Co-administration of St. John’s wort and hormonal contraceptives: a systematic review^,

Abstract

Objectives:

St. John’s wort (SJW) is a known strong inducer of the cytochrome P450 (CYP) 3 A4 enzyme, and both the ethinyl estradiol and progestin components of hormonal contraceptives are substrates of CYP3A4. This systematic review examined whether the co-administration of SJW and hormonal contraceptives leads to significant safety or efficacy concerns.

Study design:

Systematic review.

Methods:

PubMed and Cochrane Library databases were searched for articles of any comparative study design (clinical or pharmacokinetic) that examined potential interactions between SJW and hormonal contraceptives in women of reproductive age.

Results:

Of the 48 identified articles, four studies met inclusion criteria and compared use of combined oral contraceptives (COCs) alone to the use of COCs co-administered with SJW. Two studies demonstrated no change in markers of ovulation, but one study demonstrated increased follicular growth and probable ovulation when COCs were co-administered with SJW. Three studies demonstrated an increased risk of breakthrough bleeding with COCs and SJW. Three studies showed changes in at least one pharmacokinetic parameter that suggested a significantly decreased exposure to hormone concentrations when COCs were co-administered with SJW. The only study that did not demonstrate any significant pharmacokinetic differences examined a SJW product containing a low amount of hypericin.

Conclusion:

Limited evidence showing increased risk of ovulation and breakthrough bleeding raises concern for decreased contraceptive efficacy when COCs are co-administered with SJW. The pharmacokinetic evidence is mixed but suggests that SJW administration may be associated with weak to moderate induction of the metabolism of COCs.

1. Introduction

Use of complementary and alternative medicine (CAM) has been growing in the United States for decades [1–3]. Surveys of US adults report that 42% to 63% of respondents use some form of CAM therapy [2–5], and about 18% reported use of at least one herbal product in the previous 12 months [1]. Use of herbal preparations and CAM therapies is more common in women than men [1–4]. Approximately 20% of reproductive-age women in the United States report taking herbal products [1,3]. The herbal product known as St. John’s wort (SJW) (Hypericum perforatum) has become a common therapy for the treatment of depression in the United States and around the world [6]. However, SJW has been associated with numerous adverse drug interactions, likely due its ability to induce the cytochrome P450 (CYP) enzyme 3 A4 [7]. The US Food and Drug Administration classifies SJW as a strong inducer of CYP3A4 and weak inducer of CYP2C9 enzymes. A strong inducer for a specific CYP is defined as one that decreases the area under the curve (AUC) of a substrate for that CYP by ≥80% and a weak inducer translates to a 20–50% decrease in AUC [8].

SJW is commonly used along with prescription medications [9]. One study found that, for nearly 28% of office visits in which the use of SJW was documented by the provider, at least one other drug was prescribed that could lead to a potentially harmful combination, including oral contraceptives [9]. Contraceptive steroids are metabolized by both CYP3A4 and CYP2C9 enzymes [10]. Thus, coadministration with CYP3A4 and/or CYP2C9 inducers, such as SJW, may lead to rapid metabolism of steroid hormones, potentially leading to decreased steroid hormone concentrations and increased risk for unintended pregnancy.

Case reports have linked the use of SJW in users of combined oral contraceptives (COCs) with breakthrough bleeding [11] and unintended pregnancy [12]. In addition, the Medicines and Healthcare Products Regulatory Agency of the United Kingdom released a drug safety update about SJW in March 2014 describing 19 reports of potential drug interactions between SJW and hormonal contraceptives. The reporting physicians suspected that 15 cases of unintended pregnancies (4 with etonogestrel implants and 11 with oral contraceptives) and 4 cases of breakthrough bleeding (all with oral contraceptives) were linked to co-administration of the contraceptive and SJW [13]. Although previous published reviews have looked at some of the evidence around SJW and drug interactions, none focused solely on SJW and hormonal contraception [7,14]. Thus, given the theoretical concerns and case reports of potentially significant interactions, this systematic review was conducted to examine the evidence on interactions between SJW and hormonal contraception. We were interested in interactions in both directions, that is, does use of SJW increase or decrease steroid hormone concentrations (from the hormonal contraceptive), possibly leading to decreased contraceptive efficacy or increased risk for adverse events? Additionally, does use of hormonal contraception increase or decrease SJW concentrations, possibly leading to SJW toxicity or decreased SJW efficacy?

