Abstract
Lesser degrees of terminal duct-lobular unit (TDLU) involution predict higher breast cancer risk; however, standardized measures to quantitate levels of TDLU involution have only recently been developed. We assessed whether three standardized measures of TDLU involution, with high intra/inter pathologist reproducibility in normal breast tissue, predict subsequent breast cancer risk among women in the Mayo benign breast disease (BBD) cohort. We performed a masked evaluation of biopsies from 99 women with BBD who subsequently developed breast cancer (cases) after a median of 16.9 years and 145 age-matched controls. We assessed three metrics inversely related to TDLU involution: TDLU count/mm2, median TDLU span (microns, which approximates acini content), and median category of acini counts/TDLU (0–10; 11–20; 21–30; 31–50; >50). Associations with subsequent breast cancer risk for quartiles (or categories of acini counts) of each of these measures were assessed with multivariable conditional logistic regression to estimate odds ratios (ORs) and 95 % confidence intervals (CI). In multivariable models, women in the highest quartile compared to the lowest quartiles of TDLU counts and TDLU span measures were significantly associated with subsequent breast cancer diagnoses; TDLU counts quartile4 versus quartile1, OR = 2.44, 95 %CI 0.96–6.19, p-trend = 0.02; and TDLU spans, quartile4 versus quartile1, OR = 2.83, 95 %CI = 1.13–7.06, p-trend = 0.03. Significant associations with categorical measures of acini counts/TDLU were also observed: compared to women with median category of <10 acini/TDLU, women with >25 acini counts/TDLU were at significantly higher risk, OR = 3.40, 95 %CI 1.03–11.17, p-trend = 0.032. Women with TDLU spans and TDLU count measures above the median were at further increased risk, OR = 3.75 (95 %CI 1.40–10.00, p-trend = 0.008), compared with women below the median for both of these metrics. Similar results were observed for combinatorial metrics of TDLU acini counts/TDLU, and TDLU count. Standardized quantitative measures of TDLU counts and acini counts approximated by TDLU span measures or visually assessed in categories are independently associated with breast cancer risk. Visual assessment of TDLU numbers and acini content, which are highly reproducible between pathologists, could help identify women at high risk for subsequent breast cancer among the million women diagnosed annually with BBD in the US.
Keywords: Lobular involution, Breast cancer risk
Introduction
Analysis of the U.S. mammography registry data from 1996 to 2008 demonstrates that 1.2 % of screening mammograms prompt a breast biopsy, of which nearly 75 % are diagnosed as benign breast disease (BBD), including approximately 70 % with nonatypical lesions and 3.4 % with atypia [1]. A recent meta-analysis demonstrated that women with proliferative BBD without atypia experience a relative risk (RR) of developing breast cancer of 1.76, which increases to 3.93 if atypia is present [2]. Thus, women with biopsy-proven BBD represent a large number of future breast cancer cases. As proliferative BBD with atypia is a factor that is included in some models to predict breast cancer risk and used to counsel patients about chemoprevention [3, 4], developing risk stratification tools for women diagnosed with BBD using breast tissue markers may improve identification of women at highest risk for breast cancer [5, 6].
Data from retrospective studies suggest that analysis of “normal-appearing” terminal duct-lobular units (TDLUs) adjacent to BBD may have value in predicting risk of breast cancer [7–9]. TDLUs are normal structures within the breast that produce milk during lactation and also represent the primary source of breast cancer precursors (Fig. 1) [10]. With physiological aging and completion of child bearing, TDLUs of most women involute, reflected in a reduction in TDLU size, acini counts/TDLU (epithelial substructures) and total TDLU counts per standard unit area [7, 11–14]. In a cohort of women with BBD within the Mayo Clinic, qualitative assessment of degrees of TDLU involution showed that women with lesser degrees of TDLU involution had an approximately two-fold increased risk of developing breast cancer compared with expected population-based rates, irrespective of the severity of BBD [7], and use of these measures in risk prediction models showed improved discrimination of breast cancer cases and noncases [4]. Similarly, in a cohort within the Nurses’ Health Study II, women with BBD whose breast tissue contained TDLUs with the lowest acini content were also at about 30 % lower risk of developing breast cancer [8]. Preliminary efforts to use quantitative measures of TDLU involution, which could further discriminate women at highest risk, have focused on acini content within TDLUs. Using image analysis to evaluate a subset of biopsies from the Mayo BBD cohort, investigators showed women with higher acini counts/TDLU were at increased risk, independent of Gail model risk prediction [15], supporting quantitative measures of involution to further discriminate women at highest risk for breast cancer.
Fig. 1.
Standardized terminal duct-lobular unit (TDLU) involution assessment in the Mayo Benign Breast Disease Cohort. Greater TDLU counts, TDLU span, and number of acini per TDLU associated with reduced levels of TDLU involution were assessed from digitized images of H&E-stained tissue sections. Panel A shows a digital H&E section with multiple TDLUs (TDLU counts), as well as the outline of the tissue area captured in mm2 using the lasso tool, to calculate the TDLU/mm2 measures. For up to ten normal TDLUs per section, the longest TDLU span was measured in microns and the counts of acini/TDLU in categories (1 ≤ 10, 2 = 11–20, 3 = 21–30, 4 = 31–50, and 5 > 50) were recorded. Panel B shows three representative TDLUs for which the longest TDLU span was measured in microns using a digital ruler. The black arrow indicates a representative acinus
Using normal breast tissues donated by volunteer participants in the Susan G. Komen Tissue Bank at the Indiana University Simon Cancer Center (KTB), we developed standardized, reproducible metrics for evaluating TDLU involution using digital images and demonstrated associations of these measures with multiple breast cancer risk factors [11, 16–20]. Acini counts are difficult to quantitate visually and image analysis tools have not yet been validated; therefore, to approximate the size and acini content within TDLUs, we recently developed TDLU span measures which are highly correlated with acini counts [4]. In addition to the acini content within TDLUs, higher TDLU counts per unit area of tissue irrespective of their acini content have been hypothesized to represent a marker of increased breast cancer risk [11, 13, 14, 21], especially after menopause [14]; yet, whether such measures are associated with subsequent breast cancer risk has not been evaluated. Here we assessed whether standardized measures of TDLU involution, specifically, TDLU counts/mm2, acini content, and TDLU span measures are associated with risk of a future diagnosis of ductal carcinoma in situ (DCIS) or invasive breast cancer in a nested comparison within the Mayo BBD cohort [15].
