Breast cancer treatment patterns by age and time since last pregnancy in the Carolina Breast Cancer Study Phase III

To describe breast cancer treatment patterns among premenopausal women by age and time since last pregnancy. Data were analyzed from 1179 women diagnosed with premenopausal breast cancer in the Carolina Breast Cancer Study. Of these, 160 had a recent pregnancy (within 5 years of cancer diagnosis). Relative frequency differences (RFDs) and 95% confidence intervals (CIs) were used to compare cancer stage, treatment modality received, treatment initiation delay (> 30 days), and prolonged treatment duration (> 2 to > 8 months depending on the treatment received) by age and recency of pregnancy. Recently postpartum women were significantly more likely to have stage III disease [RFD (95% CI) 12.2% (3.6%, 20.8%)] and to receive more aggressive treatment compared to nulliparous women. After adjustment for age, race and standard clinical tumor characteristics, recently postpartum women were significantly less likely to have delayed treatment initiation [RFD (95% CI) − 11.2% (− 21.4%, − 1.0%)] and prolonged treatment duration [RFD (95% CI) − 17.5% (− 28.0%, − 7.1%)] and were more likely to have mastectomy [RFD (95% CI) 14.9% (4.8%, 25.0%)] compared to nulliparous. Similarly, younger women (< 40 years of age) were significantly less likely to experience prolonged treatment duration [RFD (95% CI) − 5.6% (− 11.1%, − 0.0%)] and more likely to undergo mastectomy [RFD (95% CI) 10.6% (5.2%, 16.0%)] compared to the study population as a whole. These results suggest that recently postpartum and younger women often received prompt and aggressive breast cancer treatment. Higher mortality and recurrence among recently pregnant women are unlikely to be related to undertreatment.


3
Background Younger (< 40 years of age) and recently postpartum (within 5 years of cancer diagnosis) women have been found to have worse breast cancer outcomes and higher mortality compared to other women with breast cancer . Previous studies have hypothesized that tumor biology, delayed diagnosis, or treatment delay and variation contribute to poorer disease outcomes for recently postpartum women [1][2][3][4][5][6][7][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40][41]. Our recent findings, in the Carolina Breast Cancer Study (CBCS), suggest that breast tumors of recently postpartum women were more frequently node positive and had unique immune microenvironments, but it is unknown how common treatment delay is among these women [42]. Previous analyses from the CBCS indicated that younger (< 50 years of age) women had fewer treatment delays compared to older (50-74 years of age) women [43,44], but comparisons in that study were not restricted to premenopausal women or to those with recent pregnancy. It is important to understand differences in treatment patterns for women with higher risk of aggressive cancers because delays and undertreatment are linked to worse overall and breast cancer-specific survival [45].
Using data from participants diagnosed with premenopausal breast cancers in CBCS Phase III, we hypothesized that stage at diagnosis, treatment initiation delay, prolonged treatment duration, and treatment modality would vary according to time since last childbirth and age at diagnosis. Recently postpartum women were defined as those diagnosed up to 5 years after their last full-term (≥ 7 months) pregnancy. Young-onset breast cancers were defined as cancers diagnosed at < 40 years of age.

Study population
The CBCS Phase III is a population-based study of women diagnosed with breast cancer in 44 counties of central and eastern North Carolina (2008)(2009)(2010)(2011)(2012)(2013), N = 2998); study details have been described previously [46][47][48][49]. Written informed consent was obtained at baseline (approximately 5 months after diagnosis) prior to data collection. All study protocols were approved by the Office of Human Research Ethics, Institutional Review Board at the University of North Carolina at Chapel Hill (UNC). Briefly, the primary study enrolled 20-74-year-old women with first primary invasive breast cancer and oversampled for Black and younger women (< 50 years of age) using randomized recruitment.
The current analysis examined treatment time-related factors and treatment modality for premenopausal women under 50 years of age (N = 1179). Only participants with stage I-III breast cancers were included, as treatment pathways for metastatic disease are distinct from those for localized disease. Additionally, we excluded participants who did not elect surgical treatment (N = 5). Cases with missing data for last full-term birth were excluded (N = 2). Breast cancers diagnosed during pregnancy were excluded (N = 7). Figure 1 depicts participant numbers according to inclusion/exclusion criteria. In order to examine the effect of age on treatment patterns, we also considered the CBCS Phase III population as a whole (including women 20-74 years of age at diagnosis).