The United States Medical Eligibility Criteria for Contraceptive Use, 2010 (US MEC) does not include recommendations for the safe use of contraception with SJW [15]. This systematic review was conducted to prepare for a meeting held at the Centers for Disease Control and Prevention in August 2015 to update the US MEC.

2. Methods

A systematic review was conducted according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [16].

2.1. Search strategy

We searched PubMed and Cochrane Library databases from database inception through December 8, 2015 (Appendix A).

2.2. Selection criteria

We searched for studies that addressed two research questions:

• #1: Among women using hormonal contraception, does use of SJW increase adverse outcomes (due to increased concentrations of the hormonal contraception) or decrease contraceptive efficacy (due to decreased concentrations of the hormonal contraception) compared with non-use of SJW?

• #2: Among women taking SJW, does use of hormonal contraception increase adverse mental or physical health outcomes (e.g., due to decreased or increased concentrations of SJW) compared with non-use of hormonal contraception?

Articles were included if they examined the co-administration of SJW (of any dose) with any type of hormonal contraceptive (COCs, transdermal patches, or vaginal rings; progestin-only implants, injectables, or pills; emergency contraceptive pills; or levonorgestrel intrauterine devices) among women of reproductive age. Articles in any language and of any comparative study design were included; however, case reports and case series, abstracts, and unpublished data, such as theses or dissertations, were excluded. Clinical outcomes of interest were pregnancy, ovulation, breakthrough bleeding, and adverse events. As we anticipated limited availability of clinical data, we also included articles that examined pharmacokinetic (PK) outcomes of either SJW or hormonal contraceptive steroid hormones. All PK parameters measured were included. A common method for assessing potential clinical significance of significant changes in PK parameters is to calculate geometric mean ratios for various parameters (e.g., geometric mean ratio for area under the curve_{drug A} [AUC]=AUC_{drug A} in users of drug B/AUC_{drug A} in non-users of drug B ×100), construct 90% confidence intervals (CIs) around that ratio, and set a predefined range that would suggest a lack of interaction (bioequivalent). For this review, the predefined range was a 90% CI of 80–125%. Geometric mean ratios with 90% CIs falling outside that range were said to suggest bioinequivalence. We excluded CYP genotyping outcomes and studies that examined the effects of a third co-administered drug.

2.3. Study selection and quality rating

One author (E.B.B.) searched all titles and abstracts and identified articles that required full-text review. Two authors (E.B.B. and K.M.C.) identified full-text articles that met inclusion criteria. Clinical studies were assigned a quality rating according to the US Preventative Services Task Force rating system. Because a standard rating system for PK studies has not been established, we developed and applied one (Appendix B). Two authors (E.B.B. and K.C.) independently assigned a quality rating to each article. The ratings were then compared, and a third author (M.J.K.) resolved any differences.

2.4. Data synthesis

Evidence tables were constructed to assist in evaluating and summarizing the data. Meta-analysis could not be conducted because of the heterogeneity of study designs and outcomes.

3. Results

The search yielded 48 articles, of which four studies met inclusion criteria. No studies addressed whether use of hormonal contraception among women taking SJW increases adverse mental or physical health outcomes compared with non-use of hormonal contraception (research question #2). All four studies addressed whether use of SJW among women using hormonal contraception increases adverse events or decreases contraceptive efficacy, compared with non-use of SJW (research question #1). Three studies examined clinical and PK outcomes [17–19], and one included only PK outcomes [20]. All studies examined the co-administration of COCs with SJW. None of them examined use of progestin-only contraceptives or non-oral contraceptive methods with SJW. One study used a COC containing 35 μg of ethinyl estradiol (EE) [17], while the other three used a COC with 20 μg of EE. Currently, SJW products are standardized based on hypericin levels; one study used a 0.2% hypericin SJW product [20], and all others used a 0.3% hypericin SJW product. None of the studies examined pregnancy rates. In an effort to measure ovulation, three studies evaluated progesterone concentrations [17–19] and two studies evaluated follicle maturation [17,19]. All four studies evaluated relevant PK parameters of the estrogen and progestin components of the COC. No studies evaluated the PK parameters of compounds found in SJW.