Methods
Study population
We utilized a previously designed nested case–control study within the Mayo BBD cohort, which includes women (N = 9376) who had an open breast biopsy between January 1, 1967 and December 31, 1991 with benign findings: medical record review for future breast cancer diagnoses was obtained through January 2014, for a median of 16.9 years [9]. Risk factor information and subsequent breast pathology diagnoses were obtained from a study-specific questionnaire which was sent after benign biopsy diagnosis and medical records [9]. All necessary consent from any patients involved in the study, including consent to participate in the study, was obtained. All patient contact materials and procedures were reviewed and approved by the Mayo Clinic Institutional Review Board.
As previously described, cases were randomly selected from 823 cases stratified by 5-year categories of year of benign biopsy to represent the entire spectrum of the cohort. One hundred cases were matched on age and year of benign biopsy with two controls [15]. Adequate tissue for assessment was available for 99 cases and 145 controls.
Risk factor information
Risk factors were captured on the questionnaire or from medical records, as previously described [9]. Family history of breast cancer was categorized as (1) strong, if at least one first-degree relative was diagnosed with breast cancer before the age of 50 years or if two or more relatives had breast cancer, with at least one being a first-degree relative; (2) weak, if any lesser degree of family history of breast cancer; or (3) none. To clarify, the family history would not be considered strong if only a first-degree relative was above age 50 years, unless there was a second affected relative. Other exposures examined were body mass index (BMI, kg/m2) at biopsy (< 25, 25–29, 30+); age at menarche (<12, 13, 14, 15 years); age at first live birth (<20, 20–24, 25–29, 30+ years); parity (yes/no); breastfeeding (ever/never); use of menopausal hormones (MH) (ever/never); and type of BBD (nonproliferative, proliferative without atypia, and proliferative with atypia).
Histologic assessment of TDLU involution
Qualitative measures of TDLU involution were assessed microscopically by a pathologist (DWV), as previously described [7]. Briefly, samples containing one or more normal TDLUs were assessed for the degree of involution classified as none/mildly involuted (0–24 %), partially involuted (25–74 %), or completely involuted (≥75 %).
For standardized measures of TDLU involution, stained H&E tissue sections were digitized at 20X using a Hamamatsu NanoZoomer 2.0HT (Hamamatsu, Bridgewater NJ), and managed for web-based viewing and annotation with Digital Image Hub software (Slidepath/Leica, Dublin, Ireland), as previously described [11]. Only TDLUs that appeared entirely normal or showed focal benign changes (duct dilatation, metaplasia, and hyperplasia) were assessed. TDLUs were not considered normal if more than half the acini were dilated 2–3 times the normal diameter or if there were metaplastic changes involving more than half the acini. TDLUs showing ductal hyperplasia, defined as ducts or acini lined by more than a single epithelial cell layer, were not considered normal and therefore not assessed for involution analysis. In addition, the numbers of TDLUs with proliferative changes, and therefore not considered normal or suitable for assessment of involution, were recorded.
Images of sections were reviewed masked to other data to estimate percentage of fat (0–25, 26–50, 51–75, 76–100 %) and to enumerate the total number of normal TDLUs (as defined above). Up to ten normal TDLUs were reviewed sequentially to assess (1) TDLU span, measured with an electronic ruler (microns) and (2) number of acini per TDLU (1: ≤10; 2: 11–20; 3: 21–30; 4: 31–50; 5: 51+, Fig. 1). To determine standardized number of TDLUs, total tissue area was measured using the lasso drawing function in Slidepath to outline the perimeter of the tissue (in mm2). Prior reports have found that assessment of at least six TDLUs per section per patient provides stable representative measures of TDLU characteristics [15, 22, 23]. For samples with multiple tissue pieces, the area was summed across the number of tissues represented on the H&E image. For acini counts/TDLU and TDLU span measures, we used the median of the values obtained across the multiple TDLUs measured for each woman. More than 75 % of cases and controls had ten TDLUs reviewed.
Statistical analysis
Spearman correlations between measures of involution, such as TDLU counts/100 mm2 (referred to as TDLU counts), median TDLU span, median category of acini counts/TDLU, as well as qualitative TDLU involution assessment (none, partial, and complete involution), were compared. Quartile levels of standardized TDLU involution measures were created based on the distribution in the controls to determine associations with the outcome of breast cancer risk. To identify potential confounders associated with TDLU measures, we fitted ordinal logistic regression models using quartile levels of TDLU involution measures as the outcome and risk factor variables as the independent predictors (p < 0.05). Multivariable conditional logistic regression models were used to estimate odd ratios (ORs) and 95 % confidence intervals (CI) to determine associations between TDLU metrics and breast cancer risk, adjusted for significantly (p < 0.05) associated risk factors for breast cancer, which were family history of breast cancer, MH use, and BBD severity. Subjects missing data on covariates were considered a separate group and retained in all models. Analyses were performed using SAS V9.3.
