Effect of oral contraceptives and doxycycline on endometrial MMP-2 AND MMP-9 activity

Abstract

Objectives

To describe the effect of combined oral contraceptives (COC) on Matrix Metalloproteinase (MMP) -2 and -9 activity and compare MMP activity in women taking aCOC with or without doxycycline.

Study Design

Subjects (n=20) underwent endometrial biopsies1) in the late luteal phase of a baseline cycle prior to initiating COCs, 2) on day 19 to 21 while taking COCs in a standard 28-day cycle (7-day hormone free interval), 3) on day 26 to 28 while taking active COCs continuously for a 28 day cycle. During the continuous COC cycle, they were randomized to receive daily sub-antimicrobial dose doxycycline 40 mg or placebo.

Results

Compared to baseline, COC treatment increased MMP-2 (p<0.001) and -9(p<0.001). MMP activity was lower in subjects taking a COC with doxycycline compared to those receiving placebo, although only significantly lower for MMP-2LF (p=0.002).

Conclusions

Unscheduled bleeding with COCs may be the result of increased endometrial MMPs. Sample size limitations prevent us from determining how doxycycline affects MMP activity in COC-users.

1. Introduction

Unscheduled bleeding is a common side effect of cyclic, extended and continuous combined oral contraceptives (COC)[1].Recent investigations aimed at decreasing unscheduled bleeding with hormonal contraceptives have focused on preventing endometrial degradation on a molecular level by altering matrix metalloproteinases (MMPs)[2–5]. MMPs are present throughout the endometrium with higher MMP activity associated with greater endometrial degradation and subsequent bleeding[6, 7].

Understanding the molecular effects of COCs and MMP inhibitors like doxycycline on the endometrium may contribute to our understanding of why women experience unscheduled bleeding and what can be done to decrease this unwanted side effect. In a randomized placebo-controlled trial we found doxycycline did not decrease bleeding when it was administered as a treatment measure once bleeding had started [12]. We hypothesized that once endometrial degradation began, inhibition of MMP activity with doxycycline was not sufficient to stop bleeding but if MMP activity could be inhibited prior to the onset of endometrial degradation, this could result in less unscheduled bleeding. In a follow-up randomized placebo-controlled trial, we demonstrated that prophylactic administration of sub-antimicrobial doxycycline (40 mg daily) allowed women taking a COC in a continuous manner to achieve menstrual suppression sooner than those who took a placebo[13].

Despite containing both an estrogen and a progestin, all COCs have a progestin-dominant effect on the endometrium. Exogenous progestin, like those found in COCs, may up-regulate MMP activity [4]. Other medications like the tetracyclines, inhibit MMPs. Independent of its antimicrobial properties, doxycycline, at low, sub-antimicrobial doses, is a potent MMP inhibitor[8, 9].

We elected to study MMP-2 and MMP-9 based on their known roles in endometrial degradation and menstruation and their hypothesized role in unscheduled bleeding associated with hormonal contraceptive use. MMPs are grouped according to their dominant structure into collagenases, gelatinases, stromelysins and membrane-bound types and correspondingly degrade those extracellular matrix components [10]. MMP-9 and MMP-2 are a gelatinases that degrade collagens IV and V, elastin, and gelatin components of the extracellular matrix, and the subendothelial basement membrane thereby increasing vessel fragility. The hypothesized outcome is increased endometrial bleeding [11].

The purpose of the current study was to describe MMP-2 and -9 activity in endometrial biopsies collected 1) prior to initiating COCs, 2) while taking COCs in a standard 28-day cycle (7-day hormone free interval), 3) while taking 28 days of hormonally active COCs, and 4) while taking 28 days of hormonally active COCs with doxycycline. We hypothesized the progestin-dominant effect of COCs would result in increased in MMP activity from baseline. Based on doxycycline’s actions as an MMP-inhibitor, we also hypothesized MMP activity would be lower in the group taking a COC with doxycycline compared to those taking a COC alone.

