Impact of gonadotropin selection on risk of ovarian hyperstimulation syndrome in predicted high responders: a Menopur in Gonadotropin-releasing hormone Antagonist Single Embryo Transfer-High Responder trial analysis.

All subjects who received any dose of either gonadotropin (n = 619) were included in the safety analysis cohort ( 26 ). Overall, 96 (15.5%) subjects developed any OHSS (inclusive of early and late OHSS). Significantly fewer developed OHSS in the HP-HMG group (30 of 310 [9.7%]) compared with the rFSH group (66 of 309 [21.4%]; difference: −11.7%; 95% CI: −17.3%, −6.1%). The incidence of early OHSS was significantly lower in HP-hMG–treated subjects, 19 of 310 (6.1%), compared with 54 of 39 (17.5%) who were treated with rFSH (difference: −11.3%; 95% CI: −16.4%, −6.3%). Incidence of late OHSS was similar between the treatment groups: 11 of 310 (3.5%) in the HP-hMG group vs. 12 of 309 (3.9%) in the rFSH group (difference: −0.3; 95% CI: −3.3%, 2.6%). Cycle cancellation for excessive response occurred in 3 (1.0%) subjects in the HP-hMG group vs. 6 (1.9%) subjects in the rFSH group, and leuprolide acetate was used for OHSS prophylaxis in 4 (1.3%) subjects in the HP-hMG group vs. 12 (3.9%) subjects in the rFSH group ( 26 ). No subjects received dopamine agonist as a preventive intervention for early OHSS in the study. Baseline characteristics did not significantly differ between treatment groups for the whole trial population ( 26 ). In subjects who developed any OHSS compared with those who did not, mean subject age was significantly younger (29.6 vs. 30.4; P =.0158) and duration of infertility was significantly shorter (29.8 months vs. 38.2 months; P =.0017). Fewer subjects with a diagnosis of endometriosis developed any OHSS ( P =.0085; Supplemental Table 1 , available online). In subjects who developed early OHSS compared with those who did not, mean subject age was significantly younger (29.5 vs. 30.3; P =.0232), mean weight (63.8 kg vs. 66.2 kg; P =.0406) and BMI (23.6 vs. 24.5; P =.0319) were significantly lower, and duration of infertility was significantly shorter (31.0 vs. 37.7 months; P =.0217). Baseline serum estradiol levels were also significantly higher in subjects who developed early OHSS vs. those who did not (162.68 pmol/L vs. 137.60 pmol/L; P =.0299; Table 1 ). Baseline AMH and antral follicle count (AFC) did not differ between subjects who developed early OHSS and those who did not. Table 1 Demographics and clinical characteristics of subjects by incidence of early OHSS. Parameter No Early OHSS Early OHSS P value a N n (%) N n (%) Baseline demographics  Age 546 30.3 (3.03) 73 29.5 (3.15) .0232  Weight (kg) 546 66.2 (10.33) 73 63.8 (11.87) .0406  BMI (kg/m 2 ) 546 24.5 (3.31) 73 23.6 (3.45) .0319  Duration of infertility (mo) 545 37.7 (27.58) 73 31.0 (22.46) .0217 Cause of infertility b  Oligoovulation 90 (16.5%) 16 (21.9%) .2487  Endometriosis 44 (8.1%) 1 (1.4%) .0501  Male factor 226 (41.4%) 39 (53.4%) .0588  Tubal factor 81 (14.8%) 6 (8.2%) .1520  Idiopathic 197 (36.1%) 20 (27.4%) .1531  Other 46 (8.4%) 11 (15.1%) .0819  AFC 544 30.4 (12.93) 71 33.8 (19.97) .1340  Baseline AMH (ng/mL) 546 7.6 (2.95) 73 8.4 (3.83) .2161  ≥5.0 ng/mL 546 (100.0%) 73 (100.0%)  ≥6.0 ng/mL 380 (69.6%) 53 (72.6%)  ≥7.0 ng/mL 248 (45.4%) 36 (49.3%)  ≥8.0 ng/mL 164 (30.0%) 24 (32.9%)  Luteinizing hormone (U/L) 536 6.5 (3.50) 70 7.4 (5.43) .1886  Follicle-stimulating hormone (U/L) 