Pregnancy after cancer: Does timing of conception affect infant health?

Abstract

Background

The aim of this retrospective cohort study was to determine whether women who conceive soon after treatment have higher risks of adverse pregnancy outcomes.

Methods

Vital records data were linked to cancer registry diagnosis and treatment information in three U.S. states. The first pregnancy conceived after diagnosis between ages 20–45 years with any invasive cancer or ductal carcinoma in situ was eligible. Log-binomial models were used to compare risks in cancer survivors who conceived in each interval to the risks in matched comparison births to women without cancer.

Results

Women who conceived ≤1 year after starting chemotherapy for any cancer had higher risks of preterm birth than comparison women (RR for chemotherapy alone=1.9, 95% CI: 1.3, 2.7; RR for chemotherapy with radiation=2.4, 95% CI: 1.6, 3.6); women who conceived ≥1 year after starting chemotherapy without radiation or ≥2 years after chemotherapy with radiation did not. In analyses imputing treatment end date for breast cancer survivors, those who conceived ≥1 year after finishing chemotherapy with or without radiation had no higher risks than women without cancer. The risk of preterm birth in cervical cancer survivors largely persisted but was somewhat lower in pregnancies conceived after the first year (RR for pregnancies conceived ≤1 year after diagnosis=3.5, 95% CI: 2.2, 5.4; RR for pregnancies conceived >1 year=2.4, 95% CI: 1.6, 3.5).

Conclusions

In women who received chemotherapy, the higher risk of preterm birth was limited to those survivors with short intervals between treatment and conception.

Precis: Cancer survivors who conceived their pregnancies more than 1 year after starting chemotherapy without radiation and 2 years after starting chemotherapy with radiation had no higher risks of preterm birth than comparison women without cancer. The preterm birth risk was slightly higher in cervical cancer survivors who conceived within a year of diagnosis than for those who conceived >1 year after diagnosis.

Introduction

Women who want to have children after a cancer diagnosis face difficult decisions about pregnancy timing. Organizations including the American Cancer Society, American Society of Clinical Oncology, and National Comprehensive Cancer Network offer advice on how long women should wait after treatment before attempting to conceive, but caution that more evidence is needed.^1–3 Although conceiving after a cancer diagnosis does not appear to increase risk of cancer recurrence,^{4, 5} it is unknown whether short intervals between treatment and conception increase the risks of poor pregnancy outcomes.

Many organizations suggest that women postpone pregnancy for 6-12 months after finishing chemotherapy, so that they have time to recover and do not conceive with an oocyte that was maturing during treatment. Because chemotherapy kills rapidly-dividing cells, it might damage the oocytes being recruited for ovulation, resulting in higher risks of miscarriage and birth defects in pregnancies conceived soon after treatment. This advice is rooted in the hypothesis that oocytes are most vulnerable to damage by chemotherapeutic agents during the period of rapid development before ovulation, but has not been well-tested in human studies.^{2, 3} Other side effects of chemotherapy, including immunosuppression, might increase the risk of having an infant born preterm or small for gestational age. Some have observed an increased risk of preterm birth and/or growth restriction in infants born to cancer survivors,^{6, 7} but it is not clear whether these risks depend on the time since treatment.

Cervical cancer survivors are at high risk of early delivery.⁸ Although some studies have found higher risks of preterm birth and miscarriage in women with shorter intervals between cervical surgeries and conception,⁹ others have not.^{10, 11} Pregnancy timing after cancer may also be important to thyroid cancer survivors, who require lifelong thyroid hormone replacement. Because hypothyroidism increases the risk of adverse pregnancy outcomes including miscarriage and preterm birth,¹² the American Thyroid Association recommends that women postpone conception for 6-12 months after starting hormone replacement therapy.¹³

Our aim was to determine whether the risks of adverse pregnancy outcomes differ by time since diagnosis and treatment for different cancer types.

Methods

Study populations

Two different populations were used for this study. In order to assess whether pregnancy timing after cancer is associated with adverse pregnancy outcomes, we used diagnosis and treatment start dates from state cancer registries linked to birth data from vital records in three U.S. states, with a comparison group selected from birth records. To impute treatment end dates for breast cancer survivors, we used information from cancer survivors who participated in the Furthering Understanding of Cancer, Health, and Survivorship in Adult (FUCHSIA) Women’s Study.

