Introduction

Nowadays intrauterine devices (IUDs) are accepted as highly effective, long-term methods of contraception. The normal usage period of an IUD is 4–5 years depending on the manufacturer’s recommendation. The use of these devices is usually limited by the concerns about upper-genital-tract infections and forthcoming reproductive health complications [1]. Pelvic inflammatory disease (PID) is the most common gynecological infection, which is associated with high morbidity resulting from inflammation and damage of the reproductive tract. These may lead to severe sequel, such as tubal infertility, ectopic pregnancy and chronic pelvic pain. Many studies have been carried out to determine adequate microbiological aetiology of IUD associated upper genital tract infection [1,2,3]. The role of classical pathogens of sexually transmitted infections (STI) such as Chlamydia trachomatis and Neisseria gonorrhoeae in PID was shown not to be higher among women with the use of IUD [4]. However, where STI microbes are not detected before the insertion of the IUD, several authors emphasize the possible role of aerobic and anaerobic bacteria commonly associated with bacterial vaginosis (BV) as the causative agents behind inflammation of the upper genital tract [5, 6]. BV can be described, as a complex alteration of the vaginal flora with the decrease of the hydrogen peroxide-producing lactobacilli replaced by various anaerobic and facultative anaerobic bacteria including some BV-associated bacteria such as Gardnerella vaginalis, Atopobium vaginae, Mobiluncus spp., Mycoplasma spp. and Ureaplasma urealyticum [7]. IUDs, similarly to other medical implants, as foreign materials may provide a site for biofilm formation and may be a threat for upper genital tract infection [8, 9]. The removal of the IUD, in the case of symptoms of infection, may be advisable [5]. Earlier studies have also reported about different consequences regarding the association of IUD use and BV acquisition: some studies have found a positive correlation, whereas others have not confirmed an increased risk for acquiring BV among IUD users [6, 10]. The possible association between the development of BV and IUD use could be explained by the increased volume and duration of menstrual flow because of the presence of the device [6].

The aim of this study was to detect aerobic/anaerobic bacteria and fungi present in the biofilm cleared away from the surface of the removed IUDs, using classical culture methods and PCR, and to compare these data with the duration of the IUDs in place and the clinical symptoms at the time of the IUD removal.

Patients and methods

One hundred women using IUDs for different duration of time, were involved in this comparative study. These patients visited the Department of Obstetrics and Gynecology of the University Hospital Szeged, Hungary between January 2014 and December 2015. Beside recording the symptoms and past medical history, all patients had physical examination and pelvic ultrasound prior to the decision of removing the IUD. A written consent was also requested from every participant. Reviewing the number of patients, we set up three groups following the age of the IUD in situ (< 5 years, 5–10 years, > 10 years). As the fewest patient number was among those who wore the device for more than 10 years, we tried to set comparable groups in size up. After the gentle insertion of speculum, IUDs were removed carefully to prevent contamination with the vaginal microflora and were immediately sent to the Institute of Clinical Microbiology.

Conventional culture methods

All cultures started within 1 h after sampling. Each removed IUD was placed in 10 ml reduced BHI broth (Brain Heart Infusion broth pH 7.2) (Oxoid, Bakingstoke, UK) and mixed on a Vortex shaker for 30 s. After gentle homogenization, half of the suspension was placed in − 80 °C for further investigation by PCR. From the other part of the suspensions, 100 μl was plated on selective (Levine-EMB agar [Conda, Madrid, Spaine], Sabouraud-Chloramphenicol agar, [Bio-Rad, Marnes-la-Coquette, France]) and non-selective (Columbia agar + 5% sheep blood, Chocolate-PolyvitapleX agar [bioMérieux, Lyon, France]) media for isolation of facultative anaerobic bacteria and fungi and Columbia agar supplemented with 5% sheep blood [bioMérieux, Lyon, France] for isolation of strict anaerobic bacteria. For the isolation of anaerobic organisms, plates were incubated in an anaerobic workstation with an atmosphere of 90% N2, 5% H2, and 5% CO2 (Concept 400, Baker-Russkin Co., USA) for 5 days at 37 °C. Mycoplasma IST 2 (bioMérieux, Lyon, France) assay was applied to detect the presence of Mycoplasma hominis and Ureaplasma spp. after 48 h of incubation using 200 μl inoculum. Selective Thayer-Martin agar (Oxoid, UK) was used to isolate N. gonorrhoeae and the plates were incubated for 72 h at 37 °C in 10% CO2 atmosphere. Isolated bacteria and fungi were identified by ATB/VITEK (bioMériux, Lyon, France) and/or Microflex LT MALDI-TOF mass spectrometer (Bruker Daltonik, Bremen, Germany).