One study enrolled 12 healthy women (27±7 years) who had been taking any COC for at least 3 months before study initiation (without breakthrough bleeding) [17]. This fixed-order study took place over three sequential 28-day menstrual cycles. During cycle 1, participants took a COC (EE 35 μg/norethindrone [NE] 1 mg) for 21 days followed by the inactive pills for days 22–28. In cycles 2 and 3, the same COC schedule was used and participants took SJW (0.3% hypericin) 300 mg three times daily for 28 days. SJW was not associated with significant effects on either the peak-to-mean ratios for the serum concentration of follicle-stimulating hormone (FSH), luteinizing hormone (LH) or Day 21 progesterone concentrations. More subjects reported breakthrough bleeding with SJW co-administration (7/12) when compared with the COC alone cycle (2/12, no p-value reported). On Day 7 of cycles 1 and 3, serial blood samples were used to calculate AUC, clearance (CL), half-life (t_1/2), maximum concentration (C_max), and time to maximum concentration (t_max) of EE and NE. The geometric mean ratios of AUC and C_max (COC+SJW [cycle 3] vs. COC alone [cycle 1]) were calculated. For EE, SJW administration plus COC use was associated with a significantly reduced t_1/2 (23.4±19.5 h [COC alone] vs. 12.2±7.7 [COC+SJW]; p=.023), but no significant changes were seen for the AUC, CL, C_max or t_max (Table 1). For NE, SJW administration plus COC use was associated with significant increase in CL (L/h) (8.2±2.7 vs. 9.5±3.4; p=.042) and a significant decrease in C_max (μg/L) (17.4±5.1 h vs. 16.4±75.2; p=.045). However, the 90% CIs for each of these significant differences (for both steroid hormones) fell at least partially within 80–125% range, thus failing to demonstrate bioinequivalence (Table 1). Therefore, while this study identified some statistical differences which overall demonstrated a pattern of decreased exposure to the contraceptive hormones among SJW users, these changes were not great enough to firmly establish bioinequivalence and thus are of uncertain clinical significance.

Table 1.

