The Frequency of Ovarian Hyperstimulation Syndrome and Thromboembolism with Originator Recombinant Human Follitropin Alfa (GONAL-f) for Medically Assisted Reproduction: A Systematic Review

This manuscript reports the frequency of OHSS and thromboembolism with GONAL-f in both the Global Safety Database of Merck KGaA, Darmstadt, Germany and in the published scientific literature. GONAL-f has a long history of usage; overall, there were an estimated 18 million treatment cycles conducted between the IBD of 20 October 1995 and 19 October 2019. This is the first cumulative report of these safety events specifically for a single gonadotropin preparation. There have been other analyses, including real-world studies, evaluating the frequency of OHSS and thromboembolism; however, the majority of previous published reports described pooled data for several gonadotropins, rather than for a single preparation. It is important to analyse data for all gonadotropin preparations separately (e.g. originator r-hFSH alfa and its biosimilars, as well as for r-hFSH beta and delta), as differences have been observed with regard to biochemical properties, in vitro activity, pharmacokinetics, pharmacodynamics and reproductive outcomes observed in clinical trials [1‐3, 67‐69]. In addition, other external safety databases such as the Uppsala Monitoring Center Vigibase and the FDA Adverse Event Reporting System (FAERS), also do not differentiate between GONAL-f and other gonadotropin preparations, and do not provide the data on the total number of cycles or the number of patients treated, which makes it impossible to calculate frequency rates.

A large retrospective chart review conducted in 2010 (using data from 2004 onwards) reported on 22,665 cycles using r-hFSH (GONAL-f or Puregon) for the indications licensed in the summary of product characteristics [70]. OHSS (any severity) was identified in 3812/20,144 (18.9%) cycles and hospitalization due to OHSS (grade not specified) was reported in 215/20,144 (1.1%) cycles. A Cochrane review covering the period from inception to 2017, including 20 trials of 6088 women treated with r-hFSH, reported OHSS rates of 2.5% for moderate or severe OHSS and 0.8% for severe OHSS for patients treated with standard dose FSH (150 IU), with rates between 0.8% and 2.5% reported for patients for whom the dose was adjusted according to the results of ovarian reserve tests [71]. However, the studies included in the Cochrane review were heterogeneous and the evidence quality of these studies ranged from low to moderate and was potentially subject to bias. The evidence from older studies indicates that OHSS is predominantly mild to moderate, with mild cases occurring in an estimated 20–33% of ART fertility treatment cycles; moderate in 3% to 6% treatment cycles and severe in 0.2–5.0% of assisted conception cycles [72‐78]. However, because the classification of OHSS has changed over time, any comparisons of the occurrence of OHSS in older and more recent studies should be interpreted with caution. Furthermore, there has been a steady decline of OHSS rates with the refinement of ovarian stimulation protocols with the use of gonadotropin-releasing hormone (GnRH) antagonist to prevent a premature LH surge [79] and use of a GnRH agonist to trigger final oocyte maturation [80], which might reflect more in recent studies.

There are internationally recognised classifications for OHSS, which have been updated several times over the years. The first OHSS classification was proposed in 1967 dividing OHSS into three categories (mild, moderate and severe) and with two degrees of severity for each category [81]. This was subsequently amended to include an additional category (‘critical’) and subcategories (grades A–C for severe OHSS) [81, 82]. More recently a new classification system has been proposed in which OHSS is classified as either ‘self-limited’ or ‘with significant comorbidities’ to simplify reporting [81]. The International Committee for Monitoring Assisted Reproductive Technologies (ICMART) defined OHSS as mild, moderate or severe, according to the degree of abdominal distention, ovarian enlargement and respiratory, haemodynamic and metabolic complications. The committee also included a distinct definition of severe OHSS in the glossary: ‘a systemic response as a result of ovarian stimulation interventions that is characterized by severe abdominal discomfort and/or other symptoms of ascites, hemoconcentration (Hct > 45) and/or other serious biochemical abnormalities requiring hospitalization for observation and/or for medical intervention (paracentesis, other)’ [15]. These historic changes in definitions make it difficult to compare OHSS risk over time.

