A randomized clinical trial to evaluate the effect of granulocyte macrophage colony stimulating factor GM CSF in embryo culture medium for in vitro fertilization

In a prospective randomized study, addition of recombinant granulocyte-macrophage colony-stimulating factor to IVF embryo culture medium was found to increase ongoing implantation rate at gestational week 12 and live birth rate.

Søren Ziebe, Ph.D., D.Sc., Anne Loft, M.D., Betina B. Povlsen, M.Sc., Karin Erb, M.Sc., Inge Agerholm, Ph.D., Michael Aasted, M.D., Anette Gabrielsen, M.Sc., Christina Hnida, Ph.D., Dorit P. Zobel, Ph.D., Bibi Munding, M.Sc., Susanne H. Bendz, Ph.D., Sarah A. Robertson, Ph.D.

Volume 99, Issue 6, Pages 1600-1609.e2, May 2013


To evaluate the effect of granulocyte-macrophage colony-stimulating factor (GM-CSF) in embryo culture medium on ongoing implantation rate (OIR).

Multicenter, randomized, placebo-controlled, double-blinded prospective design.

Fourteen Scandinavian fertility clinics.

A total of 1,332 women with indication for in vitro fertilization or intracytoplasmic sperm injection; 1,149 received embryo transfer (GM-CSF: n = 564; control: n = 585).

Oocytes were fertilized, and embryos cultured and transferred in control medium or test medium containing 2 ng/mL GM-CSF.

Main Outcome Measure(s):
OIR at gestational week 7, with follow-up at week 12 and birth.

At week 7, OIRs were 23.5% (GM-CSF), and 20.0% (control) (odds ratio [OR] 1.26, 95% confidence interval [CI] 0.91–1.75). At week 12, OIRs were 23.0% (GM-CSF) and 18.7% (control) (OR 1.35, 95% CI 1.06–1.72), and live birth rates were 28.9% and 24.1%, respectively (OR 1.35, 95% CI 1.03–1.78). The effect of GM-CSF was influenced by the human serum albumin concentration in the medium. Birth weight and abnormality incidence were similar in both groups. Exploratory analyses showed that GM-CSF increased OIR in women with previous miscarriage, especially in women with more than one miscarriage.

Addition of GM-CSF to embryo culture medium elicits a significant increase in survival of transferred embryos to week 12 and live birth. Our results are consistent with a protective effect of GM-CSF on culture-induced embryo stress. GM-CSF may be particularly efficacious in women with previous miscarriage.

Clinical Trial Registration Number:

  • Cindy Farquhar

    Dear Sirs

    We are writing with regard to the following paper as we have concerns that the statistical analysis and subsequent conclusions are incorrect.

    Ziebe S, Loft A, Povlsen BB, Erb K, Agerholm I, Aasted M, Gabrielsen A, Hnida C, Zobel DP, Munding B, Bendz SH, Robertson SA. A randomized clinical trial to evaluate the effect of granulocyte-macrophage colony-stimulating factor (GM-CSF) in embryo culture medium for in vitro fertilization. Fertil Steril. 2013 May;99(6):1600-9.

    Our concerns are as follows:

    1. The analyses adjusted for many of the baseline factors. The more conventional approach in randomised controlled trials would be to report the odds ratio (OR) using logistic regression adjusted for only the stratification criteria in the allocation process. This would mean adjustment for ‘centre’ would be the only stratification and there
    would have been no need for other adjustments. Furthermore, it is not clear
    that this adjusted analysis was planned. The registration website
    (clinicaltrials.gov) does not provide any details of the planned analysis.
    While it is not ‘wrong’ to perform the analyses with the additional adjustment,
    it should have been clearly pre-specified in the protocol. If it is not
    pre-specified, it is possible, especially with borderline results, to seek the
    ‘best’or most favourable adjustment from thousands of possibilities. We
    consider that the list of variables that were used for adjustments is lengthy
    and not clearly defined.

    2. The primary outcome, ‘ongoing implantation rate’, erroneously counts multiple transferred embryos as independent observations. The authors have
    utilised results from these non-independent observations to conclude that there is evidence of GM-CSF being beneficial, despite acknowledging
    the large variation in results between centres. Regardless, it is always more
    useful to look at ongoing pregnancy or live birth rates although these were
    only a secondary outcome.

    3. The authors have excluded from all the analyses and following randomisation, those participants for whom embryos of sufficient quality for transfer failed to develop in the allocated medium. No definition is provided in the report on the definition of
    ‘sufficient quality’. Excluding patients after randomisation breaches the intention to treat principle and risks introducing attrition bias. The total numbers who were not included in the analysis are not inconsiderable – 90 of 654 (13.8%) in the GM-CSF group and 93 of 678 (13.7%) in the control group.

