OBJECTIVE: To estimate the incidence of pregnancy-associated cancer during a 30-year period in Denmark on the basis of both births and abortions with and without age adjustment and age standardization.
METHODS: Using data from the nationwide Danish registries, we investigated the occurrence of pregnancy-associated cancer (defined as cancer diagnosed during pregnancy and up to 1 year after pregnancy has ended) in pregnancies resulting in a birth (live birth or stillbirth) or an abortion. We included patients with primary cancer diagnosed at ages 15–44 years during 1977–2006 according to the Danish Cancer Registry. The patients were linked to the Medical Birth Registry, the Registry for Induced Abortions, and the National Patient Registry to obtain information on pregnancies.
RESULTS: We identified 2,426 patients with pregnancy-associated cancer. The three most frequent types were melanoma, cervical cancer, and breast cancer. We found an increase in the total numbers of all pregnancy-associated cancers from 572 cases during 1977–1986 to 1,052 cases during 1997–2006. The proportions of all pregnancy-associated cancers among all cancers increased from 5.4% to 8.3% during the same periods. The overall crude incidence rate of all pregnancy-associated cancer was 89.6 out of 100,000 pregnancies. The crude incidence increased over time, with an average annual percentage change of 2.9% (95% confidence interval [CI] 2.4–3.3). After age adjustment, the increase was still statistically significant (1.6%, 95% CI 1.1–2.1).
CONCLUSION: We found a statistically significant increase in pregnancy-associated cancer during the 30-year study period. The tendency to postpone childbirth only partly explains the increase in incidence rates.
LEVEL OF EVIDENCE: III
In Denmark, the incidence of pregnancy-associated cancer increased during 1977&#x2013;2006, and this cannot be explained only by an increase in maternal age.
Virus, Lifestyle, and Genes, Danish Cancer Society Research Center, and the Department of Obstetrics and Gynecology, the Juliane Marie Center, Rigshospitalet, Copenhagen, Denmark.
Corresponding author: Lene Mellemkjær, Virus, Lifestyle, and Genes, Danish Cancer Society Research Center, Strandboulevarden 49, DK-2100 Copenhagen Ø, Denmark; e-mail: email@example.com.
Supported by Rigshospitalets Forskningspulje.
The authors thank Thor Schütt Svane Nielsen and Peter Bo Aarslev for assistance with the statistical analyses, Kirsten Frederiksen for statistical advice, and Christian Dehlendorff for comments on the manuscript.
Financial Disclosure The authors did not report any potential conflicts of interest.
Pregnancy-associated cancer is usually defined as cancer diagnosed during pregnancy and up to 1 year after end of pregnancy1–3 (Fig. 1). However, the length of the postpartum period varies between studies, as do the types of pregnancies included. Most studies include only pregnancies resulting in a birth,2–4 whereas a single study specific to pregnancy-associated breast cancer also included pregnancies resulting in an abortion.5 Because the risk of cancer increases with age, the distribution of maternal age influences the incidence rates of pregnancy-associated cancer. Few studies have provided age-standardized incidence rates, so the differences seen in crude incidence rates between various populations may be attributable to disparities in age at pregnancy (Fig. 1). Information regarding incidence of pregnancy-associated cancer over time is sparse. A Norwegian study on cancer during pregnancy showed a crude annual increase of 2.5% during 1967–2002, but an age-adjusted result was not presented.4 However, a recent study performed in Australia reported that age-standardized incidence rates of pregnancy-associated cancer increased during 1994–2007.2
Mean age at first pregnancy among Danish women has increased from 24.1 years in 1973 to 28.6 years in 2003,6 which is likely to influence the incidence of pregnancy-associated cancer during this time period. Data from the national comprehensive health registries in Denmark provide us an opportunity to identify patients with cancer and to search for a pregnancy by use of the Danish unique personal identification numbers. Thus, the aim of the present study was to estimate the incidence of pregnancy-associated cancer during a 30-year period (1977–2006) in Denmark on the basis of both births and abortions with and without age adjustment and age standardization.
