No patients were directly involved in the design of this study. The public was involved in the design of the study as part of the UKOSS Steering Committee.

All newly diagnosed cases of breast cancer during pregnancy in the UK were reported through the UK Obstetric Surveillance System (UKOSS) in women who delivered their babies or had a termination of pregnancy or miscarriage between October 1, 2015, and September 30, 2017. The total number of maternities in the UK over the 2 years of the study was used as the denominator to calculate the incidence. The incidence was calculated with 95% confidence intervals.

Women whose breast cancer had been diagnosed before pregnancy or who had recurrence of previously diagnosed disease were excluded.

UKOSS collects information on specific rare events that occur during pregnancy, agreed on by a national steering committee, from all obstetric units in the UK. The reporters submit monthly returns for the current list of conditions, including zero values, notifying UKOSS when an event has occurred (in this case, the diagnosis of breast cancer in pregnancy). UKOSS staff sent the data collection form ( https://www.npeu.ox.ac.uk/assets/downloads/ukoss/forms/UKOSS-Breast-Cancer-V1.pdf) for completion when notified of a case. In addition, we notified oncologists around the UK of the study to gain their assistance with the completion of information on oncological diagnosis and management. The UKOSS methodology ensures that no patient identifiable data are included for analysis while excluding duplicate submissions ¹³ . Cases were allocated UKOSS identification numbers, and all data were anonymised. Once received, duplicate cases were excluded by comparison of key clinical information. Reminders were sent if forms were not returned or were incomplete.

Ethics committee approval was obtained (Wales Research Ethics Committee 5, January 15, 2015, EC 14/WA/1267, IRAS project ID 165517). Consent was not required for the collection of anonymous routine data. Betsi Cadwaladwr University Health Board sponsored and funded this study.

The information requested included maternal demographics, details of the current and past pregnancies, BRCA (BReast CAncer) gene mutation status, and details of the diagnosis and staging when known. Information about staging, surgical management, chemotherapy use, and pregnancy outcomes was also collected. Babies were categorised as small for gestational age (SGA) if the birth centile was less than the 10 ^th centile ¹⁴ .

Categorical data are summarised as percentages. Continuous data are presented as medians, IQR and ranges. Group comparisons were performed using Fisher’s exact test.

We used the STROBE cross sectional reporting guidelines ¹⁵ .

The estimated total maternities in the UK during the study period were 1,544,700 ^{16, 17} ( https://www.opendata.nhs.scot/dataset/births-in-scottish-hospitals), giving an estimated incidence of 5.4 per 100,000 maternities (95% CI [4.37, 6.70]).

The characteristics of the women are shown in Table 1. The median maternal age at diagnosis was 36 years (IQR: 33–38 years; range: 26–44 years). The median BMI at booking was 25 (IQR: 22–28.5; range: 18–48). Nine women (11%) had in vitro fertilisation (IVF) pregnancies.

Data on presentation and diagnosis are presented in Table 2. Two women were asymptomatic; their tumours were identified at specific screening arising from known familial increased risk. The majority of symptomatic women (n=50, 61%) presented solely with a lump. Fourteen women (17%) presented with a lump associated with other breast symptoms, including pain, skin changes, discharge, or tenderness. The remaining 18 women (22%) presented with different combinations of symptoms but not a lump, including nipple inversion, pain and fullness, skin changes, breast enlargement, discharge from the nipple, and erythema. The median duration of symptoms before diagnosis was three weeks (IQR, 2–9 weeks; range, 0–56 weeks) and the median gestational age at diagnosis was 25 weeks (IQR: 16–33 weeks; range: 2–40 weeks) ( Table 2).

Staging investigations performed during pregnancy were reported fully for 76 women. Staging involved multiple combinations of modalities and included breast ultrasound (n = 73), mammogram (n = 62), chest X-ray (n = 30), liver ultrasound (n = 28), chest and/or abdominal computed tomography (n = 24), breast MRI (n = 21), echocardiogram (n = 15), bone scan (including radionuclear) (n = 13), non-contrast abdominal, skeletal or brain MRI (n = 10), and PET (n = 3). Twenty-four women received at least breast ultrasound, chest X-ray, and mammogram.

