Effects of hypertension and use of antihypertensive drugs in pregnancy on the risks of childhood cancers in Taiwan

Abstract

Background:

Childhood cancers are associated with high mortality and morbidity, and some maternal prescription drug use during pregnancy has been implicated in cancer risk. There are few studies on the effects of hypertension, preeclampsia, and the use of antihypertensives in pregnancy on children’s cancer risks.

Objective:

This population-based cohort study analyzed the relationship between hypertension, preeclampsia, and antihypertensives taken during pregnancy and the risks of childhood cancers in the offspring.

Methods:

Data on all children born in Taiwan between 2004 to 2015 (N=2,294,292) was obtained from the Maternal and Child Health Database. This registry was linked with the National Health Insurance Database and Cancer Registry to get the records of maternal use of diuretics or other antihypertensive in pregnancy and records of children with cancer diagnosed before 13 years. We used Cox proportional hazard modeling to estimate the influence of maternal health conditions and antihypertensive drug exposure on the risks of developing childhood cancers.

Results:

Offspring of mothers with hypertension (chronic or gestational) had a higher risk of acute lymphocytic lymphoma [hazard ratio (HR)= 1.87, 95% Confidence Interval (CI) 1.32 – 2.65] and non-Hodgkin’s lymphoma (HR= 1.96, 95% CI 1.34 – 2.86). We estimated only a weak increased cancer risk in children whose mothers used diuretics (HR=1.16, 95% CI 0.77 – 1.74) or antihypertensives other than diuretics (HR=1.15, 95% CI 0.86 – 1.54) before birth.

Conclusions:

In this cohort study, children whose mothers had chronic and/or gestational hypertension had an increased risk of developing childhood cancer.

Introduction

There is increasing evidence that the pathogenesis of various types of childhood cancers is related to exposures in-utero [1]. For some childhood cancer types, diagnoses peak at an early age, indicating that mutations in utero likely contribute to the development of childhood cancers [2, 3]. Childhood cancer risk factors include environmental exposures, maternal infections, and genetic and gestational factors [4–7]. In addition, the relation between diethylstilbestrol and adenocarcinoma of the vagina provides proof of principle that maternal pharmaceutical drug intake in pregnancy can cause cancer in offspring [8].

Hypertension may exist before pregnancy (chronic hypertension) or newly occur 20 weeks after conception as gestational hypertension. A study in the US estimated the prevalence of hypertension in women aged between 20 and 44 years as 7.7%, and 4.9% of all women in this age group are receiving antihypertensive pharmacotherapy [9]. Hypertension can also co-occur with edema and proteinuria with or without end-organ damage after 20 weeks of gestation as preeclampsia. Preeclampsia affects 2–8% of pregnancies globally, and the prevalence has increased in recent years [10]. The two most common categories of medications used by women of childbearing age with hypertension are diuretics (47.9%) and angiotensin-converting enzyme inhibitors (44.0%) [9]. The National Institute for Health and Care Excellence and the World Health Organization guidelines on the management of hypertension in pregnancy issued in 2010 recommended that neither group of medications should be used to treat gestational hypertension due to its ability to induce maternal hypovolemia and the potential fetal side effects of thrombocytopenia, hyponatremia, and hypokalemia [11, 12]. Diuretics taken during pregnancy have shown adverse effects on the fetus, including reduced plasma volume, decreased placental blood flow, intrauterine growth restriction, reduction of platelet formation, and preterm delivery [13]. Nevertheless, the potential benefits of diuretics may sometimes necessitate their use in pregnant women despite potential risks, e.g., it may not be possible to strictly follow the recommendation to avoid diuretics. Also, sometimes a patient may not be motivated to change a medication regimen they consider effective [14]. In addition, women already on antihypertensive medications may unintentionally continue these antihypertensive medications at the start of pregnancy before they know they are pregnant.

Studies have reported associations between preeclampsia in mothers and risk for childhood cancers, with hepatoblastoma most consistently associated [15–18]. Other studies have focused on hypertension, either chronic or gestational, although the number of studies of any specific cancer type has typically been sparse [19–25]. Increased risk of childhood cancer was reported with antihypertensive medication use [7, 19], particularly diuretics and neuroblastoma [18, 26–28]. Other cancer outcomes, including Wilms tumor, brain tumors, and acute lymphoblastic leukemia (ALL), were evaluated in fewer studies with less consistent results reported [6, 18, 19, 29–32]. Likely due to sample size limitations, some studies grouped all antihypertensives together, making it harder to determine which medications were associated with childhood cancer. In addition, due to the rarity of these diseases several of these studies were hampered by small sample sizes. It is unclear if any risk increases seen with hypertension may be related to the disease itself or to the antihypertensives used by mothers (confounding by indication), or whether, alternatively, maternal use of antihypertensives may lower offspring cancer risk due to controlling maternal disease [19].