MATERIAL AND METHODS
We identified patients aged 15–44 years at first primary cancer diagnosis (except nonmelanoma skin cancer and choriocarcinoma) registered in the nationwide Danish Cancer Registry from 1977 to 2006. Patients with multiple cancer diagnoses registered on the same date were excluded (n=89). The Danish Cancer Registry was established in 1943; until end of 2003, the registry received notifications from general practitioners, specialists, and hospital departments. Since 2004, the Danish Cancer Registry has been based on recordings in other health registries.7 The Cancer Registry relies on multiple sources,7 and the cancer data have high completeness and validity.8
The identified patients were linked to the Medical Birth Registry (1976–2006) to obtain information regarding live births and stillbirths. Registrations of induced abortions were obtained from the Registry for Induced Abortions during 1976–1994; thereafter, they were obtained from the National Patient Registry.9 Spontaneous abortions and abortions attributable to abnormal pregnancies (eg, ectopic pregnancies, molar pregnancies, and missed abortions) also were obtained through the National Patient Registry during 1977–2006. All linkages were performed using the unique personal identification numbers as key identifiers to secure unequivocal linkage of information between registries. Abortions in the National Patient Registry were identified through specific International Classification of Diseases (ICD) codes (ICD, 8th Revision, codes 631, 632.39, 634.29, 634.60, 634.61, 634.69, 640-645; ICD, 10th Revision, codes O00–O06) or procedure codes (63420-63680, KLCH, BKHD4-5, BKHE0, BKHE8, BKKG, and BKXG). If there was less than 60 days between the conception dates of two abortions, we assumed that this was the same event and only counted the first pregnancy registered (5.9% of the abortions). If a patient was included in more than one registry with the same pregnancy, a registration in the Medical Birth Registry triumphed.
From these registries, we obtained information regarding gestational age and end date of pregnancy (date of birth or abortion). If the gestational age was missing (44.3%) in the Medical Birth Registry, we used a gestational age of 281 days, because this was the mean value reported in a previous registry-based Danish study of singleton births during 1998–2001.10 If gestational ages for abortions were unavailable, missing, or invalid, we calculated and applied appropriate means for gestational age based on valid gestational ages. Abortions from the Registry of Induced Abortions and the National Patient Register with gestational ages less than 14 days and more than 28 weeks (or 22 weeks after 1 April 200411) were considered invalid gestational ages. From valid gestational ages, means for gestational age were calculated separately for abortions from the Registry for Induced Abortions and the National Patient Register and were divided into a mean for induced abortions before 12 weeks of gestation and a mean for induced abortions after 12 weeks of gestation. From registrations from the National Patient Register, means were furthermore calculated separately for spontaneous abortions and abortions attributable to abnormal pregnancies. These mean values were used whenever the gestational age was unavailable, missing, or invalid (Registry of Induced Abortions: no abortions with unavailable or missing gestational ages and 40 abortions [0.3%] with invalid gestational ages; National Patient Register: 19,219 abortions with unavailable gestational ages during 1977–1994 and 607 abortions [12.7%] with missing gestational ages and 16 abortions [0.3%] with invalid gestational ages during 1995–2006).
Date of conception was calculated as end date of pregnancy (abortion or birth) minus gestational age and plus 14 days. The timeframe from conception to end of pregnancy was used to determine whether the patients were pregnant, postpregnancy, or neither at time of the cancer diagnosis. We defined pregnancy-associated cancer as cancer diagnosed within 30 days before the date of conception until 1 year after end date of pregnancy. We included 62 extra patients with pregnancy-associated cancers diagnosed during a 30-day period before conception because only registrations of month and year of cancer diagnoses were entered in the Danish Cancer Registry before 2004. The population was stratified into patients who had a cancer diagnosis during pregnancy or 0–12 months after pregnancy ended. Furthermore, the pregnancies were stratified by trimesters (first trimester: week 3 to 13 6/7 weeks; second trimester: week 14 to 26 6/7 weeks; third trimester: week 27 or later).12 The 62 cases before conception mentioned herein were included in the first trimester category, because they actually may have occurred during pregnancy because of the inaccuracy of the cancer diagnosis dates.
Numbers of induced abortions (n=569,051) and live births (n=1,858,619) among Danish patients aged 15–44 years during 1977–2006 were obtained from the Statistics Denmark13 and the National Board of Health in 5-year age groups for each calendar year.14 The total number of cancers (except nonmelanoma skin cancer and choriocarcinoma) among patients aged 15–44 years was obtained from the Danish Cancer Registry15 during 1977–2006.
