BI-RADS Ultrasound Lexicon Descriptors and Stromal Tumor-Infiltrating Lymphocytes in Triple-Negative Breast Cancer

Abstract

Purpose:

Increased levels of stromal tumor-infiltrating lymphocytes (sTILs) have recently been considered a favorable independent prognostic and predictive biomarker in triple-negative breast cancer (TNBC). The purpose of this study was to determine the relationship between BI-RADS (Breast Imaging Reporting and Data System) ultrasound lexicon descriptors and sTILs in TNBC.

Materials and Methods:

Patients with stage I-III TNBC were evaluated within a single-institution neoadjuvant clinical trial. Two fellowship-trained breast radiologists used the BI-RADS ultrasound lexicon to assess pretreatment tumor shape, margin, echo pattern, orientation, posterior features, and vascularity. sTILs were defined as low <20 or high ≥20 on the pretreatment biopsy. Fisher’s exact tests were used to assess the association between lexicon descriptors and sTIL levels.

Results:

The 284 patients (mean age 52 years, range 24-79 years) were comprised of 68% (193/284) with low-sTIL tumors and 32% (91/284) with high-sTIL tumors. TNBC tumors with high sTILs were more likely to have the following features: (1) oval/round shape than irregular shape (p = 0.003), (2) circumscribed or microlobulated margins than spiculated, indistinct, or angular margins (p = 0.0005); (3) complex cystic and solid pattern than heterogeneous pattern (p = 0.006); and (4) posterior enhancement than shadowing (p = 0.002). There was no significant association between sTILs and descriptors for orientation and vascularity (p = 0.06 and p = 0.49, respectively).

Conclusion:

BI-RADS ultrasound descriptors of the pretreatment appearance of a TNBC tumor can be useful in discriminating between tumors with low and high sTIL levels. Therefore, there is a potential use of ultrasound tumor characteristics to complement sTILs when used as stratification factors in treatment algorithms for TNBC.

INTRODUCTION

The tumor immune microenvironment, with its impact on response to therapy and patient prognosis, is a subject of active research and clinical interest in breast cancer. Triple-negative breast cancer (TNBC) has been shown to be the most immunogenic breast cancer subtype, displaying more prominent lymphocytic infiltration in the stroma surrounding the tumor than other subtypes (1). Recently published results from a pool of nine large studies demonstrated that the presence of stromal tumor-infiltrating lymphocytes (sTILs) at the time of diagnosis is a robust and independent prognostic biomarker in early-stage TNBC; specifically, patients with high-sTIL TNBC were found to have lower rates of disease recurrence and death from disease after standard adjuvant chemotherapy (2). In addition, several trials have shown that higher levels of sTILs are associated with higher rates of pathologic complete response (pCR) in TNBC patients undergoing neoadjuvant therapy (3–7).

The quantification of tumor infiltrating lymphocytes (TILs) requires the visual assessment of mononuclear immune cells on routine hematoxylin-and-eosin–stained slides from the tumor biopsy. Because sTILs have been found to more reproducible than intratumor tumor-infiltrating lymphocytes, the International Immuno-Oncology Biomarker Working Group on Breast Cancer has issued guidelines recommending that TILs be evaluated in the stromal compartments (8). When sTILs are evaluated in the pretreatment neoadjuvant setting, the tissue is typically obtained by core needle biopsy of the tumor. There are recognized pitfalls that affect the accuracy of assessing sTILs, including heterogeneity in sTIL distribution, limited tumor sampling, and technical factors, such as poor fixation and crush artifacts (9). Because of these limitations, the sample obtained may not be completely representative of the tumor and the surrounding stroma. A complementary method using imaging to evaluate sTILs may be beneficial in overcoming these limitations.

Ultrasound is a readily available, conventional imaging modality that is used in breast cancer diagnosis and assessment. The American College of Radiology (ACR) BI-RADS (Breast Imaging Reporting and Data System) lexicon is a widely accepted standard for describing breast lesions. In contrast to the limited sample from a core needle biopsy, ultrasound imaging allows for visualization of the entirety of a breast tumor as well as the tissue surrounding the tumor. If strong associations between ultrasound tumor characteristics and sTILs can be established, then ultrasound could provide relatively inexpensive and reliable imaging surrogates to sTILs. The coupling of these imaging surrogates with sTILs can be used as stratification factors when designing research trials that would further explore its clinical utility in the treatment of TNBC. Therefore, the objective of this study was to investigate the relationship between the BI-RADS ultrasound lexicon descriptors and sTILs in TNBC.

