BMI and Breast Subcutaneous Fat: Potential Indicators for Ultrasound Diagnosis of Breast Cancer?

Methods

Study Population
This study included patients who underwent preoperative breast ultrasound examinations between September 2017 and August 2020. The inclusion criteria were as follows: (1) clear visualization of nodules with multiple ultrasound images available, and (2) a histopathological diagnosis of breast cancer confirmed by core needle biopsy or surgical pathology. Exclusion criteria were: (1) unclear or undetectable nodules on ultrasound, (2) any prior treatments received before the ultrasound examination, and (3) incomplete data. This retrospective study was approved by the institutional ethics committee, and informed consent was waived due to its retrospective design.

Pathological and Immunohistochemical Analysis
Patient data, including age, BMI, menopausal status, palpable axillary lymph nodes (ALN), family history of breast cancer, ultrasound ALN status, pathological type, histological grade, estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor‐2 (HER2) status, was collected. ER or PR positivity was defined as ≥1% immunohistochemistry staining.
11
HER2 positivity was defined as: (1) immunohistochemistry 3+ or >30% of invasive cancer cells exhibiting strong membranous staining; (2) HER2/CEP17 ratio >2.2/2.0 via fluorescent in‐situ hybridization; or (3) HER2 gene copy number >6.0 per nucleus via chromogenic in‐situ hybridization.
12

Ultrasound Image Evaluation
Most images were acquired using the SuperSonica Aixplorer US scanner (SuperSonic Imagine S.A., Aix‐en‐Provence, France) equipped with a 7–15 MHz linear‐array transducer. Other images were obtained using the Mindray Resona 5S US scanner (Mindray Bio‐Medical Electronics Co., Shenzhen, China) with a 5–14 MHz linear‐array transducer. Standardized image acquisition included capturing 12 evenly spaced conventional images within a 180° clockwise range from the maximum cross‐sectional view of the tumor. Ultrasound features such as shape, orientation, margins, boundary, echo pattern, calcification, and posterior acoustic features were evaluated according to the Breast Imaging Reporting and Data System classification. Vascular distribution was assessed using the Adler index (0, I, II, or III).
13
All ultrasound images were reviewed independently by 2 experienced radiologists blinded to the pathological results. Discrepancies were resolved through discussion.

Measurement of Breast SFT Via Ultrasound
For each patient, 3 clear ultrasound images were selected. Breast SFT was measured at 3 fixed locations: the tumor center and 2 points located 18 mm horizontally from the tumor center on either side. The mean value was calculated. All measurements were performed by a radiologist blinded to the pathological results (Figure 1).

Measurement of Breast SFT Via Mammography
For mammographic evaluation, images from the craniocaudal and mediolateral oblique views were selected. Breast SFT was measured at 3 fixed locations: the edge of the nipple‐areolar complex and 2 points forming a 45° angle above and below the nipple. The mean value was calculated. All measurements were performed by a radiologist blinded to the pathological results (Figure 2).

Gene Mutation
Tumor and matched blood DNA samples were sequenced using a custom 511‐gene panel specifically designed for breast cancer, as described in previous studies.
14
This panel was developed by integrating data from The Cancer Genome Atlas,
15
Memorial Sloan Kettering Cancer Center
16
and the Fudan University Shanghai Cancer Center Triple Negative Breast Cancer (TNBC) dataset.
17
The captured DNA fragments were pooled and sequenced, achieving a median coverage of 1000× for tumor genomic DNA and 400× for blood genomic DNA. This sequencing strategy effectively minimized the interference of germline variants, ensuring precise detection of somatic mutations. Mutations were included in the analysis if they involved shift deletions and insertions, in‐frame deletions and insertions, missense mutations, nonsense mutations, or splice site alterations.

Data Analysis
Descriptive statistical methods were used to summarize the data. Categorical variables were presented as absolute and relative frequencies, while numerical variables were expressed as medians and interquartile ranges for non‐normal distributions. The Shapiro–Wilk test was used to assess the distribution of numerical variables, followed by non‐parametric tests as appropriate. Continuous variables were analyzed using either the Wilcoxon rank‐sum U test or Student's t‐test. Categorical variables were assessed using the χ2 test or Fisher's exact test. Pearson's r test was used to evaluate the relationships between normally distributed variables. All p‐values were 2‐tailed, with a significance level set at α = 0.05. Statistical analyses were performed using SPSS version 27.0 or R version 4.3.0.

