Antimicrobial Resistance and Virulence Characterization among Escherichia coli Clinical Isolates Causing Severe Obstetric Infections in Pregnant Women

Abstract

The virulence markers and the antimicrobial resistance profiles of 78 Escherichia coli isolates causing obstetric infections accompanied by sepsis or not were studied. Adhesion-related virulence factors were the most prevalent markers. Low rates of resistance to the antimicrobial agents used as first-line therapy suggest their correct implementation in stewardship guidelines.

TEXT

Escherichia coli is the enteric Gram-negative bacillus most frequently found in the genital tract of women. Despite its commensal role, this microorganism can become pathogenic, colonizing new environments. Extraintestinal E. coli is the second most prevalent etiologic agent causing obstetric infections (1). E. coli possesses several virulence factor genes (VFG) that enhance vaginal and/or endocervical colonization in pregnant women. This colonization can lead to different infections in obstetric patients, such as intra-amniotic infection (IAI) or endometrial and urinary tract infections (UTIs), sometimes accompanied by sepsis. In addition, these microorganisms can cause neonatal infections, leading to maternal and fetal morbidity and mortality (2, 3). It has been estimated that 15% of pregnant and 12% of nonpregnant women in our hospital present E. coli in the genital tract (4).

The treatment of choice for maternal sepsis includes the administration of different antimicrobial agents, depending on the infection focus, being limited by the low number of antimicrobial agents considered to be safe to the fetus (5). In our hospital, the treatment of choice in patients with IAI consists of ceftriaxone, ampicillin-gentamicin, or ampicillin-cefoxitin, while the treatment of endometritis involves the use of ampicillin-gentamicin-metronidazole.

Briefly, among the virulence factors involved in UTIs, it is well known that adhesins, fimbriae, and toxins are the most important, as they allow the bacteria to adhere to the uroepithelium and cause tissue damage. However, further knowledge is necessary regarding their prevalences and the roles of other families of virulence factors in the specific field of obstetric infections derived from UTIs.

For this purpose, 78 E. coli isolates obtained from pregnant women attending the Hospital Clinic of Barcelona from 1987 to 2010 were included in the study; 56 were isolated from the blood samples of patients with sepsis from a genital or urinary origin, and 22 were isolated from amniotic fluid or placenta samples of patients with nonbacteremic IAI.

The resistance profiles were determined using the disk diffusion method. The antimicrobial agents tested are listed in Table 1 and include the first therapeutic options used to treat UTIs and genital infections. The results were interpreted according to CLSI guidelines (6), and the E. coli ATCC 25922 strain was used as the control.

TABLE 1.

Resistance to antimicrobial agents in E. coli isolates, according to the clinical features

Antimicrobial agent	No. (%) resistant in isolate group			P value
	Non-sepsis causing (n = 22)	Sepsis causing (n = 56)	Total (n = 78)
Ampicillin	13 (59)	36 (64)	49 (63)	0.6692a
Amoxicillin-clavulanic acid	3 (14)	7 (12)	10 (13)	1.0000b
Cefazolin	3 (14)	14 (25)	17 (22)	0.3680b
Cefuroxime	1 (5)	2 (4)	3 (4)	1.0000b
Cefoxitin	0 (0)	1 (2)	1 (1)	1.0000b
Cefotaxime	1 (5)	1 (2)	2 (3)	0.4872b
Ceftazidime	0 (0)	0 (0)	0 (0)
Imipenem	0 (0)	0 (0)	0 (0)
Piperacillin-tazobactam	0 (0)	1 (2)	1 (1)	1.0000b
Nalidixic acid	5 (23)	9 (16)	14 (18)	0.5216b
Ciprofloxacin	0 (0)	2 (4)	2 (3)	1.0000b
Chloramphenicol	2 (9)	5 (9)	7 (9)	1.0000b
Gentamicin	0 (0)	3 (5)	3 (4)	0.5547b
Amikacin	0 (0)	1 (2)	1 (1)	1.0000b
Kanamycin	1 (5)	5 (9)	6 (8)	0.6697b
Tetracycline	7 (32)	20 (36)	27 (35)	0.7448a
Trimethoprim-sulfamethoxazole	6 (27)	16 (29)	22 (28)	0.9087a

^aChi-square test.

