Prognostic Factors and Antibiotics in Vibrio vulnificus Septicemia

Background  Immunocompromised patients with Vibrio vulnificus septicemia are at high risk for fatality. When a hemorrhagic bullous necrotic cutaneous lesion (HBNCL) and decreased blood pressure develop, approximately 50% of V vulnificus septicemic patients die within 48 hours. This study aimed to evaluate the risk factor(s) for fatality among patients with V vulnificus septicemia, emphasizing the role of prescribed antimicrobial agents in general and the therapeutic efficacy of the combination of a third-generation cephalosporin and tetracycline or its analogue in particular.

Methods  Patients with the diagnosis of V vulnificus infection admitted to 5 large medical centers in Taiwan between 1995 and 2003 were included in this retrospective study. Patients were divided into 2 groups: those with HBNCLs and those without HBNCLs. Patients were further divided into subgoups without fatalities (fatal subgroup) and those without fatalities (nonfatal subgroup).

Results  A total of 93 patients participated in the study. In group 1, the fatal subgroup had higher Acute Physiology and Chronic Health Evaluation II (APACHE II) scores (P = .006) and a higher proportion of shock at arrival at the medical center (P = .015) than the nonfatal subgroup. In group 2, the effect of a first- or second-generation cephalosporin plus an aminoglycoside was negative (P = .01) and that of combined third-generation cephalosporin and tetracycline or its analogue was positive (P<.001); significant differences were found between the fatal and nonfatal subgroups in the APACHE II score (P<.001), number who were in shock at arrival at the medical center (P = .02), delayed surgical intervention (P = .03), and peripheral leukocytosis (P = .03). Shock at arrival at the medical center (odds ratio [OR], 19.25; 95% confidence interval [CI], 1.768-209.54; P = .02) was an independent risk factor for fatality in patients without HBNCLs. Use of a third-generation cephalosporin and tetracycline or its analogue significantly reduced fatality rates in patients with HBNCLs (OR, 0.037; 95% CI, 0.007-0.192; P<.001).

Conclusion  Septic shock is a determinant of fatality in patients with V vulnificus septicemia without HBNCLs; our data suggest that the combination of a third-generation cephalosporin and tetracycline or its analogue may be a better choice in antimicrobial treatment of V vulnificus septicemic patients with HBNCLs.

Vibrio vulnificus is a halophilic gram-negative bacillus recovered from marine and brackish environments.¹ It was first identified as a new Vibrio species pathogenic for humans in 1976.²Vibrio vulnificus is distributed worldwide, favorably growing in water with temperatures above 20°C and salinities between 0.5% and 2.5%.³ A large number of cases of infection caused by V vulnificus have been reported from a limited number of geographic areas, such as the gulf coastal communities of the United States⁴ and Taiwan^5,6; this may in part result from environments favorable to the growth of V vulnificus in these geographic localities. The incidence of infections caused by V vulnificus is on the rise in both the United States and Taiwan.^7,8 Almost all patients with sepsis due to V vulnificus are immunocompromised, and the common underlying conditions that lead to their immunoincompetence include viral hepatic diseases (hepatitis B or C or virus-related cirrhosis), alcoholic liver disease, diabetes mellitus, and steroid use.^5,8,9 Primary septicemia and soft tissue infection are 2 common clinical manifestations of V vulnificus infections^5,6,8-11; other occasionally reported clinical manifestations include gastroenteritis,³ pneumonia,¹² and spontaneous bacterial peritonitis.¹³ The portal of entry in patients with V vulnificus septicemia may be injured soft tissue or the gastrointestinal tract. The typical picture of severe V vulnificus septicemia in immunocompromised hosts is an abrupt onset of fever and chills, followed by decreased blood pressure and development of metastatic cutaneous lesions, which rapidly evolve into hemorrhagic bullae and then necrotic cutaneous ulcers.^5,11,14 More than 50% of V vulnificus septicemic patients with complicated hemorrhagic bullous necrotic cutaneous lesions (HBNCLs) and decreased blood pressure die, and the median interval from the time of admission to death is approximately 2 days.^5,11 A characteristic HBNCL in a patient with V vulnificus septicemia is illustrated in Figure 1.