Population for main analysis: Cancer registry data linked to vital records in three states

Cancer registry diagnosis and treatment data from Aug. 23, 1994-2012 in Georgia, Aug. 23, 1999-2013 in North Carolina, and Jan. 1, 2004-2013 in Tennessee were linked to vital records. Births to women ages 20-45 diagnosed with any reportable invasive cancer¹⁴ or ductal carcinoma in situ (DCIS) were eligible. We identified the first birth at greater than 20 weeks gestation that was conceived after a cancer diagnosis reported in vital records from Jan. 1, 1994-2012 in Georgia, Jan. 1, 2000-2013 in North Carolina, and May 20, 2004-2013 in Tennessee. Women diagnosed during pregnancy were excluded.

Births were eligible for the comparison group if they did not link to a cancer diagnosis in the same state as the birth. Within each state, a random sample of births to women without a record of cancer diagnosis were matched 25:1 to births to cancer survivors using four variables from vital records: mother’s exact age at delivery (single-year categories), race and ethnicity (7 categories: Hispanic ethnicity of any race, non-Hispanic white, African American, Asian, Pacific Islander, Native American, and multiracial any ethnicity), parity (0, 1, 2, and ≥3), and maternal education (college graduate yes or no).

We identified 4,922 eligible births to cancer survivors before excluding multiple births (n=162), deliveries of <20 or >44 weeks gestation (n=6), improbable combinations of gestational age and birth weight (n=6), births where records indicated the delivery was not the mother’s first after diagnosis (n=166), stillbirths (n=30), and records missing values for matching variables (n=349) and cancer treatment (n=237). Comparison births were also limited to live, singleton births between 20 and 44 weeks gestation to mothers aged 20-45 at delivery.

Study population for imputation of treatment end dates: FUCHSIA Women’s Study

For a subset of breast cancer survivors who participated in the FUCHSIA Women’s Study, treatment type, start date, and end date were abstracted from medical records (n=283). Participants in the study were diagnosed at ages 20-35 during 1990-2009 in metro Atlanta or 1999-2009 in the rest of Georgia. The study was limited to women who were ages 22-45 at recruitment and survived at least two years.

Cancer Treatment and Timing

Treatment type and start date were based on data corresponding to the first course of cancer-directed therapy as captured by the cancer registry. To calculate the date of pregnancy conception, the clinical estimate of gestational age was subtracted from the infant’s birth date.

Because the main treatments of interest for thyroid and cervical cancer were surgeries around the time of diagnosis, the exposure for these cancers was categorized time between diagnosis and conception. For women who received chemotherapy and/or radiation, the exposure for the main analysis was categorized time from treatment start date in the registry to conception. For women treated with both chemotherapy and radiation, treatment start was defined as the day that the patient initiated chemotherapy or radiation, whichever came first. For the subset of survivors who participated in the FUCHSIA Women’s Study, treatment type and start date from medical records were used if they differed from the registry, although concordance between medical records and registry data was high.

In a secondary analysis limited to breast cancer survivors, we used time since treatment completion. Because the date of treatment completion is not captured in cancer registries, we imputed this date for breast cancer patients by assigning each woman the median treatment duration length (105 days for either adjuvant or neoadjuvant chemotherapy without radiation, 188 days for adjuvant chemotherapy with radiation, and 258 days for neoadjuvant chemotherapy with radiation) using data abstracted from medical records in the FUCHSIA Women’s Study. This allowed an estimation of pregnancy risk by time since treatment completion in breast cancer survivors.

Outcomes

Outcomes from birth certificate data were preterm birth (<37 weeks gestation), low birthweight (<2,500g), low birthweight at term (<2,500g at ≥37 weeks gestation), small for gestational age (<10% of birthweight for gestational age and sex based on a national distribution¹⁵), and Cesarean section.