Data are available on the histology type in 71 women. The histological type was invasive ductal carcinoma in 62 (74%) patients, poorly differentiated in 2 (2%), invasive lobular carcinoma in 3 (4%), and ductal carcinoma in situ (DCIS) in 2 (2%). Metaplastic, inflammatory, and invasive ductal and lobular carcinomas were reported in the remaining cases.

Of the 61 cases in which the grade was recorded, grade 3 tumours were found in 45 (74%), grade 2 in 15 (25%), and grade 1 in 1 case (1%).

The TNM (tumour size, lymph nodes, metastasis) stage was known in 58 women ( Table 3). The clinical tumour size at diagnosis was greater than 2 cm in 39 patients (68%). Just over one in six women (N=9) had a tumour greater than 5 cm. Using the American Joint Committee on Cancer (AJCC) pathological staging system, 14 (24%) were stage 1, 27 (46%) were stage 2, 13 (22%) were stage 3, and 4 (7%) were stage 4. Of the 58 women with complete TNM staging, 24 (41%) were node-negative, 34 (59%) were node-positive, and 4 (7%) had distant metastasis to the liver and bones.

Tumour receptor status in relation to other tumour characteristics and staging is shown in Table 3. Overall, 41 (59%) patients were ER (+), 24 (35%) were Her2 (+) and 20 (29%) were triple-negative (ER, PR, HER2). Tumour characteristics and staging in relation to hormonal receptors are shown in Table 3. Of the nine women who conceived by IVF, four had ER (+) tumours, four had ER (-) tumours, and ER status information was missing for one.

The frequencies of surgery and chemotherapy by trimester of diagnosis are presented in Table 4. Data on the use of systemic chemotherapy were available for 79 women. This was administered during pregnancy in 37 cases (47%), was not recommended in 7 cases (9%), and was delayed until the end of pregnancy in 35 cases (44%). The timing was unknown in 5 cases. Of the 37 women who received chemotherapy during pregnancy, 18 (49%) received neoadjuvant chemotherapy and 17 (46%) received adjuvant chemotherapy. One woman received palliative chemotherapy for metastatic disease.

The median gestational age at the start of chemotherapy was 23 weeks (IQR, 19–27.5; range, 13–35 weeks).

Of the 18 women diagnosed during the first trimester, 11 received chemotherapy during pregnancy (all after 14 weeks), two received chemotherapy after miscarriage and in five cases chemotherapy was not indicated. Of these 18 women, 13 underwent surgery during pregnancy, all between 6 and 15 weeks.

Of the 23 women diagnosed during the second trimester, 20 had chemotherapy during pregnancy and 3 had the start of chemotherapy delayed until after delivery. In all three cases, delivery occurred after 37 completed weeks. In the same group of women, 11 had surgery during pregnancy and 8 had it delayed until the end of pregnancy (in two of these cases, the delivery was between 34 and 36 weeks); all other cases delivered after 37 weeks.

Breast cancer was diagnosed in 37 women during their third trimester. Thirty-five of them had induction of labour or pre-labour caesarean section, in 50% of the cases before 37 completed weeks (between 32 and 36 weeks). Of these women diagnosed during the third trimester, 6 underwent surgery during pregnancy and 25 had delayed surgery until the end of pregnancy. Five patients received chemotherapy during pregnancy, and 30 had it delayed until the end of the pregnancy.

Overall, thirty women out of the 77 with data (39%) underwent surgery during pregnancy: 14 underwent breast conservation surgery, and 16 underwent mastectomy. One woman underwent both procedures during pregnancy. The median gestational age at breast conservation surgery was 17 weeks (range: 6–34 weeks). The median gestational age at mastectomy was 18 weeks (range: 9–35 weeks). Surgery was delayed until the end of pregnancy in 34 (44 %) women. Of the 34 women, 11 received neoadjuvant chemotherapy during pregnancy. The reasons for the delayed surgery were not collected.