Results
Patient characteristics
The median age of BBD diagnosis for cases and controls was 51 years of age. Although not statistically significant, cases were more likely to have younger ages at menarche ≤12 (55 vs. 37 %, p = 0.095), be current MH users (31 vs. 21 %, p = 0.098), have a strong family history of breast cancer (14 vs. 8 % p = 0.064), and have proliferative disease with atypia (11 vs. 6 %, p = 0.159).
Relationships between standardized TDLU involution measures and demographic data
TDLU counts were significantly and positively correlated with TDLU span (Spearman rho = 0.27, p < 0.0001) and with acini/TDLU (Spearman rho = 0.37, p < 0.0001); however, TDLU span and acini counts/TDLU were more strongly correlated (Spearman rho = 0.66, p < 0.0001). All standardized TDLU measures were significantly and inversely correlated with qualitative TDLU involution assessment: TDLU counts (Spearman rho = −0.39, p < .0001), TDLU span (Spearman rho = −0.32, p < 0.0001), and acini/TDLU (Spearman rho = −0.31, p < 0.0001), consistent with studies using normal breast tissues donated by volunteers [11].
Younger age at biopsy diagnosis and lower percentage of fat on digital images of biopsies diagnosed as BBD were significantly associated with higher TDLU counts and larger TDLU spans (Supplemental Table 1). In addition, having given birth, younger age at first birth and lower BMI were significantly associated with higher TDLU counts (Supplemental Table 1). Although based on only 14 in situ cases (DCIS), analysis by case status showed significantly higher median number of TDLU counts in DCIS compared to invasive cases (53 vs. 24 p = 0.005, Supplemental Table 2). Risk factors, including age at BBD, age at DCIS/invasive diagnosis, did not differ significantly between DCIS vs. invasive cancer, and these were combined in a single endpoint (Supplemental Table 3).
Measures of TDLU involution by case status are presented in Table 1. No statistically significant difference could be seen between cases and controls for qualitative measures of involution. In contrast, compared to controls, women who subsequently developed breast cancer had statistically significant higher TDLU counts per 100 mm2 area (median of 28 vs. 20, p = 0.029), a nonstatistically significant increase in TDLU spans (median of 300 vs. 267 microns, p = 0.136), and categorical measures of acini/TDLU (median acini category 2.5 (approximately 25 acini/TDLU), 11 vs. 5 %, p = 0.134). Median and IQR of quantitative measures of TDLU involution by qualitative measures (none, partial, and complete) are presented in Supplemental Table 4 and showed, as expected, that those classified as completely involuted had many fewer TDLUs, smaller TDLU spans, and fewer acini counts/TDLU as compared with those classified as partial or complete.
Table 1.
Standardized TDLU involution assessment by outcome status
| TDLU involution measures | Cases N = 99 | Controls N = 145 | p |
|---|---|---|---|
| Qualitative TDLU involution, N (%) | |||
| Complete (>75 %) | 12 (11) | 17 (13) | |
| Partial (1–74 %) | 67 (68) | 91 (67) | |
| None | 20 (21) | 28 (20) | 0.972** |
| Quantitative measures of TDLU involution, Median (IQR) | |||
| Standardized number of TDLUs/100 mm2 | 28 (12–51) | 20 (7–35) | 0.029* |
| Median TDLU span (microns) | 300 (221–387) | 267 (201–362) | 0.136* |
| Median Acini/TDLU in categories for up to 10 sequential TDLUs N (%) | 0.134** | ||
| 1 | 50 (51) | 90 (64) | |
| 1.5 | 10 (10) | 14 (10) | |
| 2 | 27 (23) | 30 (21) | |
| 2.5 | 11 (11) | 7 (5) |
The categories of acini counts/TDLU were recorded as: 1 ≤ 10, 2 = 11–20, 3 = 21–30, 4 = 31–50, and 5 > 50
Kruskal-Wallis test
χ2 test
Exact test
TDLU involution assessment and breast cancer risk
Unadjusted and multivariable associations of TDLU involution assessment and breast cancer are shown in Table 2. Compared with women in the lowest quartile of TDLU counts, women in the highest quartile had significantly elevated risk (OR = 2.41, 95 %CI 1.01–5.76, p-trend = 0.026), and estimates were similar when further adjusted for potential confounding factors. Compared to women in the lowest quartile of median TDLU span, women in the highest quartile were at a nonsignificantly 65 % increased risk (OR = 1.65, 95 %CI 0.75–3.62, p-trend = 0.212); however, increasing quartile levels of median TDLU span measures were significantly associated with risk when further adjusted for covariates (OR = 2.83, 95 %CI 1.13–7.06, p-trend = 0.032). Findings for median acini counts/TDLU were similar to TDLU span and statistically significantly associated with the risk of subsequent breast cancer diagnosis (Table 2).
Table 2.