Most V vulnificus isolates are susceptible in vitro to a great variety of antibiotics.^15,16 As a result, great arrays of antibiotics have been administered to treat infections caused by V vulnificus based on the in vitro susceptibility testing of this pathogen.^5,16 In 1983, Bowdre et al¹⁷ reported a study using a murine model of V vulnificus septicemia induced by intraperitoneal inoculation of the culprit bacteria. The results of that study indicated that tetracycline was superior to cefotaxime and thereby the drug of choice for V vulnificus infections. However, later observations in Taiwan suggested that a third-generation cephalosporin might be clinically superior to tetracycline in the treatment of infections due to this pathogen.^5,18 Sanford et al¹⁹ proposed a combination of tetracycline and a broad-spectrum cephalosporin for the treatment of V vulnificus infections; because the proposal was not based on solid evidence, whether it works in improving the unacceptably high mortality rate in patients with V vulnificus sepsis is uncertain. Ensuing reports^16,20 from Taiwan, in which necrotizing soft tissue was created by inoculating the bacteria in the thighs of the animals, have clearly demonstrated the in vitro synergism between cefotaxime and minocycline against V vulnificus and the superiority of these combined antibiotics compared with either agent alone in the treatment of experimental murine V vulnificus infection. Since the late 1990s, it has been common practice in Taiwan to prescribe a third-generation cephalosporin alone or in combination with tetracycline or its analogue for patients with V vulnificus septicemia.⁸ However, clinical-based evidence that supports the superiority of the combination of a third-generation cephalosporin and tetracycline is not available. To elucidate the important information regarding the clinical efficacies of such combined antibiotics, retrospective analyses of patients with V vulnificus septicemia diagnosed at various centers in Taiwan were performed. The objective of this study was to identify the risk factor(s) for fatality in patients with V vulnificus septicemia, emphasizing the role of prescribed antimicrobial agents in general and the therapeutic efficacy of the combination of a third-generation cephalosporin and tetracycline or its analogue in particular. The information from this series may be valuable in improving treatment of severe V vulnificus infections, thereby reducing the fatality rate.

Patients with a diagnosis of V vulnificus infection admitted to 5 large medical centers in Taiwan between 1995 and 2003 were included in this retrospective study. These medical centers and their capacities are as follows: Chang Gung Memorial Hospital–Kaohsiung Medical Center (2300 beds), Chi-Mei Medical Center (1325 beds), Kaohsiung Veterans General Hospital (1100 beds), National Cheng Kung University Hospital (1000 beds), and National Taiwan University Hospital (1800 beds). Patients with V vulnificus infections were identified from the records of the clinical microbiology laboratories of the participating medical centers.

All included septicemic patients fulfilled the criteria of sepsis, as previously described.²¹ Staff members of these clinical microbiology laboratories in Taiwan where V vulnificus infections are endemic were experienced in identifying this pathogen. Each V vulnificus isolate was a halophilic gram-negative rod identified by test results positive for cytochrome oxidase, glucose fermentation, citrate use, indole production, ornithine decarboxylase, and hydroxylase of ortho-nitrophenyl galactoside.¹ The V vulnificus isolates identified by conventional methods were further verified by one of the following automated detection systems: API-20E System (bioMérieux Vitek Inc, Hazelwood, Mo), ID 32 GN System (bioMérieux Vitek Inc), and Vitek 2 ID-GNB identification card (bioMérieux Inc, Durham, NC).