536 6.3 (1.53) 70 6.2 (1.67) .7036  Estradiol (pmol/L) 455 137.60 (75.41) 64 162.68 (86.36) .0299  Estradiol (pg/mL) 455 37.5 (20.5) 64 44.3 (23.5) .0299  Progesterone (nmol/L) 534 0.7 (2.26) 70 0.5 (1.07) .6007  Progesterone (ng/mL) 534 0.2 (0.7) 70 0.2 (0.3) .6007  Testosterone (nmol/L) 525 1.0 (0.44) 70 1.1 (0.65) .1063  Testosterone (ng/dL) 525 29.1 (12.7) 70 32.9 (18.7) .1063 End of stimulation:  Estradiol (pmol/L) 479 10,672.95 (7,002.79) 61 13,934.53 (6,288.09) <.0001  Estradiol (pg/mL) 479 2,907.4 (1,907.6) 61 3,795.8 (1,712.9) <.0001  Number of follicles observed/subject 504 28.8 (14.65) 70 34.6 (15.78) .0018  No. of follicles ≥12 mm observed/subject c 504 16.4 (7.58) 70 21.2 (7.48) <.0001  No. of oocytes retrieved/subject 526 17.5 (10.53) 72 27.9 (13.44) <.0001 Note: Unless otherwise noted, data are presented as mean ± SD or n (%). n = number of subjects with observation; AFC = antral follicle count; AMH = anti-Müllerian hormone; BMI = body mass index. a P values were calculated using the Kruskal-Wallis test for continuous variables and the Fishers exact test for categorical variables. b Subjects may have had more than one cause of infertility. c On day of trigger. Demographics and clinical characteristics of subjects by incidence of early OHSS. Note: Unless otherwise noted, data are presented as mean ± SD or n (%). n = number of subjects with observation; AFC = antral follicle count; AMH = anti-Müllerian hormone; BMI = body mass index. P values were calculated using the Kruskal-Wallis test for continuous variables and the Fishers exact test for categorical variables. Subjects may have had more than one cause of infertility. On day of trigger. Of the subjects who developed any OHSS, those treated with HP-hMG had higher baseline testosterone levels (1.3 nmol/L) than those treated with rFSH (1.0 nmol/L; P =.0219). All other baseline demographics of subjects with any OHSS were not significantly different between treatment groups ( Supplemental Table 2 ). In subjects with early OHSS, those treated with HP-hMG had a higher mean weight (67.9 kg) than those treated with rFSH (62.3 kg; P =.0368). All other baseline demographics in subjects with early OHSS did not differ significantly between treatment groups ( Table 2 ). Table 2 Demographics and clinical characteristics of subjects with early OHSS by treatment group. Parameter rFSH HP-hMG P value a N n (%) N n (%) Baseline demographics  Age 54 29.6 (3.12) 19 29.2 (3.29) .6584  Weight (kg) 54 62.3 (12.23) 19 67.9 (9.97) .0368  BMI (kg/m 2 ) 54 23.1 (3.43) 19 24.8 (3.27) .0673  Duration of infertility (mo) 54 30.3 (20.43) 19 32.9 (28.01) .8402 Cause of infertility b  Oligoovulation 10 (18.5%) 6 (31.6%) .3330  Endometriosis 1 (1.9%) 0 (0.0%) 1.0000  Male factor 30 (55.6%) 9 (47.4%) .5994  Tubal factor 5 (9.3%) 1 (5.3%) 1.0000  Idiopathic 14 (25.9%) 6 (31.6%) .7659  Other 9 (16.7%) 2 (10.5%) .7167  AFC 53 32.5 (13.57) 18 37.7 (32.50) .8739  Baseline AMH (ng/mL) 54 7.8 (3.22) 19 9.8 (4.99) .1205  ≥5.0 ng/mL 54 (100.0%) 19 (100.0%)  ≥6.0 ng/mL 38 (70.4%) 15 (78.9%)  ≥7.0 ng/mL 24 (44.4%) 12 (63. 