^bFisher's exact test.

The VFG profiles of the isolates were analyzed by PCR using gene-specific primers for the virulence genes coding for the adhesins, toxins, and invasins most prevalent in the uropathogenic E. coli (UPEC) isolates described, from which the isolates causing the obstetric infections studied potentially come. The isolates were also screened for 5 specific virulence markers for extraintestinal pathogenic E. coli (ExPEC) or non-ExPEC classification (7). The PCR conditions used were 94°C for 4 min, followed by 30 cycles of 94°C for 30 s, with the corresponding annealing temperature (55 to 63°C) for 30 s, 72°C for 2 min, and a final elongation cycle of 72°C for 5 min. The samples were run in 1.5% agarose gels and stained with SYBR Safe DNA gel stain (Invitrogen, Spain). The E. coli phylogenetic group was determined using the 3-locus PCR-based method described previously (8). In order to determine if any isolate belonged to sequence type 131 (ST131), serotype O25b was identified in the collection, according to the methodology proposed by Clermont et al. (9), and the multilocus sequence typing (MLST) methodology was carried out with these isolates using the University of Warwick database for assigning sequence types (ST).

Statistical analysis was performed using Stata version 13.1 (Stata Corp., TX, USA). P values of <0.05 were accepted as significant, and statistical correction for multiple comparisons was applied.

Twenty isolates (26%) were resistant to three or more antimicrobial classes, presenting a multidrug-resistant (MDR) phenotype. Sixty-three percent of all the isolates were resistant to ampicillin, whereas only 13% were resistant to amoxicillin-clavulanic acid. Most of the isolates were susceptible to second- and third-generation cephalosporins, imipenem, aminoglycosides, ciprofloxacin, and chloramphenicol, with higher rates of resistance for tetracycline, trimethoprim-sulfamethoxazole, cefazolin, and nalidixic acid. The isolates causing sepsis had a lower prevalence of resistance to nalidixic acid, with a higher percentage of resistance to cefazolin being observed (Table 1).

The most prevalent VFG found among the isolates were adhesion related, with prevalences between 56 and 86%. The isolates harboring the greatest number of VFG were those causing sepsis, with a significantly higher percentages of hlyA, cnf1, papA, iha, fyuA, or papGII, all of them contained in pathogenicity islands. Regarding virulence factors related to iron recruitment, the iutA gene was found significantly more frequently in IAI-causing isolates (P = 0.0001), whereas the iroN gene was the most common in sepsis-causing isolates (P = 0.0284). A multivariate analysis of VFG showed the presence of the fimA, iucC, iroN, iutA, iha, and hra genes as being independent predictors of sepsis-causing isolates (Table 2). Seventy-eight percent of the isolates (with no significant differences between the sepsis- and non-sepsis-causing isolates) were classified as ExPEC according to the virulence markers harbored, and only two of these isolates belonged to ST131.

TABLE 2.

