Prognostic role of pregnancy occurring before or after treatment of early breast cancer patients aged <35 years

A GET(N)A Working Group analysis

Authors


  • Isabelle Chapelle-Marcillac provided editorial assistance in the preparation of the manuscript.

Abstract

BACKGROUND:

Usual practices recommend waiting at least 2 years between diagnosis of early breast cancer (EBC) and pregnancy. Few data highlighted a harmful effect of an early pregnancy for low-risk patients. The authors analyzed retrospectively data from women younger than 35 years who became pregnant before or after treatment of EBC.

METHODS:

Between 1990 and 1999, 908 consecutive EBC patients were analyzed. The primary endpoint was to compare overall survival (OS) between pregnant and nonpregnant patients. The secondary endpoint was to establish a score index laying down the risk of distant recurrence.

RESULTS:

Within the year before the diagnosis, 105 (11.6%) patients became pregnant and 118 (13%) were pregnant after treatment. In a multivariate model, a pregnancy before the diagnosis was not predictive of death but of local relapse. A pregnancy subsequent to breast cancer therapy resulted in a 77% decrease of death (P < .001). In good-prognosis score index patients, the annual risk of relapse remained low. In patients having the higher score, recurrences occurred mainly during the first years after the treatment. Beyond 80 months, the annual risk of relapse seemed to be similar to those of lower-risk subgroups.

CONCLUSIONS:

In women aged younger than 35 years, a pregnancy occurring before or after the diagnosis of breast cancer was not an independent prognostic factor of death. In the subset of patients having a high risk of relapse, it may be preferable to postpone a pregnancy beyond 5 years after the breast cancer therapy. Cancer 2009. © 2009 American Cancer Society.

The pregnancy appears to have a dual effect on the risk of breast cancer. There is a transient increased risk between 3 and 15 years right after childbirth, which decreases during following years.1, 2 One hypothesis could be that pregnancy increases the short-term risk of breast cancer by stimulating the growth of cells having undergone early stages of a malignant transformation. However, pregnancy confers a long-term protection by inducing the differentiation of normal mammary parental cells that have the potency for a subsequent neoplastic development.3 When controlling for other prognostic factors, the relationship between a worse breast cancer outcome and the proximity of a pregnancy remains unclear.4, 5 Because of the high level of estrogens and progesterone during pregnancy, it was proposed that the pregnancy after breast cancer may stimulate the occurrence of micrometastases, thus increasing the risk of recurrences, especially for women who had presented a hormone receptor positive disease.3, 6-9 On the other hand, a recent study did not support the current medical advice postponing pregnancy, at least 2 years after breast cancer.10 Authors have found an improved overall survival (OS) in women who waited at least 24 months to conceive, and a nonsignificant protective effect was seen for women who waited at least 6 months to become pregnant. In addition, women who became pregnant had a trend towards a more favorable prognosis, such as smaller tumor size and less axillary lymph nodes involved. In this setting, the interpretation of results is often limited by significant biases, including selection biases and the so-called “healthy mother effect.” Indeed, only women alive, healthy, and disease-free are likely to become pregnant after breast cancer.11 In women wishing for a pregnancy after a diagnosis of breast cancer, few data were available supporting the decision for both women and physicians. The lack of information about effects of a subsequent pregnancy on breast cancer outcome remains questionable for many women, especially because of an uncertain future.

A usual wisdom recommends waiting at least 2 years between diagnosis of the breast cancer and the pregnancy to be beyond the high-risk period of early relapse. Noteworthy, there are no data describing a harmful effect of an early pregnancy for low-risk patients. We retrospectively analyzed data of women aged younger than 35 years, who became pregnant either during the year before breast cancer or after the treatment of their breast cancer, taking into account the “healthy mother effect.” The purposes of the analysis were to compare OS between pregnant and nonpregnant women and to evaluate the annual risk of distant relapse using a risk score, subsequently dedicated for advising patients wishing to conceive.

