SEARCH

SEARCH BY CITATION

Keywords:

  • oral contraceptives;
  • extra-tumoral breast tissue;
  • ductal hyperplasia;
  • ductal carcinoma in situ;
  • risk factors

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

The association between oral contraceptive (OC) use and benign breast changes in extra-tumoral breast tissue was studied histologically in 1,503 breast cancer patients from The WHO Collaborative Study of Neoplasia and Steroid Contraceptives. The occurrence of ductal hyperplasia, ductal atypia, sclerosing adenosis, cysts, apocrine metaplasia, apocrine hyperplasia, apocrine atypia, adenosis, lobular atypia, duct ectasia, calcifications, inflammatory reaction, lactational metaplasia and a high epithelial-stromal ratio was graded semi-quantitatively. Prevalence odds ratio (POR) for each histologic variable was calculated by logistic regression analyses. Patients who had ever used OC had lower occurrence of ductal hyperplasia than never users (POR 0.72 (95% CI 0.52–0.99)). Current use and more than 8 years of use was also associated with a lower prevalence of ductal hyperplasia (POR 0.40 (0.20–0.81) and POR 0.33 (0.17–0.64), respectively). Age > 35 years at first use was associated with increased prevalence of ductal carcinoma in situ (POR 2.15 (1.05–4.40)), but not of atypical ductal hyperplasia. Our results show that the effects of OC use on ductal hyperplasia in non-neoplastic breast tissue of breast cancer patients are similar to what others have found in patients with benign breast disease only. The increased prevalence of extra-tumoral ductal carcinoma in situ in breast cancer patients who started OC use at high age may possibly be explained by a longer preinvasive phase in these patients. © 2005 Wiley-Liss, Inc.

Cancer-bearing breasts have been shown to contain more benign breast changes than autopsy control breasts,1 and some types of benign changes are associated with an increased risk of subsequent breast cancer.2 Epidemiological studies of benign breast disease and of breast cancer have identified multiple risk factors, some of which are common to both benign and malignant breast disease.3, 4 These observations are consistent with the predominant theory of the pathogenesis of breast cancer as a multi-step process which is, at least in part, morphologically expressed by the sequence of epithelial hyperplasias without atypia, atypical hyperplasia and carcinoma in situ.5

The use of steroid contraceptives has in some studies been found to increase the risk of breast cancer in the general group of ever users. Other studies have shown an increased risk in certain subgroups of users, and some have shown no increased risk. A meta-analysis of 54 studies concluded that an increased risk of having breast cancer diagnosed is present among current users and during the first 10 years after cessation of use. The findings were taken to suggest either an earlier diagnosis of breast cancer in users, or a promoting effect of steroid contraceptives on late stages of the development of breast cancer or a combination of both.6

The relationship between the occurrence of benign changes in cancer-bearing breasts and oral contraceptive (OC) use has been studied previously in 2 smaller studies by Miller et al.7 and Hulman et al.,8 the former focusing on the immediate vicinity of the carcinoma and the latter restricted to patients less than 36 years of age. Miller et al. found no association between epithelial hyperplasia and OC use, whereas Hulman et al. found a reduction of ductal hyperplasia and blunt duct adenosis among users.

Comparisons of benign breast changes in breast cancer patients exposed and not exposed to a particular breast cancer risk factor may help to explain how that risk factor influences the carcinogenic process. We have studied benign changes in the extra-tumoral tissue in cancer-bearing breasts taken from women participating in the WHO Collaborative study of Neoplasia and Steroid Contraceptives.9

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

This study was based on data and histologic material from the WHO Collaborative study of Neoplasia and Steroid Contraceptives.9 In that study, the collaborating centers were requested to send to the reference pathologist (H.S.) the following histologic material from each breast cancer patient, if available: 2 slides from neoplastic mass, 2 from non-neoplastic surrounding breast tissue and 1 from the nipple. The slides were stained with eosin and hematoxylin.