Standardized TDLU involution assessment and subsequent breast cancer risk (invasive or in situ) among women with benign breast disease
| Cases | Controls | Unadjusted Model |
Multivariable Model |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95 %CI | P-trend | OR | 95 %CI | P-trend | |||||
| #TDLUs/100 mm2 | N | N | ||||||||
| Q1 | 19 | 34 | 1.00 | 1.00 | ||||||
| Q2 | 17 | 36 | 0.87 | 0.38 | 1.99 | 0.71 | 0.28 | 1.76 | ||
| Q3 | 23 | 36 | 1.52 | 0.64 | 3.57 | 1.50 | 0.57 | 3.93 | ||
| Q4 | 40 | 36 | 2.41 | 1.01 | 5.76 | 0.026 | 2.44 | 0.96 | 6.19 | 0.023 |
| Median TDLU span microns | ||||||||||
| Q1 | 18 | 35 | 1.00 | 1.00 | ||||||
| Q2 | 23 | 36 | 1.11 | 0.49 | 2.48 | 1.56 | 0.62 | 3.91 | ||
| Q3 | 23 | 36 | 1.19 | 0.52 | 2.75 | 1.59 | 0.62 | 4.06 | ||
| Q4 | 34 | 35 | 1.65 | 0.75 | 3.62 | 0.212 | 2.83 | 1.13 | 7.06 | 0.032 |
| Median category of acini/TDLU | ||||||||||
| 1 | 50 | 90 | 1.00 | 1.00 | ||||||
| 1.5 | 10 | 14 | 0.97 | 0.37 | 2.56 | 1.40 | 0.48 | 4.04 | ||
| 2 | 27 | 30 | 1.54 | 0.77 | 3.08 | 1.91 | 0.87 | 4.19 | ||
| 2.5 | 11 | 7 | 2.50 | 0.88 | 7.13 | 0.081 | 3.40 | 1.03 | 11.17 | 0.032 |
The categories of acini counts/TDLU were recorded as: 1 ≤ 10, 2 = 11–20, 3 = 21–30, 4 = 31–50, and 5 > 50. The median category of acini count was 2.5 (approximately 25 acini/TDLU)
The cutoffs for TDLU count quartiles were: 1: ≤7.11, 2: ≤19.35, 3: ≤35.46, 4: >35.46
The cutoffs for TDLU span quartiles were: 1: ≤ 199.5, 2: ≤ 266.5, 3: ≤363.5, 4: >363.5
ORs and 95 % CI estimated using conditional logistic regression model. Multivariable models adjusted for MH use, family history of breast cancer, and BBD histology
Breast cancer risk associations for participants cross-classified by several TDLU metrics (above or below medians) are shown in Table 3. Women above the median for both TDLU span and TDLU counts were at increased risk compared with women below the median for all of these characteristics (OR = 3.75; 95 %CI 1.40–10.00, p-trend = 0.008). Associations with risk based on cross-classification by TDLU counts and acini counts were similar to associations seen with TDLU span measures. There were too few cases to reliably estimate the risk for those subjects with atypical hyperplasia (ADH) or subjects with proliferative diseases without atypia (Supplemental Table 1). We did however remove those with a diagnosis of ADH, which is presented in Table 3 and show that even among lower risk BBD lesions, TDLU involution measures show a significant association with risk.
Table 3.
Cross classification of TDLU counts and TDLU span or acini counts/TDLU and breast cancer risk among women with benign breast disease
| Cases | Controls | Unadjusted |
Multivariable |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| N | N | OR | 95 %CI | P-trend | OR | 95 %CI | P-trend | ||||
| In situ and invasive cases | |||||||||||
| TDLUs | Span | ||||||||||
| ≤Median | ≤Median | 19 | 43 | 1.00 | 1.00 | ||||||
| ≤Median | >Median | 16 | 27 | 1.37 | 0.58 | 3.25 | 1.71 | 0.68 | 4.36 | ||
| >Median | ≤Median | 22 | 28 | 2.18 | 0.89 | 5.36 | 2.38 | 0.90 | 6.30 | ||
| >Median | >Median | 41 | 44 | 2.50 | 1.07 | 5.83 | 0.026 | 3.75 | 1.40 | 10.00 | 0.008 |
| TDLUs | Acini category | ||||||||||
| ≤Median | ≤2 | 28 | 58 | 1.00 | 1.00 | ||||||
| Median | ≥2 | 7 | 11 | 1.54 | 0.49 | 4.91 | 1.56 | 0.12 | 5.74 | ||
| >Median | ≤2 | 32 | 46 | 1.89 | 0.90 | 4.01 | 2.05 | 0.92 | 4.59 | ||
| >Median | ≥2 | 31 | 26 | 3.07 | 1.27 | 7.45 | 0.014 | 3.75 | 1.41 | 10.01 | 0.009 |
| Excluding subjects with BBD classified as proliferative disease with atypia | |||||||||||
| TDLUs | Span | ||||||||||
| ≤Median | ≤Median | 19 | 40 | 1.00 | 1.00 | ||||||
| ≤Median | >Median | 14 | 24 | 1.44 | 0.53 | 3.86 | 1.74 | 0.61 | 4.90 | ||
| >Median | ≤Median | 16 | 26 | 1.67 | 0.62 | 4.57 | 2.22 | 0.78 | 6.34 | ||
| >Median | >Median | 38 | 44 | 2.44 | 0.98 | 6.05 | 0.054 | 3.11 | 1.14 | 8.53 | 0.027 |
| TDLUs | Acini | ||||||||||
| ≤Median | ≤2 | 27 | 54 | 1.00 | 1.00 | ||||||
| ≤Median | ≥2 | 6 | 9 | 2.07 | 0.50 | 8.51 | 1.92 | 0.40 | 9.20 | ||
| >Median | ≤2 | 25 | 45 | 1.61 | 0.70 | 3.74 | 1.81 | 0.76 | 4.31 | ||
| >Median | ≥2 | 29 | 25 | 3.46 | 1.28 | 9.31 | 0.028 | 3.73 | 1.35 | 10.30 | 0.018 |
| Restricted to cases with a subsequent diagnosis of invasive cancer only | |||||||||||
| TDLUs | Span | ||||||||||
| ≤Median | ≤Median | 16 | 37 | 1.00 | 1.00 | ||||||
| ≤Median | >Median | 15 | 22 | 1.51 | 0.61 | 3.76 | 1.66 | 0.62 | 4.47 | ||
| >Median | ≤Median | 18 | 24 | 2.12 | 0.82 | 5.46 | 1.96 | 0.69 | 5.63 | ||
| >Median | > Median | 30 | 40 | 1.93 | 0.76 | 4.89 | 0.15 | 2.20 | 0.76 | 6.38 | 0.15 |
| TDLUs | Acini category | ||||||||||
| ≤Median | ≤2 | 24 | 51 | 1.00 | 1.00 | ||||||
| ≤Median | >2 | 7 | 8 | 2.00 | 0.59 | 6.78 | 1.76 | 0.46 | 6.77 | ||
| >Median | ≤2 | 25 | 42 | 1.63 | 0.74 | 3.62 | 1.48 | 0.62 | 3.53 | ||
| >Median | >2 | 23 | 22 | 2.73 | 1.05 | 7.08 | 0.05 | 2.70 | 0.94 | 7.70 | 0.10 |
The cutoff for TDLU count median was: ≤19.35
The cutoff for TDLU span median was: ≤266.5
Multivariable model adjusted for family history of breast cancer, MH use and BBD histology
To determine if the number of TDLUs evaluated influences associations with risk, we performed sensitivity analysis restricted to only those subjects with five or more TDLUs evaluated. Associations with subsequent breast cancer risk remained significant, where those above the median of TDLU counts and span measures had OR = 7.06 (95 % CI 1.75–28.46) compared to those below the median for TDLU counts and span measures. Similarly, results for combinatorial metrics for median acini and TDLU counts, where those above the median of TDLU counts and acini measures, had OR = 12.89 (95 % CI 2.58–64.41) compared to those below the median for TDLU counts and span measures. Restricting analyses to invasive cases or stratification of data by time interval between BBD and cancer diagnoses (0–5, 6–10, or >10 years) yielded similar results. Tests for interactions between TDLU measures and time since diagnosis, MH use, family history of breast cancer, and BBD histology were not significant (pvalues ≥0.20).