2%)  ≥8.0 ng/mL 14 (25.9%) 10 (52.6%)  Luteinizing hormone (U/L) 52 6.7 (3.54) 18 9.2 (8.79) .4721  Follicle-stimulating hormone (U/L) 52 6.0 (1.58) 18 6.8 (1.82) .0716  Estradiol (pmol/L) 47 156.84 (78.18) 17 178.80 (106.86) .6373  Estradiol (pg/mL) 47 42.7 (21.3) 17 48.7 (29.1) .6373  Progesterone (nmol/L) 52 0.5 (1.24) 18 0.3 (0.27) .3404  Progesterone (ng/mL) 52 0.2 (0.4) 18 0.1 (0.1) .3404  Testosterone (nmol/L) 52 1.1 (0.58) 18 1.4 (0.78) .0552  Testosterone (ng/dL) 52 30.3 (16.8) 18 40.2 (22.5) .0552 End of stimulation  Estradiol (pmol/L) 45 14,362.97 (6,025.14) 16 12,729.54 (7,040.89) .3172  Estradiol (pg/mL) 45 3,912.6 (1,641.3) 16 3,467.6 (1918.0) .3172  No. of follicles observed/subject 52 34.9 (12.98) 18 33.7 (22.44) .2088  No. of follicles ≥12 mm observed/subject c 52 23.3 (6.83) 18 15.4 (6.27) .0002  No. of oocytes retrieved/subject 54 30.1 (11.62) 18 21.1 (16.41) .0010 Note: Unless otherwise noted, data are presented as mean ± SD or n (%). n = Number of subjects with observation; HP-hMG = highly purified human menotropin; rFSH = recombinant follicle-stimulating hormone; AFC = antral follicle count; AMH = antimüllerian hormone; BMI = body mass index. a P values were calculated using the Kruskal-Wallis test for continuous variables and the Fishers exact test for categorical variables. b Subjects may have had more than one cause of infertility. c On day of trigger. Demographics and clinical characteristics of subjects with early OHSS by treatment group. Note: Unless otherwise noted, data are presented as mean ± SD or n (%). n = Number of subjects with observation; HP-hMG = highly purified human menotropin; rFSH = recombinant follicle-stimulating hormone; AFC = antral follicle count; AMH = antimüllerian hormone; BMI = body mass index. P values were calculated using the Kruskal-Wallis test for continuous variables and the Fishers exact test for categorical variables. Subjects may have had more than one cause of infertility. On day of trigger. At the end of stimulation, subjects who developed any OHSS had significantly greater serum estradiol levels (13,021.98 pmol/L vs. 10,691.88 pmol/L; P =.0004), total follicles (33.4 vs. 28.7; P =.0069), follicles ≥12 mm (20.2 vs. 16.4; P <.0001), and number of oocytes retrieved (25.0 vs. 17.6; P <.0001) compared with subjects who did not develop OHSS ( Supplemental Table 1 ). Similarly, subjects with early OHSS had significantly greater serum estradiol levels (13,934.53 pmol/L vs. 10,672.95 pmol/L; P <.0001), total follicles (34.6 vs. 28.8; P =.0018), follicles ≥12 mm (21.2 vs. 16.4; P <.0001), and oocytes retrieved (27.9 vs. 17.5; P <.0001) compared with subjects who did not develop early OHSS ( Table 1 ). In subjects who developed any OHSS, those treated with rFSH had a greater number of follicles ≥12 mm (21.9 vs. 16.1; P =.0008) and oocytes retrieved (27.7 vs. 19.0; P =.0001) compared with HP-hMG–treated subjects. Estradiol levels at the end of stimulation and total number of follicles were similar between treatment groups in subjects who developed any OHSS ( P =.5195 and P =.2414, respectively; Supplemental Table 2 ). In subjects with early OHSS, rFSH-treated subjects had a greater number of follicles ≥12 mm (23.3 vs. 15.4; P =.0002) and oocytes retrieved (30.1 vs. 21.1; P =.0010) compared with HP-hMG–treated subjects. Estradiol levels at the end of stimulation and total number of follicles were similar between treatment groups in subjects who developed early OHSS ( P =.3172 and P =.2088, respectively; Table 2 ). Gonadotropin-releasing hormone agonist trigger was reserved for subjects with potential excessive ovarian response per the trial protocol. A total of 91 subjects received GnRH agonist trigger, with more subjects in the rFSH group, although the difference