To the Editor:
With great interest we read the article by Hill et al. (1) on the use of recombinant luteinizing hormone (r-LH) in patients of 35 years and older undergoing assisted reproductive techniques. The use of r-LH supplementation during ovarian stimulation as part of assisted reproduction treatment is controversial and literature shows conflicting results in women of 35 years and older. The available evidence does not support the hypothesis that the addition of r-LH increases the pregnancy rate in unselected patients treated in an IVF/ICSI protocol with recombinant FSH and a GnRH agonist or antagonist. But there might be a potential benefit of the use of r-LH supplementation during ovarian stimulation in women with a poor response and women of advanced age (2). An update and systematic review and meta-analysis focused on women of 35 years and older is therefore timely and of great value with regard to possibilities to individualize treatments to improve outcomes.
The authors conducted a literature search and included seven trials that met their inclusion criteria. They made two forest plots, one concerning embryo implantation and the other concerning clinical pregnancy rate (CPR). Both endpoints are nowadays in the international literature no longer acceptable in a meta-analysis of this nature. Embryo implantation rates are of no use to anyone and CPR should be replaced by at least ongoing pregnancy rate or live birth rate.
We have two concerns regarding the forest plot of clinical pregnancy. In the first place the authors used the ongoing pregnancy data of Bosch et al.(3) instead of clinical pregnancy data (which are not mentioned in that article in absolute numbers). This large study with high weighting cannot be part of the meta-analysis on clinical pregnancies. Second, the authors used incorrect data concerning the CPR reported in the study of Marrs et al. (4). Instead of using data of the CPR based on patients of ≥ 35 undergoing their first assisted reproduction cycle, they should have used all available data including patients >35 who had undergone previous attempts. By making this selection the odds ratio of the trial of Marrs et al. became unjustifiably high. By including all seven studies, of which five studies included patients who had previous attempts, the fixed effect model would have shown a non-significant odds ratio of 1.21 (0.85 – 1.72) instead of 1.37 (1.03 – 1.83). Furthermore, if the authors would have added several recently published works (5, 6), including this data in the meta-analysis, they would find that LH supplementation for women with advanced maternal age is not beneficial (OR 1.03 (0.80 – 1.34)) (Figure 1).
Regarding the forest plot on embryo implantation we are facing two miscalculations. Firstly, the authors recalculate implanted embryos to an odds ratio that was originally reported as a rate. This creates inaccuracy and could lead to over- or underestimation of the odds ratio. For instance when analyzing the data by Nyboe Andersen et al.(7) we can illustrate this error. If we use the number of gestational sacs (N = 42) and number of embryos transferred (N = 147) reported in the article by Nyboe Andersen et al. the odds ratio is 0.79 (0.39 -1.63) and not 0.82 (0.4 – 1.67) as reported by the authors using the formula to re-calculate wrongly the number of gestational sacs (N = 47) and number of embryos transferred (N = 170).
More importantly, the authors used an incorrect definition of implantation rate; the classical mistake they made is that they counted all gestational sacs of the total study group and divided it by the total number of transferred embryos. This is methodologically wrong in the case of multiple embryo transfer since embryos from the same couple are not independent cases, which is a requirement for the statistical methods employed. One should calculate the implantation rate per subject; individual implantation rate = number of gestational sacs per subject divided by the number of embryos transferred per subject. With the individual implantation rates one could calculate the “mean” implantation rate (sum of all individual implantation rates divided by the total number of subjects)(8) and differences should be tested with a nonparametric test. In our opinion, the results of the forest plot on embryo implantation are incorrect and not useful.
We appreciate the authors’ hard work on this meta-analysis, however we cannot agree with the drawn conclusions.
We would like to acknowledge the work of P.G.A. Hompes, M.D., Ph.D., Division of Reproductive Medicine, Department of Obstetrics and Gynaecology, VU University Medical Center, Amsterdam, the Netherlands; M.H. Mochtar, M.D., Ph.D., Division of Reproductive Medicine, Department of Obstetrics and Gynaecology, Academic Medical Center, Amsterdam, the Netherlands; and R. Homburg, M.B., B.S., FRCOG, Division of Reproductive Medicine, Department of Obstetrics and Gynaecology, VU University Medical Center, Amsterdam, the Netherlands.
Tamar E. König, M.D.
Lisette van der Houwen, M.D.
