Hyperemesis Gravidarum and the Potential for Cancer: A Longitudinal Cohort Study over Three Decades

The association between hyperemesis gravidarum and cancer is poorly understood. Hyperemesis gravidarum is a severe form of nausea and vomiting that occurs in up to 2% of pregnant women (1). Although elevated levels of human chorionic gonadotropin (hCG) are thought to play a role in the development of hyperemesis gravidarum (2), recent data suggest that the GDF15 gene may also be an important contributor (1, 3). Women with hyperemesis gravidarum have increased expression of GDF15 (1, 3), a cell-cycle regulator known to promote carcinogenesis (4, 5). The GDF15 gene is associated with several types of cancers, including bladder, renal, and colorectal neoplasms (5). Yet the possibility that women with hyperemesis gravidarum have an elevated risk of cancer has been understudied.

Research on the relationship between hyperemesis gravidarum and cancer is sparse (6–8). A nested case–control study that was large enough to study specific tumors found conflicting evidence of an association with hyperemesis gravidarum (6). Among four cancer types studied, hyperemesis gravidarum was associated with an increased risk of thyroid cancer, but decreased risk of rectal, lung, and cervical cancers (6). Hyperemesis gravidarum has not been associated with breast cancer or cancer-related mortality in smaller studies (7, 8). A number of studies found that less severe types of nausea and vomiting were associated with a decreased risk of breast cancer, with the exception of an increased risk of HER2-enriched breast cancer (9, 10). However, women with less severe nausea and vomiting were classified as having hyperemesis gravidarum. As the association with different types of neoplasm was not clear, our objective was to assess whether hyperemesis gravidarum was associated with a range of cancers in a large population-based cohort of women with three decades of follow-up.

We conducted a longitudinal cohort study of 1,343,040 women who delivered at least one infant between April 1989 and March 2019 in Quebec, Canada. We obtained data on each pregnancy from hospital discharge summaries in the Maintenance and Use of Data for the Study of Hospital Clientele registry (11). The registry contains all deliveries that occurred in hospitals in Quebec (99% of births), as well as all day admissions or hospitalizations for cancer or other medical conditions. The data have high coverage as the management of cancer requires hospital services in Quebec. Validation analyses indicate that the data provide equivalent or superior coverage of cancers compared with the tumor registry (12).

Using encrypted health insurance numbers, we followed women through time to capture hospitalizations for cancer up to 31 years after pregnancy. Follow-up began at the first delivery and ended at the time of cancer hospitalization, death, or the end of study on March 31, 2020. We restricted the cohort to residents of Quebec with valid health insurance numbers who could be followed over time. The cohort includes most of the population as all nontemporary residents are covered under provincial health insurance. We excluded women with a history of cancer before their first pregnancy to ensure that we only captured incident tumors.

The main exposure measure was hyperemesis gravidarum, defined as extreme or excessive vomiting during pregnancy with onset before the 23rd week of gestation, identified using diagnostic codes in the 9th and 10th revisions of the International Classification of Diseases (ICD-9 643.0, 643.1, 643.9; ICD-10 O21.0, O21.1, O21.9; refs. 6, 7, 13). We classified women without hyperemesis gravidarum as unexposed until their first pregnancy with hyperemesis. As a secondary exposure measure, we included a multicategoric variable capturing the severity of hyperemesis gravidarum (mild 643.0/O21.0, severe with metabolic disturbance 643.1/O21.1, mixed 643.9/O21.9). Severe hyperemesis included patients with metabolic disturbances, comprising electrolyte imbalance, dehydration, or ketonuria. Mild hyperemesis was defined as excessive vomiting without metabolic disturbance. Mixed severity included excessive vomiting of varying intensity that warranted hospitalization (6).

The main outcome measure was hospitalization for malignant cancer any time after the first delivery. We used morphology and topography codes in the International Classification of Disease for Oncology-3 to identify malignant neoplasms (Supplementary Table S1; ref. 11).