Patients with pregnancy-associated cancer were stratified into 5-year age groups and calendar periods, and the three most frequent types of pregnancy-associated cancer were identified. The proportion of pregnancy-associated cancer among cancers was calculated based on induced abortions, spontaneous abortions, and abortions attributable to abnormal pregnancies, as well as live births and stillbirths. Incidence rates were calculated as number of pregnancy-associated cancers per 100,000 pregnancies or live births with or without the 1-year postpregnancy period. The 95% confidence intervals (CIs) were computed assuming that the number of pregnancy-associated cancers was log-linear Poisson-distributed. Then, the age groups 15–19 years and 20–24 years were combined because of a limited number of cases in the age group 15–19 years. Incidence rates were only calculated based on pregnancies resulting in induced abortions and live births because available numbers for the Danish population were restricted to these categories. To enable results from future studies to be compared with our results, we age-standardized the overall incidence rates using the Segi Standard World Population.16,17 To evaluate trends over time, we used Poisson regression models to calculate crude and age-adjusted incidence rates by 5-year calendar periods (15–24 years, 25–29 years, 30–34 years, 35–39 years, and 40–44 years, with the age group 30–34 years as reference). Furthermore, we estimated the average annual percentage change using Poisson regression models with and without adjustment for age. A negative average annual percentage change is equivalent to a decreasing trend, whereas a positive average annual percentage change is equivalent to an increasing trend. P<.05 was considered statistically significant. All analyses were performed using SAS 9.2. The study was approved by the Danish Data Protection Agency.
We identified a total of 2,426 patients with cancer diagnosed during pregnancy and up to 1 year after pregnancy ended. Mean age at diagnosis was 31.7 years (standard deviation, 5.3). Almost one-third of the patients had an abortion close in time to their cancer diagnosis, and among these abortions 67.9% were induced (Table 1). Of the 262 patients who had pregnancy-associated cancer during the first trimester, 113 (43.1%) had an induced abortion (data not shown). Pregnancy-associated cancer was most frequent in the age group 30–34 years at 33.3%, whereas in the age groups 25–29 years and 35–39 years the frequencies were 28.1% and 21.8%, respectively. The absolute number of pregnancy-associated cancer cases increased from 572 during 1977–1986 to 1,052 during 1997–2006 (Table 1).
We found an increase in proportions of all pregnancy-associated cancers out of all cancers over time from 5.4% during 1977–1986 to 8.3% during 1997–2006 (data not shown). Comparable increases over time were seen in all age groups from 25 to 44 years. During 1977–2006, the total proportion of all pregnancy-associated cancers based on all pregnancies was 6.8%, and it was 4.7% when only including live births (Table 2). In the age group 25–29 years, one of every five cancer cases was diagnosed in relation to a pregnancy.
Incidence rates were based on pregnancies resulting in an induced abortion and live birth. The crude overall incidence rate of pregnancy-associated cancer was 89.6 out of 100,000 pregnancies, whereas the corresponding age-standardized incidence rate was 107.1 out of 100,000 pregnancies (Table 3). We found similar incidence rates if we restricted our calculation to pregnancies resulting in a live birth.
Assessing the overall crude incidence rates of pregnancy-associated cancer by 5-year calendar periods, we found an increase over time, with a statistically significant average annual percentage change of 2.9% (95% CI 2.4–3.3) (Fig. 2A), which decreased to 1.6% (95% CI 1.1–2.1) after age adjustment. We stratified the incidence rate by age and found statistically significant average annual percentage changes for the age groups 15–24 years (3.2; 95% CI 1.6–4.9) and 25–29 years (3.0; 95% CI 2.1–4.0) and a borderline statistically significant average annual percentage change for age group 30–34 years (0.8; 95% CI 0.0–1.6) (Fig. 2B).
