Patient-tailored ovarian stimulation for in vitro fertilization Bart C. J. M. Fauser, M.D., Ph.D. Department of Reproductive Medicine and Gynecology, University Medical Center, Utrecht, the Netherlands
At present, much attention in medicine is being directed toward individualized or patient-tailored care with the use of novel tools such as biomarkers or genomics. Research tools different from randomized controlled trials, focusing on the heterogeneity of patients rather than the intervention per se, are required to develop this concept further. In infertility care, few examples of individualized approaches with the use of multivariate prediction models can be found, such as the prediction of spontaneous conception chances in infertile patients or regarding the medical treatment of anovulatory infertility. Few prospective studies have been published in recent years concerning individualized dosing based on response prediction for ovarian stimulation in IVF. Potentially much may be gained by such methods because at present ovarian response to stimulation varies greatly, with distinct implications for both efficacy and safety of IVF treatment. (Fertil SterilÒ 2017;108:585–91. Ó2017 by American Society for Reproductive Medicine.) Key Words: Ovarian stimulation, IVF, individualized care, patient centered care Discuss: You can discuss this article with its authors and with other ASRM members at https://www.fertstertdialog.com/users/ 16110-fertility-and-sterility/posts/19131-24396
C
urrently, many new developments are reported in personalized medicine—also referred to as individualized medicine, stratified medicine, prediction medicine, or precision medicine—where the main focus of attention is to tailor a given medical intervention as much as possible to the individual needs of the patient. All major medical journals now publish widely on this topic, emphasizing that strategies complementary to the current standard of evidence-based medicine, i.e., the randomized controlled trial (RCT), are required to develop the concept of individualized medicine further (1, 2). RCTs are performed in fairly homogeneous patient populations (dependent on the predefined inclusion and exclusion criteria), where usually two different interventions or drugs are compared
and patients are allocated by chance to one treatment or the other. Subsequently, the mean responses related to the predefined primary end point of the study of both interventions are compared and the response difference tested for significance. Therefore, the average response is the primary focus of interest, and interindividual response differences in each arm of the study are usually ignored. As an example, a well designed RCT compared the aromatase inhibitor letrozole versus the antiestrogen clomiphene citrate (CC) for ovulation induction in women with polycystic ovary syndrome (PCOS) (3). That trial concluded that letrozole is superior and should be considered the first-line drug of choice for ovulation induction in PCOS. However, the overall live birth
Received May 23, 2017; accepted August 10, 2017. B.C.J.M.F., during the past 5 years, has received fees and grant support from (in alphabetic order) Actavis/Watson/Uteron, Controversies in Obstetrics and Gynecology, Dutch Heart Foundation, Dutch Medical Research Counsel (ZonMW), Euroscreen/Ogeda, Ferring, London Womens Clinic, Merck Serono (GFI), Netherland Genomic Initiative, Ovascience, Pantharei Bioscience, Preglem/Gedeon Richter/Finox, Reproductive Biomedicine Online, Roche, Teva, and the World Health Organization. Reprint requests: Bart C.J.M. Fauser, M.D., Ph.D., Department of Reproductive Medicine and Gynecology, University Medical Center, Utrecht 3584 CX, The Netherlands (E-mail: b.c.fauser@ umcutrecht.nl). Fertility and Sterility® Vol. 108, No. 4, October 2017 0015-0282/$36.00 Copyright ©2017 American Society for Reproductive Medicine, Published by Elsevier Inc. http://dx.doi.org/10.1016/j.fertnstert.2017.08.016 VOL. 108 NO. 4 / OCTOBER 2017
rate in the clomiphene arm was very low, and a post hoc analysis revealed that the significant difference in outcomes was present only in the very obese subpopulation of women. So the question seems to be justified of how useful these findings are for less obese populations outside of the United States (4). In contrast, prospective cohort follow-up studies in heterogeneous patient populations exposed to a single medical intervention allow the assessment of potential associations between initial patient characteristics and treatment outcomes with the use of multivariate prediction analyses (5). Such models should first be validated in similar but independent patient populations. Modified treatment modalities based on patient characteristics can subsequently be developed and tested with the use of RCTs comparing standard versus individualized treatment approaches in such targeted patient populations. Especially in cancer and cardiovascular disease, so-called ‘‘companion diagnostics’’ are now being introduced into the clinic, where a drug is marketed along with a diagnostic tool (usually assessing a biomarker relevant for drug dose or type of drug) (6). 585
VIEWS AND REVIEWS
FIGURE 1
Nomogram predicting chances for ovulation and live birth for clomiphene ovulation induction in women with polycystic ovary syndrome, based on initial screening characteristics such as amenorrhoea/oligomenorrhoea, body mass index (BMI), free androgen index (FAI), and age (Imani et al., 2002 [15]). CI ¼ confidence interval. Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
In summary, in an RCT the intervention per se is the primary focus, with much less attention (if any) for the patient and the context where the intervention is applied. In contrast, in cohort follow-up studies also involving prediction analysis, the focus of attention is to assess which patient features or environmental factors influence response to a given intervention. These are two very different approaches that should be considered to be complementary.