Information on use of oral and injectable steroid contraceptives, as well as data on other known or possible risk factors and confounding variables, were collected by personal interview. The great majority of OC users used combined OCs. A detailed description is given elsewhere.9

From a series of cases from the WHO study included in a previous study on histologic types of breast cancer,10 2,476 randomly selected cases were screened for inclusion in the present study. Included were only slides free from invasive carcinoma, but containing recognizable non-neoplastic epithelial tissue. One to 9 slides that fulfilled these criteria were available from 1,513 patients (61%). Three cases were omitted from the analyses because of missing information on steroid contraceptive use, and 5 cases from Nigeria and 2 from Kenya were omitted because of small numbers. The number of cases at each participating center for the remaining 1,503 patients are given in Table I. Only patients who had been at reproductive age after the introduction of steroid contraceptives in the respective countries were included, and most patients were under 55 years.

Table I. Use of Oral or Injectable Steroid Contraceptives by Country
CountryEverNeverTotal
Israel167352519
GDR248146394
Australia363066
Thailand96159255
China195877
Philippines185977
Mexico192746
Colombia51217
Chile133952

The slides from each case were marked with a random number, masking the origin of the case. The slides were then examined in the order of random numbers. The slide readings were performed without any knowledge of patients' age, or other clinical data. For each patient, the following features were scored semi-quantitatively on a scale of 1–4 (absent/normal–mild–moderate–severe/florid): adenosis, sclerosing adenosis, lobular atypia, ductal hyperplasia, ductal atypia, apocrine metaplasia, apocrine hyperplasia, apocrine atypia, duct ectasia, lactational metaplasia, cysts and inflammatory reaction. For atypias, the score indicated none, uncertain, atypical hyperplasia and carcinoma in situ. Cysts were classified as absent, <3 mm, 3–9 mm, and >9 mm. Calcification was scored as absent or present. The epithelial-stromal ratio was scored 1–6 as lower than normal, normal, mild, moderate or marked increase and epithelial predominance. The technical quality of the slides was scored 1 (poor)–5 (excellent).

Proliferative breast disease was defined according to Dupont and Page11 to include moderate and florid ductal hyperplasia, atypical ductal hyperplasia, sclerosing adenosis and atypical lobular hyperplasia, but with the exclusion of carcinoma in situ and the addition of atypical apocrine hyperplasia. Non-proliferative breast disease was defined to include lactational metaplasia, cysts, apocrine metaplasia, calcification and duct ectasia. Since adenosis and the epithelial-stromal ratio showed variations with age similar to those of glandular volume in the general population,12 they were considered to reflect physiologic changes and not included with either proliferative or non-proliferative changes.

Data were analyzed with SAS software.13 For logistic regression analyses, the scores for each histologic variable were grouped into 2 levels, the lower level including either absent/normal only or normal plus mild changes. Comparisons were expressed by the prevalence odds ratio (POR), which is the ratio of odds for finding a particular change in patients exposed to a factor in question to odds for finding the same change in patients not exposed to the same factor. All analyses were adjusted for age at diagnosis (by 5-year age groups) and country of residence. Adjustments for number and quality of slides were conducted in the analyses of features for which the recorded prevalence was previously shown to be influenced by these factors.12 Patients who were currently pregnant or had given birth or nursed within the last year before surgery were excluded from the analyses of adenosis, epithelial-stromal ratio and lactational metaplasia.

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

OCs had ever been used by 610 patients (40.6%) and only 33 (2.2%) had ever used injectable contraceptives. All users of injectable contraceptives came from low-risk countries; 20 from Thailand, 5 from Mexico, 5 from Chile and 3 from China. Distribution of steroid contraceptive use according to country of residence is shown in Table I.

Women who had ever used oral steroid contraceptives had a lower prevalence of ductal hyperplasia than never-users, but the association was of borderline significance (POR 0.72 (0.95% CI 0.52–0.99)). All other histologic changes showed no significant association with ever use of OCs (Table II). Similar analyses on injectable contraceptives (not shown) were based on small numbers and no significant associations were found.

