Risk of ovarian cancer in women who give birth after assisted reproductive technology (ART)-a registry-based Nordic cohort study with follow-up from first pregnancy.

There is concern that assisted reproductive technology (ART) may increase ovarian cancer risk, but previous studies are inconclusive. We compared ovarian cancer risk for women who gave birth after ART vs natural conception.

Through linkage of nationwide registry data, we followed 3,303,880 initially nulliparous women in Denmark (1994–2014), Finland (1990 –2014), Norway (1984 –2015) and Sweden (1985 –2015) from first pregnancy ≥22 weeks to ovarian cancer, emigration, death or end of follow-up (2014/2015). We estimated hazard ratios (HRs), adjusting for age, parity, maternal birth year and country, and for body mass index and smoking in subsamples.

Mean age at first birth was 27.7 years. During a mean follow-up of 14.4 person-years, 2683 participants (0.08%) developed ovarian cancer; 135 after ART and 2548 after natural conception only (incidence rates 11.6 and 5.5 per 100,000 person- years, respectively). The risk was higher for women who ever gave birth after ART (HR 1.70, 95% con fidence interval 1.42 –2.03) compared to natural conception. Associations were stronger for conventional in vitro fertilisation than for intracytoplasmic sperm injection.

Among parous women, ART-conception was associated with a higher risk of ovarian cancer than natural conception. Further studies should decipher whether this is causal or confounded by infertility or other factors. British Journal of Cancer (2023) 128:825–832; https://doi.org/10.1038/s41416-022-02097-7

Ovarian cancer is the seventh most common cancer among women and has five-years survival rates below 45% [ 1]. The hormonal aetiology of ovarian cancer has led to concerns about whether assisted reproductive technology (ART) increases risk in women treated for infertility [ 2]. ART treatment involves several potentially carcinogenic exposures, such as supraphysiological levels of estradiol and exogenous gonadotropins, as well as multiple ovarian punctures [3]. Following the first reports from the early 1990s on the higher risk of ovarian cancer in women treated with fertility drugs [4, 5], results have been diverging [ 6–14], and a systematic review from 2019 concluded that the heterogeneity between studies was too pronounced to conduct a meta-analysis [15]. This heterogeneity may result from differences in comparison groups (e.g. untreated infertile women, the general population, and naturally conceiving women), treatment (e.g. different stimulation regimens, ovulation induction with or without ART), and duration of follow-up, but also from random error since the number of ART-exposed cases in most studies were very small. A recent systematic review and meta- analysis included data from nineteen studies, with a total of 370 ovarian cancers in ART-exposed women (range 1 –66 in individual studies), found no association with fertility treatment [ 16]. Furthermore, previous reports suggest that infertility itself is associated with a higher risk of ovarian cancer [ 17–19], and that risk may differ according to the cause of infertility [ 17], also in women who give birth after ART [ 12, 20]. Over the last decades, the use of ART has increased steadily [ 21]. Only in Europe, more than 1 million ART cycles, resulting in more than 215,000 children, are performed each year [ 21]. In the Nordic countries, the availability and use of ART treatment is among the highest in Europe, and 3 –6% of recent birth cohorts were conceived using ART [ 21]. Received: 5 April 2022 Revised: 21 November 2022 Accepted: 29 November 2022 Published online: 22 December 2022 1Department of Public Health and Nursing, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology, Trondheim, Norwa y. 2The Cancer Clinic, St Olavs Hospital, Trondheim University Hospital, Trondheim, Norway. 3The Fertility Clinic, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark. 4Department of Knowledge Brokers, THL Finnish Institute for Health and Welfare, Helsinki, Finland. 5Department of Molecular Medicine and Surgery, Karolinska Institute, Stockholm, Sweden. 6Academic Primary Health Care Centre, Region Stockholm, Sweden. 7Department of Obstetrics and Gynaecology, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Sahlgrenska University Hospital, Gothenburg, Sweden. 8Spiren Fertility Clinic, Trondheim, Norway. 9Centre for Fertility and Health, Norwegian Institute of Public Health, Oslo, Norway. 10Department of Epidemiology, Netherlands Cancer Institute, Amsterdam, The Netherlands. 11Department of Obstetrics and Gynecology, University of Helsinki and Helsinki University Hospital, Helsinki, Finland. ✉email: [email protected] www.nature.com/bjcBritish Journal of Cancer 1234567890();,: Considering the increasing use of ART, the poor prognosis of ovarian cancer, and the inconsistency of previous results, we conducted a population-based cohort study including all parous women in four Nordic countries, with detailed baseline and follow-up information from nationwide registries, to assess whether giving birth after ART is associated with a higher risk of ovarian cancer than giving birth after natural conception (NC).

Data sources and study factors The Committee of Nordic ART and Safety (CoNARTaS) study population includes all women who gave birth after ART and NC from Denmark (1994–2014), Finland (1990 –2014), Norway (1984 –2015) and Sweden (1985–2015). Data were obtained from each country ’s national ART registry or database, linked with data from the respective Medical Birth Registry (MBR), and pooled into a Nordic cohort, described in detail previously and shown in Supplementary Table 1 [ 22, 23]. Because the Medical Birth Registries were used to de fine the study population, women who never gave birth were not included. The national identity number assigned to all residents in each Nordic country enabled follow-up through linkage at an individual level to data from the National Cancer Registries, National Patient Registries, Cause of Death Registries and Population Registries. We considered deliveries without registration of ART conception to