MATERIALS AND METHODS

Study Population

We retrospectively analyzed data from consecutive patients aged younger than 35 years having a nonmetastatic, noninflammatory, invasive, unilateral breast cancer, treated in 8 French institutions between January 1990 and December 1999. In this unique cohort of patients, previous pregnancy corresponded to a pregnancy within 1 year before the breast cancer diagnosis, and subsequent pregnancy corresponded to a pregnancy occurring after the treatment period. Patients who are pregnant during the treatment period were excluded from the analysis because of the specific management required for this population.

The data collected from medical records were as follows: demographics; epidemiologic risk factors (including date and outcome of previous pregnancies); family history of breast cancer; tumor characteristics (tumor size, histological subtype, histoprognostic grade, axillary lymph node involvement, estrogen and progesterone receptors [PGR] status); treatment modalities (surgery, chemotherapy, radiotherapy, hormonotherapy with duration and dose); recurrence characteristics (locoregional recurrence, first distant recurrence); and death or last contact date. The protocol was declared to the appropriate French authorities supervising individual computerized data files (Commission Nationale Informatique et Liberté [CNIL]).

Statistical Analysis

The primary endpoint was to compare distant recurrence free interval (DRFI) and OS between pregnant and nonpregnant patients. To summarize results, qualitative data were presented as a percentage of sample size, and quantitative data were described using mean, median, standard deviation, and range. The chi-square test was used to compare proportions, and it was replaced by the Fisher exact test if the expected frequency in any cells of the contingency table is less than 5. Ordinal variables were compared using analysis of variance or Kruskal-Wallis test in the absence of variance equal terms.

The computerized survival data were defined as the time from diagnosis of breast cancer until the occurrence of event, ie, distant relapse or death. Distant relapse-free and alive patients were censored at the date of their last known contact. The DRFI and OS rates were calculated by the Kaplan-Meier method, and groups were compared using a log-rank test. A multivariate analysis was carried out to assess the relative influence of prognostic factors on DRFI and OS, using Cox proportional hazards models in a forward stepwise procedure. The clinical tumor size was preferred to the pathological tumor size because patients were likely to receive a neoadjuvant therapy. A pregnancy occurring after the treatment was considered as a time-dependent covariate, taking into account the time between the diagnosis of cancer and conceiving. The influence of pregnancy, adjusted on the other prognostic factors, was assessed by a Cox model.

The secondary endpoint was to establish a score index predicting the risk of distant relapse in early breast cancer patients aged younger than 35 years. A specific score index was developed to predict distant recurrences, which comprised the summation of independent prognostic factors selected from the multivariate model, thus respectively multiplied by their regression coefficients. According to these regression coefficients, a weight was assigned to each variable allowing the computation of a score (Table 1). Based on this score, 3 groups were defined, according to tertiles cutoff, as follows: low risk for a score ≤6, intermediate risk for a score between 7 and 10, and high risk for a score ≥10.

Table 1. Weight Assigned to Prognostic Factors
Prognostic FactorsWeight Assigned
  1. T, tumor; UICC, Union International Contre le Cancer; SBR, Scarff, Bloom, and Richardson; PGR, progesterone receptor.

T (UICC) 
 T0-T10
 T21
 T3-T44
PGR 
 Positive0
 Negative2
SBR grade 
 SBR I0
 SBR II3
 SBR III4
 SBR not gradable/missing−1
Axillary lymph nodes involvement 
 00
 No axillary dissection2
 1-35
 >38

All the analyses were performed using S-Plus 2000 software (MathSoft, Seattle, Wash). Differences were considered statistically significant when the p value was 0.05 or less.

RESULTS

Patient Characteristics

Between January 1990 and December 1999, the data from 908 nonmetastatic breast cancer patients aged younger than 35 years were collected and analyzed. The median follow-up was 87 months (range, 7 to 166 months).

The baseline characteristics of patients and tumors at the time of breast cancer diagnosis, according to the pregnancy within the previous year, are summarized in Table 2 and concerned 893 patients (15 because of missing values concerning pregnancy). The median age of the 908 patients was 32 years (range, 29 to 35 years). Among these patients, 105 (11.8%) were pregnant within the year before the diagnosis of breast cancer. These patients had significantly higher clinical tumor size and nodal involvement than nonpregnant patients, and they more frequently have a negative hormone receptor status (Table 2). No differences were observed in terms of age at diagnosis, family history of breast cancer, histoprognostic grade, or axillary lymph nodes involvement. As regards histological subtype, there was a trend towards a higher rate of invasive lobular carcinoma in previously pregnant patients, although not significant (P = .06).

