Quantitative Mammographic Density Measurements and Molecular Subtypes in Chinese Women With Breast Cancer

Abstract

Background

Studies investigating associations between mammographic density (MD) and breast cancer subtypes have generated mixed results. We previously showed that having extremely dense breasts was associated with the human epidermal growth factor receptor-2 (HER2)-enriched subtype in Chinese breast cancer patients.

Methods

In this study, we reevaluated the MD-subtype association in 1549 Chinese breast cancer patients, using VolparaDensity software to obtain quantitative MD measures. All statistical tests were 2-sided.

Results

Compared with women with luminal A tumors, women with luminal B/HER2- (odds ratio [OR] = 1.20, 95% confidence interval [CI] = 1.04 to 1.38; P = .01), luminal B/HER2+ (OR = 1.22, 95% CI = 1.03 to 1.46; P = .03), and HER2-enriched tumors (OR = 1.30, 95% CI = 1.06 to 1.59; P = .01) had higher fibroglandular dense volume. These associations were stronger in patients with smaller tumors (<2 cm). In contrast, the triple-negative subtype was associated with lower nondense volume (OR = 0.82, 95% CI = 0.68 to 0.99; P = .04), and the association was only seen among older women (age 50 years or older).

Conclusion

Although biological mechanisms remain to be investigated, the associations for the HER2-enriched and luminal B subtypes with increasing MD may partially explain the higher prevalence of luminal B and HER2+ breast cancers previously reported in Asian women.

Epidemiological studies have shown that associations between breast cancer risk and risk factors vary by tumor pathology such as estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor-2 (HER2) status (1-3). Mammographic density (MD) is a strong risk factor for breast cancer (4); however, its association with intrinsic tumor subtypes remains elusive and inconsistently reported (5-12). Furthermore, less is known of the association of MD with clinicopathological factors in relation to racial and ethnic differences. Asian women tend to have denser breasts (13-17) but have a lower overall incidence of breast cancer (18), compared with Western populations. Additionally, Asian breast cancer is characterized by an earlier age at onset and higher prevalence of luminal B and HER2-enriched tumors (19-21). In our previous study of unscreened Chinese patients with breast cancer, we found that high MD, as measured by the Breast Imaging Reporting and Data System (BI-RADS), was associated with the HER2-enriched subtype (22). However, because BI-RADS is subjectively measured and prone to suboptimal reproducibility (23, 24), in the present study, we reevaluated the MD-subtype association using a quantitative (objective) measure of MD in a similarly unscreened sample of Chinese breast cancer patients.

Methods

Study Population

We analyzed 1549 invasive breast cancer cases diagnosed and treated at the National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College (CHCAMS), Beijing, China, between March 2016 and July 2017. Most patients were symptomatic, and less than 10% were detected through physical exams or screening. Eligible cases had complete immunohistochemical (IHC) marker status, quantitative MD assessments, and information on established risk factors. The project was approved by the CHCAMS ethics committee and was exempted from review by the Office for Human Research Protections at the National Institutes of Health because it did not involve interaction with human subjects and/or use of personal identifying information (exempt no. 11751).

Mammographic Density

Images from full-field digital diagnostic mammograms were retrospectively scored for MD by trained radiologists (YT, EL) using BI-RADS guidelines recommended by the American College of Radiology (5th edition). MD was classified into 4 categories: a = almost entirely fat; b = scattered fibroglandular dense; c = heterogeneously dense; and d = extremely dense. VolparaDensity software 5th edition (Volpara Health Technologies, Wellington, New Zealand), an automated quantitative density assessment tool, was used to assess total breast and fibroglandular volume (cm³) and Volpara density grade (VDG). Percent dense volume was calculated by dividing the fibroglandular volume by the total breast volume. Nondense volume was assessed as the absolute difference between the 2 measured volumes. VDG was based on preset cut-points from percent dense volume (VDG a = <3.5%, VDG b = ≥3.5% and <7.5%, VDG c = ≥7.5% and <15.5%, and VDG d = ≥15.5%), which were optimized based on the median visual BI-RADS readings from a reader dataset that included five Mammography Quality Standards Act–accredited US radiologists and showed high correlations with visual BI-RADS categories from American College of Radiology. VDG categories a and b were combined because of a small number of participants in VDG category a (n = 14). All MD readings used in analyses were from the unaffected contralateral breast.

