Birth defects in infants conceived by intracytoplasmic sperm injection: an alternative interpretation
BMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7118.1260 (Published 15 November 1997) Cite this as: BMJ 1997;315:1260
- a TVW Telethon Institute for Child Health Research, PO Box 855, West Perth, WA 6872, Australia
- b Western Australian Birth Defects Registry, King Edward Memorial Hospital, Bagot Road, Subiaco, WA 6008, Australia
- Correspondence to: Dr Kurinczuk
- Accepted 24 June 1997
Abstract
Objective: To test the hypothesis that liveborn infants conceived by intracytoplasmic sperm injection are at an increased risk of having a major birth defect.
Design: Reclassification of the birth defects reported in infants born after intracytoplasmic sperm injection in Belgium and comparison with prevalence estimated in Western Australian population by means of same classification system.
Setting and subjects: 420 liveborn infants who were conceived after intracytoplasmic sperm injection in Belgium and 100 454 liveborn infants in Western Australia delivered during the same period.
Main outcome measures: Estimates of birth prevalence of birth defects and comparisons of odds ratios between cohort conceived after intracytoplasmic sperm injection and Western Australian infants.
Results: Infants born after intracytoplasmic sperm injection were twice as likely as Western Australian infants to have a major birth defect (odds ratio 2.03 (95% confidence interval 1.40 to 2.93); P=0.0002) and nearly 50% more likely to have a minor defect (1.49 (0.48 to 4.66); P=0.49). Secondary data-led analyses, to be interpreted with caution, found an excess of major cardiovascular defects (odds ratio 3.99), genitourinary defects (1.33), and gastrointestinal defects (1.84), in particular cleft palate (5.11) and diaphragmatic hernia (7.73).
Conclusions: These results do not confirm the apparently reassuring results published by the Belgian researchers of intracytoplasmic sperm injection. Further research is clearly required. Meanwhile, doctors practising intracytoplasmic sperm injection should bear this alternative interpretation in mind when they counsel couples and obtain informed consent for the procedure.
Key messages
Reports of data from Belgium concluded that there was no increase in the occurrence of major birth defects among infants conceived by intracytoplasmic sperm injection
However, the Belgian researchers used a narrow definition of what constituted a major birth defect, and they might have underestimated the comparative prevalence of major defects
We reclassified the birth defects in the infants born after intracytoplasmic sperm injection using a standard classification system and compared the results with data from Western Australia that were classified with the same system
We found that the infants born after intracytoplasmic sperm injection were twice as likely to have a major birth defect and nearly 50% more likely to have a minor defect
Further research is required to elucidate these results, but, meanwhile, they should be born in mind when counselling infertile couples about intracytoplasmic sperm injection
Introduction
Intracytoplasmic sperm injection, the selection and injection of a single spermatozoon into an oocyte, is probably the most important development in assisted reproduction since the birth of the first “test tube baby” in 1978. This procedure offers, for the first time, real prospects of genetic parenthood for men with profound oligospermia and, by means of testicular biopsy and epididymal aspiration, even for those men with azoospermia. There are, however, several theoretical concerns about the safety of intracytoplasmic sperm injection and potential risks for the offspring.1 2 3 These concerns can be categorised broadly into four groups: the risks of using sperm that potentially carry genetic abnormalities, the risks of using sperm with structural defects, the potential for mechanical and biochemical damage and of introducing foreign material into the oocyte, and the risks associated with overcoming processes of natural selection by injecting a single spermatozoon.
This procedure was developed by a group at the Brussels Free University.4 This group had the foresight to institute, at the outset, follow up of the infants conceived by intracytoplasmic sperm injection at their centre. Their findings represent the first substantive data set available with which we can start to investigate the concerns about the procedure. As their number of infants has increased, they have published their findings in a series of papers.5 6 7 8 9 To date, they have closely examined 76% of the 420 infants born alive between April 1991 and September 1994 after intracytoplasmic sperm injection at their centre and have collected some information about the remainder.9 They classified birth defects as follows: “Malformations that generally cause functional impairment or require surgical correction were defined as major, and the remaining malformations were considered minor.” They concluded that the prevalence of major birth defects in liveborn infants of 3.3% was “within the expected range.”9
As the Belgian group did not have any specific data for comparison, they used published data from elsewhere in the world to compare with their findings. However, the problem with these comparisons is that different definitions of what constitutes a major birth defect were used in the comparison data. For example, the Belgian group cited data for births after in vitro fertilisation from the Australian National Perinatal Statistics Unit.9 10 However, this unit uses the five digit British Paediatric Association's ICD-9 system to classify birth defects and registers only those birth defects diagnosed by 28 days of age.10 11 This classification system has a much broader definition of what constitutes a major defect than does the classification used by the Belgian group. It is therefore possible that the Belgian group underestimated the comparative prevalence of major birth defects in their cohort of infants.
