Susceptibility of Endometrial Isolates Recovered from Women with Clinical Pelvic Inflammatory Disease or Histological Endometritis to Antimicrobial Agents

Abstract

The CDC recommended outpatient treatment of pelvic inflammatory disease (PID) is an intramuscular dose of ceftriaxone plus 14 days of doxycycline, with or without metronidazole. European guidelines (2017) include moxifloxacin plus ceftriaxone as a first line regimen, particularly for women with Mycoplasma genitaliu-massociated PID. However, the susceptibility of bacteria recovered from the endometrium of women with PID to moxifloxacin is unknown. The in vitro antibiotic susceptibility of facultative and anaerobic bacteria recovered from endometrial biopsy samples were evaluated from 105 women having symptomatic PID and/or histologically confirmed endometritis. A total of 342 endometrial isolates from enrollment visits were identified using a combination of biochemical tests and sequencing. Isolates were tested for antimicrobial susceptibility using agar dilution against ceftriaxone, clindamycin, doxycycline, metronidazole and moxifloxacin according to the Clinical and Laboratory Standards Institute (CLSI) guidelines. Neisseria gonorrhoeae was susceptible to ceftriaxone with all isolates having an MIC of 0.03 μg/mL. All the other endometrial isolates were susceptible to ceftriaxone, except for Prevotella species, only half of which were susceptible. The in vitro susceptibility profile for BV-associated bacteria (Gardnerella vaginalis, Atopobium vaginae, Prevotella species, Porphyromonas species and anaerobic gram-positive cocci) revealed greater susceptibility to moxifloxacin compared to doxycycline. Moxifloxacin was superior to metronidazole for G. vaginalis and A. vaginae, and either metronidazole or moxifloxacin was needed to cover Prevotella species. Based on in vitro susceptibility testing, the combination of ceftriaxone plus moxifloxacin provides similar coverage of facultative and anaerobic pathogens compared to the combination of ceftriaxone, metronidazole and doxycycline. Head to head clinical studies of these treatment regimens are needed to evaluate clinical efficacy and eradication of endometrial pathogens following treatment.

Introduction

Pelvic inflammatory disease (PID) is a clinical syndrome characterized by inflammation of the endometrium, fallopian tubes and/or peritoneum [1]. An estimated 2.5 million American women aged 18–44 have received a diagnosis of PID in their lifetimes [1], and delay in diagnosis or lack of treatment can result in infertility, ectopic pregnancy, and chronic pelvic pain [2]. Women with PID can present for care with a wide variety in symptoms ranging from mild to severe [3]. Diagnosis of PID is defined by Centers for Disease Control and Prevention (CDC) and the European guidelines as having one or more of the following criteria present during pelvic examination: cervical motion tenderness, uterine tenderness or adnexal tenderness [3,4].

A wide range of microorganisms cause PID. Sexually transmitted organisms such as Neisseria gonorrhoeae [5,6], Chlamydia trachomatis [5,6] and Mycoplasma genitalium [7] have been implicated in the etiology of PID. Anaerobic and facultative organisms associated with bacterial vaginosis, including Gardnerella vaginalis, Atopobium vaginae and Prevotella species, have also been linked with histologic endometritis and symptomatic PID [6,8,9]. Haemophilus influenzae, a respiratory pathogen, has been recovered from the fallopian tubes of women with PID [10].

For treatment to provide coverage against the range of pathogens associated with PID, the 2015 CDC STD outpatient treatment guidelines recommend a 250 mg intramuscular single dose of ceftriaxone or a third-generation cephalosporin plus oral doxycycline (100 mg twice a day for 14 days) with or without oral metronidazole (500 mg twice a day for 14 days) [3]. The 2017 European outpatient guidelines for PID recommend a 500 mg intramuscular single dose of ceftriaxone, plus oral doxycycline (100 mg twice a day for 14 days) plus oral metronidazole (500 mg twice a day for 14 days) [4]. The most notable difference between the two outpatient treatment guidelines is the recommendation to uniformly include metronidazole, which provides expanded coverage against anaerobic gram-negative rods including Prevotella and Porphyromonas species. In 2017, the European guidelines added ceftriaxone plus moxifloxacin as an alternative two drug regimen to treat women with Mycoplasma genitalium-associated PID [4].