Given the small number of studies and limited statistical power, we examined the influence of (i) preeclampsia and hypertension and (ii) the use of diuretics and other antihypertensives in pregnancy on childhood cancer risks in a large-scale nationwide prospective study.

Methods

As previously described, in this study, we linked multiple health registries from Taiwan [4] and accessed all the databases from the Health and Welfare Data Science Center, a part of Taiwan’s Ministry of Health and Welfare. We used the Taiwan Maternal and Child Health Data to obtain records of births between 2004 to 2015, and this dataset provided the unique identifiers of children and their parents. We linked the Taiwan Maternal and Child Health Dataset to the Taiwanese Birth Notification, the Death Registry, and the National Health Insurance Database. The National Health Insurance Database holds maternal and paternal medical claims with International Classification of Diseases (ICD)-9 and ICD-10 codes for diagnoses from each medical visit and information on every prescription filled at a pharmacy in Taiwan. We also linked these data to the Cancer Registry (diagnoses from 2004 to 2017). The analysis included 2,780 cancer cases diagnosed before age 13 and 2,291,512 non-cases (Figure 1).

Figure 1-.

Flow chart of selection of study participants

We identified mothers with chronic hypertension, gestational hypertension, preeclampsia, and eclampsia using ICD-9 and ICD-10 codes as previously identified [33, 34]. The ICD-9 codes used were hypertension (401x-405x), gestational hypertension (642.3x), mild preeclampsia (642.4x), severe preeclampsia (642.5), eclampsia (642.6x), preeclampsia or eclampsia superimposed on pre-existing hypertension (642.7x), and other hypertension complicating pregnancy, childbirth, and the puerperium (642.0, 642.1, 642.2, 642.3, 642.9). We estimated the last menstrual period from the child’s birthdate subtracting gestational week information obtained from the birth notification (i.e., birthday – gestational weeks*7 – 14 days).

Medications used by mothers were ascertained using the following Anatomical Therapeutic Chemical (ATC) classification codes: (C02) antihypertensives, (C03) diuretics, (C07) beta-blockers, (C08) calcium channel blockers, and (C09) agents acting on the renin-angiotensin system.

Conducting a complete case analysis, we assessed the associations of maternal hypertension type or maternal antihypertensive drug type use in pregnancy with risks of multiple pediatric cancers using Cox proportional hazard modeling to estimate the hazard ratios (HR) and their 95% confidence intervals (CI). We estimated associations with the risks of cancers in children born to mothers with hypertension (before or during index pregnancy), preeclampsia/eclampsia, and gestational hypertension (during the index pregnancy). Adjustment variables were chosen based on the literature, and we adjusted for parity, family income, maternal infections (using ICD coding previously identified [35]), preeclampsia/eclampsia in pregnancy, and hyperthyroidism in pregnancy [4, 36–38]. Preeclampsia/eclampsia was removed from the adjustment variable when it was analyzed as the variable of interest. We also considered adjusting for smoking and alcohol use, diabetes, and gestational diabetes, but we left the variables out of the final models because the effect estimates did not change after inclusion. All analysis was done using SAS 9.4 (SAS Institute, Cary, NC).

Results

The distribution of cancer types and their relation with covariates was previously reported [4]. The characteristics of the study population are shown in Table 1. Cases and non-cases were similar with respect to birth year, maternal and paternal age, family income, urbanization level of the inhabited area. Parity was higher in the families of cases.

Table 1.

