Prognosis of pregnancy-associated breast cancer: a meta-analysis

Background Pregnancy-associated breast cancer (PABC) is defined as breast cancer that is diagnosed during pregnancy and/or the postpartum period. Definitions of the duration of the postpartum period have been controversial, and this variability may lead to diverse results regarding prognosis. Moreover, evidence on the dose-response association between the time from the last pregnancy to breast cancer diagnosis and overall mortality has not been synthesized. Methods We systematically searched PubMed, Embase, and the Cochrane Library for observational studies on the prognosis of PABC published up to June 1, 2019. We estimated summary-adjusted hazard ratios (HRs) and the corresponding 95% confidence intervals (CIs). Subgroup analyses based on diagnosis time, PABC definition, geographic region, year of publication and estimation procedure for HR were performed. Additionally, dose-response analysis was conducted by using the variance weighted least-squares regression (VWLS) trend estimation. Results A total of 54 articles (76 studies) were included in our study. PABC was associated with poor prognosis for overall survival (OS), disease-free survival (DFS) and cause-specific survival (CSS), and the pooled HRs with 95% CIs were 1.45 (1.30–1.63), 1.39 (1.25–1.54) and 1.40 (1.17–1.68), respectively. The corresponding reference category was non-PABC patients. According to subgroup analyses, the varied definition of PABC led to diverse results. The dose-response analysis indicated a nonlinear association between the time from the last delivery to breast cancer diagnosis and the HR of overall mortality (P < 0.001). Compared to nulliparous women, the mortality was almost 60% higher in women with PABC diagnosed at 12 months after the last delivery (HR = 1.59, 95% CI 1.30–1.82), and the mortality was not significantly different at 70 months after the last delivery (HR = 1.14, 95% CI 0.99–1.25). This finding suggests that the definition of PABC should be extended to include patients diagnosed up to approximately 6 years postpartum (70 months after the last delivery) to capture the increased risk. Conclusion This meta-analysis suggests that PABC is associated with poor prognosis, and the definition of PABC should be extended to include patients diagnosed up to approximately 6 years postpartum.


Background
Breast cancer is the second most common cancer worldwide and the most commonly occurring malignancy in women [1]. Due to the trend of delayed delivery, the number of women with breast cancer during a pregnancy or in the subsequent few years after a pregnancy is expected to increase [2]. Breast cancer occurring during pregnancy is a challenging clinical situation since the welfare of both the mother and the foetus must be considered in any treatment plan. Conventionally, pregnancy-associated breast cancer (PABC) is defined as breast cancer that is diagnosed during pregnancy or the postpartum period. Definitions of how many years after delivery breast cancer can be diagnosed under this definition have ranged from 0.5 to 5 years, and sometimes even longer [3,4]. PABC is viewed as a clinically and biologically special type of breast cancer and only comprises 0.2-0.4% of all breast cancers [5,6]. However, it is the most common cancer in pregnancy and is diagnosed in approximately 15 to 35 per 100,000 births, and the number of breast cancer cases diagnosed during pregnancy is less than after delivery [7][8][9][10].
Pregnancy itself may temporarily increase the risk of developing breast cancer, although it has a long-term protective effect on the development of breast cancer [11,12]. However, whether PABC has a worse prognosis is currently controversial. A meta-analysis published in 2016 showed that the risk of death increased in women with PABC compared with women with non-PABC (pooled hazard ratio (HR), 1.57; 95% confidence interval (CI), 1.35-1.82) [13]. However, other recent studies found no significant difference in the prognosis of PABC and non-PABC [14][15][16][17]. Meanwhile, the specific definition of PABC has varied and this variability may lead to diverse results on the relationship among pregnancy, postpartum and breast cancer. Therefore, it is necessary to specify the definition of PABC by summarizing epidemiological evidence. This study was initiated to understand the prognosis of PABC and examine the dose-response relationship to provide quantitative evidence for defining PABC.

Search strategy
This meta-analysis was performed in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines. We did our best to include studies published to date regarding the prognosis of PABC. Eligible studies were found by searching PubMed, Embase, and the Cochrane Library for relevant reports published before June 1, 2019. The keywords used for the search were ("pregnan*" OR "gestation*" OR "childbirth" OR "postpartum" OR "parity") AND "breast" AND ("cancer" OR "neoplasia" OR "carcinoma"). The references lists of all retrieved articles and previous systematic reviews were manually searched.

