Incidence of cancer in children born after in-vitro fertilization

Fiona Bruinsma1,5, Alison Venn1, Paul Lancaster2, Andrew Speirs3 and David Healy4

1 Centre for the Study of Mothers' and Children's Health, La Trobe University, 251 Faraday Street, Carlton, 3053, Victoria, 2 Australian Institute of Health and Welfare National Perinatal Statistics Unit, University of New South Wales, Sydney, 2031, NSW, 3 Reproductive Biology Unit, Royal Women's Hospital, Carlton, 3053, Victoria and 4 Monash IVF & Department of Obstetrics and Gynaecology, Monash University, Monash Medical Centre, Clayton, 3168, Victoria, Australia


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Evaluation of the long-term health of children born using in-vitro fertilization (IVF) provides important information to clinicians and consumers. Until very recently, there have been no published data on the incidence of cancer in children conceived as a result of IVF, despite a number of case reports of neuroblastoma in children conceived using fertility drugs. This study used a record-linkage cohort design to investigate the incidence of cancer in children born after IVF. The study included all conceptions using assisted reproductive technologies between 1979 and 1995 at two clinics in Victoria, Australia that resulted in a live birth. Data on births were linked with a population-based cancer registry to determine the number of cases of cancer that occurred. The standardized incidence ratio (SIR) was calculated by comparing the observed number of cases to the expected number of cases. The final cohort included 5249 births. The median length of follow-up was 3 years, 9 months (range 0–15 years). In all, 4.33 cases of cancer were expected and six were observed, giving a SIR of 1.39 (95% CI 0.62–3.09). This study found that children conceived using IVF and related procedures did not have a significantly increased incidence of cancer in comparison to the general population.

Key words: cancer incidence/children/IVF/long-term follow-up


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Utilization of assisted reproductive technologies has rapidly increased over the past decade. During 1995, 1.1% of all births in Australia were the result of assisted reproductive technologies (Hurst et al., 1997Go). The introduction of a new procedure brings with it a responsibility to evaluate and monitor both the short and longer term outcomes of the intervention (Royal Commission on New Reproductive Technologies, 1993Go; Buitendijk, 1999Go; Doyle, 1999Go). Short-term outcomes of assisted reproductive technology, for example the number of live births and neonatal deaths, the rate of multiple birth, preterm deliveries and babies born at low birth weight, have been well documented. The health of children beyond the neonatal period has been less well evaluated. Many of the phases of assisted reproductive technology treatment, for instance, the use of fertility drugs to stimulate multiple folliculogenesis, the process of oocyte retrieval and preparation of spermatozoa, the growth of embryos in culture medium, and the freezing and thawing of embryos, have the potential to have a harmful effect on the developing embryo (Buitendijk, 1999Go).

Case-control studies have reported a significantly increased risk of neuroblastoma with prenatal exposure to fertility drugs (Kramer et al., 1987Go; Michalek et al., 1996Go). There have also been case series reports of embryonal tumours in children born to women who conceived as a result of fertility treatment (Melamed et al., 1982Go; White et al., 1990Go; Toren et al., 1995Go). Neuroblastomas are tumours of the sympathetic nervous system that are derived from the embryonic neural crest. They are the most common type of cancer in children <1 year of age in Victoria, Australia (Giles et al., 1993Go).

While case series reports are useful for suggesting possible adverse effects of treatment and generating hypotheses for further investigation, studies that measure associations between exposure and disease (such as case-control and cohort studies) are needed to substantiate the reports. Because case-control studies are often prone to recall bias, a well-designed cohort study is generally believed to provide a higher level of evidence of an association than that obtained from a case-control study (Greenhalgh, 1997Go).

Until very recently there were no population-based data on the incidence of cancer in children conceived as a result of in-vitro fertilization (IVF). The two published studies, conducted in the UK and Sweden, of children born after IVF did not find a significantly increased incidence of cancer (Doyle et al., 1998Go; Bergh et al., 1999Go). The present study investigated the incidence of cancer in 5249 children born as a result of assisted reproductive technologies in Australia.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The study included all conceptions using assisted reproductive technologies between 1979 and the end of 1995 at either the Royal Women's Hospital, Melbourne IVF or Monash IVF, Victoria, Australia that resulted in a live birth. It included conceptions by IVF, gamete intra-Fallopian transfer (GIFT) and intracytoplasmic sperm injection (ICSI) but excluded conceptions by artificial insemination or ovulation induction without IVF or GIFT. Pregnancies that ended in a stillbirth were excluded, as were those where the parents resided overseas or interstate.

Data were collected from clinics and the National Perinatal Statistics Unit (NPSU), Australian Institute of Health and Welfare to ensure that we had the most complete coverage of births possible. Data collected from the clinics were obtained from computerized records (where available) or from medical records. The contribution of clinics to the register of births held by the NPSU is not mandatory; however, accreditation from the national accrediting body requires participation in the registry (Lancaster et al., 1995Go).

Data items collected on each birth included parents' names, mother's date of birth, address, baby's date of birth and sex, type of infertility, drugs used for ovarian stimulation in the conception cycle, drugs used in the luteal phase, obstetric complications and pregnancy outcome.