As a hypothesized cause of hyperemesis gravidarum is abnormally high hCG (2), we grouped tumors based on the presence or absence of hCG receptors in the organ of origin (14–16). Categorization was based on the presence of hCG receptors in the organ, rather than the tumor itself. We further categorized tumors originating from hCG receptor⁺ organs by embryonic germ cell layer (endoderm, mesoderm, and ectoderm), as recent analyses suggest that carcinogenic pathways in different organs may vary depending on the embryonic germ cell layer from which they develop (17). A genetic analysis of tumors organized by germ cell layer found that endoderm tumors had more mutations in GATA6, GATA4, AFP, and Hedgehog signaling pathway genes compared with mesoderm and ectoderm tumors (17).

As secondary outcomes, we analyzed specific subtypes of endodermal, mesodermal, ectodermal, and hCG receptor⁻ cancers, including central nervous system, oral cavity, upper gastrointestinal tract and small bowel (esophagus, stomach, and small intestine), colorectal, lung, bladder, kidney, breast, uterine, cervical, ovarian, other reproductive cancers (vulva, vagina, uterine ligaments), thyroid, hematolymphopoietic, connective and soft tissues, bone, and melanoma and other malignant skin cancers.

We adjusted for potential confounders associated with both hyperemesis gravidarum and cancer, including maternal age (<25, 25–29, 30–34, ≥35 years), multiple gestation (yes, no), fetal congenital anomaly (yes, no), maternal comorbidity (yes, no), socioeconomic deprivation (yes, no, unknown), and time period (1989–1998, 1999–2008, 2009–2019). We measured these covariates at the first delivery. Maternal comorbidity included obesity, type 1, 2, or gestational diabetes, dyslipidemia, hypertensive disorders of pregnancy, and alcohol, tobacco, or substance use disorders during pregnancy (11). Socioeconomic deprivation represented the most materially disadvantaged fifth of the population based on neighborhood income, education, and employment (11).

We calculated cancer incidence rates per 10,000 person-years and the cumulative incidence of cancer after 31 years of follow-up. We plotted incidence curves using the cumulative incidence function.

Using Cox regression models, we computed unadjusted and adjusted hazard ratios (HR) and 95% confidence intervals (CI) for the association of hyperemesis gravidarum and the risk of cancer. In adjusted models, we accounted for baseline age, multiple gestation, fetal congenital anomaly, maternal comorbidity, socioeconomic deprivation, and time period. We treated hyperemesis gravidarum as a time-varying exposure using the counting process method. Women were thus considered unexposed until their first pregnancy complicated by hyperemesis gravidarum. The time scale was measured as the number of days between delivery and hospitalization for cancer, death from other causes, or the last day of the study. We censored women who did not develop cancer at the end of the study and treated death from other causes as a competing event using the Fine and Gray method. We used log (−log survival) plots to verify the proportional hazards assumption.

In sensitivity analyses, we tested whether associations persisted when hyperemesis gravidarum was measured as a time-fixed exposure, with women considered exposed if they had at least one affected pregnancy any time during the study. We also reran the analysis with data restricted to women with one pregnancy only, excluding women with mixed hyperemesis, adjusting for maternal comorbidities individually, and excluding cancers that developed within one year of delivery. We performed the analysis in SAS version 9.4 (SAS Institute Inc.) and received an ethics waiver from the institutional review board of the University of Montreal Hospital Centre because the data were anonymized and informed consent was not needed. This study complied with the ethical standards of the Declaration of Helsinki.

The data that support the study findings are available from the Quebec Statistics Institute following standard access procedures (https://statistique.quebec.ca/research/#/demarche/etape-par-etape). The syntax used to run the analysis can be accessed from the corresponding author upon request.