Evidence table for the use of COCs and SJW

Study and Funding Source		Design		Study population		Exposure		Outcomes	Results			Strengths	Limitations			Quality
Studies with PK and clinical outcomes
Hall, 2003. FDA Office of Women’s Health and NIH		PK; sequential fixed-order		Healthy premenopausal women (using COCs for at least the previous 3 months) (n=12)		Month 1: COC (EE 35 μg/NE 1 mg) for 28 days Month 2: COC+SJW 300 mg three times daily for 28 days Month 3: COC+SJW 300 mg three times daily for 28 days (SJW: 0.3% hypericin)		Day 7: AUC, Cma, t1/2, tmax for EE and NE; Day 12–16: ratio of peak to mean concentration FSH and LH for each individual; Day 21: serum [progesterone] Breakthrough bleeding	Progesterone: “unaffected” Breakthrough Bleeding: OCP alone: 2/12 (month 1) COC+SJW 7/12 (month 3) FSH and LH: “SJW had no significant effects on ratios for FSH and LH at mid-cycle” 6/12 however had increase in peak-to-mean ratios affected. Norethindrone AUC ([μg · h]/L) 131.8±35.1 vs. 118.3±41.1 (p=15) Ratio and 90% CI 88% (76% to 100%) CLoral (L/h) 8.2±2.7 vs. 9.5±3.4 (p=.042) 114% (100% to 127%) T1/2 (h) 12.6±7.2 vs. 12.1±4.9 (p=.799) 102% (85% to 120%) Cmax (μg/L) 17.4±5.1 vs. 16.4±5.2 (p=.045) 93% (88% to 98%) tmax (h) 1.8±0.8 vs. 1.7±0.5 (p=.614) 96% (76% to 117%) Ethinyl estradiol AUC ([ng · h]/L) 2177±1543 vs. 1661±1324 (p=.162) 68% (14% to 123%) CLoral (L/h) 63.3±71.6 vs. 93.1±127.9 (p=.411) 85% (57% to 114%) T1/2 (h) 23.4±19.5 vs. 12.2±7.1 (p=0 .023) 62% (15% to 110%) Cmax (ng/L) 97.3±74.6 vs. 103.6±78.9 (p=0 .809) 111% (33% to 189%) tmax (h) 2.7±2.9 vs. 2.2±1.5 (p=.131) 63% (29% to 99%)			Sequential design, single COC, standardized SJW dose tested for accuracy, adequate sample size, appropriate PK parameters, outcome measures clearly defined		Clinical data presented but not powered to detect difference Self: reported compliance (daily diaries) with COC and SJW (100 and 90% respectively) Single progesterone measure to assess ovulation	Fair
Murphy, 2005. National Center for Complementary and Alternative Medicine and the Hypercium Buyers club	PK; Single blinded sequential trial		Healthy, normal weight, reproductive aged (18–35) women a) n=16		COC (EE 20 μg/NE 1 mg)+Placebo × 2, 28-day cycles. then COC (EE 20 μg/NE 1 mg)+SJW 300 TID × 2, 28-day cycles. (SJW 0.3% hypericin)		Cmax, tmax, AUC, volume of distribution (V/F), apparent clearance (Cl/F) for EE and NE Progesterone ≥3 (measured every 7–10 days) breakthrough bleeding favorable cervical mucus and follicle size		Placebo vs. SJW Follicle size >20 mm: 37.5% vs. 57% (p>.05) >30 mm: 6% vs. 44% (p<.05) Progesterone ≥3: 6% vs. 38% (p>.05) Breakthrough bleeding Any-31% vs. 56% (p<.05) Not related to missed pills 19% vs. 50% (p<.05) % favorable cervical mucus 12.5% vs. 31% (p>.05) median(25–75th percentile); p-value NE Cmax (ng/mL) 14.9 (12.2–18.4) vs. 15.2 (10.9–19.3); p=.632 tmax (h) 2.0 (1–3) vs. 2.0 (1.38–2.25); p=.497 AUC (ngd h/mL) 113.2 (91.3–150.6) vs. 98.9 (78.6–124.9); p=.021 (↓12.6%) Cl/F (L/h) 7.6 (6.6–9.8) vs. 9.97 (8.0–12.0); p=.001 ↑31.2% V/F (L) 87.5 (77.9–117.8) vs. 104.7 (86.2–129); p=0 .052 EE Cmax (pg/mL) 90 (82–98) vs. 84 (73–98); p=0 .268 tmax (h) 2.0 (1.5–3) vs. 2.0 (1.38–3.0); p=0 .470 AUC (pgd h/mL) 994 (874–1118) vs. 854 (712–1071); p=0.016↓14.0% Cl/F (L/h) 18.2 (15.4–20.9) vs. 23.0 (18.7–26.6); p=0 .003 ↑26.4% V/F (L) 352 (296–410) vs. 400 (319–466); p=0 .051			Sequential design, Single COC, Standardized SJW dose tested for accuracy, outcome measures clearly defined, multiple measures of progesterone to detect ovulation, adequate sample size, appropriate PK parameters, outcome measures clearly defined		Measure of compliance with both drugs not reported	Fair
Pfrunder, 2003. Senglett-Foundation for Young Pharmacists	Randomized, crossover		Healthy women age 18–35 with regular menses and no use of OCs for at least the last 3 months (n=17)			COC (0.02 mg EE/desogestrel 0.15 mg) for 1 28-day cycle then COC+SJW 300 mg twice daily (A) then COC+ SJW 300 mg three times daily (B) (SJW 0.3% hypericin)	ovarian activity (follicle maturation, Day 11 serum estradiol and Day 23 progesterone) breakthrough bleeding AUC and Cmax for EE and 3-ketodesogestrel			COC alone vs. A vs. B; 95% Cl Ovarian activity Mature follicle 1 vs. 3 vs. 3 (no p-value) Estradiol (pmol/l): 46 (35.2,57.4) vs. 237 (−17.1, 490.4) vs. 151 (5, 297.3) Mature follicle +elevated Estradiol: 0 vs. 2 vs. 3 Progesterone (nmol/l): 1.3 (1, 1.5) vs. 1.3 (1, 1.5) vs. 1.4 (0.9, 1.7); no p-value Breakthrough bleeding 35% vs. 77% vs. 88%; p<.001 For EE % difference in A vs. COC alone (95% CI; p-value) AUC: −5 (164.4, 6.6: NS) Cmax:0.3 (−11.1, 11.7; NS) % difference in B vs. COC alone (95% CI; p-value) AUC: 4.8 (−7.6, 17.2: NS) Cmax: 3.2 (−7.0, 13.5; NS) For 3-ketogestrel % difference in A vs. COC alone(95% CI; p-value) AUC: −43.9 (−49.3, −38.5: p=.001) Cmax:−17.8 (−29.9, −5.7; p=.005) % difference in B vs. COC alone(95% CI; p-value) AUC: −41.7 (−47.9, −35.6: p=.001) Cmax: −22.8 (−31.3, −13.3; p<.001)	Sequential/crossover design, Drug compliance in electronically monitored package, outcome measures clearly defined, adequate sample size, appropriate PK parameters			Reported compliance for COC but compliance with SJW not reported, single measure of progesterone to detect ovulation		Good
PK studies without clinical outcomes
Will-Shahab, 2008. Max Zeller Söhne AG	PK, sequential		Healthy, reproductive age women on low-dose COC for at least 3 months (n=16)			COC (ethinylestradiol 20 μg+desogestrel 150 μg) daily × 21 days plus SJW 250 mg BID on days 7–21 (SJW 0.2% hypericin)	Ratio Day-7-to: Day 21 AUC, Cmax and tmax for EE and 3-ketodesogestrel -LH, FSH, SHBG on Day 7 vs. Day 21			Day 7-to-21 Ratios (90% CI) EE AUC: 0.951 (0.915–0.986) Cmax: 0.968 (0.939–0.997) tmax-(h) 1.25 vs. 1.50 (NS) 3-ketodesogestrel AUC: 0.968 (0.944–0.992) Cmax: 0.983 (0.893–1.073) tmax-(h) 4.70 vs. 4.58 (NS) Day 7 vs. Day 21; p-value LH: 7±5.9 vs. 3.6±2.5; p=.0009 FSH: 2.6±1.5 vs. 2.1±1.9; p=.093 SHBG-289.7±90.8 vs. 319.8±101.8; p=.032	Sequential design, Serum hypericin level drawn to verify compliance, single COC used, standard SJW dose, adequate sample size, appropriate PK parameters, outcome measures clearly defined			No measure of COC compliance		Fair

