Matthew G. Retzloff, M.D., and Mark D. Hornstein, M.D.

FERTILITY AND STERILITY VOL. 80, NO. 4, OCTOBER 2003 Copyright 2003 American Society for Reproductive Medicine Published by Elsevier Inc. Printed on acid-free paper in U.S.A. MODERN TRENDS Edward E. Wallach, M.D. Associate Editor Is intracytoplasmic sperm injection safe? Matthew G. Retzloff, M.D., and Mark D. Hornstein, M.D. Department of Obstetrics and Gynecology, Brigham and Women s Hospital, Harvard Medical School, Boston, Massachusetts Objective: To review the pregnancy and birth outcomes of patients undergoing intracytoplasmic sperm injection (ICSI). Design: Selective review of the literature. Setting: Couples undergoing either conventional IVF or ICSI and their resulting offspring. Patient(s): None. Main Outcome Measure(s):None. Intervention(s): None. Result(s): These observational studies are unfortunately inherently biased. Regardless, the information and experiences from these large centers cannot be ignored. These experiences reveal that birth weight may be decreased by an amount that is generally not clinically relevant. Chromosomal and genetic abnormalities are increased probably only as a direct corollary to the underlying parental risk. Congenital malformations reveal no clustering of any single specific abnormality. Developmental assessment is available for up to 2 years, and no major delays have been identified in either motor or mental function. Conclusion(s): Intracytoplasmic sperm injection appears to be a safe alternative for couples who otherwise would be unable to achieve pregnancy. The inherent risks associated with these genetically at risk couples mandate thorough evaluation and counseling before undertaking ICSI. (Fertil Steril 2003;80:851 9. 2003 by American Society for Reproductive Medicine.) Key Words: ICSI, IVF, infertility, birth outcomes Received February 20, 2003; revised and accepted April 29, 2003. Retzloff is a major in the U.S. Air Force. The opinions and assertions contained herein are the expressed views of the authors and are not to be construed as official or reflecting the opinions of the U.S. Department of Defense or the U.S. Air Force. Reprint requests: Mark D. Hornstein, M.D., Brigham and Women s Hospital, Harvard Medical School, 75 Francis Street ASBI-3, Boston, Massachusetts 02115 (FAX: 617-566-7752; E-mail: mhornstein@partners.org). 0015-0282/03/$30.00 doi:10.1016/s0015-0282(03) 01014-8 Intracytoplasmic sperm injection (ICSI) has allowed thousands of otherwise infertile couples to attain pregnancy through in vitro fertilization (IVF). The majority of these couples would otherwise never have conceived without sperm donation. As this technology moves into mainstream of infertility treatment, it has become even more critical to reassess its safety. Intracytoplasmic sperm injection was first described by Palermo et al. (1) in 1992. Their report documented four ICSI pregnancies in women who had otherwise failed to conceive with other existing assisted reproductive techniques. Their promising results were hailed as a major breakthrough in recalcitrant male factor infertility. This initial success has been repeated at thousands of IVF programs worldwide, and ICSI has become part of mainstream assisted reproductive technology (ART). Intracytoplasmic sperm injection involves the injection of a preselected spermatozoon into a mature oocyte after ovarian superovulation and oocyte retrieval. This in vitro insertion of sperm into the cytoplasm of the egg bypasses the process of natural selection of sperm/egg and their interaction. Initially, the major indication for ICSI was therefore severe male factor infertility. Today, however, the indications have expanded to include multiple failed IVF cycles, poor fertilization, and many others. As the indications have expanded and the number of children born from ICSI procedures has increased, greater attention has been focused on the safety of the procedure. Despite the clinical successes, controversy in this area remains regarding adverse outcomes for children born from such assisted conceptions. The potential concerns regarding ICSI offspring relate to four general areas of investigation: the obstetrical outcomes of pregnancies resulting from ICSI, chromosomal abnormalities associated with the offspring of ICSI pregnancies, congenital malformations of the newborns resulting from the ICSI procedures, and developmental abnormalities in children born as a re- 851

sult of ICSI. Therefore, this review article assesses the existing literature to understand better the risks to offspring conceived by this assisted reproductive technique. POTENTIAL RISKS Potential risks associated with the ICSI procedure can be divided into two main groups. As nicely outlined by Patrizio (2), the risks include both those that are independent of, and those that are dependent on the ICSI process. Risks independent of ICSI include potential fertilization of male gametes that carry either genetic anomalies or structural defects. In addition, there is the potential for incorporating sperm mitochondrial DNA or fertilizing anomalous female gametes that would otherwise be bypassed by natural selection. Either process could result in congenital malformations or male-related infertility in resulting offspring. The second group includes those risks that are dependent on the ICSI procedure itself. These include injection of foreign substances or contaminants, disruption of the ooplasm or the meiotic spindle apparatus, and the embryologist s improper selection of the incompetent sperm for injection. The introduction of these risks could result in birth defects or genetic abnormalities in offspring. The first place that ICSI-associated abnormalities could reveal themselves is the earliest point in development: the preimplantation embryo. Dumoulin et al. (3) addressed this issue by looking at ICSI-produced embryonic development in vitro. Their most significant finding was that development of the embryo to blastocyst in vitro was highly techniciandependent. Those technicians who tended to aspirate a higher volume of oocyte cytoplasm during ICSI had embryos with the poorest development. In this study, no aneuploidy was seen in any of the 33 