Clinical Characteristics and Immunohistochemical Subtype

Clinical characteristics and IHC marker status were extracted from pathology reports. ER and PR tumor expressions were considered positive for IHC with at least 1% nuclear staining. HER2 expression was determined by IHC and fluorescence in situ hybridization (FISH). An IHC score of 3 or a FISH-positive test result was defined as HER2+. HER2 IHC-2+ cases without FISH data were considered negative. Ki-67 was considered high if 25% or more of the cells showed nuclear staining based on a previous meta-analysis showing Ki-67 score of 25% to provide the best survival discrimination (25). We used Ki-67 status (low and high) to discriminate luminal A and B and used tumor grade as a surrogate for patients with missing Ki-67 (23). Molecular subtypes were defined as follows: luminal A: ER+ or PR+, HER2-, and low Ki-67 or histologic grade (I or II); luminal B/HER2+: ER+ or PR+, and HER2+; luminal B/HER2-: ER+ or PR+, HER2-, and high Ki-67 or histologic grade (III); HER2-enriched: ER-, PR-, and HER2+; and triple-negative (TN): ER-, PR-, and HER2-. Similar results were obtained when using a different cutoff point to define Ki-67 positivity (20% vs 25%) (Supplementary Table 1, available online).

Statistical Analysis

Differences in the distribution of each risk factor and tumor feature by each Volpara density measurement were assessed using one-way analysis of variance for continuous variables or χ² tests for categorical variables. Multinomial logistic regression was used to test the association between molecular subtypes (outcome) and each quantitative MD measure (per standard deviation [SD] increase), adjusting for age, body mass index (BMI), menopausal status, and parity. We also assessed the associations in patients stratified by BMI and age at diagnosis (younger than 50 and 50 years or older) as a proxy for menopausal status due to missing data for menopausal status (260 patients with unknown status). We conducted a sensitivity analysis using actual menopausal status and observed similar results (Supplementary Table 2, available online). Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated by comparing each tumor subtype to the luminal A subtype. Adjusted means modeling was performed on square root–transformed MD measures. All P values were 2-sided, and P values less than .05 were considered as statistically significant. Analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC).

Results

Among the 1549 breast cancer patients, the mean age at diagnosis was 52.4  (SD = 11.2) years and the average BMI was 24.3  (SD = 3.7) kg/m². With increasing age and BMI, the percent dense volume decreased, and the total breast and nondense volumes increased (P < .05). Women with HER2+ status and high Ki-67 had higher fibroglandular dense and percent dense volume (P < .05) (Supplementary Table 3, available online).

In multivariable regression models, fibroglandular dense volume was associated with higher frequencies of the HER2-enriched (OR = 1.30, 95% CI = 1.06 to 1.59; P = .01), luminal B/HER2- (OR = 1.20, 95% CI = 1.04 to 1.38; P = .01), and luminal B/HER2+ (OR = 1.22, 95% CI = 1.03 to 1.46; P = .03) subtypes (Table 1). Similarly, higher VDG and BI-RADS density categories were positively associated with the HER2-enriched subtype (VDG = c + d: OR = 2.44, 95% CI = 1.36 to 4.37; P = .003; BI-RADS = c + d: OR = 1.80, 95% CI = 1.04 to 3.14; P = .04) (Tables 2 and 3). Additionally, HER2-enriched (adjusted mean = 56.8 cm³; P = .01), luminal B/HER2- (adjusted mean = 53.8 cm³; P = .02), and luminal B/HER2+ (adjusted mean = 54.4 cm³; P = .05) subtypes also showed statistically significantly higher fibroglandular dense volume than luminal A (adjusted mean = 49.8 cm³; Table 4). Interestingly, TN breast cancer (TNBC) patients presented with lower total breast volume, driven by lower nondense volume (OR = 0.82, 95% CI = 0.68 to 0.99; P = .04) (Table 1).

Table 1.

