Streptomycin, an antimicrobial with limited availability, is the treatment of choice for plague, a fulminating and potentially epidemic disease that poses a bioterrorism concern. We evaluated the efficacy of gentamicin and tetracyclines for treating human plague. A medical record review was conducted on all 75 patients with plague who were reported in New Mexico during 1985–1999. Fifty patients were included in an analysis that compared streptomycin-treated patients (n = 14) with those treated with gentamicin and/or a tetracycline (n = 36). The mean numbers of fever days, hospital days, and complications and the number of deaths did not differ between patients treated with streptomycin and those treated with gentamicin. One patient who received tetracycline alone experienced a serious complication. Gentamicin alone or in combination with a tetracycline was as efficacious as streptomycin for treating human plague. The efficacy of a tetracycline alone could not be determined from the study.
Plague, a severe and often fatal disease responsible for 3 pandemics, has recently emerged as a possible microorganism that may be used by bioterrorists [1–3]. More than 400 cases of plague have been reported in the United States since 1950, and ∼1700 cases are reported each year worldwide . Along with smallpox and anthrax, plague is considered to be one of the greatest bioterrorist threats, and there is now renewed interest in its treatment and prevention and in controlling an outbreak .
Naturally occurring plague is endemic in New Mexico and surrounding southwestern states, where the enzootic cycle involves numerous species of rodents and small mammals and where humans frequently intrude into the natural foci of infection . More than 50% of all human cases of plague in the United States occur in New Mexico. Consequently, physicians there have greater experience in the diagnosis and treatment of plague than do physicians elsewhere in the country [5, 6]. Despite this, diagnosis is frequently delayed, which can lead to complications of illness and to deaths that could have been prevented [2, 7]. Furthermore, treatment regimens often do not follow standard guidelines. It is common for plague patients to be treated with ⩾3 antimicrobials during the course of their illness, which are often not approved by the US Food and Drug Administration (FDA) for this purpose.
Since 1948, streptomycin has been considered to be the drug of choice for bubonic, septicemic, and pneumonic plague [1, 8–13]. However, streptomycin is manufactured by 1 pharmaceutical company and is available only in modest supplies and by request . Alternative antimicrobial treatments for plague include chloramphenicol, tetracyclines, aminoglycosides other than streptomycin, fluoroquinolones, and various sulfonamides [1, 8, 15]. The rationale for the use of these alternative antimicrobials to treat human plague has been based on results of in vitro evaluations, trials in experimental animals, and varying clinical experiences [16–22]. However, no systematic evaluations of efficacy and safety have been conducted in humans. Chloramphenicol is often recommended for use in cases of plague meningitis because of its ability to cross the blood-brain barrier .
Recently, there has been increased interest in the use of gentamicin, doxycycline, and ciprofloxacin against plague, particularly for treatment or prophylaxis in response to a bioterrorist event [1, 23–25]. Of these 3 antimicrobials, only doxycycline is labeled for use in plague. Gentamicin is attractive as an alternative treatment because it is readily available and can be used intravenously in a single daily dose. Doxycycline has rapid and reliable gastrointestinal absorption, relatively mild side effects, and is easily administered. Ciprofloxacin offers the advantages of bactericidal and intracellular actions and ease of administration. Use of each of these 3 antimicrobials has been recommended by the Working Group on Civilian Biodefense in the event of plague bioterrorism .
Trimethoprim-sulfamethoxazole is recommended for plague prophylaxis when tetracyclines are contraindicated, but no sulfonamide is approved by the FDA for the treatment of plague . Other antimicrobials that are not currently recommended for plague treatment are penicillins and β-lactams [8, 26]. According to expert consensus, penicillins and cephalosporins are considered to be ineffective against plague, although in vitro and animal studies have shown that ceftriaxone has variable activity [16, 18, 27–29].
We compare the results of various regimens of antimicrobials used to treat recent human plague in New Mexico. It is, perhaps, surprising that this review provides the first retrospective, population-based evaluation of the clinical efficacy of gentamicin and tetracyclines against plague.
Patients and Methods
Case definitions and study population. We reviewed New Mexico Department of Health clinical and laboratory records for all cases of human plague contracted in New Mexico from 1 January 1985 through 31 December 1999. Using Centers for Disease Control and Prevention criteria, we defined cases as either “confirmed” or “probable” . A confirmed case was defined as a clinical illness consistent with plague plus either isolation of Yersinia pestis from a clinical specimen or a 4-fold increase in serum antibody titer to Y. pestis. A probable case was defined as a clinical illness consistent with plague plus either a positive direct florescent antibody stain of a clinical specimen that was positive for Y. pestis or an elevated serum antibody titer to Y. pestis in a patient with no history of plague vaccination.