embryos, which were tested for chromosomes X, Y, and 18, suggesting an all-or-none type of phenomenon: the embryos either would survive the micromanipulation or degrade, but their development would not be affected. Another potential procedural related genetic risk of ICSI includes epigenetic abnormalities of imprinting. The mechanism of genetic imprinting involves methylation of DNA and allows for only one of the parental copies of a gene to be expressed. Dysregulation of this system can lead to the congenital malformation syndromes such as Beckwith- Wiedemann and Angelman syndromes. Beckwith-Wiedemann syndrome (BWS) is characterized by somatic overgrowth and a predisposition to pediatric embryonal tumors. Angelman syndrome is a neurodevelopmental disorder characterized by severe learning difficulties, ataxia, seizure disorder, subtle dysmorphic facial features, yet a happy, sociable disposition. Sutcliffe et al. (4) first reported on a single identified case of BWS among 91 cases of ART in 1995. Since that time, two case series have identified BWS children resulting from ART: 7 out of 344 (5 after ICSI) (5) and 6 out of 149 (3 after ICSI) (6). Also, a report by Cox et al. (7) has described two patients with Angelman syndrome after ICSI. The fact that both disorders occur very rarely in spontaneously conceived pregnancies has given rise to concern. The relationship between imprinting defects and ICSI is thought to involve a disruption in the normal process of DNA-methylation of parental genes. Attempts to detect specific abnormalities of known DNA-methylation patterns revealed no abnormalities in 92 children produced from ICSI (8). This same group subsequently investigated whether males with abnormal semen parameters may inherently have an abnormal paternal imprint. This study failed to find any differences in spermatozoa imprinting status when compared with men with normal semen parameters (9). These results do not support a higher risk of DNA-methylation defects in children born after ICSI. As with any other disorder that occurs with such a low incidence, large-scale studies must be undertaken to see if ICSI plays a causative role in this group of disorders. OBSTETRIC OUTCOME The possibility remains that despite apparently normal in vitro development, delayed aberrations from ICSI may result in birth defects. The closest control group with which to compare ICSI births would be conventional IVF pregnancies without ICSI. Therefore, comparing obstetric outcomes from these two groups would be the best measure of the potential effects of ICSI on embryonic development. One such outcome is the rate of miscarriage. An increased rate of pregnancy loss may be indicative of an ICSI-related abnormal outcome. The 1999 Society for Assisted Reproductive Technology data for the United States reported rates of spontaneous abortion of ICSI pregnancies at 17.6%, which was similar to the reported rate for IVF non-icsi pregnancies of 16.7% (10). This would suggest no clinical effect severe enough to cause a loss of the pregnancy from ICSI in the first trimester. Several retrospective series have looked at specific obstetrical delivery data related to ICSI. Wisanto et al. (11) reviewed the first consecutive 424 pregnancies resulting from ICSI for severe male factor infertility at their center in Belgium. Evaluation of singleton gestations (69%) revealed a prematurity rate of 7.6%, a rate of low birth weight of 10.3%, and a perinatal mortality of 13.5/1,000. These rates were reported to be similar to those in a comparable IVF patient population and slightly higher than those for spontaneous pregnancies when controlled for multiple pregnancies. Another observational study by Wennerholm et al. (12) evaluated 175 ICSI pregnancies, and described the obstetric and perinatal outcomes of the subsequent births. The rate of multiple births was 20%: none were triplets or higher. Overall, 17% of deliveries were preterm (9% for singletons). 852 Retzloff and Hornstein ICSI and birth outcomes Vol. 80, No. 4, October 2003

These investigators reported a 17% incidence of low birth weight (6.4% for singletons). These risks were reported to be lower than the published data on non-icsi IVF. These studies reported a range of mean birth weights for singleton gestations resulting from ICSI from 3,168 to 3,470 grams as compared to 3,540 grams for natural conceptions (11, 12). Despite this reassuring information, we felt that the issue of maternal age and multiple birth were not adequately addressed by these relatively small studies. Shieve et al. (13) reported on a large, population-based retrospective analysis of births in the United States to assess more completely the relationship between ART and low birth weight. This study had the power of over 42,000 ART deliveries and calculated expected odds ratios from over 3 million naturally conceived deliveries in 1997. Examining both singleton and multiple births, they determined that the overall risk for low birth weight of term infants conceived by ART was 2.6 (95% confidence interval [CI], 2.4 to 2.7) compared with the risk for infants conceived naturally. It is interesting that this increased risk was not further increased for multiple births. Although the mean birth weights for each group were not listed in the article, the percentage of singleton infants with low birth weights (13.2%) was similar to that in previously published data (6.4% to 16%) (12). When the group was stratified by ICSI and IVF, low-birth-weight infants were less common in the ICSI than the IVF group. This interesting subanalysis suggests that some aspect of ART independent of ICSI (i.e., medications, infertility history, or the IVF procedure itself) may put these infants at greater risk for low birth weight. These studies have addressed the potential effects of ICSI on obstetrical outcomes from early pregnancy to late in the gestation. The results are complicated by the patients older age and increased multiple gestations in those undergoing both IVF and ICSI. When attempts are made to control for these confounding variables, the overall rate of low birth weight appears to be the single most consistent risk to offspring from both ICSI and IVF. Intracytoplasmic