The medical records of the included patients were reviewed, and their demographic, clinical, and laboratory information was retrieved and collected for analysis. Most patients with severe V vulnificus septicemia who presented with a distinctive HBNCL have an unequivocal history of recent exposure to saltwater or marine creatures or consumption of raw or undercooked seafood, which is explicitly suggestive of V vulnificus infections, and patients with HBNCLs are subject to rapid clinical deterioration and high risk of fatality^5,11; therefore, establishing the effective antibiotic regimen for a timely initiation of therapy for V vulnificus sepsis cannot be overemphasized. Surgical debridement may be additionally indicated for patients with HBNCLs when necessary. Based on this rationale, the included V vulnificus septicemic patients were separated into the following groups for further analyses: patients without HBNCLs (group 1) and patients with HBNCLs (group 2).

Variables in group 1 and group 2 were compared with each other using univariate analyses. To disclose the prognostic factors, including the antimicrobial modality for fatality in each group, patients in group 1 and group 2 were further divided into a subgroup with fatalities and one without (fatal and nonfatal subgroups, respectively). Within the same group, demographic, clinical, and laboratory data of patients in the fatal and nonfatal subgroups were compared with each other using univariate analyses. In univariate analyses, the χ² test or Fisher exact test was used for comparison of dichotomous variables, whereas the t test or Mann-Whitney U test was used for comparison of continuous variables when applicable. Statistically significant differences between the fatal and nonfatal subgroups in group 1 and group 2 in univariate analyses were separately entered into a multiple logistic regression model to identify independent prognostic factor(s) in each group. A 2-tailed P<.05 was considered statistically significant. All comparisons were performed using the SPSS software package, version 11.0 (SSPS Inc, Chicago, Ill).

In total, 93 V vulnificus septicemic patients were included in the study. Some of these patients had additional specimens obtained from infection sites other than blood (mostly from soft tissue) that were culture positive for this culprit pathogen. The demographic, clinical, and laboratory information of the included patients is summarized in Table 1. Among the included patients, males were predominant (male-female ratio = 67:26), and most were elderly (mean ± SD age, 62.2 ± 13.0 years) and immunocompromised (94%). Liver cirrhosis (47%), steroid use (25%), and diabetes mellitus (23%) were the 3 leading underlying conditions that rendered these patients immunocompromised. Thirty patients (32%) did not develop HBNCLs (group 1), whereas 63 (68%) did (group 2). Between groups 1 and 2, the differences in septic shock at arrival (11 [37%] vs 44 [70%]; P = .003) and leukopenia (7 [23%] vs 4 [6%]; P = .03) were statistically significant. In total, 31 patients died, accounting for an overall mortality rate of 33%. Although the same mortality rate was found in both groups, a significantly higher early (within 48 hours after arrival) mortality rate (30% vs 10%; P = .04) was found in group 2.

A variety of antibiotics, including penicillins, cephalosporins, tetracycline and its analogues, aminoglycosides, clindamycin, and ciprofloxacin, were prescribed to these patients. A significantly higher proportion of a combination of a third-generation cephalosporin (ceftazidime, cefotaxime, ceftriaxone, or moxalactam) and either tetracycline or its analogue (minocycline, doxycycline, or oxytetracycline) was prescribed for patients in group 2 (48% vs 7%; P<.001). Fatalities among V vulnificus septicemic patients who received various antimicrobial treatments are shown in Figure 2.

Comparisons between the fatal and nonfatal subgroups within group 1 and group 2 are given in Table 2. In group 1, the fatal subgroup had significantly higher Acute Physiology and Chronic Health Evaluation II (APACHE II) scores²² (median, 21.0 vs 11.0; P = .006) and a higher proportion with septic shock at arrival at the medical center (70% vs 20%; P = .02); multivariate analysis disclosed that shock at arrival at the medical center (odds ratio [OR], 19.25; 95% confidence interval [CI], 1.768-209.54; P = .02) was an independent risk factor for fatality in patients without HBNCLs. Between the fatal and nonfatal subgroups of group 2, significant differences were found in APACHE II score (median, 20.0 vs 12.5; P<.001), septic shock at arrival at the medical center (19 [91%] vs 25 [60%]; P = .02), delayed (later than 24 hours after arrival) surgical intervention (10 [67%] vs 11 [31%]; P = .03), and peripheral leukocytosis (4 [19%] vs 20 [48%]; P = .03), as well as use of the combination of a first- or second-generation cephalosporin plus an aminoglycoside (gentamicin, tobramycin, netilmicin, or amikacin) (24% [5/21] vs 2% [1/42]; P = .01) and use of a combination of a third-generation cephalosporin and tetracycline or its analogue (14% [3/21] vs 64% [27/42]; P<.001; see Figure 2). Of note, the effect of a combination of a first- or second-generation cephalosporin and an aminoglycoside was negative. Multivariate analysis disclosed that use of a combination of third-generation cephalosporin and tetracycline or its analogue was an independent factor (OR, 0.037; 95% CI, 0.007-0.192; P<.001) for lower mortality in patients with HBNCLs.