from NC. ART was classi fied as fertilisation by either conventional in vitro fertilisation (IVF) or intracytoplasmic sperm injection (ICSI), with either fresh or frozen embryo transfer. Ovulation induction and intrauterine inseminations were not considered ART treatment. In contrast to other European countries, ICSI is mainly reserved for couples with a male component cause of infertility in the Nordic countries [ 22, 24]. Fertilisation with both IVF and ICSI in the same cycle was categorised as IVF since we assumed male factor to be less pronounced in these couples. Fertilisation with non-ejaculated sperm was categorised as ICSI. Details on speci fic ART treatments were not available from Finland. Causes of infertility were recorded in the ART registries in Norway and Denmark, whereas in Sweden, information on diagnoses associated with infertility was extracted from the national patient registry. Relevant International Classi fication of Diseases (ICD) codes were ICD-10, N46; N80; N97; E28.2, ICD-9, 256.4; 606; 617; 628, ICD-8, 606; 628. All participants who gave birth after ART and had no recorded cause of infertility were classified as having ‘unknown’ cause of infertility. No information on the cause of infertility was available from Finland. We categorised causes of infertility as (a) female factor only, including endometriosis, polycystic ovary syndrome or anovulation, tubal, cervical or uterine factors; (b) male factor only; (c) mixed female and male factors, if any of the female factors and male factor were recorded; (d) unexplained infertility if no cause of infertility was found after the medical assessment; and (e) ‘other or unknown factors’ if specific causes not included in (a) or (b) were recorded (e.g. same sex couples), if no results from medical assessment were reported or if no diagnostic work-up had been performed. The MBRs collect data on all deliveries that take place in the respective countries, both live births and stillbirths. For Denmark and Sweden, stillbirths <28 weeks were included from 2004 and 2008, respectively. The MBRs provided information on maternal age (years), year and month of delivery, parity, and gestational age at delivery (days). For subsamples, we also had data on smoking in pregnancy, harmonised across the countries as ‘any smoking during pregnancy ’ versus ‘no smoking during pregnancy ’, height and pre-pregnancy or first-trimester weight. For all participants, data on all cancer diagnoses were available from the establishment of the national cancer registries (1943 in Denmark, 1953 in Finland and Norway and 1958 in Sweden). We de fined ovarian cancer according to the ICD-10 codes used by the NORDCAN project (C56 (malignant neoplasm of ovary) and C57.0 –4 (malignant neoplasm of fallopian tube, broad ligament, round ligament, parametrium and uterine adnexa, unspeci fied)) [ 25], a collaboration between the Nordic cancer registries which allows comparison of cancer occurrence between the Nordic countries over time. Data on borderline ovarian tumours were not available in the data linkages. Information on the date and cause of death was obtained from the National Cause of Death Registries, whereas data on emigration were obtained from national Population Registries. Data on emigration were not available from Finland. Deaths from ovarian cancer without a preceding registration of ovarian cancer in the cancer registries were also considered as events. Study population and follow-up All participants were followed from the conception of their first pregnancy with delivery in gestational week 22 or later. The date of conception was estimated by subtracting the gestational age at delivery from the date of birth. If gestational age was missing or registered as >340 days, we used the mean duration of pregnancy (282 days) [ 26]. We chose the conception date as the start of follow-up to allow the inclusion of cancers diagnosed during a pregnancy resulting in delivery since these could, in theory, be initiated or promoted by controlled ovarian stimulation in ART. We followed participants until the date of any first cancer diagnosis (except non-melanoma skin cancer), emigration, death, or December 31, 2014 (Finland and Denmark) or December 31, 2015 (Norway and Sweden), whichever occurred first. Cases of ovarian cancer diagnosed within two months (62 days) after the woman ’s first cancer diagnosis, were not censored at the date of first cancer diagnosis, but included as ovarian cancer cases, as the disease was likely present at the date of diagnosis of first cancer. All women who had their first delivery during the study period were eligible ( n = 3,319,187). We excluded women with a cancer diagnosis before the conception of their first child ( n = 15,479). These women more often had at least one ART-conceived pregnancy than those included in the study (6.7% vs 3.1%). Further, we excluded women with a period of emigration before their first birth ( n = 314), and participants with an unreliable date of birth or death ( n = 14), resulting in a final study population of 3,303,380 women (Fig. 1). Of these, 2,193,684 had information on BMI and 2,166,347 had information on both BMI and smoking. Parts of the study population were included in previous national studies from Denmark [ 20], Norway [ 13], and Sweden [ 11]. Statistical analysis We used Cox proportional hazards models to estimate hazard ratios (HRs) with confidence intervals (CIs). We used attained age as the time scale and treated ART-conception as a time-dependent exposure, considering women as exposed from their first ART conception resulting in delivery. Time-dependent exposure was used to ensure correctly allocated person- time and avoid ‘immortal time bias ’ [27]. Unsuccessful ART cycles and pregnancies resulting in miscarriages were included neither in the ART group nor in the NC group. We adjusted for parity (time-dependent), age at first birth, mother ’s year of birth in 5-year birth cohorts, and country. Missing values were handled through complete case analysis. We also adjusted for BMI and smoking in a subsample of women with available information, starting follow-up at the first recorded data. For BMI, the most recently available registration was used (time-dependent), whereas, for smoking, we expected the status to vary more between pregnancies, and included only pregnancies with non-missing smoking status. Due to a large proportion of missing data on these variables (32%), statistical power was not suf ficient to adjust for them in subgroup analyses. In a sensitivity analysis, we repeated the models above after excluding the first year of follow-up to examine if the associations were in fluenced by cancers diagnosed shortly after treatment. We used several complementary approaches to assess whether a higher level of exposure was associated with higher risk. First, we estimated risk according to combinations of conception methods for women with two or more deliveries, comparing women with two ART-conceived pregnancies, as well as both ART and natural conception, to women with two naturally conceived pregnancies. In these analyses, women were followed from their second pregnancy. Second, we assessed risk according to the number of pregnancies for women with natural or ART conceptions separately, starting follow-up at the first conception and censoring whenever a pregnancy with the other method was conceived. Third, we assessed the risk associated with pregnancies by fresh embryo transfer only, censoring when a pregnancy with frozen embryo transfer was conceived, to ensure that each additional pregnancy represented a new ovarian stimulation. We estimated cumulative incidence using a competing risk approach, treating other cancers, death and emigration as competing outcomes. In these analyses, we selected a subsample using a matched cohort design due to limited computational capacity. For each ART-conceiving mother, we selected five naturally conceiving mothers with the same year of birth, age at first birth, country of residence and parity. The proportional hazards assumption was tested with Schoenfeld residuals and by inspection of log-log plots, and there were no clear violations. All analyses were performed in Stata, version 15. M.S. Sandvei et al. 826 British Journal of Cancer (2023) 128:825 – 832