Recency of last childbirth
In-person interviews were conducted by trained nurses to collect medical history including detailed information on pregnancy history. Date of breast cancer diagnosis was collected by medical record abstraction. Time since last fullterm birth was calculated by subtracting date of last full-term (≥ 7 months pregnancy) birth from date of breast cancer diagnosis. Women were grouped according to their time since last full-term birth: 0-5 years postpartum (N = 160), 5.1-10 years postpartum (N = 207), 10.1-20 years postpartum (N = 438), and 20.1-30 years postpartum (N = 164). Women who never had a full-term (≥ 7 months) pregnancy prior to their diagnosis were assigned to the "nulliparous" group (N = 210). Women who were up to 5 years postpartum were referred to as recently postpartum. Women who were 10.1-20 years postpartum were referred to as remotely postpartum.

Treatment initiation, treatment modalities, and prolonged treatment duration
Time to treatment initiation (in days) was defined as the time between breast cancer diagnosis and first treatment (defined as surgery, adjuvant or neoadjuvant chemotherapy, or radiation); this information was abstracted from medical records. Treatment initiation was categorized as occurring ≤ 30 vs. > 30 days after diagnosis, based on a recent publication by Bleicher et al. that reported better overall-survival among invasive non-metastatic breast cancer patients who received treatment within 30 days of diagnosis compared to longer wait to treatment initiation [50]. Information on treatment type, including type of surgery (mastectomy vs. breast-conserving surgery), chemotherapy receipt (yes vs. no), radiation therapy receipt (yes vs. no), and hormone therapy (yes vs. no), was abstracted from medical records. For prolonged treatment duration, participants were sorted in four treatment groups: surgery only, surgery and radiation, surgery and chemotherapy, and surgery, chemotherapy, and radiation. Treatment duration was categorized within each treatment group by subtracting the date of last treatment from date of first treatment. Prolonged treatment duration (yes vs. no) was defined using American Cancer Society [51,52] treatment recommendations within strata of treatment modality as follows: (1) surgery only, prolonged treatment duration was "yes" if surgery was performed ≥ 30 days after diagnosis, in-line with treatment initiation delay; (2) surgery and radiation, prolonged treatment duration was "yes" if treatment duration > 2 months [52]; (3) surgery and chemotherapy, prolonged treatment duration was " yes" if treatment duration > 6 months [51]; (4) surgery, radiation, and chemotherapy, prolonged treatment duration was "yes" if treatment duration > 8 months [51,52]. Information on breast cancer stage at diagnosis was abstracted from medical records.

Covariate assessment
Race was determined by self-report and categorized as Black or non-Black. Less than 2% of non-Black participants selfidentified as multiracial, Hispanic, or other race/ethnicities. Age at diagnosis was obtained from the baseline survey and used as a continuous variable in models. Information on parity was obtained from baseline survey and categories as nulliparous, or 1, 2, and ≥ 3 full-term (≥ 7 months pregnancy) births. Self-reported income (USD < $20 K, $20 K-$50 K, and > $50 K), education (≤ high school education/GED, some college education/college degree, and post-graduate/ professional degree), marital status (married vs. not married), and health insurance status (yes vs. no) were obtained from the baseline survey.

Statistical analyses
Descriptive statistics for patient sociodemographic and tumor characteristics of recently postpartum (0-5 years postpartum) and 5.1-10 years postpartum women were compared with nulliparous or remotely postpartum (10.1-20 years postpartum) women using chi-square tests or Fisher's exact tests when cell count < 5. Generalized linear models were used to estimate relative frequency differences (RFDs) and 95% CIs as a measure of association for recency of last childbirth and age at diagnosis with respect to treatment initiation delay, prolonged treatment duration, and treatment modalities (type of surgery, chemotherapy receipt, radiation therapy receipt, and hormone therapy receipt) [53]. Models were adjusted for age, race, tumor stage, size, grade, lymph node status, hormone receptor status, and human epidermal growth factor receptor 2 status. Due to the limited cell size for recently (0-5 years) postpartum women without chemotherapy (n = 20) and hormone therapy (n = 14) treatment groups, these models were only adjusted for age and race. We minimally adjusted stage models for age, race, income, education, marital status, and health insurance status at baseline. All analyses were conducted in SAS version 9.4 (SAS Institute, Cary, NC). P-values were two-sided with an alpha of 0.05 for statistical significance.  Table 1 shows the distribution of age at diagnosis, race, parity, income, education, marital status, and health insurance at baseline according to time intervals from most recent childbirth. Recently postpartum women had a significantly younger age at diagnosis (median = 37 years) compared to those who were remotely postpartum (median = 44 years) or were nulliparous (median = 42 years). Although no significant difference in race was observed between recently postpartum and nulliparous group, the 20.1-30 years postpartum group had a higher frequency of Black compared to non-Black participants, consistent with a trend toward younger age at first birth among Black women in this study population.