The OHSS reporting rate in our systematic review is higher than in the analysis of the Global Safety Database of Merck KGaA, Darmstadt, Germany. Firstly, this may be because the reports of OHSS and thromboembolism were obtained from the Global Safety Database of Merck KGaA, Darmstadt, Germany from 20 October 1995 onwards and, although awareness of AE reporting has increased during the past decade, the period covered by the database also includes a period when there was less awareness of AE reporting. The extent of under-reporting of ADRs to spontaneous reporting systems and the differences among different types of ADRs up to April 2004 were addressed in a systematic review by Hazell and Shakir [83]. In this systematic review, the median under-reporting rate across 37 studies in 12 countries was 94% (interquartile range 82–98%), providing evidence of significant and widespread under-reporting of ADRs to spontaneous reporting systems, including serious or severe ADRs, during the first decade covered by our database. These under-reporting rates come from well-developed countries where a pharmacovigilance system is reasonably established, and under-reporting of ADRs in spontaneous databases for products that have been on the market for a long time and have well-known ADR profiles (such as GONAL-f) is recognised [83]. The reporting of AEs to GONAL-f in countries with less-developed pharmacovigilance systems may also contribute to the under-reporting. In addition, in the search strategy used in this analysis, only the PTs of OHSS were selected, meaning that cases which reported individual symptoms suggestive of mild OHSS (mild abdominal swelling, discomfort and nausea) may not be included. Further, the low reporting rate in the Global Safety Database of Merck KGaA, Darmstadt, Germany may relate to the reporting of only more serious cases of OHSS, with mild OHSS less frequently reported. Indeed, the proportion of severe OHSS cases reported out of all OHSS cases reported was about three times higher (11.7%) in the Global Safety Database of Merck KGaA, Darmstadt, Germany when compared to the systematic review (3.7%). We hypothesize that some OHSS cases occurring in daily practice may not be considered as “serious enough” to be reported in the post-marketing setting [84], whereas they would be captured systematically in a typical phase III RCT evaluating efficacy and safety. In some of these RCTs, OHSS might have been an adverse event of special interest (AESI), ensuring that serious and non-serious AEs will be collected and reported in an expedited manner. The selection criteria used for our systematic review favoured clinical trials, which are likely to have included proactive and enhanced monitoring for ADRs (often specified in study protocols and clinical research files). However, the reporting rate in our systematic review was within the range observed in other studies (which also included a mixture of study designs) [71‐74].

Finally, the characteristics of the patients included in clinical trials may not fully reflect those seen in clinical practice, as it is well established that only 37% of patients treated in the real world would be eligible for inclusion in randomized clinical trials [85].

With the exception of pharmaceutical company-sponsored trials, OHSS is rarely reported as a primary (4.4% of studies included in our systematic review) or even secondary (17.8% of studies) endpoint, and there is a lack of transparency and a high degree of variability/heterogeneity on how OHSS was assessed and defined in clinical trials. Indeed, only 45.5% of the studies that we assessed used a classification system specifying which parameters were used to classify OHSS. Broad summary statements often used in publications, such as “the product was generally safe” or “it was well tolerated” do not provide clinically meaningful information and should be avoided [86]. Furthermore, there is no consistent guidance provided by health authorities, journals or scientific communities on the reporting of fertility-related AEs and ADRs in publications; therefore, there is a need for a unified system for the reporting of safety outcomes in fertility studies. A task force has therefore been set up within Merck KGaA, Darmstadt, Germany to assess the way safety outcomes are reported in fertility trials and to develop guidance that can be used in future publications. The guidance will include a comprehensive checklist of safety parameters that should be assessed and reported in fertility-related studies. Generally reporting of OHSS and other safety results in fertility publications should be in line with the clinical study protocol and the clinical study report.