EXAMPLES OF PATIENT-TAILORED INFERTILITY CARE Multivariate prediction models have been developed in the context of infertility care, especially related to assessing chances for a pregnancy without medical intervention in couples with unexplained infertility (referred to by some as subfertility). It is generally know that such a patient population is notoriously heterogeneous, and it would help patient care to be able to more accurately predict spontaneous pregnancy chances for a given couple. A model for individualized prediction has been constructed based on initial characteristics of the couple such as female age, duration of infertility, infertility being primary or secondary, sperm motility, and referral status (7, 8). Based on such an assessment, the most appropriate treatment option can be truly personalized, e.g., expectant management in couples with relatively high chances of spontaneous pregnancy, mild intervention, such as sperm insemination, for 586
intermediate-prognosis couples, or in vitro fertilization (IVF) in cases of poor prognosis. In IVF, implantation rates of a transferred embryo of 50% represent current practice. Next to embryo quality, endometrial receptivity may be the most determining factor in failed embryo implantation. In recent years specific gene expression profiles in endometrial tissue (9), along with a distinct vaginal microbiome profile (10), have been reported to be linked to nonreceptive endometrium and therefore poor IVF pregnancy rates. Prediction models can also be used concerning ovarian response to exogenous stimulation, in the context of both ovulation induction in women diagnosed with PCOS and ovarian stimulation for IVF. Ovarian stimulation for IVF is usually referred to as controlled ovarian stimulation (COS) or controlled ovarian hyperstimulation (COH) (11). This term really represents wishful thinking, because reality teaches us that ovarian response to standard stimulation under those circumstances is far from controlled. Accordingly, this term has been modified to ‘‘ovarian stimulation’’ without the preface ‘‘controlled’’ in the updated glossary of the International Committee for Monitoring Assisted Reproductive Technology. Instead, we are often taken by surprise by a response either too low (with distinct implication for efficacy of treatment) or too high (with major implications for treatment safety), and we aim to counterbalance this undesired response by changing the dose during stimulation with no proven effect on outcome. VOL. 108 NO. 4 / OCTOBER 2017
Fertility and Sterility® Potentially there would be much to be gained by knowing the presumed ovarian response before we start stimulation, allowing adjustment of the stimulation dose (or type of drug) based on the individual needs of the woman, and for this—again—we need multivariate prediction-model cohort follow-up studies (12).
INDIVIDUALIZED OVULATION INDUCTION IN PCOS Clomiphene is usually applied as first-line treatment for ovulation induction in PCOS. This drug, clinically available since the early 1960s, is still the most prescribed fertility drug although the mode of action is not fully elucidated. Costs are low and side-effects are mild, but at the expense of limited efficacy. Overall 60%–70% ovulation rates with live birth rates of 20%–40% are reported for women who start treatment. During the 1990s, we aimed to address the question whether it would be possible with the use of multivariate prediction models to identify patient features associated with chances for ovulation (13) or pregnancy (14) for CC ovulation induction. We eventually developed a nomogram—combining predictors for chances to ovulate and then chances to get pregnant once ovulating—assessing chances for a live birth after clomiphene ovulation induction for a given woman with PCOS, based on initial screening characteristics such as amenorrhea or oligomenorrhea, body mass index, free androgen index (testosterone 100/SHBG), and female age (Fig. 1) (15). Our findings were subsequently confirmed in a post hoc analysis of a multicenter trial in 626 PCOS patients from the United States, where findings similar to ours were reported (16, 17). Rather than performing head-on comparative RCTs of different first-line interventions for ovulation induction, our observations allow the design of RCTs in a subgroup of women with predicted poor chances for pregnancy with the use of clomiphene. This would save precious time by skipping a full year of low-efficacy intervention in poor-prognosis women.
of individual patient data convincingly demonstrated the superiority of AMH (and AFC) in predicting both hyporesponse (arbitrarily defined as the retrieval of fewer than five oocytes) (21) or excessive response (defined as >15 oocytes) (22) (Fig. 2). The challenge is how to design a clinically useful dosing algorithm, based on a single or a combination of initial screening parameters, and how to test prospectively the potential merits of individualized dosing. For such an approach the desired outcome should first be defined in terms of the most appropriate ovarian response, clinical outcomes, cost, and complications. Individualized dosing approaches should be designed for both conventional (11) and mild-stimulation (23) IVF. Table 1 provides a brief summary of the few RCTs performed so far evaluating the clinical utility of individualized dosing for ovarian stimulation in IVF. A single multicenter Dutch trial (25) involving individualized dosing of exogenous FSH based on initial AFC and using cumulative live birth rates as the primary end point has just
FIGURE 2
INDIVIDUALIZED OVARIAN STIMULATION FOR IVF It has been known for many years that some factors are associated with reduced ovarian response to stimulation by exogenous FSH and poor IVF outcomes. The most commonly known factor is female age, later substituted by early follicular-phase serum FSH concentrations. Other measures subsequently developed to assess ovarian reserve include inhibin B levels and all sorts of ovarian challenge tests (18). In more recent years the focus of attention has shifted toward the antral follicle count (AFC) in ovaries assessed by means of transvaginal ultrasound, along with serum assays of antim€ ullerian hormone (AMH) closely associated with preantral and early-antral follicle development and ovarian aging (19, 20). Numerous studies over the past 10 years have clearly established an association between pretreatment AMH concentrations and ovarian response to standard stimulation. The most definitive meta analysis VOL. 108 NO. 4 / OCTOBER 2017
Receiver operating characteristic curves based on meta-analysis of individual patient data, regarding the capacity of patient characteristics to predict ovarian response to exogenous stimulation for in vitro fertilization: (A) poor response (defined as 15 oocytes retrieved) involving 57 studies and 4,786 women (Broer er al., 2013 [22]). €llerian hormone. AFC ¼ antral follicle count; AMH ¼ antimu Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
587
VIEWS AND REVIEWS
TABLE 1 Summary of RCTs evaluating individualized dosing for ovarian stimulation for IVF. Publication
Study design
Algorithms tested
Popovic, 2003 (24)
RCT, single center; n ¼ 262
van Tilborg, 2012 (25)
RCT, multicenter; n >1,300
Olivennes, 2015 (26)
RCT, multicenter; n ¼ 200
Nyboe Andersen, 2017 (27)
RCT, multicenter; n ¼ 1,329
