Cancer risk in children, adolescents, and young adults conceived by ART in 1983-2011.

Since the introduction of ART in 1978, more than 10 million children have been conceived using ART and born over the intervening decades ( European Medical Journal, 2018 ; Faddy et al. , 2018 ; Spaan et al. , 2019 ). Currently, an estimated 1 million children are expected to be conceived using ART and born every year after successful IVF, ICSI, and/or frozen–thawed embryo transfer (FET) procedures ( European Medical Journal, 2018 ). Each phase of the ART procedure is substantially different from natural conception ( Buitendijk, 1999 ) and these processes occur in the same timeframe as epigenetic programming ( Iliadou et al. , 2011 ). It seems plausible that ART could influence early stages of embryogenesis and thereby permanently influence the development and health of individuals conceived through these techniques. With the increasing number of children conceived using ART, even subtle increased health risks become important from a public health perspective ( Roseboom, 2018 ). Previous studies have shown that children born after conception through ART have a higher risk of adverse perinatal outcomes such as pre-term birth ( Pinborg et al. , 2013 ; Berntsen et al. , 2019 ), a lower birthweight ( Berntsen et al. , 2019 ), and congenital malformations ( Davies et al. , 2012 ). In addition, there is evidence that ART may have long-term consequences for later health ( Hart and Norman, 2013a , b ). An increasing body of evidence suggests that ART may have consequences for cardiovascular and metabolic risk factors. In contrast, evidence regarding risk of cancer in children born after conception through ART is less consistent ( Gilboa et al. , 2019 ; Hargreave et al. , 2019 ; Spaan et al. , 2019 ; Spector et al. , 2019 ; Sargisian et al. , 2022 ; Weng et al. , 2022 ). Some large population-based studies have observed an increased overall cancer risk after ART ( Spector et al. , 2019 ; Weng et al. , 2022 ) while other studies did not ( Hargreave et al. , 2019 ; Spaan et al. , 2019 ; Sargisian et al. , 2022 ). In addition, increased risk estimates have been observed for children born after FET ( Hargreave et al. , 2019 ; Sargisian et al. , 2022 ) and non-significantly increased risks have been observed in children born after ICSI ( Hargreave et al. , 2019 ; Spector et al. , 2019 ). In our previous study, based on 21 246 children conceived after IVF, 3023 by ICSI, and 669 by FET, with a median follow-up of 21 years, slight increased risks of cancer, although not statistically significant, were found in children born after ICSI (hazard ratio (HR) = 1.52, 95% CI = 0.81–2.85) and FET (HR = 1.80, 95% CI = 0.65–4.95) ( Spaan et al. , 2019 ). After this publication, four other large cohorts have published their results. In a Danish cohort, including 19 448 children conceived by IVF, 13 417 by ICSI, and 3356 by FET, with a mean follow-up of 11 years, the risk of cancer was significantly increased in children born after FET compared to naturally conceived offspring from subfertile couples (HR = 2.43, 95% CI = 1.44–4.11). The risk of cancer in ICSI-conceived children was HR = 1.31, 95% CI = 0.90–1.92 ( Hargreave et al. , 2019 ). A Scandinavian cohort study, with data from Denmark, Finland, Norway, and Sweden, included 171 774 children conceived by ART (2.2%) and over 7.7 million children who were born after natural conception ( Sargisian et al. , 2022 ). After a mean follow-up of 10 years, overall cancer risk was not increased after ART (HR = 1.08, 95% CI = 0.96–1.21), but risk of cancer was significantly increased in children born after FET (HR = 1.65, 95% CI = 1.24–2.19). Cancer risk after ICSI was not reported ( Sargisian et al. , 2022 ). In a study from the USA, which included 275 686 ART children with median follow-up of 4.6 years, overall cancer risk was borderline significantly increased compared with the general population (HR = 1.17, 95% CI = 1.00–1.36) ( Spector et al. , 2019 ). Risk of cancer in children born after ICSI was not significantly increased (HR = 1.29, 95% CI = 0.96–1.74) compared to children born after IVF. In a population-based Taiwanese study including 47 152 ART children, the overall risk of cancer was increased (HR = 1.42; 95% CI = 1.04–1.95) after a median follow-up of 6 years when compared to children from subfertile parents not conceived by ART. No association between use of FET and cancer was found. No data were available to allow research into the risk of cancer following ICSI conception ( Weng et al. , 2022 ). As the results from previous studies investigating risk of cancer in ART children are inconsistent, the aim of the current study was to investigate the risk of cancer in an expanded cohort of children born after conception through ART in 1983–2011. The study included a much larger group of children born after ICSI and FET than in our previous publication, as well as a comparison group of children conceived without the use of ART from subfertile couples. This allowed for the investigation of cancer risk in children born after different types of ART (IVF, ICSI, and FET) with more statistical power.