The cohort included 3,303,380 women who were nulliparous at the start of follow-up and had 6,605,024 deliveries during follow- up. Of these, 119,437 women (3.6%) gave birth after ART at least once during the study period, and 3,183,943 women gave birth after natural conception only (Table 1). The number of women who gave birth after ART increased throughout the study period. ART-mothers were followed for a mean of 9.7 years (standard deviation, SD, 6.8), while naturally conceiving mothers were followed for a mean of 14.5 years (SD 8.6). The mean age at first birth was 32.4 (SD 4.8) years in ART-mothers and 27.5 (SD 4.9) years in naturally conceiving mothers. The mean age at diagnosis of ovarian cancer was 43.0 years (SD 6.8) for ART-mothers, and 42.3 years (SD 9.0) for naturally conceiving mothers. The mean age at the end of the follow-up was 41.4 (SD 9.0) years. During >47 million person-years of follow-up, 2683 participants (0.08%) were diagnosed with ovarian cancer. Of these, 135 were ART-mothers, giving an incidence rate of 11.6/100,000 person- years (95% CI 9.8 –13.7), whereas 2548 had naturally conceived children only (incidence rate 5.5/100,000 person-years, 95% CI 5.3–5.7). Throughout follow-up, unadjusted hazard rates according to time since first birth were higher for ART-conceiving compared to naturally conceiving mothers (Fig. 2), and proportionality testing indicated no clear variation in association during follow-up.

also remained similar after excluding the first year of follow-up (92 cases excluded, Supplementary Table 2). Consistent with these observations, the cumulative risk of ovarian cancer was higher in ART-conceiving mothers throughout the entire follow-up period (Fig. 3). Overall, the risk of ovarian cancer was higher for ART- conceiving than for naturally conceiving mothers (HR 1.70 95% CI 1.42 –2.03) after adjustment for attained age, mothers 5-year birth cohort, age at first birth, parity and country (Table 2). The estimates were somewhat higher in sub-populations with information about BMI and smok ing, but the estimates did not change substantially when adjusting for these factors (HR 1.95, 95% CI 1.49 –2.55). Country-specific analyses were largely comparable with the pooled results (Table 3). Women with two or more deliveries were at higher risk of ovarian cancer if at least one of their first two deliveries were after ART, as compared to women whose two first deliveries were after NC (Table 3). The risk was somewhat higher for women whose two first deliveries were after ART than in women with one delivery after NC and one after ART. In analyses of the fertilisation method, IVF was associated with a higher risk of ovarian cancer (HR 2.05, 95% CI 1.60 –2.64) compared to naturally conceiving mothers, whereas the association with ICSI was weaker and less clear (HR 1.44, 95% CI 0.93 –2.22). When restricting ART- conception to fresh transfers, age-adjusted estimates were higher (HR 2.80, 95% CI 2.26 –3.48), but fully adjusted estimates were similar to those from the main sample (HR 1.85, 95% CI 1.49 –2.31). Among women with deliveries only after natural conception, giving birth two or three times was associated with a lower risk of ovarian cancer as compared to giving birth once (Table 4), but additional pregnancies beyond three was not associated with a further decline in risk. For women with ART-conception in their first pregnancy, giving birth to at least one more child after ART, was not associated with a different risk of ovarian cancer compared to those with one delivery only (HR 1.26, 95% CI 0.75–2.12). Among women with only one conception, ART was associated with a higher risk of ovarian cancer as compared to NC conception (HR 1.33, 95% CI 1.04 –1.70) (Table 4). The analysis sample for causes of infertility comprised 98,750 women with deliveries after ART (Supplementary Table 3). Causes of infertility were characterised as female factors (endometriosis, polycystic ovary syndrome/anovulation, tubal, cervical or uterine factors) in 2 1,293 (16%), male factor alone in 15,840 (13%), mixed female and male factors in 12,831 (13%), unexplained infertility in 11,339 (12%) and other factors (including unknown) in 37,404 (38%). ART-conceiving mothers in all categories were at higher risk of ovarian cancer as compared to naturally conceiving mothers, but precision was low due to few cases in each category (Supplementary Table 3). Among women with the female causes of infertility, All women who gave birth in Denmark, Finland, Norway and Sweden during the study period n = 4,160,039 Women with first delivery during the study period n = 3,319,187 Main analysis sample, all countries n = 3,303,380 Excluded: Not primiparous n = 841,039 Excluded: Cancer diagnosis before conception (n = 15,479) Period of emigration before first birth (n = 314) Unreliable date of birth or death (n = 14) Sample with available data on BMI and smoking, all countries n = 2,166,347 Sample with availa ble data on ART treatment details and causes of infertility n = 96,562 Excluded: Finland n = 597,590 Fig. 1 Study population and analysis samples. M.S. Sandvei et al. 827 British Journal of Cancer (2023) 128:825 – 832 endometriosis was associated with a higher risk of ovarian cancer as compared to naturally conceiving mothers than were other female causes of infertility (Supplementary Table 3). The distribution of registered causes of infertility differed substan- tially between countries, especially due to a large proportion of unknown or “other cause of infertility ” in Sweden (Supplemen- tary Table 4). In addition, the distribution of causes of infertility in our study differed from those of two large, similar studies [12, 20], as shown in Supplementary Table 4.