MATERIALS AND METHODS

Study Population

This sub-study, which received approval from the Institutional Review Board at The University of Texas MD Anderson Cancer Center, was part of an ongoing randomized clinical trial (ClinicalTrials.gov Identifier NCT02276443) in patients with stage I-III TNBC who were prospectively monitored for response to neoadjuvant therapy at a single comprehensive cancer center. In support of an institutional program to accelerate progress in treatment options for TNBC, our breast medical oncologists offered this neoadjuvant clinical trial to all patients who had a provisional diagnosis of an untreated TNBC tumor measuring at least 1.5 cm in greatest dimension. All patients in the clinical trial gave informed consent before enrollment. Complete eligibility details for enrollment in the clinical trial including inclusion and exclusion criteria can be found at https://clinicaltrials.gov/ct2/show/NCT02276443. The timeframe of this imaging study was from November 2015 through April 2019 at which time 393 patients consented to the clinical trial. The sub-study included only the 284 patients for whom sTILs were available from evaluation of tumor specimens that had been obtained by core needle biopsy prior to initiating systemic therapy at our institution. Figure 1 shows a flowchart of the patient inclusion process that was used in this imaging analysis.

Figure 1.

Flowchart shows the TNBC patient inclusion process used in this analysis. TNBC, triple-negative breast cancer.

Imaging Equipment and Interpretation

Diagnostic breast ultrasound was performed on all patients using Epiq 5G scanners with 12- to 18-MHz high-frequency linear array transducers (Philips Healthcare, Andover, MA, USA). Transverse and longitudinal images of the index breast tumor were captured during the diagnostic ultrasounds. For patients with multiple foci of disease, the index tumor was defined as the dominant or largest tumor in the affected breast. Two fellowship-trained breast radiologists with 10 years (RPC) and 2 years (DAS) of experience, respectively, performed the imaging review of the diagnostic ultrasounds while blinded to the imaging reports and to the histopathologic data; they jointly evaluated the index tumor for each patient retrospectively to arrive at a consensus. Tumor shape, margin, echo pattern, orientation, posterior features, and vascularity were assessed using the ACR BI-RADS Atlas 5th edition ultrasound lexicon (10).

Histopathologic Characteristics

Estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) data were extracted from the pathology reports of the diagnostic core needle biopsy in the patients’ electronic medical records. Each biopsy was performed within 28 days of the diagnostic ultrasound. ER and PR receptor were evaluated using immunohistochemical scoring, expressed as the percentage of cells with positive nuclear staining. ER and PR receptor positivity were defined as nuclear staining of ≥10% (11). HER2 status was determined by immunohistochemical analysis or fluorescence in situ hybridization according to the College of American Pathologists/American Society of Clinical Oncology HER2 guideline recommendations (12). sTILs were analyzed according to recommendations provided by the International Immuno-Oncology Biomarker Working Group in Breast Cancer (13). Specimens from pretreatment 14-gauge core needle biopsies were evaluated for sTILs. The sTIL levels for this study were categorized as low <20% and high ≥20%. Recursive partitioning statistical analysis was applied to identify this optimal cut-off value of 20% that was best associated with pCR within the same clinical trial cohort (Abuhadra et al. 2021, manuscript submission under review). The histomorphology of tumors with high sTILs and low sTILs is shown in Figure 2.

Figure 2.

Histomorphology of tumors with high sTILs and low sTILs in pretreatment core biopsies. (a) H&E (hematoxylin and eosin) stain of a representative tumor with high sTILs showing dense infiltration of lymphocytes in the stroma. (b) H&E stain of a representative tumor with low sTILs showing sparse lymphocytes in the stroma. sTILs, stromal tumor-infiltrating lymphocytes.

