Reducing the risks associated with in vitro fertilization
Richard H. Reindollar, MD
Owen K. Davis, MD
Renee H. Martin, MD

Richard H. Reindollar, MD

I'd like to welcome all of you today to this symposium entitled, “Reducing the Risks Associated with In Vitro Fertilization (IVF)”. I'm Richard Reindollar, the moderator, and I'm from Dartmouth Medical School. And with me I have Dr. Renee Martin from the University of Calgary and Dr. Owen Davis from Weill Cornell Medical Center. As we all know, many thousands of couples each year are helped with the assistance of assisted reproductive technology in vitro fertilization. And our goal, of course, is to increase the pregnancy rate so that as many of them can achieve pregnancy as possible. But the at the same time our goals are to reduce the risks associated with IVF. So today we're going to address three areas of risks associated with IVF.

Our learning objectives are here. We're going to begin first with Dr. Owen Davis who will talk to us about the serious and common risks, including ovarian hyperstimulation syndrome (OHSS) and multiple births. I will then follow with a discussion about the serious and fortunately uncommon risks and primarily I'll talk for a few moments about cancer and spend a good bit of my time on the risk of death from pregnancy and women with Turner syndrome. And then Dr. Martin will finish by talking to us today about the genetic risks associated with ART.

Our learning objectives are here. We're going to begin first with Dr. Owen Davis who will talk to us about the serious and common risks, including ovarian hyperstimulation syndrome (OHSS) and multiple births. I will then follow with a discussion about the serious and fortunately uncommon risks and primarily I'll talk for a few moments about cancer and spend a good bit of my time on the risk of death from pregnancy and women with Turner syndrome. And then Dr. Martin will finish by talking to us today about the genetic risks associated with ART.

There may be discussion about off-label or otherwise non-approved use of products or devices.

These are our faculty disclosures.

And so with no further adieu we'll begin with our first discussion by Dr. Owen Davis on reducing the risks of OHSS and multiple births in IVF. I'd also like to recognize the fact that this program has been made possible by an educational grant from Schering-Plough.

Owen K. Davis, MD

I'd like to start by thanking Dr. Reindollar and the organizers and sponsors of this symposium for inviting me to participate.

With IVF as with medicine in general there's a delicate balance between risk and benefit. And as a physician one of our primary responsibilities to our patients is to optimize this balance. I've been asked in this segment of the symposium to talk about two widely accepted risks of assisted reproductive technology (ART) and to talk about procedures and strategies to minimize these risks, namely ovarian hyperstimulation syndrome and multiple gestation.

Starting with OHSS, this indeed a serious and potentially life-threatening complication of ovulation induction and of ART and this can have very severe short-term and long-term sequelae.

It is characterized by ovarian enlargement, which can be massive and by third spacing, which can also be massive due to vascular permeability resulting in extravasation of fluid out of the intravascular space.

And I think most people in this room who practice ART or ovulation induction are familiar with sonograph images such as this one showing an enlarged multi-cystic ovary floating in acidic fluid

Now in my opinion there really aren't any entirely satisfactory classification or grading systems for OHSS. Rather I view it as sort of a continuum for the most mild forms, which are essentially clinically trivial up to severe and critical forms. In it's mildest form it is characterized by simply by enlargement of the ovaries and possibly minor amounts of increased peritoneal fluid and certainly high sex steroid levels peripherally. In its more moderate form there can be more extensive acidities, more massive enlargement of the ovaries and overt patient symptomatology such as abdominal discomfort or pain, nausea, vomiting and diarrhea. As it progresses to its most severe or critical form, OHSS can lead to intense ascites, hydrothorax, dysemia and the severe depletion of the intravascular compartment can lead to hemoconcentration, increasing the risk of both arterial and venous thrombosis, renal dysfunction and electrolyte imbalance.

Now a number of reports have been reviewed to look at the incidence of OHSS and I think this is probably pretty representative. Mild to moderate forms were seen in about 20% to 30% of IVF. Clearly relevant OHSS, which will nonetheless be managed as an outpatient, perhaps 10%. And the most severe forms that require active intervention and/or hospitalization perhaps .5% up to 4%, depending upon who you read.

Now there are a number of strategies for prevention because the key to treating OHSS is, of course, prevention, management is only supportive. First and foremost identify the patient at risk. If you have the luxury of previous stimulation history with the patient that's going to be your most valuable tool. But in some cases you're confronted with a treatment naïve patient. The gentlest stimulation that does the trick should always be employed in ART. One has to be willing to make a decision to cancel a cycle altogether and withhold hCG in cycles that are deemed to be an excessive risk and try again in a subsequent month. I'll talk a little bit about coasting. Reducing the hCG dose for trigger is reasonable rather than the conventional 10,000 units one can readily give 5,000 units or 4,000 or even 3,300 units with satisfactory oocyte maturation. And finally I'll talk a little bit about the use of a GnRH agonist trigger, which is an unique option in antagonist cycles.

Now, how can we identify a patient who is a priori at risk even if they are treatment naïve? Well these are a few characteristics. First of all youth is a risk factor, so young age. Secondly, the appearance of the ovaries, polycystic type ovaries, whether being a patient with a full-blown syndrome or someone who simply has PCO-like ovaries is a risk factor. And finally a lean habitus or a low BMI is a risk.

There are also risk factors that can be identified during the progression of a stimulation cycle. Certain high or very high serum estradiol levels are a risk. A very high rate of rise of the serum estradiol. The presence of numerous, small and intermediate follicle, particularly on the day of hCG. The occurrence of conception. Pregnancy because of endogenous hCG increases the incidence, severity and the duration of OHSS and similarly Luteal phase hCG support is a risk.

This is a table from a paper by Riccardo Asch from several years ago where he looked at absolute estradiol cutoffs in the incidence of severe OHSS. In this series they found no cases with estradiols of under 3500. 1.5% severe OHSS between 3500 and about 6000 and above 6000 a clearly unacceptable risk of 38%.

Notably, in women who had estradiol levels that were above 6000 in the setting of the recovery of 30 or more oocytes the rate of severe OHSS was 80%.

