Streptococcal necrotising fasciitis from diverse strains of Streptococcus pyogenes in tropical northern Australia: case series and comparison with the literature

Demographics

We describe a series of 14 cases of GAS necrotising fasciitis in a region with a high background incidence of GAS disease[9]. Our incidence of NF in 2001 of 3.8/100,000 and 5.8/100,000 in the Aboriginal population is higher than previously published rates which range from 0.4/100,000 in Canada[5] to 2/100,000/yr in Norway [4]. However the population denominators in the Canadian and Norwegian studies are far greater than in our study. Nevertheless, previous data from our region showed that rates of GAS bacteremia in the Aboriginal population were five times those in the Caucasian population[9]. Therefore NF in our population parallels an overall greater rate of GAS infection, most notably in the Indigenous population where conditions reflect those of developing countries[8].

The mean age of our patients (41.5 years) is lower than other reports showing an average age of 56–58 years, with notable increases in incidence with increasing age[4–6]. Unlike all other series which had a male predominance, 57% of our cases were female, with 71% females amongst the Aboriginal cases. The age and sex pattern seen in our series reflects the high rates of chronic diseases in the Indigenous population, with mortality from all causes being higher in all age cohorts, but especially so in females[14, 15]. 79% of our patients had concurrent chronic disease, compared with 46–71% in previous series[4–6, 16]. In this series 64% had pre-existing wounds, consistent with reported rates of 47–66%[4–6, 16]. None of our patients had preceding varicella infection.

Laboratory results

Two studies have looked at the value of early blood test results in predicting NF as opposed to non-NF soft tissue infection. One found that CK >600 U/L achieved PPV of 58% for NF[17]. However only 1/9 cases in this series had CK >600 U/L. CRP of >16 mg/L was also thought to be indicative of NF[16]. CRP was over 100 in all those measured in our series. Raised WCC has also been suggested as useful in a predictive model of early NF[18], with rates of 66–73% in other series[4–6]. However, only 36% of our patients had a raised WCC. We found lymphopenia to be a more consistent factor, with 100% of patients having a lymphocyte count of <1.0 × 10⁹/L, 64% of those being <0.5 × 10⁹/L. Low albumin was also universal.

Clinical features

STSS occurred in 64% of our cases in comparison to 40–46% in other series[4–6]. 71% of our patients were hypotensive at presentation, with 86% having renal impairment in comparison to 35–61% in other series. Similarly to the other series, 93% of our patients were admitted to the ICU but more of our patients required mechanical ventilation (79%) and inotropic support (71%). Positive blood cultures are not a prerequisite for GAS NF, as stated by Meleney in the initial descriptions[2], and reinforced in subsequent studies showing bacteremic rates of 38–60%[4–6, 16], consistent with 50% in this study.

Case-fatality rates for NF previously reached 50%, with mortality of 80–100% in GAS myositis[3]. Recent series report lower case-fatality rates of 20–34%[4–6], with our mortality being 28%. As in previous studies, deaths were more common in those with underlying illness, truncal infection, bacteremia, STSS, myonecrosis and delay in diagnosis and appropriate therapy.

Therapy

Immediate, extensive surgical debridement of all necrotic tissue is considered essential for optimal treatment of NF, with one early observational study showing a mortality of 0% with aggressive early surgery in comparison to 50% in historical controls[19, 20]. Surgery was conducted on the day of presumptive diagnosis of NF in all our patients, but none of the four fatal cases underwent amputation and one with upper limb involvement had only a fasciotomy performed.

Combination therapy with benzylpenicillin and clindamycin is now recommended treatment for proven GAS NF and STSS[21–23]. Experimental data suggest that penicillin is not as bactericidal when there are large numbers of GAS organisms present[24], with decreased expression of penicillin binding proteins when large inocula reach a stationary growth phase[25]. The potential benefits of clindamycin have been supported by a murine model of streptococcal myositis[26]. Clindamycin is a protein synthesis inhibitor, potentially also suppressing bacterial toxin and M protein production[25]. Furthermore, clindamycin may modulate the host immune response, by reducing lipopolysaccharide-induced monocyte production of TNFα[27]. A combination of β-lactam antibiotic and clindamycin has been shown to be superior to β-lactams alone in a retrospective review of invasive GAS infection in 56 children[28]. A recent retrospective review of notified invasive GAS infections in Florida, USA showed the use of clindamycin to be associated with lower mortality in NF cases, but not in other invasive GAS infections[29]. However there have been no randomised clinical trials.