Results

Baseline Characteristics
A total of 1670 patients were included in the study. According to the BMI threshold for overweight (≥24 kg/m2) in China, 686 patients (41.1%) were categorized as overweight, while 984 (58.9%) were non‐overweight. The baseline characteristics of overweight and non‐overweight patients are shown in Table 1. The mean BMI of the non‐overweight group was 21.52 kg/m2 (range: 13.22–23.95 kg/m2), whereas the mean BMI of the overweight group was 26.57 kg/m2 (range: 24–58.27 kg/m2). No significant difference was observed between groups regarding family history of breast cancer, ER status, PR status, or HER2 status. Overweight patients were more likely to be postmenopausal (p < .001) and have clinically palpable lymph nodes (p = .013).

Analysis of Ultrasonographic Features of Breast Cancer Patients Grouped by BMI (24 Kg/m2)
Significant differences in ultrasound features were observed between the high‐ and low‐BMI groups, including orientation (p = .001), margin (p = .05), boundary (p < .001), posterior acoustic features (p < .001), calcifications (p < .001), and vascular invasion (p = .001) (Table 2). Subgroup analysis based on breast cancer subtypes revealed distinct patterns. In the TNBC group, significant differences were observed in clinically palpable lymph nodes (p = .031), boundary (p = .002), and vascular degree (p = .028) (Table S1). For HER2‐positive breast cancer patients, orientation (p = .05) and calcifications (p = .044) differed between the BMI groups (Table S2). Among Luminal A breast cancer patients, palpable lymph nodes (p = .044), margin (p = .016), boundary (p < .001), posterior acoustic features (p = .008), and calcifications (p = .002) exhibited significant variation (Table S3). Similarly, in the Luminal B breast cancer group, differences were identified in orientation (p = .002), boundary (p < .001), posterior acoustic features (p < .001), and vascular invasion (p = .008) (Table S4).

Analysis of Mammographic and Ultrasound‐Measured Breast SFT
Among the 1670 patients included in the study, 470 had mammographic images. The breast SFT measured on mammographic images was analyzed for differences based on BMI (Figure 1). It was observed that patients with BMI ≥24 kg/m2 exhibited significantly higher breast SFT (Table 3). Additionally, breast SFT was measured on corresponding ultrasound images of these 470 patients and compared to the mammographic breast SFT (Figure 2). A moderate correlation was identified between the 2 measurements (r = 0.565, p < .001) (Figure 3). Similarly, patients with BMI ≥24 kg/m2 showed higher breast SFT values on ultrasound as well (Table 3). BMI was also correlated with breast SFT measured by MG (r = 0.578, p < .001) and ultrasound (r = 0.485, p < .001) (Figure 3). These findings indicate that, regardless of the imaging modality used (MG or ultrasound), the measured breast SFT differs significantly between patients with different BMI categories.

Relationship between Ultrasound‐Measured Breast SFT and Ultrasound Features
Ultrasound analysis revealed that breast SFT was higher in cases of irregular tumor shape, halo sign, and posterior shadowing (Table 4). Predictive analyses of these features yielded area under the curve (AUC) values of 0.605, 0.666, and 0.632, respectively. The optimal cutoff points for SFT based on the Youden index were 8.65, 8.35, and 8.35 mm for predicting irregular shape, halo, and posterior shadowing, respectively (Figure 4).

Relationship between Ultrasound‐Measured Breast SFT and Pathological Features
Ultrasound‐measured breast SFT showed a correlation with Ki67 expression (r = 0.098, p = .033). Patients with breast SFT ≥8.6 mm exhibited significantly higher Ki67 levels compared to those with breast SFT <8.6 mm (Figure 5).