Now in addition to absolute estradiol level the number of oocytes recovered, one could also look at the size and number of follicles as a risk factor. I don't think there's really any general consensus at this point as to what constitutes a critical number or follicular size but most reports would say somewhere between 15 and 30 follicles ranging between about 12 and 15 millimeters.

Now one of the most important ways to try to prevent OHSS is to carefully choose and tell your stimulation protocol to the individual patient. If a patient has had a history of being an excessive responder or has these other characteristics that put her at risk, one should always choose the lowest effective dose of gonadotropins, typically between 100 and 150 units a day to start. We have found in PCO and PCO-like patients that the utilization of a dual suppression strategy where you pre-suppress with a birth control pill overlapping with a GnRH agonist such as leuprolide acetate and then using a low-dose of stimulation has been very helpful. Similarly this can be done with an antagonist. And finally, again, in antagonist cycles you have the unique opportunity to use an agonist as your ovulatory trigger.

Now when a patient's cycle is underway and she appears to be having an overly exuberant response, I think it's always wise to give the lower hCG dose, 4,000, 5,000 units. Coasting can be undertaken. I'll be talking about this in a little bit but basically withholding gonadotropins for a few days while the estradiol drifts down to a more acceptable level. One can retrieve and then freeze all embryos to obviate the possibility of conception in the treatment cycle. And finally, borderline cases, one can actually observe the patient's clinical progression from retrieval to day 5 and freeze the embryos if the risk is evidently too high.

Now this schematic summarizes our dual suppression protocol, which we actually described several years ago. This was designed initially for polycystic ovary patients who were a very difficult bunch to treat. Typically high cancellation rates, high rates of OHSS and a fairly high yield of immature oocytes. When you pretreat with 28 days of a typically monophasic OCP, overlap it for seven days with an agonist such as leuprolide and then use a low-dose stimulation. We find that this very readily attenuates and dampens the response. Result in low cancellation rate, low rates of OHSS and very satisfactory pregnancy rates.

A similar approach can be taken overlapping the birth control pills with an antagonist and using a low-dose stimulation approach.

Now I'm going to talk a little bit about coasting. This was initially described by a number of groups many years ago. I think this is a pretty wide practice around the world. Basically what it consists of is withholding gonadotropins for one up to several days while continuing the GnRH agonist to allow the estradiol eventually to drop t a level at which it is safe to trigger with hCG. Now coasting is not going to be effective if it is started too early in the cycle. You really have to have your lead cohort of follicles, achieve a mean diameter of about 13 to 14 millimeters for this to work out.

Now this is a graph from a paper that was written by Claudio Benadiva, who at the time was a fellow at Cornell. This shows the typical estradiol pattern during the coasting interval. Initially as you continue the agonist, the estradiol will rise for a day or two and hopefully within the first few days it will drop to a level at which you could trigger.

Now in this paper, coasting was compared in one group of patients to a group in which cryopreservation up front was undertaken. In both groups, the peak E2 was achieved during stimulation was around 4000. Now when the coasting group, which is shown in the left column, they coasted down to an average estradiol level of 2200 for the day of trigger. In the cryopreserve group they were trigged at 4300. Now one thing that you will see consistently with prolonged coasting especially is a marked reduction in the number of oocytes that are retrieved and I think this is simply due to the elimination of the intermediate size follicles. So you'll typically get about half the number of oocytes that you will in a non-coasted cycle, but still an acceptable numbers. Cancellation rate is still about 22% in the coasting group. But the delivery rate per transfer was a very acceptable 59%.

Now can you coast for too long? I think this is an area of some debate. This is a paper that was published in 2002 that showed that you really don't have any decrement in the implantation or pregnancy rates even if you coast for up to three days. But this study showed that at four days or more of coasting that the pregnancy rate and implantation rates fell by about a half. This is somewhat controversial. There are actually some presentations at this annual meeting that suggest that up to six days actually is as effective.

Now antagonists are actually very well suited to prevention of OHSS for a double reason. First antagonist cycles intrinsically have a somewhat rate of low OHSS than conventional agonist cycles do.

And secondly you can use a GnRH such as Leuprolide as your trigger to affect an endogenous LH surge, affect oocyte maturation without having a long half-life of exogenous hCG and this is shown schematically here.

This is a study that was published this year comparing patients undergoing an antagonist protocol with an agonist trigger using Leuprolide on the left column and a conventional agonist based protocol using hCG as the trigger on the right. Both groups were relative high responders. They were triggered at a mean E2 of about 2600. Had comparable numbers of oocytes retrieved, similar number of embryos transferred. What was very striking in this study was that the investigators brought the patients back in their mid-luteal phase for a reassessment and they found a very striking reduction in the volume of the ovaries when the GnRH agonist was used to trigger averaging 36 cubic centimeters compared to 129 in the group that had the hCG trigger.

And importantly, although it was a small study, there were no cases of OHSS in the GnRH agonist group, whereas there was about a 34% incidence with the hCG trigger more than half of which were deemed to be more than trivial. Ongoing pregnancy rate was good and acceptable in both groups. I don't have a lot of time to go into this but you do have to very aggressively replace the luteal phase in these patients. If you use an agonist trigger you really need to replace estradiol as well as progesterone.

Now cryopreservation is, of course, an alternative to cycle cancellation in high-risk patients. Non-conception reduces the risk of OHSS has been long known in oocyte donors. And subsequent transfer of thawed embryos can result in excellent cumulative pregnancy rates.

Other punitive strategies for preventing OHSS, which have been described for a number of years include the use of oncotic agents such as IV albumin or hydroxyethyl starch. The proposed mechanism of action for these agent is either the maintenance of oncotic pressure, although you would think that that would be rather short lived. And possibly the binding of ovarian vasoactive compounds such as VEGF.

As least one meta analysis of prophylactic albumin at the time of retrieval might, in fact, reduce the risk of OHSS but there are several other studies in the literature that call this into question and at this meeting there's at least one presentation of an updated meta analysis suggesting that it is really not efficacious.

Similarly hydroxyethyl starch is a high molecular weight compound comparable physical properties to albumin. It is usually administered at retrieval and two days later.

And, again, there are some studies in the literature that suggest that this may be effective also when you compare hydroxyethyl starch to normal saline given in a comparable volume they found a lower rate of hyperstimulation 2% versus 14%.