Statistical analyses

The risks of adverse pregnancy outcomes were estimated for each time period after cancer. Log-binomial models were used to estimate risk ratios comparing risk of adverse outcomes in pregnancies conceived during each time interval after cancer with the risk in matched comparison women without a history of cancer. Analyses were conducted using SAS 9.4 (SAS Institute, Inc., Cary, NC). The study was approved by the North Carolina State Center for Health Statistics and the North Carolina Central Cancer Registry, and institutional review boards at Emory University, the Tennessee Department of Health, and the Georgia Department of Public Health.

Results

Cancer survivors in the linked registry study were likely to be married (80%), have a 4-year college degree (44%) and be in their 30s at the time of the first birth after cancer diagnosis (61% between ages 30 and 39). Pregnancy timing after cancer was strongly associated with age at diagnosis. Among births to women 40 or older at diagnosis, 55% were conceived within a year, compared with 21% among women who were 20–24 at diagnosis (Table 1). Pregnancy timing also differed by cancer type. Cervical cancer patients were the most likely to conceive soon after diagnosis, with 32% of births in this study conceived within a year.

TABLE 1.

Characteristics of the first eligible live singleton birth to women ages 20-45 conceived after cancer diagnosis, by time between diagnosis and conception.

	ALL CANCERS		≤1 YEAR		>1-2 YEARS		>2-5 YEARS		>5 YEARS
CHARACTERISTICS	NO.	COLUMN PCT.1	NO.	ROW PCT.2	NO.	ROW PCT.	NO.	ROW PCT.	NO.	ROW PCT.
Cancer type
Breast	754	18	168	22	212	28	273	36	101	13
Cervical	131	3	42	32	33	25	41	31	15	11
Hodgkin lymphoma	293	7	55	19	67	23	114	39	57	19
Melanoma	981	23	282	29	252	26	321	33	126	13
Thyroid	970	23	263	27	244	25	352	36	111	11
Other	1,074	26	291	27	270	25	376	35	137	13
Age at diagnosis
20-24	910	22	192	21	188	21	332	36	198	22
25-29	1,412	34	337	24	359	25	511	36	205	15
30-34	1,283	31	336	26	365	28	457	36	125	10
35-39	532	13	200	38	146	27	167	31	19	4
40-45	66	2	36	55	20	30	10	15	0	-
Maternal age at birth
20-24	251	6	128	51	76	30	47	19	0	-
25-29	1,084	26	305	28	299	28	390	36	90	8
30-34	1,480	35	359	24	396	27	525	35	200	14
35-39	1,089	26	237	22	257	24	408	37	187	17
40-45	299	7	72	24	50	17	107	36	70	23
Maternal race and ethnicity
White, non-Hispanic	3,074	73	782	25	786	26	1101	36	405	13
African American, non-Hispanic	810	19	234	29	212	26	262	32	102	13
Other non-Hispanic	191	5	53	28	52	27	64	34	22	12
Hispanic, any race	128	3	32	25	28	22	50	39	18	14
Maternal education
Less than high school	259	6	89	34	68	26	75	29	27	10
High school or GED	801	19	207	26	239	30	270	34	85	11
Some college or associate degree	1,278	30	348	27	298	23	462	36	170	13
At least 4 years of college	1,865	44	457	25	473	25	670	36	265	14

	ALL CANCERS		<1 YEAR		1-2 YEARS		2-5 YEARS		>5 YEARS
CHARACTERISTICS	NO.	COLUMN PCT.	NO.	ROW PCT.	NO.	ROW PCT.	NO.	ROW PCT.	NO.	ROW PCT.
Mother married
Yes	3,380	80	860	25	878	26	1206	36	436	13
No	819	19	241	29	199	24	270	33	109	13
Missing	4	0	0	-	1	-	1	-	2	-

¹Percents across maternal characteristics

²Percents across years between diagnosis and conception

Among survivors of any cancer treated with chemotherapy but not radiation, the risks of preterm birth and low birthweight were highest in pregnancies conceived within a year of starting treatment (Table 2). The preterm birth risk in these pregnancies was twice as high as in comparison women, with a Risk Ratio (RR) of 1.9 (95% CI: 1.3, 2.7) for chemotherapy without radiation and 2.4 (95% CI: 1.6, 3.6) for chemotherapy with radiation. The risks in breast cancer patients who had chemotherapy with and without radiation mirrored the risks for all survivors, with the highest risks of preterm birth and low birthweight in pregnancies conceived within a year of starting treatment (Table 2). In contrast, survivors who conceived at least one year after starting chemotherapy without radiation and more than two years after chemotherapy with radiation did not have higher risks of having an infant born preterm, low birthweight, or small for gestational age (SGA) than women without a cancer history. This was true in both analyses that included all survivors and those limited to breast cancer survivors. The results did not change when we controlled for method of delivery (Cesarean section vs. vaginal birth).

