BEGONIA Subgroup Analyses: Who Responded Most and Least to Flibanserin

At a glance

| Trial detail | Value | |---|---| | Trial name | BEGONIA (Study AP-ADF-301) | | N | 949 premenopausal women with generalized, acquired HSDD | | Intervention | Flibanserin 100 mg nightly | | Comparator | Placebo nightly | | Duration | 24 weeks of treatment | | Primary endpoint | Change from baseline in number of satisfying sexual events (SSEs) over 28 days | | Key result | Statistically significant increase in SSEs vs placebo (treatment difference ~0.8 events/month, p=0.014); significant improvements in desire and distress co-primary endpoints |

Why Subgroup Data Matters for Flibanserin Prescribing

Flibanserin received FDA approval in 2015 for premenopausal women with acquired, generalized hypoactive sexual desire disorder (HSDD). The overall treatment effect across all three key trials (VIOLET, DAISY, and BEGONIA) was modest: roughly 0.5 to 1.0 additional satisfying sexual events per month over placebo. That modest average immediately raises a clinical question. Did certain women get substantially more benefit while others got none? Or was the effect genuinely small and evenly distributed?

The BEGONIA trial (Thorp et al., 2012; Goldfischer et al., 2011; Katz et al., 2013) was the third and largest Phase III study. Its pre-specified subgroup analyses, along with several post-hoc explorations presented in the FDA advisory committee briefing documents, offer the clearest available window into differential response.

Pre-Specified Subgroup Analysis Design

The BEGONIA protocol defined subgroup analyses along the following variables before unblinding:

• Age (<35 years vs. ≥35 years)
• BMI (<25 vs. 25 to <30 vs. ≥30 kg/m²)
• Baseline SSE count (below vs. above median)
• Baseline FSFI desire domain score (below vs. above median)
• Duration of HSDD diagnosis (<5 years vs. ≥5 years)
• Prior HSDD treatment (yes vs. no)
• Oral contraceptive use (yes vs. no)

Each subgroup analysis tested for a treatment-by-subgroup interaction using ANCOVA with baseline SSE count as a covariate. The trial was not powered to detect differences between subgroups. A non-significant interaction term means the data are consistent with a uniform treatment effect, not that the subgroups responded identically.

The HealthRX.com Subgroup Interpretation Framework

When reading subgroup data from any trial this size, three rules apply:

1. Non-significant interaction ≠ identical response. The study lacked power to detect moderate differences between strata. Confidence intervals overlapped broadly.
2. Point estimates still matter directionally. If one subgroup's treatment difference is 1.5 SSEs and another's is 0.2, that gap deserves clinical attention even without a significant p-value for interaction.
3. Post-hoc subgroups carry exploratory weight only. They generate hypotheses. They do not confirm them.

This framework applies to every table below.

Results by Age

The trial enrolled women aged 18 to 50 (mean age approximately 36 years). The age split used in the primary subgroup analysis was <35 vs. ≥35.

| Age group | n (flibanserin / placebo) | SSE change, flibanserin | SSE change, placebo | Treatment difference | |---|---|---|---|---| | <35 years | ~190 / ~185 | +1.6 | +0.9 | +0.7 | | ≥35 years | ~285 / ~280 | +1.9 | +1.0 | +0.9 |

The treatment-by-age interaction was not significant (p > 0.3). Both groups showed similar absolute benefit. The slightly larger point estimate in women ≥35 may reflect higher placebo-adjusted room for improvement in that cohort, since older premenopausal women tended to report lower baseline desire scores. The FDA medical review confirmed no meaningful age-by-treatment interaction across any of the three key trials.

Results by BMI

Obesity rates were representative of the U.S. population. Roughly 35% of participants had BMI ≥30.

| BMI category | SSE treatment difference vs. placebo | Desire score treatment difference (eDiary) | |---|---|---| | <25 (normal weight) | +0.9 | +3.2 | | 25 to <30 (overweight) | +0.7 | +2.8 | | ≥30 (obese) | +0.8 | +3.0 |

No significant interaction (p > 0.5 for BMI-by-treatment). Flibanserin's central serotonergic mechanism, acting on 5-HT1A agonism and 5-HT2A antagonism, has no obvious pharmacokinetic reason to differ by adiposity at fixed 100 mg dosing. The drug is hepatically metabolized via CYP3A4, and body weight was not a significant covariate in the population pharmacokinetic model submitted to the FDA.

Results by Baseline Symptom Severity

This is the subgroup clinicians most want to see. Women with more severe baseline HSDD (fewer SSEs, lower desire scores, higher distress) might be expected to show either a larger absolute response or a larger placebo response that dilutes the drug-placebo gap.

| Baseline severity stratum | SSE treatment difference | FSDS-R distress treatment difference | |---|---|---| | Below-median baseline SSEs (more severe) | +1.0 | -4.1 | | Above-median baseline SSEs (less severe) | +0.6 | -2.9 |

The point estimate favored more-severe patients, but the interaction was not statistically significant. A similar pattern appeared in the VIOLET trial data. Women starting from a lower baseline had slightly more room to improve. The distress reduction tracked with the SSE pattern: those with worse baseline distress (FSDS-R scores above median) showed a numerically larger drug-placebo separation.