Demographic Characteristics of Children in the Study Born in Taiwan Between the Years 2004 to 2015

	Cases (n=2,780)	Non-cases (n=2,291,512)
	N (Percent)	N (Percent)
Maternal age
Mean (SD)	30.11 (4.8)	30.35 (4.8)
<20	64 (2.3%)	44,569 (1.9%)
20–29	1,282 (46.1%)	1,006,576 (44.0%)
30–34	1,013 (36.5%)	855,183 (37.4%)
35–39	358 (12.9%)	332,975 (14.6%)
40+	61 (2.2%)	47,974 (2.1%)
Missing	2	4,235
Paternal age
Mean (SD)	33.16 (5.4)	33.30 (5.4)
<20	11 (0.4%)	7,293 (0.3%)
20–29	739 (27.9%)	572,462 (26.3%)
30–34	990 (37.4%)	845,920 (38.8%)
35–39	628 (23.7%)	531,910 (24.4%)
40+	280 (10.6%)	221,324 (10.2%)
Missing	132	112,603
Family income
<30759	719 (27.0%)	550,116 (25.0%)
30759–48200	690 (25.9%)	547,313 (24.9%)
48200–73317	627 (23.5%)	553,088 (25.1%)
≥73317	630 (22.7%)	550,151 (25.0%)
Missing	114	90,844
The urbanization level of the inhabited area
High	1,404 (50.5%)	1,220,284 (53.3%)
Middle	1,103 (39.7%)	860,711 (37.6%)
Low	271 (9.8%)	209,766 (9.1%)
Missing	2	751
Parity
1	1,082 (39.0%)	922,722 (40.3%)
2	1,313 (47.2%)	1,124,641 (49.0%)
3 or more	385 (13.8%)	244,149 (10.7%)
Preeclampsia during the index pregnancy
No	2,701 (98.0%)	2,221,659 (98.0%)
Yes	53 (2.0%)	44,971 (2.0%)
Missing	26	24,882
Hypertension before or during pregnancy
Maternal hypertension (Any hypertension before or during pregnancy)
No	2,664 (96.8%)	2,194,415 (96.8%)
Yes	92 (3.3%)	72,656 (3.2%)
Missing	24	24,441
Chronic hypertension (Only with hypertension before 20 weeks of gestation)
No	2,715 (98.9%)	2,228,385 (99.0%)
Yes	28 (1.1%)	22,285 (1.0%)
Missing	37	40,842
Gestational hypertension (hypertension in pregnancy after 20 weeks of gestation)
No	2,725 (98.9%)	2,247,883 (99.2%)
Yes	29 (1.1%)	18,490 (0.8%)
Missing	26	25,139
Chronic hypertension during pregnancy (chronic hypertension before pregnancy and hypertension during pregnancy
No	2739 (99.8)	2,243,334 (99.6)
Yes	<5	7,333 (0.4)

Table 2 shows the relationship between maternal hypertension before or during pregnancy and cancer risk in offspring stratified by cancer type. There was an increased risk of acute lymphoblastic leukemia (HR = 1.87, 95% CI 1.32–2.65) and non-Hodgkin lymphoma (HR = 1.96, 95% CI 1.34 – 2.86) in the offspring of mothers with any hypertension and with chronic hypertension only. No increased risk was seen in children of mothers with preeclampsia.

Table 2.

Association between Maternal Hypertension, Preeclampsia before or during Pregnancy and Risk for Offspring Cancer, Stratifying by Cancer Types

				Crude	Adjusted *
		N (%)	N	HR (95%CI)	HR (95%CI)
	Total N	yes	no
	Maternal hypertension (All hypertension including chronic and gestational)
Non-cases	2,291,512	72,656 (3.17)	2,194,415	1 (ref)	1 (ref)
All cancers	2,780	92 (3.31)	2,664	1.14 (0.93–1.40)	1.12 (0.90–1.41)
Acute lymphoblastic leukemia	802	37 (4.61)	760	1.62 (1.17–2.26)	1.87(1.32–2.65)
Non-Hodgkin lymphoma	658	33 (5.02)	623	1.81 (1.27–2.57)	1.96 (1.34–2.86)
Central nervous system tumors	389	9 (2.31)	378	0.81 (0.42–1.56)	0.91(0.46–1.81)
	Maternal chronic hypertension (pre-pregnancy or diagnosis before 20 weeks gestational age)
Non-cases	2,291,512	22,285 (0.97)	2,228,385	1 (ref)	1 (ref)
All cancers	2,780	28 (1.01)	2,715	1.14 (0.78–1.64)	1.07 (0.72–1.59
Acute lymphoblastic leukemia	802	13 (1.62)	778	1.86 (1.08–3.22)	1.95 (1.12–3.37)
Non-Hodgkin lymphoma	658	13 (1.98)	640	2.33 (1.34–4.03)	2.25 (1.27–3.99)
	Maternal gestational hypertension during the index pregnancy
Non-cases	2,291,512	18,490 (0.81)	2,247,883	1 (ref)	1 (ref)
All cancers	2,780	29 (1.04)	2,725	1.36 (0.94–1.96)	1.29 (0.88–1.90)
	Maternal preeclampsia during pregnancy
Preeclampsia during the index pregnancy (any)**
Non-cases	2,291,512	44,971 (1.96)	2,221,659	1 (ref)	1 (ref)
All cancers	2,780	53 (1.91)	2,701	1.03 (0.79–1.35)	1.06 (0.80–1.40)
Mild preeclampsia during the index pregnancy
Non-cases	2,291,512	32,009 (1.40)	2,234,621	1 (ref)	1 (ref)
All cancers	2,780	35 (1.26)	2,719	0.96 (0.69–1.33)	0.99 (0.70–1.38)