Table II. The Prevalence Odds Ratio (POR) for the Occurrence of Benign Changes in Extra-Tumoral Tissue in Breast Cancer Patients: Variation with Ever Use of Oral Contraceptives
 Never use (Ref.; POR=1.00)Ever used oral contraceptives
Pos.TotalPos.TotalPOR95% CI
  • 1

    Atypical ductal hyperplasia and ductal carcinoma in situ.

  • 2

    Adjusted for age, country, and number and quality of slides.

  • 3

    Moderate and florid hyperplasia. Patients with ductal atypia excluded.

  • 4, 4,

    Adjusted for age, country and number of slides.

  • 5, 5,

    Adjusted for age and country.

  • 6

    Ductal hyperplasia (moderate and florid), atypical ductal hyperplasia, sclerosing adenosis, atypical lobular hyperplasia, atypical apocrine hyperplasia. Patients with carcinoma in situ excluded.

  • 7, 7,

    Patients who were currently pregnant or had given birth or nursed during the last year before surgery excluded.

  • 8

    Adjusted for age, country and quality of slides.

  • 9

    Lactational metaplasia, cysts, apocrine metaplasia, duct ectasia and calcification.

Ductal atypia1,284882495880.990.66–1.48
Ductal hyperplasia3,4168798825390.720.52–0.99
Sclerosing adenosis535882255881.060.59–1.88
Any proliferative disease4,62108141145460.840.63–1.12
Lactational change5,726837215391.300.69–2.46
Cysts24548823305881.110.87–1.41
Apocrine metaplasia22488821975881.090.84–1.42
Apocrine hyperplasia21968821545880.960.72–1.28
Calcification52388285880.620.27–1.44
Duct ectsia81588255880.520.18–1.47
Any non-proliferative disease2,7,94768373255391.000.78–1.29
Adenosis4,71718371355391.120.84–1.49
High epithelial-stromal ratio4,72358371725391.110.85–1.46

Among users of OCs, the occurrence of ductal hyperplasia decreased significantly with increasing duration of use and was significantly lower in women with more than 8 years of use than in never-users (Table III). This trend remained significant when the analysis was repeated after the exclusion of current users (p = 0.034).

Table III. The Prevalence Odds Ratio (POR) for the Occurrence of Ductal Hyperplasia in Extra-Tumoral Tissue of Breast Cancer Patients: Variations with Total Duration of Oral Contraceptive Use, Time Since Last Use and Age at First Use of Oral Contraceptives
Variables of OC useDuctal hyperplasia1,2POR95% CIp for trend
YesNo
  • 1

    Moderate or florid hyperplasia. Patients with ductal atypia excluded.

  • 2

    Adjusted for age, country and number of slides.

Total duration of use (years)
 Never use1686301.00Ref.0.011
 <1281300.860.54–1.36
 1–217740.940.53–1.68
 3–8191040.760.43–1.33
 >8121250.330.17–0.64
Time since last use (months)
 Never use1676291.00Ref.N.S.
 >108231200.740.45–1.21
 36–108291400.800.50–1.27
 3–3514660.840.44–1.61
 All previous   users663260.780.56–1.10
 Current use111130.400.20–0.81
Age at first use (years)
 Never used1415581.00Ref.N.S.
 <28241600.680.40–1.17
 28–34321600.740.47–1.17
 >3418990.710.41–1.25

Current users had significantly less ductal hyperplasia than never-users. The findings for previous users were intermediate between those for current users and never-users, but the difference between previous users and never-users was not statistically significant (all previous users: POR = 0.78 (0.56–1.10)). There was no significant trend with time since last use among previous users (p = 0.801). The exclusion of women with more than 8 years duration of use from the analyses of time since last use gave no significant change in these results. No significant association was found between ductal hyperplasia and age at first use. Separate analyses for moderate and florid hyperplasia (not shown) revealed no significant difference in their relationship to either duration of use, time since last use or age at first use.

The prevalence of any proliferative disease showed variations with OC use that were closely similar to those of ductal hyperplasia alone shown in Table III, with statistically significance of corresponding items in the analyses.

