Using Pharmacokinetic Results to Investigate Treatment Compliance As a Potential Contributor for Contraceptive Failure

Pearl Index Study With Levonorgestrel-releasing Intravaginal Rings

R. Nave; J. Höchel; U. Mellinger; A. Kohnke; J. Elliesen; H. Schmitz


Hum Reprod. 2020;35(11):2515-2523. 

In This Article



The study was conducted at 105 recruiting study centers (82 sites in the USA and 23 sites in Japan). In total, 2315 participants were screened for inclusion, and of those 1471 participants were treated (1166 participants in the USA and 305 participants in Japan) (Table I).

At the time the study was terminated, none of the participants had completed the 12-month treatment. Therefore, for most of the participants (70.9%, 1043 participants), termination of the study by the sponsor was the reason for discontinuing the treatment. After the study termination, 1067 participants completed the follow-up visit. A total of 404 participants did not follow the invitation to attend a follow-up visit. The mean treatment duration was 134.9 days (121.1 days in the USA and 187.7 days in Japan).

The full analysis set (FAS) consisted of 1471 participants (1166 participants (79.3%) in the USA and 305 participants (20.7%) in Japan) and included all participants admitted to the treatment phase. The PK subgroup consisted of 136 participants (106 in the USA and 30 in Japan) with at least one valid PK value during treatment.

The largest race groups were White (54.5%), Black (21.8%) and Asian (21.4%). In total, 15.1% of the study participants were Hispanic or Latino by their ethnic origin. The age of the participants at enrollment ranged from 18 to 35 years. At baseline, the mean body weight was 72.85 kg and the mean BMI was 27.34 kg/m2. However, these characteristics differed between the USA and Japan subpopulations. Overall, the demographics and baseline characteristics of the PK subgroup reflected well the study population (Table II).

Over half of the participants (59.5%) were nulliparous, and most of the remaining participants had had 1 or 2 births (18.7% and 14.5%, respectively). The average length of the natural menstrual cycle was 28.5 days. At screening, barrier methods and oral hormonal contraceptives were the most frequently reported current contraceptive methods (reported by 56.6% and 17.1% of the participants, respectively).

Treatment Duration and Compliance

The extent of exposure was evaluated based on the eDiary entries, which were to be completed by the participants every day throughout the study. The extent of exposure to the study drug was defined as the number of days from the start of treatment until the last day the IVR was worn. Overall, the mean (SD) exposure was 139.2 (83.2) days. In the USA, it was 125.4 (83.6) days, which consisted on average of 4.0 treatment cycles and the use of 4.5 IVRs. In Japan, the mean (SD) exposure to treatment was 192.1 (56.1) days, consisting on average of 6.5 treatment cycles and the use of 7.1 IVRs.

Furthermore, treatment compliance was also evaluated based on the eDiary entries using predefined overall compliance criteria (worn an IVR for ≥80% of the treatment period). As a result 95.3% of participants were considered compliant (USA, 94.8% and Japan, 97.4%). In order to be effective as a contraceptive, the IVR needs to be worn continuously, and therefore, the compliance should be 100%. A total of 75.9% of participants were considered 100% compliant (all details are provided in the Supplementary Table SI).

Contraceptive Efficacy

In total, 41 pregnancies were observed; of these, nine pregnancies occurred with the EDC before the start of treatment, and two pregnancies occurred after the participant had stopped the study treatment. In the remaining 30 pregnancies, the date of conception was during treatment: 28 in the USA and 2 in Japan (see Supplementary Table SII). Of the 30 participants with an unintended pregnancy during treatment, 10 had reports of unplanned backup contraception in their eDiary records. However, only one participant had a documented wearing break dated close to the EDC.

By the time the study was terminated, the total exposure time was 560.8 woman-years and the relevant exposure time (excluding time with additional contraceptive methods) was 495.7 woman-years. This led overall to a PIU of 6.1 (95% CI (4.1; 8.6)). When adjusted for non-compliance, the relevant exposure time was 465.0 woman-years and the PIA was 6.5 (95% CI (4.4; 9.2)). Due to the high number of pregnancies in the USA, the PIU for the USA subgroup was 8.2 (95% CI (5.4; 11.8)) and the PIA was 8.8 (95% CI (5.9; 12.8)), whereas in Japan, both PIs were below 1.4 (1.3, 95% CI (0.2; 4.7) and 1.3, 95% CI (0.2; 4.9), respectively). However, the result from the Japanese subgroup is based on a relatively small number of participants (305 participants) with on average 6.5 treatment cycles and 154.0 woman-years of relevant exposure.

As a secondary analysis, the cumulative failure rate, i.e. the probability of getting pregnant was calculated using the Kaplan–Meier method. Overall, the cumulative failure rate was 3.8%. However, this was mainly driven by the US population, in which the failure rate was 5.4% (95% CI (3.5; 8.3)), whereas in Japan the failure rate was 0.7% (95% CI (0.2; 2.7)).