In this large registry-based cohort study of all parous women in four Nordic countries across almost three decades, we found that delivery after ART was associated with 70% higher risk of ovarian cancer than delivery after natural conception. However, the absolute risk of ovarian cancer was small (0.08% during a mean follow-up of 14.4 years). Adjusting for BMI and smoking did not substantially change the results. The association was stronger for women treated with IVF compared with ICSI fertilisation, and somewhat stronger after two 0 10 20 30 40 50 Cases per 100,000 person-years 0 5 10 15 20 25 30 35 Years since first birth Natural conception 95% CI ART 95% CI Legend Fig. 2 Ever ART and risk of ovarian cancer. 0 0.1 0.2 0.3 0.4 0.5 Cumulative risk, percent (%) 0 5 10 15 20 25 30 Years since giving birth Natural conception 95% CI ART 95% CI Legend Fig. 3 ART and cumulative risk of ovarian cancer. Table 1. Characteristics of study cohort of 3,303,380 parous women from the Nordic countries Denmark, Finland, Norway and Sweden 1984 –2015 according to mode of conception. Women registered with ART- conceptiona Women registered with natural conception only All participants Participants 119,437 3,183,943 3,303,380 Participants with ovarian cancer b 135 2548 2683 Follow-up, person years, mean (SD) 9.7 (6.8) 14.5 (8.6) 14.4 (8.6) Age at first birth, mean (SD) 32.4 (4.8) 27.5 (4.9) 27.7 (5.0) Age at first birth, categories, n (%) <25 6564 (6) 899,146 (28) 905,710 (27) 25–29 24,610 (21) 1,235,430 (39) 1,260,040 (38) 30–34 48,607 (41) 776,135 (24) 824,742 (25) 35–39 32,124 (27) 229,254 (7) 261,378 (8) ≥40 7532 (6) 43,978 (1) 51,510 (2) Parity at start of follow-up 0 0 0 Parity at end of follow-up, n (%) One 57,004 (48) 915,589 (29) 972,593 (29) Two 47,592 (40) 1,509,113 (47) 1,556,705 (47) Three 11,970 (10) 581,527 (18) 593,497 (18) Four or more 2871 (2) 177,714 (6) 180,585 (5) Country and study period, n (%) Denmark 1994 –2014 29,559 (25) 541,493 (17) 571,052 (17) Finland 1990 –2014 20,665 (17) 576,925 (18) 597,590 (18) Norway 1984 –2015 23,194 (19) 746,652 (23) 769,846 (23) Sweden 1985 –2015 46,019 (39) 1,318,873 (41) 1,364,892 (41) a18,793 of the ART-conceiving women also contributed person-time in the natural conception group, as they had one or more deliveries from natural conception before their first delivery after ART. b53 of cases were diagnosed with ovarian cancer within 62 days of their first cancer diagnosis: 4 ART-conceiving mothers and 49 naturally conceiving mothers. M.S. Sandvei et al. 828 British Journal of Cancer (2023) 128:825 – 832 ART-conceived pregnancies compared with both ART and natural conception in their first two pregnancies. Our results correspond reasonably well with results from a British cohort, the largest study to date on the association between ART and risk of invasive ovarian tumours (standardised incidence ratio (SIR) 1.40, 95% CI 1.24 –1.53) [ 12]. However, those

