Clinical Characteristics and Molecular Subtyping of Vibrio vulnificus Illnesses, Israel

The genetically distinct biotype 3 has penetrated Israeli freshwaters and is causing severe illness in persons who handle tilapia or carp.


CME ACTIVITY
Medscape, LLC is pleased to provide online continuing medical education (CME) for this journal article, allowing clinicians the opportunity to earn CME credit. Medscape, LLC is accredited by the Accreditation Council for Continuing Medical Education (ACCME) to provide CME for physicians. Medscape, LLC designates this educational activity for a maximum of 0.75 AMA PRA Category 1 Credits™. Physicians should only claim credit commensurate with the extent of their participation in the activity. All other clinicians completing this activity will be issued a certificate of participation. To participate in this journal CME activity: (1) review the learning objectives and author disclosures; (2) study the education content; (3) take the post-test and/or complete the evaluation at http://www.medscape.com/cme/eid; (4) view/print certifi cate.

Learning Objectives
Upon completion of this activity, participants will be able to: List the predisposing factors for infections caused by During 1996-1997, a new biotype, V. vulnifi cus biotype 3, emerged as a cause of severe soft tissue infection and bacteremia in Israel (11,12). Several important features differentiate the illness caused by the new V. vulnifi cus biotype from previously described V. vulnifi cus infections. First, a new vector, a pond fi sh (tilapia) grown in fresh water, has been associated with V. vulnifi cus infection. Second, infection is caused by direct injury from the fi sh backbone while purchasing, cleaning, or handling the live fi sh, as opposed to contamination of a prior injury by immersion in seawater or ingestion of contaminated seafood. Bacteriologically, the new V. vulnifi cus biotype differs from other V. vulnifi cus strains by its biochemical features (salicin-, cellobiose-, citrate-, and lactose-negative, plus delayed reaction for o-nitrophenyl-β-D-galactopyranoside [ONPG]). These biochemical differences initially prevented correct identifi cation of the strain by routine laboratory methods (11,12). Furthermore, molecular analyses using several methods have shown that V. vulnifi cus biotype 3 is genetically distinct from biotypes 1 and 2 (11)(12)(13)(14)(15). Bisharat et al. suggested that biotype 3 is a recombinant clone that may have emerged as a result of hybridization of 2 V. vulnifi cus populations (14). Currently, biotype 3 is geographically restricted to Israel; biotypes 1 and 2 have a worldwide distribution (16).
The 1996-1997 Israeli cluster involved 62 persons, with a slight male predominance (58%) and a median age of 56 years. Although no deaths were reported, 41 persons (66%) had conditions that required surgical debridement, 1 had total limb amputation, and 7 had fi nger amputations (11). A new, aggressive, live-fi sh marketing initiative in the northern part of Israel was implicated in the outbreak, and the outbreak was followed by new instructions from the Israeli Ministry of Health that recommended selling only precleaned, ice-chilled tilapia (17).
We studied the epidemiology and the trends in illnesses associated with the new V. vulnifi cus biotype 3 during a 7-year period (1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) following the initial 1996-1997 cluster. Our study assesses the effects of infection, risk factors for death, possible spread to other fi sh species, and molecular relatedness of the V. vulnifi cus biotype 3 strain.

Clinical Data
Clinical data were obtained from the records of the Infectious Diseases Department, Epidemiology Unit, Israel Ministry of Health, Jerusalem. Subdistrict health offi ces reported data collected from persons with V. vulnifi cus infections; a standardized questionnaire was used. Investigation was initiated after passive reporting from primary physicians who treated patients with suspected V. vulnificus infection or when clinical isolates were positively iden-tifi ed as V. vulnifi cus biotype 3 by the Vibrio Reference Laboratory in the Government Central Laboratories, Israel Ministry of Health, Jerusalem. The data collected were patient's age, sex, and place of residence; underlying diseases; circumstances of exposure and type of fi sh involved; time lapse from exposure to seeking treatment in the emergency department; site of infection; length of hospitalization; clinical symptoms; source of isolation; antimicrobial drug treatment; and outcome.
Cases were classifi ed as laboratory-confi rmed when a patient with suggestive history had V. vulnifi cus or V. vulnifi cus biotype 3 isolated from blood or soft tissue or as suspected when a patient had suggestive history without positive cultures. Suggestive history was considered to be the development of soft tissue infection or septicemia after recent (within 7 days) fi sh exposure or immersion in a water pool.