So in summarizing this part of my talk, OHSS is a largely preventable complication of ART. I do thing that the practice-specific incidence, which is widely variable in the literature, will depend largely on the individual philosophy of stimulation. How aggressive one is about stimulating patients. And also what your go/no go threshold is for canceling a cycle or cryopreserving embryos.

I'm now going to turn to a discussion on the risk of multiple birth following IVF, which is quantitatively a much more major problem, certainly in the order of magnitude more common than severe OHSS.

There are several well-known perinatal risks of multiple pregnancies. NICU admissions are seen in about 25% of twins and about 75% of triplet gestations. Risks of long-term sequelae such as cerebral palsy are markedly elevated compare to single gestations. Twins fourfold higher, triplets seventeen fold higher. And neonatal death by one year of age is sevenfold higher in twins and twenty five times higher in triplets than in singletons.

The etiology of perinatal mortality and morbidity is due in part to these factors. Preterm birth obviously is much more common. Intrauterine growth restriction. There is a higher incidence of congenital anomalies. Abnormal placentation such as previa and placental abruption. Higher rates of preeclampsia and moderate to severe preeclampsia can lead to iatrogenic pre-term delivery and finally cord accidents.

There also, although this really isn't discussed quite as widely, there are definite and real material risks which can also be fatal in conjunction with multiple gestation. Higher rates, again of preeclampsia and eclampsia and variance including the HELLP syndrome. Acute fatty liver of pregnancy is still rare but much elevated with multiple gestations and can be fatal. Thromboembolic events are more common with multiple pregnancy and so to is postpartum hemorrhage, largely due to uterine atony.

Now I think we would all agree of the option of multi-fetal pregnancy reduction is an important clinical tool to have in our armamentarium but in the same breath I think we'd have to agree that this is a far less and optimal solution, far better to prevent multiple gestations up front. I think depending on the patient's age and the number of gestations she's carrying, it's imperative to counsel regarding the pre-procedure of chorionic villus sample to check for euploidy. And patients and clinicians need to be aware that there is an associated loss rate. The most common type of selective reduction following ART is probably triplet to twin reduction and most series suggest a loss rate that is associated between 4% and 6%.

So in some the only effective way to minimize the incidence of multiple pregnancies is to minimize the number of embryos that are replaced, ideally transferring only a single embryo.

Even here, there will still be a high rate of multiple gestations. It's been pretty well established that ART has about a 3.2% incidence of monozygotic twinning, which is eight fold higher than spontaneous conceptions which were about .4% of pregnancies.

Now in our society there is a not subtle pressure to transfer multiple embryos, and some of this is certainly patient generated and some of this is also clinician generated. First of all, there is a widely held suggestion that twins are indeed a desirable outcome. And this is an area where I think patient education is imperative and a very important part of the counseling process. There is clearly an economic incentive from the patient's standpoint. Many lack or have limited insurance coverage for ART, which is obviously very costs and so there's a tremendous economic incentive and imperative to make every fresh cycle work. And I would also question whether this is in part an unattended consequence of published clinic-specific registry data, which I think are invaluable in many respects but can be subject to misinterpretation. If a patient looks at a table and sees that Clinic A has a much higher birthrate per transfer than Clinic B across the street there's going to be a flow of patients to Clinic A. And the clinicians at Clinic A and B are acute aware of this and this may influence their practice.

Now a number of studies have looked at elected single embryo, this is both a mix of blastocyst and day three embryo transfer to double embryo transfer in the literature. And some of these have shown very satisfactory rates of pregnancy with single embryo transfer in selected patients.

Most of them, not surprisingly, show higher pregnancy rates, however, with double embryo transfer.

One well designed study that received a lot of attention was the Thurin study in the New England Journal of Medicine in 2004 in which they essentially randomized a group of young patients, the average age was 30 and I think that's notable, to either single fresh embryo transfer, or SET, or double embryo transfer. Now the patients in the single embryo transfer group as an incentive were told that they would have financially free of charge subsequent thaw cycle of their fresh cycle failed. Now the implantation rates were of course the same in the two groups. The live birth rate per fresh transfer was not surprisingly higher with the transfer of two embryos, 43% as compared to about 29%. When you look at the cumulative live birth rate per retrieval in the SET group when a subsequent frozen cycle could be incorporated it was 38.8%, which was still lower than 43% but really closing the gap very significantly. And most importantly there was a huge difference in multiple pregnancies. About a third of the pregnancies in the double transfer group, less than 1% with elective single embryo transfer.

So if one of our goals with IVF and perhaps indeed the Holy Grail of ART is to be able to select an embryo or embryos that have the potential to implant, how do we choose the embryo? Well I think still the gold standard and probably for years to come will be a morphologic developmental assessment of the embryo. So embryo morphology through grading systems on day three or day five I think are important and useful. And I think that's part of the efficacy of watching an embryo progress to a blastocyst and that there is a process of self selection. Some authors and practitioners would argue that aneuploidy assessment may actually help to choose the embryo to replace, I think this is controversial. And there are some areas of active research that may prove, provide us with viable alternatives which are less invasive in the future such as proteomics.

This is a slide of a series of patients at Cornell who under went elective single blastocyst transfer. It's stratified by age, because of time constraints I'll just summarize. The overall success rate per single blastocyst transfer in an admittedly selected group of patients was 48%. I think there wouldn't be too much excuse for transferring more than one for those patients.

Now looking at aneuploidy screening, the technique that's used of course is PGS using fluorescence in situ hybridization. And if you look at the literature, even though in theory this should improve implantation rate and therefore pregnancy rates, it's actually very mixed.

Studies by Munne, Gianaroli and others have suggested that implantation rates are indeed improved with PGS and that live birth rates may also be improved. However the studies of Stetson and more recently Mesterbrook showed the opposite. They actually showed in a randomized trial that the live birth rate was lower when PGS was undertaken then when not performing PGS in a controlled group. One might surmise that this is due to the invasiveness of the biopsy procedure and possibly error rates including false positives.