Associations between molecular subtype and Volpara quantitative density measurements among 1549 Chinese women with breast cancer

Molecular subtype	No.	Total breast volume			Fibroglandular dense volume			Nondense volume			Percent dense volume
		Mean (SD)	OR (95% CI)	P a	Mean (SD)	OR (95% CI)	P a	Mean (SD)	OR (95% CI)	P a	Mean (SD)	OR (95% CI)	P a
Luminal A	565	454.40 (224.82)	Referent	—	52.37 (29.32)	Referent	—	402.04 (218.26)	Referent	—	13.40 (7.43)	Referent	—
Luminal B/HER2-	487	471.41 (235.63)	1.05 (0.93 to 1.19)	.41	57.56 (35.75)	1.20 (1.04 to 1.38)	.01	413.85 (222.00)	1.03 (0.91 to 1.17)	.62	13.64 (7.03)	1.07 (0.92 to 1.25)	.40
Luminal B/HER2+	182	457.11 (234.46)	1.02 (0.86 to 1.21)	.81	61.39 (38.28)	1.22 (1.03 to 1.46)	.03	395.72 (222.32)	0.99 (0.83 to 1.19)	.95	15.05 (7.36)	1.17 (0.95 to 1.44)	.13
HER2-enriched	121	459.02 (196.21)	1.04 (0.85 to 1.27)	.72	59.48 (31.53)	1.30 (1.06 to 1.59)	.01	399.54 (183.50)	1.00 (0.81 to 1.23)	.99	13.83 (6.60)	1.10 (0.86 to 1.40)	.47
TN	194	428.30 (213.23)	0.84 (0.70 to 1.01)	.06	53.28 (36.79)	1.05 (0.86 to 1.27)	.65	375.02 (197.26)	0.82 (0.68 to 0.99)	.04	13.63 (6.94)	1.12 (0.91 to 1.37)	.30

^aModeling tumor subtype as the outcome variable and Volpara density per 1-unit standard deviation increase as the explanary variable, adjusting for age, body mass index, menopausal status, and parity. Age at menarche did not change the results after its addition to the model, therefore we elected to remove it from the final model. “—” is used instead of the P value for the reference category. CI = confidence interval; OR = odds ratio; TN = triple negative.

Table 2.

Associations between molecular subtype and Volpara density grade (VDG) among 1549 Chinese women with breast cancer

Molecular subtype	No.	VDG = a&b (n = 354)	VDG = c (n = 645)		VDG = d (n = 550)		VDG = c&d (n = 1195)
			OR (95% CI)	P a	OR (95% CI)	P a	OR (95% CI)	P a
Luminal A	565	Referent	Referent	—	Referent	—	Referent	—
Luminal B/HER2-	487	Referent	1.37 (0.98 to 1.91)	.06	1.25 (0.84 to 1.86)	.28	1.34 (0.97 to 1.86)	.08
Luminal B/HER2+	182	Referent	1.30 (0.79 to 2.14)	0.3	1.67 (0.95 to 2.92)	.08	1.40 (0.87 to 2.26)	.17
HER2-enriched	121	Referent	2.66 (1.48 to 4.78)	<.001	1.70 (0.84 to 3.44)	.14	2.44 (1.36 to 4.37)	.003
TN	194	Referent	1.47 (0.93 to 2.31)	.1	1.67 (0.98 to 2.86)	.06	1.51 (0.98 to 2.35)	.07

^aModeling tumor subtype as the outcome variable and VDG categories as the explanatory variable (a&b as the reference group); models were adjusted for age, body mass index, menopausal status, and parity. Missing value indicators were used in the place of missing values on covariates. VDG: a = 0-3.5, b = 3.6-7.5, c = 7.6-15.5, and d  ≥ 15.6 percent dense volume. Categories a&b were combined because of the small numbers of VDG category a (n = 14). “—” is used instead of the P-value for the reference category. CI = confidence interval; OR = odds ratio; TN = triple negative.

Table 3.

Associations between molecular subtype and BI-RADS categories among 1549 Chinese women with breast cancer

Molecular subtype	No.	BI-RADS = a&b (n = 369)	BI-RADS = c (n = 916)		BI-RADS = d (n = 264)		BI-RADS = c&d (n = 1180)
			OR (95% CI)	P a	OR (95% CI)	P a	OR (95% CI)	P a
Luminal A	565	Referent	Referent	—	Referent	—	Referent	—
Luminal B/HER2-	487	Referent	0.97 (0.71 to 1.34)	.87	1.30 (0.84 to 2.02)	.24	1.01 (0.74 to 1.38)	.95
Luminal B/HER2+	182	Referent	1.45 (0.90 to 2.33)	.13	1.31 (0.69 to 2.47)	.41	1.43 (0.89 to 2.30)	.14
HER2-enriched	121	Referent	1.80 (1.03 to 3.14)	.04	1.79 (0.84 to 3.81)	.13	1.80 (1.04 to 3.14)	.04
TN	194	Referent	1.28 (0.83 to 1.97)	.26	1.32 (0.72 to 2.42)	.37	1.28 (0.84 to 1.97)	.25

^aModeling tumor subtype as the outcome variable and BI-RADS categories as the explanatory variable (a&b as the reference group); models were adjusted for age, body mass index, menopausal status, and parity. Missing value indicators were used in the place of missing values on covariates. BI-RADS categories a&b were combined because of the small numbers of VDG category = a (n = 14). “—” is used instead of the P value for the reference category. BI-RADS = Breast Imaging Reporting and Data System; CI = confidence interval; OR = odds ratio; TN = triple negative.