Source of Isolates and Laboratory Diagnosis
Initial identifi cation of V. vulnifi cus was performed in the microbiology laboratories of hospitals where infected patients had been admitted, using the API 20E strip (bio-Mérieux, Marcy-l'Etoile, France). The laboratories submitted the isolates for confi rmation and further identifi cation to the Vibrio Reference Laboratory, Government Central Laboratories in Jerusalem. Identifi cation of V. vulnifi cus biotype 3 was performed solely in this laboratory by using biochemical tests (failure to ferment citrate, lactose, salicin, cellobiose, and a negative test result for ONPG).
Random isolates were also submitted for molecular analysis of the cytotoxin-hemolysin gene. This test was developed in the Vibrio Reference Laboratory and can differentiate between V. vulnifi cus biotype 3 and biotypes 1 and 2 by demonstrating the unique restriction fragment length polymorphism (RFLP) patterns of biotype 3 (Figure 1). Briefl y, the V. vulnifi cus cytotoxin-hemolysin gene (vvhA) is amplifi ed from crude bacteria lysate (boiling a loop full of bacteria suspended in 100 μL Tris-EDTA buffer for 10 min) using primers with the sequences RRCTH 5′-CAGCTCCAGCCGTTAACCGAACCACCCGC-3′ and LCTH 5′-TTCCAACTTCAAACCGAACTATGAC-3′. This step is followed by RFLP analysis. For the enzymatic restriction reaction, 2 μL of the PCR-amplifi ed DNA was added to a reaction mixture to give a fi nal volume of 20 μL, according to the manufacturer's instructions (New England Biolabs, Ipswich, MA, USA), in 2 separate reactions. The restricted DNA was separated by electrophoresis in a 2% gel that was stained with ethidium bromide and visualized for the specifi c biotype 3 restriction patterns. Restriction sites for 2 enzymes, KpnI and PstI, existed in the sequence of vvhA of V. vulnifi cus biotype 3 but not in the corresponding vvhA gene of biotypes 1 and 2 ( Figure 1).

Molecular Subtyping
Twenty randomly selected laboratory-confi rmed isolates of V. vulnifi cus biotype 3 from study years (1998)(1999)(2000)(2001)(2002)(2003) plus 4 retrospective isolates from 1997 were sent to Centers for Disease Control and Prevention (CDC), Atlanta, Georgia, USA, for molecular subtyping. The isolates were recovered from the blood or wound sites of patients with a history of exposure to various fi sh. An additional isolate was recovered from the remnants of a tilapia fi sh found in the refrigerator of 1 of the infected patients (the fi sh remnants were diced and inoculated onto Vibrio-selective agar with and without enrichment). These 25 isolates were available for further molecular studies. Eight submitted human V. vulnifi cus biotype 1 isolates stored at CDC were used for comparison.
V. vulnifi cus isolates were subtyped by pulsed-fi eld gel electrophoresis (PFGE) in accordance with the PulseNet protocol for V. cholerae (18) using Sfi I for primary enzyme digestion with 1 modifi cation: thiourea (Sigma-Aldrich, St. Louis, MO, USA) was routinely added to the electrophoresis running buffer at a fi nal concentration of 50 μmol/L (19) to prevent DNA degradation, which was commonly found during the initial runs. All DNA fi ngerprints were captured using a Gel Doc EQ system (Bio-Rad, Hercules, CA, USA). The PFGE fi ngerprints were analyzed in BioNumerics, version 4.0 (Applied Maths, Sint-Martens-Latem, Belgium). Gels were normalized by aligning the bands of the PulseNet universal standard Salmonella enterica serotype Braenderup (H9812) placed in every fi fth lane on the gels (20). Dendrograms were made of the similarities of the DNA fi ngerprints by using the Dice similarity coefficient and unweighted pair group method with averages (unweighted pair group method with arithmetic mean) clustering. An optimization of 1.5% and tolerance window of 1.5% were used. The 25 isolates sent to CDC as described above were submitted for susceptibility tests using the Etest method according to Clinical Laboratory Standards Institute standards for Enterobacteriaceae (21).