2. Materials and Methods

2.1 Subjects and sample collection

A prospective randomized study (Committee on Human Subjects #19375, Clinicaltrials.gov NCT01469585)was conducted in Honolulu, Hawaii between December 2011 and May 2012. Ovulatory women (progesterone level ≥3.0 ng/ml, day 18–20), 18 to 45 years of age with no contraindications to COCs or doxycycline were invited to participate. Exclusion criteria included use of an intrauterine device or contraceptive implant (within four weeks) or depot medroxyprogesterone acetate (within 9 months), smoking, or irregular menstrual cycles.

Each participant had three endometrial biopsies over three 28-day cycles (84 days total). A baseline biopsy was done prior to initiating any medications on day 19 to 21 of the menstrual cycle (baseline cycle). On the first day of spontaneous menses of treatment cycle 1, participants started a daily COC (20 μg ethinyl estradiol/100 μg levonorgestrel, Lutera®, Watson Pharma, Inc., Corona, CA). The second biopsy was performed on day 19 to 21 (treatment cycle 1). On day 21 of treatment cycle 1, participants began the 7-day placebo week as they would with typical COCs.

During treatment cycle 2, participants took hormonally active pills for all 28 days and were randomized to either daily controlled-release sub-antimicrobial dose doxycycline 40 mg (Oracea®, Galderma Laboratories L.P., Fort Worth, TX) or no additional medication using simple randomization. The third endometrial biopsy was performed on day 26 to 28 (treatment cycle 2).The snap frozen samples were stored in a −80 °C freezer and shipped to Meharry Medical College (Nashville, TN) for analysis.

2.2 Extraction and Activity Assay of MMPs

Samples were homogenized on ice in buffer containing50mM Tris-HCl, 500mM NaCl, 5 mM CaCl₂, 1% Triton X-100, pH 7.6.The homogenates were centrifuged at 11,000 × g (30 min, 4° C) and the supernatants were assayed for MMP activity using gelatin zymography. An aliquot of the supernatant was assayed for protein concentration using the BCA kit (Pierce, USA).Gelatinase zymography was performed using 10% Novex pre-cast polyacrylamide gel (Invitrogen) in the presence of 0.1% gelatin. Equal amounts of total protein were loaded per well and gels were electrophoresed under non-reducing conditions. In addition, a sample containing 2.5 ng each of MMP-2 and -9 (Enzolifesciences, USA) as positive control were run together with experimental samples on each gel. After electrophoresis, gels were treated with renaturing buffer (Invitrogen), followed by incubation in developing buffer (Invitrogen) at 37°C. Gels were stained with Simply Blue Safe Stain (Invitrogen)and washed. Densitometry analyses were carried out using a Biorad gel documentation system with Image Lab software.

2.3 Statistical Analysis

To describe differences in MMP-2 and -9 activity between subject taking a COC along with doxycycline (doxycycline group, biopsy 3) and those taking a COC alone (control group biopsy 3) at-test was used. To describe the effects of COCs on MMP activity, we compared MMP activity between the baseline biopsy (biopsy 1) and the biopsy while taking COCs (treatment cycle 1, biopsy 2) using a paired t-test. All analyses were performed with Statistical Package for the Social Sciences (SPSS) version 16.0 (Chicago, Illinois).A sample size of 20 was selected to give reasonable estimates of relative MMP activity.

3. Results

Of the 37 subjects screened, 20 were randomized (Figure 1). Most women were Asian (30%) or Caucasian (35%), nulliparous (85%) and normal weight (mean BMI 23.9 kg/m²). No significant differences in demographic characteristics emerged between the groups. MMP activity increased significantly when subjects started a COC (baseline biopsy 1 versus treatment cycle 1, biopsy 2) for MMP-9 [1.33 (0.41) versus 2.49 (0.72)], p<0.001), MMP-2 active form (AF) [1.76 (0.47) versus 2.62 (0.72), p<0.001] and MMP-2 latent form (LF) [2.35 (0.42) versus 3.19 (0.88), p<0.001] (Figure 2). Mean MMP activity at the time of the third biopsy was lower in the group randomized to a COC with doxycycline versus a COC alone [MMP-9 1.52 (0.63) versus 1.98 (0.87), (p=0.39), MMP-2AF 1.65 (0.50) versus 2.14 (0.79), (p=0.31)] though only significantly lower for MMP-2LF [2.03 (0.07) versus 3.15 (0.48), p=0.002)] (Figure 2).

