulinum toxin. Otherwise, the specimen was determined to be positive if a pair of mice was protected with 1 of the specific monovalent antitoxins (A or B) and the other mice showed typical symptoms and/or death. The botulinum toxin type was recorded as the type of monovalent antitoxin that was protective against symptoms or death. Statistical analyses. We calculated the sensitivity of the mouse bioassay as the proportion of clinical WB cases that had positive serum toxin test results by mouse bioassay. All serum toxin negative patients in the cohort were considered to have false-negative test results. Variables that were presented as check boxes on BCRs were analyzed as positive for patients for whom boxes were checked and negative for patients for whom the boxes were unchecked. We used SAS statistical software, version 9.1 (SAS Institute), to perform univariate analyses to compare serum toxin positive with serum toxin negative patients. We 1670 CID 2009:48 (15 June) Wheeler et al.

3 used x 2 tests to compare categorical variables and Wilcoxon 2- sample testing to compare continuous variables. For all calculations, we set the P value for statistical significance at.05. RESULTS Table 1. Comparison of serum toxin positive and serum toxin negative patients with suspected wound botulism, California Department of Public Health, Characteristic From 1 January 2005 to 31 December 2007, there were 129 patients with WB for whom the CDPH authorized antitoxin and for whom adequate serum specimens were provided for testing. Of these, 73 fit our case definition for having clinical WB, including 50 serum toxin positive and 23 serum toxin negative patients. The sensitivity of the mouse bioassay, therefore, was calculated as 68%. The remaining 56 patients (25 [45%] of whom were toxin positive) had many characteristics of WB but did not meet all the specific criteria of our clinical case definition. All patients in the study cohort were IDUs. Patients with positive serum samples did not differ significantly from those with negative serum samples on the basis of demographic or risk factor characteristics, including sex, age, Hispanic ethnicity, heroin use, or the presence of infected wounds (table 1). The time lapse in days from a patient s symptom onset to collection of serum for testing did not differ between serum toxin positive and serum toxin negative patients. In terms of clinical characteristics, most patients in both groups had similar proportions of ptosis, dysarthria, dysphagia, and dyspnea but differed in that serum toxin positive patients were more likely than serum toxin negative patients to have required mechanical ventilation during their hospital stay (74% vs. 43%; P p.01). One serum toxin positive patient and no serum toxin negative patients died. Five serum toxin positive patients and 1 serum toxin negative underwent electromyography. Of the 5 serum toxin positive patients, 2 had electromyography results consistent with botulism, 2 had inconclusive results, and 1 had a result that was not consistent with botulism. For the single serum toxin negative patient for whom electromyography testing was reported, results were described as negative for Guillain-Barré and myopathy. Of the 23 serum toxin negative patients, 17 (74%) had botulism or suspected botulism in the discharge diagnosis list of their discharge summaries. Of the 6 summaries in which botulism was not mentioned as one of the discharge diagnoses, 3 gave lists of cranial nerve palsies without any identified cause, 2 offered no diagnosis to explain the descending flaccid paralysis, and 1 listed respiratory failure secondary to muscular weakness secondary to hypokalemia as the final diagnosis for Serum toxin positive patients (np 50) Serum toxin negative patients (n p 23) P Male sex 38 (76) 17 (77) a.91 b Age, median years (range) 46 (29 61) 47 (23 60).60 c Hispanic ethnicity d 30 (60) 14 (61).94 b Injection drug user d 50 (100) 23 (100) Heroin user 49 (100) 21 (100) e Uses subcutaneous route of injection d 42 (84) 22 (96).26 f Infected wound 36 (75) g 11 (58) h.17 b Ptosis d 44 (88) 18 (78).28 b Dysarthria d 41 (82) 18 (78).71 b Dysphagia d 43 (86) 19 (83).73 f Alert and oriented d 30 (60) 14 (61).94 b Dyspnea d 33 (66) 12 (52).26 b Required mechanical ventilation during hospitalization 34 (74) 10 (43).01 b Median no. of days from onset of symptoms to collection of specimen (range) 3 (0 14) 3 (1 18).83 c NOTE. Data are no. (%) of patients, unless otherwise indicated. a One value with sex listed as transgender excluded. b Determined by x 2 test. c Determined by Wilcoxon 2-sample test. d The proportion given is number of patients for whom box was checked over the total number of patients. e Two missing values. f Determined by Fisher s exact test. g Two unknown values. h Two unknown values and 2 missing values. Mouse Bioassay Sensitivity in Wound Botulism CID 2009:48 (15 June) 1671

