In this large register-based study of breast cancer cases, histological type and grade of tumors were found to differ significantly by age, parity, and maternal age at first birth. The difference between nulliparous and parous women by histological type was most pronounced among women at pre- and perimenopausal ages. Ductal tumors represented the most common histological type in all subgroups defined by time interval since birth. Nevertheless, certain histological types appeared to be more susceptible to an adverse effect of a pregnancy. Further knowledge on this issue may provide important information in the treatment of women diagnosed with breast cancer shortly after a childbirth.
The results from this study need to be interpreted with some caution. Our study comprised information on breast cancer cases diagnosed in Norway in the period 1955-99. The histological classification system and procedures have varied over time, and may also have varied at the individual as well as institutional level. Still, the distribution of histological types across the two main calendar periods considered (before/after 1993), was roughly the same. Moreover, the observed age distribution of histological types appeared to be rather similar to that reported by others [23, 29], except for a slightly higher proportion of ductal tumors. Histological grading of breast cancer tumors was not commonly carried out in Norway before 1993. Although the missing rate was high, there was no reason to believe that the information on histological grade was not missing at random. In some cases, however, the missing rate differed between the groups that were compared, making interpretation of results more difficult. Nevertheless, it is unlikely that our results can be completely explained by misclassification, or incompleteness of data.
Established associations between reproductive factors and the risk of breast cancer will mostly reflect associations with ductal tumors, which comprise about 80% of all breast cancer cases. A complete risk analysis of each subtype, in particular when considering time-related effects of a pregnancy, is difficult due to the low number of cases of certain histological types. However, the population at risk for each histological type in our cohort is the same, and information from the present study with respect to contrasting histological types, is thus comparable to the few previous studies that have calculated risk estimates specific to histological type, despite different measures of effect. Most of these studies, however, have focused on traditional reproductive factors, rather than time-related effects of a childbirth.
Histological type of tumor by parity and age at first birth
Consistent with results from previous studies [1, 7], our results indicated that the overall protective effect of a childbirth in relation to breast cancer risk may be particularly pronounced for mucinous tumors. In our study, a similar pattern was seen for tubular tumors. Our results also give some support to previous findings [3, 7] of a less pronounced protective effect of parity in relation to medullary tumors, perhaps even an association in the opposite direction. In our study, Paget disease also appeared to be more common in women of high parity. Few previous studies have considered this disease.
Results from some previous studies have indicated that lobular tumors may be more strongly associated with age at first birth than other histological types of breast cancer [1–3, 7]. Our result is consistent with these findings. Tubular and medullary tumors, however, were less strongly associated with maternal age at first birth, as compared to ductal tumors.
Histological type of tumor by time interval since a childbirth
The transient increase in risk of breast cancer shortly after a childbirth is assumed to be related to a growth enhancing effect of pregnancy-related factors on pre-malignant breast cells [8, 21]. Biological features of malignant or pre-malignant breast cells may affect susceptibility to pregnancy-related exposure and progression of disease into a clinically detectable phase [26, 30, 31]. A main objective of the present study was therefore to examine whether breast cancer tumors diagnosed shortly after a childbirth tend to be of a particular histological type. Women with a familial predisposition are more likely to have pre-cancerous lesions at a young age . The number of cases in our study, however, was too low to perform an analysis separately for this subgroup. Nevertheless, results from overall associations between histological type and familial risk are briefly discussed.
The proportion of ductal tumors seemed to be particularly high among women diagnosed <2 year after first childbirth, but no clear trend with time since birth was seen. In one previous study of breast cancer diagnosed during pregnancy , all tumors were ductal carcinomas and 32 of 38 (84.2%) were poorly differentiated. Malignant sarcomas (mainly phyllodes tumor) occurred at significantly higher frequency in women diagnosed <2 years after birth. A rapid growth of a phyllodes tumor during pregnancy has been reported previously . Patients with phyllodes tumors have in general been found to have a good prognosis [27, 34], except those with large and poorly differentiated tumors [35, 36].
The crude proportion of lobular tumors increased with increasing time since birth, but the age-adjusted trend estimate indicated an association in the opposite direction, with a decreasing trend in risk stronger than for ductal tumors. Lobular tumors have been found to be particularly sensitive to hormonal exposure [4, 5], and may thus be susceptible to exposure to pregnancy hormones. On the other hand, such tumors tend to be more common among older women [3, 37, 38]. An excess familial risk has been noted [37–40], but was not confirmed in the present study.
Medullary tumors occurred with significantly higher frequency 2-5 years after last birth in our study. The unadjusted trend estimate indicated a decrease in the proportions of medullary tumors with increasing time since birth. The age-adjusted analyses, however, suggested an association in the opposite direction. The observed non-linear pattern with time since birth, with a peak 2-5 years after last birth, made interpretation of estimated linear trends somewhat difficult though. Thus, our data may indicate that medullary tumors are particularly susceptible to pregnancy-related exposure. Medullary tumors have previously been noted to progress rapidly from a pre-clinical to a clinically detectable phase . The tumors are often associated with negative prognostic markers [22, 23, 29, 41], but a favorable prognosis has been observed in most studies [24–27, 38, 41–43]. Medullary tumors have previously been found to be associated with a familial and/or hereditary risk, though not consistently [3, 38, 39, 42–44]. No association with familial risk was seen in the present study.
Results from the present study also suggested that a pregnancy may have particular impact on the development of Paget disease. As far as we know, no previous studies have reported that this disease may be more common in the first years following a childbirth. Consistent with previous suggestions [45, 46], most women diagnosed with Paget disease in the present study had an underlying tumor of ductal type. The tumor is often high grade and estrogen and progesterone receptor negative [45, 47], and has been found to be positive also for other molecular markers that are frequently associated with more aggressive tumor behavior . A high degree of inflammatory infiltrate in tumor-surrounding tissue has also been noted . It has previously been suggested that inflammation-like processes in connection with the rebuilding of breast tissue after pregnancy and lactation may play a role for growth and spread of breast cancer tumors [48, 49]. In the present study, Paget disease appeared to be more common in women with a familial risk.
The proportion of unspecified carcinomas decreased significantly with increasing time since birth in our study, both in unadjusted and adjusted analyses. The rather high proportion of unspecified carcinomas in women diagnosed shortly after birth, however, may reflect difficulties with classification of the tumor. About half the tumors within this subgroup were poorly differentiated.
Histological grade of tumor by reproductive factors
The present finding that younger women, and women with recent childbirths more often have poorly differentiated tumors, is consistent with results from previous studies [20, 50]. Results from our analyses with mutual adjustment, however, indicated that the high proportion of poorly differentiated tumors in women with a recent childbirth may be related to their generally younger age. On the other hand, the borderline significant association with time since last birth in the age-adjusted analyses suggested an effect of pregnancy-related factors rather than a confounding effect of age. Additional adjustment for order of birth, however, weakened the association with time interval since most recent birth.
Consistent with results from two previous studies [20, 51], we observed a larger proportion of poorly differentiated tumors among women with an early first birth. No clear trend with age at first birth remained in our age-adjusted analyses, however. Rather unexpectedly, we observed an increase in the proportion of high-grade tumors with increasing number of births among parous women. The trend estimate did not change after adjustment for age at diagnosis and maternal age at first birth. A similar pattern was observed in another study , but the positive though non-significant trend with parity in that study was mainly related to a low proportion of high-grade tumors in nulliparous women, in contrast to that seen in our study.