And finally proteomics in ART I think is potentially an exciting area. I mean briefly this is analyzing the embryonic secretome by employing single embryo culture. And sensitive techniques like mass spectrometry can be used to identify candidate proteins. And in theory one might emerge with a profile of proteins that might, for example, be a marker for euploidy. If this emerges as a viable approach, this may be a non-invasive viability assay. I think it's probably not going to be ready, however, for clinical practice for the next few years.

Finally in closing I'm going to talk about what are the current ASRM guidelines for number of embryos to transfer. And how important is the United States with these guidelines, at least as of 2006. First of all ASRM and SART I think have rightly and correctly divided patients into more favorable and less favorable prognostic categories. Favorable entails one or more of the following criteria. First IVF attempt good morphologic embryo quality. Excess embryos to sufficient quality to freeze and previous IVF success. Others would either not meet any of the above criteria or have two or more previous failed cycles.

So looking at the most favorable group of patients, the guidelines suggest that only one blastocyst should be transferred in patients under 35 and at most two day three embryos with an emphasis on trying to perform elective single embryo transfer in a really good prognosis patient. Between 35 and 37 up to two embryos, 38 to 40 up to three day three embryos or two blastocysts and over the age of 40, 41 and up, up to five day three embryos or three blastocysts.

For the patients of less favorable prognosis, up to two embryos again under 35. Up to three to two blasts, 35 to 37 and up to four embryos or three blasts between 38 and 40.

Now this is the 2006 SART registry data, which is the most recent year that it has been published, this just came out this spring. And this is looking at the mean numbers of embryos transferred at the various age groups and the incidence of elective single embryo transfer. In most age categories practice does seem to be pretty concordant with the guidelines if you include good and poor prognosis patients. But the outliers are really are the youngest patients, under 35 a mean of 2.3 embryos were replaced when the guidelines for even less favorable prognosis cap it at 2. And I was particularly surprised to see that in donor oocyte cycles the mean number of embryos replaced was 2.2. This should be the target audience for elective single embryo transfer, certainly no more than two embryos. And if you look at elective, voluntary single embryo transfer even in the youngest patients only 3.3%.

I think this is my final slide summarizing what is the incidence of multiple births in the United States at least with IVF performed in 2006. We've made great inroads on the high order multiples. So triplets under 35, 2%, 41 to 42 less than 1%. But twins are still a major problem. Under 35, 32% and 41-to- 42, 15.5%.

So I thank you for your attention and I'll turn the podium back to Dr. Reindollar.

Richard H. Reindollar, MD

Thank you very much Dr. Davis.

So for the next few moments I would like to talk to you about some serious risks of IVF but fortunately they're very infrequent. I'll spend just a few moments on cancer. I've given you some references which I think are the best review articles that you can get more details about this. I'm going to spend the bulk of the time on an area that I think is very concerning and that is the risk of death in pregnancy for patients with Turner syndrome. And then I'm going to end with just a few moments about an observation about a potential risk that I think has gone unnoticed in the past.

Now when we talk about the risk for cancer it's been long known that different etiologies have been associated in the literature with an increased risk for ovarian cancer.

There's extensive literature on the risks of endometriosis for causing ovarian cancer in the future. And as you can see here these other etiologies have been implicated as well.

But I think that the risks that we are most concerned with, of course, are the risk of ovulation induction or super ovulation and the risk for ovarian cancer down the road. Now we know that by inference there are some theoretical epidemiologic data that would suggest that that risk might occur because we all know that when ever we reduce the number of ovulations or put the ovary to rest whether it's by pregnancy, whether it's by the use of combined hormonal contraception or breast feeding we do reduce the risk of ovarian cancer.

So it would make sense to us that if we over stimulated the ovary that in fact we might increase that risk. There are some other data and some basic science data, some molecular markers that have been shown to correlate with the number of ovulations that patients have. For example, women with p53 over expression tumors are associated with more ovulations and patients with lower levels of anti-MUC1 antibodies, a protective antibody, also have higher numbers of ovulations.

So I think though that the proof in the pudding is what do we know from the literature. These next two tables actually come from a review article published several years ago in Modern Trends and the first lists the large number of cohort studies that have been done examining this question.

And the second case controlled studies.

And we know that in the early studies, the very earliest studies for both the case controlled and the cohort studies there was the suggestion of an increased risk of ovarian cancer. And interestingly enough it was primarily and most noted with an increased duration of time of clomiphene and particularly over 12 month use of clomiphene and also most noticed for an increased risk of a borderline ovarian tumors. But these early studies that demonstrated that risk were plagued to some degree by a smaller number overall of cases of ovarian cancer. And they lacked information regarding the known confounders of ovarian cancer.

The recent studies have consistently been reassuring, although, of course, not decisive. The evidence of modest increases in rick suggested with extended followup were increased exposure is in the literature, which suggests to us that while we may be overall reassured we can't bury our heads in the sand and we have to continue to have these patients followed so that we know down the road whether or not it's true. Also, if an increased risk between ovulation induction agents and ovarian cancer exist, it appears that it will be greatest for patients who are nulligravid and also, again for producing borderline ovarian tumors.

However, studies consistently show no increased risk specific to IVF. And I think most of us would agree that that makes a lot of sense. As it's the ovulation inducing agents that have been most implicated.

Now for breast cancer there are a number of studies shown here and here that are overall very very reassuring. It's interesting one study actually suggested that with clomiphene use there seemed to be a reduced risk of breast cancer suggesting that maybe there was a tamoxifen like effect on the breast.

However most studies show really do not show an increased risk for breast cancer.

And as we would realize, endometrial cancer is most increased in patients with chronic anovulation and unexposed estrogen. And the large multi-center U.S. cohort study demonstrated that it was clomiphene usage for the highest doses and the longest durations that seemed to be associated with the highest risk. Now they actually, individuals have actually suggested will maybe this is also a tamoxifen like effect but I think most of us would believe that this is because these are the patients who have the longest term unopposed estrogen and are at the highest risk.

So there are a number of questions that remain unanswered. And overall when we look at reducing the risks, I think that we need long-term followup studies.

The studies that we have are clearly reassuring, although not decisive. We need to limit the overall exposure should that risk exist. In other words, limit our treatments to those studies that are most successful. Last year at this meeting we reported the outcome of the NIH trial, the FASTT trial in which we randomized 503 couples to two different treatment paradigms and demonstrated that there was no added benefit to gonadotropin IUI.

