Moving towards agreement in ovarian stimulation

Published 21 November 2019

A well attended Campus meeting in October reviewed themes of debate in ovarian stimulation following publication of ESHRE's new guidelines. Not all were simple matters of consensus, especially in those areas where evidence was thin or 'conditional'.

The title given to a recent Campus meeting on the newly published ESHRE guidelines was 'Do we have agreement?', a question which one assumes was rhetorical. For, as so many other guidelines and systematic reviews now attest, the development of universally accepted recommendations today is limited by a paucity of strong evidence-based data and by heterogeneous studies. So it's a sign of the times that so many of the recommendations in this new guideline on ovarian stimulation for IVF and ICSI were described as 'conditional' (with 'little confidence in the effect estimate' or 'probably not recommended'), or at best acknowledged as a 'good practice point'. And all this despite 2741 abstracts and/or full texts screened by the guideline development group.(1)

The presentations at this very well attended meeting (more than 180 participants) did not simply summarise the headline recommendations, but considered each of the major themes in their development, particularly where there was still some contention and not necessarily 'agreement'. But it was of some gratification to development group chairman Frank Broekmans, who described the guidelines as a 'living document', that a snapshot algorithm derived from the guidelines for low, normal and high responders had almost complete support from the audience (with just two dissenters from a full show of hands).

The pre-stimulation phase
The guidelines define as 'strong' the evidence in favour of AMH or AFC for predicting likely response. However, in her presentation Simone Broer from UMC Utrecht, where much of the landmark work on ovarian reserve testing has been performed, was unable to recommend one over the other, but she did emphasise that just one test - AMH or AFC - would be sufficient. AMH test results, she said, take longer than AFC, and are subject to a range of assays, but results are reliable as a predictor of response to stimulation. Other tests - age, BMI, basal estradiol, inhibin B or FSH - were assessed for the guideline but deemed insufficiently reliable.

In terms of scheduling an individual stimulation programme Ernesto Bosch noted the development group's recognition that 'estrogen or progesterone are widely used' and that, as a good practice point, this is 'probably acceptable' given the data. However, the data were less than encouraging for pre-treatment with estrogen, progesterone or a combined oral contraceptive pill in GnRH antagonist cycles.

Agonist or antagonist?
It was Bosch who also reviewed the relative performance of an agonist or antagonist for LH suppression and explained the guidelines' strong recommendation for an antagonist in predicted normal responders. Their evident advantages lie in immediate suppression with no initial flare-up or estrogen deficiency symptoms, a short treatment cycle, and, what was a noticeably recurring theme at the meeting, the potential to remove the risk of OHSS with an agonist trigger. Moreover, Bosch explained, any early fears that antagonists were associated with lower pregnancy and live birth rates than agonists had been largely superseded, notably in a Cochrane review of 2016 and meta-analysis by Lambalk the following year. Bosch left little doubt that in his view antagonists were well favoured over agonists.

Roy Homburg in his review of predicted high responders to stimulation emphasised how quickly and how far practice has moved on in less than decade. In 2011 when Paul Devroey and colleagues proposed their OHSS-free clinic based on the segmentation model, prevention of OHSS was the paramount strategy and FSH overdosing its usual iatrogenic cause. But today, as these new guidelines reaffirm, management relies on a clear prediction by AMH or AFC, conservative dosing and avoidance of hCG (both as a trigger and in pregnancy). Hence the guidelines' 'strong' recommendation for an agonist trigger in women at risk of OHSS (with a freeze-all option in very high risk cases). Overall, therefore, the guidelines also strongly recommend antagonists with agonist trigger 'with regards to improved safety and equal efficacy'.

The low responder
However, there was not universal agreement that an antagonist cycle is preferred in a predicted low responder. The guidelines themselves indicate that agonists and antagonists 'are equally recommended' in terms of efficacy and safety, but with the added 'strong' evidence that a FSH dose of 300 IU (or more) cannot be recommended in these cases. There was evidence that a gonadotrophin dose above 150 IU may result in a higher number of oocytes in poor responders, and greater chances of having an embryo for transfer. However, there was no difference in live birth/ongoing pregnancy rates at these higher doses.

Frank Broekmans in his presentation on low responders asked if the outlook for the predicted 'poor' responder (as defined by the Bologna criteria of 2011) was always poor and noted, from a systematic review in 2011, that pregnancy rates were indeed lower in poor responders, especially in older subjects and when fewer eggs were retrieved (0–7% with one oocyte versus 11.5–18.6% with four). However, Broekmans argued that, based on the model of the Poseidon group, the idea of 'poor' prognosis (as based on the Bologna criteria) in such responders does not reflect an accurate pregnancy potential, which, he said, is mainly affected by patient age. So Broekmans' claim was that a low ovarian response has little impact on cumulative LBR - 'There is no low prognosis,' he said - which is only evident in older subjects. Thus, the key to Broekmans case was that low quantity does not necessarily mean low quality, and that prognosis for the poor responder need not always be poor. There has been recent controversy in the literature over the claims of Broekmans' group, who, after applying the Poseidon criteria to their own data, concluded that cumulative LBR in low responders was indeed primarily affected by female age. However, even the Poseidon group itself responded that oocyte quantity, as well as oocyte quality, still plays a significant role for the cumulative LBR of a Poseidon criteria patient. What does seem agreed to all groups, however, is that high FSH dosages will not
change the prognosis in expected low responders.

So if increased dosage will not help the low responder, what will? Sesh Sunkara, whose study of 2011 set a benchmark in correlating the chance of live birth with the number of eggs retrieved (up to a plateau of around 15), considered pituitary suppression, stimulation approaches, and adjuvant treatments. First, despite their favour in normal and high responders, Sunkara found no favourable evidence for antagonists in low responders, reaffirming the guidelines that both antagonists and agonists are 'equally recommended'. A Cochrane review of 2018 found more eggs retrieved with FSH doses over 150 IU, but no benefit over 300 IU. These findings were reflected in the guidelines - that a recommendation over 150 IU in low responders is 'unclear', but that more than 300 IU is 'not recommended'. A recommendation for clomiphene citrate alone or in combination with gonadotrophins, or gonadotrophins alone was similarly cited as 'strong' for predicted poor responders. However, said Sunkara, there are no reliable studies of modified natural cycle approaches in low responders, which 'probably could not be recommended'. Poor quality studies were also associated with adjuvant treatments - such as growth hormone or androgens - and none was recommended.

Sunkara's 2011 study on the number of eggs retrieved was performed in a first fresh cycle as evident in the HFEA database, and its results, although widely cited, also attracted some discussion at this meeting. In 2018, for example, a similarly large observational study (14,000 cycles) from Nikolaos Polyzos and colleagues at centres in Spain and Belgium found a significantly progressive increase of cumulative LBR with the number of oocytes. The authors reported - as did Polyzos at this meeting - that ovarian stimulation may not have a detrimental effect on oocyte or embryo quality in good prognosis patients, although he did recognise that many patients did not need large numbers of oocytes and that stimulation should be modified according to ovarian reserve and individual characteristics. Moreover, the guidelines note specifically that in GnRH agonist cycles an ovarian response of 18 or more follicles strongly suggests an increased risk of OHSS, with preventative measures recommended. Conversely, low response to ovarian stimulation is not a reason to cancel a cycle.

1. Full details of the guidelines are available on the ESHRE website at