were not adjusted for age at first birth and parity but compared to standardised incidence rates (SIRs) in the general population. Also, similar estimates were reported in a meta- analysis [ 16] and a recent, large Dutch study [ 28], when comparing risk in ART-treated women as compared to the general population (relative risk (RR) 1.50, 95% CI 1.17 –1.92, and SIR 1.43, 95% CI 1.18 –1.71, respectively). However, in the Dutch study, parous women were not at higher risk of ovarian cancer as compared to women from the general population (SIR 1.10, 95% CI 0.83 –1.43), in contrast to our results. However, all the nine included studies in the meta-analysis were small, with a total of 76 exposed cases, whereas the Dutch study included 115 ovarian cancer cases exposed to ART, of which 54 were parous women. Furthermore, infertility itself seems to be associated with a higher risk of ovarian cancer, with associations of comparable magnitude to the associations for ART treatment from this study and other studies [ 17, 18, 29]. Interestingly, the meta-analysis and the Dutch study mentioned above found no clear associations between ART treatment and the risk of ovarian cancer when the comparison group was other infertile or subfertile women [ 28, 30]. Consistent with these observations, several studies indicate stronger associations between ART treatment and ovarian cancer among women treated for female factor infertility, especially endometriosis, than other causes of infertility [ 12, 20, 31]. Thus, it may be hypothesised that the higher risk of ovarian cancer after ART treatment might be due to underlying infertility rather than the ART procedures. Our finding of a weaker association in mothers who conceived after ICSI, an indicator of the male component cause of infertility in the Nordic countries [ 23, 24], compared to mothers who conceived after IVF, as well as the suggestive stronger association in mothers with endometriosis registered as the cause of infertility, compared to women with natural conception, is compatible with this hypothesis. However, we cannot rule out that mothers who conceived after ICSI may also have been exposed to fewer ART cycles, and women with endometriosis to more ART cycles, and precision was limited compared to the main analyses. A major strength of the CoNARTaS data is the large sample size, with the inclusion of all women who gave birth during the entire registration period of ART treatment in the Nordic countries. The linkage to national cancer and population registries offers essentially complete and long-term follow-up [ 32]. The linkage to the national birth registries allows for adjustment for age at first birth and parity, and for BMI and smoking in a subset of participants, in contrast to studies that compare incidence rates in treated patients with standardised incidence rates only [7, 8, 12, 33]. The prospective data collection and population- based design reduce the risk of non-differential misclassi fication and selection. A major limitation is that we had no information about women who did not give birth (regardless of the reason) and no information on ART cycles that did not result in delivery. Previous studies show a higher risk of ovarian cancer in women who remain nulliparous after ART, compared to untreated nulliparous women and the general population [ 12, 14, 28]. However, based on cumulative success rates from Denmark [ 34], we estimate that on a Nordic level, about 80,000 women would remain nulliparous after ART treatment and therefore not be included and that around 0.8% of the naturally conceiving mothers may have had unsuccessful ART treatment at some point (Supplementary Material). Similarly, although we had no data on reproductive tourism and most pregnancies conceived after ART abroad would be registered as NC, we expect that they comprise a very small proportion of our reference group. Despite the large cohort size, the number of ovarian cancer cases was still relatively small, because the proportion of women who gave birth after ART was small, and they are still relatively young. The median age at diagnosis in our study was only 42 years, compared to 64 years in the general population [ 35] (data from Denmark only). The follow-up to older age ranges will be possible in the future and is needed to assess if ART is associated with risk during typical onset ages. Differences in registration practice between the countries, combined with a high number of couples with unknown causes of infertility, limited our ability to directly take underlying infertility into account. The cause and severity of infertility may also in fluence the total number of ART cycles needed, the stimulation regimens and their responses, Table 2. Risk of ovarian cancer among 3,303,380 parous women from the Nordic countries Denmark, Finland, Norway and Sweden 1984 –2015. Analysis sample Mode of conception Cases/participants Age-adjusted HR (95% CI) a Full model HR (95% CI) b Full sampleb Natural conception 2546/3,202,736 1 (Ref.) 1 (Ref.) ART 135/119,390 1.75 (1.47 –2.08) 1.70 (1.42 –2.03)c BMId Natural conception 1050/2,116,692 1 (Ref.) 1 (Ref.) ART 63/85,559 1.97 (1.53 –2.54) 1.93 (1.48 –2.51)e Smoking and BMI f Natural conception 1012/2,090,238 1 (Ref.) 1 (Ref.) ART 61/84,196 1.97 (1.52 –2.56) 1.95 (1.49 –2.55)g ART assisted reproductive technology, BMI body mass index, CI confidence interval, HR hazard ratio. aAdjusted for attained age. bParticipants with information about age, parity and age at first birth. Eighteen thousand seven hundred ninety-three of the ART-conceiving women had one or more deliveries after natural conception before their first delivery after ART, and therefore, they also contributed person-time in the natural conception group up to their first ART conception resulting in delivery. cAdjusted for attained age, mothers 5-year birth cohort, age at first birth, parity and country. dParticipants with information about age, parity, age at first birth and body mass index. Followed from the conception of the first pregnancy where pre- pregnancy or first-trimester body mass index was recorded. eAdjusted for attained age, mothers 5-year birth cohort, age at first birth, parity, country and body mass index. fParticipants with information about age, parity, age at first birth, body mass index and smoking status (any smoking in pregnancy). Followed from the conception of the first pregnancy where pre-pregnant body mass index and smoking status was recorded. gAdjusted for attained age, mothers 5-year birth cohort, age at first birth, parity, country, body mass index and smoking status. M.S. Sandvei et al. 829 British Journal of Cancer (2023) 128:825 – 832 neither of which were available. In consequence, residual confounding from infertility cannot be excluded. We had no information on oral contraceptive use, which may confound our