It was planned to collect PK samples for LNG concentration at baseline and under treatment after 6 months and 12 months (planned EOT) in a total of 150 participants. In total, 108 samples were taken in the USA and 57 samples in Japan during treatment. The 6-month PK samples were taken at a median (min–max) on days 182 (168–208) and 169 (166–190) in USA and Japan, respectively. Due to the early termination of the study, some PK samples were collected at the participants actual EOT visit independent from duration of treatment (between 10 and 337 days after start of treatment). Therefore, the time variability with respect to the treatment duration was relatively high for the EOT sample (Supplementary Table SIII).

At baseline, 5 out of 96 participants had LNG concentration ≥LLOQ (10 ng/l). This is not relevant for the PK evaluation as this reflects the use of hormonal contraception during screening, which was allowed prior to the start of treatment.

The PK data indicated that the steady-state concentration of plasma LNG was, with mean values of 91–115 ng/l, markedly lower in the participants in the USA compared to Japan with about 260 ng/l. The summary statistics for plasma LNG concentration are presented in Table III. Overall, the mean values of LNG concentration were similar at both visits (Figure 1). However, it is to be noted that altogether in 19 samples taken during treatment the LNG concentration was below LLOQ (17 out of 108 samples in the USA and in 2 out of 57 samples in Japan, Table III).

Figure 1.

Individual plasma LNG concentrations under treatment for participants with two samples (Japan left, n = 27 and USA right, n = 24). The box ranges from 25th to 75th percentile. The marker inside the box indicates the arithmetic mean value. The line inside the box indicates the median value. The whiskers are drawn from the box indicate the 10th to 90th percentile range. EOT, end of treatment; LNG, levonorgestrel.

It is worthwhile to note that two of the participants (both in the US study population) with on-treatment pregnancies belong to the PK subgroup. In both cases, the LNG concentration in plasma was below LLOQ. In one case, the EDC was 7 days after the start of treatment. Based on the eDiary, the PK sample was taken 16 days after the EOT. (Supplementary Table SIV) In the other case, the EDC was approximately 6 months after the start of treatment, the PK sample was taken 5 days after the EDC, and the EOT was about 3 weeks thereafter. Interestingly, 106 days are recorded as relative days to start of actual IVR (all details are provided in the Supplementary Table SIV).

Another observation related to the PK samples taken in the USA was a longer duration of use than the intended wearing period of 28 ± 7 days. For all PK samples, the relative day to the insertion of the actual IVR was recorded. These data indicated extended wearing periods of more than 35 days in 19 cases; thereof, more than 84 days in 9 cases and a maximum of 175 days (Supplementary Table SIII). However, in these cases, the participant may have forgotten to document the IVR replacements in the eDiary. Furthermore, an evaluation of the timing of PK (EOT) sampling related to IVR insertion and removal of the samples with low LNG concentrations (<50 ng/l including LLOQ) indicated for at least 7 out of 12 samples an IVR free interval of at least 4 days.


Median SHBG concentrations in serum at baseline were 61 nmol/l. After the start of treatment SHBG concentrations decreased slightly in all dose groups and remained at approximately 42 nmol/l. There were no differences between the regions (Supplementary Figure S1).


Safety analysis was based on the FAS, including all 1471 participants who received treatment. The treatment was not associated with any new or unexpected safety events and was generally well-tolerated in the study population.

From a total of 1471 participants treated with LNG, 756 participants (51.4%) reported at least one treatment-emergent adverse event (TEAE) (49.5% of participants in the USA and 58.7% of participants in Japan). The majority reported TEAEs of mild (30.8%) or moderate (18.5%) intensity. The most common TEAE was ovarian cyst (227 participants, 15.4%), while all other TEAEs were reported in <4% of participants. TEAEs of severe intensity were reported in 31 participants (2.1%). The TEAEs most frequently classified as severe were migraine (3 participants, 0.2%), and headache and abdominal pain (2 participants (0.1%) each).

Study drug-related TEAEs were reported in 382 (26.0%) participants. The majority reported AEs of mild (17.1%) or moderate (8.1%) intensity. Most frequent drug-related AEs were ovarian cyst (117 participants, 8.0%), metrorrhagia (48 participants, 3.3%) and vaginal hemorrhage (33 participants, 2.2%), all typical side effects of a progestogen-only hormonal contraceptive (Tayob et al., 1986; Broome et al., 1995; Jadelle® Prescribing Information)

A total of 135 participants (9.2%) discontinued the study due to TEAE(s), and for 112 participants (7.6%) these TEAE(s) were considered drug-related. The most common TEAEs that led to discontinuation were metrorrhagia (21 participants, 1.4%) and vaginal hemorrhage (18 participants, 1.2%). A total of 8 participants (0.5%) experienced at least one serious TEAE during the study. One participant had a serious TEAE (pelvic mass—not further specified) that was considered related to the study drug. All serious AEs were considered resolving or resolved by the investigator.

One death occurred during screening (homicide).

Majority of participants did not have clinically meaningful changes in the laboratory values, or vital signs or body weight.

By the time of the study termination, no major changes had been observed in the vaginal bleeding pattern; slight decreases were seen in the bleeding/spotting intensity, length of the bleeding episodes and the number of bleeding and spotting days. However, due to the limited observation time, the interpretation of these data is limited.