Individualized dosing (based on nomogram involving AFC, ovarian volume, Doppler, age, and smoking; 100–250 IU/d) vs. standard (150 IU/d) in standard IVF patients Individualized dose (based on AFC) tested separately in predicted poor (225–400 IU/d) and hyperresponders (100–150 IU/d) vs. standard 150 IU/d Consort calculator (age, height, weight, FSH, AFC; 125–450 IU/d), vs. ‘‘standard’’ (150 IU/d) in standard IVF patients Individualized (AMH, body weight; 0.10 mg/kg/d– 12 mg/d) vs. standard 150– 450 IU/d
Allegra, 2017 (28)
RCT, single center; n ¼ 191
Individualized dosing (nomogram involving age, FSH, AMH; 75–225 IU/d) vs. age based standard dose (150–225 IU/d)
Conclusion concerning individualized dosing Increased proportion of appropriate response Decreased need for dose adjustment Higher ongoing pregnancy rate Cumulative live birth rate as primary study outcome No results published as yet
Lower daily and total FSH doses Fewer oocyte retrieved Same clinical pregnancy rate Less OHSS More targeted ovarian response Less poor response Less excessive response Less OHSS Same oocyte number Same ongoing pregnancy rate More often optimal oocyte number retrieved Same clinical pregnancy rate
Note: AFC ¼ antral follicle count; AMH ¼ antim€ ullerian hormone; OHSS ¼ ovarian hyperstimulation syndrome; RCT ¼ randomized controlled trial. Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
been completed (Broekmans, personal communications), but results have not yet been published. That study randomized a total of >1,500 women undergoing IVF to receiving either 100 or 150 IU/d FSH in presumed high responders and either 225 or 450 IU/d in presumed low responders based on initial AFC. Another recently published RCT, in 191 women in their first IVF/ICSI cycle in a single center (28), compared the efficacy of different FSH staring doses according to a nomogram using multiple features such as female age and FSH and AMH levels (Fig. 3) compared with FSH dosing based on female age only. The total dose of FSH administered was similar in both arms of the trial. An optimal ovarian response (arbitrarily defined as 8–14 oocytes) was observed more often in the nomogram arm of the study (63% vs. 42%; P¼ .004). A new recombinant FSH was tested with the use of individualized dosing (based on initial AMH levels and body weight) compared with 150 IU/d FSH starting dose in the standard arm in a multicenter (involving major IVF centers in different countries) RCT involving a total of 1,329 women undergoing IVF (27). Live birth rates were similar (30% vs. 31%), but individualized dosing resulted in higher chances for targeted ovarian response (also defined as 8–14 oocytes) (43% vs. 38%; Fig. 4), fewer women with poor response (12% vs. 18%), fewer with excessive response 588
(28% vs. 35%), and finally fewer measures needed to prevent ovarian hyperstimulation syndrome (OHSS) (2.3% vs. 4.5%).
CONCLUSION In several areas in medicine (especially oncology), individualized or patient-tailored strategies have been implemented in everyday clinical care. Treatment modality and type or dose of drug are modified based on specific patient characteristics. Although several patient-tailored strategies have been developed, they are not yet widely implemented in clinical infertility management. A major challenge in IVF treatment represents the unpredictable ovarian response to standard stimulation. We have to face the fact that it is impossible to design a standard protocol—sophisticated as it may be—that works for every woman. On the contrary, we are often confronted with suboptimal ovarian response, either too low (with implications for pregnancy chances) or too high (affecting safety of treatment). Clinical investigators should first reach an agreement regarding the optimal outcome of ovarian stimulation in terms of number of oocytes being retrieved. This optimal oocyte number should be viewed in the context of burden of treatment (and related drop-out rates affecting cumulative VOL. 108 NO. 4 / OCTOBER 2017
Fertility and Sterility®
FIGURE 3
€ llerian hormone (AMH) concentrations, Nomogram for calculating individualized FSH starting dose based on female age and FSH and antimu subsequently tested in a randomized controlled trial compared with an age-based standard FSH dose (Allegra et al., 2017 [28]). Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
pregnancy chances involving multiple IVF cycles), cost (and related access to IVF care), chances for complications (especially OHSS), and IVF success rates in terms of healthy children born. The optimal oocyte number is still heavily debated, but some recent studies aimed for 8–14 oocytes, whereas an increasing number of publications concerning mild ovarian stimulation aim for 3–8 oocytes. Many well designed clinical trials have demonstrated in recent years that more oocytes do not necessarily result in increased pregnancy rates. Rather than being taken by surprise by individual ovarian response differences to standard ovarian stimulation, it seems that much may be gained by assessing the extent of ovarian response before the start of stimulation. In this way, ovarian stimulation protocols can be truly individualized by altering the type or dose of medication in relation to specific patient features. The predicted ovarian response is to a great extent VOL. 108 NO. 4 / OCTOBER 2017
related to the ovarian reserve status of a given woman. The ovarian follicle pool slowly decreases with increasing age, but major individual variability also occurs. This difference is also demonstrated by a distinct individual variability in age of menopause, which is closely associated with follicle pool exhaustion. Next to chronologic age, the most predominant recently identified and fairly robust markers for ovarian aging (and resulting extent of ovarian response to stimulation) include AFC and AMH. Altogether, five RCTs (involving a total of >3,280 women undergoing IVF) prospectively assessed the usefulness of individualized (based on a variety of ovarian reserve biomarkers and other patient characteristics) versus standard dosing. It may be concluded at this point in time that subtle but distinct advantages of individualized dosing have been disclosed by studies undertaken so far in terms of safety and cost, with similar pregnancy rates. 589
VIEWS AND REVIEWS
FIGURE 4
3.
4. 5.
6. 7.
8. 9.
10.
11. 12.
13.
14.
15.
Data obtained from a multicenter randomized controlled trial in 1,329 women comparing individualized and conventional FSH dosing (Nyboe Andersen et al., 2017 [27]). (A) Proportion of women achieving the targeted number of oocytes (8–14), €llerian hormone comparing both study arms in relation to antimu (AMH) levels. (B) Proportion of women requiring ovarian hyperstimulation syndrome (OHSS)–preventive interventions and or experiencing OHSS in relation to AMH levels.
16.
17.
Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
18.
Much additional work needs to be done to identify the optimal ovarian response to stimulation, identify the most clinically relevant combination of ovarian response markers along with patient characteristics, develop a robust dosing algorithm (with the use of multivariate prediction analysis), and perform large-sample-size RCTs with the use of all relevant end points, such as burden of treatment, cost, patient preference, and cumulative chances for a healthy child.
19.
20.
21.
22.
REFERENCES 1. 2.
590
Mega JL, Sabatine MS, Antman EM. Population and personalized medicine in the modern era. JAMA 2014;312:1969–70. Bothwell LE, Greene JA, Podolsky SH, Jones DS. Assessing the gold standard—lessons from the history of RCTs. N Engl J Med 2016;374:2175– 81.