sperm injection thus does not appear to impose any additional obstetrical detriment over conventional IVF to the developing fetus. CHROMOSOMAL ABNORMALITIES It has long been recognized that the prevalence of chromosomal abnormalities is increased in both azoospermic and severely oligospermic men. These aneuploidies have, as expected, heavily favored sex-chromosome abnormalities (14, 15, 16). A number of investigators have reported a twofold to 12-fold increased risk of abnormal karyotypes in infertile men (15, 17, 18), and the percentage with abnormal karyotypes may be a high as 25% among completely azoospermic men (18). Unfortunately, even morphologically normal sperm from infertile men have a twofold to threefold risk of genetic aneuploidy (19), increasing the potential for passing the affected genotype to each male offspring. The concerns about ICSI-related chromosomal aberrations were first raised in 1995 by In t Veld et al. (14) and Liebaers et al. (15) from Von Steirtegham s group in Belgium. In t Veld et al. (14) first reported on 5 (33%) of 15 pregnancies after ICSI in which abnormal sex chromosomes were identified through amniocentesis. There was a significant selection bias in this group, as it was based on a referral for advanced maternal age for prenatal diagnosis and involved only a small sample of the clinic s patients. This relatively high rate of aneuploidy, however, led Liebaers et al. (15) to report on an expanded cohort of 585 patients from the same clinic reported on by In t Veld. They found a total of six (1%) chromosomal anomalies. Five of the six abnormalities noted in this group were of the sex chromosome. Although the risk of autosomal aneuploidy was not increased, the relatively high rate of sex chromosome aberration brought to the forefront the concerns regarding sex chromosome aneuploidy in ICSI offspring. Several other investigators subsequently followed up on these initial reports and further examined the issue of sex chromosome aberrations following ICSI. To explain the correlation, Gekas et al. (19) recently explored the candidates for ICSI to assess the baseline frequency of parental chromosomal aberrations. They reported on karyotypes of 2,196 men and 1,012 women undergoing ICSI for male factor infertility from 20 French centers over a 3-year period. They found a rate of karyotypic abnormality of 6.1% in the men compared with a 3.0% prevalence of abnormalities in fertile male controls. It is interesting that 4.8% of the women partners of infertile men also showed an abnormality. This finding suggested that both members of a couple in which male factor infertility is diagnosed have an increased prevalence of chromosomal abnormalities. In addition, the results highlighted the potential existence of a chromosomal female infertility factor in some ICSI couples. The cytogenetic results from couples undergoing ICSI at the center in Belgium that first reported a successful ICSI outcome were reported sequentially. Bonduelle et al. (18) in 1995 found no abnormal maternal karyotypes out of 130 ICSI patients and no abnormal prenatal karyotypes in 100 offspring. Although there was one fetus with chromosomal mosaicism at amniocentesis, this was not confirmed at birth. The same group published a follow-up study the next year with 293 prenatal karyotypes of 423 children resulting from ICSI procedures that revealed a 0.3% rate of abnormal chromosomes (20). Parental karyotyping of the 320 pregnancies in this group revealed eight chromosomal aberrations (six of the eight chromosomal abnormalities were paternal in origin). In 1999, again expanding their series, they reported a 2.6% rate of abnormal karyotype (28 out of 1,082) in which FERTILITY & STERILITY 853

nine sex-chromosome abnormalities were found (21). Ten of these appeared to be inherited from the male. This suggested a de novo rate of karyotypic abnormality of 1% unrelated to abnormalities in the male. Another group reviewed all of the Danish ICSI pregnancies from January 1994 through July 1997 and published their report in 1999 (22). They found seven (3.3%) of 209 abnormal karyotypes on prenatal diagnosis, but interestingly, none of these involved the sex chromosomes. Six of the seven abnormalities were de novo autosomal abnormalities, four of which were in women 35 years old or greater. Age, therefore, may have played a role in the aneuploidy seen. Only one of these autosomal abnormalities was subsequently determined to be inherited. The possibility remains that even in couples whose karyotypes are normal, their gametes are disproportionately aneuploid. Several studies have documented a significant rise in the frequency of sex chromosome abnormalities in the sperm from males with abnormal spermatogenesis. These infertile men with a normal karyotype and low sperm concentration have been shown to carry a significantly increased risk of producing aneuploid spermatozoa, primarily of the sex chromosomes (23 26). These studies are consistent with the increased incidence of sex chromosome abnormalities in ICSI offspring. The causes remain independent of the ICSI process itself. Specific gene mutations in infertile men may also be passed to their offspring. It has been well established that gene-specific mutations may underlie male factor infertility. The effect of these microdeletions on offspring has only recently been assessed. Specifically, microdeletions of the Y chromosome have been of the greatest concern. In 1996, Reijo et al. (16) reported microdeletions on the long arm of the Y chromosome in the DAZ cluster of genes. They found that 13% of 89 men with nonobstructive azospermia have these deletions. Silber (27) reported at the 2001 American Society for Reproductive Medicine meeting that transmission of Y microdeletions to male progeny does not reduce the pregnancy rate. This suggests that these abnormalities may easily be incorporated into the genome of the male offspring of ICSI. This is of special concern because these microdeletions are not picked up on typical karyotype analysis and may not show up until the male offspring attempts to reproduce. It is not known whether the sons of fathers with Y chromosome microdeletions will have the same reproductive