Analysis of Somatic Mutations
Among the 470 patients with ultrasound‐measured breast SFT, 234 patients had data available for somatic mutations in specific genes. Our study found no significant difference in the mutation frequencies of genes closely associated with malignant phenotypes of breast cancer, such as proliferation and invasion, including fibroblast growth factor receptor (FGFR), erb‐b2 receptor tyrosine kinase 2 (ERBB2), phosphatase and tensin homolog (PTEN), epidermal growth factor receptor (EGFR), ataxia‐telangiectasia mutated (ATM), AKT serine/threonine kinase 1 (AKT1), phosphatidylinositol‐4,5‐bisphosphate 3‐kinase catalytic subunit alpha (PIK3CA), and Janus kinase (JAK), between patients with breast SFT <8.60 mm and those with SFT ≥8.60 mm.
We further analyzed the relationship between BMI and somatic mutations in these genes. Using 24 kg/m2 as the BMI threshold, we observed that patients with BMI ≥24 kg/m2 had a significantly higher mutation frequency only in the ATM gene compared to those with BMI <24 kg/m2, while other genes showed no significant difference.
Based on the linear regression equation between breast SFT and BMI, we determined that the BMI cutoff corresponding to predictive thresholds for irregular shape, echogenic halo, and posterior acoustic shadow was approximately 22 kg/m2. When patients were grouped by this BMI threshold, those with BMI ≥22 kg/m2 exhibited significantly higher mutation frequencies in these genes. Although breast cancer gene 1 (BRCA1) (p = .06), breast cancer gene 2 (BRCA2) (p = .12), and signal transducer and activator of transcription (STAT) (p = .08) also showed higher mutation frequencies in the higher BMI group, these differences did not reach statistical significance (Figure 6).

Discussion
Our study revealed that a higher BMI is indeed associated with malignant ultrasound features in breast cancer patients. When patients were grouped by subtype, only those with luminal A and B subtypes exhibited more malignant ultrasound features in the higher BMI group, while those with TNBC and HER2‐positive subtypes exhibited no significant difference in most ultrasound features between the high‐ and low‐BMI groups. This suggests that patients with a BMI ≥24 kg/m2 might have more aggressive tumors compared to those with a BMI <24 kg/m2.
In addition to BMI, we sought to explore whether breast SFT was associated with overweight and obesity. Since ultrasound‐measured breast SFT represents a cross‐sectional measurement, whereas MG‐derived breast SFT reflects the entire breast, we respectively examined the relationship between BMI and MG‐measured breast SFT, BMI and ultrasound‐measured breast SFT, as well as the correlation between the 2 methods. Both MG‐ and ultrasound‐measured breast SFT showed a positive correlation with BMI. Additionally, the breast SFT values measured by the 2 methods were significantly correlated, indicating that while ultrasound measures local breast SFT, it can still represent overall breast subcutaneous fat to a certain extent. These findings elucidated that breast SFT is relevantly related to BMI and obesity.
Analysis of the relationship between ultrasound features and breast SFT revealed that higher breast SFT was observed only in tumors with representative malignant ultrasound features including irregular shape, halo, or posterior acoustic shadowing. Using breast SFT to predict these features yielded AUC values >0.6, indicating moderate predictive performance. Interestingly, the optimal Youden indices for these predictive models corresponded to breast SFT thresholds of 8.35–8.65 mm, suggesting that nodules with breast SFT exceeding this range are more likely to exhibit irregular and malignant ultrasound features. Furthermore, we found a positive correlation between ultrasound‐measured breast SFT and Ki67 expression. The percentage of Ki67 expression indicates the proportion of cells in the proliferative phase and serves as a crucial clinical marker of tumor proliferation and invasiveness.
18
Thus, we propose that higher breast SFT is associated with more aggressive proliferation and invasion of breast cancer.
Somatic mutations caused by gene mutations occur in most tumor types and are highly prevalent in breast cancer.
19
FGFR1, a member of the FGFR family, activates downstream signaling pathways such as phosphoinositide 3‐kinase (PI3K)/AKT, mitogen‐activated protein kinase (MAPK), phospholipase C gamma, and JAK/STAT upon binding with FGF.
20
FGFR1 mutations, more common than FGFR2‐4 mutations, led to hyperactivation of FGFR signaling, which transformed normal cells into cancer cells and contributed to tumor growth, metastasis, and reduced survival in breast cancer.
21
Similarly, ERBB2 amplification is commonly associated with poor prognosis and severe invasiveness in breast cancer.
22
,
23
The PI3K/AKT/PTEN pathway is frequently activated in tumors, with approximately 35% of TNBC cases exhibiting alterations in PIK3CA, AKT1, and/or PTEN. PTEN deficiency drives tumorigenesis and invasiveness via PI3K/AKT activation.
24
,
25
EGFR, highly expressed in TNBC, is linked to lower survival rates and serves as a biomarker and therapeutic target.
23
Its dysregulation activates downstream pathways influencing tumor survival, proliferation, metastasis, and invasion.
26
ATM kinase is an essential protein involved in the DNA damage response pathway. Mutations in the ATM gene often result in reduced cellular capacity to repair DNA damage.
27
Activation of ATM can regulate interleukin‐8, thereby enhancing the migratory and invasive capabilities of breast cancer cells. Furthermore, oxidized ATM has been shown to promote the abnormal proliferation of cancer‐associated fibroblasts in breast tissue.
28
BRCA1/2 mutations also confer a significantly increased lifetime risk of breast cancer.
29