Table 4.

Adjusted least square means for Volpara quantitative density and tumor subtype

Molecular subtype	No.	Total breast volume		Fibroglandular dense volume		Nondense volume			Percent dense volume
		Mean (SD)	P a ,b	Mean (SD)	P a ,b		Mean (SD)	P a .b	Mean (SD)	P a ,b
Luminal A	565	429.7 (0.05)	—	49.8 (0.01)	—		375.5 (0.04)	—	13.0 (0.001)	—
Luminal B/HER2-	487	441.0 (0.05)	.39	53.8 (0.01)	.02		383.2 (0.05)	.53	13.4 (0.001)	.24
Luminal B/HER2+	182	433.8 (0.14)	.82	54.4 (0.02)	.05		375.0 (0.14)	.97	13.8 (0.003)	.09
HER2-Enriched	121	443.1 (0.21)	.53	56.8 (0.03)	.01		383.1 (0.21)	.70	13.7 (0.01)	.22
TN	194	396.7 (0.13)	.05	49.7 (0.02)	.97		343.5 (0.13)	.04	13.6 (0.003)	.17

^aVolpara variables were square root–transformed in models. Results were back-transformed for ease of interpretation. “—” is used instead of the P value for the reference category. TN = triple negative.

^bModeling each quantitative mammographic density measurement as the outcome variable; models were adjusted for age, body mass index, parity, and menopausal status.

In analyses stratified by age, BMI, and tumor size, the associations of increasing fibroglandular dense volume with luminal B and HER2-enriched subtypes were only seen in women with normal BMI and smaller tumors (tumor size <2 cm), and the formal tests for interactions were statistically significant for tumor size (Table 5;  Supplementary Table 4, available online). The lower total and nondense breast volumes associated with TNBC were only seen among older women (Table 5), and the interactions between these associations and age were statistically significant (Supplementary Table 4, available online).

Table 5.

Associations between molecular subtype and Volpara density measurements stratified by age, BMI, and tumor size^a