phenotype and thus require these same procedures themselves once they reach maturity. Taken together, these studies imply that offspring from ICSI indeed carry an increased rate of chromosomal aberrations. These abnormalities seem to correlate with the underlying parental risk of abnormality and not with the ICSI procedure itself. Therefore, genetic counseling must incorporate pre-icsi screening of couples. Discussion of both macro-abnormalities (karyotype) and micro-abnormalities (gene microdeletions) should be included. CONGENITAL MALFORMATIONS Several centers have evaluated the implication that congenital malformations may arise secondary to ICSI. A large series representing all children born from ICSI from a clinic in Brussels has been analyzed (Table 1). The first report, by Bonduelle et al. (18), prospectively followed 130 ICSI offspring and compared the results with those for 130 IVF offspring. The rate of congenital malformations was 3.2% for the ICSI group vs. 4.6% for the IVF group. Reassuringly, the ICSI malformations appeared to be distributed evenly without clustering in any organ system. These investigators followed their initial report with a second one in 1996 of 423 children born from ICSI (20). They performed physical examinations and psychomotor evaluations within 2 months of birth, at 1 year, and at 2 years. Major malformations in the study were defined as those causing functional impairment or requiring surgical correction; minor malformations were defined as those with a prevalence of less than or equal to 4% in children in the same racial group. They reported an overall rate of major malformations of 3.3% (1.9% in singletons, 4.8% in twins, and 6.7% in triplets) in these ICSI offspring. A rate of minor malformations of 20.5% was noted; 57% were associated with just three malformations: angiomas, congenital nevi, and umbilical hernias. The investigators concluded that the malformations discovered were within the expected range relative to those offspring from other assisted reproductive treatments and population registries. Kurinczuk and Bower (28) reanalyzed the 1996 Belgian data. They alleged that the Belgian group inaccurately compared data on ICSI births with data from various birth registries that used different definitions of what constitutes a major birth defect. They claimed that this resulted in an inaccurate comparison and possibly an underestimation of the comparative prevalence of major birth defects in the ICSI offspring. Therefore, Kurinczuk and Bower reclassified the malformations reported by the Belgian group according to the criteria of British Paediatric Association (ICD-9) criteria and the U.S. Centers for Disease Control and Prevention (CDC). They then used the Western Australian Birth Defect Registry as their control group for comparison. They found that with their definition of malformations, the Belgian data would have reported a 7.4% rate of major malformations and a 0.7% rate of minor malformations. The controls from Western Australia showed a 3.8% rate of major malformations and a 0.5% rate of minor malformations. Comparing the recalculated malformation rate of ICSI vs. controls, the data revealed that infants born after ICSI were twice as likely as infants from the Australian birth registry to have a major birth defect (odds ratio [OR], 2.03; 854 Retzloff and Hornstein ICSI and birth outcomes Vol. 80, No. 4, October 2003

TABLE 1 Congenital malformations. Malformation Country Author Year Study population Major Minor Belgium Bonduelle 1996 ICSI (n 423) 3.3% 20.5% Australia/Belgium Kurinczuk 1997 ICSI (n 420) Belgian 7.4% 0.7% Natural (n 100,454) Australian 3.8% 0.5% Belgium Bonduelle 1999 ICSI (n 1,987) 2.3% N/A IVF (n 130) 4.6% N/A Europe ESHRE 1998 ICSI (n 807) 2.0% N/A Denmark Loft 1999 ICSI (n 730) 2.2% 1.2% England Sutcliffe 1999 ICSI (n 123) 4.9% 11% Natural (n 123) 4.1% 7.3% UK registry 5% N/A Sweden Wennerholm 2000 ICSI (n 1,139) 3.2% 3.5% USA Palermo 2000 ICSI (n 3,573) 1.1% 0.8% IVF (n 3,277) 1.7% 1.3% Germany Ludwig 2000 ICSI (n 2,809) 9.1% N/A Registry 7.2% N/A Australia Hansen 2002 ICSI (n 301) 8.6% 0.3% IVF (n 837) 9.0% 0.8% Natural (n 4,000) 4.2% 0.6% Netherlands Anthony 2002 IVF ICSI (n 4,224) 0.7% 1.3% Registry (n 314,605) 0.5% 1.1% Note: N/A, Not available. Retzloff. ICSI and birth outcomes. Fertil Steril 2003. 95% CI, 1.4 2.93). The minor malformation odds ratio was only 1.49 (95% CI, 0.48 4.66), which was not statistically significant. Specifically, the authors found an increased risk of cleft palate and diaphragmatic hernia in the ICSI population as compared with controls. However, only two patients with cleft palate and one patient with diaphragmatic hernia were identified in the Belgian series. In response to the paper by Kurinczuk and Bower, Bonduelle et al. (29) recalculated their 1996 data. They concluded that the main difference between their data and that of the Australian group had to do with the classification of the cardiac malformations. They claimed that the atrial septal defects and ventricular septal defects, which were classified as major malformations in the Australian paper, should not have been so classified. They pointed out that the majority of the cardiac malformations in these children were discovered on routine ultrasound examination as part of their scientific study. These studies would not have been done routinely in controls and should therefore have not been used for comparison. In addition, the patients with atrial septal defects had no evidence of a shunt at 6 months of life, suggesting these abnormalities were minor according to their original definition. In two of the three patients with ventricular septal defects, the defects closed spontaneously by 1 year of age. In addition, in all six of the patients with a patent ductus arteriosis, the abnormality closed by 6 months of age. When the Belgians recalculated the cardiac abnormalities in the ICSI patients vs. those in controls, they noted no difference in the incidence of these abnormalities. When they recalculated the incidence of all major malformations without the cases of atrial septal defects, they again found no statistically significant difference as compared with controls. The