Statistical Methods
Proportions were compared by using the Fisher exact test or the χ 2 test, and continuous variables were compared by using the Kruskal-Wallis test. Variables associated with death at the 0.05 signifi cance level were entered into a stepwise forward logistic regression model for mortality rate. Analyses were performed by using SPSS version 15 software (SPSS, Inc., Chicago, IL, USA).

Results
A total of 134 cases of V. vulnifi cus infection were identifi ed during the 8-year study period from 1998 through 2005 ( Figure 2). Most cases (96, 71.6%) were laboratoryconfi rmed; 70 (52%) were submitted to the Vibrio Reference Laboratory and identifi ed as V. vulnifi cus biotype 3. PCR-RFLP analysis performed on 34 of these 70 isolates (49%) showed the unique KpnI, PstI pattern. Two patients with laboratory-confi rmed V. vulnifi cus biotype 3 infection were excluded from further analyses because clinical data were missing. The median age of the remaining 132 patients was 66 years (mean 58.9 years, range 10-93 years) ( Figure 3). Overall, infection rates for women were only slightly higher than those for men (1.1:1); this predominance increased in those >70 years of age (1.7:1). The sites of V. vulnifi cus isolation included wounds in 61 patients (65% of 94 patients with laboratory-confi rmed infection), blood in 24 (26%), or both in 7 (7%). The source was not recorded in 2 patients.
Clinical characteristics of the patients and outcomes are summarized in Table 1. Compared to patients with laboratory-confi rmed infection, patients with suspected infection were more likely to be men, with a longer incubation period after exposure, less severe clinical symptoms, a shorter hospitalization time, and a more favorable outcome. Information regarding the type of exposure was available for 93 patients. Most of these patients (75, 81%) were injured while purchasing or preparing fi sh for cooking. Fish were purchased from fi sh stores (52, 56%), mobile selling units (9, 10%), or stands near fi shponds (14, 15%). Other types of exposures included involvement in fi sh marketing (14, 15%), either in selling (11/14) or cleaning the fi shponds (3/14). One person was infected after fi shing at a fi shpond. Four persons had no connection to the pond fi sh industry. Three persons with suspected infection became ill after fi shing in the Sea of Galilee, and 1 with laboratory-confi rmed V. Peter's fi sh) was the primary fi sh involved in 86 of these cases (83%), common carp (Cyprinius carpio) in 14 cases (13%), and both in 2 cases (2%). Two other patients reported exposure to saltwater fi sh: gilt-head sea brim (Sparus aurata) that was purchased in a fi sh store (laboratory-confi rmed infection) and common gray mullet (Mugil cephalous) that was purchased at a fi sh stand near a fi shpond (suspected infection).
Ten patients (7.5%) died and 9 (6.8%) underwent amputation of fi ngers or part of the arm (7 patients) or a leg (2). Two of the patients who underwent amputation subse-quently died. Seven of the patients who died (70%) were >70 years of age, 2 were in their 50s, and 1 was in his 60s. All patients who died or underwent amputation had laboratory-confi rmed infection.
Ninety patients with laboratory-confi rmed infections were included in the analyses of risk factors associated with death. Factors that were signifi cantly associated with death by a univariate analysis are outlined in Table 2. We could not show that any of the antimicrobial drug regimens infl uenced outcomes. Variables associated with death were entered into a stepwise forward logistic regression model.

