Undiagnosed cases of fatal Clostridium-associated toxic shock in Californian women of childbearing age

Materials and Methods 

Results 

There were 1115 deaths among women aged 15-44 years between 2000-2003. We determined that 38 of these women had death certificate ICD-10 diagnoses highly suspicious for CSTS (Table 1). To assess possible coinfections with Clostridium species, decedents diagnosed with other bacterial infections were not excluded. Fourteen women met the case definition, but 3 women did not have autopsy reports or tissues available, and 24 were excluded because of factors such as underlying immunocompromised condition, noninfectious cause of death, or falling outside the case definition for CSTS (Table 2 and Figure).

TABLE 1. Number of study subjects with primary cause of death ICD-10 codes highly suspicious for Clostridium sordellii-associated toxic shock (n = 38)

	ICD-10	Diagnosis	No. of women
	A419	Septicemia unspecified	16
	A483	Toxic shock syndrome	3
	O85	Puerperal sepsis	4
	O998	Other specified diseases and conditions complicating pregnancy, childbirth, and the puerperium	4
	O045	Therapeutic abortion, complete or unspecified, complicated by genital tract and pelvic infection	2
	N710	Acute inflammatory disease of uterus	2
	N739	Female PID unspecified	2
	N72	Inflammatory disease of cervix uteri	1
	O040	Therapeutic abortion, incomplete, complicated by genital tract and pelvic infection	1
	O068	Therapeutic abortion, complete or unspecified, with other and unspecified complications	1
	O988	Other maternal infectious and parasitic diseases complicating pregnancy, childbirth, and the puerperium	1
	I469	Cardiac arrest unspecified	1

ICD-10, International Classification of Disease, 10th revision; PID, pelvic inflammatory disease. Ho. Undiagnosed cases of fatal Clostridium-associated toxic shock. Am J Obstet Gynecol 2009.

TABLE 2. Diagnoses of excluded cases

	Diagnosis	Organism
	Septic shock (15) resulting from:	Sepsis, unspecified (9) Group A Streptococcus (2) Gangrenous ulcer Candida, Actinomyces, Streptococcus Enterococcus faecalis status Posttherapeutic abortion Beta-hemolytic streptococci
	Toxic shock syndrome (2) resulting from:	Streptococcus species Staphylococcus aureus (tampon retention)
	Uterine rupture (2)
	Polymicrobial pelvic abscess
	Pulmonary embolus secondary to DVT
	Pneumonia with sepsis
	Idiopathic cardiomyopathy
	Dilated cardiomyopathy
	Cardiac arrest

DVT, deep venous thrombosis; ICD, International Classification of Diseases. Ho. Undiagnosed cases of fatal Clostridium-associated toxic shock. Am J Obstet Gynecol 2009.

View Large Image Download to PowerPoint

FIGURE. Flowchart CA, California; CSTS, C sordellii toxic shock; ICD-10, International Classification of Diseases, 10; IHC, immunohistochemistry; PCR, polymerase chain reaction. Ho. Undiagnosed cases of fatal Clostridium-associated toxic shock. Am J Obstet Gynecol 2009.

Of the remaining 11 women, tissues from 5 women were positive for Clostridium species by molecular testing: 3 for C perfringens, 1 for C sordellii, and 1 for both C perfringens and C sordellii (Table 3). Evidence of infection with multiple pathogens was identified in an additional case: Clostridium species and group A Streptococcus were identified by IHC on the surface of the episiotomy wound/perineum. Because the 16S ribosomal RNA gene sequences amplified from tissues showed 99% identity with Fusobacterium nucleatum, infection from Clostridium species was not deemed to be the cause of this death. IHC and PCR assays were negative for the remaining 5 women, including 3 who had other bacterial pathogens listed as their cause of death.

TABLE 3. Characteristics of study subjects with positive molecular assays for Clostridium