Patients who met the case definition for plague were further categorized as having 1 of 3 clinical plague syndromes. Bubonic plague was diagnosed in patients with a swollen and tender lymph node or bubo. Primary pneumonic plague was diagnosed in patients who acquired the disease by the respiratory route. Septicemic plague was diagnosed when patients without buboes or primary pneumonic plague otherwise met the case definition . Complications included systemic inflammatory response syndrome (SIRS), meningitis, and secondary pneumonia.
Fever was defined as a documented temperature of >38.0°C (>100.4°F), and a fever day was defined as a 24-h period, starting at midnight, during which a patient had a documented fever. A hospital day was defined as any part of a 24-h day, starting at midnight, during which a patient was hospitalized. The day of defervescence was defined as the first day during which a patient had no documented fever.
Analysis. Patients who received streptomycin or an alternative antimicrobial regimen (gentamicin and/or a tetracycline) for plague were included in the analysis. A total of 4 antimicrobial treatment groups were compared: those who received streptomycin, gentamicin alone, gentamicin-tetracycline, or a tetracycline alone. Any patient who received streptomycin in combination with any other antimicrobial was categorized as a member of the streptomycin group.
No analysis was performed that involved patients treated with fluoroquinolones, because only a few patients had received these antimicrobial drugs. All patients who received sulfonamides and β-lactam antimicrobials alone or simultaneously with alternative antimicrobials were excluded from further analysis. However, patients who were initially treated with a β-lactam or sulfonamide without showing signs of clinical improvement and whose therapy was changed to one of the alternative antimicrobial regimens or streptomycin were included in 1 of the 4 treatment groups listed above in the analysis.
Antimicrobials that were initiated >24 h after a patient defervesced were excluded from the analysis. Antimicrobials that were prescribed as a single dose or as multiple subtherapeutic doses were excluded from the analysis.
Patient characteristics, including age, race, and sex, were determined for the 4 treatment groups. To assure that the 4 groups were comparable, measures for severity of illness were determined. The number of symptom days before the initiation of antibiotic treatment was used as a measure for severity of illness because previous reviews of plague have identified an association between delayed initiation of treatment and increased mortality [2, 7]. Other severity of illness measures used in the analysis included the diagnosis of septicemic plague, hospitalization, and complications (SIRS, meningitis, and secondary pneumonia) before the initiation of treatment.
Statistical analysis was performed using JMP IN software (SAS Institute). Analysis of covariance (ANCOVA), adjusting for patient characteristics and severity of illness markers, was used to compare treatment groups according to the number of fever days after the initiation of antimicrobial treatment and the number of hospitalization days. Because of the possibility of shorter hospital stays as a result of the effects of managed care in the 1990s, we adjusted for the date of hospitalization in the ANCOVA that compared the number of hospital days.
Complications and deaths after the initiation of antimicrobials were described for all patients who received streptomycin or an alternative regimen. Deaths that involved patients who were not included in the final analysis also were described.
Demographic and clinical characteristics. Seventy-five cases of human plague were reported in New Mexico during 1985–1999, including 70 confirmed and 5 probable cases. Clinical syndrome and treatment data were available for 74 of 75 cases, including 71 complete and 3 partial records.
Patients presented with symptoms typically associated with plague, including fever (99%), painful bubo (56%), gastrointestinal complaints (53%), myalgias (24%), headache (21%), altered mental status (12%), cough (11%), and pharyngitis (9%). In the entire review of 75 cases, a mean of 2.5 antimicrobials (range, 1–6) were used to treat each patient. Fifty patients were identified who were treated with either streptomycin or an alternative regimen; 14 patients were in the streptomycin group, 18 were in the gentamicin group, 10 were in the gentamicin-tetracycline group, and 8 were in the tetracycline group. Twenty-four patients received other treatment with chloramphenicol, β-lactam, or sulfonamide antimicrobials alone or simultaneously with alternative antimicrobials; data were missing for 1 patient. Data for these latter 25 patients were excluded from the comparative analysis.
Of the 75 patients, only 3 were treated with fluoroquinolones. The first patient was treated successfully with a third-generation cephalosporin simultaneously with ciprofloxacin, but data for this patient were excluded from the comparative analysis because the patient received a β-lactam simultaneously with an alternative antimicrobial. Fluoroquinolone therapy was initiated for the remaining 2 patients only after they had defervesced while receiving other antimicrobial regimens, and, consequently, both patients were excluded from analysis.