Table 2. Patient and Tumor Characteristics at the Diagnosis of Breast Cancer (893 patients)
 Pregnancy During The Previous Year (15 Missing Values) 
No (n=788)Yes (n=105)
  1. UICC indicates Union International Contre le Cancer; SBR, Scarff, Bloom, and Richardson; ER, estrogen receptors; PGR, progesterone receptor; CT, chemotherapy; BCS, breast-conserving surgery; RT, radiotherapy.

CharacteristicsNo.%No.%P
 Age, y     
 ≤3024330.83937.1.19
 30 to 3333342.34643.8 
 >3321226.92019.1 
Family history of breast cancer   .97
 Yes22328.43028.6 
 No56371.67571.4 
 Missing40.500.0 
Tumor classification (UICC)    .0002
 T0-T134543.82624.8 
 T233742.85451.4 
 T3-T49712.32422.8 
 Missing91.111.0 
Node classification (UICC)    .009
 N050664.25451.4 
 N1-N227835.35148.6 
 Missing50.500.0 
 Histological type    .06
 Ductal68887.39186.7 
 Lobular293.798.6 
 Medullary131.721.9 
 Other536.732.8 
 Missing50.600.0 
SBR grade    .41
 I8510.887.6 
 II32240.83836.2 
 III29637.64643.8 
Not gradable719.01211.4 
Missing141.811.0 
Axillary lymph nodes involvement    .37
 No axillary dissection313.976.6 
 039149.64946.7 
 1-323329.62725.7 
 > 313316.92221.0 
ER    .03
 Positive38248.54441.9 
 Negative28235.85249.5 
 Missing12615.798.6 
 PGR    .03
 Positive40251.04744.8 
 Negative26133.14946.6 
 Missing12515.998.6 
Treatments    .04
 Neoadjuvant CT21427.24341.0 
 BCS followed by RT9712.376.7 
 BCS followed by RT and adjuvant CT27434.83533.3 
 Mastectomy first16821.32019.0 
 Missing354.40 

After the treatment, data on menstrual periods were not available for 102 (11.2%) patients. Among the 806 patients who have well-informed menses, 522 (64.8%) were not amenorrheic. Eighty-two (10.2%) patients developed a permanent amenorrhea, and 202 (25.1%) had a transient amenorrhea, defined as the absence of menstrual periods for at least 6 months, followed by the reappearance of regular menstruations.12 Among the 880 patients with known data, 118 (13.4%) patients were pregnant after adjuvant or neoadjuvant treatment of their breast cancer, of which 15 (1.7%) were pregnant within the year before the breast cancer diagnosis. There was no significant difference in the pattern of baseline tumor characteristics between patients who became pregnant after treatment and nonpregnant patients, except for the histological type showing that pregnant patients had significantly more tumors of medullary and other subtype tumors (Table 3). As expected, the patients who were pregnant after treatment were significantly younger (P < .0001).

Table 3. Patient and Tumor Characteristics of Patients Pregnant After Treatment Compared With Nonpregnant Patients (Total of 880 patients)
 Pregnancy After Treatment (28 Missing Values)
No (n=762)Yes (n=118)
  1. UICC indicates Union International Contre le Cancer; SBR, Scarff, Bloom, and Richardson; ER, estrogen receptor; PGR, progesterone receptor; CT, chemotherapy; BCS, breast-conserving surgery; RT, radiotherapy.