Patient and Tumor characteristics
Recently postpartum women were significantly more likely to have health insurance at baseline (94.4% vs. 88.6%) and were more commonly married (73.1% vs. 40.0%) compared to nulliparous women. Compared to remotely postpartum women, recently postpartum women had higher income and education, and were more likely to be primiparous. There were no differences in income and education between recently postpartum and nulliparous women.  Fig. 2). Similarly, younger women (< 40 years) were significantly less likely to experience prolonged treatment duration compared to older women (≥ 40 years), but no significant difference was observed with respect to tumor stage and delayed treatment initiation (Table 3; Fig. 3).

Treatment modalities
Recently postpartum women were more likely to receive more aggressive treatments compared to nulliparous women (Table 4). Considering type of surgery, the relative frequency difference for recency of birth was attenuated after adjusting for age, race, and standard clinical tumor characteristics, but women who were up to 10 years postpartum remained significantly more likely to get mastectomy (vs. breast-conserving surgery) compared to nulliparous women. Considering chemotherapy, associations with recency of birth were strongly related to tumor stage; we were unable to adjust for clinical tumor characteristics due to positivity violations, with all stage III cases in the recently pregnant group receiving chemotherapy. Thus, the association between postpartum status and chemotherapy is not independent of tumor characteristics. With respect to radiation therapy and hormone therapy (among ER positive and borderline cases only), no significant differences were observed between the recently postpartum and nulliparous group. Similarly, treatment patterns among younger vs. older women were strongly related to tumor clinical characteristics (especially, stage at diagnosis and lymph node status), with younger women being significantly more likely to receive mastectomy and chemotherapy and less likely to receive radiation and hormone therapy compared to older women (Table 3; Fig. 3).
Supplemental Tables 1 and 2 show the distribution of tumor characteristics by postpartum status and age, respectively. Sensitivity analyses adjusting for Area Health Education Center (AHEC) region or including cases diagnosed during pregnancy (n = 7) in the recently postpartum group (n = 160) did not substantially alter these results.

Discussion
In the Carolina Breast Cancer Study Phase III, conducted between 2008 and 2013, recently postpartum women had prompt and aggressive post-diagnostic treatment, at least in part due to more aggressive clinical tumor characteristics (i.e., later stage and lymph node positivity). Recently pregnant women tended to have later stage at diagnosis, prompt treatment initiation, shorter treatment duration, and were more likely to receive mastectomy (vs. breast-conserving surgery) and chemotherapy. These trends were mirrored by patterns in young-onset (< 40 years of age) breast cancer cases. These findings suggest that patterns of poorer outcomes for recently pregnant and younger women are not driven by undertreatment or treatment delays. Only one previous study has examined treatment initiation delay with respect to recency of pregnancy; however, that study compared pregnant vs. postpartum breast cancer cases and reported delayed initiation among pregnant women [3]. No previous studies have evaluated treatment timelines among postpartum women or comparing postpartum women to nulliparous women. Similarly, few studies have investigated the treatment modality among recently pregnant women [1][2][3][4]. The majority of these studies found no significant association between treatment modality such as receipt of chemotherapy [1][2][3], radiation [1,4] and surgery [1][2][3][4], and recency of pregnancy. However, previous studies included pregnant women [54,55], for whom treatment plans must address risk to both mother and child, and therefore are difficult to interpret relative to our analysis of premenopausal breast cancer among postpartum and nulliparous women. The current findings separating the postpartum period add resolution to the unique experience of this group.
Recent pregnancy and younger age both were associated with treatment timeliness and modality, but the higher incidence of more aggressive tumors appeared to drive the shift in treatment patterns for both groups. We were not able to examine these relationships separately by race due to limited number of recently postpartum cases. We were also not able to evaluate the differences in specific chemotherapy regimens. Future research should address whether social support, childcare needs, or other treatment-related "workload" experiences by younger women influence their reported quality of life and stress levels. Our results show that worse   breast cancer outcomes among younger and recently postpartum women are unlikely to be related to post-diagnostic undertreatment.