The three most frequent types of pregnancy-associated cancer were melanoma (n=507; 150 [30.0%] cases diagnosed during pregnancy), cervical cancer (n=493; 108 [21.9%] cases diagnosed during pregnancy), and breast cancer (n=489; 94 [19.2%] cases diagnosed during pregnancy) (Table 1). The proportion of pregnancy-associated cancers diagnosed in relation to an abortion was higher for breast cancer (41.7%) than for cervical cancer (33.7%) and for melanoma of skin (19.9%). During the first trimester, 12.5% of patients with melanoma diagnosed, 40.0% with cervical cancer diagnosed, and 81.0% with breast cancer diagnosed had an induced abortion (data not shown).
The proportion of pregnancy-associated melanomas out of all melanomas was 11.8% (Table 2). The corresponding proportions for cervical and breast cancer were 8.7% and 4.5%, respectively. These proportions were slightly smaller for all three site-specific pregnancy-associated cancers when only considering live births. The proportions were highest among women aged 25–29 years, with every fifth cervical cancer and breast cancer and every fourth melanoma diagnosed in relation to a pregnancy.
The crude overall incidence rate for breast cancer was 17.5 out of 100,000 pregnancies (induced abortions and live births), whereas the age-standardized incidence rate was 27.2 out of 100,000 pregnancies (Table 3). Similar crude overall incidence rates were seen for melanoma and cervical cancer. There were small differences between rates based on pregnancies and rates based on live births for all three sites. The crude incidence rates of melanoma increased from 1977 to 2006, supported by a statistically significant average annual percentage change of 5.3% (95% CI 4.1–6.4) (Fig. 3) and an age-adjusted average annual percentage change of 4.1% (95% CI 2.9–5.2) (data not known). The crude average annual percentage change for breast cancer was 3.1% (95% CI 2.1–4.0) (Fig. 3), whereas an average annual percentage change of 0.9% (95% CI −0.1, 2.0) was seen after age adjustment. The crude average annual percentage change for cervical cancer was 0.8% (95% CI −0.6 to 2.1) and was −0.5% (95% CI −1.7 to 0.7) after age adjustment. After excluding melanomas from the combined group of pregnancy-associated cancers, the crude average annual percentage change was 2.3% (95% CI 1.7–2.8) and the age-adjusted average annual percentage change was 0.9% (95% CI 0.4–1.5).
We found that every fifth cancer in the age group 25–29 years was diagnosed in relation to a pregnancy, and for melanomas this proportion was even higher. Absolute numbers of pregnancy-associated cancer, proportions, and crude incidence rates increased during the study period. An increase in incidence rates persisted, although it was less pronounced, after adjustment for age at cancer.
Cancer cases in the present study were obtained from the nationwide Danish Cancer Registry, which is considered to have a high degree of completeness.7,8 Most cancer diagnoses only had registrations of month and year; therefore, to capture some additional cases that were potentially diagnosed during pregnancy, we also accepted cancer diagnoses within 30 days before conception. This way, we added 62 cases (2.6%) to our population. A study performed in Australia also included instances when the cancer was diagnosed up to 1 month before pregnancy.5 We searched for pregnancy in various registries to make the ascertainment as complete as possible. The conception date was assessed as accurately as possible using the gestational age that was available for the majority of pregnancies (62.2%). Our calculation of incidence rates did not include person-years, and therefore it can be argued that the term proportion of pregnancies is more appropriate; however, we chose to use the term incidence because this was the term used in previous studies.1,2,4,18,19
By including abortions, we added approximately 30% more cases of pregnancy-associated cancers. Of all patients with pregnancy-associated cancer diagnosed during the first trimester, approximately 43% had an induced abortion; however, this percentage varied greatly for melanomas, cervical cancers, and breast cancers, with breast cancer being the highest (81%). In a study performed in Japan concerning prognosis after pregnancy-associated breast cancer, 93% of women with cancer diagnosed during the first trimester had an induced abortion.20 Although we have little information regarding the recommendations for abortions in pregnancy-associated cancer during our study period, our findings probably reflect that medical recommendations for pregnancies associated with cancer differ according to cancer type. Multidisciplinary evaluations regarding diagnostic strategy and treatment strategy as has been described for pregnancy-associated breast cancer21 possibly could be helpful in the effort to continue pregnancy. None of the previous studies except one specific to breast cancer5 has included abortions. By including abortions, ascertainment of cases becomes more complete. We reached a higher estimate of the absolute number and the proportions of pregnancy-associated cancer, whereas the incidence rates were quite similar with and without abortions.
