Background. Pyomyositis is typically caused by gram-positive bacteria, especially Staphylococcus aureus. Few cases of Escherichia coli pyomyositis have been reported, including only 1 involving a patient with a hematologic malignancy.
Methods. The clinical microbiology database at The M. D. Anderson Cancer Center (Houston, TX) was reviewed for the period January 2003 through December 2007 to identify cases of E. coli pyomyositis. Clinical characteristics, laboratory and radiologic findings, treatment, and outcomes were recorded. Available isolates underwent phylogenetic group determination, virulence genotyping, multilocus sequence typing, repetitive-element polymerase chain reaction, and pulsed-field gel electrophoresis.
Results. Six cases of E. coli pyomyositis were identified. All patients were receiving chemotherapy for a hematologic malignancy; 5 were severely neutropenic. Three patients became hypotensive, 2 required intensive care, and 2 (33%) died, despite receiving carbapenem therapy. All E. coli isolates were fluoroquinolone resistant; 55% produced an extended-spectrum β-lactamase (ESBL). Five of 6 available isolates belonged to phylogenetic group B2, had similar virulence factor profiles, exhibited > 95% similar repetitive-element polymerase chain reaction profiles, and represented sequence type ST131; however, all had unique pulsed-field gel electrophoresis profiles.
Conclusions. E. coli pyomyositis has emerged as a serious problem among our patients with hematologic malignancy. It usually is caused by members of E. coli ST131, a recently identified cause of fluoroquinolone-resistant, ESBL-positive E. coli infection worldwide. Awareness of this emerging syndrome and the usual causative agent is important to ensure appropriate management when febrile, neutropenic patients with hematologic malignancy exhibit signs of localized muscle infection.
Escherichia coli is the most common cause of gram-negative bacteremia , including among patients with cancer , and constitutes a serious threat to compromised hosts. E. coli sepsis has substantial associated morbidity, mortality, and health care costs, with an estimated 40,000 deaths per year in the United States and a calculated cost between 1.1 and 2.8 billion dollars per year .
Pyomyositis, an acute bacterial infection of skeletal muscle, is caused by Staphylococcus aureus or other gram-positive organisms in >90% of cases. Extraintestinal pathogenic E. coli (ExPEC) can cause infections at a wide variety of anatomical sites, including the urinary tract, bloodstream, meninges, lungs, and other organs and tissues. However, to date, only 14 cases of E. coli pyomyositis in adults have been reported [4–13], only one of which involved a patient with a hematologic malignancy. Moreover, the characteristics of the causative bacteria have not been studied.
We recently encountered a case of E. coli pyomyositis in a neutropenic patient with acute myeloid leukemia. This prompted us to assess the frequency of E. coli pyomyositis in patients with cancer at our institution, to describe the clinical features of this syndrome, and to determine the phenotypic and molecular characteristics of the responsible E. coli strains.
Materials and Methods
To capture recent cases of E. coli pyomyositis at our institution, we screened the medical records of all 5793 consecutive patients known to have had E. coli isolated from any site at The University of Texas-M. D. Anderson Cancer Center (MDACC) during the preceding 5-year period (January 2003 to December 2007), as identified via the institution's clinical microbiology laboratory database. The diagnosis of pyomyositis was based on a typical clinical presentation (ie, fever and pain in the involved muscle[s]), confirmed by compatible radiologic findings, such as edema in the muscle tissue noted by computerized tomography or magnetic resonance imaging. E. coli was considered to be the causative agent if E. coli was isolated from a muscle abscess or, if no abscess culture was done, from blood.
Data collection. Medical records of patients with pyomyositis were reviewed retrospectively. Demographic data, underlying disease, clinical features of the pyomyositis episode, laboratory and radiologic findings, treatment, and outcomes were recorded.
Clinical microbiology. Species identity was confirmed by use of API 20E (bioMérieux). Antimicrobial susceptibility was determined according to Clinical and Laboratory Standards Institute guidelines using the Vitek legacy system (bioMérieux, Durham NC) and Clinical and Laboratory Standards Institute-specified interpretive criteria. Extended-spectrum β-lactamase (ESBL) production was determined using Vitek cards containing cefotaxime and cefotaxime plus clavulanate.