Molecular Subtyping
Analysis of the PFGE subtyping results from 25 isolates (  1997) showed that the selected isolates represented 18 unique but similar (>88% pattern similarity; 1-3 fragment differences) PFGE fi ngerprint patterns regardless of the type of fi sh exposure (tilapia vs. carp) and year of isolation ( Figure 5). A tilapia isolate from 2003 generated an Sfi I PFGE pattern, which was indistinguishable from the PFGE patterns of 3 patient isolates (1 from 2002, 2 from 2003) with reported exposure to tilapia from wound sites. When compared with biotype 1 strains reported from the United States, the biotypes separated into 2 distinct clusters with ≈70% pattern similarity ( Figure 5).

Discussion
Our study showed that infections caused by V. vulnifi cus biotype 3 continued to occur after the initial cluster during 1996-1997, with an average of ≈16 cases annually. Although the annual rate of infection during 1998-2005 was half the rate of infection during 1996-1997 (62 cases) (11), outcomes were more grave. Our study found that 10 persons (7.6%) died; no deaths were reported during the 1996-1997 outbreak (11). Possible explanations for this disparity could be that patients in our study were older (median age 66 vs. 56 years, respectively) with a higher proportion of laboratory-confi rmed infections (70.1% vs. 53%, respectively). We have shown that patients without laboratory confi rmation had a much milder form of the disease and a more favorable outcome.
Notably, our fi ndings show a high proportion of infected women (52%), including 8 (80%) of the 10 patients who died. Previous studies have stressed a male predominance (7,9,10,(22)(23)(24)(25), and have even argued that female sex hormones protect against contracting the disease (26). With V. vulnifi cus biotype 3 infections in Israel, it was more likely that women purchased and prepared the fi sh before cooking, and thus were more likely to be exposed to fi shbone injuries.
There are 3 major clinical syndromes of V. vulnifi cus illnesses, including primary bacteremia (mostly related to raw seafood consumption), wound infection (mostly related to immersion in contaminated water or to injury by seafood preparation), and gastroenteritis (after consumption of seafood or swallowing contaminated water) (9,23,27). Patients with primary bacteremia caused by non-biotype 3 strains are more likely to have predisposing conditions, particularly liver diseases in >80% of patients (7-9), whereas patients with wound infection are more likely to be previously healthy and have a more favorable outcome (9,10,23,24,(28)(29)(30). The death rate may exceed 50% in the most seriously ill patients (7-9,23). Our study describes a large, uniform group of patients who acquired infection through percutaneous exposure. All sought treatment for a wound infection, which in 18% was complicated by secondary bacteremia. In 38 patients (29%), laboratory confi rmation of infection was lacking. These patients tended to delay seeking treatment and to have a milder form of infection, compared with patients who had laboratory-confi rmed infection. Some of the patients who lacked laboratory confi rmation may not have had V. vulnifi cus infection. However, patients with milder forms of V. vulnifi cus infection may have been less likely to undergo extensive microbiologic workup and the yield of cultures, if taken, would have been lower.
Only 45% of the patients in the group had underlying diseases. Liver disease occurred in 14% of patients overall, and in 19% of patients with confi rmed cases, but was not a statistically signifi cant risk factor for death. The casefatality rate was 7.6% for the entire study population and 10.6 % for patients with laboratory-confi rmed infection. These characteristics concur with prior reports of wound infection syndrome caused by non-biotype 3 V. vulnifi cus (9,10,23,24,(27)(28)(29)(30).
Of concern is our fi nding that V. vulnifi cus infection was not limited to tilapia exposure. Fourteen persons (13%) reported exposure to the common carp, or to both tilapia and common carp (2 persons, 2%). Tilapia and common carp are co-cultivated in several fi shponds in northern Israel, which may have resulted in cross-contamination. For those exposed to salt water fi sh (gilt-head sea brim and common gray mullet, 1 person each), contamination may have occurred at the place of purchase. V. vulnifi cus infection after exposure to the common carp caused the death of 1 patient and the amputation of a fi nger in another. One person infected after exposure to the gilt-head sea brim also died. Notably, orthopedic surgeons from hospitals in northern Israel have also pointed out severe soft tissue infections, including those caused by V. vulnifi cus, after fi shbone injury from the common carp (31).
The V. vulnifi cus biotype 3 strains studied were uniformly susceptible to all tested antimicrobial agents. These results concur with those of similar studies of clinical and environmental V. vulnifi cus biotype 1 isolates from the United States (32,33) and Taiwan (22). Only a few isolates from Taiwan showed resistance to ceftazidime and moxalactam.
Molecular subtyping of the V. vulnifi cus biotype 3 strains by PFGE showed no specifi c association between fi sh species and PFGE pattern. The results indicate that the biotype 3 strains are homogeneous with limited heterogeneity between the isolates but cluster distinct from biotype 1 strains. PFGE has been shown (34) to offer suffi cient discrimination when subtyping biotype 1 strains, but published fi ndings that evaluate the utility of PFGE to sufficiently discriminate between biotype 3 strains are limited. Modifi cations in restriction sites may alter the number and size of DNA fragments, which defi ne the PFGE pattern and result in observable true differences.

