Consistent with expectation,6,9,13
the incidence of pregnancy-associated cancer increased from 1994 to 2007. Although maternal age was a strong risk factor for cancer, increasing maternal age explained only 14% of the increase. Improved diagnostic techniques, detection and increased interaction with health services during pregnancy might also contribute to higher incidence rates. The cancer incidence was higher than expected among women of reproductive age, and over two-thirds of pregnancy-associated cancers were diagnosed in the 12 months after delivery. The genetic and environmental orgins of pregnancy-associated cancers are likely to pre-date the pregnancy; however, the hormones and growth factors necessary for fetal growth may accelerate tumour growth.
Between 1994 and 2008, the overall incidence of pregnancy-associated cancer in NSW was 137.3 per 100 000 maternities, which is higher than the 100 per 100 000 generally reported in the literature.1–5
Our reported estimates for the most common types of cancer are also higher than the published international estimates: melanoma of skin (45.7 versus 8.7 per 100 000), breast cancer (28.8 versus 19.3 per 100 000) and thyroid cancer (17.4 versus 14.4 per 100 000), except for cervical (8.4 versus 12.0 per 100 000) and ovarian cancers (3.6 versus 5.2 per 100 000).1
Our overall higher incidence partly arises from the predominance of melanoma, for which Australia has the highest incidence in the world.31
Slightly higher rates of the other cancers probably reflect our older and recent study population.
The observation that the majority of cancers (72.2%) were diagnosed in the postpartum period has been noted elsewhere.1,9
Some of these cancers may have been suspected given the high rate of planned preterm delivery for women whose cancer was diagnosed postpartum, but this hypothesis could not be verified in the current data set. Another possible explanation is that the physiologic changes of pregnancy may make cancer more challenging to diagnose, leading to a delay in diagnosis. The distribution of the timing of initial diagnosis is consistent with the contention that postpartum cancers are part of the cancer-in-pregnancy continuum, and are appropriate to be included in the overall incidence.1
The steady increase in the number of cancer diagnoses as the duration of pregnancy advanced, which peaked in the 2 months postpartum, may be opportunistic because of pregnancy surveillance.1
The increased rate of cancer associated with pregnancy above that expected (based on the rates in the general female population) may be explained in two-fold. First, antenatal and postnatal care, involving a standard protocol of history, physical examination, cervical cytology and blood pressure monitoring, is available for Australian women during and after their pregnancy. On the basis of these routine care visits a screening effect might be expected, thereby increasing the chance of detection of cancer in association with pregnancy. That the melanoma rate is over two times higher than expected in the general population is consistent with such an effect. Second, pregnancy is a proangiogenic state, and it is plausible that the angiogenesis (driven by factors such as placental growth factor and vascular endothelial growth factor) required for successful placentation and pregnancy outcomes contributes to tumerogenesis or growth. In vitro
placental growth factor increases the proliferation of melanoma cells,32
and animal models demonstrate that metastasis is associated with the increased expression of vascular endothelial growth factor receptor.33
Furthermore, women with cancer diagnosed during pregnancy were more likely to have multiple pregnancies, and to have large-for-gestational-age infants, even after adjustment for pre-existing or gestational diabetes. High birthweight is an established risk factor for childhood cancers (Wilms’ tumour, infant and childhood leukaemia, osteosarcoma and astrocytoma),34
and twins have increased risks of endocrine, bone and breast cancers.35,36
The postulated mechanism is elevated levels of maternal hormone factors during pregnancy, such as estrogens and insulin-like growth factor I levels, or the aforementioned angiogenic factors.37
This mechanism could also predispose maternal cancer, and deserves further investigation. There are few studies that have examined these associations for mothers. We found only one study that examined infant size and breast cancer during pregnancy, which like us, found no increased risk in infant size for breast cancer.37
Furthermore, although studies of maternal cancer later in life have examined the risk for women who had multiple pregnancies, the findings have been contradictory, and none have examined pregnancy-associated cancers.38
The timing of cancer diagnosis had an effect on adverse pregnancy outcomes. The risks of thromboembolic events, sepsis and severe morbidity, which are recognised cancer complications, were higher among women with cancer diagnosed during pregnancy. Among postpartum cancers, we only demonstrated a significant risk of postpartum sepsis. This is consistent with the fact that these women are more prone to infections and malignancy-related immune suppression, or may have had cancer treatments prior to delivery. The higher rate of caesarean delivery is likely to reflect the standard management plan of certain types of cancer (e.g. cervical cancer) in pregnant women.9
Higher rates of planned preterm delivery, to allow the postpartum initiation of cancer treatment, has previously been reported to be common for women with cancer in pregnancy.9
The timing of obstetric delivery is a controversial issue surrounding the management of cancer associated with pregnancy. It has been reported that a deliberate delay in treatment is not associated with poor survivorship for pregnant women with early-stage cancer,5
and therefore early elective delivery should be carefully considered to ensure the best outcomes for both the mother and the neonate. However, this general finding needs to be assessed by cancer type.
Our findings of higher risks of caesarean delivery and premature birth are consistent with the findings of the population-based Californian studies of pregnancy-associated cancer.9
Similar results have also been reported from cancer registry studies on breast, cervical and colorectal cancers, independently.7,10,11
These studies had methodological limitations, however. The Californian studies have assessed pregnancy outcomes solely based on hospitalisation records, in which gestational age and perinatal death are poorly ascertained. Consequently, size at birth was defined by birthweight, which does not differentiate size and maturity. Other limitations included restriction to a specific cancer type, resulting in small sample sizes, and no adjustment for socio-economic status and maternal clinical conditions (hypertensive disorders and diabetes) in assessing pregnancy outcomes.
The strength of our study is the population-based incidence and outcomes of cancer associated with pregnancy derived from large, validated and contemporary data sources. Importantly, there is a complete registration of cancers and births in statutory data collections in NSW. The linkage to cancer registry dating back to 1972 provided the opportunity to assess history of cancer as a risk factor for pregnancy-associated cancer. However, several limitations of our study warrant consideration. First, as early pregnancy loss (miscarriage or abortion) was not registered in the birth data, the number of pregnancy-associated cancers will be somewhat underestimated, and the average gestational age at diagnosis will be over-estimated. Second, we could not examine cancer treatment, as chemotherapy and radiotherapy are primarily provided in outpatient clinics, and are under-ascertained in the hospitalisation records.39
Third, history of smoking, alcohol consumption and maternal obesity were not available to provide adjustment for the potential confounding of pregnancy outcomes. Finally, the exact date of cancer diagnosis was not available (month and year only), so the uncertainty between the timing of diagnosis and birth or subsequent adverse outcomes may be as much as 4 weeks.