Phylogenetic analysis and virulence genotyping. E. coli phylogenetic groups (A, B1, B2 and D) and virulence genotypes for 51 ExPEC-associated virulence genes and 13 papA (P fimbrial structural subunit) alleles were determined by multiplex polymerase chain reaction (PCR) [14, 15]. Isolates were categorized as ExPEC if positive for ⩾2 of: papA and/or papC (P fimbriae; counted as one), sfa/foc (S and F1C fimbriae), afa/dra (Dr-binding adhesins), iutA (aerobactin system), and kpsM II (group 2 capsules). The O25b rfb (lipopolysaccharide) variant and blaCTX-M-15 (which encodes the CTX-M-15 ESBL) were detected by PCR .
Clonal analysis. Broad clonal group relationships were assessed by repetitive-element-based PCR profiling (rep-PCR) and multilocus sequence typing (MLST). Rep-PCR was done using the DiversiLab microbial typing system (Bacterial BarCodes) . Six E. coli bloodstream isolates from patients without pyomyositis were included as controls. MLST was performed based on allelic profiles for 7 housekeeping genes (http://mlst.ucc.ie). To resolve distinct strains, pulsed-field gel electrophoresis (PFGE) analysis of XbaI-restricted total DNA was done according to a standardized protocol .
Six cases of E. coli pyomyositis were identified among MDACC patients from 2003 through 2007 (table 1). Significantly more cases of E. coli pyomyositis per total number of E. coli unique bloodstream infections occurred during the final year of the study period (ie, 2007; 4 [2.4%] of 163) than during the first 4 years (ie, 2003–2006; 2 [0.3%] of 622; P=.01).
The mean age of the 6 case patients (4 men and 2 women) with pyomyositis was 53 years (range, 38–67 years). All 6 patients had an underlying hematologic malignancy (leukemia in 5 and lymphoma in 1) and were receiving chemotherapy or high-dose steroids plus fluoroquinolone prophylaxis. Five of six patients had been severely neutropenic (neutrophil count, <500 neutrophils/µL) for ⩾3 days (range, 3–13 days) before the clinical onset of pyomyositis. The single nonneutropenic patient was receiving high-dose steroids (equivalent to 120 mg of prednisone daily) for graft-vs-host disease from a recent allogeneic stem cell transplant.
When pyomyositis was diagnosed, 5 of the 6 patients had involvement of the calves, whereas 2 additionally had involvement of the thigh. Magnetic resonance imaging showed diffusely abnormal T2 signal intensity within the musculature, consistent with edema and suggestive of myositis (Figure 1A). Three patients had frank abscesses detected in the involved muscle(s) on days 5, 14, and 16, respectively, after the initial pyomyositis diagnosis (Figure 1B). Neutrophil counts had already recovered in these 3 patients at the that time the abscesses appeared. In each instance, the abscess was aspirated and yielded E. coli as the only pathogen.
Three patients (50%) developed hypotension and required transfer to the intensive care unit. All 6 patients were treated with a carbapenem; 3 also received amikacin. Two patients (33%) died; 1 had fulminant sepsis, whereas the other never recovered from his underlying malignancy and the infection.
E. coli was isolated from blood (all 6 subjects) and, if cultured, abscess contents (all 3 subjects with abscesses). In each instance, E. coli was the only organism isolated. All 9 E. coli isolates (100%) were fluoroquinolone resistant (table 2), significantly greater than the overall prevalence of fluoroquinolone resistance among all E. coli isolated from any site at our institution during the study period (2551 [44%] of 5793; P<.001). Moreover, 5 (55%) of the isolates (3 blood and 2 abscess isolates), all from 2007, were ESBL positive (table 2). Thus, the prevalence of ESBL-positive isolates was significantly greater among the pyomyositis isolates than among other clinical E. coli isolates, both during the overall study period (5 [55%] of 9 vs 144 [2.5%] of 5793; P<.001) and especially during the most recent year, 2007 (5 [83%] of 6 vs 85 [7%] of 1194; P<.001).