    We have undertaken intention to treat (ITT) analyses, using the number randomised to each arm of the study as the denominator, for the following clinical outcomes:

    For live birth rate: GM-CSF 163/654, and control 141/678. Unadjusted OR (95% CI 1.3
    (0.98 to 1.6). (numbers taken from Figure 1 in the paper)

    For ongoing pregnancy rate GM-CSF 165/654 and control 145/ 678 the unadjusted OR (95% CI = 1.2 (0.96, 1.6) (numbers taken from Figure 1 in the paper)

    For clinical pregnancy: GM-CSF (214-29)/654, and control (218-44)/678. Unadjusted OR (95% CI) 1.1 (0.90 to 1.5). This was calculated using positive HCG data minus
    biochemical pregnancy data.

    These results do not support the statements of ‘significant increase’ in ongoing pregnancy and livebirth in those with the addition of GM-CSF to embryo culture medium.

    4. Concern about the change in intervention following interim analysis when
    the human serum albumin (HSA) concentration was increased without
    providing a reason why it wasn’t the higher concentration used initially,
    except for the sentence “because that concentration was commonly used in
    culture media”.

    5. We note an interim analysis was undertaken and that the sample size was adjusted after this. Although the paper states under the statistical analysis section that no other changes were made it is clear from other sections of the paper that there was a change in the concentration of the human serum albumin at the time of the interim
    analysis. It is our view that the pregnancy outcomes before and after the
    change should have been provided.

    As with all real-life trials there are other potential sources of bias and points for methodological debate. However, our overriding concern relates to the potentially misleading analysis, especially given that the conclusions of the trial are being used as evidence of efficacy in marketing and promotional material for GM-CSF.

    We suggest that this paper should either be reconsidered or have the pre-specified analysis attached to it. If no specific approach was included prospectively in the statistical analysis plan, analysis under intention to treat, adjusting only for the stratification criterion, would be the standard

    Yours sincerely

    Professor Cindy Farquhar, Prof Andy Vail, Dr
    Selma Mourad and Dr Sarah Armstrong

  • Marcos Meseguer

    First I would like to congratulate Soren for the large RCT performed. This is really difficult to undergo and the conclusions are at least quite intriguing.
    I am embryologist and just wondering how the addition of GMCSF is affecting embryo quality. It would be very interesting to evaluate the features of those embryos. I am an enthusiastic of the time-lapse technology and I am sure that future papers will come to find an explanation of how embryo quality may be improved by the addition of GMCSF or other protective molecules added to the culture media. I also recomment the paper of my colleague D. Morbeck published this year in HR which may help us to understand my thoughts.
    Marcos Meseguer PhD
    Scientific Supervisor IVI Valencia

  • Nina Desai

    The GM CSF work is interesting but I still can’t get a handle on why GMCSF supplementation would help the ongoing pregnancy rate after the point of a fetal heart , In the literature and in my own studies GMCSF appears to work by reducing apoptosis especially in the inner cell mass, essentially functioning as a “survival” factor for inner cell mass cells in the blastocyst. My endometrial coculture cell line that I use for patients with failed implantation secretes GMCSF in low levels and we believe this is helpful in jumpstarting patient embryos and promoting blastocyst formation.

    Most of the published clinical studies that I am aware of are however supplementing until Day 3 and then transferring the embryos. Is this really enough to “jumpstart” the embryo, affect the endometrium and most importantly affect ongoing implantation? Sure would be nice if it was just that easy to improve outcomes!

    Nina Desai, PhD, HCLD
    IVF Lab Director
    Cleveland Clinic

    • Harold Moreno-Ortiz

      Harold Moreno-Ortiz

      I will give you a recently published case report. This is very interesting topic.

      colony-stimulating factor (G-CSF): a mediator in endometrial
      receptivity for a patient with polycystic ovary (PCO) undergoing in
      vitro maturation (IVM).
      Lucena E, Moreno-Ortiz H.BMJ Case Rep. 2013 Apr 18;2013.

  • Claudio Benadiva, MD, HCLD

    This is an interesting paper that despite its large size, I don’t believe provides a convincing argument for the use of media supplemented with GM-CSF. After the HSA concentration was increased to 5mg/ml, the differences with the control group disappeared. If I understood Supplemental Table 2 correctly, even in the subgroup of previous miscarriage patients, the beneficial effect on ongoing IR after 12 weeks was only in the low HSA group.

    Transferring embryos and assessing embryo quality at the blastocyst stage could have been a better model, in addition to allow more time for the potential benefit of GM-CSF (if any) to show an effect on embryo development.

    Another problem is the transfer of embryos in the same culture media , adding a potential effect of GM-CSF on the endometrium as another variable, since intrauterine infusion of GM-CSF has recently been used to treat women with thin endometrium.

    In summary, this is an intriguing paper that left me with many unanswered questions. I wonder if ORIGIO is planning to undertake further studies to demonstrate the benefit of using this media.

Translate »