TABLE 2.

Risk and risk ratios for pregnancy outcomes by time between treatment start and conception of the first live birth after cancer, compared with the risk in matched women without cancer.

		PRETERM BIRTH			LOW BIRTH WEIGHT			SMALL FOR GESTATIONAL AGE
	LIVE BIRTHS	NO.	RISK (PERCENT)	RISK RATIOS	NO.	RISK (PERCENT)	RISK RATIOS	NO.	RISK (PERCENT)	RISK RATIOS
All cancers, chemotherapy without radiation
≤1 year	121	26	21 (15, 29)	1.9 (1.3, 2.7)	22	18 (12, 26)	2.0 (1.4, 3.0)	15	12 (7, 20)	1.1 (0.7, 1.7)
>1-2 years	163	20	12 (8, 18)	1.1 (0.7, 1.7)	18	11 (7, 17)	1.4 (0.9, 2.2)	21	13 (8, 19)	1.1 (0.8, 1.7)
>2-5 years	179	17	10 (6, 15)	0.9 (0.6, 1.4)	17	10 (6, 15)	1.3 (0.8, 2.1)	21	12 (7, 17)	1.2 (0.8, 1.9)
>5 years	68	5	7 (2, 16)	0.6 (0.3, 1.4)	8	12 (5, 22)	1.4 (0.7, 2.7)	16	24 (14, 35)	2.0 (1.3, 3.1)
All cancers, chemotherapy and radiation
≤1 year	72	19	26 (17, 38)	2.4 (1.6, 3.6)	16	22 (13, 34)	2.7 (1.7, 4.2)	12	17 (9, 27)	1.5 (0.9, 2.5)
>1-2 years	95	14	15 (8, 23)	1.5 (0.9, 2.4)	9	9 (4, 17)	1.4 (0.7, 2.6)	10	11 (5, 19)	1.2 (0.6, 2.1)
>2-5 years	158	19	12 (7, 18)	1.2 (0.8, 1.8)	15	9 (5, 15)	1.4 (0.8, 2.3)	15	9 (5, 15)	1.0 (0.6, 1.6)
>5 years	76	5	7 (2, 15)	0.6 (0.2, 1.3)	5	7 (2, 15)	0.6 (0.3, 1.5)	13	17 (9, 27)	1.4 (0.9, 2.4)
Breast cancer, chemotherapy without radiation
≤1 year	37	11	30 (16, 47)	2.4 (1.4, 4.0)	10	27 (14, 44)	3.3 (1.9, 5.8)	7	19 (8, 35)	1.7 (0.9, 3.4)
>1-2 years	60	6	10 (4, 21)	1.0 (0.4, 2.0)	5	8 (3, 18)	1.3 (0.6, 3.1)	4	5 (2, 16)	0.8 (0.3, 1.8)
>2 years	77	7	9 (4, 18)	0.8 (0.4, 1.7)	9	12 (5, 21)	1.3 (0.7, 2.5)	14	18 (10, 29)	1.7 (1.0, 2.7)
Breast cancer, chemotherapy and radiation
≤1 year	36	11	31 (16, 48)	2.9 (1.7, 5.0)	10	28 (14, 45)	3.4 (1.9, 6.0)	8	22 (10, 39)	1.7 (0.9, 3.2)
>1-2 years	59	12	20 (11, 33)	2.1 (1.2, 3.6)	8	14 (6, 25)	1.9 (1.0, 3.7)	8	14 (6, 25)	1.2 (0.6, 2.4)
>2 years	113	12	11 (6, 18)	1.0 (0.6, 1.7)	11	10 (5, 17)	1.2 (0.7, 2.1)	15	13 (8, 21)	1.3 (0.8, 2.1)
Cervical cancer*
≤1 year	42	15	36 (22, 52)	3.5 (2.2, 5.4)	13	31 (18, 47)	3.5 (2.1, 5.7)	5	12 (4, 26)	1.2 (0.5, 2.7)
>1 year	89	22	25 (16, 35)	2.4 (1.6, 3.5)	15	17 (10, 26)	2.6 (1.6, 4.2)	5	6 (2, 13)	0.6 (0.3, 1.4)
Thyroid cancer*
≤1 year	263	21	8 (5, 12)	0.8 (0.6, 1.3)	21	8 (5, 12)	1.2 (0.8, 1.9)	25	10 (6, 14)	1.0 (0.7, 1.4)
>1-2 years	244	28	11 (8, 16)	1.3 (0.9, 1.8)	17	7 (4, 11)	1.1 (0.7, 1.7)	24	10 (6, 14)	1.1 (0.7, 1.5)
>2-5 years	352	37	11 (8, 14)	1.1 (0.8, 1.6)	14	4 (2, 7)	0.6 (0.4, 1.1)	19	5 (3, 8)	0.6 (0.4, 0.9)
>5 years	111	15	14 (8, 21)	1.4 (0.9, 2.3)	12	11 (6, 18)	1.5 (0.9, 2.6)	14	13 (7, 20)	1.2 (0.7, 2.0)