Discussion
Studies employing visual qualitative assessment of TDLU involution among women biopsied for BBD have demonstrated associations between decreased involution and increased risk of developing breast cancer [7, 8, 15]. Using standardized measures of TDLU involution, we found that women with the lowest levels of TDLU involution, as reflected in higher TDLU counts, and measures of acini content noted by larger TDLU span (which approximate acini content of TDLUs) or visually assessed acini counts/TDLU, had approximately a doubling of risk of developing breast cancer. Further, cross-classification using TDLU counts and either TDLU span or acini number per TDLU showed even higher risk. Thus, we propose that further evaluation of these metrics in clinical studies assessing risk of breast cancer development in additional cohorts is warranted.
Using quantitative as compared with qualitative measures of TDLU involution could enable the development of potentially clinically applicable markers of breast cancer risk among women undergoing benign biopsies [7]. In this analysis, higher TDLU counts/mm2, acini counts per TDLU, and TDLU span, were associated with risk of developing breast cancer, and a combinatorial metric combining TDLU counts and acini content seemed to demonstrate a stronger association. We hypothesize that the latter combination may reflect the at-risk epithelial content of the breast. As found previously, TDLU counts and specific features of TDLUs were only weakly correlated [11]. This could suggest that separate mechanisms control these two features of involution-reduced size of individual TDLUs and decrease in numbers of TDLUs; for example, TDLU counts may reflect differences in baseline breast development. Alternatively, involution may occur in a dynamic continuum in which individual TDLUs undergo simplification and obsolescence at differing rates.
Prior analyses demonstrating associations between breast cancer risk factors and the TDLU involution metrics evaluated here support that these are biologically plausible markers of risk [11, 16, 17]. Levels of TDLU involution increase with age and menopause, paralleling the slower rise in breast cancer incidence rates around age 50 years, a surrogate for menopause [24]. Further, mammographic density, a strong breast cancer risk factor, is also associated with levels of TDLU involution [12, 25–27], and data suggest that IGF signaling may be interrelated with both of these characteristics [16], [28]. These data support further studies using standardized measures of TDLU involution and mammographic density, which might enhance risk assessment among women with BBD.
Degrees of TDLU involution have been consistently shown to be lower among parous women and those with short intervals since childbirth [7, 11], suggesting that pregnancy-associated proliferation may possibly relate to the possible transient increase in cancer risk after a delivery. However, TDLU involution has not informed questions about the protective effect of pregnancy and breast cancer diagnosed at older ages, supporting a role for other mechanisms, including molecular changes, which may be related to the protective effect of parity [29, 30]. Additionally, data suggest that involution is inversely related to circulating concentrations of factors associated with elevated breast cancer risk, such as prolactin, testosterone, and postmenopausal estrogens, as well as use of menopausal hormone therapy [11, 16, 17]. Exposures such as smoking, which may reduce estrogenic effects, have also shown an association with increased TDLU involution [4].
The biologic mechanisms underlying age-related TDLU involution remain unknown. Studies of postpartum involution in animal models have been related to complex signaling pathways, including cytokine signaling (such as FAS-L, interleukins 6 and 10, and TGF-beta 3), apoptosis (including BAX and BCL-2), inflammatory process (including COX-2), and STAT proteins [31]. Ultimately, understanding the mechanisms that govern the distinctive processes of postweaning involution and age-related involution may have value for breast cancer risk reduction [32], if TDLU measurement can be shown to represent both a marker of cancer risk and a biological intermediate. Further, biologic processes related to postpartum involution may be distinct from those involved in age-related involution, and understanding these mechanisms could help identify potentially relevant pharmacological targets for chemoprevention.
In our analysis, higher TDLU counts were associated with increased breast cancer risk independent of family history of breast cancer, parity, and MH use, possibly suggesting that involution may be downstream of these factors on a common etiological pathway. The magnitude of the associations for number of acini per TDLU and TDLU span showed more variable relationships with breast cancer risk in multivariate models; however, the reasons for these findings are unclear.