The objective of this study was to evaluate the in vitro antibiotic susceptibility of facultative and anaerobic bacteria from the endometrium of women having symptomatic PID and/or histologically confirmed endometritis against ceftriaxone, clindamycin, doxycycline, metronidazole and moxifloxacin. We hypothesized that the combination of ceftriaxone and doxycycline would be inadequate to cover anaerobic bacteria, and that the same combination with the addition of metronidazole or the two-drug combination of ceftriaxone plus moxifloxacin would cover all endometrial pathogens.

Materials and Methods

Bacterial Isolates

The Anaerobes and Clearance of Endometritis (ACE) cohort was comprised of symptomatic women with clinically diagnosed PID, according to CDC criteria [11], who participated in a clinical trial (NCT01160640) comparing antibiotic regimens with or without metronidazole for PID treatment [12]. The T-cell Response Against Chlamydia (TRAC) cohort was comprised of asymptomatic women at high risk for STI [13]. Both cohorts were recruited from clinics and emergency departments in Pittsburgh, PA between 2010–2015.

This secondary analysis included 90 women from the ACE study having a clinical diagnosis of PID and 15 women from the TRAC study without symptoms of PID. All 105 women had confirmed histological endometritis. Endometrial biopsy tissues were collected using a Pipelle® endometrial sampling device. In order to minimize contamination, the external sheath of the sampling device was wiped with an alcohol wipe in the laboratory prior to expulsion of tissue into a sterile petri plate for processing. As a further precaution to avoid contamination by bacteria present in the cervical canal at the time of the sampling, only the mid portion of the tissue sample was used for bacterial cultivation and the remainder was evaluated by histology. The tissues were evaluated by nucleic acid amplification testing (NAAT) for N. gonorrhoeae, C. trachomatis and M. genitalium, and cultivation of aerobic and anaerobic bacteria. A subset of 342 isolates recovered from 105 endometrial biopsy samples from 105 women were used in this secondary analysis.

G. vaginalis was isolated on human bilayer agar with Tween (HBT) (Becton Dickinson, Rockville, MD) and incubated at 37°C for 48 hours in anaerobic jars. G. vaginalis was identified based on characteristic colony morphology, beta hemolysis on HBT, Gram stain showing gram-variable pleomorphic rods, and negative catalase reaction. N. gonorrhoeae and H. influenzae were isolated on Chocolate agar (Becton Dickinson, Rockville, MD) and incubated in CO₂ at 37°C for 48 hours. Both microorganisms were identified by colony, Gram stain morphology and the API® NH (Biomérieux, Marcy-l’Etoile, France). Anaerobic bacteria were isolated on Brucella agar (Hardy Diagnostics, Santa Maria, CA) and incubated at 37°C for 48 hours in anaerobic jars. Anaerobic bacteria were identified based on phenotypic characteristics including Gram stain and colony morphology, and DNA was extracted using PrepMan™ Ultra Sample Preparation Reagent (Applied Biosystems, Foster City, CA). 16S rDNA for restriction fragment length polymorphism (RFLP) analysis was completed using HaeIII (Promega, Madison, WI) restriction enzyme for identification of A. vaginae and Prevotella species. Hinf1 (Promega, Madison, WI) restriction enzyme was used to identify anaerobic gram-positive cocci species. Both HaeIII and Hinf1 restriction enzymes were used to identify Megasphaera phylotypes I and II, novel species which have not yet been named. Porphyromonas species were identified using both HaeIII and TaqI (Promega, Madison, WI) restriction enzymes. The RFLP patterns for each species were confirmed by 16S rDNA sequences compared to the GenBank data library using the nucleotide BLAST program.

