Individualized recombinant human follicle-stimulating hormone dosing using the CONSORT calculator in assisted reproductive technology: a large, multicenter, observational study of routine clinical practice

In women undergoing controlled ovarian stimulation (COS) during assisted reproductive technology (ART), daily doses of recombinant human follicle-stimulating hormone (r-hFSH) to induce multifollicular development typically range from 100–350 IU. 1 However, women differ greatly in their ovarian response to gonadotrophin stimulation, and there is currently no established consensus for determining the optimal follicle-stimulating hormone (FSH) dose, with starting doses often based on patient characteristics, such as age, combined with a physician’s clinical experience and judgment. 1 Various factors have been identified that may be associated with ovarian response to COS, including: patient age, 2 – 5 body mass index (BMI), 6 estradiol, 7 basal FSH, 7 – 9 inhibin-B, 7 , 8 anti-Müllerian hormone, 2 , 7 ovarian stromal blood flow, 10 and antral follicle count (AFC). 7 , 8 , 10 , 11 The CONsistency in r-FSH Starting dOses for individualized tReatmenT (CONSORT) dosing algorithm was developed to determine the optimal starting dose of r-hFSH (follitropin alfa) in normo-ovulatory women aged 18–34 years undergoing COS in a long gonadotrophin-releasing hormone (GnRH) agonist protocol. 1 , 12 Four baseline variables were included in the CONSORT calculator, 1 namely age, BMI, early follicular phase serum FSH level, and AFC. In a multinational pilot study to clinically validate the CONSORT calculator, 76.4% (123/161) of patients were allocated a lower dose than the physician would have prescribed in routine practice. 12 Despite achieving good clinical pregnancy rates (34.2% per started cycle), cycle cancellation owing to an inadequate response often occurred in the lowest starting dose group (75 IU). 12 The CONSORT algorithm was subsequently modified so that the lowest starting dose that could be recommended was 112.5 IU. A multinational, randomized study was then conducted to compare the ovarian response in normo-ovulatory women (number [n] =200) aged 18–34 years who received a daily dose of r-hFSH allocated by the CONSORT calculator versus a standard starting dose of 150 IU. 13 Mean daily and total doses of r-hFSH were significantly lower in the CONSORT group than in the standard-dosing group ( P <0.001 for both). In the CONSORT group, fewer oocytes were retrieved when compared with the standard-dosing group ( P =0.037; primary endpoint); however, the number of embryos transferred (ET) in each group ( P =0.674) and the clinical pregnancy rates (CONSORT 36.0% versus standard dosing 35.5%; estimated difference [95% confidence interval {CI}]: 0.6 [−13.5, 14.6]) were similar. 13 The postmarketing surveillance study reported here evaluated the starting dose of r-hFSH recommended by the treating physician, the dose recommended by the CONSORT calculator, and the dose actually received by the patient in routine clinical practice in Germany.