And as I said I think is our study and other studies are presented in the literature will be able to limit the stimulation that patients have and get to those treatments that are most successful in the shortest period of time.

Perhaps there's some benefit of oocyte retrieval without ovarian stimulation and in vitro maturation as these studies also appear to be coming more and more successful although are still considered research at this time.

Now I'd like to spend the next few moments talking about an area of which I have concern. And I think it's an area that all of us have known about but there are some developments that raise my concerns even further. And that is the risk of donor oocyte for women with Turner syndrome.

We know that about 85% of all patients with Turner syndrome will present at puberty with delayed puberty and will, no doubt, have streak ovaries. They have lost their oocytes prior to that period of time. About 15% will have enough oocytes to mount a pubertal response and about 5% will have cyclic menses.

But the vast majority of patients that we see are those who had delayed puberty and have streak ovaries as shown here and a normal uterus.

In the early days of oocyte donation, in the mid-1980s, as oocyte donation was being developed and offered and found to be successful it made sense that Turner syndrome would be a very reasonable indication for donor oocyte. And this was one of the, well it was actually the first largest study from the UK of patients with Turner syndrome who had had donor oocyte and gone through pregnancy. And as you can see there it was successful although they actually reported a 50% spontaneous abortion rate and actually suggested it was related to the uterus. And I'll tell you that subsequent studies I'll show you have normal spontaneous abortion rate and I think was likely related to the early days of donor oocyte and the hormone regimens that were used for support for those pregnancies.

But during the time that donor oocyte was being offered to the patient in those early years with Turner syndrome and up through that time there were a number of reports that were gathering in the non-OB/GYN literature which demonstrated individual cases of women with Turner syndrome who were developing dilation, dissection and rupture of the ascending aorta in the non-pregnant state. Now it's interesting, Angela Lin in 1986 actually compiled the literature at that time. Most of these patients presented to the emergency departments with nondescript pain, it may have been shoulder pain, it may be chest wall pain and were misdiagnosed and were sent home to die. But it was found that of these patients, the vast majority of them had risk factors which included congenital heart defect primarily by cuspid aortic valve, less commonly coarctation of the aorta and pseudocoarctation and also another risk factor was hypertension. However, Angela Lin reported that there were a number of these patients, who, in actuality, did not have any seeming risk factors. All of them at post were found to have cystic medial necrosis of the vessel walls, that is the histopathology associated with the vessels of Marfan syndrome that dissect and rupture. So in those early days of a donor oocyte I had tremendous concerns that when these patients became pregnant, the increased blood volume, the increased pulse pressure would hammer away at the ascending aorta and would increase the risk for dissection and rupture for these patients.

And, in fact, subsequently there were four such patients reported in the literature, three of whom had known risk factors for developing dissection and rupture. And it was the fourth of these reports that really alarmed me. It was the study from Mayo Clinic in which they reported the pregnancy outcomes of 50 of their patients who had coarctation of the aorta and had become pregnant either with or without prior repair of the coarct. Only one patient died in that series. She was the only patient to have Turner syndrome. And so it made me realize that there must be something different about the women with Turner syndrome that may place them at a high risk.

So at that time I was at Beth Israel I had our fellows and Megan Kharnis, who is here today, to do a survey of the donor oocyte programs across the U.S. And we contacted these programs over and over again to get the best response and ultimately 134 programs responded, which was 52%. You can see here that only half of the patients at that time had been offered a donor oocyte and had been who had screening of their aorta with an echocardiography.

Of these patients there were 101 pregnancies, 65% of couples became pregnant and you can see here the spontaneous abortion rate was only 7%.

Now we did a back of the envelope calculation. That's all it is. But what we said was, if there were 101 pregnancies in 52% of the programs and if all of the other 48% of the programs had this same number of Turner syndrome patients who became pregnant, and my guess is they didn't respond many of them because they didn't have Turner patients, but if they had the same number then there were be an estimated 194 pregnancies with a death rate of about 2%. And I will tell you my non-OB/GYN colleagues who are Turner syndrome specialists believe that this a conservative figure. Now our patients may say well that's 98% of the patients are going to be fine. What that means is, for example, at the BI with 6,000 deliveries a year, 120 women would die. It's a hundred fold greater than maternal mortality rates for all other causes.

Now I'll tell you that subsequent to this the ASRM practice guideline published their recommendations. What they recommended first of all was that Turner syndrome should be considered a relative contraindication. If you have any of the risk factors it would be an absolute contraindication. They suggested that all individuals have cardiac consultation and that patients have monitoring with echocardiogram if, in fact, they had no abnormalities and decide to become pregnant.

In 2007, the NIH Consensus Study Group had met and they published their findings or their recommendations in JCEM. Now these are for non-pregnant patients. But what they published was that all patients should have a baseline echocardiogram and if it is normal that they then have an MRI because there are patients who have dilation of the aorta found by MRI but not found by echocardiogram.

Now, the last year, in 2007, there have been two articles that have shed more light on this. Michael Silverbach and Carlson, who are from Oregon updated the literature to 85 patients who have had subsequent rupture, and these are all patients and the majority non-pregnant. And what they did was report that of all patients who have had rupture, 11% had not had any risk factors. No congenital heart defect, nor hypertension. And in fact they added two patients to the literature that when they did MRI, these patients who had died from dissection, had aortic measurements that were less than one would expect for risk factor for rupture.

And what they suggested was that maybe patients unlike Marfan syndrome who dilate first and then dissect, may because of something else associated with Turner syndrome may actually dissect and rupture in the absence of prior dilation.

However, Carolyn Bondy at the NIH who studied many hundreds of patients with Turner syndrome reported in circulation 166 patients who had been evaluated with MRI. And what they demonstrated was that in fact that they felt that the reason that patients may have dissection and dilation of the aorta is that their aortic size was not corrected for body surface size. It's known that larger individuals have larger aortic sizes, that it's a direct relationship. So in this study they actually measured the ascending aorta and then they corrected for body mass size and when they did that they demonstrated that a third of the Turner syndrome patients had aortic sizes that were greater than the 95% percentile for the normal population. So patients with Turner syndrome may actually rupture at a lower diameter because it's larger to them.