Agar Dilution Susceptibility Testing

The anaerobic endometrial isolates were evaluated for susceptibility to ceftriaxone, clindamycin, doxycycline, metronidazole and moxifloxacin (Sigma-Aldrich, St. Louis, MO) using the agar dilution methods described by the Clinical and Laboratory Standards Institute [14] as previously described [15]. N. gonorrhoeae and H. influenzae were isolated to purity and suspended in Mueller-Hinton broth (Becton Dickinson, Rockville, MD) at a 0.5M McFarland suspension. For N. gonorrhoeae isolates, GC Medium agar (Becton Dickinson Difco, Rockville, MD) plates with supplement VX with reconstituting fluid (Becton Dickinson Difco, Rockville, MD) and Hemoglobin solution 2% (Becton Dickinson, Rockville, MD) and varying concentrations of test agent alongside a no drug growth control were inoculated and incubated at 35°C for 24 hours. For H. influenzae isolates, a microdilution tray was inoculated with Haemophilus test medium [16] and varying concentrations of test agent alongside a no drug growth control and incubated at 35°C for 24 hours. N. gonorrhoeae ATCC 49226 and H. influenzae ATCC 49247 were used as controls. The lowest antibiotic concentration yielding marked reduction to no growth was read as the Minimum Inhibitory Concentration (MIC). The microbiological susceptibility and resistant breakpoints used for interpretation of MIC results for the anaerobic endometrial isolates are as follows: ceftriaxone (≤16μg/mL and ≥64μg/mL), clindamycin (≤2μg/mL and ≥8μg/mL), doxycycline (≤4μg/mL and ≥16μg/mL), metronidazole (≤8μg/mL and ≥32μg/mL) and moxifloxacin (≤2μg/mL and ≥8μg/mL) as defined by CLSI [14]. The microbiological breakpoints for N. gonorrhoeae are as follows: ceftriaxone (≤0.25μg/mL), clindamycin (not defined), doxycycline (≤0.25μg/mL and ≥2μg/mL) and moxifloxacin (≤0.125μg/mL and ≥0.5μg/mL) as defined by CLSI [16]. The microbiological breakpoints for H. influenzae are as follows: ceftriaxone (≤2μg/mL), clindamycin (not defined), doxycycline (≤2μg/mL and ≥8μg/mL) and moxifloxacin (≤1μg/mL) as defined by CLSI [16].

Results

The 342 endometrial isolates from 105 women with histologically confirmed endometritis were evaluated for in vitro susceptibility to ceftriaxone, clindamycin, doxycycline, metronidazole and moxifloxacin. C. trachomatis was detected in the endometrium of 17 (16%) women, of whom 12 had symptomatic PID and 5 had histologic evidence of endometritis but did not meet the criteria for PID. N. gonorrhoeae was detected in 6 (6%) women, all of whom had symptomatic histologically proven PID, and 5 of these women had N. gonorrhoeae by culture. M. genitalium was detected in 10 (10%) women, nine of whom had symptomatic PID. Of the 105 women having histologically proven endometritis, 78 (74%) were negative for all three sexually transmitted pathogens but had endometrial cultures positive for facultative and/or anaerobic bacteria.

As shown in Table 1, G. vaginalis and A. vaginae accounted for over half of the endometrial isolates. Based on in vitro susceptibility testing, ceftriaxone was active against all endometrial isolates tested except for Prevotella species. With an MIC₅₀ of ≤16 ug/mL, only half of the Prevotella isolates were susceptible to this drug with variability by species. While none of the P. disiens, P. melaninagenica and Prevotella group 1 isolates were susceptible to ceftriaxone, 11 (92%) P. amnii, 4 (83%) P. bivia, 1 (100%) P. buccalis, 2 (100%) P. denticola, 1 (100%) P. massiliensis and 11 (52%) P. timonensis isolates were susceptible. For doxycycline, susceptibility ranged from 43% in A. vaginae to 100% in novel Megasphaera species and H. influenzae. The MIC₉₀ values for G. vaginalis, A. vaginae, Prevotella species, Porphyromonas species and anaerobic gram-positive cocci isolates were 16 ug/mL. Doxycycline was less effective against these same endometrial isolates compared to moxifloxacin. Prevotella species, novel Megasphaera species, Porphyromonas species and anaerobic gram-positive cocci isolates were susceptible to metronidazole. However, both G. vaginalis and A. vaginae had a MIC₉₀ of ≥128 ug/mL for metronidazole. Moxifloxacin was not fully effective against novel Megasphaera and N. gonorrhoeae species.