	Case no.	1	2	3	4	5
	DEMOGRAPHICS
	Age (y)	32	40	40	41	37
	Underlying condition	Abnormal Papanicolaou smear	Cervical dysplasia	No	Methamphetamine abuse	Asthma, methamphetamine abuse
	Recent pregnancy	No	No	Yes	Yes	Yes
	Recent abortion	No	No	No	No	Yes
	Precipitating event	Cervical conization procedure	Cervical laser treatment	Term vaginal birth	Third-trimester stillbirth	Therapeutic surgical abortion
	Time from event to symptom onset	12 d	3 d	4 d	Immediate	2 wk
	Time from admission to death	6 h	2 d	28 h	11 h	N/A
	CLINICAL PRESENTATION
	Fevera	No	No	No	N/R	N/R
	Tachycardiab	Yes	No	Yes	N/R	N/R
	Hypotensionc	Yes	Yes	Yes	Yes	N/R
	Presence and location of effusion	Peritoneal	Pleural, peritoneal, and pericardial	Pleural and peritoneal	Pleural and peritoneal	Pleural
	Highest white blood cell count (K cells/μL)	75.7	70.2	131.2	N/R	N/R
	Highest hematocrit (%)	59.7	58.7	56.9	N/R	N/R
	PATHOLOGY RESULTS
	Focus of infection	Cervix/endometrium	Cervix/myometrium	Endometrium	Cervix/uterus	Endometrium
	Edema	Yes	Yes	Yes	Yes	Yes
	Acute inflammation	Yes	Yes	Yes	Yes	Yes
	Hemorrhage	Yes	Yes	Yes	Yes	Yes
	Necrosis	No	Yes	Yes	No	Yes
	IDPB IHC	Clostridium species	Clostridium species	Clostridium species	Clostridium species	Negative
	IDPB PCR	Clostridium sordellii	C sordellii and Clostridium perfringens	C perfringens	C perfringens	C perfringens

IDPB, infectious disease pathology branch; IHC, immunohistochemical assay; N/A, not applicable; N/R, not recorded; PCR, polymerase chain reaction. Ho. Undiagnosed cases of fatal Clostridium-associated toxic shock. Am J Obstet Gynecol 2009.

a Temperature >38.0°C; b Heart rate >120 beats/min; c Blood pressure <90/60 mm/Hg.

Ages ranged from 32-41 years for the 5 women with molecular evidence of Clostridium species infection (cases 1-5). Two women had undergone a medical procedure for cervical dysplasia (1 by laser therapy and 1 by conization); 1 had an abortion that used cervical dilatation with laminaria, followed by curettage; 1 woman had a stillborn delivery; and 1 woman delivered an infant 4 days earlier. The time interval from their precipitating event to the onset of symptoms varied, with both acute and delayed presentations (median 4 days; range, immediately to 2 weeks). The time interval from hospital admission to death ranged from 6 hours to 2 days. One woman died before hospital admission; 2 weeks had elapsed from the time of her abortion until her death. Two of the women had a history of methamphetamine abuse.

Because of rapidly deteriorating conditions, patient clinical and laboratory information was not always available. The available clinical characteristics, nonetheless, were consistent with other reports of CSTS: tachycardia, hypotension, edema formation, hemoconcentration, leukemoid reaction, and the absence of fever. For those with available medical history, all had abdominal pain, but none had vomiting, diarrhea, or rash. Available laboratory results showed a marked leukemoid response (white blood cell count range, 70.2-131.2 [×103/μL]) and hemoconcentration (hematocrit range, 56.9-59.7[%]). Hospital cultures of blood, peritoneal, and cervical specimens from 3 of the women were all negative.

The histopathology examination for the 5 Clostridium-positive women showed inflammation with either a cervical or uterine (endometrium or myometrium) focus. All had evidence of severe hemorrhage, and 3 of the 5 women had diffuse necrosis. There was no evidence of gas production in any of the cases.

Between 2000-2003, an average of 16.5 million women (range, 15,995,796-16,910,486) aged 14-44 years were living in California.18 Including the 4 California women who were reported previously8 and the 2 women found to have C sordellii infection in this investigation, this gives an estimated incidence of 0.036 episodes of CSTS per 100,000 persons. The proportion of deaths caused by CSTS is 0.54% or 5.4 per 1000 deaths for women who died in this age group during this 4-year span.

Comment 

Although an exceedingly uncommon cause of death in the general population, CSTS is a relatively common cause of death among young women; our investigation found that approximately 1 in 200 deaths in women of childbearing age in California is due to CSTS. In our study, 3 of 5 cases of CSTS, normally attributed to C sordellii, were actually C perfringens. All Clostridium-positive cases shared clinical characteristics consistent with past cases of CSTS. Many of the symptoms unique to CSTS—the toxic shock presentation, edema formation, hemoconcentration, and leukemoid reaction—are attributable to the highly potent large clostridial toxins: lethal toxin (LT) and hemorrhagic toxin (HT) from C sordellii and alpha toxin from C perfringens.17, 19, 20, 21 The cytopathic effects of C sordellii are enhanced in acidic environments, such as the vagina.22 Two of the women were not pregnant but had undergone recent instrumentation of the cervix. Women selected through our death certificate query were older (median age, 40 years) than women described in previously published similar cases of clostridial infection. In contrast to previously published reports, we observed a wider range of time between symptom onset and death, and none of the women in this investigation had nausea or vomiting when they sought medical attention. In 10 of 16 previously published accounts of women with CSTS, gastrointestinal symptoms of nausea and vomiting were reported.