In total, the cohort comprised 51 417 children born after the use of ART and 37 832 children conceived without the use of ART. After a median follow-up of 17.8 years (interquartile range (IQR) = 12.1–24.6), 358 cancers were observed; 157 in the ART group and 201 in children conceived without the use of ART. Follow-up was shorter in children conceived using ART (median 16.3 years) than in children conceived without the use of ART (median 20.3 years) ( Table 1 ). Age of mothers of children born after the use of ART was substantially older than those conceived without the use of ART. As expected, ART children had a shorter mean gestational age, lower mean birthweight, and were more often part of a multiple birth than children conceived without the use of ART. Characteristics of the OMEGA-offspring cohort * by conception method. a  Includes children conceived by fertility drugs with/without IUI or naturally conceived children from subfertile couples. b  Follow-up ended at date of any cancer diagnosis or date of completeness of cancer registry, whichever came first. c  The causes of subfertility were categorized according to the factor that was identified as the most important one within a couple. If several diagnoses had been registered, without mention of the main diagnosis, the following order was applied: male factor, tubal factor, hormonal factor, other factor or unexplained for main diagnosis. Male factors include azoospermia, oligozoospermia, asthenozoospermia, and/or teratozoospermia. Tubal factor was reported for all types of tubal pathology. Hormonal factors include ovulation disorders, PCOS, and premature menopause; unexplained and other factors such as endometriosis and cervical factors. d  Other factors include factors such as endometriosis, cervical factors, and hormonal factors (such as ovulation disorders, PCOS, and premature menopause). *  The OMEGA-offspring cohort consists of all live-born offspring from subfertile couples who were treated with ART between 1983 and 2011 in The Netherlands and all offspring of subfertile couples who were not treated with ART between 1980 and 2001. Compared to the incidence rates in the general Dutch population, overall cancer risk was not increased in the entire OMEGA-offspring cohort (SIR = 0.95, 95% CI = 0.86–1.06), the ART group (SIR = 0.96, 95% CI = 0.81–1.12), or children conceived without the use of ART (SIR = 0.95, 95% CI = 0.83–1.09) ( Table 2 ). Risks were also not increased in children conceived by ICSI and FET (SIR = 1.11, 95% CI = 0.80–1.51 and SIR = 1.22, 95% CI = 0.61–2.18, respectively) compared to the general population. Furthermore, the risk of cancer did not increase in any of the groups with an older attained age ( Table 3 ). Risk of cancer according to conception method compared with the Dutch general population. SIR: standardized incidence ratio. a  Includes naturally conceived children and children conceived by fertility drugs (with/without intrauterine insemination) from subfertile couples. b  Total cohort, includes children born after the use of ART and without the use of ART from subfertile couples (and includes naturally conceived children and children conceived by fertility drugs (with/without IUI)). c  Only first cancers were included in the analyses. Risk of cancer according to conception method and attained age compared with the Dutch general population. a SIR: standardized incidence ratio. a  Only first cancers were included in the analyses. b  Includes naturally conceived children and children conceived by fertility drugs (with/without IUI) from subfertile couples. Children born after the use of ART were at an increased risk of parotid gland cancer (SIR = 6.42, 95% CI = 1.32–18.77, based on ≤3 cases) compared to the general Dutch population. Risk of melanoma was non-significantly increased in children born after the use of ART (SIR = 1.66, 95% CI = 0.95–2.69, based on 16 cases), and not in children conceived without ART (SIR = 0.93, 95% CI = 0.58–1.41, based on 22 cases) ( Table 2 ). ICSI-conceived children were at a significantly increased risk of melanoma (SIR = 4.67, 95% CI = 1.27–11.95, based on four cases) compared to the general population. In comparison to children from subfertile couples conceived without the use of ART, the risk of cancer in children born