1880
Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 14, No. 12, December 2008 A high degree of homogeneity among the V. vulnifi cus biotype 3 strains and distinction from the biotypes 1 and 2 has been also observed by other authors applying various methods, including random amplifi ed polymorphic DNA, (13), multilocus sequence typing (14,16), and analysis of variations in simple sequence repeat loci (35). Notably, the latter method was able to demonstrate small-scale variations among the biotype 3 strains (35). The PFGE results also support the conclusion that this biotype 3 is distinct from the other V. vulnifi cus biotypes. The high degree of homogeneity is another indicator that the emergence of biotype 3 is a recent evolutionary event (14,16,36).
We identifi ed independent risk factors for death in our group, including bacteremia, altered immune status, and history of ischemic heart disease. Septic shock was also found to be a strong predictor of death; however, septic shock may also be an outcome variable. Nonetheless, prior reports have also identifi ed septic shock or hypotension as important risk factors for death (8,23). A recent large study from Taiwan (29) found that treatment with thirdgeneration cephalosporins combined with tetracycline was an independent predictor of lower death rates in a subgroup of patients with hemorrhagic bullous necrotic cutaneous lesions. We did not demonstrate any correlation between a specifi c antimicrobial drug regimen and death rate in our study population. The predisposing diseases that were associated with death in previous reports were liver disease and neutropenia (8,23). Ischemic heart disease was not previously recognized as a classic predisposing factor predict-ing death; however, in the Vibrio-associated wound infections after Hurricane Katrina in Louisiana, USA, ischemic heart disease occurred in 7 of 13 patients (54%) with more severe illness (25).
Before the introduction of V. vulnifi cus biotype 3 into the fi sh aquaculture no infections were reported in Israel (37). Also, no reports have been made of V. vulnifi cus infection acquired through marine activities in the Mediterranean Sea. Almost the entire impact of V. vulnifi cus infections in Israel is associated with the freshwater fi sh industry. In response to the new threat, the Israeli Ministry of Health has issued regulations forbidding the selling of live, uncleaned tilapia (11,38). Fish stores and fi shpond workers have been instructed to use protective gloves when handling fi sh, to keep fresh fi sh packed in ice, and to prevent direct contact between buyers and live fi sh. The public has been instructed accordingly. Apparently, compliance with these regulations is not universal.
Our fi ndings outline the substantial effects of V. vulnifi cus illnesses in Israel and support a call for more strict regulations of fresh fi sh marketing as well as public education. Research efforts should focus on how V. vulnifi cus has penetrated the freshwater aquaculture in Israel and ways in which this trend can be reversed.