Data Collection and Statistical Analysis

The electronic medical records of patients were reviewed to determine age, histologic type, immunohistochemical results, T stage, N stage, and tumor focality. Fisher’s exact tests were used to assess the association between sTIL level and BI-RADS ultrasound lexicon descriptors. All analyses were performed using the statistical software R version 3.6.1. A p value of < 0.05 was regarded as statistically significant between groups. No adjustments for multiple testing were made.

RESULTS

Study Population

The mean patient age was 52 years, with a range of 24 to 79 years. Ninety-one patients of 284 had high-sTIL tumors (32%) and 193/284 (68%) had low-sTIL tumors.

Clinical and Pathologic Characteristics

The clinical and pathologic characteristics of the 284 patients are summarized in Table 1. There was no significant difference in histologic subtype (p = 0.40) or focality (p = 0.88) between tumors with high and low sTILs. Invasive ductal carcinoma was the most common histologic type among tumors with both low (162/193 [83.9%]) and high sTILs (83/91 [91.2%]). Most patients had unifocal disease: 144/193 (74.6%) with low-sTIL tumors and 69/91 (75.8%) with high-sTIL tumors. There was a more substantial difference in N stage between the two groups, approaching significance level (p = 0.05). N0 disease was identified in 116/193 (60.1%) tumors with low sTILs and in 48/91 (52.7%) with high sTILs. There was a significant association between T stage and sTIL level (p = 0.01). More specifically, 146 of193 (75.6%) low-sTIL tumors were T1-T2 and 47/194 (24.2%) were T3-T4. In comparison, 82/91 (90.1%) high-sTIL tumors were T1-T2 and 9/91 (9.9%) were T3-T4.

TABLE 1.

Clinical and Pathologic Characteristics of 284 Patients with TNBC Stratified by Low versus High sTILs

Variable	Descriptor	Low (n)	Low (%)	High (n)	High (%)	All (n)	All (%)
Histologic type	Apocrine	3	1.6	0	0.0	3	1.1
	Invasive ductal carcinoma	162	83.9	83	91.2	245	86.3
	Invasive lobular carcinoma	2	1.0	1	1.1	3	1.1
	Invasive mammary	1	0.5	0	0.0	1	0.4
	Mixed lobular and ductal	0	0.0	1	1.1	1	0.4
	Metaplastic	22	11.4	5	5.5	27	9.5
	Neuroendocrine	1	0.5	0	0.0	1	0.4
	Other	2	1.0	1	1.1	3	1.1
p = 0.40	All	193	100.0	91	100.0	284	100.0
T stage	T1	26	13.5	21	23.1	47	16.5
	T2	120	62.2	61	67.0	181	63.7
	T3	37	19.2	7	7.7	44	15.5
	T4	10	5.2	2	2.2	12	4.2
p = 0.01	All	193	100.0	91	100.0	284	100.0
N stage	N0	116	60.1	48	52.7	164	57.7
	N1	45	23.3	28	30.8	73	25.7
	N2	2	1.0	5	5.5	7	2.5
	N3	30	15.5	10	11.0	40	14.1
p = 0.05	All	193	100.0	91	100.0	284	100.0
Focality	Multifocal	49	25.4	22	24.2	71	25.0
	Unifocal	144	74.6	69	75.8	213	75.0
p = 0.88	All	193	100.0	91	100.0	284	100.0

TNBC, triple-negative breast cancer; sTILs, stromal tumor-infiltrating lymphocytes

BI-RADS Ultrasound Lexicon Descriptors and sTILs

The summary statistics of the BI-RADS lexicon descriptors by sTIL levels are provided in Table 2. TNBC tumors with high sTILs were more likely to be an oval or round shape (65/91 [71.4%]) than an irregular shape (26/91 [28.6%]), (p = 0.003). TNBC tumors with high sTILs had a greater likelihood of having circumscribed or microlobulated margins (70/91 [76.9%]) and less likely to have spiculated, indistinct, or angular margins (21/91 [23.1%], p = 0.0005). TNBC tumors with high sTILs were more likely to have complex cystic and solid echo patterns (22/91 [24.2%]) and less likely to have a heterogeneous pattern (16/91 [17.6%], p = 0.006). TNBC tumors with high sTILs were more likely to have posterior enhancement (36/91 [39.6%]) and less likely to have posterior shadowing (6/91 [6.6%], p = 0.002). There was no significant association between sTIL levels and lexicon descriptors for orientation and vascularity (p = 0.06 and p = 0.49, respectively). Examples of tumors with high and low sTILs are shown in Figures 3 and 4, respectively.