Table 1:

The comparative in vitro activity of ceftriaxone, clindamycin, doxycycline, metronidazole and moxifloxacin against endometrial isolates from women having histological evidence of endometritis

Endometrial Species	Na	Antimicrobial Agents	Range	MIC (ug/mL)		nb (%)c
				50%	90%
Gardnerella vaginalis	130	Ceftriaxone	0.03–32	0.05	2	0
		Clindamycin	0.03–1	0.125	0.5	0
		Doxycycline	0.125–32	1	16	18 (14)
		Metronidazole	0.03–>128	32	>128	83/125 (66)
		Moxifloxacin	0.06−16	1	2	7 (5)
Atopobium vaginae	60	Ceftriaxone	0.03–8	0.5	2	0
		Clindamycin	0.03–>128	0.03	0.125	1 (2)
		Doxycycline	0.25–32	16	16	32 (64)
		Metronidazole	0.25–>128	32	128	29/57 (51)
		Moxifloxacin	0.125–>128	0.5	16	15/59 (25)
Prevotella speciesd	56	Ceftriaxone	0.125–>128	16	64	24 (43)
		Clindamycin	0.03–>128	0.06	>128	20 (36)
		Doxycycline	0.06–32	8	16	13 (23)
		Metronidazole	0.25–>128	2	8	2/51 (4)
		Moxifloxacin	0.03–4	1	4	0
Anaerobic gram-positive cocci species e	74	Ceftriaxone	0.03–16	0.25	2	0
		Clindamycin	0.03–>128	1	>128	20 (27)
		Doxycycline	0.125–32	4	16	19/73 (26)
		Metronidazole	0.03–>128	2	4	5/67 (8)
		Moxifloxacin	0.03–128	0.5	16	21 (28)
Novel Megasphaera species f	6	Ceftriaxone	0.06–1	0.5	1	0
		Clindamycin	0.03–1	0.03	1	0
		Doxycycline	0.125–4	4	4	0
		Metronidazole	0.03–1	0.25	1	0
		Moxifloxacin	0.5–16	0.5	16	1 (17)
Porphyromonas species g	6	Ceftriaxone	0.03–0.25	0.125	0.25	0
		Clindamycin	0.03−1	0.03	0.06	0
		Doxycycline	1–16	4	16	2 (33)
		Metronidazole	0.03–>128	1	1	1 (17)
		Moxifloxacin	0.25–32	0.5	0.5	0
Neisseria gonorrhoeae	5	Ceftriaxone	0.03	0.03	0.03	0
		Clindamycin	0.5–8	2	2	NDi
		Doxycycline	0.12–4	2	2	3 (60)
		Metronidazoleh	NT	NT	NT	NT
		Moxifloxacin	0.03–8	2	4	3 (60)
Haemophilus influenzae	5	Ceftriaxone	0.03	0.03	0.03	0
		Clindamycin	4–32	8	8	NDi
		Doxycycline	1–2	1	2	0
		Metronidazoleh	NT	NT	NT	NT
		Moxifloxacin	0.06–0.25	0.06	0.125	0

^aNumber of isolates tested

^bNumber of resistant isolates

^cPercent resistant (n/N)

^dPrevotella species includes 12 P. amnii, 12 P. bivia, 1 P. buccalis, 2 P. denticola, 3 P. disiens, 1 P. massiliensis, 3 P. melaninogenica, 21 P. timonensis and 1 Prevotella species