Molecular subtype	No.	Total breast volume		Fibroglandular dense volume		Nondense volume		Percent dense volume
		OR (95% CI)	P b	OR (95% CI)	P b	OR (95% CI)	P b	OR (95% CI)	P b
By age
Among women younger than 50 y (n = 670)c
Luminal A	238	Referent	—	Referent		Referent	—	Referent	—
Luminal B/HER2-	221	1.25 (1.03 to 1.52)	.02	1.22 (1.01 to 1.49)	.04	1.23 (1.02 to 1.49)	.04	0.99 (0.81 to 1.20)	.89
Luminal B/HER2+	96	1.09 (0.85 to 1.41)	.51	1.18 (0.92 to 1.52)	.18	1.06 (0.82 to 1.38)	.64	1.01 (0.79 to 1.31)	.93
HER2-enriched	44	1.14 (0.81 to 1.59)	.47	1.19 (0.87 to 1.65)	.28	1.11 (0.79 to 1.57)	.55	0.93 (0.66 to 1.31)	.69
TN	71	1.09 (0.82 to 1.45)	.56	1.21 (0.92 to 1.58)	.17	1.06 (0.79 to 1.41)	.71	0.98 (0.74 to 1.30)	.87
Among women older than 50 y (n = 878)
Luminal A	326	Referent	—	Referent	—	Referent	—	Referent	—
Luminal B/HER2-	266	0.95 (0.81 to 1.12)	.57	1.08 (0.91 to 1.29)	.40	0.94 (0.80 to 1.11)	.50	1.07 (0.89 to 1.29)	.48
Luminal B/HER2+	86	1.03 (0.81 to 1.30)	.83	1.22 (0.96 to 1.54)	.10	1.01 (0.79 to 1.27)	.96	1.24 (0.97 to 1.59)	.08
HER2-enriched	77	0.98 (0.76 to 1.26)	.89	1.35 (1.07 to 1.70)	.01	0.95 (0.73 to 1.22)	.67	1.22 (0.94 to 1.58)	.13
TN	123	0.71 (0.56 to 0.90)	.004	0.88 (0.69 to 1.13)	.32	0.71 (0.56 to 0.90)	.004	1.24 (0.99 to 1.55)	.06
By BMI
Among women with underweight/normal BMId (n = 800)
Luminal A	305	Referent	—	Referent	—	Referent	—	Referent	—
Luminal B/HER2-	234	1.06 (0.89 to 1.27)	.50	1.36 (1.11 to 1.66)	.003	1.03 (0.86 to 1.23)	.79	1.22 (0.98 to 1.53)	.08
Luminal B/HER2+	91	1.03 (0.81 to 1.32)	.81	1.48 (1.15 to 1.89)	.002	0.97 (0.75 to 1.26)	.83	1.39 (1.03 to 1.87)	.03
HER2-enriched	71	1.04 (0.80 to 1.36)	.76	1.40 (1.06 to 1.86)	.02	1.00 (0.76 to 1.31)	.99	1.17 (0.84 to 1.65)	.36
TN	99	0.94 (0.74 to 1.21)	.64	1.05 (0.79 to 1.40)	.73	0.93 (0.73 to 1.20)	.59	0.95 (0.70 to 1.29)	.76
Among women with high BMId (n = 473)
Luminal A	158	Referent	—	Referent	—	Referent	—	Referent	—
Luminal B/HER2-	159	1.05 (0.84 to 1.32)	.67	1.04 (0.81 to 1.33)	.79	1.05 (0.84 to 1.32)	.68	0.90 (0.68 to 1.18)	.45
Luminal B/HER2+	59	1.03 (0.75 to 1.40)	.88	0.98 (0.70 to 1.37)	.91	1.03 (0.75 to 1.41)	.85	0.97 (0.68 to 1.39)	.88
HER2-enriched	29	1.31 (0.89 to 1.94)	.18	1.21 (0.84 to 1.75)	.30	1.30 (0.87 to 1.94)	.20	0.79 (0.48 to 1.32)	.37
TN	68	0.81 (0.59 to 1.10)	.18	0.96 (0.69 to 1.35)	.83	0.80 (0.58 to 1.09)	.16	1.16 (0.83 to 1.62)	.40
By tumor size
Among tumors <2 cm (n = 636)
Luminal A	262	Referent	—	Referent	—	Referent	—	Referent	—
Luminal B/HER2-	188	1.23 (1.01 to 1.51)	.04	1.39 (1.10 to 1.74)	.01	1.20 (0.98 to 1.47)	.08	1.06 (0.83 to 1.36)	.62
Luminal B/HER2+	58	1.13 (0.84 to 1.53)	.42	1.40 (1.03 to 1.90)	.03	1.09 (0.80 to 1.48)	.59	1.14 (0.80 to 1.63)	.46
HER2-enriched	46	1.31 (0.96 to 1.81)	.09	1.58 (1.16 to 2.16)	.004	1.26 (0.90 to 1.75)	.18	1.13 (0.75 to 1.68)	.56
TN	82	0.93 (0.70 to 1.23)	.60	1.21 (0.90 to 1.64)	.22	0.90 (0.68 to 1.19)	.45	1.10 (0.80 to 1.52)	.55
Among tumors ≥2 cm (n = 590)
Luminal A	176	Referent	—	Referent	—	Referent	—	Referent	—
Luminal B/HER2-	203	0.92 (0.74 to 1.13)	.41	1.10 (0.88 to 1.38)	.41	0.90 (0.73 to 1.11)	.32	1.27 (0.98 to 1.64)	.08
Luminal B/HER2+	84	1.10 (0.85 to 1.43)	.48	1.12 (0.85 to 1.48)	.42	1.09 (0.84 to 1.42)	.51	1.14 (0.82 to 1.59)	.44
HER2-enriched	53	0.79 (0.56 to 1.13)	.19	1.02 (0.71 to 1.47)	.93	0.77 (0.54 to 1.11)	.16	1.19 (0.80 to 1.77)	.39
TN	74	0.87 (0.65 to 1.16)	.34	0.97 (0.70 to 1.33)	.83	0.86 (0.64 to 1.16)	.33	1.15 (0.81 to 1.63)	.43

^aAge at diagnosis (younger than 50 years and older than 50 years) was used as a proxy for menopausal status because of missing data for menopausal status (260 patients with unknown data). “—” is used instead of a P value for the reference category. BMI = body mass index; CI = confidence interval; OR = odds ratio; TN = triple negative

^bModeling tumor subtype as the outcome variable and Volpara density per 1-unit standard deviation increase and adjusting for age, BMI, menopausal status, and parity.