after ART was not increased (age-adjusted HR = 1.03, 95% CI = 0.82–1.30). After adjustment for year of birth, the risk was also not increased (adjusted HR = 1.06, 95% CI = 0.84–1.33, Table 4 ). When stratifying ART into IVF and ICSI, IVF children were not at higher risk compared with children conceived without the use of ART (adjusted HR = 0.97, 95% CI = 0.76–1.23) but ICSI-conceived children were (adjusted HR = 1.58, 95% CI = 1.08–2.31) ( Table 4 ). Cancer risk in children born after FET was slightly but nonsignificantly increased, both when compared to children not conceived by ART (HR = 1.61, 95% CI = 0.86–3.03) and when compared to children born after fresh embryo transfers (HR = 1.56, 95% CI = 0.83–2.91). The HRs for cancer in children born after IVF–FET and in children born after ICSI–FET were 1.28 (95% CI = 0.47–3.30) and 1.50 (95% CI = 0.68–3.29), respectively, compared to children not conceived by ART ( Table 4 ). Analyses stratified according to attained age, sex, and multiple birth did not show different risks of cancer between ART children and those conceived without the use of ART ( Table 5 ). Risk of cancer according to specific conception methods; multivariable Cox regression analyses. FET: frozen–thawed embryo transfer; HR: hazard ratio. Each bold row represents a separate regression analysis. a  Additionally adjusted for birth year. Parental subfertility cause and maternal age did not confound the results and were therefore not included as confounders. b  Includes naturally conceived children and children conceived by fertility drugs (with/without IUI) from subfertile couples. Cancer risk for ART conceived children and adolescents versus children and adolescents not conceived by ART; multivariable Cox regression analyses. HR: hazard ratio. a  Includes naturally conceived children and children conceived by fertility drugs (with/without IUI) from subfertile couples. b  Not all numbers add up to 100%, because of missing values. c  Cox regression analyses: models with age (in years) as time scale and adjusted for birth year. Analyses include 89 249 persons, 51 417 ART, and 37 832 without ART. Parental subfertility cause and maternal age did not confound the results and were therefore not included as confounders. d   P value of interaction terms. There were no significantly increased site-specific cancer risks in children born after the use of ART when compared with children conceived without the use of ART ( Table 6 ). However, risk of melanoma was significantly increased in ICSI-conceived children compared with children conceived without ART (adjusted HR = 6.43, 95% CI = 1.59–25.94, based on 4 versus 22 cases). Additional adjustment for household income, as a surrogate for social economic status, did not alter the results. The risk of testicular carcinoma was not increased among ART boys (adjusted HR = 0.67, 95% CI = 0.30–1.50, based on 11 versus 20 cases). Risk of selected malignancies according to conception method; multivariable Cox regression analyses. HR: hazard ratio. a  Includes naturally conceived children and children conceived by fertility drugs (with/without IUI) from subfertile couples. b  Acute myeloid leukemia and lymphoblastic leukemia. c  Additionally adjusted for birth year. Parental subfertility cause and maternal age did not confound the results and were therefore not included as confounders. d  Hodgkin lymphoma and non-Hodgkin lymphoma. Sensitivity analyses excluding children born before 1989 (starting date NCR) yielded a HR for ART conception versus conception without ART of 0.95 (95% CI = 0.74–1.21). Excluding children born from mothers never treated with ART resulted in comparable risk estimates for ART children versus children conceived without the use of ART (adjusted HR = 1.02, 95% CI = 0.78–1.34). To assess the influence of non-independency owing to the inclusion of siblings in our study, we performed an analysis restricted to the first child with cancer in a sibship. There were only three sibships with two cancer cases each in the cohort. Exclusion of the second child with cancer in these sibships did not alter the overall cancer risk (HR = 1.05, 95% CI = 0.84–1.32). Risk of cancer was also assessed separately in the original and expanded cohort; these risk estimates are shown in Supplementary Table S3 .