Ascertainment of cases of invasive cancer was by record-linkage with the population-based Victorian Cancer Registry (VCR). Cancer notification to the VCR became a statutory requirement of all hospitals and pathology laboratories in 1981. Ascertainment of childhood cancer cases diagnosed before this time is thought to be practically complete because the major treatment centre for the State had been part of a voluntary notification scheme since 1959. Data held by the cancer registry include full name, sex, date of birth, date of diagnosis, histological type of tumour, date of last known contact and vital status (Giles et al., 1993Go). Follow-up for each child was from the date of birth until (i) diagnosis of cancer, or (ii) the date of death, or (iii) the end of 1996 (if the child was <15 years old at that time) or, (iv) the date of the child's 15th birthday.

The expected number of cases was calculated by applying the Victorian age-specific population-based cancer incidence from the years 1982–1995 to the person-years follow-up in each age group of children. The standardized incidence ratio (SIR) was determined by calculating the ratio of the observed number of cancer cases to the expected number.

Prior to commencement of the study, calculations indicated that there was 80% power ({alpha} = 0.05) to detect a SIR of >6 for neuroblastoma and a SIR of 2.8 for all childhood cancers combined. The expected numbers of cancers were derived from the estimated age-specific person-years and age-specific population cancer rates.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Characteristics of the cohort
The data set included 5579 live births, of which 330 were subsequently excluded because the parents resided overseas or interstate, giving a final cohort of 5249 births. The 5249 births were from 4357 pregnancies. The annual number of births increased substantially over the study period. Figure 1Go illustrates the number of conceptions resulting in a live birth for each year of the study. Over 50% of births were conceived in the last 4 years of the study.



View larger version (13K):
[in this window]
[in a new window]
 
Figure 1. Number of conceptions resulting in live births for each year of study (1979–1995).

 
Multiple pregnancy occurred in 19% of pregnancies (Table IGo). Over two-thirds of pregnancies resulted from the transfer of fresh embryos (IVF) or from the transfer of oocytes and spermatozoa for GIFT (Table IGo). The use of frozen embryos has increased during the 1990s: 60.6% of pregnancies conceived during 1993–1995 resulted from the transfer of frozen embryos.


View this table:
[in this window]
[in a new window]
 
Table I. Pregnancy characteristics
 
Information on the drugs used to stimulate multiple folliculogenesis was available from both the clinic and NPSU. If a discrepancy occurred, records were checked and the appropriate correction made. If frozen embryos were transferred, or a woman was the recipient of donated oocytes, the stimulation details refer to the drugs that were administered at the time of transfer. Details are not given on drugs used at the time of oocyte collection. Gonadotrophin-releasing hormone agonists (GnRHa) combined with human menopausal gonadotrophin (HMG) or follicle stimulating hormone (FSH) was the most commonly used stimulation regime (Table IIGo). The use of this regime increased rapidly in the late 1980s and continued into the 1990s. In the early to mid-1980s, clomiphene and HMG or FSH was used more frequently. The number of natural cycles reflects the increasing use of frozen embryo transfers during the 1990s.


View this table:
[in this window]
[in a new window]
 
Table II. Drug usage
 
Human chorionic gonadotrophin (HCG) was the most commonly used preparation for luteal support, followed by progesterone pessaries and then oestrogen and progesterone (Table IIGo). A very small number of women received HCG and progesterone (19) or other unspecified luteal support (three) during their pregnancy. Prior to 1990 the majority of pregnant women did not receive luteal support, by 1993–1995 this had fallen to 29% of pregnancies.

Approximately 13% of singleton births and 60% of multiple births occurred prior to 37 weeks gestation. Only 15 deliveries occurred after 43 weeks. Overall, 10.4% of singleton births and 57.1% of multiple births were of low birthweight (<2500 g). A total of 12.9% of multiple births were very low birthweight (<1500 g) (Table IIIGo).


View this table:
[in this window]
[in a new window]
 
Table III. Infant characteristics
 
The median age at the end of follow-up was 3 years, 9 months (range 0–15 years). 87.4% of the cohort had at least 1 year of follow-up. Of those who had <1 year of follow-up, 8% (54/660) died in the first year of life and one had a diagnosis of cancer. The total number of person-years contributed was 23854.6.

Cancer cases
The expected number of cases of childhood cancer was calculated as 4.33, compared with an observed of six. This gave a SIR for all childhood cancers of 1.39 (95% CI 0.62–3.09). There were two cases of acute lymphoid leukaemia (ICD-9 = 204), one case of unspecified leukaemia (ICD-9 = 208), one malignant neoplasm of the brain (ICD-9 = 191), one malignant neoplasm of the soft tissue (ICD-9 = 171) and one malignant neoplasm of the salivary gland (ICD-9 = 142). When coded according to morphology, the soft tissue neoplasm was classified as a ganglioneuroblastoma (ICD-0 9490). Age at diagnosis ranged from 8 months to 6 years, 9 months (Table IVGo). One child died 13 months after diagnosis. The remaining five children were alive at the end of follow-up. The malignant neoplasm of the salivary gland was one of only two cases reported to the Victorian Cancer Registry between 1982 and 1995.