Including the 32 cases known that had axillary clearance during pregnancy, the median number of lymph nodes removed was 12 (IQR: 4.8–17.3 range 2 to 34). In these cases, the median number of positive nodes was 1 (IQR: 0–3.3 range: 0–16).

Chemotherapy regimens used during pregnancy and outcomes are given in Table 5. The systemic regimen that most patients received was anthracycline-based chemotherapy (31 of 37). This was associated with cyclophosphamide use in 26 of the 37 women who received chemotherapy during pregnancy. Taxanes were used in 15 women. The most widely used combination was FEC (fluorouracil, epirubicin, and cyclophosphamide) in ten women. Five women received trastuzumab during pregnancy, including in combination with pertuzumab. These women generally had advanced disease, evidenced by one or more of large primary tumour, multiple affected lymph nodes, or metastatic disease. No congenital abnormalities, NICU admissions or stillbirths were reported in the infants of any of the four who received trastuzumab during pregnancy and did not subsequently terminate the pregnancy. No other HER2-targeted therapy was received during pregnancy. A further woman who received doxorubicin and cyclophosphamide during pregnancy gave birth at 28 weeks to start Herceptin (trastuzumab) postnatally.

There were 81 babies born to the 80 women who continued with pregnancy, with a median gestational age at delivery of 37 completed weeks (IQR: 35–38, range 28–41). In total, 41% of the babies were delivered preterm and 59% after at ≥37 completed weeks. Outcomes in neonates born preterm and at ≥37 completed weeks are summarised in Table 6. All 16 women who gave birth before 36 completed weeks received steroids for fetal lung maturation; two babies died, both prior to 30 weeks. Sixteen babies (20%) were admitted to the neonatal care unit; 13 of these (81%) were admitted because of symptoms directly related to their prematurity. At birth, 11 of 78 (14%) neonates with known birth weight were classified as small for gestational age (SGA; weight below the 10 ^th centile) ¹⁴ . Eight of the babies who were SGA were born to the 35 mothers who had received chemotherapy, and three SGA babies were born to the 39 mothers who did not receive chemotherapy during pregnancy. This difference was not statistically significant (22.8% vs. 7.7%, P = 0.142; Fisher’s exact test).

Thirty women were induced, 28 (93%) for reasons related to their cancer. The mode of birth was known for 78 of the 80 women who continued pregnancy beyond the second trimester; 37 (47.4%) had pre-labour caesarean section, 21 (57%) after 37 weeks and 16 (43%) preterm. Of the 41 women who were planning a vaginal birth, 27 (66%) had a spontaneous vaginal birth, four (10%) had a ventouse birth, three (7%) had a forceps birth, one woman had a breech birth and six women (15%) had an emergency caesarean section after the onset of labour.

Three women were diagnosed later in pregnancy with metastatic cancer in the liver, thoracic spine, and lung, respectively. Two maternal deaths were reported. Two additional women had major maternal morbidity during pregnancy or puerperium; the morbidity in one of these cases was thought to be directly related to chemotherapy.

Thirty-seven percent of women (27 out of 73 known) breastfed, and 19 had lactation suppression.

The incidence of breast cancer diagnosed during pregnancy in the UK (5.4 per 100,000 maternities) is similar to the figures of 2.4 to 7.8 cases per 100,000 births estimated in the USA, Sweden, and Western Australia ^{4, 10– 12} .

The median maternal age was 36 years, which is the same as that found in the Prospective Study of Outcomes in Sporadic versus Hereditary Breast Cancer (POSH Breast Cancer Study) in the UK between 2000 and 2008, which included 2956 young women aged 18–40 years with breast cancer ¹⁸ , with a similar median ¹⁹ and mean ^{20– 23} maternal age found in other studies during pregnancy: 35–37. In contrast, during the study period, the maternal age was less than 35 years in 78.5% of births in the UK ²⁴ . This may partially explain the disproportionate number of women with IVF pregnancies who tend to be older. However, the percentage of IVF pregnancies in the UK in 2015 and 2016 was 2.57% and 2.6% ²⁵ , respectively, compared to 11% in this study. This observation requires further investigation. While there is uncertainty, there is a need to discuss this potential additional risk, and special attention should be paid to breast symptoms and examinations during the antenatal period in women who conceive by IVF until the position is clarified.

