Chairman of Urology
Weill Cornell Medical Center
Former President, ASRM
NYU School of Medicine
Director, Male Reproductive Medicine
Mount Sinai Medical Center
If you are going to a fertility clinic this year, be prepared to have your clinic strongly encourage you to do ICSI. As you can see below, nearly all patients with a male factor diagnosis will have it performed, as will two thirds without such a diagnosis. In the next section, will look into which patients are proven to benefit by ICSI, versus who can probably avoid it.
Generally speaking, the trials to measure ICSI’s value are unsatisfying. If you are interested in reading how we select and calibrate these trials, we have more detail below. Otherwise, feel free to skip below to the chapter on “Male Factor” to learn about who benefits, and who is hindered by ICSI.
First, we are often dealing with small, single center studies. As you saw above, embryologist skill varies and to generalize about ICSI’s effectiveness based upon a few embryologists working at one clinic requires caution.
Second, many of these studies are poorly-designed and do not measure the important variable: live birth rates. Rather, many compare (and poorly at that) what percentage of eggs fertilize with CI or ICSI or what percentage of the time a patient had none of their eggs fertilize (again, known as Total Fertilization Failure, or TFF). Again, improving fertilization rates often does not translate into improving live birth rates.
Speaking of TFF, this endpoint is highlighted in studies because doctors worry that if a cycle ends in “no fertilization” the patient will be angry with them. However, oftentimes when we use ICSI to mitigate TFF, live birth rates are not improved. If our goal is to improve live birth rates, we think patients (and thus doctors) should focus less on the TFF endpoint.
Finally, we consider “fertilization rate” to be the least important endpoint because most investigators do not calculate it consistently. For CI, fertilization rate is calculated using the “total number of oocytes retrieved” as the denominator. For ICSI, they use “mature oocytes” as the denominator, which is an inherently smaller number and thus makes ICSI look more potent.
Below is our hierarchy and definitions for the endpoints you should care about:
Next, we rely on CDC data (and population studies) which has the benefit of capturing data from nearly all U.S. centers, but suffers from the risk the populations receiving ICSI or conventional insemination (CI) may not be the same and so comparing effectiveness is hard.
Finally, we rely upon “meta-analysis” where a researcher looks at all of the studies addressing a specific question (e.g. does ICSI cause birth defects). While these studies may incorporate data from many centers, they suffer from “investigator discretion”: the researcher gets to choose which studies to exclude and how much to weight the studies they include.
ICSI was developed to treat severe forms of male factor infertility and in these cases it is necessary and effective. The data is poor in characterizing if ICSI helps improve live birth rates in cases of “mild” or “moderate” male factor infertility, which encompasses most male factor patients. Below are the three main measures to characterize a man’s male factor issues:
At a high-level, when the CDC reviewed how male factor patients fared with ICSI versus CI, ICSI had lower rates of cycle cancellation (6% vs 14%) likely on account of TFF and higher live birth rates (37% sv 34%), but the difference was surprisingly narrow to us.
For patients with no ejaculated sperm (azoospermia) or sperm that does not swim forward, ICSI clearly increases rates of live birth and in our opinion, ICSI is a requirement
For example, at Cornell, when patients with non-obstructive azoospermia (considered the hardest to treat) have a testicular biopsy and sperm is discovered, over 36% of transfers lead to a pregnancy. Most fertility doctors will tell you these patients would have otherwise terrible odds of success without ICSI.
In cases where semen parameters are otherwise good but fewer than 5% of the sperm have normal shape, the better-run studies indicate ICSI does not improve live birth rates. A team from the University of Utah concluded in their recent meta-analysis that in this population ICSI is ineffective.
What do we do in cases where men have low, but not severely low, scores on the semen analysis? In this context, the quality of our data is poor, especially since the majority of these men will be told they must do ICSI. The studies available are conducted on small volumes of patients (often less than 50 cycles) and do not measure the important statistic: live birth rates.
The best study (in an otherwise unimpressive batch) demonstrates using ICSI reduces the risk of TFF in a meaningful way. However the study is small for our liking (76 cycles) and overly-broad in whom it includes (patients with count, motility and morphology issues) and does not elucidate if any one group especially benefits, or doesn’t, from ICSI’s use. The authors concluded that for every three patients that underwent ICSI, one was saved the heartbreak of having none of their eggs fertilize.
The CDC looked at all patients treated without a male factor issue. They noted cancelled cycles (likely on account of TFF) were similar between ICSI and CI and so were live birth rates. In effect, using ICSI did not help this broad group of patients.
Similarly, the best studying comparing ICSI vs CI outcomes showed no statistical difference in pregnancy rates. However, the analysis may not extend to women over 37 or couples that previously had low rates of fertilization (<20% of eggs fertilized), as these groups were excluded.