The OMEGA-offspring cohort consists of all live-born offspring from subfertile couples who were treated with ART between 1983 and 2011 in The Netherlands and all offspring of subfertile couples who were not treated with ART between 1980 and 2001 ( Fig. 1 ). Couples were defined as subfertile if they were not able to conceive after 1 year or more of unprotected sex. The exposed group consisted of children conceived and born after ART in 1983–2011 while the comparison group included children conceived naturally (or after fertility drugs with/without IUI) born to subfertile women who did or did not receive ART treatment. Structure of the OMEGA-offspring cohort. The OMEGA-offspring cohort consists of all live-born offspring from subfertile couples who were treated with ART between 1983 and 2011 in The Netherlands and all offspring of subfertile couples who were not treated with ART between 1980 and 2001. The original OMEGA-offspring cohort study included all live-born children of women treated with ART in 1983–2000. To obtain a large enough comparison group of children not born after ART, we identified women who were diagnosed with fertility problems shortly before ART became a routine procedure for subfertile patients, i.e. all live-born children of subfertile women never treated with ART in 1980–2000 were included ( Fig. 1 ).The children of the women were identified through the Personal Records Database. In brief, in the Netherlands, the personal record of a woman also includes information about her children. Information on subfertility treatments (including ART) and patient characteristics were retrieved from the clinic’s (paper or electronic) records of the mothers from 12 Dutch IVF clinics and two regional fertility centers. Information on maternal characteristics and perinatal outcomes was available from the mothers’ questionnaires (62% response). A more detailed description of the cohort is given in our previous paper ( Spaan et al. , 2019 ) and in the Supplementary Data . In 2018, the cohort was expanded with children born in 2000–2011 to women treated with ART in 2000–2010, in order to study with more power the risk of cancer in children born after ART, and especially the risk after ICSI and FET ( Supplementary Table S1 ). In brief, all Dutch IVF clinics (n = 13, including the same 12 as in the original cohort plus one additional center) were requested to provide retrospective data regarding ART treatment cycles that led to a pregnancy between 1 January 2000 and 1 January 2011. Pregnancy was defined as the presence of hCG hormone tested in urine or blood samples following ART treatment. Information on patient characteristics and ART treatments was retrieved from the clinic’s electronic patient records system. Information on maternal characteristics and perinatal outcomes was extracted from the Dutch Perinatal Registry. To create one dataset, including IVF data and perinatal data, the two databases were combined by probabilistic linkage, as described in detail previously ( Pontesilli et al. , 2021 ). In order to identify all children of the women, including those not conceived by ART (as a comparison group), a linkage based on birthdate and postal code(s) of the mother was performed with the Personal Records Database ( Fig. 1 ). The overlap between children in the original cohort and the expanded cohort was excluded (n = 4441). The conception method of children from couples treated with or without ART between 1980 and 2001 (original cohort) was ascertained using information from the medical record and data regarding pregnancies (≥24 weeks) from the mother’s questionnaires, and was described in more detail previously ( Spaan et al. , 2019 ) and in the Supplementary Data . For children from couples treated with ART between 2000 and 2011 (expanded cohort), the conception method was available from the clinic’s electronic patient record of the mother. All children that resulted from a successful ART cycle were classified as born after the use of ART (and subdivided into IVF or ICSI and into fresh or FET). All children born to ART-treated couples between 2000 and 2011 that did not result from a successful ART cycle were classified as conceived without the use of ART, i.e. naturally conceived, by hormonal stimulation, or IUI. Children born before 1 May 2000 or after 31 December 2011, i.e. that could not be born as a result of the included ART cycles in the expanded cohort, were excluded because their conception method was unknown ( Supplementary Table S2 ). Additionally, adopted children were excluded ( Fig. 2 ). Identification of the OMEGA-offspring cohort. The OMEGA-offspring cohort consists of all live-born offspring from subfertile couples who were treated with ART between 1983 and 2011 in The Netherlands and all offspring of subfertile couples who were not treated with ART between 1980 and 2001. The following exclusions from the cohort were applied: a Children born before 1975; b Stillborns; c Children with incomplete birth