Molecular diagnostic methods

Until DNA isolation, the ultrasonicated samples were stored at − 80 °C. DNA was purified using the QIAamp DNA Mini Kit (QIAgen, Hilden, Germany) tissue protocol according to the manufacturer’s instruction. Extracted DNA was subjected to human β-globin real-time PCR using primers PC04 and GH20 to monitor the presence of PCR inhibitors [11]. Traditional species-specific PCR assays were performed using a Veriti 96 Well Thermal Cycler (Applied Biosystems, US) for the detection of N. gonorrhoeae, C. trachomatis, U. urealyticum, G. vaginalis, A. vaginae, and Mobiluncus spp. The sequences of the primers were described earlier [11,12,13,14,15,16]. Target genes and PCR conditions are shown in Table 1. We used American Type Culture Collection reference strains as positive controls for the PCR assays (Table 1).

Table 1 Reference strains, target genes, primer sequences and conditions used for the detection of bacteria with PCR
Full size table

Statistical evaluation

The comparison of the PCR data was statistically evaluated using the Chi squared test with the Sigmaplot 12.0 program (Systat Sofware, Inc.)

Results

Clinical characteristics of patients

The mean age of the patients involved in this study was 44.9 years, ranging between 24 and 73 years. No patient was treated by antibiotics at the time of the removal of the IUD. Due to the bad patient’s compliance towards gynecology services in Hungary, we found some extremely “old” IUDs (nine patients had the IUD in place for ≥ 20 years). The average duration time the IUDs in place was 8.2 years, with a range between 0.5 and 40 years. The patients were put into three groups according to the duration time of the IUD in place: < 5 years (32 patients), 5–< 10 years (36 patients) and ≥ 10 years (32 patients). The mean age of the patients in the three groups at the time of removal of the IUDs are shown in Table 2 together with the clinical symptoms. According to the clinical data at the removal of the IUDs 43% of the patients (equally distributed in the three groups 43.75%, 47.2% and 37.5%, respectively) had no symptoms of infection, but irregular bleeding (in 24 patients), and leiomyoma of the uterus (in 7 patients) were the reasons of removal of the IUD. In all three groups, only few patients visited the gynecologist for the removal of the IUD due to the time in place (Table 2). There were six patients without any symptoms, who wanted to be pregnant and the IUD was removed for this reason and two who became pregnant despite IUD use. Postmenopausal condition was the reason of removal of the IUD in the case of 12 patients. In this cases the insertion of the devices was much earlier, and the patients went through the perimenopausal period with intact IUDs (Table 2). Altogether 57% of the patients had any symptoms of possible infection of the genital tract with or without fever or irregular bleeding, such as acute vaginal/cervical inflammation (n = 30) or severe lower abdominal pain (n = 29) at the time of the removal of the IUD. Postmenopausal conditions of the uterus and cervix could be observed in two patients.

Table 2 Clinical symptoms of the patients at the time of removal of the IUD
Full size table

Culture results of the removed IUDs

Table 3 shows the culture results of the IUDs according to their different time periods in place. Altogether 62% of the IUDs had positive culture results. Out of the culture positive samples, 30 proved to be positive only for aerobic bacteria with one or two different species (Table 3). The most frequent aerobic bacteria were Staphylococcus aureus (10 isolates) Enterococcus faecalis (9 isolates) Streptococcus spp. excluding Streptococcus agalactiae (8 isolates) and Escherichia coli (6 isolates). The prevalence of some of these aerobic species are listed separately in the Table 3. S. agalactiae colonization of the 14 IUDs was observed, with high colony forming units (CFU), independent how long they were in place. They were also isolated as the only pathogen or together with other aerobic or anaerobic bacteria.