CharacteristicsNo.%No.%
 Age, y    
 ≤3021328.06252.5
 30 to 3333544.04033.9
 >3321428.01613.6
Family history of breast cancer    
 Yes22930.02521.2
 No52969.49378.8
 Missing40.500.0
Tumor classification (UICC)    
 T0-T130740.25546.6
 T234445.14840.7
 T3-T410213.41512.7
 Missing101.300.0
Node classification (UICC)    
 N046961.58471.2
 N1-N228937.93428.8
 Missing50.600.0
Histological type    
 Ductal68189.49177.1
 Lobular314.065.1
 Medullary101.354.2
 Other374.81512.7
 Missing40.510.9
SBR grade    
 I7710.11411.9
 II30640.14941.5
 III30339.83832.2
 Not gradable648.41613.6
 Missing121.610.8
Axillary lymph nodes involvement    
 No axillary dissection334.354.2
 036848.36454.2
 1-322028.93832.2
 >314118.5119.3
ER    
 Positive37148.75344.9
 Negative28737.74235.6
 Missing10413.62319.5
 PGR    
 Positive39151.35445.8
 Negative26634.94134.7
 Missing10513.82319.5
Treatments    
 Neoadjuvant CT22028.93428.8
 BCS followed by RT8711.41512.7
 BCS followed by RT and adjuvant CT26434.74336.5
 Mastectomy first16221.22117.8
 Missing293.854.2

Treatment Regimens

One hundred and sixty-six (18.3%) patients had not received systemic therapy (Table 4). Among patients who had received a systemic treatment, 259 (28.5%) were treated with neoadjuvant chemotherapy, 382 (42.1%) with adjuvant chemotherapy, 8 (0.9%) with adjuvant hormonotherapy, and 93 (10.2%) with adjuvant chemoendocrine therapy (Table 3). The small amount of adjuvant hormonotherapy was because the standard of care in the 1990 seconds. In the whole population, 21 (2.3%) patients were not operated (20 in the neoadjuvant chemotherapy subgroup, and 1 in the adjuvant chemoendocrine therapy subgroup), and 102 (11.2%) did not receive locoregional radiotherapy. Otherwise, the surgery comprised breast-conserving surgery (BCS) in 63.4 % of cases and mastectomy in 34.2%. In patients receiving chemotherapy, the median number of cycles delivered was 4 (range, 1 to 12) for neoadjuvant chemotherapy and 6 (range, 1 to 7) for adjuvant chemotherapy. The chemotherapy regimens are summarized in Table 4. Overall, there was no significant difference in treatment protocols used between previously pregnant and nonpregnant women, taking into account the initial status of the disease.

Table 4. Treatment Characteristics (908 patients)
CharacteristicsNo.%
  • BCS indicates breast-conserving surgery; RT, radiotherapy.

  • *

    Six patients received radiotherapy alone without surgery.

 No systemic treatment16618.3
 Surgery alone495.4
 Surgery and RT11712.9
 Neoadjuvant chemotherapy25928.5
 Plus locoregional therapy alone11112.2
 Surgery alone70.8
 Radiotherapy alone141.5
 Surgery and RT909.9
 Plus locoregional and systemic treatments148*16.3
 Adjuvant chemotherapy889.7
 Adjuvant hormonotherapy293.2
 Adjuvant chemoendocrine therapy313.4
Type of neoadjuvant chemotherapy, n=259  
 Anthracycline-based18972.9
 Taxane-based20.8
 Anthracycline plus taxane166.2
 Other5019.3
 Missing20.8
 Adjuvant chemotherapy38242.1
 After BCS26228.9
 After modified mastectomy12013.2
 Adjuvant hormonotherapy80.9
 After BCS70.8
 After modified mastectomy10.1
 Adjuvant chemoendocrine therapy9310.2
 After BCS647.0
 After modified mastectomy283.1
 After RT alone10.1
Type of adjuvant chemotherapy, n=594  
 Anthracycline-based39065.7
 Taxane-based122.0
 Anthracycline plus taxane122.0
 Other16828.3
 Missing122.0

Prognostic Role of Pregnancy on OS

At the cutoff date for analysis, 232 patients died, and the 5-year OS rate was 82% (95% confidence interval [CI], 80% to 85%).

Regarding pregnancy within the year before the breast cancer diagnosis, 33 pregnant women and 196 nonpregnant died, resulting in 5-year OS rates of 77% and 83%, respectively (P = .028) (Table 5, Fig. 1). Regarding pregnancy after breast cancer treatment, 6 pregnant women and 222 nonpregnant women died. The univariate analysis showed that significant prognostic factors of death were prior pregnancy, no subsequent pregnancy, clinical tumor size, axillary lymph nodes involvement, Scarff-Bloom-Richardson grade, estrogen status, and PGR status (Table 5).