^eAnaerobic gram-positive cocci includes 2 novel species, 3 Anaerococcus species, 18 Anaerococcus tetradius, 1 Anaerococcus vaginalis, 9 Dialister micraerophilus, 5 Finegoldia magna, 9 Peptoniphilus harei, 1 Peptoniphilus ivorii/coxii, 5 Peptoniphilus lacrimalis, 6 Peptoniphilus species, 11 Peptostreptococcus anaerobius, 2 Peptostreptococcus stomatis, 2 Veillonella montpellierensis

^fNovel Megasphaera species includes 1 novel Megasphaera type 1 and 5 novel Megasphaera type 2

^gPorphyromonas species includes 1 Porphyromonas asacch/uenonis, 3 Porphyromonas asacch, 2 Porphyromonas uenonis

^hNeisseria gonorrhoeae and Haemophilus influenzae isolates were not tested (NT) against metronidazole

ⁱResistance is not defined (ND) by CLSI

Even though clindamycin is not currently part of the outpatient regimen for treatment of PID in the US or Europe, clindamycin is included as part of the parenteral therapies for inpatient treatment of PID [3]. As shown in Table 1, clindamycin was active against G. vaginalis, A. vaginae, novel Megasphaera species and Porphyromonas species. However, only 35 (63%) of Prevotella isolates and 51 (69%) of anaerobic gram-positive cocci isolates were susceptible to clindamycin.

Because women usually had more than one isolate present in each endometrial sample, we next calculated the proportion of all isolates which would be covered by the various treatment regimens. In Figures 1A–1C, the results are summarized to show the proportion of women having endometrial isolates susceptible to each antibiotic alone and in combination. In 84% of women (Figure 1A), all isolates were susceptible to ceftriaxone whereas the remaining 17 women had at least one isolate resistant to this drug. In 65% of women, all isolates were susceptible to doxycycline; and in 93% of women, all isolates were susceptible to a combination of ceftriaxone and doxycycline (Figure 1A). The remaining 7 women had at least one isolate, all of which were Prevotella species, resistant to this combination of antibiotics. Figure 1B displays the proportion of isolates susceptible to the drug regimens when metronidazole is added. Because of the activity of metronidazole against Prevotella species, the combination of ceftriaxone, doxycycline and metronidazole provided full coverage against all endometrial isolates. As shown in Figure 1C, moxifloxacin alone provided activity against all endometrial isolates for 78% of women but when combined with ceftriaxone, full coverage was provided against these isolates.

Figure 1.

Proportion of women having endometrial isolates susceptible in vitro to four antibiotics: ceftriaxone, doxycycline, metronidazole and moxifloxacin. The proportion of women with endometrial isolates susceptible to: Figure 1A ceftriaxone, doxycycline and the combination of ceftriaxone and doxycycline; Figure 1B ceftriaxone, doxycycline, metronidazole and the combination of ceftriaxone, doxycycline and metronidazole; and Fgure 1C ceftriaxone, moxifloxacin and the combination of ceftriaxone and moxifloxacin.

Discussion

The question of whether metronidazole should be added to outpatient regimens for treatment of PID has been in debate for many years. The CDC guidelines have noted that there is no definitive evidence that inclusion of metronidazole to cover anaerobic bacteria prevent the long-term sequelae of PID including ectopic pregnancy and infertility [3]. However, the 2017 European guidelines have recommended the addition of metronidazole to include activity against obligate anaerobes [4]. The present study was designed to evaluate which of the anaerobic bacteria recovered from the endometrium of women with histologically confirmed endometritis would not be covered if metronidazole was omitted from PID treatment. Our study found that Prevotella species, which accounted for 56 (16%) of the 342 endometrial isolates, was the primary group of anaerobic bacteria where resistance to ceftriaxone and doxycycline was observed. Eleven (20%) of the Prevotella isolates were resistant to both antibiotics but were susceptible to metronidazole.