^*For cervical and thyroid cancer, the timing categories reflect time from diagnosis to conception, rather than treatment start to conception.

In analyses with imputed treatment end dates, infants born to breast cancer survivors who conceived within a year of completing chemotherapy with or without radiation had higher risks of preterm birth and low birthweight (preterm birth RR for chemotherapy without radiation=2.4, 95% CI: 1.5, 3.9; RR for chemotherapy with radiation=2.0, 95% CI: 1.2, 3.1) than comparison women (Appendix Table A1). Infants born to breast cancer survivors who conceived at least one year after the estimated treatment end date had no higher risk than comparison women without a history of cancer.

We observed a slightly higher RR for preterm birth in cervical cancer survivors who conceived within a year of diagnosis (RR=3.5, 95% CI: 2.2, 5.4) than in those who conceived after 1 year (RR=2.4, 95% CI: 1.6, 3.5), but because the confidence intervals for these RRs were wide and overlapping, this may be a chance finding. Cervical cancer patients similarly had higher risks of Cesarean section, which was more common in preterm deliveries, than matched comparison women. The RR for Cesarean section was 1.7 (95% CI: 1.3, 2.3) among cervical cancer survivors who conceived within one year and 1.4 (95% CI: 1.1, 1.7) in those who conceived after one year (Appendix Table 2).

Thyroid cancer survivors did not have higher risks of any adverse outcome, regardless of when they conceived.

The risk of having an infant born SGA was highest in women with the longest intervals between treatment start and conception (Table 2). The SGA risk was twice as high in births to chemotherapy patients who waited at least 5 years to conceive than those without cancer.

Low birthweight at term was a rare outcome in infants born to cancer survivors. Among term infants born to survivors of any cancer treated with chemotherapy, the risk of low birthweight was 5% (7/148; 95% CI: 2%, 10%), in infants conceived within a year of treatment initiation, 2% (12/525; 95% CI: 1%, 4%) in infants conceived in >1-5 years after treatment, and 6% (8/134; 95% CI: 3%, 11%) among infants conceived >5 years after treatment, compared with 3% among matched women without a history of cancer.

We excluded stillbirths from our primary analyses due to small numbers and concerns that missing data in fetal death records could cause us to underestimate stillbirths in cancer survivors due to missed links between fetal death records and cancer registry diagnosis data. However, we did not observe an increased risk of stillbirth in cancer survivors. Among eligible pregnancies after any cancer diagnosis that reached 20 weeks, 0.7% (30/4,582) ended in stillbirth, which was the same as the risk (also 0.7%) in all unmatched eligible pregnancies to women without a cancer diagnosis. Of the 9 cancer survivors treated with chemotherapy whose first pregnancy after treatment ended in stillbirth, 3 survivors conceived within a year of starting treatment.