dates; d Children with a cancer diagnosis or those who died before 1989; e Adopted children (i.e. not born in the Netherlands); f Children with an unknown conception method; g Children already identified in the OMEGA I–II cohort were excluded from the OMEGA-III cohort. The OMEGA-offspring cohort was linked with the Netherlands Cancer Registry (NCR), under strict privacy regulations. The NCR is a national population-based registry, with 96–98% completeness from 1989 ( Schouten et al. , 1993 ). For each cancer among OMEGA children until 1 November 2019, the NCR electronically provided information on date of diagnosis, topography, morphology, and stage (International Classification of Diseases for Oncology (ICD-O)). Deceased children were identified through linkage with the Central Bureau for Genealogy, which keeps electronic data about the vital status of all Dutch citizens from October 1994 onwards. Through the mother’s record in the Personal Records Database the date of birth of the mother and the child was used to create the following three variables: maternal age, child’s birth year, and multiple birth. Information about the sex of the child was not available through the mother’s record. Therefore, sex was based on the first name(s) of the child, using information about popularity of first names among males and females obtained from The Corpus of Given Names in The Netherlands (online database that includes information about given names in the Netherlands). Information on parental subfertility cause was obtained from the medical records. If missing, it was supplemented from the women’s questionnaire (only available for women in the original cohort). The participating IVF clinics and fertility clinics, the Institutional Review Board, legal counsel, and the disease registries gave permission for the performance of this study according to the General Data Protection Regulation. The dataset with the children’s names has been encrypted by a Trusted Third Party (ZorgTTP). Because of concerns about potential birth cohort effects on childhood cancer risk, children born before 1975 were excluded, providing a more equal age distribution between children born after the use of ART and those conceived without the use of ART, leaving 89 249 children in the analytical cohort ( Fig. 2 ). As the NCR did not fully cover the Netherlands before 1989, the observation time for each child started on 1 January 1989 or date of birth, whichever came last. Person-years of the observation were calculated until 1 November 2019, the date of first cancer, or the date of death, whichever came first. Children with a known cancer diagnosis or those who died before 1989 were excluded. Cancer incidence in the OMEGA-offspring cohort was compared with that in the Dutch general population by determining the standardized incidence ratio (SIR), defined as the ratio of the observed and expected number of cancers in the study population. Expected numbers were calculated by applying the person-year distribution in the cohort to sex-, age-, and calendar year-specific cancer incidence rates from NCR. Multivariable Cox regression models, with attained age on the X -axis, were used to directly compare cancer risk between children conceived using ART and those conceived and born to subfertile couples without using ART, while adjusting for confounding. Confounders were identified as factors that changed the risk estimate for the exposure of interest by ≥10%. Based on a priori knowledge about risk factors for childhood cancer and perinatal factors associated with ART, we tested the following variables for confounding: parental subfertility cause, maternal age, and child’s birth year. Effect modification of the association between ART and cancer risk was tested for parental cause of subfertility, sex, multiple birth, and different attained age groups. Missing data on sex was imputed with multivariate imputation by chained equations ( van Buuren et al. , 1999 ). The variables adjusted for in the analysis are provided in the footnotes to each table. Risk was assessed according to various ART aspects (IVF, ICSI, and FET (and IVF–FET and ICSI–FET)), follow-up period, and different cancer sites. Sensitivity analyses were performed to evaluate the influence of: inclusion of children born before 1989 (starting date NCR); inclusion of children from subfertile mothers never treated with ART in the comparison group (i.e. restricting the non-ART comparison group to children conceived without the use of ART from ART-treated women (sibling analysis)); non-independency caused by inclusion of siblings in the cohort; we excluded the second child with cancer in one sibship (i.e. only including the first child with cancer); and the expansion of the cohort (i.e. separate analysis for the original and expanded cohort). All tests were two-sided and a P -value below 0.05 was considered statistically significant. All statistical analyses were performed with Stata version 15.0 ( StataCorp, 2017 ).