Throughout the period of the study, pregnant women in the UK were overweight or obese in more than 47% of cases ²⁶ . We found a median BMI of 25 in women presenting with breast cancer during pregnancy, with 42% of cases above the normal BMI. This finding is similar to the median found in the POSH study ¹⁸ . This tendency may reflect previously reported findings of lower BMI among premenopausal women with breast cancer ²⁷ .

The high percentage of symptomatic women before diagnosis (97%) is similar to that found in the POSH study (98%) ¹⁸ and in the CARING study (93%), a retrospective study of cancer diagnosed in pregnancy at 14 UK sites between 2016 and 2020 ¹⁹ . This high percentage is as might be expected outside a screening program population’s demographic. In postmenopausal women, two-thirds are asymptomatic at the time of diagnosis, with diagnosis at an earlier stage of screening. This may contribute to the worse prognosis in young women with breast cancer ²⁸ . Three-quarters of the cases presented with a breast lump, two-thirds on its own and the remainder mostly with a lump and/or other symptoms. These other presentations, such as nipple inversion, pain and fullness, skin changes, breast enlargement, nipple discharge, and erythema, need to be highlighted to women and midwives because symptoms are easily dismissed as part of the normal pregnancy breast changes. This emphasises the importance of increasing awareness of this diagnosis when discussing breast symptoms at any stage of pregnancy.

Pregnancy often masks symptoms and signs of breast cancer, leading to delayed diagnosis. The interval between the start of symptoms and diagnosis reported here (median 3 weeks; IQR: 2–9) is shorter than the median time of 4 weeks (range 1–104 weeks) found in the Australian study of Ives ¹⁰ , which included postpartum diagnosis, and the mean time of 3.9 months during pregnancy found by Langer et al. ²⁹ .

The median gestational age at diagnosis, 25 weeks, had a wide range, similar to the findings of other, smaller studies ²² . A large number of women (44%) were diagnosed in the third trimester. Of these 37 women, 35 (95%) underwent induction of labour or pre-labour caesarean section. In most cases, treatment (surgery or chemotherapy) was delayed until the end of pregnancy with iatrogenic delivery between 32 and 36 weeks in nearly half (49%). This finding could reflect guidelines regarding safe delivery time for women receiving chemotherapy to allow both maternal and fetal bone marrow recovery ^{30, 31} . The effects of granulocyte-colony stimulating factor on treating neutropenia in pregnancy may modify this decision, avoiding iatrogenic prematurity, and seems to be safe ^{32, 33} .

The UK confidential inquiry into maternal mortality, Mothers and Babies: Reducing Risk through Audit and Confidential Enquiries (MBRRACE-UK), has recently reported on a subset of the cases identified through the UKOSS reporting system as part of this UK breast cancer in pregnancy (UKBCiP) study, with direct access to anonymised patient records for assessment. This suggests new guidance: in general, early delivery to avoid delays in chemotherapy should not be recommended ³⁴ . For women diagnosed with breast cancer in the third trimester, the risk–benefit is likely to favour both mother and baby if a woman can receive at least two cycles of chemotherapy prior to a term (39–40 weeks) birth ³⁰ .

Overall, we found that women were more likely to be induced (57% vs. 39%) or undergo elective caesarean section (47% vs. 15%) than the general population in England ³² . This is similar to the 42% caesarean section rate in women with breast cancer observed in the CARING study ¹⁹ .