Strengths of this study are the use of a nested case–control study design to determine associations with subsequent breast cancer risk and standardized measures of TDLU involution that have been shown to have good intraand interobserver reproducibility. Previous data suggest involution takes place at a relatively uniform rate throughout the breast tissue [22]; therefore, associations are likely robust despite review of only a portion of tissue from the entire breast; sampling error would likely bias results to the null. The generalizability of our findings is limited to women with BBD who have undergone an excisional biopsy. Sampling for histopathology of tissues was based on pathologists’ gross visual inspection of tissue, suggesting that oversampling of nonfatty areas is a potential issue. We did evaluate models adjusted for percentage fat in the tissue sections, and this did not substantially change associations observed for breast cancer risk suggesting associations are robust. Additional limitations include a relatively small sample size and lack of tumor marker data, particularly estrogen receptor (ER) expression on subsequently diagnosed tumors. A prior case–case analysis found that among women less than 55 years, levels of TDLU involution are lower for ER-negative breast cancers compared with tumors that express hormone receptors [23].
In summary, these data show that the lack of TDLU involution, as measured by increased standardized number of TDLU counts, larger median TDLU spans, higher categories of acini counts/TDLU, and cross-classification using these measures, is associated with increased breast cancer risk among women with BBD. This work extends prior research in this cohort linking decreased TDLU involution to increased risk of developing breast cancer [7, 15]. We note that several groups have developed automated methods to assess TDLU histology and to estimate levels of involution [15, 17, 18, 33, 34] that agree well with visual assessment. Hence, we view visual assessment of involution levels as a temporary approach to advance this area as automated algorithms are developed. Future studies are warranted to identify determinants of TDLU involution and to assess its role as a risk marker or potential intermediate endpoint in different clinical contexts. Given that there are approximately one million women diagnosed annually with BBD in the US, visual assessment of TDLU numbers and acini content, which are highly reproducible between pathologists, could help identify women at high risk for subsequent breast cancer.
Supplementary Material
Acknowledgments
JDF would like to thank Montserrat Garcia-Closas for editorial comments on this work.
Funding suppport This research was supported by Mayo Clinic Breast Specialized Programs of Research Excellence Grant NCI CA116201 (D.W.V., D.C.R., and L.C.H.), the Jimmy V Foundation (D.C.R. and L.C.H.), and ROI CA132879 (MHF). This research was supported in part by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research and Division of Cancer Epidemiology and Genetics.
Abbreviations
- CI
95 % confidence intervals
- BBD
Benign breast disease
- BMI
Body mass index
- ER
Estrogen receptor
- MH
Menopausal hormones
- ORs
Odds ratios
- RR
Relative risk
- KTB
Susan G. Komen Tissue Bank at the Indiana University Simon Cancer Center
- TDLU
Terminal duct-lobular unit
- DCIS
Ductal carcinoma in situ
Footnotes
Electronic supplementary material The online version of this article (doi:10.1007/s10549-016-3908-7) contains supplementary material, which is available to authorized users.
Author contribution JDF conceived the study, led its design and execution of the study, and interpretation, and drafted the manuscript. RMF provided expertise for the statistical analysis and interpretation of data, and drafted the manuscript. LAB provided expertise for the breast cancer epidemiology and drafted the manuscript. MMP performed statistical analysis and drafted the manuscript. ACD, DR, LCH, MHF, MLM, and DV conceived the study, and participated in its design and coordination and critical revision of the manuscript for intellectual content. DV, SMH and MES provided the expertise for pathology. DP and SMH helped to obtain pathologic annotation, coordination of design and execution of study, and participated in revision of the manuscript. MES conceived the study and study execution, interpreted the data, and drafted the manuscript. All authors read and approved the final manuscript.
Compliance with ethical standards
Conflict of interest The authors declare that they have no competing interests.
References
- 1.Allison KH, Abraham LA, Weaver DL, Tosteson AN, Nelson HD, Onega T, Geller BM, Kerlikowske K, Carney PA, Ichikawa LE, Buist DS, Elmore JG. Trends in breast biopsy pathology diagnoses among women undergoing mammography in the United States: a report from the Breast Cancer Surveillance Consortium. Cancer. 2015 doi: 10.1002/cncr.29199. doi:10.1002/cncr.29199. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Dyrstad SW, Yan Y, Fowler AM, Colditz GA. Breast cancer risk associated with benign breast disease: systematic review and meta-analysis. Breast Cancer Res Treat. 2015;149(3):569–575. doi: 10.1007/s10549-014-3254-6. doi:10.1007/s10549-014-3254-6. [DOI] [PubMed] [Google Scholar]