Figure 1.

Subjects enrolled, randomized, and analyzed.

*Discontinued because of pain from EMB (n=1), loss to follow up (n=1), scheduling difficulty (i.e. patient could not come in during specified time) (n=1).

**Discontinued because of side effects (mood, fatigue) (n= 2) and scheduling difficulty (n=1)

^†Discontinued because of pain from EMB (n=1)

^‡Discontinued because of scheduling difficulty (n=1)

Figure 2.

Mean MMP-9, MMP-2 AF (active form), and MMP-2 LF (latent form) activity. Error bars depict standard deviation. Baseline biopsy performed on day 19 to 21, cyclic oral contraceptive (COC) biopsy performed on day 19 to 21 of treatment cycle 1 and COC +/− Doxycycline (Doxy) biopsy performed on day 26 to 28 days of treatment cycle 2 where subjects took 28 days or hormonally active pills (no HFI).

4. Discussion

In the current study we sought to describe the effect of COCs on MMP-2 and -9 as well as the effect of doxycycline on these same MMPs in women taking a COC.MMP-9 is present in the endometrium throughout the menstrual cycle with levels highest in the mid to late luteal and peri-menstrual phases [5, 7, 14]. MMP-2 is predominantly expressed in endometrial stromal cells [5, 7]. Similar to MMP-9, expression of MMP-2 is high during the menstrual phase [5]. As gelatinases, MMP-2 and -9 degrade the extracellular matrix and major components of the subendothelial basement membrane which increases vessel fragility [15].High endogenous progesterone concentrations correspond with higher levels of MMP-9 [18]. Exogenous progestins also affects endometrial MMPs with increased MMP activity noted with depot medroxyprogesterone acetate, the levonorgestrel intrauterine system, and levonorgestrel implants [4, 6, 16, 17].A study by Skinner et al investigating the effect of a levonorgestrel intrauterine device on the endometrium reported increased MMP-9 with levonorgestrel IUD use. Focal expression of MMP-2 has been noted in areas of tissue degradation in levonorgestrel implant users [5]. In vitro, estrogen also appears to increase MMP-2 and -9 expression through a vascular endothelial growth factor (VEGF) [19].

The effect of COCs, which contain estrogen and progestin, on endometrial MMP activity has not previously been described. We took our baseline biopsy in the late luteal phase (day 19 to 21) when MMPs are relatively high[14] yet we noted COCs increased MMP-2 and -9 activity above this baseline. With continued exposure to progestin, MMP-2 and -9 levels are thought to decrease. Labied et al demonstrated MMP-2 and -9 decreased over time with LNG-IUD use [15]. With the levonorgestrel implant, mean MMP-9 levels decreased between one and six months of use though the same decrease was not seen with MMP-2 [5]. We also noted a general decrease in MMP activity during treatment cycle 2 with continued COC exposure.

Zhao et al noted a decrease in MMP activity with doxycycline administration. In a 2009 study they found endometrial MMP-9 was significantly lower in women using a levonorgestrel implant who were randomized to receive doxycycline 20 mg twice a day compared to placebo one- and six- months following implant insertion though a change in MMP-2 was not noted [5].Though we intended to describe MMP-2 and -9 activity in women taking a COC alone versus those taking a COC with subantimicrobial doxycycline, drop-out prior to the third biopsy and the resultant small sample size, prevent us from drawing conclusions on this outcome.

Abbreviations

COC

Combined Oral Contraceptives

MMP

Matrix Metalloproteinase