The Prospective Study of Outcomes in Sporadic versus Hereditary Breast Cancer (POSH Breast Cancer Study) ¹⁸ , although not directly comparable, provides context outside pregnancy because the demographics overlap with this UKBCiP study. In the POSH study, breast cancer was found to present at a more advanced stage in pregnancy, with higher node involvement and tumour grade. The tumour size at diagnosis was more commonly greater (> 2 cm in 68% vs. 52% of cases and > 5 cm in 15% vs. 7% in pregnancy and outside pregnancy, respectively). Node involvement was greater (59% vs. 50%), as were metastases (7% vs. 2.5%). The histological grade may be higher in pregnancy (grade 1, 1% vs. 6%; grade 2, 18% vs. 32.9%; and grade 3, 53% vs. 59%). The finding of invasive ductal carcinoma in 84% of cases is similar to that reported by other authors ^{10, 21– 23, 29, 35} . Despite 2/3 (68%) being diagnosed with stage II or III disease, most women did well in the short-term. In our study, four women (5%) presented with stage IV disease. Notably, staging was not conducted or was incomplete for around a third of women. At the time of the surveillance, UK guidelines on breast cancer in pregnancy recommended staging for metastases only in cases with high clinical suspicion ³⁶ ; however, the 2019 MBRRACE report subsequently found multiple examples of cases in which a lack of staging information impaired decision-making around management, including resulting in unnecessarily invasive surgery, and concluded that guidance is needed to ensure timely staging ³⁷ .

The finding that 59% of women with newly diagnosed breast cancer in pregnancy were oestrogen receptor (ER) positive compares to 66% in the POSH study, with triple-negative tumours present in 29% and 20% of women, respectively.

Although in England, overall for the period 2016–17 7.9% of babies were preterm ³⁸ and in Scotland 6.5% ( https://www.opendata.nhs.scot/dataset/births-in-scottish-hospitals), the preterm birth rate of 41% was lower than that observed in the CARING study (47%) ¹⁹ and that reported by Amant et al. in an international collaborative study of 311 cases of breast cancer diagnosed during pregnancy (49.6%) ⁶ . A smaller case series reported higher rates of preterm delivery; Gomez-Hidalgo in 11 cases found that 54.6% of the neonates were preterm ²⁰ and Framarino-dei-Malatesta in 54.5% of 22 women ²² . Outcome data suggest that the fetus does relatively well, even when exposed to several maternal chemotherapy regimens. Amant et al. (2015) concluded that prenatal exposure to maternal cancer, with or without treatment, did not impair the cognitive, cardiac, or general development of children in early childhood. Prematurity was correlated with worse cognitive outcomes; however, this effect was independent of cancer treatment ³⁹ .

The safety of current chemotherapy regimens during pregnancy is well documented, and breast cancer treatment during pregnancy should adhere to that of non-pregnant women. Most patients (31 of 37) received anthracycline-based chemotherapy. This was associated with cyclophosphamide use in 26 of the 37 women who received chemotherapy during pregnancy. Taxanes were used in 15 women. The most widely used combination was FEC in ten women.

It seems that later delivery (for the fetus), providing appropriate chemotherapy in pregnancy, and aiming for vaginal delivery are all reasonably safe. In a situation that is psychologically distressing for a woman and her family, the creation of ‘normality’ in relation to the birthing experience can be hugely important to their well-being and future perception.

This is the first national prospective study of breast cancer diagnosed during pregnancy in the UK and its incidence, management, and short-term outcomes. The study raised awareness of the rare condition of breast cancer in pregnancy among UK obstetricians and oncologists during the 2 years of the UKOSS data collection.

UKOSS data rely on the reporting of monthly cases, and a certain amount of under-reporting may occur; hence, the use of a 95% confidence interval has been included with our estimate of incidence.

The incidence estimate reported here must be considered a minimum estimate for several reasons. UKOSS is a system that involves all consultant-led maternity units in the UK, but some women diagnosed with breast cancer during pregnancy could have chosen to undergo termination of pregnancy or had a miscarriage before reaching these maternity units; thus, this will not be a true reflection of what may be happening in early pregnancy.

Additionally, although we wrote to clinical oncology units and contacted patient support groups such as Mummy’s Star to raise awareness of the study, the anonymised nature of UKOSS reporting precludes the study team from having women’s details, for example, through self-reporting, to cross-check with reporting units specifically to ensure their details had been included in the study. Currently, pregnancy data are not included in most cancer registries; therefore, these data cannot be used to enhance case ascertainment.