Figure 1.

Ten-year overall survival rates are illustrated according to pregnancy in the year before breast cancer diagnosis.

Table 5. Overall Survival Analysis According to a Pregnancy Before or After Diagnosis of Breast Cancer: Results of the Univariate and Multivariate Models
Prognostic factorsNo. of Patients% 5-Year DR Rate [95%CI]Univariate AnalysisMultivariate Analysis
HR [95%CI]P (LR)HR [95% CI]P (Cox)
  1. DR indicates distant recurrence; CI, confidence interval; HR, hazard ratio; LR, log-rank; OS, overall survival; NS, not significant; UICC, Union International Contre le Cancer; SBR, Scarff, Bloom, and Richardson; ER, estrogen receptors; PGR, progesterone receptor.

Prior pregnancy   .028 NS
 No78883 [80-86]1   
 Yes10577 [69-86]1.51 [1.05-2.20]   
Subsequent pregnancy   <.001 (Cox) <.001
 No7621 1 
 Yes1180.25 [0.11-0.56] 0.23 [0.10-0.52] 
Tumor classification (UICC)   <.001 .012
 T0-T137787 [84-91]1 1 
 T240073 [68-78]1.60 [1.19-2.16] 1.21 [0.89-1.65] 
 T3-T412169 [60-78]2.66 [1.85-3.83] 1.92 [1.30-2.84] 
Axillary lymph nodes   <.001 <.001
 044990 [87-93]1 1 
 1-326481 [76-86]2.29 [1.64-3.19] 2.37 [1.68-3.35] 
 >315762 [55-70]4.48 [3.21-6.25] 4.24 [2.96-6.09] 
 No dissection3886 [75-98]2.16 [1.19-3.94] 1.53 [0.82-2.86] 
SBR grade   .002 NS
 I9493 [87-98]1   
 II36581 [77-86]2.56 [1.41-4.65]   
 III35079 [74-83]2.83 [1.55-5.16]   
ER   .028 NS
 Positive43386 [82-89]1   
 Negative34075 [70-80]1.35 [1.03-1.78]   
PGR   .003 NS
 Positive45686 [82-89]1   
 Negative31674 [69-80]1.51 [1.15-1.99]   

In the multivariate Cox regression model, clinical tumor size and axillary lymph nodes involvement remained significantly associated with death (Table 5). In this model, a pregnancy subsequent to breast cancer therapy resulted in a 77% decrease of death, with a relative risk (RR) of death of 0.23 in favor of pregnant women (95% CI, 0.10-0.52; P < .001).

Impact of Pregnancy on Recurrences

Overall, three hundred and eighteen patients developed a distant relapse, with a 5-year DRFI rate of 69% (95% CI, 66-72). Regarding pregnancy after breast cancer treatment, 16 pregnant women and 297 nonpregnant relapsed, with a RR of recurrence of 0.65 in favor of pregnant women (95% CI, 0.36-1.17; not significant). In the multivariate model, independent prognostic factors of distant recurrences were clinical tumor size and axillary lymph nodes involvement (Table 6). In the 576 patients who underwent a breast-conserving surgery, 129 patients developed a local recurrence. Pregnancy within the year before the breast cancer diagnosis was significantly associated with the risk of local recurrence with an adjusted hazard ratio (HR) of 1.75 (95% CI, 1.10-2.80; P = .022).

Table 6. Distant Recurrence Analysis According to Pregnancy Before or After Breast Cancer Treatment: Results of the Univariate and Multivariate Models
Prognostic factorsNo. of Patients5-Year DR Rate [95% CI]Univariate AnalysisMultivariate Analysis
HR [95%CI]P (LR)HR [95%CI]P (Cox)
  1. DR indicates distant recurrence; LR, log-rank; HR, hazard ratio; CI, confidence interval; NS, not significant; UICC, Union International Contre le Cancer; SBR, Scarff, Bloom, and Richardson; ER, estrogen receptors: PGR, progesterone receptor.