All the above genes are closely related to the proliferation and invasion of breast cancer. In our study, higher BMI patients exhibited a significantly elevated proportion of somatic mutations in these genes. This may result from various factors, including adipose accumulation disrupting lipid metabolism.
30
Globally increasing high‐fat diets induce cancer through fat tissue accumulation, altered gut microbiota, and chronic inflammation. Excessive adiposity also modulates the immune microenvironment, promoting inflammation and immune cell dysfunction.
31
Additionally, metabolic syndromes such as hyperglycemia, hypertension, and hyperlipidemia are independently linked to cancer initiation and progression, frequently associated with high BMI.
32
Finally, hyperinsulinemia in obesity may activate rat sarcoma virus/rapidly accelerated fibrosarcoma/MAPK and PI3K/Akt/mechanistic target of rapamycin pathways, enhancing cell proliferation and tumorigenesis.
33

When the breast SFT was grouped by 8.60 mm, no significant difference was observed in the mutation frequency of the above genes between the 2 groups. According to the results of this study, 8.60 mm may be the best predictive cut‐off point of ultrasound characteristics, but not necessarily the best critical value for gene mutation rate grouping. In the future, the optimal grouping cut point of breast SFT on gene mutation rate can be reevaluated, and the sample size can be increased and possible confounding factors can be corrected to improve the statistical power and robustness of the conclusions. When the BMI threshold is set at 22 kg/m2, the difference in gene mutation frequencies between the 2 groups is more pronounced compared to a BMI of 24 kg/m2. Based on our findings, although a BMI of 24 kg/m2 is defined as overweight according to domestic guidelines, it may not accurately reflect the metabolic state or biological differences in patients with breast cancer. Thresholds established based on public health standards may not be appropriate for molecular research specific to certain cancers.
Our results suggest that a BMI of 22 kg/m2 may have greater biological significance in the differential analysis of gene mutation rates in breast cancer. Compared to a BMI of 24 kg/m2, a BMI of 22 kg/m2 may more accurately represent the physiological state of patients regarding the impact of adipose tissue on breast cancer progression. Our study emphasizes the potential diagnostic value of a BMI of 22 kg/m2 in predicting gene mutation rates associated with breast cancer. Particularly when combined with breast SFT and ultrasound characteristics, a BMI of 22 kg/m2 could represent a more biologically meaningful cutoff. In future clinical applications, a BMI of 22 kg/m2 could be considered a key threshold for stratifying breast cancer patients (for example, in genetic testing or treatment strategies). This may help optimize treatment decisions and monitoring plans.
This study has several limitations. First, being a retrospective analysis, it is subject to inherent biases, and the interpretation of ultrasound images is somewhat subjective. However, all cases included in the study adhered to standardized imaging protocols, with multiple ultrasound images collected to ensure a thorough assessment of breast lesions. Second, although the dataset is sizable, it is derived from a single institution, and data from additional centers would be necessary to improve the generalizability and objectivity of the findings. Third, only a subset of patients with available mammogram and ultrasonographic data on breast SFT also had information on somatic mutations, which may have introduced selection bias, potentially affecting the validity of the comparisons and limiting the broader applicability of the results.

Conclusions
Our study demonstrated that in terms of the ultrasound diagnosis of breast cancer, BMI and breast SFT under ultrasound have a certain auxiliary diagnostic value for malignant ultrasound features, while in predicting the proliferation and invasion of breast cancer, BMI has a higher diagnostic value than breast SFT, and BMI of 22 kg/m2 has a higher diagnostic value than BMI of 24 kg/m2. These metrics can aid radiologists in interpreting imaging results.

Supporting information

Table S1. Clinical and ultrasound features of the participants with TNBC.

Table S2. Clinical and ultrasound features of the participants with HER2‐positive breast cancer.

Table S3. Clinical and ultrasound features of the participants with luminal A breast cancer.

Table S4. Clinical and ultrasound features of the participants with luminal B breast cancer.