This 3-year, multicenter, open-label, observational, post-marketing surveillance study ( ClinicalTrials.gov identifier: NCT01100333 ) evaluated data from women treated at 31 German ART centers. Standard practice in Germany for ART during the study period was followed; therefore, up to three embryos could be transferred and supernumerary two pronuclei oocytes could be cryopreserved for future use. The primary objective of this noninterventional study was to compare the starting dose of r-hFSH recommended by the treating physician for COS with the starting dose selected by the CONSORT calculator and the dose actually prescribed to the patient. The study collected data from women who were scheduled to undergo COS with follitropin alfa (GONAL-f ® ; Merck Serono SA, Geneva, Switzerland, a subsidiary of Merck KGaA, Darmstadt, Germany) to induce multifollicular development in an ART cycle. Key inclusion criteria included baseline age ≤35 years, BMI ≤30 kg/m 2 , and early follicular phase (cycle days 2–4) basal serum FSH levels ≤12 IU/L. Patients were excluded if they required combination treatment with clomiphene citrate, follitropin beta, urine-derived human FSH, human menopausal gonadotrophin, or luteinizing hormone during COS. One ART cycle per patient was included in this analysis; in 67.3% of patients, this was their first ART cycle and in 13.6%, this was their second ART cycle. For the remaining patients, this cycle number was three or above. Prior to the initiation of COS, physicians were asked to record their recommended starting dose of r-hFSH for each patient. Each patient’s age, BMI, early follicular phase serum FSH level, and AFC were then entered into the online CONSORT calculator, and the CONSORT-recommended starting dose of r-hFSH was recorded. The CONSORT calculator could select one of seven r-hFSH starting doses: 75 IU; 112.5 IU; 150 IU; 187.5 IU; 225 IU; 262.5 IU; or 300 IU. However, the treating physician could prescribe any starting dose of r-hFSH, and they were asked to recommend this before they were aware of the CONSORT-recommended dose. After this, they were free to follow the CONSORT-recommended dose, their original recommended dose, or a different prescribing dose. Dose reduction was possible if the treating physician considered a patient to be at risk of ovarian hyperstimulation syndrome (OHSS). Also, cycles could be cancelled in cases of inadequate or excessive ovarian response. The physician-planned starting dose of r-hFSH, four baseline factors, CONSORT-calculated starting dose, and the starting dose actually prescribed for the patient were recorded and faxed to ANFOMED GmbH (Möhrendorf, Germany). Additional routine clinical and laboratory data were entered prospectively by the physician into a standardized electronic data collection system (RecDate 14 ). All data were recorded anonymously. The primary endpoint was a comparison of the starting doses of r-hFSH recommended by the treating physician, selected by the CONSORT calculator, and those actually administered to the patient. Secondary endpoints were dose of r-hFSH on the last day of COS, total dose of r-hFSH, duration of COS, and the number of oocytes retrieved. Clinical pregnancy (ultrasound identification of an intrauterine gestational sac with fetal cardiac activity) rates were also reported. Safety outcomes included the incidence of serious adverse events (SAEs). OHSS was considered to be an SAE if it resulted in hospitalization. Grade 1 OHSS was considered to be a normal response to COS. No formal sample size calculation was performed, owing to the exploratory and observational nature of the study. All statistical analyses were unplanned, and therefore conducted post hoc. The differences between the physician-recommended, CONSORT-calculated, and prescribed starting doses of r-hFSH were compared using the two-sided Wilcoxon signed-rank test, with P -values of ≤0.05 considered to be significant. Secondary outcomes were compared between patients who received the CONSORT-calculated starting dose (CONSORT group) and those who did not receive the CONSORT-calculated dose (non-CONSORT group) using descriptive statistics. Clinical pregnancy rates per started cycle and per embryo transfer were also compared between groups using a two-sided chi-square test, with P -values of ≤0.05 considered to be significant. There was no imputation for missing data; therefore, for each variable, the number of cycles included differed. The planned inclusion criteria restricted enrollment to patients aged ≤35 years, with a BMI ≤30 kg/m 2 and who used a long GnRH agonist protocol; however, patients aged ≥35 years (n=87) with a BMI >30 kg/m 2 (n=16), or who used a GnRH antagonist protocol (n=707), were included in the analyses, as data on the use of the CONSORT algorithm in these patient groups were considered to be of scientific value. Additional comparisons of pregnancy rates in each group were performed for patients aged <35 years and ≥35 years. A two-sided chi-square test was used to compare the clinical pregnancy rates of patients aged <35 years, while a two-sided Fisher’s exact test was used to compare clinical pregnancy rates in patients aged ≥35 years owing to the small sample size.