reporting of the Belgian data, the recalculation of their data by the Australian group, and finally the rebuttal from the Belgian group highlight several important points in the design of these studies. First, it is likely that the follow-up of malformations in patients who have undergone ICSI is more rigorous than that in controls. The use of ultrasound to assess abnormalities in ICSI offspring after hospital discharge is not routinely performed and therefore not included in birth registries. This extra testing imparts a biased outcome assessment. Physical examinations are often not blinded, and the examiner may be influenced by the child s history. In addition, there is an excess of multiple gestations in the Belgian data (48%) vs. the Western Australian controls (2.7%). There are clearly more malformations associated with multiple gestations than with singleton pregnancies. Advanced maternal age has also been associated with increased chromosomal aberrations and congenital malformations. The mean maternal age was approximately 32 years in the Belgian ICSI group as compared with 28 years in the Australian control group. FERTILITY & STERILITY 855

There is also likely to be an increased prevalence of chromosome abnormalities in the offspring of children who went through the ICSI procedure, as most of these were done for severe male factor infertility. As noted above, the infertile men themselves have an increased risk of karyotypic abnormalities. Lastly, the absence of a control group of IVF alone further confounds any comparison. All of these factors suggest the introduction of a selection bias that could cast doubt on the validity of the conclusions of most of these studies. In 1998, the European Society for Human Reproduction and Embryology (30) reviewed the follow-up of 807 children born from ICSI through 1993 reported from 90 centers in 24 countries. They found a rate of major malformations of 2.0%, which was not different from that in the general population. When the Danish consortium (22) looked at the malformation rates from ICSI procedures done in Denmark between 1994 and 1997, they found a rate of major malformations of 2.7% and a rate of minor malformations of 1.2%. The malformations were classified by ICD-10 from the Danish National Board of Health. These results, too, were considered in line with other published ICSI outcomes (11, 12, 21). A study reported from England by Sutcliffe et al. (31), used a case-control model to assess 123 ICSI offspring and 123 controls. Information was gathered by means of a questionnaire for couples. They found six major malformations (4.9%) in the ICSI group vs. five (4.1%) in the controls. The rate of minor malformations was 11.3% in ICSI offspring vs. 7.3% in controls. This difference was not statistically significant. Using controls from the Swedish Medical Birth Registry and the Registry of Congenital Malformations, Wennerholm et al. (32) examined 1,139 infants born from ICSI procedures in Sweden. These investigators stratified their data by delivery hospital, year of birth, and maternal age, and they reviewed medical records and the birth registry from Sweden. They found an overall OR for ICSI offspring for any major or minor malformations of 1.75 (95% CI, 1.19 2.58). The OR after stratification for singletons and twins was only 1.19 (95% CI, 0.79 1.81), which was not statistically significant when compared with controls. The only specific malformation they found at increased prevalence was hypospadias. The relative risk for this specific malformation was 3.0 (95% CI, 1.09 6.5). The investigators believed that this malformation in affected male infants might have been associated with paternal subfertility. Others have implicated progesterone support used in IVF as the cause of an increased incidence of hypospadias in IVF offspring (33). The investigators also noted that a substantial number of congenital malformations were not included in the medical birth registry, suggesting that malformations in the ICSI offspring may have been overrepresented as compared with controls. Palermo et al. (34) reviewed the Cornell experience with ICSI. His group reviewed the malformation rates in ICSI offspring compared with those from non-icsi, male factor infertility IVF patients. The data in this study came from all of the 3,573 cycles of ICSI and 3,277 cycles of IVF associated with male factor infertility at Cornell from September 1993 to March 1999. In the ICSI group, 38 (1.8%) of 2,059 had malformations compared with 54 (3.0%) of 1,796 in the IVF group. They noted that 53% of the ICSI malformations observed occurred in multiple gestations. Most striking in this study was the significantly increased rate of malformations in offspring of conventional IVF as compared with ICSI offspring (P.05). Ludwig et al. (35) reported on the German ICSI experience at the American Society for Reproductive Medicine 2001 meeting. They looked at 2,809 pregnancies following ICSI in 59 German IVF centers in a prospective fashion and classified malformations according to the European Registry of Congenital Anomalies and Twins (EUROCAT). They found a rate of major malformations in controls of 7.2% from a national birth registry and a rate in ICSI offspring of 9.1%. This difference was not statistically significant. They found no clustering of defects. Specifically, there was no increase in hypospadias. They did find that maternal risk (i.e., occupational exposure or family history of malformation) statistically increased the malformation rate independent of ICSI. Multivariate analysis confirmed that ICSI was not an independent risk factor for congenital malformations. Most recently, Hansen et al. (36) reported a case-control study of ART-associated birth defects in Western Australia. These investigators identified four common areas of concern in the existing literature: surveillance, data comparison (controls), definition of birth defects, and sample size. They reported that they overcame these obstacles in their study by collecting data without reference to the mode of conception, using the same source of data for each group, and comparing a large group of ICSI (n 301), IVF (n 837), and natural conceptions (n 