Hyperbaric oxygen (HBO) therapy has been advocated as adjuvant therapy for both microbiological types of NF. Although there are no randomised trials, NF is an approved Undersea and Hyperbaric Medical Society indication for HBO therapy[30]. It is suggested that increased tissue oxygen partial pressures increases bacterial killing by increased respiratory burst and increased formation of oxygen free radicals. HBO is postulated to facilitate wound healing by supporting fibroblast proliferation and angiogenesis[30, 31]. In a murine model of GAS myositis, the combination of penicillin and HBO therapy exerted at least additive effects in decreasing bacterial counts in vivo and increasing survival[32]. An observational study of 29 patients with NF showed significantly lower mortality, and need for fewer debridements in those receiving adjunctive HBO, despite the HBO group being more seriously ill[33]. Patients with non-clostridial fasciitis appeared to have greater benefit. However more recently a retrospective evaluation of 37 patients treated for NF showed higher mortality with greater need for debridement in those receiving HBO[34].

The clinical picture of STSS has been attributed to superantigens produced by GAS, including pyrogenic exotoxins, streptococcal superantigen SSA and a mitogenic exotoxin[35]. This is supported by selective depletion of Vβ-bearing T cells in patients with STSS[36]. However GAS isolates from both severe and uncomplicated disease can produce large amounts of toxins. Furthermore, in comparison to patients with uncomplicated infection the sera of patients with severe GAS disease had low antibody levels against erythrogenic toxins[37], suggesting an important role for host humoral immunity. IVIG has therefore been proposed as therapy to increase neutralizing antibody levels in patients with severe GAS disease. A study of 12 patients, 11 with STSS and one with GAS NF, showed increased capacity of plasma to neutralize superantigenic activity following IVIG[38]. Post-IVIG plasma from each patient completely blocked cytokine production elicited by their respective GAS culture supernatants or by purified streptococcal pyrogenic exotoxins. GAS superantigens and cytokines suggestive of superantigen response have been isolated from tissue samples in patients with GAS NF[39]. More severe clinical disease correlated with significantly higher bacterial load in biopsy samples, and bacterial load in turn correlated with expression of superantigenic toxins. This supports a role for IVIG in GAS NF, with or without STSS. Adjunctive IVIG therapy in STSS was evaluated in a multicenter, randomized, double-blind, placebo-controlled trial[40]. Although prematurely terminated due to slow patient recruitment, there was a non-significant 3.6 fold higher mortality rate in the placebo group and a significant decrease in sepsis-related organ failure assessments on days two and three in the IVIG group. There are case reports and two case series of four and seven patients describing clinical improvements of patients with STSS following IVIG[41–43]. In a series of 20 cases of GAS NF and myonecrosis, 16 were treated with IVIG[6]. Although the case fatality rate was not significantly lower in those receiving IVIG, the overall survival rates of 80% in patients with NF and 63% in patients with myonecrosis were much higher than previously reported[6].

In summary, there is in vitro and in vivo experimental data to suggest a benefit for clindamycin, HBO therapy and IVIG in GAS NF and myositis. This is supported by limited case series but not by randomized controlled trials. The primary role of early, aggressive surgery in GAS NF remains probably the most critical factor in minimizing mortality. Clindamycin is recommended for GAS NF but whether clindamycin alone is as effective as penicillin plus clindamycin remains unclear. Benefit from HBO remains unproven, while there is emerging evidence for the use of IVIG, especially if STSS is present.

Molecular epidemiology

Following a reported resurgence in invasive GAS infections in the 1980's, epidemiological studies proposed that dissemination of a new, more virulent strain of M1 serotype GAS was the cause for a large proportion of these infections[7]. In reports from the United States, Europe and New Zealand the proportion of this strain appeared to increase, both in those with uncomplicated pharyngitis and in invasive GAS infections[37, 44–47]. The M1 serotype has also been reported to have a stronger association with STSS[48]. In contrast to these findings, other groups have noted a significant diversity amongst the organisms causing severe streptococcal infections, as well as significant clinical variation amongst patients infected with identical organism clones, suggesting that the severe infections are not related to one specific virulent clone[49–51]. Our findings are consistent with the latter, with GAS sequence typing showing all NF isolates to be different and only 1/13 to be emm type 1. This reflects previous findings of enormous diversity amongst organisms causing GAS bacteremia in our region, with no evidence of a dominant clone[9]. As well as finding no correlation between NF and specific M types, a recent Dutch study also found that no specific toxin genes or genes encoding matrix binding proteins were associated with NF[48].