Table 3 Culture results of the IUDs removed after different times in place
Full size table

In the case of 28 IUDs, almost equally distributed in the three groups according to their time in place, a complex anaerobic flora was isolated (≥ 3 anaerobic strains) with or without aerobic species. In the three groups of IUDs, ranged according to their time in place, we had positive culture for ≥ 3 anaerobic bacteria, characteristic for BV, almost in the same percentage (25.0%, 30.5% and 28.1%, respectively). The most frequent anaerobic isolates were Gram-positive anaerobic cocci (GPAC) (24 isolates), Actinomyces spp. (18 isolates), Prevotella spp. (12 isolates), Bacteroides spp. (10 isolates) and Clostridium spp. (7 isolates). However, bacteria typically present in BV such as G. vaginalis, A. vaginae, and Mobiluncus spp., were isolated from the samples of the removed IUDs only seldom, such as 1, 2, and 4 cases, respectively (data not shown). Using the MALDI-TOF MS-based identification, out of the 18 Actinomyces isolates, 16 were identified on species level as A. naeslundii (5), A. mayeri (4), A. odontolyticus (3), A. viscosus (2), and A. israelii (2). In the case of two IUDs (removed after 4 and 7 years) only A. odontolyticus or A. naeslundii were isolated in pure culture after 5 days of anaerobic incubation (data not shown). Ureaplasma urealyticum was detected in 6 samples and M. hominis and N. gonorrhoeae were not detected in any biofilm samples of the IUDs by culture method. Candida albicans was isolated only from two IUDs in high CFUs, one that was in place for 3 years and the other for 6.5 years (Table 3).

Results of PCR detection of bacteria from the ultra-sonicated IUD samples

The specificity of species-specific PCR primer pairs used during this study was successfully confirmed with ATCC reference strains for six bacteria considered to be STI pathogens (N. gonorrhoeae, C. trachomatis) or frequently found in BV (G. vaginalis, A. vaginae, Mobiluncus spp. and U. urealyticum) (Table 1). We have not found any positive PCR reaction for N. gonorrhoeae and only one C. trachomatis positive result was obtained among the IUDs, which was removed after 17 years in place from a patient who had severe lower abdominal pain. The biofilm of the same IUD was positive for A. vaginae, G. vaginalis and U. urealyticum by PCR and for A. israelii, GPACs, P. oralis and E. coli by the culture method. With the PCR method, 76% of the IUD samples were found to be positive at least for one of the four bacteria frequently found in BV with some increasing percentage according to the time of the IUDs in place (65.6%, 80.6% and 81.3%, respectively) (Table 4). G. vaginalis was detected in 62% of the examined IUD samples, A. vaginae was present in 32% of the specimens, Mobiluncus-specific gene was detected in 23% and U. urealyticum PCR was positive for 16% of the samples (Table 4). There was no significant difference in the positive results of any of the BV “signaling” bacteria even not for A. vaginae (p-0.328) or Mobiluncus spp. (p-0.192) according to the time of the IUDs in place. Although the numbers of the positive PCR results for these pathogens are increased in the older age IUD groups, no significant differences were found due to the limited number of patients. The combination of the different BV “signaling” bacteria was also evaluated. The most frequently found double positivity was the PCR detection of G. vaginalis + A. vaginae.

Table 4 PCR positivity for N. gonorrhoeae, C. trachomatis and the BV “signaling” bacteria on IUDs in place for different time periods
Full size table

Discussion

IUDs are popular contraceptive choices for women however, the possible risk of PID associated with the use of an IUD has been a long-lasting important concern throughout the world. The connection between the development of any upper genital tract infection immediately after the insertion of the IUD, or due to the extended duration of IUD in place has been studied extensively during the past decades, but controversial results were obtained [17]. Although higher rates of PID immediately after insertion of IUDs has been noted in previous studies, some recent data allow for a finer analysis of this relation [2, 18]. Sexually transmitted organisms, especially N. gonorrhoeae and C. trachomatis were implicated in many cases of PID; however, microorganisms that comprise the vaginal flora (e.g., different anaerobic species dominating the vaginal flora in BV, G. vaginalis, S. aureus, enteric Gram-negative rods, Enterococcus sp. or S. agalactiae) have also been associated with PID [19]. In addition, M. hominis, U. urealyticum, M. genitalium and in rare cases, bacteria from the oral region (Haemophilus influenza or Streptococcus pyogenes) have also been reported to be associated with some cases of PID [20,21,22]. How the frequency of the use of IUD and the bacteria, which may form a biofilm on it, influences the development of upper genital tract infection, is still a question.