C.B. Lambalk, M.D., Ph.D.
Division of Reproductive Medicine, Department of Obstetrics and Gynaecology,
VU University Medical Center, Amsterdam, the Netherlands
1. Hill MJ, Levens ED, Levy G, Ryan ME, Csokmay JM, DeCherney AH, et al. The use of recombinant luteinizing hormone in patients undergoing assisted reproductive techniques with advanced reproductive age: a systematic review and meta-analysis Fertil Steril 2012;97:1108-14.e1.
2. Mochtar MH, Van der Veen, Ziech M, van Wely M. Recombinant luteinizing hormone (rLH) for controlled ovarian hyperstimulation in assisted reproductive cycles. Cochrane Database Syst Rev. 2007;18(2):CD005070. Review.
3. Bosch E, LAbarta E, Crespo J, Simon C, Remohi J, Pellicer A. Impact of luteinizing hormone administration on gonadotrophin-releasing hormone antagonist cycles: an age-adjusted analysis. Fertil Steril 2011;95:1031-6.
4. Marrs R, Meldrum D, Muasher S, Schoolcraft W, Werlin L, Kelly E. Randomized trial to compare the effect of recombinant human FSH (follitropin alfa) with or without recombinant human LH in women undergoing assisted reproduction treatment. Reprod Biomed Online 2004;8:175-82.
5. König TE, van der Houwen LEE, Overbeek A, Hendriks ML, Beemsterboer SN, Kuchenbecker WK, et al. The influence of LH substitution for GnRH antagonist blocked endogenous pituitary LH in older IVF patients. Poster presentation ESHRE 2011, Stockholm.
6. Musters AM, van Wely M, Mastenbroek S, Kaaijk EM, Repping S, van der Veen F, et al. The effect of recombinant LH on embryo quality: a randomized controlled trial in women with poor ovarian reserve. Hum Reprod 2012;27:244-50.
7. Nyboeandersen A, Humaidan P, Fried G, Hausken J, Antila L, Bangsbøll S, et al. Nordic LH study group. Recombinant LH supplementation to recombinant FSH during the final days of controlled ovarian stimulation for in vitro fertilization. A multicentre, prospective, randomized, controlled trial. Hum Reprod 2008;23:427-34.
8. Huirne JA, van Loenen AC, Schats R, McDonnell J, Hompes PG, Schoemaker J, et al. Dose-finding study of daily GnRH antagonist for the prevention of premature LH surges in IVF/ICSI patients: optimal changes in LH and progesterone for clinical pregnancy. Hum Reprod 2005;20:359-67.
Published online in Fertility and Sterility doi:10.1016/j.fertnstert.2012.04.048
The authors respond:
Reply to comments on “The use of recombinant luteinizing hormone in ART patients with advanced reproductive age: a systematic review and meta-analysis”
We appreciate the interest Konig et al. expressed in our meta-analysis on the use of rLH in patients with advance reproductive age (1). They disagree with our conclusion that the evidence from published randomized control trials suggests a benefit for the use of rLH in this patient group. In support of their argument, they present data in the form of a forest plot of their own analysis with additional studies included, discuss specific data extracted from one of the source studies, debate the value of implantation and clinical pregnancy as trial endpoints, and express concern over how implantation rate was calculated in our trial.
In their letter, they provide a forest plot that concludes no benefit from the addition of rLH in patients with advanced reproductive age. They include three papers not included in our meta-analysis. The papers from Musters et al. and Fabregues et al. were not published at the time our meta-analysis was conducted (2, 3). While it is important to consider this new data, it is obvious we could not include studies which had not yet been published at the time we conducted our meta-analysis. Further, the authors include data from a poster they have previously presented, but which has not yet been published as a peer-reviewed paper. As stated in the methods section of our paper, abstracts published only as a meeting abstract were excluded from analysis. In an effort to include only the highest quality data, only studies that had undergone a full peer-review process for journal publication were included in our analysis. Finally, they excluded the large trial from Bosch et al. in their analysis (4). This is the single largest randomized controlled trial on this topic and we believe exclusion of this study ignores valuable data and the evidence it provides regarding the question at hand.