was not statistically significant (rFSH: 54/309, 17.5% vs. HP-hMG: 37/310, 11.9%; difference: 5.5%; 95% CI: −0.2%, 11.2%; P =.0544). Subjects who received GnRH agonist trigger had higher numbers of follicles ≥15 mm (14.2 ± 7.4 vs. 9.6 ± 4.8), oocytes retrieved (30.9 ± 13.0 vs. 16.5 ± 9.7), and serum estradiol (21,016.59 pmol/L vs. 9356.57 pmol/L) compared with subjects who received hCG trigger (n = 509). Incidence of early OHSS was 16.5% (15/91) in subjects who received a GnRH agonist trigger vs. 11.2% (57/509) in subjects who received an hCG trigger (difference: 5.3%; 95% CI: –2.2%, 14.7%; P =.1618). Late OHSS occurred in 1.1% (1/91) of subjects who received a GnRH agonist trigger vs. 4.3% (22/509) of subjects who received an hCG trigger (difference: −3.2%; 95% CI: −5.7%, 1.8%; P =.23). There was no significant difference in rates of any OHSS by trigger type ( P =.6403; Table 3 ). Mild OHSS occurred in 9.9% (9/91) of subjects who received GnRH agonist trigger vs. 2.9% (15/509) of subjects who received hCG trigger (difference: 6.9%; 95% CI: 2.0%, 15.1%; P =.0053). There was no significant difference in incidence of moderate OHSS ( P =.5497) or severe OHSS ( P =.4913) by trigger type ( Table 3 ). Of the 23 subjects who developed late OHSS, 1 received GnRH agonist trigger (in the rFSH treatment group) and 22 received hCG trigger (11 in the rFSH and 11 in the HP-hMG treatment group). Table 3 Incidence of OHSS by trigger type. OHSS GnRH agonist (N = 91) hCG (N = 509) GnRH agonist – hCG P value n (%) n (%) (95% CI) Early 15 (16.5) 57 (11.2) 5.3 (–2.2, 14.7) .1618 Late 1 (1.1) 22 (4.3) –3.2 (–5.7, 1.8) .2311 Any 16 (17.6) 79 (15.5) 2.1 (–5.3, 11.6) .6403  Mild 9 (9.9) 15 (2.9) 6.9 (2.0, 15.1) .0053  Moderate 6 (6.6) 48 (9.4) –2.8 (–7.5, 4.5) .5497  Severe 1 (1.1) 16 (3.1) –2.0 (–4.4, 3.0) .4913 Note: Unless otherwise noted, data are presented as n (%). CI = confidence interval; N = number of subjects in given treatment group; n = number of subjects with observation. Confidence intervals were generated using the Agresti-Min exact confidence interval approach. OHSS = ovarian hyperstimulation syndrome (early OHSS was defined as having an onset ≤9 days after triggering final follicular maturation, before trigger, or during stimulation when trigger was not used; late OHSS was defined as onset >9 days after trigger); GnRH = gonadotropin-releasing hormone; hCG = Human chorionic gonadotropin. Incidence of OHSS by trigger type. Note: Unless otherwise noted, data are presented as n (%). CI = confidence interval; N = number of subjects in given treatment group; n = number of subjects with observation. Confidence intervals were generated using the Agresti-Min exact confidence interval approach. OHSS = ovarian hyperstimulation syndrome (early OHSS was defined as having an onset ≤9 days after triggering final follicular maturation, before trigger, or during stimulation when trigger was not used; late OHSS was defined as onset >9 days after trigger); GnRH = gonadotropin-releasing hormone; hCG = Human chorionic gonadotropin. A statistically significant interaction ( P =.0323) was observed between the baseline AMH level and treatment group on the number of oocytes retrieved ( Supplemental Fig. 1A , available online). Subjects with higher baseline AMH had a higher ovarian response to rFSH vs. HP-hMG. For approximately every 1.54 ng/mL (11.0 pmol/L) increase in baseline AMH, rFSH stimulation resulted in