Molecular analysis of an outbreak of lethal postpartum sepsis caused by Streptococcus pyogenes

Abstract

Sepsis is now the leading direct cause of maternal death in the United Kingdom, and Streptococcus pyogenes is the leading pathogen. We combined conventional and genomic analyses to define the duration and scale of a lethal outbreak. Two postpartum deaths caused by S. pyogenes occurred within 24 h; one was characterized by bacteremia and shock and the other by hemorrhagic pneumonia. The women gave birth within minutes of each other in the same maternity unit 2 days earlier. Seven additional infections in health care and household contacts were subsequently detected and treated. All cluster-associated S. pyogenes isolates were genotype emm1 and were initially indistinguishable from other United Kingdom emm1 isolates. Sequencing of the virulence gene sic revealed that all outbreak isolates had the same unique sic type. Genome sequencing confirmed that the cluster was caused by a unique S. pyogenes clone. Transmission between patients occurred on a single day and was associated with casual contact only. A single isolate from one patient demonstrated a sequence change in sic consistent with longer infection duration. Transmission to health care workers was traced to single clinical contacts with index cases. The last case was detected 18 days after the first case. Following enhanced surveillance, the outbreak isolate was not detected again. Mutations in bacterial regulatory genes played no detectable role in this outbreak, illustrating the intrinsic ability of emm1 S. pyogenes to spread while retaining virulence. This fast-moving outbreak highlights the potential of S. pyogenes to cause a range of diseases in the puerperium with rapid transmission, underlining the importance of immediate recognition and response by clinical infection and occupational health teams.

Fig 1

Identification of group A streptococcus (GAS) cases. Cumulative number of individuals carrying or infected with the outbreak strain by date (upper axis, blue-shaded bars). Individuals with the outbreak strain included maternity patients (solid shading), household contacts (diagonal crosshatching), and health care workers (HCWs) (horizontal crosshatching). Individuals found to be carrying or infected with other GAS isolates during the period of surveillance are shown on the lower axis (gray-shaded bars). Period of overlapping hospital admissions for three maternity patients denoted by gray shaded area. During the period of intense surveillance, which started on December 24 and lasted for 2 months, two HCWs were identified to be carrying emm12 GAS (gray-shaded horizontally hatched bars) and two nonmaternity patients had community-acquired invasive infections caused by GAS strains emm89 (gray-shaded bar, January 3) and emm75 (white bar, January 23). Enhanced surveillance continued for a further 4 months, during which time no further GAS was isolated from any maternity patient.

Fig 2

Phylogenetic analysis results demonstrating that the core genomes of the maternity unit isolates were identical to each other but different from contemporaneous emm1 isolates submitted to the reference laboratory. The genomes of all 15 isolates from the maternity unit and 9 emm1 isolates from around England were sequenced and mapped to the complete genome sequence of MGAS5005, a contemporary U.S. strain (3). A maximum-likelihood phylogenetic tree was generated from core genome single nucleotide polymorphisms (SNPs) of 24 emm1 isolates. The maternity unit isolates clustered in a single clade that differed from the nearest relative by eight SNPs. The sic allele for each isolate is indicated. The unique sic allele (sic1.301) that arose within the outbreak cluster is highlighted in red. The scale bar represents substitutions per SNP site. Details of all SNPs are provided in Table S2 in the supplemental material. HCW, health care worker; URT, upper respiratory tract; LRT, lower respiratory tract; GT, genital tract.

Fig 3

Immunity to emm1 GAS among healthy pregnant women. The ability of antenatal sera to opsonize FITC-labeled M1 GAS is shown for 80 representative sera (out of 199). The percentage of neutrophils associated with serum-opsonized FITC-labeled bacteria was measured by flow cytometry and expressed relative to the percentage of neutrophils associated with FITC-labeled bacteria when IVIG (2.5 mg/ml) was used as the opsonin. Negative, below the level of the “no serum” (complement alone) control (≤10% of IVIG-equivalent opsonophagocytosis).