because it is associated with a lower risk of ovarian cancer [36] and may differ according to fertility status. Also, we had no information about ethnicity or socio-economic status. However, ART treatment is highly subsidised in the public health care systems in the Nordic countries, and treatment decision is therefore primarily based on medical indications rather than the couple’s financial situation. Unfortunately, we had no information on borderline tumours or the different histotypes, for which risk factors may differ [19]. For example, endometriosis is more strongly associated and may share a genetic basis with the clear-cell, endometroid and high-grade serous histotypes [ 37]. However, given the relatively low total number of ovarian cancer cases, we would have had little power to assess associations with the different histological types, had data been available. We also had no information about familial ris k, for instance, BRCA1 and 2 mutations, which account for 10 –15% of all ovarian cancer [ 19]. To conclude, in this large registry-based cohort study of all parous women in four Nordic countries across almost three decades, we found that delivery after ART was associated with a higher risk of ovarian cancer than delivery after natural concep- tion. However, further studies should try to decipher whether this is due to the ART treatment, underlying infertility, confounding by other factors such as less use of oral contraceptive use, or a combination. Table 3. Risk of ovarian cancer among 3,303,380 parous women from the Nordic countries Denmark, Finland, Norway and Sweden 1984 –2015, according to country of residence and fertility history in their two first pregnancies. Casesa Participantsa,f Incidence rate b,c Age-adjusted HR (95% CI) a,d Full model HR (95% CI) e Mode of conception by country Denmark Natural conception 204 545,564 3.4 (3.0 –3.9) 1 (Ref.) 1 (Ref.) ART 24 29,537 8.9 (6.0 –13.3) 1.69 (1.10 –2.60) 1.69 (1.09 –2.63) Finland Natural conception 622 580,601 8.2 (7.5 –8.8) 1 (Ref.) 1 (Ref.) ART 35 20,638 16.2 (11.6 –22.5) 1.57 (1.11 –2.21) 1.62 (1.14 –2.29) Norway Natural conception 698 751,226 5.7 (5.3 –6.2) 1 (Ref.) 1 (Ref.) ART 23 23,194 10.0 (6.6 –15.0) 1.46 (0.96 –2.21) 1.40 (0.92 –2.14) Sweden Natural conception 1022 1,325,345 5.0 (4.7 –5.3) 1 (Ref.) 1 (Ref.) ART 53 46,019 11.9 (9.1 –15.5) 2.02 (1.53 –2.67) 2.12 (1.59 –2.83) Fertility history, first two pregnanciesg Two natural conceptions 1557 2,261,392 5.2 (5.0 –5.5) 1 (Ref.) 1 (Ref.) First natural, then ART conception 16 15,884 11.3 (7.0 –18.5) 1.94 (1.19 –3.18) 1.92 (1.17 –3.14) First ART, then natural conception 24 23,965 11.7 (7.8 –17.5) 2.06 (1.38 –3.09) 2.32 (1.54 –3.50) Two ART conceptions 22 19,521 14.4 (9.5 –21.9) 2.46 (1.62 –3.75) 2.66 (1.73 –4.07) ART fertilisation method c,h,i,j Natural conception 1924 2,621,425 5.0 (4.7 –5.2) 1 (Ref.) 1 (Ref.) IVF 67 58,741 14.3 (11.3 –18.2) 2.20 (1.72 –2.80) 2.05 (1.60 –2.64) ICSI 21 36,808 6.6 (4.3 –10.1) 1.34 (0.87 –2.06) 1.44 (0.93 –2.22) Fresh embryo transfer k 87 80,280 10.9 (8.8 –13.4) 2.80 (2.26 –3.48) 1.85 (1.49 –2.31) ART assisted reproductive technology, CI confidence interval, HR hazard ratio, ICSI intracytoplasmic sperm injection, IVF in vitro fertilisation, NC natural conception. aParticipants with information about age, parity and age at first birth. bPer 100,000 person-years. cData on emigration and ART method is not available from Finland. dAdjusted for attained age. eAdjusted for attained age, mother ’s 5-year birth cohort, age at first birth, parity and country. fNumbers do not add up (Participants NC + Participants ART ≠ Participants all) because some ART women are also counted as NC if they have at least one NC conception before their first ART conception. gFollow-up from the conception of the second pregnancy. hCompared to women with natural conceptions (not other ART-conceiving mothers). iWomen with children after first IVF and then ICSI contributes with person-time in the IVF group until conception with ICSI, whereas women with first ICSI and then IVF contributes with person-time in the ICSI group only. jConceptions with both IVF and ICSI were categorised as IVF , whereas conceptions using non-ejaculated sperm and ICSI were categorised as ICSI. kCensoring when a pregnancy with frozen embryo transfer was conceived. M.S. Sandvei et al. 830 British Journal of Cancer (2023) 128:825 – 832 DATA AVAILABILITY The data that support the findings of this study were used under license for the current study and are not publicly available. Restrictions apply to data availability, but data may be accessed through Statistics Denmark upon reasonable request to the authors and with permission from the relevant authorities, ethics committees and Statistics Denmark.