one extra oocyte vs. HP-hMG. Conversely, there was no statistically significant interaction between AFC and treatment group on the number of oocytes retrieved ( Supplemental Fig. 1B ). In HP-hMG–treated subjects, early OHSS varied from 2.9% (1/34) in subjects with 1–5 oocytes retrieved, 4.1% (3/73) in subjects with 6–10 oocytes, 5.3% (4/76) in subjects with 11–15 oocytes, 9.7% (7/72) in those with 16–25 oocytes retrieved to 8.3% (3/36) in those with >25 oocytes ( Fig. 1 A). The corresponding early OHSS incidences by oocyte yield category in subjects treated with rFSH were 0% (0/6), 0% (0/31), 7.1% (4/56), 15.5% (17/110), and 32.4% (33/102), respectively ( Fig. 1 A). In subjects with >25 oocytes retrieved, the incidence of early OHSS was significantly higher in rFSH-treated subjects (32.4%) compared HP-hMG–treated subjects (8.3%; difference: 24%, 95% CI: 8.3, 36.2; Fig. 1 B). The incidence of early OHSS was directly proportional to oocyte yield for both treatment groups. The likelihood of early OHSS increased by 1.06 times (95% CI: 1.04, 1.08) for each additional oocyte retrieved. After adjusting for the number of oocytes retrieved, the odds of early OHSS in HP-hMG–treated subjects were 0.46 times (95% CI: 0.26, 0.82) that of rFSH-treated subjects ( Fig. 1 C). Figure 1 Relationship between oocyte yield and incidence of early OHSS, by treatment group. (A) Early OHSS by category of oocytes retrieved and treatment group. Subject numbers in each category and treatment group are shown inside the bars. (B) Forest plot of early OHSS by category of oocytes retrieved and treatment group. (C) Logistic regression analysis was performed to account for imbalances in subjects within oocyte number categories. Modeled probability of early OHSS by treatment group, based on continuous representation of oocytes retrieved. (D) Odds of early OHSS by treatment group, using an alternative information criterion model (Akaike information criterion, AIC). Note that no statistically significant interaction was found between oocyte yield and treatment group. HP-hMG = highly purified human menopausal gonadotropin; rFSH = recombinant follicle-stimulating hormone; OHSS = ovarian hyperstimulation syndrome (early OHSS was defined as onset ≤9 days after triggering final follicular maturation, before trigger, or during stimulation when trigger was not used). Relationship between oocyte yield and incidence of early OHSS, by treatment group. (A) Early OHSS by category of oocytes retrieved and treatment group. Subject numbers in each category and treatment group are shown inside the bars. (B) Forest plot of early OHSS by category of oocytes retrieved and treatment group. (C) Logistic regression analysis was performed to account for imbalances in subjects within oocyte number categories. Modeled probability of early OHSS by treatment group, based on continuous representation of oocytes retrieved. (D) Odds of early OHSS by treatment group, using an alternative information criterion model (Akaike information criterion, AIC). Note that no statistically significant interaction was found between oocyte yield and treatment group. HP-hMG = highly purified human menopausal gonadotropin; rFSH = recombinant follicle-stimulating hormone; OHSS = ovarian hyperstimulation syndrome (early OHSS was defined as onset ≤9 days after triggering final follicular maturation, before trigger, or during stimulation when trigger was not used).