23.
Legro RS, Brzyski RG, Diamond MP, Coutifaris C, Schlaff WD, Casson P, et al, NICHD Reproductive Medicine Network. Letrozole versus clomiphene for infertility in the polycystic ovary syndrome. N Engl J Med 2014;371:119–29. Fauser BC. Reproductive endocrinology: revisiting ovulation induction in PCOS. Nat Rev Endocrinol 2014;12:704–5. Moons KG, Altman DG, Reitsma JB, Ioannidis JP, Macaskill P, Steyerberg EW, et al. Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015;162:W1–73. Twomey JD, Brahme NN, Zhang B. Drug-biomarker co-development in oncology—20 years and counting. Drug Resist Updat 2017;30:48–62. Hunault CC, Laven JS, van Rooij IA, Eijkemans MJ, te Velde ER, Habbema JD. Prospective validation of two models predicting pregnancy leading to live birth among untreated subfertile couples. Hum Reprod 2005;20:1636–41. Habbema JD, Eijkemans MJ, Leridon H, te Velde ER. Realizing a desired family size: when should couples start? Hum Reprod 2015;30:2215–21. Koot YE, van Hooff SR, Boomsma CM, van Leenen D, Groot Koerkamp MJ, Goddijn M, et al. An endometrial gene expression signature accurately predicts recurrent implantation failure after IVF. Sci Rep 2016;6:19411. Haahr T, Jensen JS, Thomsen L, Duus L, Rygaard K, Humaidan P. Abnormal vaginal microbiota may be associated with poor reproductive outcomes: a prospective study in IVF patients. Hum Reprod 2016;31: 795–803. Macklon NS, Stouffer RL, Giudice LC, Fauser BC. The science behind 25 years of ovarian stimulation for in vitro fertilization. Endocr Rev 2006;27:170–207. Fauser BC, Diedrich K, Devroey P, Evian Annual Reproduction Workshop Group 2007. Predictors of ovarian response: progress toward individualized treatment in ovulation induction and ovarian stimulation. Hum Reprod Update 2008;14:1–14. Imani B, Eijkemans MJ, te Velde ER, Habbema JD, Fauser BC. Predictors of patients remaining anovulatory during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. J Clin Endocrinol Metab 1998;83:2361–5. Imani B, Eijkemans MJ, te Velde ER, Habbema JD, Fauser BC. Predictors of chances to conceive in ovulatory patients during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. J Clin Endocrinol Metab 1999;84:1617–22. Imani B, Eijkemans MJ, te Velde ER, Habbema JD, Fauser BC. A nomogram to predict the probability of live birth after clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. Fertil Steril 2002;77:91–7. Rausch ME, Legro RS, Barnhart HX, Schlaff WD, Carr BR, Diamond MP, et al, Reproductive Medicine Network. Predictors of pregnancy in women with polycystic ovary syndrome. J Clin Endocrinol Metab 2009;94:3458–66. Fauser BC, Eijkemans MJ. Predicting pregnancy in women with polycystic ovary syndrome. J Clin Endocrinol Metab 2009;94:3183–4. Broekmans FJ, Soules MR, Fauser BC. Ovarian aging: mechanisms and clinical consequences. Endocr Rev 2009;30:465–93. la Marca A, Sunkara SK. Individualization of controlled ovarian stimulation in IVF using ovarian reserve markers: from theory to practice. Hum Reprod Update 2014;20:124–40. Broer SL, Broekmans FJ, Laven JS, Fauser BC. Anti-m€ ullerian hormone: ovarian reserve testing and its potential clinical implications. Hum Reprod Update 2014;20:688–701. Broer SL, van Disseldorp J, Broeze KA, Dolleman M, Opmeer BC, Bossuyt P, et al, IMPORT Study Group. Added value of ovarian reserve testing on patient characteristics in the prediction of ovarian response and ongoing pregnancy: an individual patient data approach. Hum Reprod Update 2013;19:26–36. lleman M, van Disseldorp J, Broeze KA, Opmeer BC, Bossuyt PM, Broer SL, Do et al, IPD-EXPORT Study Group. Prediction of an excessive response in in vitro fertilization from patient characteristics and ovarian reserve tests and comparison in subgroups: an individual patient data meta-analysis. Fertil Steril 2013;100:420–9.e7. Nargund G, Datta A, Fauser BC. Mild stimulation for in vitro fertilization. Fertil Steril 2017;108:558–67.
VOL. 108 NO. 4 / OCTOBER 2017
Fertility and Sterility® 24.
25.
26.
Popovic-Todorovi B, Loft A, Bredkjaeer HE, Bangsbøll S, Nielsen IK, Nyboe Andersen A. A prospective randomized clinical trial comparing an individual dose of recombinant FSH based on predictive factors versus a ‘‘standard’’ dose of 150 IU/day in ‘‘standard’’ patients undergoing IVF/ICSI treatment. Hum Reprod 2003;18:2275–82. van Tilborg TC, Eijkemans MJ, Laven JS, Koks CA, de Bruin JP, Scheffer GJ, et al. The OPTIMIST study: Optimisation of Cost Effectiveness Through Individualised FSH Stimulation Dosages for IVF Treatment. A randomised controlled trial. BMC Womens Health 2012;12:29. Olivennes F, Trew G, Borini A, Broekmans F, Arriagada P, Warne DW, et al. Randomized, controlled, open-label, noninferiority study of the CONSORT
VOL. 108 NO. 4 / OCTOBER 2017
27.
28.
algorithm for individualized dosing of follitropin alfa. Reprod Biomed Online 2015;30:248–57. Nyboe Andersen A, Nelson SM, Fauser BC, García-Velasco JA, Klein BM, Arce JC, ESTHER-1 Study Group. Individualized versus conventional ovarian stimulation for in vitro fertilization: a multicenter, randomized, controlled, assessor-blinded, phase 3 noninferiority trial. Fertil Steril 2017;107:387–96. Allegra A, Marino A, Volpes A, Coffaro F, Scaglione P, Gullo S, et al. A randomized controlled trial investigating the use of a predictive nomogram for the selection of the FSH starting dose in IVF/ICSI cycles. Reprod Biomed Online 2017;34:429–38.