FDA, US Food and Drug Administration; NIH, National Institutes of Health; OC, oral contraceptive; CL_oral, oral clearance.

In a single-blinded sequential trial, 16 healthy, reproductive-age women took a COC (EE 20 μg/NE 1 mg) daily for 21 days followed by the inactive pills for Days 22–28 for four consecutive cycles [18,21]. In addition, they took placebo three times daily in Cycles 1 and 2 and then SJW (0.3% hypericin) 300 mg three times daily in cycles 3 and 4. Serum progesterone concentrations were obtained every 7–10 days, and 12 serial blood samples were drawn (during the third week of cycles 2 and 4) over 24 h and used to calculate the median AUC, C_max, t_max, apparent clearance (CL/F) and apparent volume of distribution (V/F) for EE and NE [18]. Significantly more women reported breakthrough bleeding during the SJW+COC cycle (n=9; 56%) compared with the placebo+COC cycle (n=5; 31%; p<.05). By assessing daily diaries, authors found that breakthrough bleeding episodes unrelated to missed pills were greater in women when taking SJW+COC compared with placebo+COC (50% vs. 19%; p<.05). Significantly more women had follicles >30 mm in diameter when taking SJW+COC (n=7, 44%) compared with placebo+COC (n=1, 16%; p<.05). Although the number of women with follicles>20 mm in diameter (n=9 vs. n=6), favorable cervical mucus scores (≥10) (n=5 vs. n=3), serum progesterone values ≥3 ng/mL (n=3 vs. n=1), and who experienced probable ovulations (n=6, vs. n=1) were all greater in the SJW+COC cycle compared to the placebo+COC cycle, none of these comparisons were significant (p>.05) [18]. For EE, the median AUC (pg* h/mL median; 25th–75th percentile) was decreased in the SJW+COC cycle (854; 712–1071) compared with the placebo+COC cycle (994; 874–1118) (p=.016). In addition, the CL/F (L/h) was increased in the SJW+COC cycles (23.0; 18.7–26.6) compared with the placebo+COC cycle (18.2; 15.4–20.9) (p=.003). For NE, the median AUC (pg* h/mL) was decreased in the SJW+ COC cycle (98.9; 78.6–124.9) compared with the placebo+COC cycle (113.2; 91.3–150.6) (p=.021). The CL/F (L/h) was also increased in the SJW+COC cycle (9.97; 8.0–12.0) compared with the placebo+COC cycle (7.6; 6.6–9.8) (p=.001). However, no significant difference was seen for the other parameters (C_max, t_max, or V/F), in either EE or NE, for the SJW+COC cycle compared with the placebo+COC cycle. (Table 1).