Deaths from CSTS are sudden and catastrophic, and laboratory data and tissue are usually either negative or unavailable. Microbial isolates were not available for further characterization. Our investigation yielded a high proportion of clostridial infections among the 11 cases examined. This is important, because negative microbial cultures in this toxin-mediated disease are common; thus, ascertaining the true clinical spectrum of this disease is difficult. Differences in virulence factors or strains that might determine pathogenicity could not be examined. The presence of toxins such as sordellilysin (SDL), a cholesterol-dependent cytolysin, has been found to be associated with a less lethal strain of C sordellii than in cases with lethal toxin.23 Future studies aimed at delineating the clinical and microbiologic factors that characterize fatal CSTS are needed to improve diagnosis and treatment for this fatal disease.

Reports of fatal CSTS and endometritis subsequent to receiving the medical abortion regimen of oral mifepristone and intravaginal misoprostol have raised questions regarding the safety of this regimen. In 2005, the report of 4 Californian women with fatal CSTS was followed shortly by a public health advisory from the Food and Drug Administration regarding this medical abortion regimen. In addition, Planned Parenthood of America has changed their regimen to the use of oral instead of intravaginal misoprostol. The possible association of medically induced abortion with fatal C sordellii infection is difficult to assess without a background estimate of the true incidence rate of CSTS among young women. Whether the recent cases of fatal CSTS resulted from the mifepristone-misoprostol regimen for medically induced abortion or were incidentally diagnosed by enhanced molecular-based testing has been the subject of much discussion.24, 25, 26, 27, 28, 29 Hypotheses supporting an association between medical abortion and CSTS are based on mifepristone's antagonistic effects on both glucocorticoid and progesterone receptors. Interference with the glucocorticoid-mediated stress response could compromise normal physiologic compensatory mechanisms in the setting of septic shock, resulting in cardiovascular collapse and death.22 However, women who have not undergone medically induced abortion have died of obstetric or gynecologic CSTS,1, 3, 5, 7, 30 and in 1 case, a fatal gynecologic infection occurred in the absence of pregnancy or pharmacologic intervention.2 Our investigation discovered fatal CSTS in 2 young women, neither of whom had been recently pregnant or undergone medical abortion but who had undergone medical procedures to the cervix (laser therapy [case 2] and conization [case 3]). This suggests that pregnancy and pharmacologic modulation of the host's immune response are not the sole determinants of fatal C sordellii and C perfringens infections. In each, circumstances allowed the passage of anaerobic pathogens into the uterus, an environment likely to be supportive of Clostridium replication and toxin production.

This retrospective review probably underestimates the true prevalence of fatal infections from Clostridium species and C sordellii in particular. Undiagnosed cases of fatal CSTS in women may not have had the selected ICD-10 codes listed as the primary cause of death, autopsies performed, or tissues or hospital records available for evaluation. Also, the CSTS case definition we devised was based on symptoms from published C sordellii fatalities,1, 2, 3, 4, 5, 6, 7, 8, 10, 11, 12, 16 which could have excluded other deaths caused by CSTS. Finally, this investigation would not have detected any patients with nonlethal CSTS who were hospitalized but survived. Cohen et al11 report 1 nonfatal obstetric infection with a C sordellii strain that lacked the lethal toxin gene, but the vast majority of reported cases have been fatal.

This retrospective death certificate review targeted women in the same geographic region, age, and time frame as the 4 young women who died after receiving medical abortion.8 Our search revealed previously undiagnosed cases of fatal clostridial infections, including fatal C perfringens infections in peripartum women with CSTS-like illnesses. C sordellii or C perfringens infection should be suspected in afebrile women who are postpartum, had a recent abortion, or had recent surgical procedures on the cervix, with negative bacterial cultures and rapidly deteriorating condition. This constellation of findings should alert clinicians toward CSTS as a possible diagnosis. Although CSTS has been almost uniformly fatal to date, the treatment of choice is aggressive physiologic support, surgical debridement with removal of the necrotic tissue, and appropriate anaerobic antibiotic coverage. Hopefully, earlier institution of specific treatment can save lives.

Conclusion 

A rare disease, the diagnosis of CSTS often goes undetected. Our retrospective study shows that C sordellii and C perfringens are associated with undiagnosed catastrophic infectious gynecologic illnesses among women of childbearing age who are either peripartum or have had recent cervical instrumentation. These cases did not include any women who received medically induced abortion with mifepristone or misoprostol, but the clinical spectrum of our CSTS cases was consistent with cases in previously published reports. Indeed, we suspect that treatments or conditions that allow passage of anaerobic bacterium into the uterus are important risk factors for catastrophic clostridial infections in women.