The largest proportion of pregnancy-associated cancer out of all cancers was found among women aged 25–29 years, that is, every fifth cancer in this age group was diagnosed in relation to a pregnancy; for melanoma it was every fourth case. Similar results have been reported in a study performed in Sweden during 1960–1990, although this study only included live births.1 Our proportion of all breast cancers resembles the result of 6.25% reported in a study performed in Australia that also included all pregnancies in their calculations,5 whereas our proportion of all cervical cancers differs from the result of 0.5% during 1960–2004, which was reported by a different Swedish study.22 Thus, although pregnancy-associated cancer rarely occurs, among young female cancer patients a pregnancy often occurs close in time to the cancer diagnosis.
Our crude overall incidence rate of pregnancy-associated cancer is similar to rates found in studies performed in California3 and Norway4 (Fig. 1). A study performed in Sweden1 and one performed in Australia2 found lower and higher crude incidence rates, respectively, perhaps because their results were based on earlier and later calendar periods. Our average annual percentage increase remained statistically significant after age adjustment, meaning that the increase from 1977 to 2006 in Denmark was likely explained by factors other than an increase in maternal age. Lee et al2 recently reached the same conclusion based on data for pregnancy-associated cancer during the period from 1994 to 2007 in Australia. They suggested that improved diagnostic techniques and increased interaction with health care facilities over time might be responsible for the increase.
The dominating site-specific pregnancy-associated cancers were melanoma, cervical cancer, and breast cancer, as observed elsewhere.1,2,23 We found a crude overall incidence rate of melanoma that was higher than incidence rates noted in Sweden and in California1,3 but was lower than rates in Australia2 (Fig. 1). We report the highest crude overall incidence rate of cervical cancer to date,1–3,23 probably reflecting that Denmark has a high incidence of cervical cancer.24 The crude incidence rate of breast cancer in our study was quite similar to those reported in studies performed in Australia5 and Sweden.19
In the present study, incidence rates for pregnancy-associated melanomas clearly increased during the study period both before and after adjustment for age in parallel to the clear increase in the incidence of melanomas among all Danish women aged 15–44 years during 1978–2007.25 The general increase is likely to primarily reflect changes in sun exposure habits, including increased outdoor leisure activities, traveling, and tanning bed use. We also found an increase in the incidence rate of pregnancy-associated breast cancer, but the increase in maternal age explained most of that increase. A recent study from Canada only presented crude incidence rates of pregnancy-associated breast cancer, which increased negligibly from 1999 to 2008.18 A study performed in Sweden showed an increase in age-specific incidence rates between the periods 1963–1974 and 1975–1989, but not between the periods 1975–1989 and 1990–2002.19
Other factors beside the increase in maternal age6 and the general increase in the incidence of cancer during fertile ages26 may have contributed to the increase in pregnancy-associated cancer in Denmark. Antenatal and postnatal health care has improved over time, thus increasing the likelihood of detecting cancer in women. Use of infertility drugs also has increased over time, but no conclusion has been reached concerning the effect of infertility treatment on the risk of cancer.27 Biological phenomena influencing the incidence of pregnancy-associated cancer, such as pregnancy’s causing symptoms to be missed, pregnancy’s failing to be achieved because of subclinical cancer, and pregnancy’s having a suppressive effect on cancer, probably have not changed over time. However, these may explain why fewer cancers than expected have been observed during pregnancy, based on incidences occurring in the background population.1,23
Fortunately, pregnancies rarely are associated with cancer; however, young female cancer patients often have cancer diagnosed close in time to pregnancy, that is, every fifth cancer among women aged 25–29 years seems to be diagnosed in relation to pregnancy. We found a clear increase in the absolute numbers and in the incidence of pregnancy-associated cancer over the course of 30 years; thus, clinicians are more often faced with coinciding pregnancy and cancer. The tendency to postpone childbirth only partly explains the increase in incidence rates.
1. Lambe M, Ekbom A. Cancers coinciding with childbearing: delayed diagnosis during pregnancy? BMJ 1995;311:1607–8.
2. Lee Y, Roberts C, Dobbins T, Stavrou E, Black K, Morris J, et al.. Incidence and outcomes of pregnancy-associated cancer in Australia, 1994-2008: a population-based linkage study. BJOG 2012;119:1572–82.