MEGASET-HR was an open-label, assessor-blind, parallel-group trial conducted between August 2015 and February 2018 in accordance with the Declaration of Helsinki, the International Conference on Harmonization Guidelines for Good Clinical Practice ( 26 ). A total of 31 fertility clinics in the United States participated in the trial. Institutional review board approval was provided at each participating study site. All individuals who performed ultrasound monitoring, embryologists, and central laboratory personnel were blinded to treatment allocation throughout the study. The main inclusion criteria were: women 21−35 years of age with menstrual cycles of 21−45 days; body mass index (BMI) 18−30 kg/m 2 ; infertility for ≥1 year; day 2 or 3 serum FSH levels 1−12 IU/L; total testosterone, prolactin, and thyroid stimulating hormone within normal limits; and serum AMH ≥5 ng/mL at screening. The study population, including subject demographic information and full inclusion and exclusion criteria, as well as clinical outcomes, have been previously published ( 26 ). The full study protocol has been described previously ( 26 ). Subjects were randomized 1:1 immediately before administration of HP-hMG (Menopur, Ferring Pharmaceuticals, Inc.) or rFSH (Gonal-f, EMD-Serono) for controlled ovarian stimulation. Subjects began treatment on day 2 or 3 of the menstrual cycle at a dose of 150 IU HP-hMG or rFSH for the first 5 days. Beginning on day 6, the daily dose could be adjusted in 75 IU increments up to a maximum of 300 IU/day. Treatment could not exceed 20 days, and coasting was not permitted. Gonadotropin-releasing hormone antagonist was administered when the lead follicle measured ≥14 mm and/or serum estradiol levels were ≥300 pg/mL. Oocyte maturation was induced by hCG when three follicles ≥17 mm were observed. Subjects who received an hCG trigger underwent fresh transfer of a single blastocyst; in the event of potential excessive ovarian response (>30 follicles of ≥12 mm each and/or estradiol levels ≥5,000 pg/mL), a GnRH agonist trigger was administered to induce oocyte maturation and fresh transfer was cancelled. The primary endpoint of the MEGASET-HR trial was ongoing pregnancy rate per cycle start after fresh transfer, defined as the presence of ≥1 intrauterine pregnancy with a fetal heartbeat 10–11 weeks gestation (8–9 weeks after fresh blastocyst transfer) ( 26 ). Secondary endpoints included incidence and grade of OHSS, endocrine profile, and number of oocytes retrieved. In this post hoc analysis of the MEGASET-HR trial, we examined the characteristics of subjects in each treatment arm who developed OHSS, as well as the association of OHSS incidence with ovarian response to stimulation, trigger type, and oocyte yield. Ovarian hyperstimulation syndrome was classified using Golan’s classification system ( 27 ) and all OHSS cases were graded as mild, moderate, or severe. Preventative interventions for early OHSS included cycle cancellation due to excessive ovarian response, triggering of final follicular maturation with GnRH agonist, and administration of dopamine agonist (the latter was only considered as a preventative intervention in subjects with ≥20 follicles of ≥12 mm). Early OHSS was defined as OHSS with onset ≤9 days after triggering of final follicular maturation, and late OHSS was defined as OHSS with onset >9 days after triggering of final follicular maturation. If OHSS was suspected, the subject underwent transvaginal ultrasound to document ovarian size and presence of ascites, as well as serum chemistry and hematology assessments. The MEGASET-HR trial was powered to demonstrate noninferiority in ongoing pregnancy rate for fresh cycles treated with HP-hMG vs. rFSH ( 26 ). The trial randomized 620 participants, 619 of whom received treatment (310 HP-hMG vs. 309 rFSH). This post hoc analysis was performed to evaluate the risk of OHSS in this population of potential high responders after stimulation with HP-hMG vs. rFSH and to analyze differences between those participants who did or did not experience OHSS as well as any treatment-related differences in OHSS outcomes. Continuous variables were summarized using the number of observations, mean, and standard deviation. Categorical variables were summarized using the number of observations and percentage of subjects with said observation. Non-parametric tests were used to compare demographics, baseline characteristics, and end-of-stimulation values, namely, the Kruskal-Wallis test for continuous variables and Fisher’s exact test for categorical variables. To examine whether baseline AMH is related to the number of oocytes retrieved and whether the effect is modulated by treatment, a linear regression model was used to model oocytes retrieved as a function of both treatment and baseline AMH. To assess whether the number of oocytes retrieved impacts the odds of early OHSS and whether the impact is modulated by treatment, a logistic regression model was used to model the probability of early OHSS as a function of treatment and oocytes retrieved. Both models were fit with and without an interaction term, using the interaction model if the interaction term was statistically significant at the 5% significance level. Otherwise, the main effects model was used for inference. For the overall treatment comparison of OHSS, 95% confidence intervals (CI) were derived from the t-distribution. For the more granular summaries based on type of OHSS or categories of oocytes retrieved, exact 95% CI for differences in proportions were created using the Agresti-Min methodology.

In the MEGASET-HR trial, despite the exaggerated difference in mean numbers of oocytes retrieved in high responders compared with previous trials in normal responders ( 41 , 42 , 44 ), cumulative live birth rates were comparable in the HP-hMG and rFSH groups after fresh and frozen transfer ( 26 ). However, significantly fewer subjects in the HP-hMG group developed OHSS compared with those treated with rFSH; consequently, more subjects in the HP-hMG group were able to undergo fresh transfer. When evaluating ovarian stimulation protocols of comparable efficacy, protocol individualization may consider the differential safety in patients at high risk of developing OHSS. Characterizing the differences in gonadotropin stimulation profiles offers an evidence-based approach toward therapeutic individualization in ART.

