Making sense of the risks: what to tell adolescents and young adults diagnosed with cancer during pregnancy

The diagnosis of cancer is a devastating one, even more so when it occurs during a pregnancy. Many adolescents and young adults (AYA) who are diagnosed with cancer want to be able to have a family, but their decisions about parenthood are complicated by worries about their personal health and their potential child’s health (1). Decisions about whether to continue the pregnancy, delay initiation of cancer therapy, alter therapy during pregnancy to protect the fetus, or deliver the baby prior to term present a major challenge to the medical team and considerable stress to the pregnant individual. Several key considerations shape decisions about how to balance treating the cancer with supporting the pregnancy: 1) how will the choices made impact the mother’s likelihood of cancer survival; 2) how will maternal health otherwise be impacted by the cancer and/or its treatment; 3) how might these choices impact the course or outcome of the neonate; and 4) how will they impact long-term offspring health in pregnancies that result in live births. In this issue of the Journal, Betts and colleagues (2) address the third and part of the fourth consideration in their report, “Adverse Birth Outcomes of Adolescent and Young Adult Women Diagnosed With Cancer During Pregnancy,” contributing information that health-care providers can use to counsel and care for this unique population as well as informing researchers on the necessary direction of future research.

In their study, the authors linked data from the Texas Cancer Registry with vital records and the Texas Birth Defects Registry to identify births among 1291 AYAs who were diagnosed with cancer during pregnancy over an 18-year period, representing 2.3% of all new cancer cases over this period. The rarity of cancer diagnosed during pregnancy underlies the importance of being able to use high-quality, linked administrative data to delineate the maternal and child health outcomes of this population. The study, one of the largest population-based studies of its kind to be conducted in the United States, has several strengths, including the high ascertainment rate of the Texas Cancer Registry, the use of a registry to capture birth defects, and the high representation of racial and ethnic minorities in the cohort. Compared with matched individuals without a history of cancer, AYAs with cancer had a threefold higher prevalence of low birthweight (<2500 grams) offspring, fivefold higher prevalence of premature birth before 32 weeks, and a 1.8-fold higher prevalence of newborns with low Apgar scores. The magnitudes of these unadjusted risks are clinically significant and should be disseminated to oncology, obstetrics, and neonatal providers. Encouragingly, the prevalence of birth defects detected in offspring by 12 months of age did not differ between groups—this is reassuring given that almost one-quarter of the cancers were diagnosed during the first trimester when organogenesis occurs.

In the current study, individuals treated for cancer during their pregnancy were statistically significantly more likely to have a preterm delivery than the matched cancer-free cohort (30.8% vs 8.4%). The prevalence of preterm birth in the control cohort without cancer was similar to the 7.8% rate reported for singleton US births in 2015 (3), supporting the generalizability of the Texas data. Two compelling questions arise from this data. First, were preterm births spontaneous or induced? If preterm births were more likely to be spontaneous, then a richer understanding of the mechanisms leading to premature labor in individuals undergoing cancer therapy could inform strategies for prevention. If preterm births were more likely to be induced, delineating whether this was due to maternal pregnancy–related morbidity such as pre-eclampsia or a need to start additional cancer treatments prior to reaching term would inform risk discussions and clinical care decisions. Second, was chemotherapy associated with growth restriction? Because preterm birth was three- to nearly fivefold higher among the pregnant cohort, the proportion of low birthweight neonates would be expected to be higher. A more useful metric to assess how cancer and its therapy impact fetal growth would be small for gestational age, but this measurement was not reported here.

The usefulness of this study for informing clinical decision making is tempered by several limitations, some of which are a consequence of using registry data. Key among these is missing data about cancer therapies received (for example, almost 60% of cases were missing radiation data) and the absence of data about specific types and doses of chemotherapies. Limited availability of data to inform the generation of robust treatment-specific risk estimates for perinatal and maternal outcomes during cancer therapy has been a long-standing barrier to providing accurate risk counseling to inform decision making. It is well established that reproductive and other cancer late effects vary by treatment exposures (4,5), but most traditional data sources lack sufficient detail to elucidate the specific impact of different chemotherapy agents, doses and regimens; targeted therapies; and radiation doses and fields. Advances in informatics, particularly the ability to identify and extract data elements from large data sources such as cancer registries, electronic health records, vital records, and insurance administration databases may address this limitation. Cancer treatment exposures and outcomes can often be identified in administrative claims and electronic health record data (6-8). If machine learning and natural language processing methods improve to support detailed capture of exposures and outcomes, then linkage across insurers, health systems, cancer registries, and vital records could be a powerful tool for improving estimation of risks for this population (9).

Other data elements not available to the researchers in the current study included information regarding miscarriages and pregnancy terminations, both outcomes of importance in this population. Although the study reported on the prevalence of birth defects detected by 1 year from birth, it did not address the more general health outcomes of the babies. This is particularly important given not only the higher rates of low birth weight, prematurity, and low Apgar scores in these infants but also the possibility that exposure to chemotherapy or other cancer therapies in utero could have had direct health impacts beyond the risk for congenital abnormalities. It would have been interesting to determine whether birth defect rates differed according to the trimester in which the cancer was diagnosed. Birth defects caused by exposure to cancer therapies might have led to early pregnancy loss or a decision to terminate the pregnancy (an option that was still available to pregnant individuals in Texas during the years of this study), but both outcomes were not captured by the study. Therefore, it is plausible that the risk for birth defects might be underestimated—this outcome was only captured in live births and the 20 documented stillbirths. What is also not reported are the short- and long-term survival and maternal outcomes (eg, pre-eclampsia, severe maternal morbidity) of the pregnant individuals with cancer. Ideally, knowledge about neonatal outcomes presented in this paper would be accompanied by miscarriage and termination data, maternal health outcomes, cancer course of the AYA, and longer-term health of the live-born infants. Such data could be paired with cancer and pregnancy management decisions that were made by the medical team or the patient to understand the critical intersection between decisions and outcomes.

Despite some limitations in the available data, the authors are to be commended for conducting a large and rigorously analyzed study in the AYA population, a group that is notoriously understudied (10). The development of cancer during such a critical time in the life course, when individuals are often focused on transitions between education and employment, decisions about relationships and possibly the desire to have children, and the transformation from childhood to adulthood, can have considerable and often long-lasting impacts. Studies such as the current one by Betts et al. (2) are essential for informing how best to counsel and manage the many critical decisions that must be made by patients and their medical teams to maximize the likelihood for cancer cure while minimizing the long-term impact of the cancer and its therapy on their health and quality of life, and in this case, that of their offspring.

Data availability

No new data were generated or analyzed for this editorial.

Author contributions

Paul Craig Nathan, MD, MSc (Writing – original draft; Writing – review & Editing); Irene Su, MD, MSCE (Writing – original draft; Writing – review & Editing).

Funding

No funding was used for this editorial.

Conflicts of interest

The authors have no disclosures.