- 3.Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, Schairer C, Mulvihill JJ. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst. 1989;81(24):1879–1886. doi: 10.1093/jnci/81.24.1879. [DOI] [PubMed] [Google Scholar]
- 4.Pankratz VS, Degnim AC, Frank RD, Frost MH, Visscher DW, Vierkant RA, Hieken TJ, Ghosh K, Tarabishy Y, Vachon CM, Radisky DC, Hartmann LC. Model for individualized prediction of breast cancer risk after a benign breast biopsy. J Clin Oncol. 2015 doi: 10.1200/JCO.2014.55.4865. doi:10.1200/JCO.2014.55.4865. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Nelson HD, Tyne K, Naik A, Bougatsos C, Chan B, Nygren P, Humphrey L. Screening for breast cancer: an update for the U.S. Preventive Services Task Force. Ann Intern Med. 2009;151(10):727–737. doi: 10.1059/0003-4819-151-10-200911170-00009. doi:10.7326/0003-4819-151-10-200911170-00009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Cummings SR, Tice JA, Bauer S, Browner WS, Cuzick J, Ziv E, Vogel V, Shepherd J, Vachon C, Smith-Bindman R, Kerlikowske K. Prevention of breast cancer in postmenopausal women: approaches to estimating and reducing risk. J Natl Cancer Inst. 2009;101(6):384–398. doi: 10.1093/jnci/djp018. doi:10.1093/jnci/djp018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Milanese TR, Hartmann LC, Sellers TA, Frost MH, Vierkant RA, Maloney SD, Pankratz VS, Degnim AC, Vachon CM, Reynolds CA, Thompson RA, Melton LJ, 3rd, Goode EL, Visscher DW. Age-related lobular involution and risk of breast cancer. J Natl Cancer Inst. 2006;98(22):1600–1607. doi: 10.1093/jnci/djj439. doi:10.1093/jnci/djj439. [DOI] [PubMed] [Google Scholar]
- 8.Baer HJ, Collins LC, Connolly JL, Colditz GA, Schnitt SJ, Tamimi RM. Lobule type and subsequent breast cancer risk: results from the Nurses’ Health Studies. Cancer. 2009;115(7):1404–1411. doi: 10.1002/cncr.24167. doi:10.1002/cncr.24167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hartmann LC, Sellers TA, Frost MH, Lingle WL, Degnim AC, Ghosh K, Vierkant RA, Maloney SD, Pankratz VS, Hillman DW, Suman VJ, Johnson J, Blake C, Tlsty T, Vachon CM, Melton LJ, 3rd, Visscher DW. Benign breast disease and the risk of breast cancer. N Engl J Med. 2005;353(3):229–237. doi: 10.1056/NEJMoa044383. [DOI] [PubMed] [Google Scholar]
- 10.Wellings SR, Jensen HM. On the origin and progression of ductal carcinoma in the human breast. J Natl Cancer Inst. 1973;50(5):1111–1118. doi: 10.1093/jnci/50.5.1111. [DOI] [PubMed] [Google Scholar]
- 11.Figueroa JD, Pfeiffer RM, Patel DA, Linville L, Brinton LA, Gierach GL, Yang XR, Papathomas D, Visscher D, Mies C, Degnim AC, Anderson WF, Hewitt S, Khodr ZG, Clare SE, Storniolo AM, Sherman ME. Terminal duct lobular unit involution of the normal breast: implications for breast cancer etiology. J Natl Cancer Inst. 2014 doi: 10.1093/jnci/dju286. doi:10.1093/jnci/dju286. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ginsburg OM, Martin LJ, Boyd NF. Mammographic density, lobular involution, and risk of breast cancer. Br J Cancer. 2008;99(9):1369–1374. doi: 10.1038/sj.bjc.6604635. doi:10.1038/sj.bjc.6604635. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Henson DE, Tarone RE. Involution and the etiology of breast cancer. Cancer. 1994;74(1):424–429. doi: 10.1002/cncr.2820741330. [DOI] [PubMed] [Google Scholar]
- 14.Jensen HM. On the origin and progression of human breast cancer. Am J Obstet Gynecol. 1986;154(6):1280–1284. doi: 10.1016/0002-9378(86)90713-1. doi:10.1016/0002-9378(86)90713-1. [DOI] [PubMed] [Google Scholar]
- 15.McKian KP, Reynolds CA, Visscher DW, Nassar A, Radisky DC, Vierkant RA, Degnim AC, Boughey JC, Ghosh K, Anderson SS, Minot D, Caudill JL, Vachon CM, Frost MH, Pankratz VS, Hartmann LC. Novel breast tissue feature strongly associated with risk of breast cancer. J Clin Oncol. 2009;27(35):5893–5898. doi: 10.1200/JCO.2008.21.5079. doi:10.1200/jco.2008.21.5079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Horne HNSM, Pfeiffer RM, Figueroa JD, Khodr Z, Falk RT, Pollak M, Patel DA, Linville L, Papathomas D, Geller B, Vacek PM, Weaver DL, Chicoine R, Shepherd J, Mahmoudzadeh AP, Wang J, Fan B, Herschorn S, Hewitt SM, Brinton LA, Gierach GL. Circulating insulin-like growth factor-I, insulin-like growth factor binding protein-3 and terminal duct lobular unit involution of the breast. Cancer Res. 2016 doi: 10.1186/s13058-016-0678-4. doi:10.1186/s13058-016-0678-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Khodr ZG, Sherman ME, Pfeiffer RM, Gierach GL, Brinton LA, Falk RT, Patel DA, Linville LM, Papathomas D, Clare SE, Visscher DW, Mies C, Hewitt SM, Storniolo AM, Rosebrock A, Caban JJ, Figueroa JD. Circulating sex hormones and terminal duct lobular unit involution of the normal breast. Cancer Epidemiol Biomarkers Prev. 2014;23(12):2765–2773. doi: 10.1158/1055-9965.EPI-14-0667. doi:10.1158/1055-9965. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Gierach GL, Patel DA, Pfeiffer RM, Figueroa JD, Linville L, Papathomas D, Johnson JM, Chicoine RE, Herschorn SD, Shepherd JA, Wang J, Malkov S, Vacek PM, Weaver DL, Fan B, Mahmoudzadeh AP, Palakal M, Xiang J, Oh H, Horne HN, Sprague BL, Hewitt SM, Brinton LA, Sherman ME. Relationship of terminal duct lobular unit involution of the breast with area and volume mammographic densities. Cancer prevention research. 