The aim of our study was to test the hypothesis that the infants born alive after intracytoplasmic sperm injection in Belgium are at an increased risk of having a major birth defect. To do this we reclassified the Belgian data according to the methods used by the Western Australian Birth Defects Registry and then compared the results with population data from the registry.12 13
Methods
The Western Australian Birth Defects Registry receives notifications of birth defects diagnosed in Western Australian children aged up to 6 years.12 13 Notifications of birth defects diagnosed in spontaneous abortions, terminations of pregnancy, and stillbirths are also received, although they were not included in our analysis as the comparison was with only live births after intracytoplasmic sperm injection. Multiple sources of notification are used, and data, which have been validated on several occasions, are available for births from 1980 onwards.14 15 16 17 For the purposes of the registry, birth defects are defined as abnormalities probably of prenatal origin and include structural, chromosomal, genetic, and biochemical defects.12 Each defect (up to a maximum of 10 defects per child) is coded according to the five digit British Paediatric Association's ICD-9 system and is also coded as major or minor according to a method devised by the Centers for Disease Control in the United States.11 18
A list of the birth defects reported by the Belgian group in the 420 infants born alive after intracytoplasmic sperm injection was given to the coding officer at the Western Australian Birth Defects Registry, who does all the registry's coding and has done so since 1989.9 The coding officer, who was not aware of the Belgian categorisation of the defects, coded the defects as major, minor, or not included, according to the registry criteria. When there was insufficient information available to classify the defect according to the registry criteria—for example, patent ductus arteriosus—the least severe classification was chosen (see footnotes to Table 1).
Birth defects observed in 420 liveborn infants conceived by intracytoplasmic sperm injection in Belgium classified according to Belgian classification9 and to classification of Western Australian Birth Defects Registry
Statistical analysis
We compared the reclassified Belgian data with the prevalence of birth defects diagnosed by the age of 1 year that were reported to the Western Australian Birth Defects Registry for liveborn infants delivered in 1991–4 inclusive. The primary analysis compared the birth prevalence of major and minor defects overall. Comparisons of the occurrence of system specific defects were carried out as secondary data-led analyses. We calculated birth prevalence and the odds ratios for birth prevalence (and their 95% confidence intervals) using the methods of Clayton and Hills.19 We calculated two tailed P values using the standard χ2 test for association.
Results
Table 1 details all the birth defects (major and minor) observed in the infants born alive after intracytoplasmic sperm injection in Belgium, together with the classification of the defects according to both the Belgian and Western Australian systems of classification. Under the Belgian classification system, 14 of the 420 infants (3.33% (95% confidence interval 1.99% to 5.55%)) were defined as having major birth defects and 84 (20.0% (16.4% to 24.1%)) were regarded as having minor defects. Under the Western Australian classification, 31 of the 420 infants (7.38% (5.04% to 10.31%)) were classified as having major defects and three (0.71% (0.23% to 2.19%)) were classified as having minor defects.
During 1991-4, there were 100 454 live births in Western Australia, of which 2.7% were multiple births.20 The infants were born to mothers with a mean age of 27.9 years (range 12–49 years). In this period, 3800 liveborn infants were notified as having major defects diagnosed by the age of 1 year. This represents a birth prevalence of 3.78% (95% confidence interval 3.67% to 3.90%) for major birth defects, compared with the prevalence of 7.38% in the infants born after intracytoplasmic sperm injection. The prevalence odds ratio for this comparison was 2.03 (1.40 to 2.93; P=0.0002), indicating that the infants born after intracytoplasmic sperm injection were twice as likely to have a major birth defect diagnosed as were the Western Australian infants. We recalculated the figures for the infants born after intracytoplasmic sperm injection assuming that the three ventricular septal defects not associated with other cardiac defects were trivial and therefore not classified as major (see footnotes to Table 1). This gave a prevalence of 6.67% (4.64% to 9.49%) and a prevalence odds ratio of 1.82 (1.24 to 2.67; P=0.0023).