In a crossover study designed to look at two different doses of SJW, healthy, reproductive-age women (n=17) were enrolled [19]. In Cycle 1, participants took a COC (20 μg EE/desogestrel 0.15 mg) daily for 21 days and the inactive pills on days 22–28. In cycle 2, subjects were randomly assigned to either treatment A, during which they took COC+SJW 300 mg (0.3% hypericin) twice daily, or treatment B, during which they took the COC+SJW 300 mg (0.3% hypericin) three times daily. In Cycle 3, the subjects crossed over. Compliance for both study drugs was monitored via electronically monitored packages. Serial blood samples determined the Day 10 AUC and C_max for EE and 3-ketodesogestrel. The number of women with intra-cyclic bleeding episodes was significantly higher in both treatment A (13/17; p=.037) and treatment B (15/17; p=.004) compared with the number of women in the COC alone cycle (6/17). No differences between COC-alone cycle and either treatment A or treatment B cycles were seen in the number of participants with follicle-like structures or Day 23 progesterone concentrations. Estradiol (E2) concentrations (nmol/L) were higher in both treatment A (237; −17.1 to 490.4) and in treatment B cycles (151; 5.0–297.3) compared with the COC-alone cycle (46; 35.2–57.4); however, this difference was only significant for treatment B compared with COC alone. The percentage difference in the PK parameters for 3-ketodesogestrel was significantly decreased from treatment A versus COC alone for C_max (ng/mL) (−17.8%; −29.9 to −5.7); p=.005), t_1/2 (−32.8;−55.8 to −10.0; p=.003), and AUC (ng* mL/h) (−43.9; −49.3 to −38.5; p=.001) as well as significantly decreased from treatment B versus COC alone for C_max (ng/mL) (−22.8%; −31.2 to −13.3; p<.001), t_1/2 (−30.0%; −50.9 to −9.0; p=.009) and AUC (ng* mL/h) (−41.7%; −47.9 to −35.6; p=.001). No significant differences were seen in t_max for 3-ketodesogestrel between the COC-alone cycle and treatment A or treatment B cycles. For EE, there was no significant difference for any of the PK parameters between COC alone and treatment A or treatment B cycles [19]. (Table 1).

In the one study that reported PK outcomes alone, healthy, reproductive-age women who had been taking a COC (EE 20 μg+desogestrel 150 μg) for at least the prior 3 months were enrolled (n=16) [20]. During the 21-day study period, participants continued to take their COC (EE 20 μg+desogestrel 150 μg) once daily and were given a standard SJW product (Ze 117; standardized to 0.2% hypericin) 250 g twice daily from Day 7 through Day 21. Day 7 values were used as the comparison group (women taking COC alone) compared with COC plus SJW (Day 21). Serial blood samples were used to determine Day 7 to Day 21 geometric mean ratios for AUC, C_max, and t_max for EE and 3-ketogestrel (with 90% CIs). To check for compliance of the SJW, trough values of hypericin and pseudohypericin were determined from plasma samples on Day 7 and Day 14. The day-7-to-day-21 ratios for AUC and C_max were not significantly different for either EE or 3-ketodesogestrol (Table 1). Similarly, for both EE and 3-ketodesogestrel, the Day 21 t_max values were not significantly different than the Day 7 values (Table 1). Although mean FSH values were not significantly different between Day 7 and Day 21, mean LH values were significantly lower after SJW administration on Day 21 (Day 7: 7.0±5.9 mlU/mL and Day 21: 3.6±2.5 mlU/mL; p<.05), and mean sex hormone-binding globulin (SHBG) values were significantly higher on Day 21 (Day 7: 289.7±90.8 mlU/mL and Day 21: 319.8±101.8 mlU/mL; pb.05) compared with Day 7. Although the authors did not describe how adverse events including breakthrough bleeding or spotting were measured or defined, no such events were observed throughout the study [20].