591
At present, much attention in medicine is being directed toward individualized or patient-tailored care with the use of novel tools such as biomarkers or genomics. Research tools different from randomized controlled trials, focusing on the heterogeneity of patients rather than the intervention per se, are required to develop this concept further. In infertility care, few examples of individualized approaches with the use of multivariate prediction models can be found, such as the prediction of spontaneous conception chances in infertile patients or regarding the medical treatment of anovulatory infertility. Few prospective studies have been published in recent years concerning individualized dosing based on response prediction for ovarian stimulation in IVF. Potentially much may be gained by such methods because at present ovarian response to stimulation varies greatly, with distinct implications for both efficacy and safety of IVF treatment. (Fertil SterilÒ 2017;108:585–91. Ó2017 by American Society for Reproductive Medicine.) Key Words: Ovarian stimulation, IVF, individualized care, patient centered care Discuss: You can discuss this article with its authors and with other ASRM members at https://www.fertstertdialog.com/users/ 16110-fertility-and-sterility/posts/19131-24396
C
urrently, many new developments are reported in personalized medicine—also referred to as individualized medicine, stratified medicine, prediction medicine, or precision medicine—where the main focus of attention is to tailor a given medical intervention as much as possible to the individual needs of the patient. All major medical journals now publish widely on this topic, emphasizing that strategies complementary to the current standard of evidence-based medicine, i.e., the randomized controlled trial (RCT), are required to develop the concept of individualized medicine further (1, 2). RCTs are performed in fairly homogeneous patient populations (dependent on the predefined inclusion and exclusion criteria), where usually two different interventions or drugs are compared
and patients are allocated by chance to one treatment or the other. Subsequently, the mean responses related to the predefined primary end point of the study of both interventions are compared and the response difference tested for significance. Therefore, the average response is the primary focus of interest, and interindividual response differences in each arm of the study are usually ignored. As an example, a well designed RCT compared the aromatase inhibitor letrozole versus the antiestrogen clomiphene citrate (CC) for ovulation induction in women with polycystic ovary syndrome (PCOS) (3). That trial concluded that letrozole is superior and should be considered the first-line drug of choice for ovulation induction in PCOS. However, the overall live birth
Received May 23, 2017; accepted August 10, 2017. B.C.J.M.F., during the past 5 years, has received fees and grant support from (in alphabetic order) Actavis/Watson/Uteron, Controversies in Obstetrics and Gynecology, Dutch Heart Foundation, Dutch Medical Research Counsel (ZonMW), Euroscreen/Ogeda, Ferring, London Womens Clinic, Merck Serono (GFI), Netherland Genomic Initiative, Ovascience, Pantharei Bioscience, Preglem/Gedeon Richter/Finox, Reproductive Biomedicine Online, Roche, Teva, and the World Health Organization. Reprint requests: Bart C.J.M. Fauser, M.D., Ph.D., Department of Reproductive Medicine and Gynecology, University Medical Center, Utrecht 3584 CX, The Netherlands (E-mail: b.c.fauser@ umcutrecht.nl). Fertility and Sterility® Vol. 108, No. 4, October 2017 0015-0282/$36.00 Copyright ©2017 American Society for Reproductive Medicine, Published by Elsevier Inc. http://dx.doi.org/10.1016/j.fertnstert.2017.08.016 VOL. 108 NO. 4 / OCTOBER 2017
rate in the clomiphene arm was very low, and a post hoc analysis revealed that the significant difference in outcomes was present only in the very obese subpopulation of women. So the question seems to be justified of how useful these findings are for less obese populations outside of the United States (4). In contrast, prospective cohort follow-up studies in heterogeneous patient populations exposed to a single medical intervention allow the assessment of potential associations between initial patient characteristics and treatment outcomes with the use of multivariate prediction analyses (5). Such models should first be validated in similar but independent patient populations. Modified treatment modalities based on patient characteristics can subsequently be developed and tested with the use of RCTs comparing standard versus individualized treatment approaches in such targeted patient populations. Especially in cancer and cardiovascular disease, so-called ‘‘companion diagnostics’’ are now being introduced into the clinic, where a drug is marketed along with a diagnostic tool (usually assessing a biomarker relevant for drug dose or type of drug) (6). 585
VIEWS AND REVIEWS
FIGURE 1
Nomogram predicting chances for ovulation and live birth for clomiphene ovulation induction in women with polycystic ovary syndrome, based on initial screening characteristics such as amenorrhoea/oligomenorrhoea, body mass index (BMI), free androgen index (FAI), and age (Imani et al., 2002 [15]). CI ¼ confidence interval. Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
In summary, in an RCT the intervention per se is the primary focus, with much less attention (if any) for the patient and the context where the intervention is applied. In contrast, in cohort follow-up studies also involving prediction analysis, the focus of attention is to assess which patient features or environmental factors influence response to a given intervention. These are two very different approaches that should be considered to be complementary.
EXAMPLES OF PATIENT-TAILORED INFERTILITY CARE Multivariate prediction models have been developed in the context of infertility care, especially related to assessing chances for a pregnancy without medical intervention in couples with unexplained infertility (referred to by some as subfertility). It is generally know that such a patient population is notoriously heterogeneous, and it would help patient care to be able to more accurately predict spontaneous pregnancy chances for a given couple. A model for individualized prediction has been constructed based on initial characteristics of the couple such as female age, duration of infertility, infertility being primary or secondary, sperm motility, and referral status (7, 8). Based on such an assessment, the most appropriate treatment option can be truly personalized, e.g., expectant management in couples with relatively high chances of spontaneous pregnancy, mild intervention, such as sperm insemination, for 586
intermediate-prognosis couples, or in vitro fertilization (IVF) in cases of poor prognosis. In IVF, implantation rates of a transferred embryo of 50% represent current practice. Next to embryo quality, endometrial receptivity may be the most determining factor in failed embryo implantation. In recent years specific gene expression profiles in endometrial tissue (9), along with a distinct vaginal microbiome profile (10), have been reported to be linked to nonreceptive endometrium and therefore poor IVF pregnancy rates. Prediction models can also be used concerning ovarian response to exogenous stimulation, in the context of both ovulation induction in women diagnosed with PCOS and ovarian stimulation for IVF. Ovarian stimulation for IVF is usually referred to as controlled ovarian stimulation (COS) or controlled ovarian hyperstimulation (COH) (11). This term really represents wishful thinking, because reality teaches us that ovarian response to standard stimulation under those circumstances is far from controlled. Accordingly, this term has been modified to ‘‘ovarian stimulation’’ without the preface ‘‘controlled’’ in the updated glossary of the International Committee for Monitoring Assisted Reproductive Technology. Instead, we are often taken by surprise by a response either too low (with distinct implication for efficacy of treatment) or too high (with major implications for treatment safety), and we aim to counterbalance this undesired response by changing the dose during stimulation with no proven effect on outcome. VOL. 108 NO. 4 / OCTOBER 2017
Fertility and Sterility® Potentially there would be much to be gained by knowing the presumed ovarian response before we start stimulation, allowing adjustment of the stimulation dose (or type of drug) based on the individual needs of the woman, and for this—again—we need multivariate prediction-model cohort follow-up studies (12).