Now, at the National Turner Syndrome Society meeting this past summer I met with Michael Silverbach and I met Dr. Bondy and what's concerning is the following. That there are more patients who have died. Each of us had known individuals who have died during pregnancy. Dr. Bondy has patients whom she's followed during pregnancy and shown a progressive dilation of the aorta during pregnancy and there are now patients who died down the road after a pregnancy. So I think when we look about it there are many unanswered questions and what we really need is a registry for these patients. It's something that we're looking into. In the absence of hard data we should consider that the maternal mortality of 2% it's an estimate but it may be conservative.

That patients who safely get through pregnancy may not have safely made it through pregnancy. That the physiologic changes of pregnancy may actually increase their risk for subsequent rupture down the road. And that's something that we don't have good data but more and more data is being collected.

We need to have MRIs on these individuals and as the ASRM guidelines suggest that we should get cardiac consultation, please it should not be for clearance. Unfortunately, cardiologists still may not know this risk and I know of one patient who got a go ahead for pregnancy who had two risk factors and died during pregnancy. So cardiac consultation for IVF should be for assessment alone.

I'm just going to end with two minutes and just tell you a brief story just to keep your eyes and ears open. We know that true hermaphrodites, the vast majority, over 80% of them have a 46 XX karyotype and interestingly their gonads are usually ovotestes on one or both sides. Less and rarely true hermaphrodites are 46, XX, 46, XY and interestingly about 60% of the time they have an ovary on one side and a testes on the other and the very very rare 46, XYs are just like them. And it's felt that they probably are XX, XYs. Most of the literature would suggest that or XX, XY true hermaphrodites are from amalgamation or whole-body chimerism an XX and an XY embryos together and an ovary and a testes of course.

There is some data that has suggested that this could be caused by fertilization of the polar body by say an X bearing sperm and of the oocyte by a Y bearing. However, just to leave you with a thought.

In 1998 there was report in New England Journal of a case of IVF baby that was born from two embryos that were transferred and it was a 46 XX, 46 XY true hermaphrodite and they did very extensive studies that suggested that this truly was from amalgamation of the two embryos. I was contacted this past year by a colleague of one of my fellows who had a very similar case that has yet to be reported. And it was a patient who had three embryos transferred after ICSI and at the blastocyst stage had twins, one was a normal 46 XX female and another was a 46 XX, 46 XY individual with a true hermaphrodism. So I haven't seen any other cases but I think it's something that we should keep in mind and perhaps down the road think about how we're transferring these embryos.

So I'm going to end there and turn the podium over to Dr. Martin. Thank you.

Renee H. Martin, MD

I'd like to thank the organizers for inviting me. It's my very great pleasure to be here and my instruction to day was to look at whether ART increases the risk of birth defects.

So specifically in offspring of ART pregnancies is the frequency of birth defects is increased and is the frequency of chromosomal abnormalities increased.

So there have been a number of studies looking at congenital abnormalities after ART and these are just some examples of some. So a study in 2002 that was a Finnish registry study found four times the frequency of heart defects. A study here in the United States by Hudson et al reviewed 2000 studies found 200 of them to be acceptable for criteria that they defined and found no strong evidence for an increased frequency of birth defects with the possible exception of imprinting defects, which at that time were just sort of starting to be reported. Bonduelle et all part of the Brussels group that have looked at the large cohort that they have been following looked at 5 year old children in 2006 and found that ICSI there was a 4.2% risk of major malformations increased over controls, especially for hypospadias.

There have been other studies that have highlighted hypospadias in ICSI offspring. Wennerholm et all found a relative risk of 3 in the Swedish medical study that looked at a congenital malformation registry. And Ericson and Kallen found a relative risk of 1.5. So overall I think most people now feel that there is a slight but it is a very small increase but a slight increase of relative risk of 1.3 to 1.4 and with perhaps highlighting hypospadias and imprinting disorders as being possible ones that are more increased.

When we look at chromosome disorders after ICSI, these studies have been ongoing for a number of years. One of the first was by Van Opstal et al in 1997. And they looked at 71 prenatal diagnoses, found 8% to be abnormal after ICSI and importantly the sex chromosomal abnormalities were all of paternal origin, highlighting the fact that we're talking about male infertility in this case. Aboulghar et al had a very nice study in Egypt because they looked at 430 consecutive babies and this is unusual to have a consecutive series like this and they found 3.5% to be abnormal and these included both sex chromosomes and autosomes. And 3.5% is considerably higher than the background rated controls of .6%. And Devroey & van Steirteghem, again of the Brussels in 204 looked at over 2600 fetal karyotypes and found 3% to be abnormal, 1.6% to be de novo abnormalities, again sex chromosomes and autosomes and 1.4% to be inherited, things like translocations and inversions.

So, these birth defects that we see in ART, the increased risk could reflect the procedure per se, the in vitro fertilization or the intracytoplasmic sperm injection of they could reflect the characteristics at the population that require ART. In other words ART offspring have an increased a priori risk because of their parent's infertility.

Now there are a number of technical variables that could implicate the ART. The ovulation stimulation regimes they're very variable and have changed with time so it's very difficult to see if one particular regime could be responsible for anything. The same with obtaining and handling gametes, they may be at different temperature under different atmospheres. The embryo culture has certainly changed over time and still there are variations today. And cryopreservation may have different media or different freeze/thaw cycles.

So we have multiple variabilities at each step. And it's plausible that certain variabilities could differentially affect imprinting especially embryo culture since we've shown this in animal studies.

The population that requires ART differs from the general population. Generally they have an older age and they have disorders that have implications to the offspring. And the difficulty here is that it's difficult to get a true comparison group, that is infertile couples who are not undergoing ART. Difficult but not impossible

So, when we look at anomalies in ART offspring there are heterogeneous indications, often no single known etiology for the infertility or you may have maternal and paternal factors. And some parental disorders causing infertility predispose offspring to have anomalies. So for example if you have a father with male factor infertility because of congenital bilateral absence of the vas deference, and is heterozygous or even homozygous for a CF mutation, of course that's going to have increased risk for cystic fibrosis in the child.