In this case, 33 couples had to undergo ICSI to save one couple from TFF. Compare that with the MFI patients, where only three couples needed to undergo ICSI to save one from TFF.
As you will see, “non-male factor” is a broad group and when we sub-segment it, we note the differences in rates of success between ICSI and CI vary by patient population. We’ll tackle what those sub-segments might look like below.
When there is no discernible cause for infertility, and the man’s semen parameters are within normal range, it’s common for doctors to use ICSI. That likely is unwarranted.
We have three small, single center studies that consistently show using ICSI virtually eliminates TFF. A larger meta-analysis from 2013 confirms the observation. However, only one study went so far as to compare live birth rates (in women under 40) and the study showed no significant difference between the two groups (though the study’s power was low).
The CDC data showed the rates for cancellation (likely on account of TFF) were similar between the groups and also showed the ICSI population had a slightly lower live birth rate.
We have one small, single center study from Spain that looked at ICSI in patients that retrieved six-or-fewer oocytes. Surprisingly, TFF rates were identical between ICSI and CI and, while pregnancy rates were higher in the ICSI group, this didn’t meet significance (likely because the power was so low).
The sub-segment analysis is interesting here. The findings were consistent across age groups. However, when the group looked at outcomes for women with three-or-fewer embryos, ICSI patients recorded fewer TFF issues. However, given the study’s small size, the group couldn’t declare it statistically significant.
The CDC data looks at women with four-or-fewer oocytes. The data clearly shows the ICSI group had a lower rate of cancellation (likely on account of TFF). But, the ICSI group recorded fewer live births (15% vs 13%). The implication here is that in this setting ICSI may do more harm than good. Indeed, when few oocytes are retrieved, dismissing 20 - 30% that are not mature, and then taking the risk of damaging a remaining oocyte seems ill-advised.
We have no prospective, randomized trials on whether ICSI improves outcomes for women of advanced maternal age (defined here as 38 and older).
The CDC data shows no difference in the percentage of cancellations (likely on account of TFF). From a live birth rate perspective, patients who used ICSI performed mildly worse.
Anywhere between 5 - 15% of patients will have a low percentage, or none, of their eggs fertilize. If you’ve had failed or low fertilization before, there’s a 40 - 50% chance it will happen in later cycles, so doctors err on the side of using ICSI. The data shows this is a wise choice.
One study took half of the patient population's eggs and fertilized them with ICSI and tried using CI for the other half. When ICSI was utilized, fertilization rates per egg were higher.
Another group did a similar exercise and noticed that for patients with previous TFF, ICSI not only fertilized a much higher percentage of eggs but alo led to far more live births.
The CDC did not study the subject likely because few, if any, doctors in the US feel comfortable treating such patients with CI. Hence, there is almost no data to study.
In a study out of Turkey, patients with only tubal factor issues who were under 39 were randomized to either ICSI or conventional insemination. During the study, not a single patient had TFF, pregnancy rates were identical and the live birth rates were considered within the margin of error.
A similar study done in Austria showed similar results. For both groups the rates of TFF were practically zero and, importantly, the ICSI patients had lower rates of ongoing pregnancy.
The CDC review did not investigate the issue but given the above, it would appear for patients with a solely tubal factor issue, ICSI does not improve outcomes.
ICSI is used as the default tool for fertilizing cyropreserved oocytes. The rationale is ICSI that is likely to have higher rates of success with eggs that have undergone structural changes during the cryopreservation process.
We have seen only one study that compares success rates of ICSI and CI in cryopreserved eggs. The study was small, conducted on outmoded cryopreservation technology (slow freeze, not vitrification) but demonstrates ICSI has meaningfully higher fertilization rates (46% vs 14%).
Unfortunately, we don’t expect any useful trials anytime soon. Most clinics have under 30 women per year return to fertilize their frozen eggs and most of those clinics insist on ICSI.
When sperm is frozen it often has a detrimental impact on the sperm’s motility. For this reason, not a single clinic in the US is likely to risk sperm that was “precious enough to be frozen“ with Conventional Insemination.
Your doctor and clinic are mortified that your eggs will not fertilize and you will be incensed if there existed an alternative technique with lower TFF rates. While live birth rates are often no higher with ICSI, it’s easier for patients to ascribe blame when failure occurs in the lab rather than in the uterus. This prompts many clinics to deploy ICSI as a “cover your butt” move.
While we have not seen data that ICSI improves the accuracy or PGT-A testing, or outcomes with PGT-A (see the next chapter), many clinics are afraid to be wrong on this and when PGT-A is slated to be used, default to ICSI.
Patients often pay $1,500 or more for ICSI and the clinic’s fully-burdened cost to perform ICSI is less than $250. Clinics make a lot of money on ICSI. Does this impact how often ICSI is used? We cannot say for certain, but we believe so.