A second objective of the study was to assess whether the combination of moxifloxacin, a quinolone antibiotic which has greater activity than doxycycline against M. genitalium [17], plus ceftriaxone would have adequate activity against anaerobes associated with PID. A study evaluating treatment of PID with another quinolone, ciprofloxacin, found higher persistence of anaerobes compared to women treated with a combination of clindamycin and gentamicin [18]. Boothby et al. compared treatment of uncomplicated PID with moxifloxacin or ofloxacin plus metronidazole in a comparison study where there was no clinical difference between the two treatments, but no cultures of the endometrium were performed [19]. Ross et al. reported that 14 days of moxifloxacin versus 14 days of ofloxacin plus metronidazole was associated with fewer adverse events in a randomized study although anaerobic bacteria were not included in this evaluation [20]. We found that based on in vitro susceptibility testing, the combination of ceftriaxone and moxifloxacin provides similar coverage against facultative and anaerobic bacteria compared to the combination of ceftriaxone, doxycycline and metronidazole, but improved coverage when compared to ceftriaxone and doxycycline.

The link between female reproductive tract disease and M. genitalium has been inconsistent [21]. A study evaluating the association between M. genitalium and PID and ectopic pregnancy based on serological assays found no association [22].

However, a study has documented the presence of this pathogen in the fallopian tube of a woman with PID [23], while Bjartling et al. reported M. genitalium was an independent and strong risk factor for both cervicitis and PID [7]. Lis et al. conducted a metaanalysis and reported that M. genitalium was associated with a significant increased risk of cervicitis, PID, preterm birth and infertility [24]. Others have also reported that women with endometrial M. genitalium were more likely to have histologically confirmed endometritis compared to women without M. genitalium [25, 26]. In the PEACH study, infertility, recurrent PID, and chronic pelvic pain were more common in women with endometrial M. genitalium [25]. Thus, drug regimens that are active against M. genitalium as well as C. trachomatis, N. gonorrhoeae and facultative/anaerobic pathogens may be optimal.

Our study has several strengths and important limitations. The present study included the evaluation of cultivable pathogens recovered from endometrial biopsy samples. These sensitive methods yielded several microbes which are novel species including Megasphaera phylotypes types 1 and 2 and two novel anaerobic grampositive cocci. This study is unique in its reporting of susceptibility of these cultivated bacteria against the primary agents used to treat PID, including clindamycin. It is notable that more than a third of Prevotella species are resistant to clindamycin (Table 1).

Important limitations of the present study include the inherent bias introduced by any study based on cultivation, which fails to detect microorganisms which cannot be cultivated. However, since having a purified culture isolate of each microorganism is required for performance of MIC testing, it was not possible to address this limitation. A second limitation is that endometrial sampling, even one performed with an endometrial biopsy system having a protected sheath to minimize contaminants from the vagina, may inadvertently detect bacteria from the endocervical canal. In the present study, skin contaminants including diphtheroids and staphylococci, as well as lactobacilli, were excluded from evaluation. A major limitation of the present study is the inherent limitations of in vitro susceptibility testing since it may not accurately predict clinical efficacy. Although ceftriaxone had activity against a broad range of endometrial isolates, a single dose of ceftriaxone may be inadequate to eradicate these pathogens from the endometrium of a woman with PID. In addition, lack of antimicrobial activity in vitro may not translate to microbiological treatment failure. Even though A. vaginae is known to have limited susceptibility to metronidazole, a study evaluating clearance of A. vaginae following treatment of BV demonstrated 100% clearance in a mean of 3.5 days [27]. Thus, clinical studies evaluating clearance of these pathogens and resolution of endometritis will be needed to assess whether the addition of metronidazole provides an advantage by providing activity against Prevotella species.

Highlights:

• Anaerobes are recovered from the upper genital tract of women having signs and symptoms of PID.

• Prevotella species, found in half of women with PID, are not covered by ceftriaxone/doxycycline.

• Moxifloxacin plus ceftriaxone has in vitro activity against all PID pathogens isolated in this cohort.