In this post hoc analysis of a randomized, open-label, assessor-blind, parallel-group, multicenter, noninferiority US trial, we established that predicted high-responder patients treated with HP-hMG have a significantly diminished incidence of OHSS compared with those treated with rFSH, independent of the number of oocytes retrieved, but had comparable live birth rates after fresh and/or frozen transfer. Subjects who developed any OHSS were significantly younger, consistent with prospective and retrospective studies indicating that younger age is associated with increased risk of OHSS ( 28 , 29 , 30 , 31 , 32 , 33 ). Subjects with early OHSS also had significantly lower weight and BMI. Studies investigating the link between BMI and OHSS have produced conflicting results. Two studies support a correlation between lower BMI and development of OHSS whereas four additional studies show no predictive value ( 28 , 29 , 33 , 34 , 35 , 36 ). There were no differences in baseline AMH or AFC between subjects who developed OHSS and those who did not, although elevated AMH and AFC have been shown to be predictive of OHSS in several studies ( 36 , 37 , 38 , 39 ). Likewise, no differences were observed in the proportion of patients within each AMH level stratification. This may be due to the fact that, unlike prior studies, the MEGASET-HR trial population was restricted to a patient cohort with the highest risk of OHSS, as all subjects had AMH >5 ng/mL, potentially masking any effect. No significant differences in baseline demographics based on treatment group were identified, indicating that subjects with a heightened risk of OHSS were likely to have been evenly distributed. Further, there were few differences in demographics between rFSH-treated vs. HP-hMG–treated subjects who developed any OHSS or early OHSS. Subjects in the HP-hMG group had lower rates of OHSS despite receiving a higher total dose of gonadotropin with a longer mean duration of stimulation ( 26 ). Together, this suggests that the difference in OHSS incidence observed is related to the treatment regimen rather than subject characteristics. Subjects who received an hCG trigger had a similar incidence of OHSS compared with those who received a GnRH agonist trigger, despite a higher ovarian response in subjects receiving GnRH agonist trigger per the trial protocol criteria. Incidence of mild OHSS was higher in subjects who received a GnRH agonist trigger, although rates of moderate and severe OHSS were similar between treatment groups. These results suggest that the use of GnRH agonist trigger does not eliminate the risk of OHSS, but may reduce its severity, consistent with the published literature ( 3 , 10 , 11 ). Of note, the criteria for applying GnRH agonist trigger and freeze-all were selected on the basis of clinical practice at the time of trial initiation in 2015. Current American Society for Reproductive Medicine Practice Committee Guidelines recommend preventative strategies for patients at a moderate-to-severe risk of OHSS, defined as >17 follicles 10 mm or larger and/or estradiol of >3,500 pg/mL at the time of trigger ( 40 ). Guidelines recommended strategies for these patients include use of a GnRH antagonist protocol, dopamine agonist administered on the day of and several days after the hCG trigger, GnRH agonist trigger, and/or cancellation of fresh transfer with cryopreservation of all embryos ( 40 ). The rate of early OHSS after GnRH agonist trigger in this study (16.5%; 15/91) is similar to that reported in a recent combined analysis of high responders from four clinical trials (22%; 17/77) ( 10 ). More rFSH than HP-hMG–treated subjects met the criteria for a GnRH agonist trigger (>25 follicles and estradiol levels >5000 pg/ml) and underwent freeze-all with cancellation of fresh transfer. Therefore, more subjects treated with HP-hMG were able to undergo fresh transfer. The incidence of OHSS is directly associated with peak estradiol levels and oocyte yield. Estradiol levels at the end of stimulation were similar between treatment groups in subjects who developed OHSS; VEGF levels were not assessed. The mean oocyte yield in this trial was 15.1 (±10.12) in HP-hMG–treated subjects compared with 22.2 (±11.54) in those treated with rFSH, resulting in a mean difference of approximately 7 oocytes per subject ( 26 ). This pattern is consistent with that observed in previous studies comparing HP-hMG with rFSH stimulation in normal responder subjects ( 41 , 42 , 43 , 44 ); however, the mean