The prevalence of ductal atypia was not associated with total duration of OC use or time since last use, although the point estimate for over 8 years of use was similar to that for ductal hyperplasia (0.33 vs. 0.46). Some decrease in ductal atypia was observed with increasing time since last use among previous users, but the trend was not statistically significant (Table IV). The prevalence of ductal atypia increased significantly with age at first OC use (test for trend p = 0.033). Compared with never-users, ductal atypia was somewhat less frequent in patients who started OC use before age 28, and significantly more frequent in those who started after age 34. Five of the 15 women with ductal atypia and age > 34 at first use were also included among the 12 with ductal atypia and 3–35 months since last use. Thus, the positive findings in these 2 groups (POR 1.93 and 1.87, respectively) are partly based on the same women. Because of high correlation between time since last use and age at first use, we were unable to distinguish between the possible effects between these 2 variables on ductal atypia.

Table IV. The Prevalence Odds Ratio (POR) for the Occurrence of Ductal Atypia in Extra-Tumoral Tissue of Breast Cancer Patients: Variations with Total Duration of Oral Contraceptive Use, Time Since Last Use and Age at First Use of Oral Contraceptives
Variables of OC useDuctal atypia1,2POR95% CIp for trend
YesNo
  • 1

    Atypical ductal hyperplasia and ductal carcinoma in situ.

  • 2

    Adjusted for age, country, number and quality of slides.

Total duration of use (years)
 Never use847981.00Ref.N.S.
 <1161581.000.56–1.80
 1–28910.890.41–1.94
 3–8181231.630.89–2.97
 >851370.460.18–1.23
Time since last use (months)
 Never use847961.00Ref.N.S.
 >108121430.790.42–1.51
 36–108161691.070.59–1.93
 3–3512801.870.91–3.84
 All previous   users403921.060.70–1.63
 Current use81240.810.36–1.84
Age at first use (years)
 Never used676991.00Ref.0.033
 <28121840.660.33–1.32
 28–34161920.980.53–1.80
 >34151171.931.02–3.64

When the subgroups of ductal atypia were analyzed separately, ductal carcinoma in situ (n = 81) still had a higher prevalence in women who were 35 years or more at first use (POR 2.15 (1.05–4.40)), whereas the occurrence of atypical ductal hyperplasia (n = 29) was not significantly increased (POR 1.28 (0.35–4.60)) in patients who were 35 years or more at first use. With an increasing age at first use, test for trend was significant for ductal carcinoma in situ (p = 0.0079) and non-significant for atypical ductal hyperplasia (p = 0.8538). For total duration and time since last use, separate analysis (not shown) showed no significant difference between atypical ductal hyperplasia and ductal carcinoma in situ.

The prevalence of cysts (Table V) was not associated with total duration of OC use or time since last use. The prevalence of cysts increased with an increasing age at first OC use, but the trend was not statistically significant.

Table V. The Prevalence Odds Ratio (POR) for the Occurrence of Cysts in Extra-Tumoral Tissue of Breast Cancer Patients: Variations with Total Duration of Oral Contraceptive Use, Time Since Last Use and Age at First Use of Oral Contraceptives
Variables of OC useCysts1POR95% CIp for trend
YesNo
  • 1

    Adjusted for age, country, number and quality of slides.

Total duration of use (years)
 Never use4544281.00Ref.N.S.
 <192821.120.79–1.59
 1–248510.900.57–1.40
 3–880611.140.76–1.71
 >887551.070.71–1.63
Time since last use (months)
 Never use4524281.00Ref.N.S.
 >10887681.160.81–1.68
 36–108106791.190.84–1.69
 3–3545470.730.45–1.19
 All previous users2381941.070.82–1.38
 Current use76561.110.72–1.71
Age at first use (years)
 Never used3943721.00Ref.N.S.
 <28911050.810.55–1.18
 28–34122861.090.77–1.54
 >3486461.510.99–2.29

The prevalence of adenosis (Table VI) was not significantly related to total duration of OC use or time since last use, but increased significantly with increasing age at first use.