Prevalence of virulence factor genes according to the clinical features^a

Virulence factor	No. (%) with VFG in isolate group			P value	Univariate analysis			Multivariate analysis
	Non-sepsis causing (n = 22)	Sepsis causing (n = 56)	Total (n = 78)		Odds ratiod	95% confidence interval	P value	Odds ratiod	95% confidence interval	P value
hlyA	5 (23)	28 (50)	33 (42)	0.0282b	3.40	(1.10, 10.49)	0.0332
cnf1	2 (9)	19 (34)	21 (27)	0.0261b	5.14	(1.08, 24.32)	0.0392
sat1	9 (41)	23 (41)	32 (41)	0.9895b	1.01	(0.37, 2.74)	0.9895
fimA	17 (77)	50 (89)	67 (86)	0.2757c	2.45	(0.66, 9.07)	0.1792	32.20	(1.28, 809.78)	0.0349
papA	8 (36)	40 (71)	48 (62)	0.0042b	4.37	(1.54, 12.43)	0.0056
papC	11 (50)	33 (59)	44 (56)	0.4742b	1.43	(0.53, 3.86)	0.4752
papEF	12 (55)	38 (68)	50 (64)	0.2701b	1.76	(0.64, 4.83)	0.2727
papGI	0 (0)	0 (0)	0 (0)		1.00
papGII	8 (36)	42 (75)	50 (64)	0.0014b	5.25	(1.82, 15.13)	0.0021
papGIII	9 (41)	14 (25)	23 (29)	0.1656b	0.48	(0.17, 1.37)	0.1697
prs	15 (68)	36 (64)	51 (65)	0.7448b	0.84	(0.29, 2.40)	0.7450
fyuA	4 (18)	29 (52)	33 (42)	0.0069b	4.83	(1.45, 16.10)	0.0103
hra	8 (36)	11 (20)	19 (24)	0.1216b	0.43	(0.14, 1.27)	0.1270	0.13	(0.02, 0.96)	0.0452
sfa	5 (23)	18 (32)	23 (29)	0.4118b	1.61	(0.51, 5.06)	0.4142
ibeA	5 (23)	9 (16)	14 (18)	0.5216c	0.65	(0.19, 2.22)	0.4926
iucC	14 (64)	44 (79)	58 (74)	0.1740b	2.10	(0.71, 6.16)	0.1787	53.38	(2.31, 1,233.37)	0.0130
iutA	15 (68)	10 (18)	25 (32)	<0.0001b	0.10	(0.03, 0.31)	0.0001	0.01	(0.00, 0.13)	0.0016
iha	4 (18)	26 (46)	30 (38)	0.0210b	3.90	(1.17, 13.00)	0.0267	20.61	(1.77, 240.12)	0.0157
iroN	8 (36)	36 (64)	44 (56)	0.0252b	3.15	(1.13, 8.79)	0.0284	6.47	(1.30, 32.15)	0.0225
ag43	10 (45)	27 (48)	37 (47)	0.8261b	1.12	(0.42, 3.01)	0.8262
malX	9 (41)	38 (68)	47 (60)	0.0286b	3.05	(1.10, 8.44)	0.0319

^aStatistically significant results are in bold type.

^bChi-square test.

^cFisher's exact test.

^dOdds ratio for present versus absent.

An analysis of the presence of each VFG among the resistance profiles of the isolates to each of the antimicrobial agents tested was carried out, showing that susceptible isolates had a higher carriage of VFG.

The phenotypic results of antimicrobial resistance observed in the present study indicated high levels of ampicillin-resistant isolates in the collection, in accordance with those found in E. coli isolates causing neonatal sepsis and in extraintestinal E. coli in general (10). On the other hand, the low rates of resistance to amoxicillin-clavulanic acid and second- and third-generation cephalosporins observed in the present study are in contrast with the increasing appearance of strains carrying extended-spectrum β-lactamases (ESBLs) in the last years and causing infections from other sources, suggesting that the implementation of these antimicrobial agents as first-line therapy in these types of infections is correct (11). Nonetheless, the treatment administered should still be chosen depending on the rates of resistance in each hospital to gentamicin and cephalosporins in E. coli causing obstetric infections, as well as the prophylaxis or previous treatment with these antimicrobial agents, which have led to the development of resistant bacteria.

Regarding the VFG present in E. coli involved in the obstetric infections studied, it was found that adhesins and fimbriae may play an important role in the development of these infections, allowing the bacteria to colonize different environments. The higher prevalence of hlyA and cnf1 among the isolates causing sepsis might be related to the tissue damage involved with these infections. Concerning the iron recruitment systems, the yersiniabactin receptor encoded by fyuA and the genes encoding the siderophore receptors Iha and IroN were also more prevalent among the isolates causing sepsis, due to the need for UPEC to capture iron from the host within the hostile environment of urine. These virulence factors have been largely described as characteristic of UPEC (12). On the other hand, iutA was more frequently found in isolates causing IAI, elucidating a high adaptation capacity according to the particular microenvironmnent colonized.

A specific relationship was found between tetracycline-resistant isolates and the lower presence of several VFG included in pathogenicity islands (PAIs), similar to the previously described relationship between the acquisition of quinolone resistance and the loss of VFG (13).

In conclusion, to date, E. coli isolates causing obstetric infections present similar rates of antimicrobial resistance to those described for extraintestinal E. coli infections, except for a lower prevalence of resistance to third-generation cephalosporins, thereby those not carrying ESBLs. These results demonstrate that the administration of antimicrobials in our hospital is correct. However, it is important to establish surveillance networks specific for these kinds of infections in order to adapt stewardship programs when appropriate.