During 1991-4, 484 of the Western Australian infants were notified as having a minor defect diagnosed by 1 year of age, which represents a birth prevalence of 0.48% (0.44% to 0.53%). Compared with the prevalence of 0.71% in the infants born after intracytoplasmic sperm injection, this gave a non-significant prevalence odds ratio of 1.49 (0.48 to 4.66; P=0.49).
Table 2 compares the birth prevalence of major cardiovascular, genitourinary, and gastrointestinal defects (according to the Australian classification) in the infants born after intracytoplasmic sperm injection and the Western Australian infants. The former infants showed a greater prevalence of major cardiovascular defects than did the Western Australian infants (3.33% (1.99% to 5.55%) v 0.67% (0.62% to 0.72%)), representing a fivefold excess risk (odds ratio 5.12 (2.99 to 8.77); P<0.0001). Even after excluding the three infants with ventricular septal defects that may have been trivial, there was still evidence of a fourfold excess risk (odds ratio 3.99 (2.18 to 7.30); P<0.0001).
Prevalence of major birth defects (according to classification system of Western Australian Birth Defects Registry) observed in 420 liveborn infants conceived by intracytoplasmic sperm injection in Belgium compared with population figures from the registry
The infants born after intracytoplasmic sperm injection also showed a slightly greater prevalence of major genitourinary defects than did the Western Australian infants (1.43% v 1.08%), representing a non-significant 30% excess risk (odds ratio 1.33 (0.59 to 2.98); P=0.489).
Overall, the infants born after intracytoplasmic sperm injection showed a greater prevalence of major gastrointestinal defects than did the Western Australian infants (0.95% v 0.52%), with a non-significant associated odds ratio of 1.84 (0.68 to 4.94; P=0.23). However, two of these infants with such birth defects had a cleft palate, giving a prevalence of 0.48% (0.12% to 1.88%) compared with only 0.09% (0.07% to 0.11%) in the Western Australian population. This suggests that the infants born after intracytoplasmic sperm injection had a fivefold excess risk of cleft palate (odds ratio 5.11 (1.26 to 20.80); P=0.023). One of the infants born after intracytoplasmic sperm injection had a diaphragmatic hernia, giving a prevalence of 0.24% (0.03% to 1.67%) compared with a prevalence of 0.031% (0.02% to 0.04%) in the Western Australian infants (odds ratio 7.73 (1.05 to 56.76); P=0.044).
Discussion
The findings of our reclassification and reanalysis are rather less reassuring than the conclusions published by the Belgian researchers, who reported no increase in the prevalence of birth defects in infants born after intracytoplasmic sperm injection.9 Our results suggest that these infants were twice as likely to have a major birth defect as was the general population of Western Australian infants born during the same period—a result unlikely to be due to chance. Furthermore, our estimate was a conservative one as it excluded all six infants with a patent ductus arteriosus on the basis that there was insufficient information to determine whether they met the criteria of the Western Australian Birth Defects Registry to be included as major defects (see footnotes to Table 1). The same was true for the 20 infants with angiomata and the 20 infants with congenital naevi. Three infants with ventricular septal defects were classified as having major defects, as such defects would always be defined as major according to the registry criteria unless they were notified as being “trivial.” Excluding these three infants reduced the odds ratio estimate to just less than two, but this had no impact on the substantive conclusions.
Potential for confounding and bias
Our reclassification of the Belgian data produced results for the two infant populations that were more appropriate for comparison than had been the case before. However, we recognise that the vigilance in terms of diagnosing birth defects would not necessarily have been the same in the two populations. The Belgian research group have closely examined 76% of their infants at 2 months of age, and it is therefore possible that they diagnosed defects which would not otherwise have come to medical attention in the first year of a child's life, if ever. However, all the defects defined as major according to the Western Australian criteria would, with the possible exception of the ventricular and atrial septal defects, have either been visibly obvious at birth—such as cleft palate—or so clinically serious—such as Fallot's tetralogy—that it is unlikely that they would not have been diagnosed in the general course of events. Most ventricular and atrial septal defects would also probably have been detected during the course of the routine child health surveillance that is in place in Western Australia. It seems unlikely, therefore, that differential vigilance in surveillance would account entirely for the excess of major defects seen in the infants born after intracytoplasmic sperm injection in Belgium.