Table VI. The Prevalence Odds Ratio (POR) for the Occurrence of Adenosis in Extra-Tumoral Tissue of Breast Cancer Patients: Variations with Total Duration of Oral Contraceptive Use, Time Since Last Use and Age at First Use of Oral Contraceptives
Variables of OC useAdenosis1POR95% CIp for trend
YesNo
  • 1

    Adjusted for age, country and number of slides.

Total duration of use (years)
 Never use1716661.00Ref.N.S.
 <1401151.240.82–1.88
 1–226551.550.92–2.60
 3–8321030.930.58–1.50
 >8341041.070.66–1.74
Time since last use (months)
 Never use1716641.00Ref.N.S.
 >108321141.030.66–1.61
 36–108411291.120.74–1.70
 3–3520591.080.61–1.93
 All previous users933021.070.78–1.46
 Current use40811.530.95–2.48
Age at first use (years)
 Never used1485781.00Ref.0.0257
 <28461240.920.59–1.42
 28–34521441.260.85–1.88
 >3431931.791.10–2.92

The occurrence of a high epithelial-stromal ratio was not associated with total duration of use or age at first use of OCs. Current use was significantly associated with higher epithelial-stromal ratio (POR 1.67 (1.06–2.65)) than never-users, but previous use was not. Among previous users, test for trend with an increasing time since last use was non-significant.

Total duration of OC use, time since last use and age at first use were not significantly associated with the prevalence of sclerosing adenosis, lactational change, apocrine metaplasia, apocrine hyperplasia, calcification, duct ectasia or the group of any non-proliferative disease.

No significant association with OC use was found for apocrine atypia (present in 11 patients), lobular atypia (present in 36 patients) or inflammatory reaction (76 patients).

Because pregnancy and lactation are associated with extensive temporary changes in breast histology, all of the analyses of histological changes described earlier were also run with the inclusion of years since last pregnancy and years since last breast-feeding in the models. In no case did this significantly alter the results.

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Ductal hyperplasia without atypia is the only lesion that showed a significant association with ever use of OCs in this study. Ductal hyperplasia without atypia is the major component of benign proliferative breast disease,11 and has in previous related studies been described under synonymous or largely overlapping terms as ductal atypia grade 2,14 epithelial hyperplasia, hyperplasia of usual type, epitheliosis and papillomatosis.

In a study of extra-tumoral tissue in breast cancer patients under 45 years, Miller et al.7 found no significant association between the prevalence of epithelial hyperplasia and ever use of OCs. Their study was based on much smaller numbers of patients, and was different from our study, in that they included only slides containing carcinoma and thus studied the area immediately adjacent to the tumor. In our material, the contributing centers were asked to include sections from macroscopically non-tumorous breast tissue, and because we wanted to be blinded with regard to the type of carcinoma present, we excluded all slides containing invasive carcinoma from the present study. Thus, compared to Miller et al., our findings are more representative of the general state of the breast outside the area afflicted by cancer.

Hulman et al.8 studied the extra-tumoral breast tissue of cancerous breasts in women less than 36 years of age. They found that ductal hyperplasia was marginally reduced in ever users compared to never users, and as in our study, ductal hyperplasia was reduced in current users when compared to past users. They also specifically identified the entity of “blunt duct adenosis”, which is a type of hypertrophy and hyperplasia of the intralobular ductules. Similar to our findings on ductal hyperplasia, they found blunt duct adenosis to be less common among current users than among past or never-users, and less than half as common among long-term users as among short-term users. Considering the lower age of their patients, the similarity raises the possibility that the blunt duct adenosis of younger women may develop into ductal hyperplasia of the older by continued epithelial proliferation and expansion of ductules.

Others have studied the effect of OCs in breast tissue removed from benign breast disease. LiVolsi et al.,15 Pastides et al.,16 Berkowitz et al.17 and Hsieh et al.18 all used the scoring system of Black and Chabon14 and found that lower or intermediate grades of ductal changes, which largely correspond to ductal hyperplasia in our terminology, were reduced in ever users of OCs. Pastides et al.16 and Berkowitz et al.17 also used the classification of benign breast disease proposed by Haagensen19 and found that OC use was associated with reduced prevalence of papillomatosis, which is closely related to ductal hyperplasia. All of these findings are in good agreement with our observations in cancerous breasts.