2016;9(2):149–158. doi: 10.1158/1940-6207.CAPR-15-0282. doi:10.1158/1940-6207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Horne HN, Sherman ME, Pfeiffer RM, Figueroa JD, Khodr ZG, Falk RT, Pollak M, Patel DA, Palakal MM, Linville L, Papathomas D, Geller B, Vacek PM, Weaver DL, Chicoine R, Shepherd J, Mahmoudzadeh AP, Wang J, Fan B, Malkov S, Herschorn S, Hewitt SM, Brinton LA, Gierach GL. Circulating insulin-like growth factor-I, insulin-like growth factor binding protein-3 and terminal duct lobular unit involution of the breast: a cross-sectional study of women with benign breast disease. Breast Cancer Res. 2016;18(1):24. doi: 10.1186/s13058-016-0678-4. doi:10.1186/s13058-016-0678-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Oh H, Khodr ZG, Sherman ME, Palakal M, Pfeiffer RM, Linville L, Geller BM, Vacek PM, Weaver DL, Chicoine RE, Falk RT, Horne HN, Papathomas D, Patel DA, Xiang J, Xu X, Veenstra T, Hewitt SM, Shepherd JA, Brinton LA, Figueroa JD, Gierach GL. Relation of Serum Estrogen Metabolites with Terminal Duct Lobular Unit Involution Among Women Undergoing Diagnostic Image-Guided Breast Biopsy. Horm Cancer. 2016 doi: 10.1007/s12672-016-0265-2. doi:10. 1007/s12672-016-0265-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hutson SW, Cowen PN, Bird CC. Morphometric studies of age related changes in normal human breast and their significance for evolution of mammary cancer. J Clin Pathol. 1985;38(3):281–287. doi: 10.1136/jcp.38.3.281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Vierkant RA, Hartmann LC, Pankratz VS, Anderson SS, Radisky D, Frost MH, Vachon CM, Ghosh K, Distad TJ, Degnim AC, Reynolds CA. Lobular involution: localized phenomenon or field effect? Breast Cancer Res Treat. 2009;117(1):193–196. doi: 10.1007/s10549-008-0082-6. doi:10. 1007/s10549-008-0082-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Yang XR, Figueroa JD, Falk RT, Zhang H, Pfeiffer RM, Hewitt SM, Lissowska J, Peplonska B, Brinton L, Garcia-Closas M, Sherman ME. Analysis of Terminal Duct Lobular Unit (TDLU) Involution in Luminal A and Basal Breast Cancers. Breast Cancer Res. 2012;14(2):R64. doi: 10.1186/bcr3170. doi:10.1186/bcr3170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Pike MC, Krailo MD, Henderson BE, Casagrande JT, Hoel DG. ‘Hormonal’ risk factors, ‘breast tissue age’ and the age-incidence of breast cancer. Nature. 1983;303(5920):767–770. doi: 10.1038/303767a0. [DOI] [PubMed] [Google Scholar]
- 25.Ghosh K, Hartmann LC, Reynolds C, Visscher DW, Brandt KR, Vierkant RA, Scott CG, Radisky DC, Sellers TA, Pankratz VS, Vachon CM. Association between mammographic density and age-related lobular involution of the breast. J Clin Oncol. 2010;28(13):2207–2212. doi: 10.1200/JCO.2009.23.4120. doi:10.1200/jco.2009.23.4120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ghosh K, Vachon CM, Pankratz VS, Vierkant RA, Anderson SS, Brandt KR, Visscher DW, Reynolds C, Frost MH, Hartmann LC. Independent association of lobular involution and mammographic breast density with breast cancer risk. J Natl Cancer Inst. 2010;102(22):1716–1723. doi: 10.1093/jnci/djq414. doi:10.1093/jnci/djq414. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Gierach GL, Brinton LA, Sherman ME. Lobular involution, mammographic density, and breast cancer risk: visualizing the future? J Natl Cancer Inst. 2010;102(22):1685–1687. doi: 10.1093/jnci/djq433. doi:10.1093/jnci/djq433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Rice MS, Tamimi RM, Connolly JL, Collins LC, Shen D, Pollak MN, Rosner B, Hankinson SE, Tworoger SS. Insulin-like growth factor-1, insulin-like growth factor binding protein-3 and lobule type in the Nurses’ Health Study II. Breast Cancer Res. 2012;14(2):R44. doi: 10.1186/bcr3141. doi:10.1186/bcr3141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Russo J, Balogh G, Russo IH. Breast cancer prevention. Climacteric. 2007;10(Suppl 2):47–53. doi: 10.1080/13697130701606762. doi:10.1080/13697130701606762. [DOI] [PubMed] [Google Scholar]
- 30.Rotunno M, Sun X, Figueroa J, Sherman ME, Garcia-Closas M, Meltzer P, Williams T, Schneider SS, Jerry DJ, Yang XR, Troester MA. Parity-related molecular signatures and breast cancer subtypes by estrogen receptor status. Breast Cancer Res. 2014;16(1):R74. doi: 10.1186/bcr3689. doi:10.1186/bcr3689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Haricharan S, Li Y. STAT signaling in mammary gland differentiation, cell survival and tumorigenesis. Mol Cell Endocrinol. 2014;382(1):560–569. doi: 10.1016/j.mce.2013.03.014. doi:10.1016/j.mce.2013.03.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Radisky DC, Hartmann LC. Mammary involution and breast cancer risk: transgenic models and clinical studies. J Mammary Gland Biol Neoplasia. 2009;14(2):181–191. doi: 10.1007/s10911-009-9123-y. doi:10.1007/s10911-009-9123-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Rosebrock A, Caban JJ, Figueroa J, Gierach G, Linville L, Hewitt S, Sherman M. Quantitative Analysis of TDLUs using Adaptive Morphological Shape Techniques. Proc SPIE Int Soc Opt Eng. 2013 doi: 10.1117/12.2006619. doi:10.1117/12.2006619. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Sandhu R, Chollet-Hinton L, Kirk EL, Midkiff B, Troester MA. Digital histologic analysis reveals morphometric patterns of age-related involution in breast epithelium and stroma. Hum Pathol. 2016;48:60–68. doi: 10.1016/j.humpath.2015.09.031. doi:10.1016/j.humpath.2015.09.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.