Inclusion and exclusion criteria
All eligible studies met the following criteria: (1) observational prognostic studies with a follow-up period longer than 6 months; (2) participants were diagnosed with breast cancer by clinical diagnosis and/or histologically; (3) the case group was diagnosed with PABC, and the control group was non-PABC or nulliparity; (4) the outcomes were in terms of overall survival (OS), diseasefree survival (DFS) or cause-specific survival (CSS); and (5) the risk point estimate was reported as an HR with 95% CI, or the data were presented such that an HR with 95% CI could be calculated. The exclusion criteria were as follows: (1) duplicated or irrelevant articles; (2) reviews, letters, and case reports; (3) non-human studies; and (4) studies with inappropriate data for metaanalysis, such as incomplete or inconsistent data.

Data extraction
Two reviewers extracted the data independently using a predefined data extraction form. Any disagreements were resolved by discussion. The extracted data included the first author, publication year, country, PABC definition, control definition, sample size, cancer type, stage or grade, age, matching criteria, adjusted variables, and adjusted HRs with 95% CIs.

Assessment of study quality
The methodological quality of the studies was assessed by the Newcastle-Ottawa scale (NOS) [18]. A score of 0-9 was allocated to each study, with higher scores indicating higher quality.

Meta-analysis and statistical analysis
We used adjusted HRs and 95% CIs, which are most appropriate for time-to-data events. If HRs were not reported, we estimated HRs from the raw data or Kaplan-Meier curves [19]. The I-square (I 2 ) test was performed to assess the impact of study heterogeneity on the results of the meta-analysis. If severe heterogeneity was present at I 2 > 50%, a random effects model was chosen; otherwise, a fixed effects model was used. Visual inspection of the funnel plot and Egger's and Begg's tests were performed to assess publication bias. Subgroup analyses were performed according to the diagnosis time, PABC definition, geographic region, year of publication and estimation procedure for HR.
Variance-weighted least squares regression (VWLS) model was used to evaluate the dose-response association between the time from the last pregnancy to breast cancer diagnosis and HR of overall mortality [20]. Restricted cubic splines were used to check the time from the last pregnancy as a continuous, nonlinear exposure, and the time was defined by the 5th, 35th, 65th and 95th percentiles of the distribution [21]. The time from the last pregnancy to breast cancer diagnosis reported in each study was converted to months. We used the average value of the lower and upper limits of each category. If the lowest category was open ended, the average value of the upper limit and 0 was used. If the highest category was open ended, the average value was defined as 1.5 times the lower limit. All statistical analyses were performed using STATA Version 13.0. P < 0.05 was considered significant.

Search results and study characteristics
We initially identified 12,414 articles and screened their titles and abstracts ( Fig. 1). After duplicated and irrelevant articles were excluded, 54 articles with 76 studies met the inclusion criteria and were thus included in our meta-analysis. The quality of the studies was assessed based on the NOS and ranged from 6 to 9 (mean of 7.2). The characteristics of the studies are summarized in Table 1.

Overall survival (OS)
Forty-five studies comprising 6602 PABC patients and a total of 157,657 individuals were identified for the metaanalysis of OS. There was an overall increased risk of death for PABC patients compared to controls, with a pooled hazard ratio of 1.45 (95% CI 1.30-1.63). There was significant heterogeneity (I 2 = 64.9, P<0.001). The subgroup analysis according to different follow-up durations (4 years, 5 years, 6 years, 7 years, 10 years and > 10 years) had similar results to the overall analysis (Fig. 2). However, the 6-year and 7-year OS, with few studies, showed nonsignificant results.

Disease-free survival (DFS)
Twenty studies comprising 1786 PABC patients and a total of 9762 individuals were identified for the metaanalysis of DFS. The overall HR was 1.39 (95% CI, 1.25-1.54). There was no significant heterogeneity (I 2 = 24.5, P = 0.146). The subgroup analysis according to different follow-up durations (5 years, 6 years, 10 years and > 10 years) had similar results as the overall analysis (Fig. 3). However, the 7-year DFS, with only 2 studies, showed nonsignificant results.

Cause-specific survival (CSS)
Only 6 studies provided information on CSS with 296 PABC patients and a total of 29,598 individuals. The overall HR was 1.40 (95% CI, 1.17-1.68). There was no significant heterogeneity (I 2 = 53.1, P = 0.074). The subgroup analysis (5-year CSS) had similar results as the overall analysis (Fig. 4).