Also in agreement with our findings in cancerous breasts are the observations that the prevalence of ductal hyperplasia or benign proliferative epithelial changes in benign breast disease is negatively associated with the length of OC use.18, 20 Our data further indicate that after cessation of OC use, ductal hyperplasia rapidly returns to a level intermediate between that of current users and never users.

As ductal hyperplasia is associated with increased risk of breast cancer11 and is considered to be an early step in the multi-step development of breast cancer,5 it has been a problem to explain how OC use could reduce the frequency of ductal hyperplasia, but not the risk of breast cancer. LiVolsi et al.15 suggested that OC use protects only against the lower grades of ductal hyperplasia that are not firmly associated with increased risk of breast cancer, but not against the higher grades of ductal atypia. These findings were not confirmed by Hsieh et al.18 who found all grades of ductal hyperplasia and atypia in benign breast disease to be reduced in OC users. They suggested that the discrepancy between the risk of ductal hyperplasia and the risk of breast cancer in OC users could be explained by reduced detection of benign breast disease in OC users, without any important concomitant effect of OC use on the cellular pathology, i.e., on the occurrence of hyperplasia, metaplasia or atypia. In our study, reduced detection is unlikely to have been a problem because all women had concurrent breast cancer, which would be less subject to detection bias. An alternative explanation could be that, because of differences in genetic background, the effect of OC use on ductal epithelium might be different in patients who develop breast cancer and those who do not. Our findings, however, refute this hypothesis; the associations of OC use and ductal hyperplasia that we observed in breast cancer patients were similar to those observed by others in patients with benign breast disease.

The decrease of ductal hyperplasia with OC use thus seems to be a real finding both in women who develop breast cancer and in those who do not. If usual ductal hyperplasia is an early precursor of breast cancer, as has commonly been assumed, the explanation for the absence of a concomitant decrease in breast cancer risk in OC users may be sought in qualitative rather than quantitative aspects of the hyperplastic epithelium. Genetic studies have shown a variety of changes in ductal hyperplasia.21, 22 OCs have been shown to extend the duration of proliferation of normal breast epithelium during each menstrual cycle, and may thus enlarge the “window of opportunity” during which transforming events could occur.23 It would be of interest, therefore, to study whether the hyperplastic ductal epithelium in OC users have more genetic changes that could be related to cancer development than the corresponding changes in non-users.

Alternatively, the usual ductal hyperplasia may be a mere risk indicator rather than a cancer precursor, as has been suggested in some recent studies.24, 25, 26, 27 The decrease of ductal hyperplasia with OC use may then be an event that does not directly interact with the multiple-step carcinogenic process.

Atypical ductal hyperplasia and ductal carcinoma in situ are generally considered to be the immediate precursors of invasive ductal carcinoma. When detected in cancerous breasts, these lesions may be considered as part of the neoplastic disease for which the patients were operated, and their rate of detection may be related to the duration of the preinvasive phase of the cancer and the extent to which these lesions were allowed to grow before progression to invasive carcinoma. The longer the cancer is left to grow in its preinvasive phase, the more widespread will it presumably be at the time when invasion occurs, and the greater the chance of its presence outside the invasive area when the tumor presents for diagnosis and surgery. We observed a significant increase in the prevalence of ductal carcinoma in situ with increasing age at first use of OCs. If this is not a chance finding, it could indicate a longer preinvasive phase in cancer patients who started OC use at a late age.

Although current users had reduced prevalence of ductal hyperplasia, the prevalence of a high epithelial-stromal ratio was increased. This is explained by the increased occurrence of adenosis in this group.

In previous analyses on the same study material, we found no effect on the prevalence of ductal hyperplasia in extra-tumoral breast tissue by a positive family history of breast cancer28 or in residents of countries with high risk of breast cancer when compared to those of low-risk countries,12 indicating that the effect of these breast cancer risk factors on the mammary epithelium is different from that of steroid contraceptives.

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References