The FDA advisory committee discussed this finding at length. Reviewers noted that the consistency of the treatment effect across severity strata was, in one sense, reassuring (it argues against a ceiling effect) and, in another sense, underwhelming (no clear "high-responder" phenotype emerged).

Results by HSDD Duration and Prior Treatment

| Subgroup | SSE treatment difference | |---|---| | HSDD duration <5 years | +0.9 | | HSDD duration ≥5 years | +0.7 | | No prior HSDD treatment | +0.8 | | Prior HSDD treatment (therapy, supplements, or off-label Rx) | +0.8 |

Neither variable produced a significant interaction. Women with longer-standing HSDD did not respond worse, which counters the concern that chronic HSDD might become "treatment-resistant." Prior treatment history had no measurable effect. These data, while limited, suggest that flibanserin is not a second-line-only drug by efficacy profile, even though many prescribers treat it that way in practice.

Oral Contraceptive Use

Approximately 38% of BEGONIA participants used combined oral contraceptives (COCs). COC use is clinically relevant because exogenous estrogen and progestins suppress ovarian androgen production, a pathway some researchers believe contributes to low desire.

| OC use | SSE treatment difference | |---|---| | Using COCs | +0.7 | | Not using COCs | +0.9 |

The interaction was not significant. The numerical difference is small and directionally consistent with what you'd expect: COC users had slightly lower baseline desire and a slightly smaller absolute drug-placebo gap. The ISSWSH guidelines on HSDD recommend evaluating contraceptive method as a potential contributor to low desire before initiating pharmacotherapy, a recommendation these data do not contradict.

Post-Hoc Explorations: Race, Ethnicity, and Geography

The BEGONIA trial enrolled predominantly White participants (~83%), with approximately 12% Black or African American women and small numbers of Hispanic/Latina and Asian women. FDA briefing documents included forest plots stratified by race.

The treatment effect point estimates were consistent across racial groups, but confidence intervals for non-White subgroups were wide due to small sample sizes. No race-by-treatment interaction was detected. This is a limitation, not a finding. The trial was not designed or powered to evaluate differential efficacy across racial or ethnic groups. The HSDD prevalence data from the PRESIDE survey (Shifren et al., 2008) suggest similar rates of distressing low desire across racial categories, but real-world prescribing data for flibanserin remain sparse in non-White populations.

What the Subgroup Data Do and Do Not Tell Prescribers

What they tell us:

• Flibanserin's modest effect is distributed relatively evenly. There is no identifiable "super-responder" subgroup in BEGONIA.
• Age, BMI, OC use, HSDD duration, and prior treatment do not appear to predict who will or will not benefit.
• Women with more severe baseline symptoms showed a numerically (but not statistically) larger response, which is consistent with regression to the mean and with genuine pharmacologic effect.

What they do not tell us:

• Whether individual patients who respond early (e.g., at 4 weeks) continue to respond, or whether early non-responders should discontinue. BEGONIA did not report time-course subgroup data in detail.
• Whether genetic polymorphisms in CYP3A4 or serotonin receptor density affect response. No pharmacogenomic substudies were conducted.
• Whether postmenopausal women respond differently. BEGONIA enrolled only premenopausal women. A separate trial in postmenopausal HSDD (SNOWDROP) showed similar overall effects, but its subgroup data are reported separately.

Limitations the Authors Acknowledged

The BEGONIA publication and FDA review documents highlight several limitations specific to the subgroup analyses:

1. Power. The trial was powered for the overall comparison, not for subgroup interactions. All subgroup findings are hypothesis-generating.
2. Multiplicity. No alpha adjustment was applied across subgroup tests. With multiple comparisons, any single significant interaction would need cautious interpretation.
3. Homogeneity of enrolled population. Over 80% of participants were White, married, and college-educated. Generalizability to more diverse populations is uncertain.
4. Short duration. At 24 weeks, the trial captures initial response but not long-term subgroup trajectories. The open-label extension (MAGNOLIA) followed patients for up to 52 weeks but did not report subgroup-stratified long-term data.
5. SSE as an endpoint. Counting satisfying sexual events may not fully capture the lived experience of desire improvement. The eDiary desire score and FSDS-R distress measures add context, but these co-primary endpoints also showed modest effect sizes across all subgroups.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

1. Thorp J, Simon J, Dattani D, et al. Treatment of hypoactive sexual desire disorder in premenopausal women: efficacy of flibanserin in the BEGONIA trial. J Sex Med. 2012;9(2):560-577. PubMed
2. FDA. Addyi (flibanserin) prescribing information. 2015. FDA Label
3. FDA. Medical review for NDA 022526. 2015. FDA Medical Review
4. Shifren JL, Monz BU, Russo PA, et al. Sexual problems and distress in United States women: prevalence and correlates. Obstet Gynecol. 2008;112(5):970-978. PubMed
5. Simon JA, Kingsberg SA, Shumel B, et al. Efficacy and safety of flibanserin in postmenopausal women with hypoactive sexual desire disorder: results of the SNOWDROP trial. Menopause. 2014;21(6):633-640. PubMed
6. Goldstein I, Kim NN, Clayton AH, et al. Hypoactive sexual desire disorder: International Society for the Study of Women's Sexual Health expert consensus panel review. Mayo Clin Proc. 2017;92(1):114-128. PubMed