This large-scale study with a median of 18 years of follow-up showed that overall cancer risk in children born after the use of ART is not increased, either when compared with the general population or when compared with children born to subfertile couples without the use of ART. Despite the reassuring results, an increased risk of cancer was observed in ICSI-conceived children, which was mainly driven by a higher risk of melanoma. However, as the number of melanoma cases was small, this finding may be due to chance and should be interpreted with the greatest caution. Furthermore, if this risk increase would be true, the absolute risk of developing melanoma after ART remains very low (<1%). Results from the current study are in line with those from the two Scandinavian cohort studies to the extent that no increased overall cancer risk was observed after ART ( Hargreave et al. , 2019 ; Sargisian et al. , 2022 ). However, our results are in contrast with two other large population-based cohorts that did report an increase of overall cancer risk after ART ( Spector et al. , 2019 ; Weng et al. , 2022 ). The increased risk we observed in ICSI-conceived offspring compared to children conceived without the use of ART is consistent with risk estimates reported in other cohort studies (HR = 1.29, 95% CI = 0.96–1.74 ( Spector et al. , 2019 ) and HR = 1.33, 95% CI = 0.94–1.89 ( Hargreave et al. , 2019 ), although none of these reached statistical significance ( Hargreave et al. , 2019 ; Spector et al. , 2019 ). Although our risk estimates for FET (HR = 1.61 and 1.56) are quite comparable with those observed in the Scandinavian studies (HR = 2.43 ( Hargreave et al. , 2019 ), HR = 1.65 ( Sargisian et al. , 2022 )), our results were not statistically significant. The varying outcomes across studies may be explained by differences in follow-up time (4.6 years ( Spector et al. , 2019 ), 6.0 years ( Weng et al. , 2022 ), 9.9 years ( Sargisian et al. , 2022 ), and 11.3 years ( Hargreave et al. , 2019 ) versus 17.8 years in our study), included treatment years, cohort size, comparison groups (general population ( Hargreave et al. , 2019 ; Sargisian et al. , 2022 ; Spector et al. , 2019 ; Weng et al. , 2022 ) versus offspring from subfertile women conceived without the use of ART in our study and two other cohort studies ( Hargreave et al. , 2019 ; Weng et al. , 2022 )), and the availability of confounding variables. As ever more children are born through ICSI and FET, long-term cancer risk should be investigated in cohorts comprising larger numbers of children born after these techniques. Compared with the general Dutch population, the risk of some cancer types was (non-)significantly increased among ART children and children conceived without the use of ART. Except for melanoma, the small number of cases did not enable us to further explore the association with ART within the cohort. As melanoma incidence is higher among people with a high income (possibly owing to more frequent exposure to sunlight) ( van der Aa et al. , 2011 ), we additionally adjusted for household income of the parents. The results did not change substantially, which might be explained by the fact that the children diagnosed with melanoma are relatively young and that risk factors, such as sun exposure, might not be very important yet. Therefore, early life exposures, including perinatal factors, might be of more importance. Information regarding perinatal exposures and risk of melanoma in offspring is scarce. Interestingly, in a cohort study among women who were exposed to diethylstilbestrol (DES) in utero , an increased risk of melanoma was found (risk of melanoma before age 40 years: SIR = 1.59; 95% CI = 1.08–2.26) ( Verloop et al. , 2010 ). Although the biological mechanism is unknown, it is possible that conception by ART may also induce melanoma in later life. Before definitive conclusions can be drawn, more research is warranted into perinatal factors, including ART, and risk of melanoma. The literature on the risk of melanoma after ART is scarce, possibly because most studies did not have