The predominance of male sex and old age, as well as immunoincompetence, in most V vulnificus septicemic cases in this report is consistent with previously published studies.^5,11 In agreement with other series,^11,14 chronic liver diseases in general and cirrhosis of the liver in particular were commonly encountered underlying diseases. Animal studies²³ disclosed that iron could accelerate the growth of V vulnificus to reach a lethal level with enhanced cytotoxicity in iron-overload mice. One study²⁴ of V vulnificus in whole blood from patients with hepatoma, cirrhosis of the liver, and a varied degree of chronic hepatic inflammation caused by hepatitis B or hepatitis C virus showed that high serum ferritin levels and low phagocytosis activity of neutrophils were independent predictors of survival of this microbe in blood. Another study²⁵ disclosed that when incubated with live V vulnificus, significantly lower levels of proinflammatory cytokines (including interleukin [IL] 1β, IL-6, IL-8, and tumor necrosis factor α) were produced by peripheral blood mononuclear cells of individuals with presumed chronic alcoholic liver disease, which resulted from cellular oxidative stress reflected by reduced glutathione in the peripheral blood mononuclear cells. Interleukin 1β, IL-6, and tumor necrosis factor α were important proinflammatory cytokines in innate immune response in V vulnificus–infected patients.²⁶ These findings indicated that patients with chronic liver diseases due to either chronic viral hepatitis or alcoholism are at high risk for V vulnificus septicemia.

Because illegal steroid-containing herbal drugs have long been used in Taiwan,^27-29 it is not surprising that steroid use was found to be the second leading cause (cirrhosis of the liver being the most frequent cause when chronic liver diseases was subclassified as indicated in Table 1) of immunoincompetence in this series, which was unique when compared with other reports.^9,11 The high mortality rate of sepsis made health authorities in some areas endemic for V vulnificus infections implement regulations that mandate posting warning signs concerning risks of consuming raw oysters by vulnerable people^30,31; however, one study³¹ disclosed that this is a far from effective strategy in prevention of V vulnificus infections. Once an immunocompromised person is infected with V vulnificus, the culprit pathogen produces a variety of toxins that trigger vigorous septic response in the host. The well-known toxins generated by V vulnificus include capsular polysaccharides,^32,33 metalloprotease,³⁴ lipopolysaccharides,^26,35 and cytolysin.^36,37 When rapid suppression of production of these toxins by swiftly eliminating the invading culprit V vulnificus fails, the affected immunocompromised patients are destined to experience clinically fulminant sepsis and are at high risk of mortality. Previously reported susceptibility testing on clinical V vulnificus isolates indicated that the minimum inhibitory concentrations (MIC_90%) for cephalothin and cefamandole were both 4 μg/mL,^15,38 whereas the MIC_90% for cefotaxime and ceftriaxone were both 0.03 μg/mL.^15,16,38 Using the interpretative criteria for Enterobacteriaceae recommended by the National Committee for Clinical and Laboratory Standards, V vulnificus isolates were susceptible in vitro to cephalosporins of all generations.^15,16 However, the MIC_90% of a first-generation cephalosporin or a second-generation cephalosporin is at least 130-fold higher than that of a third-generation cephalosporin.^8,16,38 When there is a low V vulnificus burden and a high concentration of the administered antibiotic achievable in the infected site, effective killing of V vulnificus by a first-generation or second-generation cephalosporin is theoretically possible. Therefore, in V vulnificus septicemic patients without HBNCLs, uses of various antibiotic regimens did not lead to significant differences in mortality rates. The prognostic factor for fatality in patients without HBNCLs is the presence of septic shock, which is reflective of the clinical severity of sepsis as shown in this report.