When we look at somatic chromosomal abnormalities these are chromosomal abnormalities that are in all of the body cells of the individual. We've known for a long time that these are increased in infertile patients. So in infertile women, even when the cause of the infertility is known to be a male factor, the frequency of constitutional chromosomal abnormalities is 5% to 8%. And in infertile men, the frequency is 2% to 20%. That big range because the 2% is for males with any sort of infertility. The 20% for men with non-obstructive azoospermia.

Approximately half of these are sex chromosomal abnormalities and the other half are structural abnormalities. And these, of course, are much higher than the newborn frequency of .6%.

When we look at meiosis in sex chromosomal trisomies theoretically we should expect that 50% of the gametes will have an extra sex chromosome when you just think of these three sex chromosomes segregating at meiosis. But in general it seems that the extra sex chromosome is eliminated during gametogenesis and when we look at offspring of 47, XYY or 47, XXX individuals, these are generally normal, although there have been few studies and certainly not with large numbers.

In our lab we looked at meiosis in a 47, XYY male a number of years ago in 1988 and used the sperm karyotyping technique in which you have a hamster egg penetrated by a human sperm to look at the actual pronuclear chromosomes. We found no sperm disomic for the sex chromosomes. So no abnormalities at all. That's based on 75 sperm karyotypes. When we studied this same man in 1997, we were able to look at 200,000 sperm using fluorescence in situ hybridization and then with this bigger number we were able to show an increased frequency of XY disomy but at a low frequency, only .6% suggesting that the extra sex chromosome is eliminated in most cells but a small yet significant increase in aneuploidy does occur. And other labs have found frequencies varying from .5% to 15% sex chromosomal aneuploidy in these men.

This just shows an example of what human sperm karyotypes look like only 23 chromosomes because these have undergone meiosis so we have separate the homologous chromosomes into just 23 in the sperm. And the important thing is that you can identify every individual chromosome, numerical as well as structural abnormalities.

And this is just an example of sperm having been analyzed by FISH with the red signal being for a Y chromosome, the green for an X chromosome and the blue a control for a number 1 chromosome. So this is an abnormal 24, XY sperm with an extra sex chromosome that would lead to Kleinfelter's if it fertilized a normal egg.

So when we look at meiosis in Kleinfelter's male, the sperm generally has to be extracted microsurgically and there have been analyses of FISH in mosaic Kleinfelter's finding a sperm aneuploidy frequency of 1.5% to 7%. And also in non-mosaic Kleinfelter's syndrome, sperm aneuploidy slightly higher at 2% to 25% but overall nowhere near the 50% we might expect theoretically.

And when we look at offspring of these men after ICSI, there have been 54 “normal” children born, normal in quotation marks because many have not been karyotyped. And of course you expect sex chromosomal abnormalities in general to look normal at birth. Two terminated fetuses with a Kleinfelter's. A number of case reports from preimplantation genetic diagnosis that had abnormal sex chromosomes. Schiff et all reported on 21 chromosomally normal live births from 42 Kleinfelter's men. And a Brussels group, Staessen et all looked at preimplantation genetic diagnosis of 113 embryos and found a significantly increased frequency of autosomal and sex chromosomal abnormalities. So, again, it seems like a slightly increased risk for these individuals but nowhere near the 50% you might expect.

When we look at the risk for Robertsonian translocation carriers, we find that 3% to 40% of the sperm are unbalanced when you look either by karyotyping or by FISH with a mean of about 15%. The frequency of abnormalities at prenatal diagnosis varies for males and females. It's quite low for males at 1% to 2% and females for 10%. And even though the risk is relatively low by the time you get to the time of prenatal diagnosis, of course the risks can be devastating to the individual families and there have been a number of abnormalities born such as trisomy 13 and a 13/14 translocation carrier. And interestingly enough quite a number of reports of 13/13/ Robertsonian translocation carriers and what this means is if you have an individual with this translocation they can only have abnormal children. So they can only have trisomy 13 or monosomy 13, which will abort, which of course oviates the reason for having ICSI because there's no possible way they can have a normal child.

Reciprocal translocations between any two chromosomes and these have a much higher risk of unbalanced offspring. And when we look at sperm karyotyping in 35 carriers, the frequency of unbalance is 43%. So quite high. And at preimplantation genetic diagnosis it's significant that the risk is the same in sperm as it is in embryos. So no selection of those sperm based on chromosomal abnormality, which is still something that a lot of people believe occurs. And in prenatal diagnosis the mean of unbalanced is both 12% and in this case it's the same in males and females.

So that's a quick summary of what happens when an individual has a constitutional chromosomal abnormality. But of course the majority of our patients are chromosomally normal of our infertile patients. And we're interested, do these individuals have an increased risk of chromosomal abnormalities? A student of mine, N. Moosani, first showed in 1995 that even these men who have ICSI and are chromosomally normal still have an increased risk of chromosomally abnormal sperm. So she looked at over 500 sperm karyotypes and found a significant increase of aneuploidy four times the control rate. And FISH on over 100,000 sperm and found a significant increase for chromosome 1 and the sex chromosomes.

Since then there have been more than 30 multicolor FISH studies done on infertile men and the great majority show an increased frequency of disomy for the autosomes and the sex chromosomes. And these have mainly been on men with severe infertility or various mixed types of infertility.

We wanted to look at rather the risks of ICSI might be different for different subsets of infertility. And my bias was that the oligozoospermia, the low sperm count, would have the highest risk of chromosomal abnormalities. Because it's well known that they're in meiosis in human males there's a very strong checkpoint that when something starts to go wrong those cells are removed from the spermatogenesis subsequent making sperm so that then you have a lower sperm count and it might refer to this abnormality. But in fact when we looked at these various types, oligozoospermia, asthenozoospermia, motility problems, abnormal forms, azoospermia, and no sperm in the ejaculate.

We found that all of them had an increased frequency of sperm chromosomal abnormalities varying from 2 to 10 times higher than controls with the highest risk being, not surprisingly in men with non-obstructive azoospermia.