4 a patient who presented with ptosis, dysphagia, dyspnea, and bilateral, descending muscle weakness. DISCUSSION The mouse bioassay had a sensitivity of 68% for cases of WB that fit our specific clinical case definition. All cases occurred in IDUs presenting to hospitals in California. Serum toxin negative patients did not differ significantly from serum toxin positive patients by demographic characteristics, risk factors, or the interval in days between onset of symptoms and specimen collection. Likewise, serum toxin negative patients had similarly high proportions of cranial nerve palsies and dyspnea as serum toxin positive patients, except that they were significantly less likely to require mechanical ventilation during their hospital stays. Few underwent electromyography, and the reported results were not uniformly helpful in establishing a botulism diagnosis. Although serum toxin positive patients did not differ significantly by most clinical criteria from serum toxin negative patients, their greater propensity to require mechanical ventilation indicates more serious illness in the serum toxin positive group. It is likely that the severity of the illness reflects a greater load of circulating toxin in the serum toxin positive patients. However, although we know that the human lethal dose for type A botulinum toxin, the most common toxin type in WB, is estimated to be 1 mg/kg [12] approximately 50 times greater than the dose lethal to mice [13] we do not know the amount of toxin that will elicit symptoms in mice, and we cannot predict how much toxin will be circulating at the time that specimens are taken from the patient for testing. In WB, toxin that is produced at the infected tissue site by C. botulinum may be released intermittently and cleared from the bloodstream as it is taken up by nerve endings. The serum toxin negative patients in this cohort could not have any other disease but WB. They had a progressive neuroparalytic presentation that fit our rigorous and specific case definition for WB and had just as severe cranial nerve palsies and dyspnea as patients with laboratory-confirmed WB, except for less mechanical ventilation, suggesting slightly milder disease. In addition, we have documented instances in California in which serum toxin negative patients were confirmed for WB by mouse bioassay testing of wound tissue or tissue culture. The CDPH Microbial Diseases Laboratory has confirmed WB cases by tissue testing in 15 serum toxin negative patients since 1980 (CDPH Microbial Diseases Laboratory, unpublished data). Although comparison of the clinical course of serum toxin positive with serum toxin negative patients was beyond the scope of this study, there was no evidence in the data and materials available to us that the lack of a confirmatory WB test result adversely affected serum toxin negative patients. However, our cohort represents the fortunate subset of WB patients for whom WB was suspected and antitoxin administered. Prompt diagnosis is critical in patients with WB, because early medical attention, including administration of antitoxin, can result in better outcomes and shorter recovery times [14 16], and respiratory failure and death can occur rapidly in unattended patients. Therefore, the recognition of clinical WB in patients even when their test results are not confirmatory may facilitate early diagnosis in subsequent patients who present with similar demographic, risk factor, and clinical profiles. The documentation of WB in the diagnosis list of an IDU may facilitate recognition of a recurrence in that patient, who is prone to continue to engage in risky activities. Our review of the hospital discharge summaries for serum toxin negative patients suggests that some physicians may be hesitant in listing the diagnosis of WB in the absence of corroborating laboratory tests. Six summaries (26%) did not list botulism as one of the discharge diagnoses. Of these, none gave a diagnosis that explained the neurologic syndrome for which the patient was hospitalized. The study has some limitations. Our definition of a WB case was based on clinical criteria and depended on the reported observations of clinicians at emergency departments across California. Nonetheless, our case definition was specific and rigorous, and it is unlikely in the 23 serum toxin negative patients that the condition fitting the definition could have been explained by a disease entity other than WB. Our study was retrospective and based on surveillance data; thus, it is subject to some biases. For serum toxin positive patients, data were derived from BCRs that were reviewed and updated by local health departments, whereas for serum toxin negative patients, data were derived from duty officer-completed BCRs in combination with hospital discharge summaries. However, with regard to the analysis of patients symptoms, only BCRs were used, and the discrepancies between local health department reviewed and duty officer-completed BCRs would be minimal. The necessity for mechanical ventilation, in contrast, may have been more completely recorded in hospital discharge summaries than on BCRs because discharge summaries cover the full course of a patient s hospitalization. This inconsistency would have biased the study toward the null; however, our results showed a significantly greater propensity of patients requiring mechanical ventilation among serum toxin positive compared with serum toxin negative patients. Finally, we had scant information on electromyography, so we could not confirm that the test was performed specifically for botulism, which requires demonstrating action potential facilitation with repetitive stimulation at high frequencies. The fact that our study materials derive solely from California and mainly from the CDPH may raise questions about the generalizability of the results. However, the crucial finding was 1672 CID 2009:48 (15 June) Wheeler et al.