difference in retrieved oocyte number was exaggerated in this sample of potential high-responder subjects. One of the most interesting findings in this study was that, despite an increase in incidence of early OHSS with increasing oocyte yield, fewer HP-hMG–treated subjects with high oocyte yields (>25 oocytes) developed OHSS compared with subjects with high oocyte yields treated with rFSH. Likewise, after adjusting for the number of oocytes retrieved, incidence of OHSS remained lower with HP-hMG treatment vs. rFSH. In contrast to rFSH, HP-hMG is comprised of an equimolar combination of human-derived FSH and hCG-driven luteinizing hormone (LH) activity. Early follicular recruitment is attributed to the actions of FSH, whereas LH and hCG have been shown to be important for follicular selection later in ovarian stimulation ( 45 ). Administration of exogenous LH or hCG during stimulation has been shown to reduce the number of small follicles, allowing preferential growth of large and dominant follicles ( 45 , 46 , 47 ). Low overall numbers of follicles reduce peak estradiol levels and subsequently diminish the risk of developing OHSS. These differences in the selection and growth of different cohorts of follicles could explain the maintenance of efficacy despite differences in oocyte yield. It does not, however, fully explain the difference in OHSS incidence among those with high oocyte yields seen in this trial. Luteinizing hormone/hCG act via LH/hCG receptors on theca cells throughout stimulation to produce androgens that are then metabolized in granulosa cells to estradiol, the key driver of follicular growth. It is likely that hCG-induced hyperandrogenemia benefits larger follicles with increased substrate to produce estradiol and grow. In smaller follicles, high androgens overwhelm the metabolic capability to inhibit estradiol production and further growth ( 48 ). Differences in the post-receptor stimulation profile of human FSH compared to Chinese hamster ovarian cell-derived rFSH may also play a role. Additional investigation is necessary to determine whether these differences in steroid output induced by stimulation with rFSH or HP-hMG could explain the differences in OHSS incidence observed between these two treatments. This study has several notable strengths. The primary strength lies in the use of data from a rigorously conducted randomized controlled trial in a highly characterized study population. The availability of comprehensive, prospectively collected data allowed for a robust exploratory analysis. However, there are important limitations inherent to both the trial design as well as to the post hoc nature of the analysis. The strict criteria used for GnRH agonist trigger and freeze-all is not representative of current practice, where a variety of trigger mechanisms are often used, including GnRH agonist only with freeze-all or dual triggers with low-dose hCG and GnRH agonist with fresh transfer. It is possible that the use of such mechanisms may have reduced both the risk of OHSS and the need for freeze-all in the trial. As a post hoc analysis, this study’s hypotheses were formulated after examining the data and should be considered hypothesis-generating rather than confirmatory. There is an increased risk of type I error, and the MEGASET-HR trial was not powered to detect specific differences in the incidence of OHSS. Furthermore, the use of post-randomization variables in statistical models may introduce bias. Future trials will need to examine the impact of modern ART protocols on OHSS incidence and the rate of fresh transfer in high responders.

R.F. was an investigator in the Menopur in Gonadotropin-releasing hormone Antagonist Single Embryo Transfer–High Responder (MEGASET-HR) trial. C.C.S. was an investigator in the MEGASET-HR trial. V.K. was an investigator in the MEGASET-HR trial. L.K.K. is a full-time employee of Ferring Pharmaceuticals, Inc. E.D.F. reports payments from Ferring Pharmaceuticals, Inc. as a statistical consultant for the submitted work, and was an employee of Ferring Pharmaceuticals, Inc. during the design, conduct, and/or analyses of the MEGASET-HR trial. P.W.H. was an employee of Ferring Pharmaceuticals, Inc. during the design, conduct, and/or analyses of the MEGASET-HR trial. G.S.D. was an employee of Ferring Pharmaceuticals, Inc. during the design, conduct, and/or analyses of the MEGASET-HR trial. S.L. was an employee of Ferring Pharmaceuticals, Inc. during the design, conduct, and/or analyses of the MEGASET-HR trial.