Among 1,343,040 women in the cohort, 16,738 women (1.2%) had at least one pregnancy affected by hyperemesis gravidarum (Table 1). Women with hyperemesis gravidarum tended to be younger at first pregnancy. A slightly higher proportion had a multiple gestation, carried a female fetus, or was socioeconomically disadvantaged compared with no hyperemesis gravidarum. Women with and without hyperemesis gravidarum had similar proportions of maternal comorbidity and fetal congenital anomaly.

Overall, women with hyperemesis gravidarum had a slightly lower cancer incidence (18.8 per 10,000 person-years; 95% CI, 17.3–20.4) than women with no hyperemesis (21.1 per 10,000 person-years; 95% CI, 20.9–21.3; Table 2). In adjusted Cox models, women with any hyperemesis gravidarum had the same risk of cancer as women without hyperemesis (HR, 1.00; 95% CI, 0.93–1.09). Confidence intervals included the null for all categories of hyperemesis.

When we plotted cumulative incidence curves of cancer by germ cell layer of origin, women with hyperemesis gravidarum had a greater incidence of endodermal cancer than women without hyperemesis gravidarum beginning around 12 years after their first pregnancy (Fig. 1). Women with hyperemesis gravidarum had a somewhat shorter median time to endodermal cancer onset (16.0 years) than women without hyperemesis (16.2 years). Cumulative incidence curves for women with and without hyperemesis gravidarum did not differ for ectodermal cancer before five years, mesodermal cancer before 18 years, and cancers that were hCG receptor⁻ before 15 years. After these time points, women with hyperemesis gravidarum had a lower incidence of ectodermal, mesodermal, and hCG receptor⁻ cancer than women without hyperemesis gravidarum.

Regression analyses confirmed that women with hyperemesis gravidarum had a higher risk of endodermal cancer than women without hyperemesis gravidarum, but not mesodermal or ectodermal cancer (Table 3). Compared with unexposed women, women with hyperemesis gravidarum and metabolic disturbance had nearly two times the risk of endodermal cancer (HR, 1.97; 95% CI, 1.51–2.58) and women with mild hyperemesis had 1.19 times the risk (95% CI, 1.03–1.37), supporting a dose–response effect. There was no dose–response trend for mesodermal, ectodermal, or hCG receptor⁻ cancers. Women with hyperemesis gravidarum had a lower risk of hCG receptor⁻ cancer (HR, 0.53; 95% CI, 0.38–0.75) than women without hyperemesis gravidarum. The reduced risk of hCG receptor⁻ cancer was most marked for mild hyperemesis (HR, 0.39; 95% CI, 0.27–0.56).

The increased risk of endodermal neoplasms among women with hyperemesis gravidarum was primarily driven by bladder, colorectal, and thyroid cancers (Table 4). Compared with no hyperemesis, women with hyperemesis gravidarum had 2.49 times the risk of bladder cancer (95% CI, 1.37–4.53), 1.41 times the risk of colorectal cancer (95% CI, 1.08–1.84), and 1.34 times the risk of thyroid cancer (95% CI, 1.09–1.64). The decreased risk of hCG receptor⁻ tumors was driven by lung cancer (HR, 0.44; 95% CI, 0.28–0.68).

In sensitivity analyses, we found that measuring hyperemesis gravidarum as a time-fixed exposure, restricting the data to women with one pregnancy, excluding mixed severity hyperemesis, adjusting for each comorbidity individually, and excluding cancer that developed within one year of delivery did not affect the results (Supplementary Table S2).

In this cohort study of 1.3 million women with three decades of follow-up after pregnancy, hyperemesis gravidarum was associated with a greater incidence of cancer originating from the endodermal germ cell layer, but not cancer arising from the mesodermal and ectodermal cell layers or organs lacking hCG receptors. Compared with no hyperemesis, women with hyperemesis gravidarum had a greater risk of bladder, colorectal, and thyroid cancers. These three types of cancers arise mainly from tissues in the endodermal germ cell layer that have hCG receptors (16, 17). There was, moreover, evidence of a dose–response relationship: severe hyperemesis gravidarum with metabolic disturbance was associated with a greater risk of endodermal cancer than mild hyperemesis gravidarum. Overall, the findings of this study suggest that hyperemesis gravidarum may be a novel risk factor for endodermal cancer in women, but not for other cancer types.