There is also little reason to believe that the Western Australian population of infants has an unusually low prevalence of major birth defects compared with European populations; indeed, the prevalence of birth defects in Western Australia is generally higher than that reported from registers in Europe.21 Furthermore, 17% of the mothers who delivered in Western Australia in 1991–4 had been born in Europe, and control data collected for another study showed that in 1981–5 only 9% of infants did not have at least one parent, grandparent, or great grandparent born in Europe.20 22
Birth defects are known to occur more often in infants from multiple pregnancies, but this is largely confined to monozygous multiples.23 24 Excess numbers of defects seen in dizygous multiples tend to be positional in origin—such as talipes—and probably relate to the reduced space in the uterine cavity. Nearly half (48%) of the 420 infants born after intracytoplasmic sperm injection were multiple births, whereas only 3% of the Western Australian births were multiple.9 However, this is unlikely to account for the excess of major birth defects seen in the infants born after intracytoplasmic sperm injection as, although not stated, nearly all of the multiples are likely, by the nature of their conception, to have been dizygous or trizygous.
The overall mean maternal age for the infants born after intracytoplasmic sperm injection was 32.1 years, while it was 27.9 years for the Western Australian infants; a difference almost certainly due to the history of infertility of the couples treated by intracytoplasmic sperm injection.9 20 Chromosomal defects are known to increase in incidence with increasing maternal age, and gastroschisis increases with decreasing maternal age. Otherwise, there is little relation between maternal age and the risk of birth defects.
Conclusions
No single system for classifying birth defects can be regarded as the gold standard, although the British Paediatric Association's classification is widely used by many registries around the world. However, the issue here is not which is the best classification system but that, in comparing two sets of data, the data must be classified with the same system. The Belgian researchers chose to use their own classification system and then compared the results with data that had been classified with more inclusive systems. While the Belgian classification has some pragmatic value in that it classifies as minor those defects that do not affect function or require surgery, this could lead us to ignore the potential underlying aetiological aspects—namely, what are the findings telling us about the effects of intracytoplasmic sperm injection and the causes and effects of the types of male infertility treated with this procedure? Both of these are vital questions if we wish to advance our knowledge and understanding of the physiology of fertilisation and embryonic development, and the causes of male infertility.2
Our reanalysis of the Belgian data suggests that infants born after intracytoplasmic sperm injection may experience an excess of major birth defects. While these findings do not exclude the possibility of other teratogenic influences, parental infertility and the procedure of intracytoplasmic sperm injection are the two exposures that all the infants born after intracytoplasmic sperm injection in Belgium had in common. These findings clearly require confirmation in further data sets of all pregnancy outcomes, not just live births, and they presently raise more questions than they answer.
Our results also suggest that there may be an excess occurrence of major cardiovascular, gastrointestinal, and genitourinary defects generally—and cleft palate and diaphragmatic hernia specifically—in infants born after intracytoplasmic sperm injection. These particular findings must be interpreted with caution as they arose from multiple significance testing in the absence of a primary hypothesis and are mainly based on small numbers. However, we report them as they raise additional hypotheses and emphasise the need for continued surveillance of birth defects in the Belgian cohort and the instigation of systematic surveillance of birth defects in other cohorts of infants conceived by intracytoplasmic sperm injection, together with the collection of appropriate data for comparison.
Clearly, further research is required to elucidate these apparently discrepant results and their implications. Meanwhile, the results from this reanalysis have important implications for counselling couples, providing information, and obtaining informed consent before performing intracytoplasmic sperm injection. These findings are unlikely to deter many would-be recipients or practitioners, nor do we believe they necessarily should. They are, however, of value, as it is vital that couples embarking on this procedure have realistic expectations of the outcome.
Acknowledgments
We thank Aandra Ryan, who codes the birth defect data at the Western Australian Birth Defects Registry and who kindly recoded the Belgian data for us. We also thank all the contributors to the registry, as without their notifications this valuable resource would not exist.
Funding: Australian National Health and Medical Research Council (NHMRC) program grant number 963209.
Conflict of interest: None.