4,000). The size of the samples of patients was relatively small for both ICSI and IVF offspring relative to the size of the control group. The demographics of the study populations also differed in several respects. The women in the natural conception group were younger than those in either the ICSI or IVF group (28.2 years vs. 32.6 years vs. 34.1 year, respectively) and had an increased parity. The investigators reported a twofold increase in major birth defects in both IVF (8.6%) and ICSI (9.0%) offspring as compared with offspring of natural conceptions (4.2%). This difference held despite controlling for preterm delivery, maternal age, multiple gestation, and early termination of pregnancy for anomalies. This observational study attempted to control for diagnostic surveillance bias; however, the investigators conceded that diagnostic studies to determine malformations may not have been equal among the three 856 Retzloff and Hornstein ICSI and birth outcomes Vol. 80, No. 4, October 2003

groups. The investigators attempted to avoid this observational bias by employing a single pediatrician to review the list of anomalies in an effort to determine those anomalies that would have been detected independent of intensity of the surveillance used. This approach was criticized in a correspondence by Sutcliffe et al. (37), who pointed out that the results rested fundamentally on this assumption that the selected pediatrician could discern whether the observed congenital anomalies were more likely to occur in ICSI offspring. In addition, the classification chosen to distinguish minor and major malformations does not allow one to make this distinction easily. Interesting as well in the Hansen study, both IVF and ICSI malformations were increased, a finding at variance with the preponderance of reports in the world literature on this subject (38 40). This fact is suggestive of other possible causes for the outcome, such as infertility diagnostic confounders or treatment-related effects. Steinkampf and Grifo (41) reported that the data of the Society of Assisted Reproductive Technology from 1996 to 2000 reveals a 1.9% incidence of major birth defects among ART offspring in the United States and Canada. This large database survey remains reassuring to both patients and their physicians. In an attempt to address the shortcomings of prior studies, Anthony et al. (42) compared the congenital malformation rates for offspring of IVF (n 4,224) and offspring of natural conceptions (n 314,605) in the Netherlands. The strengths of this study were the use of the same database for both cases and controls and a cohort that was large enough to allow subgroup analysis. Their results showed a slight increased risk of any malformation (major and minor) for IVF offspring (OR, 1.20; 95% CI, 1.01 1.43) compared with the natural conception group. This difference was, however, no longer statistically significant when confounding variables such as maternal age, parity, and ethnicity were controlled (OR, 1.03; 95% CI, 0.86 1.23). Unfortunately, the investigators were not able to separate out the ICSI children from the IVF study population. The fact that there was only a small increase in the malformation rate in a variety of organ systems makes it much less likely that some procedural aspect of IVF was responsible. The lack of a correlation between malformations and IVF treatment was further substantiated by the fact that the differences could be completely accounted for by maternal risk factors. Thus, the issue regarding congenital malformations and ICSI remains clouded by the inherent biases of each observational study. The vast majority of studies have found no increased rate of malformations associated with ICSI. In addition, it is reassuring that there is no clustering of any single specific major malformation. The inability to randomize treatments in a prospective fashion and the inherent difficulty in finding ideal controls limits the capacity to determine potential minor differences in the ICSI population given the present state of knowledge. DEVELOPMENTAL ABNORMALITIES In 1998, Bowen et al. (43) compared 89 ICSI offspring with 84 IVF offspring and 80 infants conceived naturally. This prospective study, which looked at Bayley scales of infant development at 1 year, emphasized motor and mental development. The investigators found that mental development of most ICSI offspring was within the normal range; however, scores on the Bayley mental development scale were lower for the ICSI offspring than for the IVF and natural conception controls. When the investigators further analyzed this finding, they found that the differences were statistically significant in boys but not in girls. Furthermore, 17% of ICSI babies had delayed development in memory, language, and problem-solving skills compared with only 2% in the IVF offspring and 1% in the naturally conceived infants. A major limitation of the study was the use of examiners who were not blinded to the conceptual origin of the children. In addition, the IVF offspring and naturally conceived controls were recruited initially as part of a separate study, and there were demographic differences between the groups, such as parental occupation skill level, education level, and age. The investigators emphasized that these findings, although supporting the need for follow-up, may not predict school performance or mental developmental at school age. Bonduelle et al. (44) from Belgium assessed mental development in 201 ICSI offspring as compared with 131 IVF offspring and 1,238 normal Dutch children. This was a prospective study with follow-up of 2 years. Important in this study was the use of a single, blinded pediatrician to test all of the ICSI and IVF children. Unfortunately, follow-up at 2 years was somewhat low, at 25%. The overall Bayley mental development scores were similar for all groups. Last, Sutcliffe et al. (31) looked at the mental development at 12 to 24 months of age of 123 singleton children from ICSI and 123 conceived naturally. A single observer assessed all children and follow-up was 90%. These children were matched for social class, maternal education level, religion, sex, and race but not for maternal age. Griffiths scales were used as an objective measurement of mental development. They found no difference in average mental age and no difference in the overall development as expressed by the Griffiths quotient. In addition, there was no difference in four of five Griffiths subquotients: locomotor, personal and social, hearing and speech, and performance. Offspring of ICSI, however, performed worse in high-level hand-eye coordination tasks, despite adjustment for multiple gestations. The investigators commented that this difference in hand-eye coordination is unlikely to be of functional significance. FERTILITY & STERILITY 857