Six of 9 E. coli isolates from the pyomyositis episodes were available for molecular analysis, including the blood isolates from patients 1, 3, 4, and 5, and both the blood and abscess isolates from patient 6. All 6 available isolates met the molecular definition of ExPEC, and 5 of 6 (ie, all but patient 6's blood isolate, which was from phylogenetic group D), were from group B2. The 5 group B2 isolates exhibited fairly similar virulence factor profiles that included genes for adhesins (iha and fimH, +/− afa/dra), a toxin (sat), siderophore receptors (iha, fyuA, and iutA), group 2 capsule (kpsM II, with its K2 and K5 variants), and miscellaneous other traits (usp, traT, ompT, and malX) (table 3). In contrast, the group D blood isolate from patient 6 exhibited a distinct virulence profile.
According to rep-PCR, the 5 group B2 pyomyositis isolates exhibited >95% similar profiles and, thus, clustered closely together in a dendrogram (Figure 2), suggesting a common clonal group background. The group D strain stood apart, connected to the above cluster at the 66% similarity level. Five arbitrarily selected nonpyomyositis E. coli strains exhibited diverse rep-PCR profiles, all <85% similar to one another or to the pyomyositis isolate profiles (data not shown), suggesting diverse clonal group backgrounds.
In confirmation of the rep-PCR findings, MLST showed the 5 group B2 pyomyositis isolates to exhibit the same combination of alleles across the 7 sequenced loci, in a pattern corresponding to sequence type ST131. Consistent with their ST131 background, all 5 isolates exhibited the O25b rfb allele and were resistant to 2–7 antimicrobial classes (median, 4 antimicrobial classes) each (Tables 2 and 3). Two of the ST131 isolates were ESBL positive; one of these exhibited (ST131-associated) blaCTX-M-15. However, according to PFGE analysis, each of the 5 ST131 isolates represented a distinct PFGE type, as defined at the 94% profile similarity level. Three of these ST131-associated PFGE types were unique to this study. In contrast, 2 had been encountered previously in isolates from New York and Cleveland, Ohio, in a large PFGE reference library involving 250 ST131 isolates from 32 US and international locales (J.R.J., unpublished data).
E. coli pyomyositis is a rare clinical syndrome that has emerged as a new and often quite serious or fatal infectious complication among hematologic malignancy patients at our institution. The literature through 2008 contained only a few case reports of pyomyositis due to E. coli; these occurred mostly in immunosuppressed or chronically ill patients, but not in patients with hematologic malignancies [7–13]. Moreover, a recent review of pyomyositis in patients with hematologic malignancies found the most common causative organism to be S. aureus (as for pyomyositis generally) and included no cases due to E. coli . Only in 2009 was the first case of E. coli pyomyositis reported in a patient with a hematologic malignancy (ie, acute myeloid leukemia) . Interestingly, like most of our patients, this patient had calf involvement and was neutropenic when pyomyositis occurred .
By definition, all our patients had the typical presentation of pyomyositis, with fever and tenderness over the involved muscle area and characteristic computerized tomography or magnetic resonance imaging findings. Pyomyositis traditionally has been divided into 3 stages. Stage 1 involves inflammation, but aspiration of the muscle will not yield pus; stage 2 usually occurs 2–3 weeks later, with a clinically apparent frank abscess; and stage 3 has associated signs of toxicity . In contrast to many published pyomyositis cases, none of our patients had a frank muscle abscess on initial presentation. Possible explanations include that the patients (1) presented early in the disease (ie, in stage 1) or (2) were neutropenic, making them unable to mount a full inflammatory response to infection. The latter hypothesis is support by our finding that the 3 patients who developed an abscess did so only after recovering their neutrophil counts, despite ongoing treatment with appropriate antibiotics.
The clinical course of E. coli pyomyositis is usually benign. However, in our case series, 50% of the patients required intensive care unit transfer secondary to hypotension, and 33% died. In general, pyomyositis has been reported to have an attributable mortality of 1% in immunocompetent hosts  and 11% in patients with hematologic malignancies . The mortality associated specifically with pyomyositis due to E. coli has not been reported previously. However, considering the published cases of pyomyositis secondary to E. coli [6–13], we could calculate an associated mortality of 33%. This is identical to the mortality observed in our study and higher than that of E coli bloodstream infection in patients with cancer (14%) (K.J.V., unpublished data). These findings suggest that both host immunosuppression and bacterial virulence contribute to disease outcome.