3. Smith LH, Danielsen B, Allen ME, Cress R. Cancer associated with obstetric delivery: results of linkage with the California cancer registry. Am J Obstet Gynecol 2003;189:1128–35.
4. Stensheim H, Moller B, van DT, Fossa SD. Cause-specific survival for women diagnosed with cancer during pregnancy or lactation: a registry-based cohort study. J Clin Oncol 2009;27:45–51.
5. Ives AD, Saunders CM, Semmens JB. The Western Australian gestational breast cancer project: a population-based study of the incidence, management and outcomes. Breast 2005;14:276–82.
7. Gjerstorff ML. The Danish Cancer Registry. Scand J Public Health 2011;39:42–5.
8. Jensen AR, Overgaard J, Storm HH. Validity of breast cancer in the Danish Cancer Registry. A study based on clinical records from one county in Denmark. Eur J Cancer Prev 2002;11:359–64.
9. Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient Register. Scand J Public Health 2011;39:30–3.
10. Zhu JL, Hjollund NH, Olsen J. Shift work, duration of pregnancy, and birth weight: the National Birth Cohort in Denmark. Am J Obstet Gynecol 2004;191:285–91.
12. Chmitorz A, von KR, Rasmussen KM, Nehring I, Ensenauer R. Do trimester-specific cutoffs predict whether women ultimately stay within the Institute of Medicine/National Research Council guidelines for gestational weight gain? Findings of a retrospective cohort study. Am J Clin Nutr 2012;95:1432–7.
16. Doll R. Cancer incidence in five countries. International Union Against Cancer. Berlin (Germany): Springer-Verlag, 1966.
17. Segi M. Cancer mortality for selected site in 24 countries (1950-57). Sendai (Japan): Department of Public Health, Tohoku University of Medicine; 1960.
18. Abenhaim HA, Azoulay L, Holcroft CA, Bure LA, Assayag J, Benjamin A. Incidence, risk factors, and obstetrical outcomes of women with breast cancer in pregnancy. Breast J 2012;18:564–8.
19. Andersson TM, Johansson AL, Hsieh CC, Cnattingius S, Lambe M. Increasing incidence of pregnancy-associated breast cancer in Sweden. Obstet Gynecol 2009;114:568–72.
20. Ishida T, Yokoe T, Kasumi F, Sakamoto G, Makita M, Tominaga T, et al.. Clinicopathologic characteristics and prognosis of breast cancer patients associated with pregnancy and lactation: analysis of case-control study in Japan. Jpn J Cancer Res 1992;83:1143–9.
21. Amant F, Loibl S, Neven P, Van Calsteren K. Breast cancer in pregnancy. Lancet 2012;379:570–9.
22. Pettersson BF, Andersson S, Hellman K, Hellstrom AC. Invasive carcinoma of the uterine cervix associated with pregnancy: 90 years of experience. Cancer 2010;116:2343–9.
23. Haas JF. Pregnancy in association with a newly diagnosed cancer: a population-based epidemiologic assessment. Int J Cancer 1984;34:229–35.
24. Ferlay J, Shin HR, Bray J, Forman D, Mathers C, Parkin DM. GLOBOCAN 2008 cancer incidence and mortality worldwide. Available at: http://www-dep.iarc.fr
. Retrieved December 18, 2012.
25. Fuglede NB, Brinck-Claussen UØ, Deltour I, Boesen EH, Dalton SO, Johansen C. Incidence of cutaneous malignant melanoma in Denmark, 1978-2007. Br J Dermatol 2011;165:349–53.
26. Engholm G, Ferlay J, Christensen N, Johannesen TB, Klint Å, Køtlum JE, Milter MC, Ólafsdóttir E, Pukkala E, Storm HH. NORDCAN cancer incidence, mortality, prevalence and survival in the Nordic countries. Available at: http://www.ancr.nu
. Retrieved November 21, 2012.
© 2013 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
27. Brinton LA, Sahasrabuddhe VV, Scoccia B. Fertility drugs and the risk of breast and gynecologic cancers. Semin Reprod Med 2012;30:131–45.