Developmental assessment of ICSI children has been sparse at best, with only short-term follow-up of the children. The data as stated are quite dichotomous, and no consistent conclusions can be drawn from them. It does appear that there are no major developmental delays, either motor or mental, but to decipher potential minor abnormalities a larger concerted prospective effort of many institutions will need to be undertaken. CONCLUSIONS Despite conflicting data and significant confounders and biases in some of the studies, several conclusions can be drawn from the available data. There appears to be no increase in prematurity and perinatal mortality in ICSI pregnancies compared with IVF pregnancies when studies are appropriately controlled. Some data suggest a higher incidence of low birth weight (1,500 to 2,500 grams) among both IVF and ICSI offspring even after adequately controlling for multiple gestations and maternal age. This difference, however, has not been shown to be clinically significant. There is likely an increased rate of sex-chromosome abnormalities in offspring from ICSI pregnancies. This finding appears related to either inherited paternal karyotypic abnormalities or abnormal spermatogenesis and not to de novo acquisition. In addition, a high percentage of azoospermic/severe oligospermic men have microdeletions of the Y chromosome. These patients should be informed that such deletions are passed on to ICSI male progeny, possibly resulting in their infertility later in life. There is probably no significant increase in congenital malformations. However, the large, well-designed study of Hansen et al. (36) as well as the reports of imprinting defects are cautionary in this regard. It is reassuring that there appears to be no increase in malformations of any specific organ system. Several studies have looked at both IVF and ICSI without finding any consistent abnormalities. All of these observational studies suffer from the inherent bias of retrospective analysis. Despite attempts at controlling for factors such as age, parity, and higher-order gestations, significant inherent biases remain, making it impossible to draw any firm conclusions from these data. Although there remains some uncertainty, the overwhelming majority of studies are reassuring in their findings of no specific increased risk for congenital malformations after ICSI. The effect on psychomotor development remains difficult to assess in children born from ICSI procedures. Follow-up studies were performed at early ages, and the predictive value of such early-age testing for school performance and development later in life remains questionable. Nonetheless, it appears that psychomotor development of such children is probably normal. It is possible that mental development is delayed in some categories. Here it is clear that future studies with longer follow-up are sorely needed. Finally, examination of the available data from thousands of ICSI children reveals that it is largely a safe procedure. Further study is warranted and must stand the test of separation from bias. This is a daunting task in the field of assisted reproduction. In the meantime, patient education regarding chromosomal testing, specific Y microdeletion testing in severely oligospermic men, and counseling regarding risk of low birth weight and the possibility of congenital malformations is paramount. With the evidence as it stands today, ICSI should continue to be offered to couples for whom no other method of assisted reproduction can offer success. References 1. Palermo G, Joris H, Devroey P, Van Steirteghem AC. Pregnancies after intracytoplasmic injection of single spermatozoon into an oocyte. Lancet 1992;340:17 8. 2. Patrizio P. Intracytoplasmic sperm injection (ICSI): potential genetic concerns. Hum Reprod 1995;10:2520 23. 3. Dumoulin JCM, Coonan E, Bras M, Bergers-Janssen JM, Ignoul- Vanvuchelen CM, van Wissen LCP, et al. Embryo development and chromosomal anomalies after ICSI: effect of the injection procedure. Hum Reprod 2001;16:306 12. 4. Sutcliffe AG, D Souza SW, Cadman J, Richards B, McKinlay IA, Lieberman B. Minor congenital anomalies, major congenital malformations and development in children conceived from cryopreserved embryos. Hum Reprod 1995;10:3332 37. 5. DeBaun M, Niemitz E, Feinberg A. Association of in vitro fertilization with Beckwith-Wiedemann syndrome and epigenetic alterations of LIT1 and H19. Am J Hum Genetics 2003;72:156 60. 6. Maher E, Brueton L, Bowdin S, Luharia A, Cooper W, Cole T, et al. Beckwith-Wiedemann syndrome and assisted reproduction technology (ART). J Med Genet 2003;40:62 4. 7. Cox G, Burger J, Lip V, Mau U, Sperling K, Wu B, et al. Intracytoplasmic sperm injection may increase the risk of imprinting defects. Am J Hum Genet 2002;71:162 64. 8. Manning M, Lissens W, Bonduelle M, Camus M, De Rijcke M, Liebaers I, et al. Study of DNA-methylation patterns at chromosome 15q11-q13 in children born after ICSI reveals no imprinting defects. Mol Hum Reprod 2000;6:1049 53. 9. Manning M, Lissens W, Liebaers I, Van Steirteghem A, Weidner W. Imprinting analysis in spermatozoa prepared for intracytoplasmic sperm injection (ICSI). Int J Androl 2001;24:87 94. 10. ASRM/SART. Assisted reproductive technology in the United States: 1999 results generated from the American Society for Reproductive Medicine/Society for Assisted Reproductive Technology Registry. Fertil Steril 2002;78:918 31. 11. Wisanto A, Magnus M, Bonduelle M, Liu J, Camus M, Tournaye H, et al. Obstetric outcome of 424 pregnancies after intracytoplasmic sperm injection. Hum Reprod 1995;10:2713 18. 