View this table:
[in this window]
[in a new window]
 
Table IV. Characteristics of cancer cases
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
This study was initiated to investigate findings of case-control studies (Kramer et al., 1987Go; Michalek et al., 1996Go) and case reports (White et al., 1990Go; Toren et al., 1995Go) that suggested that children who had been conceived using fertility drugs may be at an increased risk of neuroblastoma. While case-reports can be useful for highlighting potential problems and generating hypotheses, it is important that the possible associations are substantiated with large, population-based studies. This study found that children conceived using assisted reproductive technologies did not have a significantly increased incidence of cancer in comparison to the general population. One case of ganglioneuroblastoma was observed in this study population. Our findings are consistent with those reported in the UK and Sweden (Doyle et al., 1998Go; Bergh et al., 1999Go). The study conducted in the UK of 2507 children expected 3.5 cases and observed two cases (Doyle et al., 1998Go). Although the recently published Swedish study included a larger number of children (5856) than our study, a smaller number of cases (3.6) was expected (Bergh et al., 1999Go). This difference was primarily due to a difference in the distribution of year of birth and the associated person-years follow-up. When the findings of the three studies were combined, 11.4 cases of childhood cancer were expected compared with 12 observed, giving a SIR of 1.05 (95% CI 0.6–1.8).

A limitation of our study was that the median length of follow-up was relatively short (3 years, 9 months). However, the study was initiated to address concerns about the incidence of neuroblastoma in children born after IVF. Neuroblastoma most commonly occurs in the first year of life. Secondly, the calculation of the expected number of cases is based on the number of person-years each child contributes and takes into account differential lengths of follow-up. Our population-based study contains the largest number of person-years follow-up to date and contributes to knowledge about the health of children born after IVF. Further large studies will be needed to detect a small increase in the overall incidence of cancer or an increase in the incidence of individual types of cancer with statistical significance.


    Acknowledgments
 
We thank Professor Judith Lumley, Ms Lyndsey Watson and Dr Graham Giles for their assistance with the study. Special thanks to Monash IVF and the Royal Women's Hospital/Melbourne IVF, without whose co-operation this study would not have been possible.


    Notes
 
5 To whom correspondence should be addressed Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Bergh, T., Ericson, A., Hillensjö, T. et al. (1999) Deliveries and children born after in-vitro fertilization in Sweden 1982–95: a retrospective cohort study. Lancet, 354, 1579–1585.[ISI][Medline]

Buitendijk, S. (1999) Children after in vitro fertilisation. An overview of the literature. Int. J. Tech. Assist. Health Care, 15, 52–65.

Doyle, P. (1999) The UK Human Fertilisation and Embryology Authority. How it has contributed to the evaluation of assisted reproduction technology. Int. J. Tech. Assist Health Care, 15, 3–10.

Doyle, P., Bunch, K., Beral, V. et al. (1998) Cancer incidence in children conceived with assisted reproduction technology. Lancet, 352, 452–453.[ISI][Medline]

Giles, G., Farrugia, H., Thursfield, V. et al. (1993) Childhood cancer in Victoria, 1970 to 1989. Anti Cancer Council of Victoria, Melbourne.

Greenhalgh, T. (1997) How to read a paper. Getting your bearings (deciding what the paper is about). Br. Med. J., 315, 243–246.[Free Full Text]

Hurst, T., Shafir, E. and Lancaster, P. (1997) Assisted Conception, Australian and New Zealand 1996. AIHW National Perinatal Statistics Unit, Sydney.

Kramer, S., Ward, E., Meadows, A. et al. (1987) Medical and drug risk factors associated with neuroblastoma: a case-control study. J. Natl Cancer Inst., 78, 797–804.[ISI][Medline]

Lancaster, P., Shafir, E. and Huang, J. (1995) Assisted Conception, Australia and New Zealand, 1992 and 1993. AIHW National Perinatal Statistics Unit, Sydney.

Melamed, L., Bujanover, Y., Hammer, J. et al. (1982) Hepatoblastoma in an infant born to a mother after hormonal treatment for sterility. N. Engl. J. Med., 307, 820.[ISI][Medline]

Michalek, A., Buck, G., Nasca, P. et al. (1996) Gravid health status, medication use and risk of neuroblastoma. Am. J. Epidemiol., 143, 996–1001.[Abstract]

Royal Commission on New Reproductive Technologies (1993) Proceed with care: Final report of the Royal Commission on New Reproductive Technologies. Minister of Government Services, Canada.

Toren, A., Sharon, N., Mandel, M. et al. (1995) Two embryonal cancers after in vitro fertilisation. Cancer, 76, 2372–2374.[ISI][Medline]

White, L., Giri, N., Vowels, M. et al. (1990) Neuroectodermal tumour in children born after assisted conception. Lancet, 336, 1577.[ISI][Medline]

Submitted on August 19, 1999; accepted on November 12, 1999.