INDIVIDUALIZED OVULATION INDUCTION IN PCOS Clomiphene is usually applied as first-line treatment for ovulation induction in PCOS. This drug, clinically available since the early 1960s, is still the most prescribed fertility drug although the mode of action is not fully elucidated. Costs are low and side-effects are mild, but at the expense of limited efficacy. Overall 60%–70% ovulation rates with live birth rates of 20%–40% are reported for women who start treatment. During the 1990s, we aimed to address the question whether it would be possible with the use of multivariate prediction models to identify patient features associated with chances for ovulation (13) or pregnancy (14) for CC ovulation induction. We eventually developed a nomogram—combining predictors for chances to ovulate and then chances to get pregnant once ovulating—assessing chances for a live birth after clomiphene ovulation induction for a given woman with PCOS, based on initial screening characteristics such as amenorrhea or oligomenorrhea, body mass index, free androgen index (testosterone 100/SHBG), and female age (Fig. 1) (15). Our findings were subsequently confirmed in a post hoc analysis of a multicenter trial in 626 PCOS patients from the United States, where findings similar to ours were reported (16, 17). Rather than performing head-on comparative RCTs of different first-line interventions for ovulation induction, our observations allow the design of RCTs in a subgroup of women with predicted poor chances for pregnancy with the use of clomiphene. This would save precious time by skipping a full year of low-efficacy intervention in poor-prognosis women.
of individual patient data convincingly demonstrated the superiority of AMH (and AFC) in predicting both hyporesponse (arbitrarily defined as the retrieval of fewer than five oocytes) (21) or excessive response (defined as >15 oocytes) (22) (Fig. 2). The challenge is how to design a clinically useful dosing algorithm, based on a single or a combination of initial screening parameters, and how to test prospectively the potential merits of individualized dosing. For such an approach the desired outcome should first be defined in terms of the most appropriate ovarian response, clinical outcomes, cost, and complications. Individualized dosing approaches should be designed for both conventional (11) and mild-stimulation (23) IVF. Table 1 provides a brief summary of the few RCTs performed so far evaluating the clinical utility of individualized dosing for ovarian stimulation in IVF. A single multicenter Dutch trial (25) involving individualized dosing of exogenous FSH based on initial AFC and using cumulative live birth rates as the primary end point has just
FIGURE 2
INDIVIDUALIZED OVARIAN STIMULATION FOR IVF It has been known for many years that some factors are associated with reduced ovarian response to stimulation by exogenous FSH and poor IVF outcomes. The most commonly known factor is female age, later substituted by early follicular-phase serum FSH concentrations. Other measures subsequently developed to assess ovarian reserve include inhibin B levels and all sorts of ovarian challenge tests (18). In more recent years the focus of attention has shifted toward the antral follicle count (AFC) in ovaries assessed by means of transvaginal ultrasound, along with serum assays of antim€ ullerian hormone (AMH) closely associated with preantral and early-antral follicle development and ovarian aging (19, 20). Numerous studies over the past 10 years have clearly established an association between pretreatment AMH concentrations and ovarian response to standard stimulation. The most definitive meta analysis VOL. 108 NO. 4 / OCTOBER 2017
Receiver operating characteristic curves based on meta-analysis of individual patient data, regarding the capacity of patient characteristics to predict ovarian response to exogenous stimulation for in vitro fertilization: (A) poor response (defined as 15 oocytes retrieved) involving 57 studies and 4,786 women (Broer er al., 2013 [22]). €llerian hormone. AFC ¼ antral follicle count; AMH ¼ antimu Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
587
VIEWS AND REVIEWS
TABLE 1 Summary of RCTs evaluating individualized dosing for ovarian stimulation for IVF. Publication
Study design
Algorithms tested
Popovic, 2003 (24)
RCT, single center; n ¼ 262
van Tilborg, 2012 (25)
RCT, multicenter; n >1,300
Olivennes, 2015 (26)
RCT, multicenter; n ¼ 200
Nyboe Andersen, 2017 (27)
RCT, multicenter; n ¼ 1,329
Individualized dosing (based on nomogram involving AFC, ovarian volume, Doppler, age, and smoking; 100–250 IU/d) vs. standard (150 IU/d) in standard IVF patients Individualized dose (based on AFC) tested separately in predicted poor (225–400 IU/d) and hyperresponders (100–150 IU/d) vs. standard 150 IU/d Consort calculator (age, height, weight, FSH, AFC; 125–450 IU/d), vs. ‘‘standard’’ (150 IU/d) in standard IVF patients Individualized (AMH, body weight; 0.10 mg/kg/d– 12 mg/d) vs. standard 150– 450 IU/d
Allegra, 2017 (28)
RCT, single center; n ¼ 191
Individualized dosing (nomogram involving age, FSH, AMH; 75–225 IU/d) vs. age based standard dose (150–225 IU/d)
Conclusion concerning individualized dosing Increased proportion of appropriate response Decreased need for dose adjustment Higher ongoing pregnancy rate Cumulative live birth rate as primary study outcome No results published as yet
Lower daily and total FSH doses Fewer oocyte retrieved Same clinical pregnancy rate Less OHSS More targeted ovarian response Less poor response Less excessive response Less OHSS Same oocyte number Same ongoing pregnancy rate More often optimal oocyte number retrieved Same clinical pregnancy rate
Note: AFC ¼ antral follicle count; AMH ¼ antim€ ullerian hormone; OHSS ¼ ovarian hyperstimulation syndrome; RCT ¼ randomized controlled trial. Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
