Study population
We prospectively identified all women referred for investigation of cervical dysplasia at Østfold Hospital Trust, Health Region East, Norway between October 2005 and February 2007. Women planned for their first LEEP for suspected high-grade disease were eligible (Fig. 1). Women on long-term antibiotics were excluded leaving 89 eligible women in the case group, termed the “LEEP-group”. Of these, 81 women were treated for CIN3, six for CIN2 and two for persistent CIN1. Women requiring antibiotics after LEEP and pregnant women were excluded in the follow-up analysis. Two women were lost to follow up at six months and four women at 12 months leaving 77 women for the follow up analysis at six months and 72 women at 12 months.
A population of 100 healthy references, defined by the presence of normal cytology, was recruited at a private gynecological practice, and termed the “reference group” (Fig. 1).
A detailed history was taken by the consulting clinician to include sociodemographical data, including year of birth, education level, marital status, smoking habits and medical history including use of recent antibiotics and contraception.
Sample collection
Microbiota
With sterile speculum examination, serial Liquid Based Cytology (LBC) and five endocervical bacterial swabs were collected prior to treatment and at six and 12 months post-treatment as well as at baseline for the reference population. One swab was transported on Stuart medium for cultivation of anaerobic and aerobic bacteria on blood and lactose agar plates. Growth results were reported as growth or no growth for the following bacteria; Bacteroides spp., Escherichia coli, Gardnerella vaginalis, Streptococcus spp. and Lactobacillus spp. Other bacterial species detected at cultivation were also reported. The four remaining swabs were transported in Copan Universal Transport Medium (UTM-RT) for detection of Chlamydia trachomatis, using the COBAS TaqMan 48 CT test (Roche Molecular Diagnostics, Pleasanton, CA), and for detection of Mycoplasma hominis, Ureaplasma urealyticum and Ureaplasma parvum. DNA was extracted from the swabs using the BioRobot M48 (Qiagen, Hilden, Germany) and the Mag Attract DNA bact protocol. Amplification and detection of U. parvum, U. urealyticum and M. hominis were performed using the ABI Prism 7900HT Sequence Detection System (Applied Biosystems, Foster city, CA) [20, 21].
In every woman all bacterial species found were recorded. All women that had at least one type of one non-Lactobacillus bacteria were included in a constructed group called“ any non-Lactobacillus”.
HPV tests and histology
Cervical LBC specimens were collected with a Cytobrush Plus (Medscan Medical AB, Sweden) and transferred to PreservCyt medium (item no 0234005, Cytyc Corporation, Marlborough, USA) in the follow-up samples of the treated population. Nucleic acid extraction and high-risk HPV DNA testing (Amplicor HPV test, Roche Diagnostics Switzerland) was performed as previously described [22]. All biopsy and cone histological specimens were re-evaluted blindly by the same experienced pathologist and diagnosed according to WHO classification [23]. The cone depth was documented in millimeters (mm). The specimen with the most severe lesion was chosen for diagnosis.
Statistical analyses
Background characteristics of sociodemographical data and contraception in the reference group and the LEEP group were compared using chi-square or independent t-tests.
The number of different non-Lactobacillus species at each assessment was recorded. The mean number of non-Lactobacillus before and after LEEP was compared with paired t-test (at six or 12 months). The change in mean number of non-Lactobacillus after LEEP was compared between women below 46 years and women 46 years and above with independent t-test. This change in mean number of non-Lactobacillus after six and 12 months follow up was also compared between women without hormonal contraceptive and women with hormonal contraceptive and between women in a relationship (married/cohabiting) and single women. To study if LEEP resulted in a change of different bacterial species in each individual we focused on the women with a change in detection of bacteria after LEEP compared to before LEEP (those who either lost or aquired bacteria after LEEP compared to before LEEP). We compared the number of women who lost a specific species of bacteria that were present before LEEP to the number of women that aquired that particular bacteria species after LEEP. This was performed using paired data from before LEEP and from follow up (at six or 12 months) with McNemar test.
HPV DNA status post-treatment for assessment of persistence of HPV was studied. The mean number of non-Lactobacillus was compared between HPV negative and HPV positive women with independent t-test at six and 12 months follow up. Difference in detection of Lactobacillus and the constructed group any non-Lactobacillus according to HPV DNA positivity after LEEP was analysed with Fisher’s exact test. The mean aquisition or loss of detectable non-Lactobacillus bacterial species at six and 12 months follow up compared to before treatment was analysed in relation to HPV-DNA status after LEEP using independent t-test.
We compared the cone depth between women testing positive or negative for HPV DNA at both six and 12 months with independent t-test. We further assessed whether change in the microbiota (of the number of different species of non-Lactobacillus) at six and 12 months after treatment was correlated to the cone depth using the Pearson Correlation test.
The mean number of detected bacterial species in the reference group and the LEEP group before treatment was compared using an independent t-test. The differences in individual bacterial species detected and the constructed broader group“ any non-Lactobacillus” were compared between the reference group and the LEEP group before treatment and at six and 12 months follow up using Fishers exact test and logistic regression adjusted for age (< 46, 46–55, > 55 years), marital status (living together (married/cohabiting) or living alone (single)), smoking (none, 1–10/day, > 10/day) and use of hormonal contraception (yes or no).
Statistical analyses were performed for the whole population and in a subgroup including only women below the age of 46 (assumed premenopausal).
For all statistical tests a p-value < 0.05 was accepted as statistically significant. Statistical analyses were performed using IBM SPSS software, version 24.0.