^*Model adjusted for parity, preeclampsia/eclampsia in the index pregnancy, family income, infection in pregnancy and hyperthyroidism.

^**Preeclampsia during the index pregnancy adjusted for parity, and hyperthyroidism

When stratifying by medication use (Table 3), we saw only a weak increase in cancer risk in children whose mothers had hypertension and who used diuretics before birth, with wide confidence intervals (HR=1.16, 95% CI 0.77–1.74) and similar point estimates for those who used antihypertensives other than diuretics (HR=1.15, 95% CI 0.86–1.54).

Table 3.

The Childhood Cancer Risk among Offspring of Mothers with Hypertension before Birth, Stratifying by Medication Use

	Hypertension before birth (all types)
	Cases (n=2,780)	Non-cases (n=2,291,512)	Crude HR (95% CI)	*Adjusted HR (95% CI)
No diagnosis of hypertension before birth	2,664	2,194,495	1 (ref)	1 (ref)
Maternal hypertension before birth	92	72,656	1.14 (0.93–1.40)	1.12 (0.90–1.41)
No medications before birth	12	11,403	0.94 (0.53–1.65)	0.98 (0.56–1.73)
Any medications before birth other than diuretics	54	40,583	1.20 (0.92–1.57)	1.15 (0.86–1.54)
Diuretics before birth	26	20,670	1.13 (0.77–1.66)	1.16 (0.77–1.74)
Hypertension and diuretic use between conception (0 weeks) and birth	10	8,945	0.98 (0.53–1.83)	0.94 (0.48–1.83)
Hypertension but without diuretic use between conception (0 weeks) and birth	80	63,350	1.13 (0.91–1.42)	1.13 (0.89–1.43)

^*Model adjusted for parity, preeclampsia/eclampsia in the index pregnancy, family income, infections in pregnancy, and hyperthyroidism

Discussion

This study is one of the first studies in an Asian population to describe pediatric cancer risk in relation to maternal hypertension, preeclampsia, and the use of antihypertensives. Importantly, we observed increased risks for cancer (ALL and NHL) with maternal hypertension. Sample size restrictions prevented us from examining specific childhood cancer types previously associated with these maternal health conditions, most notably neuroblastoma and hepatoblastoma in relation to preeclampsia. The prevalence of antihypertensive use in pregnancy in the United States increased from 3.5% to 4.9% during the years 2000–2007 and it is mostly used in the first and third trimester [39]. The prevalence of antihypertensive medication use in our study population in Taiwan was lower at 1.9%. Our study is one of the few studies that assessed both diuretics or other antihypertensives and presented effect estimates for specific maternal drug use and childhood cancers. We observed only weakly increased risk between intake of any antihypertensive medication in relation to childhood cancer (all cancer types combined).

We found elevated risk estimates for childhood cancer, specifically acute lymphoblastic leukemia (ALL) and non-Hodgkin’s lymphoma (NHL) and maternal hypertension (chronic and gestational). When we stratified by duration of hypertension, the risk was more apparent in mothers with chronic hypertension. Our findings confirmed those of previous studies that found elevated but imprecise risk estimates for ALL and NHL in offspring of mothers with hypertension [19, 25], however not all studies have been consistent [20, 25]. The specific mechanisms of this elevated risk could not be ascertained from our study but a previous study reported increased risk of Epstein-Barr virus (EBV) reactivation in people with hypertension [40]. Hypertension can induce injury in arteries and arterioles causing endothelial damage. The body tries to repair the damage endothelium of blood vessels by releasing leading inflammatory cytokines and vascular endothelial growth factors. The increased inflammatory cytokines may enhance immune activation and lead to the reactivation of EBV [40]. EBV has been associated with increased risk of non-Hodgkin’s lymphoma in immunocompromised patients and this virus is transferable from mothers to their offspring [41]. This could be a possible etiologic mechanism of the relationship between maternal hypertension and childhood cancer.