Evaluable data on 2,579 patients were matched with cycle data saved on the RecDate database for 2,579 cycles of ART between April 2008 and July 2011. Baseline patient characteristics are shown in Table 1 . A long GnRH agonist protocol was used in 70.0% (1,649/2,356) of cycles, while the remaining 30.0% (707/2,356) of cycles used a GnRH antagonist, or a short or ultralong GnRH agonist protocol. Details of the GnRH protocol used for downregulation were missing for 208 patients; 15 patients did not undergo downregulation. In patients with both the CONSORT-calculated and the prescribed starting doses available (n=2,477), 27.3% (677/2,477) received the CONSORT-calculated starting dose and 72.7% (1,800/2,477) received a non-CONSORT-calculated starting dose. The proportion of cycles during which patients in the non-CONSORT group were prescribed r-hFSH starting doses that were higher or lower than their CONSORT-recommended dose are shown in Figure 1 . The mean doses recommended by the physician and CONSORT-calculated doses are shown in Table 2 . The CONSORT-calculated mean dose was significantly lower than the physician-recommended mean dose ( P <0.0001); however, no significant difference was found between the mean starting doses of r-hFSH recommended by the physician and the actual dose received ( Table 2 ). The mean dose of r-hFSH on the last day of COS and the mean total dose of r-hFSH were both lower in the CONSORT group when compared with the non-CONSORT group; however, the need for dose adjustment and duration of COS were similar in both groups ( Table 3 ). For patients who received an r-hFSH starting dose that was higher or lower than the CONSORT-calculated dose, the mean (standard deviation [SD]) doses of r-hFSH on the last day of COS and of total r-hFSH were higher than those in the “CONSORT-calculated dose” group ( Table 4 ). Treatment outcomes for the CONSORT and non-CONSORT starting dose groups are shown in Table 5 . Similar numbers of oocytes were retrieved and similar numbers of embryos per patient were transferred in the two groups. The clinical pregnancy rate per ET was higher in the CONSORT group than in the non-CONSORT group ( P =0.049), but there was no difference in the clinical pregnancy rates per COS cycle started between the two groups ( P =0.142). In women aged <35 years (96.6%; n=2,491), clinical pregnancy rates per ET were higher in the CONSORT than in the non-CONSORT group ( P =0.044), but clinical pregnancy rates per started cycle were similar in the two groups ( P =0.144; Table 6 ). For women aged ≥35 years (3.4%; n=87), clinical pregnancy rates per ET ( P =0.273) and per COS cycle ( P =0.296) were slighter lower in the CONSORT group versus the non-CONSORT group, but these differences were not significant ( Table 6 ). Symptoms of OHSS were reported in 11.3% (291/2,579) of cycles (OHSS data were missing for 34.6% [892/2,579] of cycles); 3.0% (78/2,579) were grade 2 (CONSORT: 2.8% [19/677]; non-CONSORT: 2.9% [53/1,800]) and 0.5% (14/2,579) were grade 3 (CONSORT: 0.3% [2/677]; non-CONSORT: 0.6% [11/1,800]). OHSS was considered an SAE (ie, required hospitalization) in 0.81% (21/2,579) of cases (CONSORT group: 0.74% [5/677]; non-CONSORT group: 0.83% [15/1,800]). Miscarriage occurred in 14.8% (132/889) of clinical pregnancies (CONSORT: 12.5% [33/263]; non-CONSORT: 15.3% [96/626]). Singleton births occurred in 55.2% (363/658) of patients, twins in 24.3% (160/658), and triplets in 0.46% (3/658). No births were reported in 20.1% (132/658) of cycles. Data were missing for 265 cycles.

Dosing models for use in COS represent an important step toward individualization of ART protocols. In this large observational study, the starting doses of r-hFSH for COS recommended by physicians in routine clinical practice were generally higher than the CONSORT-calculated doses. In addition, most patients received an actual starting dose of r-hFSH that was higher than the CONSORT-calculated dose, suggesting a lack of trust by the physician in the CONSORT-calculated dose. Further research to evaluate the full clinical impact of dosing algorithms on safety and efficacy during COS is warranted.