However, in the scenario of V vulnificus infection with HBNCLs, the collagen and elastic fibers degenerate, muscle cells become necrotized, and blood vessels become congested.^39,40 The blood supply will be seriously compromised in such a histopathological milieu, and a high tissue antibiotic level can hardly be expected. Huge numbers of V vulnificus embedded in the inflammatory and devitalized soft tissue make the situation worse.^39,40 The low concentrations of different administered antibiotics that separately reach the ongoing inflammatory site may act synergistically against the V vulnificus as suggested by in vitro and murine experiments.^16,20 Previously published experiments disclosed the superiority of combined cefotaxime and minocycline over either one used alone in severe soft tissue infection caused by V vulnificus, and in these experiments the severity of sepsis was proportional to the quantity of bacteria inoculated.²⁰ Animals used in the aforementioned experiments were immunocompetent.²⁰ Our study is consistent with the in vitro and experimental data and suggests that this antibiotic combination remains effective in immunocompromised human hosts. Although in vitro efficiency of cefotaxime and ciprofloxacin was recently reported to be superior to that of cefotaxime and minocycline,⁴¹ an in vivo study is needed to determine if the combination of cefotaxime and ciprofloxacin is justified for clinical trial.

The limitations of this retrospective study are that the performance of surgical debridement depended on an internist's decision regarding whether a consultation with a surgeon was needed and was in turn at the discretion of the consulted surgeon; as a consequence, the decision regarding debridement or the timing of debridement in the event that an operation was scheduled was not made based on standardized criteria. However, because the in vitro and in vivo experiments clearly demonstrated the superiority of the combination of a third-generation cephalosporin with tetracycline or its analogue over either one used alone in the treatment of severe V vulnificus infections, which carry a high chance of fatality,^16,20 it is, based on ethical considerations,⁴² no longer feasible to conduct a prospective randomized clinical study to determine the efficacy or superiority of the combination of a third-generation cephalosporin and tetracycline or its analogue. Therefore, the data disclosed in this study are extremely important in determining the definite treatment of severe V vulnificus septicemia or the empirical treatment of suspected V vulnificus septicemia.

In conclusion, septic shock is an independent risk factor for fatality in V vulnificus septicemic patients without HBNCLs; our data suggest that combination of a third-generation cephalosporin and tetracycline or its analogue may be a better choice in antimicrobial treatment for V vulnificus septicemic patients with HBNCLs.

Correspondence: Yin-Ching Chuang, MD, Department of Medical Research, Chi-Mei Medical Center, 901 Chung-Hwa Rd, Yung-Kang City, Tainan, Taiwan (chuangkenneth@hotmail.com).

Accepted for Publication: July 13, 2006.

Author Contributions:Study concept and design: J.-W. Liu and Chuang. Acquisition of data: Ko, H.-C Lee, Y.-C Liu, J.-W. Liu, I.-K. Lee, Tang, and Chuang. Analysis and interpretation of data: J.-W. Liu, I.-K. Lee, Hsueh, and Chuang. Drafting of the manuscript: J.-W. Liu and Chuang. Critical revision of the manuscript for important intellectual content: J.-W. Liu, I.-K. Lee, Tang, Ko, H.-C. Lee, Y.-C Liu, Hsueh, and Chuang. Statistical analysis: J.-W. Liu and I.-K. Lee. Obtained funding: Chuang. Administrative, technical, and material support: I.-K. Lee, Tang, Hsueh, and Chuang. Study supervision: Chuang.

Financial Disclosure: None reported.