sufficiently long follow-up to observe these tumours. Only one earlier study reported on the risk of skin cancer in young adults born to women evaluated for infertility (HR = 1.22, 95% CI = 0.94–1.60) ( Hargreave et al. , 2013 ). Two other studies reported the number of observed melanomas ( Kallen et al. , 2010 ) and skin cancers ( Wainstock et al. , 2017 ) but the numbers were too small for statistical analyses. To further investigate the risks of cancer types frequently occurring at young adult ages, such as melanoma, many large studies with longer follow-up periods are needed. This study has some limitations. Firstly, as cancer in children and young adults is rare, the number of cancers was rather small for subgroup analyses, despite the large size and long follow-up of the cohort. As a consequence, the observed (non-)significantly increased risks must be interpreted with caution. Although our study had long-term follow-up (median 18 years, IQR = 12.1–24.6), the risks of cancer in older adults conceived by ART remain unclear. Furthermore, analyses according to type of fertility drug used could not be performed owing to little variation in addition to a large proportion of missing data. Additionally, for 10 365 children, the conception method was unknown, mainly for children in the expanded cohort, therefore they were excluded from our main analysis. However, in sensitivity analyses including these children as either ART-conceived or conceived without the use of ART, the results were comparable, rendering bias caused by exclusion of these children unlikely. Furthermore, the cohort includes children born from 1975 onwards, while cancer registration was incomplete before 1989. Therefore, accrual of person-years and childhood malignancies started in 1989. In sensitivity analyses, exclusion of children born before 1989 did not affect the results in a meaningful way. Selection bias and misclassification bias are therefore highly unlikely. Residual confounding might be present owing to unmeasured differences (such as paternal age) between children born after the use of ART and children conceived without the use of ART. However, an analysis only including siblings from ART children in the comparison group showed comparable results, rendering such bias unlikely. Furthermore, the risk of cancer at young ages may run in families and some ART-conceived children may have had parents who became infertile after cancer treatment. Unfortunately, we had no complete data on the prevalence of cancer among parents, only among the mothers in our original cohort. Data from our original cohort showed that only 0.46% of the women had cancer prior to their subfertility treatment. This is in line with the prevalence of cancer in women from the general Dutch population, rendering bias owing to a higher proportion of parents with cancer less likely. Lastly, results are largely based on ART treatments performed in 1983–2010. Therefore, it is uncertain how the study results generalize to more contemporary ART treatments. The strengths of the current study include a long and complete follow-up, a comparison group of children from subfertile couples conceived without the use of ART, and detailed information on potential confounders. Selection bias is minimized because the Personal Records Database yielded complete information about all children from women included in the OMEGA study, and cancer incidence in the children was obtained through the national population-based NCR. In conclusion, after a median follow-up of 18 years, children born after the use of ART did not have an increased overall cancer risk. Our observation of an increased risk among ICSI-conceived children must be interpreted with caution owing to the small number of cases. Many large studies with prolonged follow-up are needed to investigate cancer risk in (young) adults conceived by different types of ART. In addition, international pooling of studies is recommended to provide sufficient power to study the risk of specific cancer sites after ART.

Click here for additional data file. Click here for additional data file.