Prior pregnancy   .20 NS
 No78870 [68-73]1   
 Yes10563 [54-73]1.25 [0.90-1.74]   
Subsequent pregnancy   .10 (Cox) .15
 No7621 1 
 Yes1180.61 [0.34-1.09] 0.65 [0.36-1.17] 
Tumor classification (UICC)   <.001 .014
 T0-T137778 [74-83]1 1 
 T240065 [60-70]1.57 [1.23-2.02] 1.19 [0.92-1.56] 
 T3-T412154 [45-64]2.28 [1.66-3.14] 1.58 [1.11-2.25] 
Axillary lymph nodes   <.001 <.001
 044982 [79-86]1 1 
 1-326460 [54-67]2.57 [1.96-3.39] 2.45 [1.83-3.27] 
 >315745 [38-54]4.25 [3.19-5.67] 3.58 [2.60-4.91] 
 No dissection3875 [62-91]1.65 [0.93-2.96] 1.40 [0.73-2.67] 
Histological type   .032 NS
 Ductal79368 [65-71]1   
 Lobular3956 [42-74]1.30 [0.80-2.13]   
 Medullary1593 [82-100]0.15 [0.02-0.96]   
SBR grade   .011 NS
 I9483 [75-91]1   
 II36565 [61-71]1.94 [1.24-3.05]   
 III35064 [59-70]1.89 [1.20-2.97]   
ER   .28 NS
 Positive43367 [63-72]1   
 Negative34069 [64-74]0.88 [0.69-1.11]   
PGR   .65 NS
 Positive45670 [65-74]1   
 Negative31665 [60-71]1.06 [0.83-1.34]   

The score obtained with prognostic factors identified in the multivariate analysis was calculated on 908 women. The median score was equal to 9 (range, −1 to 18). Two-hundred and thirty-six patients had a score ≤6, 245 had a score between 7 and 10, and 277 had a score ≥10. Compared with the lower score group, the intermediate-risk group had a relative risk (RR) of death of 2.2 (95% CI, 1.4-3.4), and the high-risk group had an RR of 3.9 (95% CI, 2.6-5.8; P < 10-4; Fig. 2A). This score obtained with the prognostic factors of survival was applied to the distant recurrence-free interval. Compared with the lower score group, the intermediate-risk group had a RR of recurrence of 1.7 (95% CI, 1.2-2.4), and the high-risk group had a RR of recurrence of 3.2 (95% CI, 2.3-4.3; P < 10-4; Fig. 2B). Annual risks of death and distant recurrence according to the predictive score are illustrated in Figure 3. In the subset of patients considered to have a low-risk score, the annual risk of relapse remained low and constant over time. On the other hand, in patients presenting the higher score, recurrences occurred mainly during first years after the diagnosis. Beyond 80 months of follow-up, the annual risk showed a trend towards a similar level to lower-risk subgroups.

Figure 2.

Shown are (A) overall survival and (B) distant disease-free survival according to the predictive score.

Figure 3.

(A) Annual risks of death (B) and distant recurrences are depicted.

DISCUSSION

In the univariate model, a pregnancy, occurring within the year before the diagnosis, increased the risk of death and local relapse. Only the relationship with a local relapse persisted in the multivariate model (hazard ratio [HR], 1.75; P = .022), and a prior pregnancy did not predict independently for a poor outcome. In patients who had a prior pregnancy, our results contrast with other reports, showing that a recent pregnancy is an independent prognostic factor of death compared with nulliparous breast cancer patients (Table 7).6, 13-17 However, these reports involved older patients than ours (younger than 45 years), and the effect of pregnancy was studied within 2 years, instead of 1 year, before the diagnosis of breast cancer. Our findings are consistent with reports stating that women giving birth within 2 years before the diagnosis of breast cancer are more likely to have axillary lymph node involvement, PGR negative, and high histoprognostic grade tumors.16, 17 We have not studied the impact of the time interval between previous pregnancy and the diagnosis of breast cancer. The breast-feeding could potentially influence the risk of local recurrence, as it is generally assumed that a delay in diagnosis, possibly related to breast-feeding, worsens the prognosis. This could require a more intense locoregional treatment.