In our study, we used extended incubation time (5 days) for isolation of aerobic and anaerobic bacteria/fungi as well as PCR detection of selected bacteria, which are known to cause female upper genital tract infection such as N. gonorrhoeae and C. trachomatis. We also detected by PCR some difficult to culture BV “signaling” bacteria, such as G. vaginalis, A. vaginae, Mobiluncus spp. and U. urealyticum. Due to the poor patient’s compliance, there were quite a few extremely old IUDs removed with or without symptoms of infection. Only 32% of the patients had an IUD in place for less than 5 years, while the manufacturer’s recommendation is to change the device within 4–5 years, depending on the type. The high number of extremely old IUDs (32% of them were in place for more than 10 years up to 40 years) can probably be explained by the fact that the majority of the examined patient had denied their minor genital complains. No differences were found in culture positivity of the IUDs removed in different time-periods in place. In all three categories of the IUDs, 65.6%, 55.6% and 65.6% of them, respectively, were positive by culture for aerobic and/or anaerobic bacteria (p-0.609) (Table 3). The negative results for N. gonorrhoeae using culture- and PCR-based methods, as well as the one positive sample for C. trachomatis found by PCR, support the opinion of others that in low risk population IUD usage, even for long time in place, may not increase the possibility of PID caused by STI pathogens [5, 6]. However, it seems that biofilm formation and the presence of different other bacteria (often characteristic for BV) in the biofilm may happen with the same frequency on IUDs in place for different time-periods. Two examples from this study (one from group < 5 years, other from group > 10 years) demonstrates the microbiological flora’s complexity we could detected. An IUD of a 34 year old women, which was inserted only 1.5 year earlier was positive for several bacteria using different method (culture positive for A. naeslundii, Mobiluncus sp., Peptostreptococcus anaerobius, Prevotella buccalis, Prevotella oralis, Prevotella loescheii and PCR positive for G. vaginalis and A. vaginae). The biofilm removed from another IUD of a 42 years old patient, which was in place for 20 years, was culture positive for A. naeslundii, Anaerococcus prevotii, Bacteroides fragilis Prevotella bivia, Bifidobacterium sp., Gemella morbillorum, S. agalactiae and PCR positive for G. vaginalis and A. vaginae (not shown).

The frequent isolation of S. agalactiae from the biofilm of all three groups of IUDs draw the attention that S. agalactiae can be a frequent colonizer of the vagina of IUD users, and screening for its presence at the end of the pregnancy is important especially in women who used IUD as birth control earlier [23]. In our study, only two IUDs being in place for 3 and 6.5 years were positive by culture for C. albicans. In a recent study, Zahran et al. [24] recovered Candida species in 46.4% of 56 IUDs. Non-albicans Candida spp. were more frequent among the isolates and strong biofilm formation could be visualized by scanning electron microscope.

The high number of Actinomyces isolates found during this study belonging to five different species is in agreement with recent findings about the association of upper-genital-tract infections due to Actinomyces spp. in women with IUD in place for different time-periods [23, 25,26,27]. Clinicians may consider this possibility in case of clinical symptoms of women using IUDs.

Very high number of positive PCR results were observed in all three groups of IUDs detecting the selected bacteria often present in BV, compared with the culture results. Despite the very careful anaerobic culture procedure and the extended incubation time up to 5 days in the anaerobic environment, the confirmation of the possible presence of BV-related bacteria by the culture method in the biofilm on the IUDs was less successful than to detect BV “signaling” bacteria by the PCR method for G. vaginalis, A. vaginae and Mobiluncus spp. It is well known that PCR detection of any bacteria raise the question that non-viable bacteria can also be detected, which may explain this finding. However, the frequent detection of BV associated bacteria or others in the biofilm, present on the surface of the IUDs independent how long time they were in place may draw the attention that in case a severe symptoms culture and PCR-based detection of bacteria and fungi of the removed IUD could help in selection of proper treatment.

Conclusions

Our study confirms the hypothesis that IUDs, independent of the type or duration of use can increase the possibility of an up-spreading gynecological infection. It is a therapeutic challenge to manage these cases because of the mixed bacterial biofilm flora found on the implant’s surface.