1. Webb PM, Jordan SJ. Epidemiology of epithelial ovarian cancer. Best Pract Res Clin Obstet Gynaecol. 2017;41:3 –14. 2. Twombly R. Too early to determine cancer risk from infertility treatments. J Natl Cancer Inst. 2012;104:501 –2. 3. Fathalla MF. Incessant ovulation-a factor in ovarian neoplasia? Lancet. 1971;2:163. 4. Rossing MA, Daling JR, Weiss NS, Moore DE, Self SG. Ovarian tumors in a cohort of infertile women. N Engl J Med. 1994;331:771 –6. 5. Whittemore AS, Harris R, Itnyre J. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 US case-control studies. II. Invasive epithelial ovarian cancers in white women. Collaborative Ovarian Cancer Group. Am J Epidemiol. 1992;136:1184–203. 6. Kessous R, Davidson E, Meirovitz M, Sergienko R, Sheiner E. The risk of female malignancies after fertility treatments: a cohort study with 25-year follow-up. J Cancer Res Clin Oncol. 2016;142:287 –93. 7. van Leeuwen FE, Klip H, Mooij TM, van de Swaluw AM, Lambalk CB, Kortman M, et al. Risk of borderline and invasive ovarian tumours after ovarian stimulation for in vitro fertilization in a large Dutch cohort. Hum Reprod. 2011;26:3456 –65. 8. Luke B, Brown MB, Spector LG, Missmer SA, Leach RE, Williams M, et al. Cancer in women after assisted reproductive technology. Fertil Steril. 2015;104:1218 –26. 9. Trabert B, Lamb EJ, Scoccia B, Moghissi KS, Westhoff CL, Niwa S, et al. Ovulation- inducing drugs and ovarian cancer risk: results from an extended follow-up of a large United States infertility cohort. Fertil Steril. 2013;100:1660 –6. 10. Brinton LA, Trabert B, Shalev V, Lunenfeld E, Sella T, Chodick G. In vitro fertili- zation and risk of breast and gynecologic cancers: a retrospective cohort study within the Israeli Maccabi Healthcare Services. Fertil Steril. 2013;99:1189 –96. 11. Kallen B, Finnstrom O, Lindam A, Nilsson E, Nygren KG, Olausson PO. Malig- nancies among women who gave birth after in vitro fertilization. Hum Reprod. 2011;26:253–8. 12. Williams CL, Jones ME, Swerdlow AJ, Botting BJ, Davies MC, Jacobs I, et al. Risks of ovarian, breast, and corpus uteri cancer in women treated with assisted repro- ductive technology in Great Britain, 1991 –2010: data linkage study including 2.2 million person years of observation. BMJ. 2018;362:k2644. 13. Reigstad MM, Larsen IK, Myklebust TA, Robsahm TE, Oldereid NB, Omland AK, et al. Cancer risk among parous women following assisted reproductive tech- nology. Hum Reprod. 2015;30:1952 –63. 14. Reigstad MM, Storeng R, Myklebust TA, Oldereid NB, Omland AK, Robsahm TE, et al. Cancer risk in women treated with fertility drugs according to parity status —a registry-based cohort study. Cancer Epidemiol Biomark Prev. 2017;26:953 –62. 15. Rizzuto I, Behrens RF, Smith LA. Risk of ovarian cancer in women treated with ovarian stimulating drugs for infertility. Cochrane database Syst Rev (Online). 2019;6:CD008215. 16. Barcroft JF, Galazis N, Jones BP, Getreu N, Bracewell-Milnes T, Grewal KJ, et al. Fertility treatment and cancers-the eternal conundrum: a systematic review and meta-analysis. Hum Reprod. 2021;36:1093 –107. 17. Lundberg FE, Iliadou AN, Rodriguez-Wallberg K, Gemzell-Danielsson K, Johansson ALV. The risk of breast and gynecological cancer in women with a diagnosis of infertility: a nationwide population-based study. Eur J Epidemiol. 2019;34:499–507. 18. Tworoger SS, Fair field KM, Colditz GA, Rosner BA, Hankinson SE. Association of oral contraceptive use, other contraceptive methods, and infertility with ovarian cancer risk. Am J Epidemiol. 2007;166:894 –901. 19. Reid BM, Permuth JB, Sellers TA. Epidemiology of ovarian cancer: a review. Cancer Biol Med. 2017;14:9 –32. 20. Vassard D, Schmidt L, Glazer CH, Lyng Forman J, Kamper-Jørgensen M, Pinborg A. Assisted reproductive technology treatment and risk of ovarian cancer-a nationwide population-based cohort study. Hum Reprod. 2019;34:2290 –6. 21. European Ivf Monitoring Consortium f.t.E.S.o.H.R., Embryology, Wyns C, De Geyter C, Calhaz-Jorge C, Kupka MS, et al. ART in Europe, 2018: results generated from European registries by ESHRE. Hum Reprod Open. 2022;2022:hoac022. 22. Henningsen AK, Romundstad LB, Gissler M, Nygren KG, Lidegaard O, Skjaerven R, et al. Infant and maternal health monitoring using a combined Nordic database on ART and safety. Acta Obstet Gynecol Scand. 2011;90:683 –91. 23. Opdahl S, Henningsen AA, Bergh C, Gissler M, Romundstad LB, Petzold M, et al. Data Resource Pro file: Committee of Nordic Assisted Reproductive Technology and Safety (CoNARTaS) cohort. Int J Epidemiol. 2020;49:365 –366f. 24. De Geyter C, Calhaz-Jorge C, Kupka MS, Wyns C, Mocanu E, Motrenko T, et al. ART in Europe, 2014: results generated from European registries by ESHRE: The Eur- opean IVF-monitoring Consortium (EIM) for the European Society of Human Reproduction and Embryology (ESHRE). Hum Reprod. 2018;33:1586 –601. 25. Engholm G, Ferlay J, Christensen N, Bray F, Gjerstorff ML, Klint A, et al. NORDCAN- a Nordic tool for cancer information, planning, quality control and research. Acta Oncol. 2010;49:725 –36. 26. Nguyen TH, Larsen T, Engholm G, Møller H. Evaluation of ultrasound-estimated date of delivery in 17,450 spontaneous singleton births: do we need to modify Naegele’s rule? Ultrasound Obstet Gynecol. 1999;14:23 –28. 27. Mansournia MA, Nazemipour M, Etminan M. Causal diagrams for immortal time bias. Int J Epidemiol. 2021;50:1405 –9. 28. Spaan M, van den Belt-Dusebout AW, Lambalk CB, van Boven HH, Schats R, Kortman M, et al. Long-term risk of ovarian cancer and borderline tumors after assisted reproductive technology. J Natl Cancer Inst. 2020;113:699 –709. 29. Murugappan G, Li S, Lathi RB, Baker VL, Eisenberg ML. Risk of cancer in infertile women: analysis of US claims data. Hum Reprod. 2019;34:894 –902. 30. Siristatidis C, Sergentanis TN, Kanavidis P, Trivella M, Sotiraki M, Mavromatis I, et al. Controlled ovarian hyperstimulation for IVF: impact on ovarian, endometrial and cervical cancer-a systematic review and meta-analysis. Hum Reprod Update. 2013;19:105–23. 31. Stewart LM, Holman CD, Aboagye-Sarfo P, Finn JC, Preen DB, Hart R. In vitro fertilization, endometriosis, nulliparity and ovarian cancer risk. Gynecol Oncol. 2013;128:260–4. 32. Pukkala E, Engholm G, Hojsgaard Schmidt LK, Storm H, Khan S, Lambe M, et al. Nordic Cancer Registries —an overview of their procedures and data compar- ability. Acta Oncol. 2018;57:440 –55. Table 4. Risk of ovarian cancer according to parity among 3,277,309 parous women with deliveries after either natural conception or ART from the Nordic countries Denmark, Finland, Norway and Sweden 1984 –2015. Natural conception ART Parity Cases/ participantsa Age-adjusted HR (95% CI) b Full model HR (95% CI) c Cases/ participants a Age-adjusted HR (95% CI) b Full model HR (95% CI) c 1st pregnancy 991/3,185,827 1 (Ref.) 1 (Ref.) 71/76,495 1.22 (0.95 –1.55) 1.33 (1.04 –1.70) 1st pregnancy 991/3,185,827 1 (Ref.) 1 (Ref.) 71/76,495 1 (Ref.) 1 (Ref.) 2nd pregnancy 1111/2,231,542 0.69 (0.63 –0.75) 0.64 (0.58 –0.70) 20/18,512 1.23 (0.74 –2.02) 1.26 (0.75 –2.12) 3rd pregnancy 343/737,296 0.56 (0.50 –0.64) 0.50 (0.43 –0.57) 4th pregnancy 80/171,401 0.64 (0.51 –0.80) 0.54 (0.43 –0.69) 5th and later pregnancies 21/74,646 0.57 (0.37 –0.87) 0.47 (0.30 –0.73) P for trend <0.001 <0.001 0.42 0.39 aParticipants with information about age, parity and age at first birth. Women with additional deliveries beyond the first, with the same conception method as in the first, will be included in more than one category/cell. bAdjusted for attained age. cAdjusted for attained age, mothers 5-year birth cohort, age at first birth, parity and country. M.S. Sandvei et al. 831 British Journal of Cancer (2023) 128:825 – 832 33. Venn A, Watson L, Bruinsma F, Giles G, Healy D. Risk of cancer after use of fertility drugs with in-vitro fertilisation. Lancet. 1999;354:1586 –90. 34. Malchau SS, Henningsen AA, Loft A, Rasmussen S, Forman J, Nyboe Andersen A, et al. The long-term prognosis for live birth in couples initiating fertility treat- ments. Hum Reprod. 2017;32:1439 –49. 35. Gottschau M, Mellemkjaer L, Hannibal CG, Kjaer SK. Ovarian and tubal cancer in Denmark: an update on incidence and survival. Acta Obstet Gynecol Scand. 2016;95:1181–9. 36. Havrilesky LJ, Moorman PG, Lowery WJ, Gierisch JM, Coeytaux RR, Urrutia RP, et al. Oral contraceptive pills as primary prevention for ovarian cancer: a sys- tematic review and meta-analysis. Obstet Gynecol. 2013;122:139 –47. 37. Mortlock S, Corona RI, Kho PF, Pharoah P, Seo JH, Freedman ML, et al. A multi- level investigation of the genetic relationship between endometriosis and ovarian cancer histotypes. Cell Rep Med. 2022;3:100542.