Table 7. Effect of Prior Childbirth on Breast Cancer Mortality Compared With Nulliparous Women: Data From Literature
StudyNo.Median Age, yPeriod of Pregnancy, HR (95% CI)
<1 Year<2 Year and >1 Years
  1. HR, hazard ratio; CI, confidence interval; NR, not reported.

Kroman65652≤45NR1.54 (1.24-2.02)
Mohle-Boetani13226<50NR2.91
Guinee14407<01.88 (0.88-3.98)1.09 (0.54-2.19)
Olson15540<45NR3.1 (1.8-5.4)
Daling161174<45NR2.7 (1.7-4.2)
Phillips17750<45NR1.9 (1.1-3.2)

This retrospective analysis carried out on a large cohort of young, early breast cancer patients showed that a pregnancy subsequent to the adjuvant therapy was not deleterious for the disease outcome, by taking into account the “healthy mother effect.” A possible adverse effect of recent childbirth could be related to the hormonal milieu of pregnancy and potential stimulation of preexisting malignant clones. In the present analysis, the rate of estrogen and PGR negative tumors was higher than 45%, although hormone receptor status was not an independent prognostic factor. Noteworthy, only 17.7% of patients received a hormonal treatment explaining the lack of difference in survival between hormone receptor positive and hormone receptor negative tumors. Our data suggest that a pregnancy subsequent to the treatment of breast cancer is not associated with an increased risk of death or recurrence. The only significant difference between both populations was the younger age of pregnant patients. Although not significant, women who became pregnant had a lower axillary lymph node involvement and a less frequent family history of breast cancer. These findings suggest that women who became pregnant were more likely to have a better outcome regardless of pregnancy (Table 8). Considering data of the literature, no detrimental effect of pregnancy has been previously reported.7, 11, 18-20 However, the bias due to the “healthy mother effect” should be taken into account. The time before pregnancy is very important. This duration is possibly involved in the selection of good-prognosis patients and corresponds to current standards to deliver hormonotherapy in hormone receptor positive patients and/or trastuzumab in HER2-positive patients.

Table 8. Effect on a Pregnancy Subsequent to Breast Cancer Survival: Data From Literature
StudyNo.HR95% CIPregnancy Impact on Death
  1. HR indicates hazard ratio; CI, confidence interval.

Gelber71370.440.21-0.96Decreased risk
Sankila11914.802.20-10.5No adverse effect
von Schoultz18500.480.18-1.29No adverse effect
Velentgas19530.800.30-2.30No adverse effect
Kroman201730.550.28-1.06Decreased risk

The women with a favorable prognostic score had a lower annual risk of distant relapse, remaining constant in the course of time. On the contrary, the women with a poor prognostic score had a risk of distant recurrence that did not level off for 80 months. Beyond this period, their risk did not seem to be greater than the risk of women with a good prognostic score, taking into account that the median follow-up was 87 months. The advice given to young women desiring to have a child is important to define prognostic markers for early recurrence of breast cancer. We have developed a decision-making tool based on a score index taking into account patient and tumor characteristics. Some additional analyses are planned to validate this model in a larger population of patients aged younger than 35 years. To date, we can conclude that patients with a good-prognosis score have a low risk of distant recurrence. The risk remains constant in the course of time, and no contraindication to pregnancy should be provided. For patients with a poor-prognosis score, the annual risk of distant recurrence remains high within 5 years after the adjuvant therapy. Beyond 80 months, their risk becomes similar to the good-prognosis subgroup. In the subset of patients having a high risk of relapse, it may be preferable to postpone a pregnancy approximately 5 years after the treatment of breast cancer.

For women aged younger than 35 years, a pregnancy was not an independent prognostic factor of death, compared with nonpregnant patients. Our results must be interpreted with the limitations of a descriptive analysis, knowing that important factors were not considered. Obviously, a randomized trial assigning patients to get pregnant, or not, is not possible. Thereby, we have tried to answer in an indirect way. This study could reassure women who plan to have children after surviving breast cancer. The patients should clearly discuss with their physician about the potential risks and some of the unknowns that remain.

Conflict of Interest Disclosures

The authors made no disclosures.

Ancillary