Our analysis of all childhood cancers did not show any increased risks in children of mothers with preeclampsia. However, we lacked statistical power to examine specific cancer types most often associated with preeclampsia in earlier studies. Further, results may be partly explained by the study’s time period. Improved preeclampsia management in recent decades may have reduced the risk of childhood cancer, resulting in lower effect estimates in more recent studies [18].

We did not have information on maternal blood pressure in pregnancy and therefore could not ascertain the severity of maternal hypertension. It is likely that antihypertensives were used by those with more severe hypertension in pregnancy. Hypertension is the major indication for antihypertensive use in pregnancy, while other indications include preeclampsia, ascites due to kidney failure, liver disease, cardiovascular conditions, and previous cancers. Apart from preeclampsia and family cancer history these conditions have not been associated with childhood cancer in offspring, however several of these conditions are understudied.

First line medications for treating hypertension include methyldopa and labetalol which have been associated with intrauterine growth restriction (IUGR) [18], and IUGR has been previously linked with childhood cancers [4]. Other medications include slow-release nifedipine and some beta blockers.

Although other studies have observed associations between childhood cancers and maternal use of diuretics, we observed only weakly elevated point estimates and no difference in results between diuretics and other antihypertensives. The increased risks of cancers in children of mothers using antihypertensives may be due to the transplacental transfer of some of the breakdown products of the medications. Some medications used to treat hypertension such as methyldopa, nitrendipine, and propranolol, diuretics like furosemide and hydrochlorothiazide, contain nitrosatable amines, constituents of N-nitroso compounds (NOC) [31, 42]. NOCs have been associated with cancer induction in diverse species, including humans, as experimental studies have shown that transplacentally administered nitrosamides are potent neuro-carcinogens. For example, experimental studies in rats show that transplacental exposure to nitrosamides produces neurogenic tumors in the offspring [43], particularly during specific developmental stages [31, 44]. In humans, a case-control study of patients in a cancer surveillance program in Los Angeles that examined the effect of N-nitroso compounds on childhood brain tumors reported a two-fold increased odds of childhood brain tumors in children of mothers who used diuretics containing N-nitroso compounds in pregnancy compared to children of mothers who did not use diuretics [32]. Other studies also reported an increased risk of tumors [42, 45, 46], with the risk increasing when the exposure happened in the first four months of pregnancy. However, some studies did not find an increased risk [7, 47].

The strengths of this study in Taiwan include its population-based design, as it included all children born in the country during the study period. Cases and non-cases were determined from nationwide cancer and birth registries covering a population of over 2 million children. Maternal medication use was ascertained by relying on pharmaceutical data from insurance records and was not self-reported, preventing recall bias. However, we could not be certain if the mothers took the medications after filling their prescriptions, which is a limitation of the study as this could have resulted in non-differential exposure misclassification that most likely biased the result towards the null. Another limitation of our study is that we could not stratify by more specific type of diuretics or other antihypertensives the mothers used due to low statistical power. We also did not have data on maternal pre-pregnancy BMI in the cohort. This paper may have been subject to live-birth bias if we consider the risk of possible fetal death from preeclampsia, which may have underestimated the associations. To check for confounding by indication, we adjusted for hypertension when examining the relation with antihypertensive medications and the increases in risk we observed was due to hypertension rather than medications used for treatment.

In conclusion, our study suggests that there may be an association between maternal hypertension and the development of childhood cancers. Although our inconclusive results for antihypertensives might provide reassurance to pregnant women taking antihypertensives, the suggestive associations seen in other studies may suggest that women should proceed with caution before taking those medications in pregnancy. Our results suggest that women should try to get their hypertension under control prior to pregnancy.

Data availability statement

This study is based on deidentified information from Taiwan national health insurance and registries, and permission is required prior to data access. Further details are available from the corresponding author upon request.