We thank the staff in Nordic fertility clinics and hospitals for taking time to complete the registry noti fications in their busy working day. The details and completeness of their work provide a solid foundation for our study. AUTHOR CONTRIBUTIONS AP, MG, CB, LBR, AT, UBW, AKH and SO conceived and designed the work that led to the submission and acquired the data. MSS and SO performed the statistical analyses. MSS drafted the manuscript. All authors played an important role in interpreting the

and revision of the manuscript, approved the final version and agree to be accountable for all aspects of the work. FUNDING The project was supported by a grant from the Norwegian Cancer Society [grant number 182356 –2016]. The establishment of the CoNARTaS cohort has additionally been supported by the Nordic Trial Alliance: a pilot project jointly funded by the Nordic Council of Ministers and NordForsk [grant number 71450], the Central Norway Regional Health Authorities [grant number 46045000], the Nordic Federation of Obstetrics and Gynaecology [grant numbers NF13041, NF15058, NF16026 and NF17043], the Interreg Öresund-Kattegat-Skagerrak European Regional Development Fund (ReproUnion project). COMPETING INTERESTS The authors declare no competing interests. ETHICS APPROVAL AND CONSENT TO PARTICIPATE This study was approved by the registry-keeping authorities in each country. Permission was granted from the regional ethics committees in Norway (REC North 2010/1909) and Sweden (Dnr 214-12, T422-12, T516-15, T233-16, T300-17, T1144-17, T121-18, T1071-18, T2019 02347). In Denmark and Finland, study-speci fic ethical approval is not required when using national registry data for research purposes. CONSENT FOR PUBLICATION Non applicable. ADDITIONAL INFORMATION Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41416-022-02097-7. Correspondence and requests for materials should be addressed to Marie Søfteland Sandvei. Reprints and permission information is available at http://www.nature.com/ reprints Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional af filiations. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. M.S. Sandvei et al. 832 British Journal of Cancer (2023) 128:825 – 832