So we were interested to know whether these chromosome abnormalities in these infertile men are related to errors of meiotic recombination. And the reason for wondering about this is that there had been many studies now in animal showing this linked between recombination defects and chromosomal abnormalities and also in humans showing these links of particularly down syndrome and sex chromosomal abnormalities reflecting lack of recombination. So a new technique has comes about an immunofluorescence techniques, which allows us to study meiosis directly by looking at the synaptonemal complex. And this is that protein goop that holds homologous chromosomes together at meiosis and facilitates recombination to occur. So we use antibodies for the synaptonemal complex. Also for recombination foci in that we're lucky that a mismatch repair protein (MLH1) exactly marks where recombination occurs because it's important in the recombination process. And then we also mark the centromeres with a CREST.

So this is what they look like, these little red worms are actually that's the synaptonemal complex with the chromosomes being outside of that. The flue is the centromere and the yellow is the exact site of recombination. So now we can know the number of recombinants in a cell. We can tell which chromosome they occurred at. We can tell the location of the recombination and then start to analyze this more exactly in human males.

And we found that the mean frequency of recombination significantly decreased in men with infertility, particularly men with non-obstructive azoospermia.

And even more importantly we found that those cells in which there was a chromosome pair, at least one chromosome pair with no recombination hugely increased from 5% in controls to 29% in men with non-obstructive azoospermia. So that means in about a third of the cells, at least one chromosome pair is not properly joined, which means it's not going to be able to separate from each other and segregate properly at meiosis.

So in non-obstructive azoospermia we've got a decreased frequency of recombination and this high risk factor of chromosome pairs with no recombination foci. And this is either going to lead to meiotic arrest if the checkpoint sees those abnormalities and moves those cells. Or an increased frequency of aneuploid sperm if the cell escapes that checkpoint. Now what about infertile women? Are they also at increased risk of having gametes with chromosome abnormalities? Of course this is much harder to study, oocytes are much harder to get than sperm, but there have been some suggestive studies.

So Sante Munne et al looked at preimplantation genetic diagnosis in infertile women 18 o 34 years of age and found that 34% of the embryos were normal. When you look at donor oocytes at exactly the same age, significantly more were normal, 43%. Again, suggesting that the infertile women have something innate causing these chromosomal abnormalities similar to the men.

Now when we look at imprinting and ART there's been a lot of talk about this in the press. And the relationship of ICSI and IVF to imprinting is biologically plausible when you consider that we are manipulating the gametes and the embryo and particularly for female the imprinting in the female occurs not just in the gamete but in the early embryonic stage. And this is also supported by the association of cases characterized by uncommon molecular basis, in particular, a disturbance of maternal imprinting that occurs right there in the early embryo.

So for almost all genes we normally have bi-allelic expression. Both the mother's chromosome and the chromosome you got from your father are transcribing genes from both. But in imprinted genes, which we now evokes 60 now in humans, we get unequal expression of the maternal and paternal alleles. So in this example, say a paternal allele is not transcribing and it's only the maternal allele is transcribing. And really we learned a lot of this from mouse work.

So, the requirements for imprinting. You have to have a paternal mark that is set in germ cells saying this chromosome will transcribe and the other one won't. It has to be reversible on passage through the opposite parental germline so that we have a chance to switch that imprint. There has to be differential expression of the new parental imprint in the offspring. And it has to be stable transmission through mitosis in somatic cells. And as I said we know now of about 60 imprinted genes but it's estimated that we have about 100.

So imprinting defects in ART babies. Angelman's syndrome, which is a syndrome of mental retardation and seizures, has been reported. Three sporadic imprinting defects on a maternal chromosome. And Beckwith-Wiedemann syndrome, which is an overgrowth syndrome of abdominal wall defects and embryonal tumors. There have now been a number of studies in the UK, France, USA and Australia showing 4 to 11 times the risk. Now you can have these imprinting defects caused by a lot of things. They can be caused by chromosome deletion, translocation, uniparental disomy, errors of the imprinting control and I think one thing that is very significant here is that 13 out of 14 analyzed Beckwith-Wiedemann syndrome cases are associated with ART with sporadic loss of methylation on the female chromosome. Now this is a very rare case normally, in less than 5% of cases. But almost all of the cases associated with ART have had this cause. Now the absolute risk is still small. There's no question that this is 1 in 3,000, 1 in 4,000 if there is an increased risk and we certainly don't know that yet but it is intriguing that we have these very specific types of defects.

As I said, we know from a lot of mouse data that culture media can have an affect on expression of imprinted genes. So for example, KSOM media leads to the normal maternal expression of the H19 locus or as a different media, Whitten's media leads to expression from both the maternal and the paternal chromosome, which is not normal, which makes it plausible that media could have an affect.

Now, again, is the imprinting an effect of infertility or is it an effect of the ART. There have been two studies, one by Ludwig et all in 2005 and one by Doornbos et all in 2007, both showing an increased risk of imprinting disorders, Angelman in the first study and a combination of Angelman, Bechwith-Wiedemann syndrome and Prader-Willi in the second study showing an increase with ART bit also showing the same increase in infertile patients without ART. So this suggests, again that it's something about the infertility that is causing this difficulty with the imprinted genes rather than the ART per se.

Now, one question that has come up is: are these imprinting disorders really the tip of the iceberg? And the reason for being concerned about this, we don't have any evidence for this but if epigenetic errors are associated with ART or infertility, not the technical aspects, they may effect methylation at other genes. So we know for example that epigenetic down regulation of the insulin growth factor 2 is correlated with intrauterine growth retardation and we know that this is increased in ART. And methylation of tumor suppressor genes are correlated with some cancers. So the worry is, do we imprinting defects of other genes, not just these ones that we know associated with various syndromes and are these going to be a problem in the future. As I said we don't have any evidence of this right now, but it's something that needs to be built in in future followup studies.

So, in conclusion, I think we can say that the ART couples are not a normal population and so it's difficult to find a control group. There are a myriad of technical variables at every step of ART so in some ways the effect is plausible. The overall malformation rate appears to be slightly increased with a relative risk of 1.3 to 1.4. Not a huge increase over the background. And no specific abnormality except for hypospadias in ICSI.

= Somatic chromosome abnormalities confer a high risk to offspring so karyotyping of couples should be considered critical. In ICSI, there's an additional increase of about 1% for chromosomal abnormalities in offspring. In imprinting perturbations are plausible but the absolute risk is still low. Thank you.