All of our pyomyositis isolates were resistant to fluoroquinolones. Both selective pressure from extensive use of antibiotics and clonal spread have been postulated to contribute to the emergence of fluoroquinolone resistance. Each of these factors was present in our patients. Supporting the selection hypothesis, all patients were receiving fluoroquinolones as prophylaxis. Supporting the clonal spread hypothesis, 5 of the 6 isolates available for molecular analysis derived from the same E. coli lineage (ie, ST131). Notably, the diversity of PFGE types noted among the 5 ST131 isolates—and the fact that 2 of these PFGE types have been encountered previously at other centers—indicate that our cases are part of the global dissemination of ST131, not just a localized point-source outbreak in our institution or community.
Recent analyses of European and Canadian fluoroquinolone-resistant E. coli urine isolates found that the most prominent subset within the study populations was E. coli ST131 [16, 22], which not long previously had been first reported as an emerging cause of CTX-M-15-positive E. coli infections on multiple continents . Our finding that ST131 accounted for 5 of 6 pyomyositis cases establishes that this globally disseminated clonal group is capable of causing not only cystitis and bloodstream infections, as previously described, but also tissue-invasive disease. Moreover, on the basis of the dissimilar (presumably, non-ST131) rep-PCR profiles of the 5 comparison nonpyomyositis E. coli blood isolates, we can infer that at our institution ST131 is specifically associated with pyomyositis (5 of 6 pyomyositis isolates vs 0 of 5 nonpyomyositis bloodstream isolates; P=.015).
Our ST131 E. coli pyomyositis isolates exhibited virulence profiles (table 3) similar to those observed among previously reported E. coli ST131 isolates, whether CTX-M-15-positive or negative [16, 23, 24]. Additional study is needed to determine whether some of these virulence genes play a direct role in pathogenesis of pyomyositis or whether they are simply markers for this clonal group.
In summary, we report the emergence of E. coli pyomyositis as a distinctive and quite serious (even fatal) infectious disease syndrome among patients with hematologic malignancy at MDACC. Almost all cases were caused by members of a recently described, virulent, multidrug-resistant E. coli lineage, ST131 (O25:H4), that has been identified as an emerging cause of fluoroquinolone-resistant and ESBL (typically CTX-M-15)-positive extraintestinal E. coli infection worldwide. Awareness of this emerging syndrome and the predominant causative clonal group is important to ensure rapid institution of early and appropriate management when febrile neutropenic patients with hematologic malignancy exhibit localized muscle pain. In addition, relevant risk factors, reservoirs, transmission pathways, and virulence mechanisms of E. coli ST131 need to be defined to inform the development of effective measures to prevent further dissemination and emergence of this threatening new pathogen.
We thank Dr. Jeffrey Tarrand, Ms. Ruth Reitzel, and Ms. Tanya Dvorak for their support.
Potential conflicts of interest. J.R.R. has been a consultant for Ortho-McNeil and has received grants or contracts from Merck, Bayer, Procter and Gamble, Wyeth-Ayerst, and Rochester Medical. D.P.K. is an investigator for Merck and a consultant for Merck and Schering-Plough; has received grants and research support from Merck, Enzon, and Fujisawa/Astellas; is in the speaker's bureau of Merck, Pfizer, Enzon, and Fujisawa/Astellas; has been a speaker for Enzon and Fujisawa/Astellas; and is on the advisory board of Merck and Schering-Plough. I.I.R. is a consultant for Clorox, Cubist, and Cook; has received grants from Cubist, Schering-Plough, Versicor, Enzon, Pfizer, Wyeth, and Cook; is on the speaker's bureau of Merck, Schering-Plough, Pfizer, and Cook; and has royalties related to patents licensed to American Medical Systems, Horizon Medical Products, and TyRx on which he is a coinventor. H.L.D. has consulted with, has received honoraria for speaking, and has received research grants from Salix Pharmaceutical Company; has received a consulting fee from Romark Institute for Medical Research; has received research grants from Optimer Pharmaceuticals and IOMAI Corporation; has received honoraria for consulting and/or speaking from McNeil Consumer Healthcare and Merck Vaccine Division. All other authors: no conflicts.