Demographic and clinical characteristics for the 50 patients in the comparative analysis were evaluated (table 1), and they were similar to those described in other case series in the literature [1, 2, 31–35]. Measures for severity of illness also are presented in table 1. Patients included in the comparative analysis experienced an average of 4.2 days of symptoms before the initiation of treatment. Forty-two (84%) of the 50 patients had bubonic plague, and 8 (16%) had primary septicemic plague. There were no cases of primary pneumonic plague. Cases of plague in New Mexico were similar to those reported in the United States during 1947–1996, of which 84% were bubonic, 13% septicemic, and 2% primary pneumonic . Forty-one (98%) of 42 patients in the streptomycin, gentamicin-tetracycline, or gentamicin groups were hospitalized, and 3 (38%) of 8 of patients in the tetracycline group were hospitalized.
A number of patients presented for medical care already having developed complications of plague. One patient in the streptomycin group and 2 patients in the gentamicin group had developed SIRS before antimicrobials were administered. One patient from each of the streptomycin, gentamicin, and gentamicin-tetracycline groups had developed secondary pneumonic plague before the initiation of antimicrobial treatment.
Eleven patients who initially did not respond to treatment with β-lactam or sulfonamide antimicrobials were then successfully treated with streptomycin or an alternative regimen (table 1). These included 5 cases of third-generation cephalosporin failure. Treatment with streptomycin (n = 1), gentamicin (n = 3), or gentamicin-tetracycline (n = 1) was later provided. The remaining 6 cases included 3 patients who initially did not respond to treatment with a first-generation cephalosporin and for whom streptomycin (1 patient), gentamicin-tetracycline (1), and tetracycline (1) was subsequently provided; 1 who initially did not respond to second-generation cephalosporin treatment and subsequently received gentamicin; 1 who initially did not respond to fourth-generation penicillin treatment and subsequently received gentamicin-tetracycline; and 1 who initially did not respond to sulfonamide therapy and subsequently received tetracycline therapy.
Clinical Outcome Data
Mean time to defervescence. The mean number of fever days after the initiation of antimicrobial treatment was 3.5 days for patients in the streptomycin group, 2.6 days in the gentamicin group, 1.9 days in the gentamicin-tetracycline group, and 2.6 days in the tetracycline group (table 2). ANCOVA that adjusted for age, race, sex, number of symptom days before antimicrobial therapy, septicemic plague, complications before the initiation of treatment with streptomycin or an alternative regimen, and prior treatment with a β-lactam or sulfonamide antimicrobial revealed no significant difference between the 4 treatment groups with respect to the number of fever days after initiating therapy (adjusted P = .23). Patients with complications before the initiation of antimicrobial treatment had a significantly greater number of fever days than did those without such complications (4.0 vs. 2.5 days, adjusted P = .033).
Hospitalization duration after the initiation of antimicrobial therapy. Because 3 of 8 patients treated only with a tetracycline were hospitalized, the tetracycline group was not included in ANCOVA that compared hospitalization duration. The mean duration of hospitalization was 6.2 days in the streptomycin group, 7.2 days in the gentamicin group, and 6.0 days in the gentamicin-tetracycline group. ANCOVA that adjusted for age, race, sex, number of symptom days before initiation of antimicrobial therapy, complications before the initiation of treatment with established or alternative antimicrobials, septicemic plague, year of hospitalization, and prior treatment with a β-lactam or sulfonamide antimicrobial showed no significant differences between the 3 groups (P = .57).
Deaths. There were no deaths among the 50 patients in the 4 treatment groups. Among the 25 patients who were excluded from the analysis, there were a total of 4 deaths. Three of these deaths occurred in septicemic patients who received penicillin therapy only, and 1 death involved a patient who presented to the hospital with SIRS and died within 24 h of receiving chloramphenicol therapy.
Complications after the initiation of therapy. There was 1 complication due to treatment failure involving a tetracycline. The patient was thought to have bubonic plague and was treated first as an outpatient with tetracycline. His symptoms worsened, and he was admitted to the hospital. Treatment was continued with oral tetracycline and gentamicin, although blood levels of gentamicin were subtherapeutic. On the sixth hospital day, he developed meningitis. The antimicrobial treatment was changed to therapeutic levels of chloramphenicol and gentamicin, and he recovered. The patient was excluded from the final comparative analysis, because he was treated with chloramphenicol.