^cOne patient aged 36 years was removed from analysis because of her non-natural menopause status.

^dUnderweight/normal BMI: BMI ≤ 25; overweight/high BMI: BMI > 25.

Discussion

In this large analysis of unscreened Chinese women with breast cancer, we replicated our previous findings (22), showing that high quantitative MD is associated with the HER2-enriched subtype. Additionally, we found that high MD was associated with luminal B subtypes and that these associations were driven by dense fibroglandular volume instead of percent density. Consistent with the previous study, we observed that the associations between increasing fibroglandular dense volume and HER2+ subtypes were stronger in women with normal BMI.

Although studies suggest that elevated breast density increases the risk of breast cancer, irrespective of intrinsic tumor subtype, associations between MD and tumor subtypes in case-only analyses are inconsistently reported in the literature. Although most studies reported no statistically significant variations in MD across different subtypes (12,26,27), some suggested that higher MD was associated with luminal A (28) and HER2-enriched subtypes (29,30), and 2 studies found that TNBC was associated with lower MD (31,32). The mixed results may be attributed to heterogeneous study populations, mode of detection, and risk factors. For example, we found that the MD-HER2 association was stronger in women with normal BMI. The lower prevalence of obesity in East Asian compared with Western women may explain why the association is more consistently seen in Asian populations. In addition, molecular and histologic features of breast tumors are thought to vary by mode of detection, and therefore, the association between MD and tumor subtypes may differ according to mode of detection (8). The vast majority (>90%) of patients were not detected through screening in our study, which may lead to different findings from those based on screened populations.

The associations between higher MD and luminal B and HER2-enriched subtypes may explain the high incidence rates of HER2-enriched and high prevalence of luminal B breast cancers observed among Asian American and indigenous Asian women (19,21,33,34). Although the etiology of these breast cancer subtypes remains unknown, our findings suggest that high MD, particularly fibroglandular breast volume, may be associated with HER2+ and luminal B subtypes. The TN subtype, however, was associated with lower nondense volume, which is consistent with findings from two large population-based studies showing stronger inverse associations of nondense area with breast cancer risk for ER- as compared with ER+ tumors, indicating reduced nondense area in ER- than ER+ patients (35,36). These results also align with findings from studies showing TNBC associated with lower MD (31,32) and our previous reports of Chinese and Malaysian breast cancer patients, where lower BMI was observed among TN, as compared with luminal A patients (33,34). The associations of lower BMI or nondense breast area and/or volume (fatty tissue) with TNBC in postmenopausal women suggest that local estrogen production through aromatization in fatty tissue may play a less important role in the risk of TNBC than that of luminal cancers.

There are some limitations to this study. The case-only analysis makes any conclusions on the temporal relationship between MD and risk of specific tumor subtypes indirect. Further, our sample is not representative of all breast cancer populations within China or across Asian populations. Nevertheless, we were able to replicate our previous results using quantitative measures of MD among Chinese breast cancer patients. Our findings may explain, in part, the higher proportion of HER2+ and luminal B breast cancers observed in Asian populations. Future studies are needed to elucidate the role of MD in the etiology of these breast cancer subtypes in Asian women.

Funding

This work was supported by the Intramural Research Program of the Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Rockville, MD.

Notes

Role of the funder: The funder played no part in the design, analysis, or drafting of this manuscript.

Disclosures: Ariane Chan is an employee of Volpara Health Technologies Ltd. Volpara Health Technologies Ltd did not provide any financial assistance for the project. The authors have no other conflicts of interest to disclose.

Author contributions: Study design: YT, JLG, GLG, JL, and XRY; data collection: YT, ENL, CG, NH, NL, and JL; data curation: HK, ET, JD; data analysis: JLG and HK; data interpretation: YT, JLG, HK, AC, GLG, and XRY; consultation: BZ, HS, HZ, AC, and GLG; manuscript preparation: JLG, HK, and XRY; manuscript review and editing: all authors; decision to submit the manuscript: all authors.

Prior presentations: Part of data included in this paper was presented at the 2019 San Antonio Breast Cancer Symposium.

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.