4. Conclusion

Overall, the markers of contraceptive efficacy measured in the review, including breakthrough bleeding, predictors of ovulation, and pharmacokinetics of contraceptive steroid hormones, suggest that SJW may weakly or moderately induce the metabolism of COCs, giving rise to contraceptive efficacy concerns. This review is limited by several factors, including the small number of articles published in this area and the absence of any published data for progestin-only or non-oral hormonal contraceptives. Additionally, the studies did not directly measure contraceptive efficacy or rates of unintended pregnancy. Thus, conclusions from this review rely on PK and more indirect clinical outcomes (e.g., breakthrough bleeding, follicular size/growth, and serum progesterone measurements). Given the concern for contraceptive efficacy, theoretically, the greatest risk for a clinically significant interaction may be with hormonal contraceptives that have the smallest margins for contraceptive efficacy, which include “very low dose” COCs or progestin-only pills. However, in this review, no study evaluated SJW administration with COCs that contain less than 20 μg of EE or progestin-only pills. Further, because the contraceptive efficacy of COCs comes primarily from the progestin component, this review is potentially limited because only two types of progestin were studied.

No adverse events were reported in the four studies; however, given the rarity of serious adverse events associated with use of COCs, the studies were not designed or powered to address our question related to the safety of COCs with SJW use (sample sizes range 12–18). Since all 4 studies suggested decreased exposure to hormone, there may be concerns about contraceptive efficacy; however, none of the studies reported increased exposure to hormones and thus did not suggest safety concerns that might be associated with hormones. No studies were identified that looked at the effect hormonal contraception might have on the safety or efficacy of SJW (research question #2). However, there is no known theoretical concern for an herb-drug interaction that would lead to hormonal contraceptives compromising the efficacy or safe use of SJW.

Clinical indicators of ovulation demonstrated mixed results. Overall, these findings raise concern for the risk of breakthrough ovulation, potentially increasing risk of unintended pregnancy, with addition of SJW in COC users. Interpretation of these findings is greatly limited by heterogeneity of methods, different COCs used and the very small sample sizes. Three of the four studies demonstrated an increased risk of breakthrough bleeding when SJW is administered with COCs. The fourth study identified in this review did not describe any methods for defining or measuring breakthrough bleeding but mentioned “no breakthrough bleeding or spotting” occurring in a single cycle in which a SJW product containing less of the proposed active compound (0.2% hypericin) was administered over a short period (14 days) [20]. Thus, this study was limited because it failed to define the outcome of breakthrough bleeding and likely was not administered long enough to get meaningful data for this outcome. However, as breakthrough bleeding is a common reason for method discontinuation or noncompliance, which puts women at risk for unintended pregnancy, anything that influences rates of breakthrough bleeding may be of clinical importance [22]. Larger clinical trials are needed to further explore the clinical implications of these findings.

The PK results of these studies overall demonstrate that the addition of SJW in users of COC is associated with fluctuations in PK parameters that suggest metabolic induction. However, although each study measured similar PK parameters, a different pattern of change was seen when SJW was co-administered with COCs. The PK methods for the studies identified in this review were well designed and of fair to good quality. However, most were limited by not addressing or inadequately addressing medication compliance with study drugs. The concentration of the active ingredients in SJW (hypericin and hyperforin) has been shown to greatly influence the degree of its interaction with other drugs [7,23]. In this review, the one study of SJW with low hypericin content was the only study that failed to demonstrate a difference between PK parameters for steroid hormone levels for women using SJW. However, this study took place over a single cycle and was not designed to evaluate clinical outcomes. The generalizability of these studies in the real-world clinical scenario may be limited because of the wide range of SJW products commercially available to consumers that likely contain a variable or unknown hyperforin or hypericin content [24].

In conclusion, the clinical data in this review are mixed but raise concern for decreased contraceptive efficacy based on evidence showing increased risk of breakthrough bleeding and ovulation in women using COCs when co-administered with SJW. The PK evidence in this review supports these clinical findings, as they show a pattern of small to moderate induction of the metabolism of contraceptive steroids (both EE and progestins) when co-administered with SJW. However, given the overall paucity of data including more specifically the lack of clinical pregnancy data, the interaction SJW administration may have on the risk of contraceptive failure, resulting in unintended pregnancy, for users of COCs or other methods of HC is unknown. Given the substantial theoretical concern posed by the ability of SJW to induce CYP 450 3 A4, combined with the concerning pattern the data in this review demonstrate, more research examining clinical outcomes for hormonal contraception users taking SJW is needed.

Appendix A. Search Strategy

Search Strategy for Research question #1 and #2.