12. Wennerholm UB, Bergh C, Hamberger L, Nilsson L, Reismer E, Wennergren M, et al. Obstetric and perinatal outcome of pregnancies following intracytoplasmic sperm injection. Hum Reprod 1996;11: 1113 19. 13. Schieve L, Meikle S, Ferre C, Petersen H, Jeng G, Wilcox L. Low and very low birth weight in infants conceived with use of assisted reproductive technology. N Engl J Med 2002;346:731 37. 14. In t Veld P, Brundenburg H, Verhoeff A, Dhant M, Los F. Sex chromosomal abnormalities and intracytoplasmic sperm injection. Lancet 1995;346:773. 15. Liebaers I, Bonduelle M, Van Assche E, Devroey P, Van Steirtegheim A. Sex chromosome abnormalities after intracytoplasmic sperm injection. Lancet 1995;346:1095. 16. Reijo R, Alagappan RK, Patrizio P. Severe oligospermia resulting from microdeletions of azospermiafactor gene on Y chromosome. Lancet 1996;347:1290 93. 17. Johnson M. Genetic risks of intracytoplasmic sperm injection in the treatment of male infertility: recommendations for genetic counseling and screening. Fertil Steril 1998;70:397 411. 18. Bonduelle M, Legein J, Devde MP, Buysse A, Schietecotte J, Wisanto A, et al. Comparative follow-up study of 130 children born after intracytoplasmic sperm injection and 130 children born after in-vitro fertilization. Hum Reprod 1995;12:3327 31. 19. Gekas J, Thepot F, Turleau C, Sittroi JP, Dadoune JP, Wasels R, et al. Chromosomal factors of infertility in candidate couples for ICSI: an 858 Retzloff and Hornstein ICSI and birth outcomes Vol. 80, No. 4, October 2003

equal risk of constitutional aberrations in women and men. Hum Reprod 2001;16:82 90. 20. Bonduelle M, Legein J, Buysse A, Van Assche E, Wisanto A, Devroey P, et al. Prospective follow-up study of 423 children born after intracytoplasmic sperm injection. Hum Reprod 1996;11:1558 64. 21. Bonduelle M, Camus M, De Vos A, Staessen C, Tournaye H, Van Assche E, et al. Seven years of intracytoplasmic sperm injection and follow-up of 1987 subsequent children. Hum Reprod 1999;14(Suppl): 243 64. 22. Loft A, Peterson K, Erb K, Mikkelsen AL, Grinsted J, Hald F, et al. A Danish national cohort of 730 infants born after intracytoplasmic sperm injection (ICSI) 1994 1997. Hum Reprod 1999;14:2143 48. 23. Ohashi Y, Mihari N, Honda H, Samura O, Ohama K. High frequency of XY disomy in spermatozoa of severe oligospermic men. Hum Reprod 2001;16:703 8. 24. Rubio C, Gil-Salom M, Simon C, Vidal F, Rodrigo L, Minguez Y, et al. Incidence of sperm chromosomal abnormalities in a risk population: relationship with sperm quality and ICSI outcome. Hum Reprod 2001; 16:2084 92. 25. Vegetti W, Van Assche E, Frias A, Verheyen G, Bianchi M, Bonduelle M, et al. Correlation between semen parameters and sperm aneuploidy rates investigated by fluorescence in-situ hybridization in infertile men. Hum Reprod 2000;15:351 65. 26. Shi Q, Martin R. Aneuploidy in human spermatozoa: FISH analysis in men with constitutional chromosomal abnormalities, and infertile men. Reproduction 2001;121:655 66. 27. Silber SJ. Spontaneous pregnancy in couples with very severe oligospermia ( 0.5 10 sperm): implications for transmission of Y chromosome deletions [abstract no. O-26]. In: Fertility and Sterility Abstracts of the Scientific Oral and Poster Sessions. Orlando, FL: American Society for Reproductive Medicine, 2001:76. 28. Kurinczuk JJ, Bower C. Birth defects in infants conceived by intracytoplasmic sperm injection: an alternative interpretation. BMJ 1997;315: 1260 64. 29. Bonduelle M, Devroey P, Liebaers I, Van Steirtegham A. Commentary: Major defects are overestimated. BMJ 1997;315:1265 66. 30. ESRHE Task Force on Intracytoplasmic Sperm Injection. Assisted reproduction by intracytoplasmic sperm injection: a survey on the clinical experience in 1994 children and the children born after ICSI, carried out until 31 December 1993. Hum Reprod 1998;13:1737 46. 31. Sutcliffe AG, Taylor B, Thornton S, Grudzinskas JF, Lieberman BA. Children born after intracytoplasmic injection: population control study. BMJ 1999;328:704 5. 32. Wennerholm UB, Bergh C, Hamberger L, Lundin K, Nilsson L, Wikland M, et al. Incidence of congenital malformations in children born after ICSI. Hum Reprod 2000;15:944 48. 33. Silver RI, Rodriguez R, Chang TS, Gearhart JP. In vitro fertilization is associated with an increased risk of hypospadias. Urology 1999;161: 1954 57. 34. Palermo GD, Neri QV, Haripraeshod JJ, David OK, Veeck LL, Rosenwaks Z. ICSI and its outcome. Semin Reprod Med 2000;18:161 69. 35. Ludwig M, Queiber-Luft A, Katalinic A. Pregnancy and birth following ICSI: a prospective, controlled multicentric German trial [abstract no. O-35]. In: Fertility and Sterility Abstracts of the Scientific Oral and Poster Sessions. Orlando, FL: American Society for Reproductive Medicine, 2001:S14. 36. Hansen M, Kurinczuk J, Bower C, Webb S. The risk of major birth defects after intracytoplasmic sperm injection and in vitro fertilization. N Engl J Med 2002;346:725 30. 37. Sutcliffe AG, Bonduelle M, Taylor BW. To the editor. N Engl J Med 2002;347:1450. 38. Friedler S, Mashiach S, Laufer N. Births in Israel resulting from in-vitro fertilization/embryo transfer, 1982 1989: National Registry of the Israeli Association for Fertility Research. Hum Reprod 1992;7:1159 63. 39. Olivennes F, Kerbrat V, Rufat P, Blanchet V, Fanchin R, Frydman R. Follow-up of a cohort of 422 children aged 6 13 years conceived by in vitro fertilization. Fertil Steril 1997;67:284 89. 40. Koivurova S, Jartikainen A, Gissler E, Sovio U, Jarvelin M. Neonatal outcome and congenital malformations in children born after in-vitro fertilization. Hum Reprod 2002;17:1391 98. 41. Steinkampf MP, Grifo J. To the editor. N Engl J Med 2002;347:1449. 42. Anthony S, Buitendijk SE, Dorrepaal CA, Lindner K, Braat DD, den Ouden AL. Congenital malformations in 4224 children conceived after IVF. Hum Reprod 2002;17:2089 95. 43. Bowen JR, Gibson FZ, Leslie GI, Saunders DM. Medical and developmental outcome of 1 year for children conceived by intracytoplasmic sperm injection. Lancet 1998;351:1529 34. 44. Bonduelle M, Javis H, Hofmans K, Liebars I, Van Steirtegham A. Mental development of 201 ICSI children at 2 years of age. Lancet 1998;351:1553. FERTILITY & STERILITY 859