been completed (Broekmans, personal communications), but results have not yet been published. That study randomized a total of >1,500 women undergoing IVF to receiving either 100 or 150 IU/d FSH in presumed high responders and either 225 or 450 IU/d in presumed low responders based on initial AFC. Another recently published RCT, in 191 women in their first IVF/ICSI cycle in a single center (28), compared the efficacy of different FSH staring doses according to a nomogram using multiple features such as female age and FSH and AMH levels (Fig. 3) compared with FSH dosing based on female age only. The total dose of FSH administered was similar in both arms of the trial. An optimal ovarian response (arbitrarily defined as 8–14 oocytes) was observed more often in the nomogram arm of the study (63% vs. 42%; P¼ .004). A new recombinant FSH was tested with the use of individualized dosing (based on initial AMH levels and body weight) compared with 150 IU/d FSH starting dose in the standard arm in a multicenter (involving major IVF centers in different countries) RCT involving a total of 1,329 women undergoing IVF (27). Live birth rates were similar (30% vs. 31%), but individualized dosing resulted in higher chances for targeted ovarian response (also defined as 8–14 oocytes) (43% vs. 38%; Fig. 4), fewer women with poor response (12% vs. 18%), fewer with excessive response 588
(28% vs. 35%), and finally fewer measures needed to prevent ovarian hyperstimulation syndrome (OHSS) (2.3% vs. 4.5%).
CONCLUSION In several areas in medicine (especially oncology), individualized or patient-tailored strategies have been implemented in everyday clinical care. Treatment modality and type or dose of drug are modified based on specific patient characteristics. Although several patient-tailored strategies have been developed, they are not yet widely implemented in clinical infertility management. A major challenge in IVF treatment represents the unpredictable ovarian response to standard stimulation. We have to face the fact that it is impossible to design a standard protocol—sophisticated as it may be—that works for every woman. On the contrary, we are often confronted with suboptimal ovarian response, either too low (with implications for pregnancy chances) or too high (affecting safety of treatment). Clinical investigators should first reach an agreement regarding the optimal outcome of ovarian stimulation in terms of number of oocytes being retrieved. This optimal oocyte number should be viewed in the context of burden of treatment (and related drop-out rates affecting cumulative VOL. 108 NO. 4 / OCTOBER 2017
Fertility and Sterility®
FIGURE 3
€ llerian hormone (AMH) concentrations, Nomogram for calculating individualized FSH starting dose based on female age and FSH and antimu subsequently tested in a randomized controlled trial compared with an age-based standard FSH dose (Allegra et al., 2017 [28]). Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
pregnancy chances involving multiple IVF cycles), cost (and related access to IVF care), chances for complications (especially OHSS), and IVF success rates in terms of healthy children born. The optimal oocyte number is still heavily debated, but some recent studies aimed for 8–14 oocytes, whereas an increasing number of publications concerning mild ovarian stimulation aim for 3–8 oocytes. Many well designed clinical trials have demonstrated in recent years that more oocytes do not necessarily result in increased pregnancy rates. Rather than being taken by surprise by individual ovarian response differences to standard ovarian stimulation, it seems that much may be gained by assessing the extent of ovarian response before the start of stimulation. In this way, ovarian stimulation protocols can be truly individualized by altering the type or dose of medication in relation to specific patient features. The predicted ovarian response is to a great extent VOL. 108 NO. 4 / OCTOBER 2017
related to the ovarian reserve status of a given woman. The ovarian follicle pool slowly decreases with increasing age, but major individual variability also occurs. This difference is also demonstrated by a distinct individual variability in age of menopause, which is closely associated with follicle pool exhaustion. Next to chronologic age, the most predominant recently identified and fairly robust markers for ovarian aging (and resulting extent of ovarian response to stimulation) include AFC and AMH. Altogether, five RCTs (involving a total of >3,280 women undergoing IVF) prospectively assessed the usefulness of individualized (based on a variety of ovarian reserve biomarkers and other patient characteristics) versus standard dosing. It may be concluded at this point in time that subtle but distinct advantages of individualized dosing have been disclosed by studies undertaken so far in terms of safety and cost, with similar pregnancy rates. 589
VIEWS AND REVIEWS
FIGURE 4
3.
4. 5.
6. 7.
8. 9.
10.
11. 12.
13.
14.
15.
Data obtained from a multicenter randomized controlled trial in 1,329 women comparing individualized and conventional FSH dosing (Nyboe Andersen et al., 2017 [27]). (A) Proportion of women achieving the targeted number of oocytes (8–14), €llerian hormone comparing both study arms in relation to antimu (AMH) levels. (B) Proportion of women requiring ovarian hyperstimulation syndrome (OHSS)–preventive interventions and or experiencing OHSS in relation to AMH levels.
16.
17.
Fauser. Patient-tailored ovarian stimulation. Fertil Steril 2017.
18.
Much additional work needs to be done to identify the optimal ovarian response to stimulation, identify the most clinically relevant combination of ovarian response markers along with patient characteristics, develop a robust dosing algorithm (with the use of multivariate prediction analysis), and perform large-sample-size RCTs with the use of all relevant end points, such as burden of treatment, cost, patient preference, and cumulative chances for a healthy child.
19.
20.
21.
22.
REFERENCES 1. 2.
590
Mega JL, Sabatine MS, Antman EM. Population and personalized medicine in the modern era. JAMA 2014;312:1969–70. Bothwell LE, Greene JA, Podolsky SH, Jones DS. Assessing the gold standard—lessons from the history of RCTs. N Engl J Med 2016;374:2175– 81.
23.