5 that the mouse bioassay failed to detect toxin in nearly onethird of patients with clinical WB. Although the accuracy of the bioassay may be laboratory dependent, it is unlikely to exceed that offered in this study because the CDPH Microbial Diseases Laboratory has performed more mouse bioassay WB testing than any other laboratory in the world. The Health Protection Agency in London, England, which now performs most WB mouse bioassays outside the United States, reports a serum toxin detection rate of 40% for specimens from IDUs suspected of having WB (K. Grant, personal communication). In this study, mouse bioassay detected botulinum toxin in only 68% of patients determined to have WB. Faster and moresensitive testing methods for botulism are urgently needed. Until such methods are developed, however, public health officials should consider counting cases of WB based on a positive mouse bioassay result or a clinical case definition. The patients suspected of having WB that fits the clinical case definition but without laboratory confirmation should be considered as having probable WB. Physicians should be aware of the test s limited sensitivity and base their final diagnosis of a suspected case of WB on clinical criteria when the mouse bioassay result is negative. From our experience in California, patients with histories of recent injection drug use, who present with sudden onsets of cranial nerve palsies and descending paralysis, are likely to have WB even if serum test results are negative. Acknowledgments We thank the clinicians and public health duty officers for their work with California WB patients. Financial support. California Department of Public Health. Potential conflicts of interest. All authors: no conflicts. References 1. Davis JB, Mattman LH, Wiley M. Clostridium botulinum in a fatal wound infection. JAMA 1951; 146: Werner SB, Passaro D, McGee J, Schechter R, Vugia DJ. Wound botulism in California, : recent epidemic in heroin injectors. Clin Infect Dis 2000; 31: Weber JT, Goodpasture HC, Alexander H, Werner SB, Hatheway CL, Tauxe RV. Wound botulism in a patient with a tooth abscess: case report and review. Clin Infect Dis 1993; 16: Passaro DJ, Werner SB, McGee J, Mac Kenzie WR, Vugia DJ. Wound botulism associated with black tar heroin among injecting drug users. JAMA 1998; 279: Centers for Disease Control and Prevention. National botulism surveillance. 9 January Available at: surveillance/botulism_surveillance.html. Accessed 8 January Centers for Disease Control and Prevention. Case definitions for infectious conditions under public health surveillance. MMWR Recomm Rep 1997; 46(RR-10): Merson MH, Dowell VR, Jr. Epidemiologic, clinical and laboratory aspects of wound botulism. N Engl J Med 1973; 289: MacDonald KL, Rutherford GW, Friedman SM, et al. Botulism and botulism-like illness in chronic drug abusers. Ann Intern Med 1985; 102: Maselli RA, Ellis W, Mandler RN, et al. Cluster of wound botulism in California: clinical, electrophysiologic, and pathologic study. Muscle Nerve 1997; 20: Hikes DC, Manoli A. Wound botulism. J Trauma 1981; 21: Cherington M, Ginsburg S. Wound botulism. Arch Surg 1975; 110: Arnon SS, Schechter R, Inglesby TV, et al. Botulinum toxin as a biological weapon: medical and public health management. JAMA 2001; 285: Schantz EJ, Johnson EA. Properties and use of botulinum toxin and other microbial neurotoxins in medicine. Microbiol Rev 1992;56: Chang GY, Ganguly G. Early antitoxin treatment in wound botulism results in better outcome. Eur Neurol 2003; 49: Sandrock CE, Murin S. Clinical predictors of respiratory failure and long-term outcome in black tar heroin-associated wound botulism. Chest 2001; 120: Tacket CO, Shandera WX, Mann JM, Hargrett NT, Blake PA. Equine antitoxin use and other factors that predict outcome in type A foodborne botulism. Am J Med 1984; 76: Mouse Bioassay Sensitivity in Wound Botulism CID 2009:48 (15 June) 1673