There are a number of potential limitations to the analyses reported in this manuscript. The data range from 1995 to 2018 and, as previously stated, owing to changes in clinical practice, the overall data provided might not fully reflect what is observed in current practice. Our understanding of OHSS has increased in recent years, and the risk can now be greatly reduced with appropriate treatment selection: individualized choice of gonadotropin type and starting dose; dose adjustment based on hormonal and ultrasound monitoring during ovarian stimulation; choice of ovulation trigger; choice between embryo transfer in fresh stimulation cycle or delayed embryo transfer after cryopreservation of all oocytes/embryos. For this reason, data from older studies and analyses are unlikely to provide an accurate representation of OHSS risk at the current time. Ovarian stimulation should be monitored via ultrasound and serum oestradiol levels, and if multiple immature follicles or high serum oestradiol levels are observed the dose can be reduced, a coasting strategy (i.e. stopping gonadotropin treatment until serum oestradiol levels return to normal levels) employed or the cycle cancelled, depending upon the perceived risk by the treating physician/healthcare professional [82]. Current guidance is that women at high risk for a high ovarian response and related OHSS risk should be identified before treatment is started, so that an appropriate treatment strategy can be selected to mitigate the risk. Risk factors for OHSS include the presence of polycystic ovarian syndrome, a prior history of OHSS, low body weight and younger age (less than 30 years old) [87, 88]. Biomarkers may also be used to identify patients at risk for OHSS, including AFC, AMH, day 3 FSH and inhibin B [87, 89]. If the patient wishes to have a maximal probability for a fresh embryo transfer, while avoiding both cycle cancellation and freeze-all due to high risk for moderate-severe OHSS, the number of oocytes should be limited to around 15, with a maximum of 20, oocytes. In order to achieve this, the r-hFSH starting dose should be individualized on the basis of existing normograms that include patient age, AFC, AMH [7, 16] and potentially other factors, such as BMI [9] or additional ultrasound markers [11], rather than adopting a “one-size-fits-all” dosing [90] such as 100 IU [10] or 150 IU [14]. This strategy will typically result in r-hFSH starting doses lower than 150 IU, as recommended before [8, 13, 90], varying between 75 and 125 IU, as shown in the example provided in the “Introduction”. In addition, GnRH antagonists should be used, as they enable gonadotropins to be used for a shorter duration and are known to reduce the OHSS risk [8, 12].

Gonadotropin dose adjustment using small dose changes (12.5–25 IU), in combination with combined ultrasound and hormonal monitoring, during ovarian stimulation is needed to adjust for both a lower or higher than expected ovarian response [8]. In case of an increased risk for OHSS, a GnRH agonist and/or low-dose hCG can be used to trigger ovulation, followed by, if possible (and in line with patient preference), a fresh embryo transfer or (if high risk for OHSS) cryopreservation of all embryos (freeze all) followed by embryo transfer in subsequent frozen embryo transfer cycles.

Furthermore, the definitions for classification of OHSS and thromboembolism have changed during the time period that was evaluated. The cases reported in the Global Safety Database of Merck KGaA, Darmstadt, Germany were either submitted voluntarily or obtained from the literature. Case reports were excluded from the systemic literature search, which helped to avoid duplication. There may, however, be some duplication, in the sense that we cannot rule out that some cases included in the systematic review were also reported as individual cases in the Global Safety Database of Merck KGaA, Darmstadt, Germany. We anticipate that any overlap, if present, will be small and would increase the incidence of OHSS in the Global Safety Database of Merck KGaA, Darmstadt, Germany, resulting in an overestimate of the OHSS risk in this database. Although the two analyses are complementary, the reported outcomes should be considered separately to avoid overestimation of the frequency of OHSS. The number of treatment cycles in the post-marketing setting was estimated from sales data, and the number of treatment cycles per patient was not available because of variability between the units used per patient. The number of treatment cycles was therefore used as an alternative, which is generally an accepted pharmaco-epidemiological approach [91] and, in our experience, one that is accepted by health authorities when companies submit product-specific Periodic Benefit Risk Evaluation Reports. In the systematic review of clinical studies, the number of patients treated was available, however, not presented to keep the consistency in reporting between the literature review and the review of Global Safety Database of Merck KGaA, Darmstadt, Germany. In the systematic review, the publications included a mixture of study designs, which included different durations of follow-up and monitoring for ADRs. There may be the potential for under-reporting, as most of the studies were powered for efficacy rather than safety, and sometimes safety events are combined or not reported in primary publications. The systematic review only included studies which stated the starting dose and where this dose was within the range stated in the prescribing information; this resulted in the inclusion of a greater number of RCTs than studies with other designs. The severity of OHSS was based upon the diagnosis in the publication and there was heterogeneity in the definitions used.