Legro RS, Brzyski RG, Diamond MP, Coutifaris C, Schlaff WD, Casson P, et al, NICHD Reproductive Medicine Network. Letrozole versus clomiphene for infertility in the polycystic ovary syndrome. N Engl J Med 2014;371:119–29. Fauser BC. Reproductive endocrinology: revisiting ovulation induction in PCOS. Nat Rev Endocrinol 2014;12:704–5. Moons KG, Altman DG, Reitsma JB, Ioannidis JP, Macaskill P, Steyerberg EW, et al. Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015;162:W1–73. Twomey JD, Brahme NN, Zhang B. Drug-biomarker co-development in oncology—20 years and counting. Drug Resist Updat 2017;30:48–62. Hunault CC, Laven JS, van Rooij IA, Eijkemans MJ, te Velde ER, Habbema JD. Prospective validation of two models predicting pregnancy leading to live birth among untreated subfertile couples. Hum Reprod 2005;20:1636–41. Habbema JD, Eijkemans MJ, Leridon H, te Velde ER. Realizing a desired family size: when should couples start? Hum Reprod 2015;30:2215–21. Koot YE, van Hooff SR, Boomsma CM, van Leenen D, Groot Koerkamp MJ, Goddijn M, et al. An endometrial gene expression signature accurately predicts recurrent implantation failure after IVF. Sci Rep 2016;6:19411. Haahr T, Jensen JS, Thomsen L, Duus L, Rygaard K, Humaidan P. Abnormal vaginal microbiota may be associated with poor reproductive outcomes: a prospective study in IVF patients. Hum Reprod 2016;31: 795–803. Macklon NS, Stouffer RL, Giudice LC, Fauser BC. The science behind 25 years of ovarian stimulation for in vitro fertilization. Endocr Rev 2006;27:170–207. Fauser BC, Diedrich K, Devroey P, Evian Annual Reproduction Workshop Group 2007. Predictors of ovarian response: progress toward individualized treatment in ovulation induction and ovarian stimulation. Hum Reprod Update 2008;14:1–14. Imani B, Eijkemans MJ, te Velde ER, Habbema JD, Fauser BC. Predictors of patients remaining anovulatory during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. J Clin Endocrinol Metab 1998;83:2361–5. Imani B, Eijkemans MJ, te Velde ER, Habbema JD, Fauser BC. Predictors of chances to conceive in ovulatory patients during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. J Clin Endocrinol Metab 1999;84:1617–22. Imani B, Eijkemans MJ, te Velde ER, Habbema JD, Fauser BC. A nomogram to predict the probability of live birth after clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility. Fertil Steril 2002;77:91–7. Rausch ME, Legro RS, Barnhart HX, Schlaff WD, Carr BR, Diamond MP, et al, Reproductive Medicine Network. Predictors of pregnancy in women with polycystic ovary syndrome. J Clin Endocrinol Metab 2009;94:3458–66. Fauser BC, Eijkemans MJ. Predicting pregnancy in women with polycystic ovary syndrome. J Clin Endocrinol Metab 2009;94:3183–4. Broekmans FJ, Soules MR, Fauser BC. Ovarian aging: mechanisms and clinical consequences. Endocr Rev 2009;30:465–93. la Marca A, Sunkara SK. Individualization of controlled ovarian stimulation in IVF using ovarian reserve markers: from theory to practice. Hum Reprod Update 2014;20:124–40. Broer SL, Broekmans FJ, Laven JS, Fauser BC. Anti-m€ ullerian hormone: ovarian reserve testing and its potential clinical implications. Hum Reprod Update 2014;20:688–701. Broer SL, van Disseldorp J, Broeze KA, Dolleman M, Opmeer BC, Bossuyt P, et al, IMPORT Study Group. Added value of ovarian reserve testing on patient characteristics in the prediction of ovarian response and ongoing pregnancy: an individual patient data approach. Hum Reprod Update 2013;19:26–36. lleman M, van Disseldorp J, Broeze KA, Opmeer BC, Bossuyt PM, Broer SL, Do et al, IPD-EXPORT Study Group. Prediction of an excessive response in in vitro fertilization from patient characteristics and ovarian reserve tests and comparison in subgroups: an individual patient data meta-analysis. Fertil Steril 2013;100:420–9.e7. Nargund G, Datta A, Fauser BC. Mild stimulation for in vitro fertilization. Fertil Steril 2017;108:558–67.
VOL. 108 NO. 4 / OCTOBER 2017
Fertility and Sterility® 24.
25.
26.
Popovic-Todorovi B, Loft A, Bredkjaeer HE, Bangsbøll S, Nielsen IK, Nyboe Andersen A. A prospective randomized clinical trial comparing an individual dose of recombinant FSH based on predictive factors versus a ‘‘standard’’ dose of 150 IU/day in ‘‘standard’’ patients undergoing IVF/ICSI treatment. Hum Reprod 2003;18:2275–82. van Tilborg TC, Eijkemans MJ, Laven JS, Koks CA, de Bruin JP, Scheffer GJ, et al. The OPTIMIST study: Optimisation of Cost Effectiveness Through Individualised FSH Stimulation Dosages for IVF Treatment. A randomised controlled trial. BMC Womens Health 2012;12:29. Olivennes F, Trew G, Borini A, Broekmans F, Arriagada P, Warne DW, et al. Randomized, controlled, open-label, noninferiority study of the CONSORT
VOL. 108 NO. 4 / OCTOBER 2017
27.
28.
algorithm for individualized dosing of follitropin alfa. Reprod Biomed Online 2015;30:248–57. Nyboe Andersen A, Nelson SM, Fauser BC, García-Velasco JA, Klein BM, Arce JC, ESTHER-1 Study Group. Individualized versus conventional ovarian stimulation for in vitro fertilization: a multicenter, randomized, controlled, assessor-blinded, phase 3 noninferiority trial. Fertil Steril 2017;107:387–96. Allegra A, Marino A, Volpes A, Coffaro F, Scaglione P, Gullo S, et al. A randomized controlled trial investigating the use of a predictive nomogram for the selection of the FSH starting dose in IVF/ICSI cycles. Reprod Biomed Online 2017;34:429–38.
591