Subgroup analyses
Several factors that may have induced differences in outcomes were investigated with subgroup analyses, including diagnosis time, PABC definition, geographic region,    year of publication and estimation procedure for HR. The results consistently showed worse prognoses in women with PABC than in those with non-PABC, except for the subgroup based on PABC definition and year of publication ( Dose-response association between the time from the last pregnancy to breast cancer diagnosis and HR of overall mortality As the meta-analysis included studies reporting the HRs with their 95% CIs of overall mortality relating to three or more categories of time since the last pregnancy, all the studies were eligible to be included in the doseresponse analysis. A total of ten studies were included in the dose-response meta-analysis, and nulliparous women were taken as the corresponding reference category ( Table 3). The analysis of departure from linearity indeed indicated a nonlinear association between the time from the last delivery to breast cancer diagnosis and the hazard ratio of PABC overall mortality (P < 0.001). The nonlinear spline showed a decreasing trend. Compared to nulliparous women, the mortality was almost 60% higher in women with PABC diagnosed at 12 months after the last delivery (HR = 1.59, 95% CI 1.30-1.82), and the mortality was not significantly different at 70 months after the last delivery (HR = 1.14, 95% CI 0.99-1.25) (Fig. 5). These results showed a higher risk of death than that in nulliparous patients, suggesting that the definition of PABC should be extended to include patients diagnosed up to approximately 6 years postpartum (70 months since the last delivery) to capture the increased risk.

Publication Bias
As shown in Fig. 6, each point represents an independent study of the indicated association, and a visual inspection of the funnel plot did not suggest evidence of publication bias among the articles (Egger's test, P = 0.451; Begg's test, P = 0.077).

Discussion
We reviewed and meta-analyzed the existing scientific literature on the prognosis of PABC to draw a powerful conclusion that PABC is associated with a poor prognosis. Our results are consistent with those of the previous meta-analysis conducted in 2016 [13]. However, the negative effect on OS and DFS appears to be less pronounced in our study overall than in the previous meta-analysis. This is the largest and latest meta-analysis in this field. It included a larger number of participants, thus reducing the small-study effect to a great degree. The studies included in our meta-analysis were of relatively high quality. The mean Newcastle-Ottawa score of the studies was 7.2.
There are two explanations that may account for the results. On the one hand, mammary gland involution following pregnancy has been suggested to explain the poor prognosis [71]. Breast degeneration is the process of tissue remodelling, until wound healing, inflammatory bowel disease and immune infiltration reach a state indistinguishable from the non-productive breast [72,73], which supposedly promotes tumour progression. On the other hand, pregnancy and breastfeeding lead to less timely detection and clinical examination. The delayed diagnosis allows more time for tumour growth, increasing the metastatic potential of the disease [52,74]. Pregnancy also makes the treatment strategy more conservative to ensure the safety of the foetus [10,75]. However, the exact reasons for the poor prognosis of PABC need to be explored in the future.
To the best of our knowledge, this is the first doseresponse meta-analysis providing comprehensive insights into the association between the time from the last pregnancy to breast cancer diagnosis and the overall mortality of PABC. The scientific value of dose-response metaanalyses is higher than meta-analyses with exposure classified as two categories [20,76]. Through the variance weighted least-squares regression with a random effects model, we found a nonlinear direct association between the time from the last pregnancy to breast cancer diagnosis and overall mortality. Compared with nulliparous women, the mortality was almost 60% higher in women with PABC diagnosed at 12 months after the last delivery, and the mortality had no significant difference at 70 months after the last delivery. We propose that the definition of PABC should include patients diagnosed up to at least 6 years postpartum to better delineate the increased risk imparted by a postpartum diagnosis. These findings also provide valuable insights into further research. Callihan's cohort demonstrated that breast cancer patients diagnosed within 5 years postpartum have a significantly higher risk of metastasis and mortality than nulliparous patients [58]. Compared to that cohort, our dose-response meta-analysis provides a higher quality of evidence to expand the definition of PABC. Understanding the differences between breast cancers diagnosed during different times postpartum would better permit the translation of informative data from basic science and epidemiologic studies into the clinical care and treatment of breast cancer in young women. The present meta-analysis has the following limitations that must be taken into account. First, if HRs and  95% CIs were not directly reported in the included studies, we estimated HRs from the crude data or Kaplan-Meier curves. This may cause bias without adjustment. However, we performed subgroup analysis based on the estimation procedure for HR. This analysis consistently showed a worse prognosis for women with PABC than for those with non-PABC. Second, the meta-analysis was based on data from observational studies; although most of the included studies adjusted for several relevant confounders (including age, year of diagnosis, tumour stage, axillary lymph node status, oestrogen receptor, hormonal receptor status, HER2 status, family history, etc.), residual confounding by other potential factors cannot be ruled out. Third, high between-study heterogeneity is another limitation of the current meta-analysis. This was likely due to significant differences in the sample sizes, definitions of PABC and/or treatment interventions. Last, the language of the studies was limited to English, which may result in potential language bias.

Conclusions
In summary, this meta-analysis suggests that PABC is associated with a poor prognosis for OS, DFS and CSS compared to non-PABC cases. The definition of PABC should be extended to include patients diagnosed up to approximately 6 years postpartum to capture the increased risk of death. Further long-term prospective cohort studies with larger sample sizes should be conducted to validate this article's findings.