Incidentally, in our review of 75 cases, we found that 6 patients were treated with chloramphenicol only. When these patients were added to the analysis, there were no significant differences between patients treated with chloramphenicol, streptomycin, gentamicin, and tetracyclines with regard to the number of fever days and duration of hospitalization. These data, however, are not reported in tables 1 or 2, because of the small number of patients treated with chloramphenicol. One patient who received chloramphenicol died, but this patient presented with SIRS and died within 24 h after the initiation of treatment.
Y. pestis, the bacterium that causes plague, is considered to be a category A critical biological agent of terrorism because it has high potential for morbidity and mortality and because it can be mass-produced and easily disseminated through aerosolization among a large population [1, 36, 37]. There also is efficient person-to-person transmission of the organism, and the fatality rate of primary pneumonic infection is close to 100% if not treated within 24 h [1, 8, 37]. In simulated bioterrorist attacks, the World Health Organization estimated that 36,000 people would die after an attack on a city of 5,000,000 people, and the US Department of Justice estimated that 1000 would die after an attack at a rock concert [38, 39].
With the use of appropriate antimicrobial drugs, the mortality rates for bubonic and septicemic plague decrease from 50% to 5%–15% [6, 8, 9, 26]. To limit mortality during a bioterrorist attack with plague, it will be necessary to have effective antimicrobials that are readily available and easily administered. With naturally occurring plague, most fatalities in the United States are associated with a delay in proper diagnosis and treatment [2, 7]. This reinforces the importance of empirical treatment of septic patients in areas of plague endemicity with appropriate antimicrobial therapy, because the diagnosis of plague is often difficult to make .
Streptomycin, currently considered to be the drug of choice for the treatment of plague, has very limited availability. The Working Group on Civilian Biodefense has recommended both gentamicin and tetracyclines for the treatment of plague during a bioterrorist event, because of the availability and ease of administration of these drugs .
This study presents the first population-based evaluation showing the clinical efficacy of gentamicin in the treatment of human plague. In our study, patients with plague who were treated with an alternative antimicrobial regimen (gentamicin alone or with a tetracycline) had similar clinical outcomes to those patients treated with streptomycin. There were no significant differences between patients who received streptomycin versus alternative antimicrobials with respect to the number of fever days, complications, and deaths, and the duration of hospitalization.
On the basis of our data, however, it is difficult to comment on the efficacy of a tetracycline alone as sufficient for the treatment of plague. Patients who were treated with tetracyclines were likely to be less severely ill than patients in other groups, because few were hospitalized and none had complications before the initiation of antimicrobial therapy. Furthermore, another patient's illness failed to respond to treatment with tetracycline; this patient developed plague meningitis and was subsequently treated with chloramphenicol.
The number of patients treated with chloramphenicol, trimethoprim-sulfamethoxazole, cephalosporins, and fluoroquinolones was extremely limited in our review, and we therefore can make no recommendations regarding the efficacy of these alternative antimicrobials for the treatment of human plague. We confirm that penicillins are ineffective and should not be used in the treatment of plague. Three out of the 4 deaths noted in this review involved patients who were treated with penicillins. This conclusion reinforces previously published observations [8, 26].
The limitations of the study are as follows. This was a small study (50 patients), and patients were treated with complicated combinations of antimicrobials, often initially with 1 class of antimicrobial and later with 2–3 different classes of antimicrobials. In our analysis, we made the assumption that patients whose illness failed to respond to initial treatment with a β-lactam or sulfonamide antimicrobial could be grouped with the antimicrobial regimen that was used to treat them successfully.
Although we evaluated and excluded antimicrobials that were administered at subtherapeutic doses, we did not evaluate the frequency of antimicrobial administration. We also did not control for the possible effect of antipyretics on the number of fever days.
No patients in our study experienced primary pneumonic plague, and thus the generalizability of our findings to patients with this clinical syndrome could be questioned. This point is particularly relevant when one considers that a bioterrorist event likely would cause a pneumonic plague outbreak . In addition, wild Y. pestis strains may not be comparable with strains used in a bioterrorist event, especially considering the possible use of multidrug-resistant strains of Y. pestis.
This analysis suggests that gentamicin alone or with a tetracycline is as efficacious as streptomycin for the treatment of human plague. The efficacy of a tetracycline alone cannot be determined from this study. Additional prospective studies are needed to confirm these findings.
We are grateful to Dr. Ed Bedrick and Kyle Smith, for assistance with statistical analysis; Dr. William Wiese, for reviewing the article at the manuscript stage; and Martha Tanuz, for input and assistance with data collection.