Our findings tend to align with the small number of studies that have examined the association between hyperemesis gravidarum and specific cancers (6–10). A Norwegian cohort study of long-term mortality found that hyperemesis gravidarum was associated with a lower risk of death from tobacco-related cancer (7), likely driven by a lower risk of lung cancer. In a Scandinavian nested case–control study that examined the association with several individual cancers, results were reminiscent of ours, with an increased risk of thyroid cancer and decreased risk of lung cancer among women with hyperemesis gravidarum (6). Contrary to our findings, the authors found a protective association with rectal cancer and no association with urinary cancers (6). However, the Scandinavian cohort had a somewhat older average age, suggesting that hyperemesis gravidarum may only increase the risk of early-onset colorectal and bladder cancers. As in our study, all prior studies of hyperemesis gravidarum have found no association with breast cancer (6, 8). Studies of general nausea and vomiting in pregnancy have also found no association with breast cancer (9, 10).

The current understanding of the pathways linking hyperemesis gravidarum with cancer is nascent. Although the pathways remain to be confirmed, one possibility is that hyperemesis gravidarum is connected to certain cancers through circulating hCG levels or GDF15 gene modulation. hCG, the hormone produced by trophoblastic cells during placental development (16), is known to be elevated in women with hyperemesis gravidarum (17). hCG elevation is a hallmark of hyperemesis gravidarum, with levels usually peaking around the time of onset of hyperemesis (18). Similarly, GDF15 is expressed at high levels by the placenta (19). Women with hyperemesis gravidarum have a greater frequency of GDF15 locus variants and increased serum levels of GDF15 (1, 3). However, hCG and GDF15 are not only implicated in the etiology of hyperemesis gravidarum but also contribute to cancer development (4, 5, 20).

hCG is involved in regulating cell growth and proliferation (20). In vitro research has shown that cells treated with hCG have increased expression of GATA4 (21), a gene involved in cell growth and tumor development (22). Murine models have demonstrated that GATA4 and GATA6 are needed for the expression of hCG receptors (23), whereas GATA6 has been shown to suppress lung cancer in both in vitro and in vivo experiments (24). Mutations in GATA4, FOXA2, GATA6, and AFP are more common in endodermal tumors than in ectodermal and mesodermal tumors (17). In our data, hyperemesis gravidarum was associated with an increased risk of cancer arising from the endodermal germ cell layer. As there was a dose–response pattern in step with hyperemesis severity, higher hCG levels seen in hyperemesis gravidarum could potentially disturb cellular signaling and increase the risk of cancer. hCG may also overstimulate the thyroid, priming the thyroid for neoplastic transformation (6). Hyperemesis gravidarum was strongly associated with thyroid cancer in our data set.

GDF15 may also play a role in the pathway between hyperemesis gravidarum and cancer. GDF15 is widely expressed in the placenta, and women with hyperemesis gravidarum commonly have elevated serum levels of GDF15 (1, 3). The GDF15 gene codes for a cytokine ligand of the tumor growth factor-beta family of proteins that is regulated by the p53 tumor suppressor gene (25). GDF15 can promote or inhibit cancer growth depending on cellular context, disease stage, and tumor microenvironment (19, 26). In vivo and in vitro studies point to a tumorigenic role of GDF15 in bladder, gastrointestinal, and thyroid cancers (19, 26–29). GDF15 may have a tumor suppressor effect and promote apoptosis in lung adenocarcinoma (26). These observations align with our findings as we found that hyperemesis gravidarum was associated with an increased risk of bladder, colorectal, and thyroid cancers, but a decreased risk of lung cancer. Epigenetic modifications in GDF15 have also been reported in bladder cancer cell experiments (26), and GDF15 polymorphisms are more frequent in colorectal cancer patients (28).