They also suggest that our study should have utilized a different value for clinical pregnancy from the Marrs et al. trial (5). This particular study was small, with a weight of 6.73% for the total clinical pregnancy analysis. In the discussion section of our paper we identified this outlier result as a potential weakness. If we utilize the data from the Marrs et al. paper as suggested and reanalyze our data, the effect on clinical pregnancy is an OR 1.30 (0.98-1.73). The estimated effect of rLH in this analysis is similar to the OR of 1.37 we report in our paper.
We agree with the letter that ongoing pregnancy and live birth are desired endpoints for randomized controlled trials and meta-analysis. However, we disagree with the statement that implantation rate is of no use to anyone. The majority of RCT reporting rLH use implantation and clinical pregnancy as their primary endpoints, and as such we believe it was an appropriate outcome on which to report for this meta-analysis. Using only live birth as an endpoint would have resulted in the exclusion of a large amount of important data on this topic.
We agree with the final and important point raised by Konig et al. Consideration of implantation rate per patient cycle rather than per embryo transferred avoids issues of dependency in the data among embryos contributed by the same couple. Given adequate individual level data, statistical methods are available to evaluate implantation rates on a per patient basis, while still addressing the amount of information contributed by each couple. Nevertheless, implantation is often reported in primary trials on a per embryo basis and this was the case in several of the RCTs we reported on. Therefore, our data had to be converted to the same single unit for statistical synthesis. We contacted the corresponding author from each trial in our review to obtain raw data to perform this calculation. In the cases where the authors were unable to provide raw data, we used the calculation described in our methods. Statistical analysis, using each patient enrolled as the unit of measure, yields an OR estimate of 1.41 (1.02-1.95) for implantation; although the confidence interval is somewhat wider, this estimate is not substantially changed from the OR 1.36 (1.05-1.78) that we reported in our paper.
The points raised by Konig et al. are interesting and highlight the importance of reporting transparent methods when performing meta-analysis so that the results can be appropriately interpreted and reproduced. We appreciate their critiques but do not believe they materially alter the conclusion of our paper.
The authors would like to acknowledge the work of Gary Levy, M.D.; John M Csokmay, M.D.; and Alan H. DeCherney, M.D. at the Program in Reproductive and Adult Endocrinology, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland.
Micah J. Hill, D.O.
Program in Reproductive and Adult Endocrinology, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland
Eric D. Levens, M.D.
Program in Reproductive and Adult Endocrinology, Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland and Shady Grove Fertility Reproductive Science Center, Rockville, Maryland
Brian W. Whitcomb, Ph.D.
Division of Biostatistics and Epidemiology, University of Massachusetts School of Public Health and Health Sciences, Amherst, Massachusetts
Financial Support: This work was supported, in part, by the Program in Reproductive and Adult Endocrinology, NICHD, NIH, Bethesda, Maryland
Disclaimer: The views expressed in this manuscript are those of the authors and do not reflect the official policy or position of the Department of the Army, Department of Defense, or the U. S. Government.
1. Hill MJ, Levens ED, Levy G, Ryan ME, Csokmay JM, Decherney AH et al. The use of recombinant luteinizing hormone in patients undergoing assisted reproductive techniques with advanced reproductive age: a systematic review and meta-analysis. Fertil and Steril 2012;97:1108-14.e1.
2. Fabregues F, Iraola A, Casals G, Creus M, Carmona F, Balasch J. Evaluation of two doses of recombinant human luteinizing hormone supplementation in down-regulated women of advanced reproductive age undergoing follicular stimulation for IVF: a randomized clinical study. Eur J Obstet Gynecol Reprod Biol 2011;158:56-61.
3. Musters AM, van Wely M, Mastenbroek S, Kaaijk EM, Repping S, van der Veen F et al. The effect of recombinant LH on embryo quality: a randomized controlled trial in women with poor ovarian reserve. Hum Repro 2012;27:244-50.
4. Bosch E, Labarta E, Crespo J, Simon C, Remohi J, Pellicer A. Impact of luteinizing hormone administration on gonadotropin-releasing hormone antagonist cycles: an age-adjusted analysis. Fertil and Steril 2011;95:1031-6.
5. Marrs R, Meldrum D, Muasher S, Schoolcraft W, Werlin L, Kelly E. Randomized trial to compare the effect of recombinant human FSH (follitropin alfa) with or without recombinant human LH in women undergoing assisted reproduction treatment. Reprod Biomed Online 2004;8:175-82.
Published online in Fertility and Sterility doi:10.1016/j.fertnstert.2012.04.049