Discussion

In this study, the elevated risks of preterm birth and low birthweight in infants born to cancer survivors were limited to pregnancies conceived soon after treatment. Infants born to women who conceived more than a year after starting chemotherapy without radiation and more than two years after chemotherapy and radiation did not have any higher risks for preterm delivery than women without a cancer history. In breast cancer survivors with imputed treatment end dates, the higher risks of adverse outcomes were only among women who conceived less than a year after finishing treatment. Breast cancer patients with at least a year after treatment before pregnancy did not have higher risks than matched women without cancer. We observed higher risks of infants born small for gestational age after chemotherapy, with or without radiation, in women who conceived more than 5 years after starting treatment. Thyroid cancer patients did not have higher risks for preterm birth or infants born low birthweight, or small for gestational age at any time after diagnosis.

Some have hypothesized that conceiving too soon after cervical cancer could increase the risk of preterm birth due to inflammation from incomplete wound healing. However it is also possible that the slightly higher risk we observed in women who conceived quickly is due to other underlying differences between the patients who conceive quickly and those who wait, such as type of cervical procedure or risk of recurrence.

The mechanism by which chemotherapy might cause a temporary increase in preterm birth might be through transient effects such as immunosuppression^{16, 17} associated with preterm birth.^{18, 19} Studies have found that immunosuppression in breast cancer patients persists months or years after chemotherapy, with CD4+ counts at half of pretreatment levels 12–14 months after treatment,^{20, 21} and weaker vaccine response in breast cancer survivors with a mean of 2.6 years since chemotherapy.²² Other possible mechanisms by which chemotherapy could cause adverse outcomes include chronic anemia, cardiovascular effects, physical stress, or insufficient weight gain in pregnancy. However, the mechanism by which long intervals between cancer treatment and delivery could result in growth restriction is unclear. It is possible that adverse pregnancy outcomes in women with long intervals between cancer and conception are not due to the wait time itself, but underlying differences such as poorer cancer prognosis or underlying reproductive conditions that cause both infertility and adverse pregnancy outcomes.

This study has limitations. If the oocytes that were maturing during treatment are most susceptible to damage from chemotherapeutic agents, women who conceive soon after treatment could have a higher risk of birth defects and miscarriage, which we could not assess. A second limitation is that while the effects of chemotherapy likely depend on regimen and dose, these are not available in cancer registry data. A third limitation is that cancer registries report only the first course of cancer treatment, which excludes treatment for relapse or treatment initiated after the first treatment failed. As a result, we underreport time to conception in women who conceived after a second course of treatment. To assess the extent of treatment misclassification, we compared registry data to medical records for the subset of breast cancer survivors who also participated in the FUCHSIA study. In this population, the sensitivity for treatment with chemotherapy at any time before conception was 91%, with 99% specificity; radiation before pregnancy had 82% sensitivity and perfect specificity. Only 1 out of 91 women for whom we had treatment start dates from both medical records and the registry was classified into the wrong category of time since treatment based on the registry date, indicating that the magnitude of misclassification is likely small.

Our study has important strengths, including its population-based cohort design and large sample size. This allowed us to match precisely on important potential confounders, including the mother’s exact age at delivery. Studies have shown that vital record accuracy is excellent for birthweight and our matching variables,^23–26 and generally good for clinical estimate of gestational age.

The best pregnancy timing after cancer is a complex and individual question that depends on factors beyond the scope of this study, including whether the woman needs long-term hormone treatment. Some clinicians advise women not to conceive within two years of diagnosis, when the risk of relapse is highest, to reduce the risk of needing more cancer treatment during pregnancy. Others may not have time to wait, because treatments like alkylating chemotherapy can accelerate ovarian aging.²⁷ Women diagnosed at older ages have to decide whether the risks of declining fertility with time outweigh the potential risks of a short interval between treatment and conception. In this population, survivors who postponed conception for 1 year after starting chemotherapy without radiation, 2 years after starting chemotherapy with radiation, and 1 year after cervical cancer diagnosis had the lowest risks of preterm birth. Although additional studies are needed to confirm our results, this evidence can help guide clinicians in counseling women diagnosed with cancer during their reproductive years.