The potential involvement of hCG and GDF15 in the pathway between hyperemesis gravidarum and cancer may explain why we found associations only with cancers arising from the endodermal germ cell layer. GATA transcription factors, primarily those coded by GATA4/5/6, are needed for the development of endoderm and mesoderm tissues (22). However, mutations in GATA4 and GATA6 are more common in endoderm-derived tissues (17) and may explain why we did not find an increased risk of mesodermal cancers. Other GATA genes (GATA1/2/3) are involved in the development and differentiation of the ectoderm and mesoderm (22), and these genes do not seem to have the same potential for modulation by hCG (23).

Other potential pathways linking hyperemesis gravidarum with cancer have been speculated but are less convincing (6, 7). For example, some researchers have proposed that tobacco use may explain the link owing to the strong association between smoking and several types of cancers (7). However, women who smoke are less likely to develop hyperemesis gravidarum (7). Moreover, hyperemesis gravidarum was associated with a lower risk of lung cancer in our study, yet increased risk of other smoking-related tumors such as colorectal and bladder cancers. As there was a low number of lung cancers, the protective association with lung cancer should be interpreted with caution. Estrogen exposure has been considered another potential route, as women with hyperemesis gravidarum have higher estrogen levels (6, 8). However, prior research has not found a consistent association between hyperemesis gravidarum and reproductive cancers (6, 8, 9), and we found no association with breast or ovarian cancer. It is unlikely that closer surveillance explains the findings for thyroid cancer, as thyroid screening is not part of recommendations for women with hyperemesis gravidarum.

This study has several limitations. We used registry-based data in which there may be misclassification of exposures or outcomes. As it is unlikely that misclassification was differential, the associations that we measured may be underestimated. Although we adjusted for confounders such as obesity, tobacco, and alcohol use, these covariates may be underreported. We did not have detailed information on smoking status and could not model covariates as time-varying. Thus, we cannot completely rule out residual bias. We did not have information on prepregnancy body mass index and nutrition, which may be risk factors for hyperemesis gravidarum and some cancer types. We did not have data on medications used to treat hyperemesis gravidarum and could not assess if treatment modified later cancer risk. We did not have information on women with mild hyperemesis gravidarum who did not require treatment, although we had complete coverage of severe cases that are the most likely to be pathologic.

This study found that hyperemesis gravidarum is associated with an increased risk of cancer arising from the endodermal germ layer of hCG receptor⁺ organs, primarily the bladder, colorectum, and thyroid. Our results support the possibility that mechanisms connecting hyperemesis gravidarum and cancer involve an interplay between elevated hCG levels, variants of GDF15, and tissues originating from the endodermal germ cell layer. Subsequent research would benefit from determining the absolute impact of hyperemesis gravidarum on risks of endoderm-derived tumors, as well as how hyperemesis compares with traditional risk factors for cancer. Further epidemiologic studies should consider the embryonic germ cell layer of origin of tumors to help identify novel molecular pathways involved in cancer development.

N. Auger reports grants from the Canadian Institutes of Health Research and Fonds de Recherche du Québec–Santé during the conduct of the study. No disclosures were reported by the other authors.

S. Marcoux: Conceptualization, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. V. Leduc: Writing–review and editing. J. Healy-Profitós: Conceptualization, formal analysis, methodology, writing–original draft, writing–review and editing. M. Bilodeau-Bertrand: Conceptualization, writing–review and editing. N. Auger: Conceptualization, resources, supervision, funding acquisition, validation, methodology, writing–original draft, writing–review and editing.

S. Marcoux was supported by the Fonds de recherche du Québec-Santé (284477). N. Auger was funded by the Canadian Institutes of Health Research (PCC-170244) and the Fonds de recherche du Québec-Santé (296785). V. Leduc, J. Healy-Profitós, and M. Bilodeau-Bertrand did not have specific funding.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.