This postmarketing surveillance study compared the starting dose of r-hFSH recommended by the CONSORT calculator with that planned by the treating physician and the starting dose actually received by the patient. Overall, approximately one-third of patients were prescribed the CONSORT-calculated starting dose of r-hFSH. Most of the patients in the non-CONSORT group (81.5%) received a starting dose of r-hFSH that was higher than the CONSORT-calculated dose. Indeed, the mean CONSORT-calculated starting dose (134.5 IU) was lower than both the starting dose recommended by the physician (164.6 IU) and the actual starting dose (162.2 IU). Although similar proportions of patients in the CONSORT and non-CONSORT groups required dose adjustments during COS, the mean r-hFSH dose on the last day of COS and mean total r-hFSH dose were higher in the non-CONSORT group. The mean number of oocytes retrieved (11.4 oocytes versus 10.6 oocytes, respectively) and clinical pregnancy rates per started cycle (34.8% versus 38.8%, respectively [ P =0.142]) were similar in the non-CONSORT and CONSORT groups. However, clinical pregnancy rates per ET were lower in the non-CONSORT group (39.5%) than in the CONSORT group (45.0% [ P =0.049]). This may have been due to differences in procedures across the 31 centers. Furthermore, this study was not powered to analyze secondary endpoints. Nevertheless, the clinical pregnancy rates per ET in our study compare favorably with those reported from the German IVF Registry, which includes women aged >35 years using in vitro fertilization (29.5%–30.0%) and 28.4% using intracytoplasmic sperm injection. 15 , 16 Severe OHSS (grade 3) was reported in 0.5% of cycles, with a lower frequency in the CONSORT group than in the non-CONSORT group (0.3% versus 0.6%, respectively). Miscarriage occurred in 14.8% of clinical pregnancies, with a numerically lower rate in the CONSORT group than in the non-CONSORT group (12.5% versus 15.3%, respectively). The rates of miscarriage per clinical pregnancy observed in this study were lower than those calculated using data (fresh cycles) from the 2008 and 2010 reports of the German IVF Registry (18.9% and 18.8% per clinical pregnancy, respectively). 15 , 16 We acknowledge the limitations associated with the observational nature of this postmarketing surveillance study, which include a lack of randomization, potential patient selection bias, and the reliance on the accurate reporting of events. One of the main criticisms of observational data is the selection bias that may arise during assignment of a treatment by the physician and the lack of matched patient populations in treatment groups. 17 In this study, the treating physicians may have based their choice of r-hFSH starting dose on nonstudied covariates, or they may have been influenced by the CONSORT calculator, altering their opinion of an appropriate recommended dose for consecutive patients treated. In addition, patients at greater risk of poor or excessive ovarian response were more likely to have cycles cancelled; the withdrawal of these patients could bias the results, for example, when calculating clinical pregnancy rates per ET. Also, owing to reliance on the accurate reporting of events, the numbers of cycles for which evaluable data were available for outcomes differed and, therefore, these results may not be generalizable to all populations of women undergoing ART. Furthermore, since the primary population in this study comprised women aged ≤35 years with a BMI of ≤30 kg/m 2 – a population of women less commonly seen in United States clinical practice, for example – these data may not be applicable to all populations of women. Although SAEs have been described here, the reporting of adverse events is not required in German postmarketing surveillance studies. Finally, it should also be noted that all statistical analyses were carried out post hoc, so any conclusions from these data should be made with caution. In the present study, the CONSORT calculator only selected 75 IU as the starting dose in three patients. Interestingly, the CONSORT calculator pilot study found an inadequate response using starting doses of 75 IU, and this dose was subsequently excluded from the CONSORT randomized controlled study. 13 Despite the original CONSORT algorithm being developed for women aged <35 years (undergoing COS using a long GnRH agonist protocol), 1 , 12 no significant difference was found in clinical pregnancy rates (per ET or per started cycle) for the CONSORT and non-CONSORT groups for the subset of older (≥35 years) women in this study. However, owing to the small patient number, these findings should be treated with caution. Also, short or ultralong GnRH agonist or antagonist protocols were used for some patients, again representing scenarios that the CONSORT calculator was not developed for.