PubMed: “hypericum”[MeSH Terms] OR “hypericum”[All Fields] OR (“st”[All Fields] AND “john”[All Fields] AND “wort”[All Fields]) OR “st john wort”[All Fields] AND (“Contraceptives, Oral, Combined”[Mesh] OR “Contraceptives, Oral”[Mesh] OR “Contraceptives, Oral, hormonal”[Mesh] OR “Contraceptives, Oral, Combined”[Pharmacological Action]) OR (contracept* AND (oral OR pill OR tablet)) OR ((combined hormonal) OR (combined oral) AND contracept*) OR (contracept* AND (ring OR patch)) OR “ortho evra” OR NuvaRing OR (progestin* OR progestins[MeSH] OR Progesterone[MeSH] OR progesterone OR progestogen* OR progestagen* OR “Levonorgestrel”[Mesh] OR Levonorgestrel OR “Norgestrel”[Mesh] OR norgestrel OR etonogestrel AND contracept*) OR dmpa OR “depot medroxyprogesterone” OR “depo provera” OR “net en” OR “norethisterone enanthate” OR “norethindrone enanthate” OR (contracept* AND (inject* OR implant)) OR ((levonorgestrel OR etonogestrel) AND implant) OR implanon OR nexplanon OR jadelle OR norplant OR uniplant OR sino-implant OR (levonorgestrel-releasing two-rod implant) OR “Intrauterine Devices”[Mesh] OR “Intrauterine Devices, Copper”[Mesh] OR “Intrauterine Devices, Medicated”[Mesh] OR ((intrauterine OR intra-uterine) AND (device OR system OR contracept*)) OR IUD OR IUCD OR IUS OR mirena OR Skyla OR paragard OR “Copper T380” OR CuT380 OR “Copper T380a” OR “Cu T380a”).

Cochrane Library: “Contraception AND St John’s Wort” and “Contraception AND Hypericium”

Appendix B. Quality rating System for Pharmacokinetic Studies

Three Overall Quality Categories:

Good: No important limitations. Well done study that meets all criteria for an adequate pharmacokinetic (PK) study (below). Reviewer feels confident the results are internally valid.

Fair: Clear limitations to study design but no fatal flaws.

Poor: One or more fatal flaws that likely invalidates results.

Criteria	Good (meet all criteria)	Fair	Poor (has one or more)
Design	Crossover design (or parallel design with appropriate justification)	Parallel design
Sample Size	Cross-over n≥12; if parallel design, n should be higher	n is 8–12	n<8
Exposure	Clear definition of exposure (clearly defined drug(s), dosages, and frequency). Clearly stated exposure assessment accounting for ensured exposure to drug (in d-d-I studies, exposure to both drugs clearly defined).	Clear definition of exposure. Adequate but less than ideal exposure assessment (self-report alone).	Exposure not defined. No exposure assessment.
Outcome	Appropriate PK parameter measured for desired outcome (e.g., for hormonal contraception Cmax, AUC or Cavg; for non-oral formulations Caverage or AUC) and the measured out come has clinically meaningful relevance (known or theoretical).	PK parameter less than ideal but still gives some potentially useful information.	Clinically irrelevant PK parameter.
Timing	Time of the blood draw(s)/testing appropriate for the desired outcome. Repeated measures taken (unless steady state demonstrated to be achieved, then one-time measurement OK).	Time of blood draw not ideal but still yields useful information. One-time measurements.	Time of blood draw out of range to yield meaningful information in relation to desired outcome.
Intersubjective variability	Methods minimize possibility for intersubjective variability. (e.g., range of timing for blood draws). There is adequate control in studies for factors known to impact metabolism (age, BMI, other medications, or other known risk factors) as appropriate/needed.	Moderate intersubjective variability. Some controlling for factors known to impact metabolism or no theoretical factors known to impact so no controlling done.	Very large intersubjective variability. No control and clear presence of factors that very likely impacted metabolism between subjects.
Population	Appropriate population chosen (e.g., reproductive-aged women).	Less than ideal population but not fatally so.	Completely wrong population chosen that has proven or likely will have different metabolism/effect of the drugs.
Steady state of perpetrator drug (Victim drug OK for onetime dose)	Clearly allowed for perpetrator drug to be in steady state at time of evaluation.	Likely that perpetrator drug was in steady state; however, methods not clearly defined or uncertain of SS actually reached.	Perpetrator drug clearly NOT in steady state.
Assay/analyses and validation	Study described methods for analysis and validation of analyses.	Study did not describe methods for analysis and validation of analyses.	Methods described for analysis or validation described but methods used known to be erroneous.