TABLE 2.

Ultrasound Imaging Features in 284 Patients with TNBC Stratified by Low versus High sTILs

Variable	Descriptor	Low (n)	Low (%)	High (n)	High (%)	All (n)	All (%)
Shape	Irregular	91	47.1	26	28.6	117	41.2
	Oval/round	102	52.9	65	71.4	167	58.8
p = 0.003	All	193	100.0	91	100.0	284	100.0
Margin	Angular	9	4.7	0	0.0	9	3.2
	Circumscribed	15	7.8	17	18.7	32	11.3
	Indistinct	88	45.6	21	23.1	109	38.4
	Microlobulated	70	36.3	53	58.2	123	43.3
	Spiculated	11	5.7	0	0.0	11	3.9
p = 0.0005	All	193	100.0	91	100.0	284	100.0
Orientation	Not parallel	45	23.3	12	13.2	57	20.0
	Parallel	148	76.7	79	86.8	227	80.0
p = 0.06	All	193	100.0	91	100.0	284	100.0
Echo pattern	Complex cystic-solid	25	12.9	22	24.2	47	16.5
	Heterogeneous	64	33.2	16	17.6	80	28.2
	Hypoechoic	104	53.9	53	58.2	157	55.3
p = 0.006	All	193	100.0	91	100.0	284	100.0
Posterior features	Combined pattern	10	5.2	2	2.2	12	4.2
	Enhancement	43	22.3	36	39.6	79	27.8
	No posterior features	100	51.8	47	51.6	147	51.8
	Shadowing	40	20.7	6	6.6	46	16.2
p = 0.002	All	193	100.0	91	100.0	284	100.0
Vascularity	Absent	55	28.5	21	23.1	76	26.8
	Internal vascularity	128	66.3	67	73.6	195	68.7
	Vessels in rim	10	5.2	3	3.3	13	4.6
p = 0.49	All	193	100.0	91	100.0	284	100.0

TNBC, triple-negative breast cancer; sTILs, stromal tumor-infiltrating lymphocytes

Figure 3.

A 40-year-old woman with high-sTIL TNBC. Pretreatment breast ultrasound shows that the mass has an oval shape with circumscribed margins and posterior acoustic enhancement. TNBC, triple-negative breast cancer.

Figure 4.

A 42-year-old woman with low-sTIL TNBC. Pretreatment breast ultrasound shows that the mass has an irregular shape with indistinct margins. TNBC, triple-negative breast cancer.

DISCUSSION

In this study, our aim was to determine the relationship between BI-RADS ultrasound lexicon descriptors and sTILs in TNBC. The key findings of our study indicate that BI-RADS ultrasound descriptors for tumor shape, margin, echo pattern, and posterior enhancement are associated with the sTIL level. More specifically, tumors with high sTILs are more likely to have an oval or round shape, circumscribed or microlobulated margins, posterior enhancement, and complex cystic and solid echo patterns compared to tumors with low sTILs. The predictive and prognostic significance of sTILs in TNBC has recently been recognized (14–20). Indeed, increased sTILs have been shown to predict response to neoadjuvant chemotherapy and be associated with increased invasive disease-free survival, distant disease-free survival, and overall survival rates in TNBC patients (2,4).

To date, we are aware of only two studies that have evaluated the association between sTILs and ultrasound tumor features of the primary tumor, both published in 2019: one by Fukui et al., in which 8 of 191 breast cancer patients in the cohort had TNBC, and another by Celebi et al., in which 19 of 158 patients had TNBC (21,22). In contrast to our study, which used the ACR BI-RADS ultrasound lexicon descriptors, Fukui et al. used the Japan Association of Breast and Thyroid Sonology classification system (22). Despite using different lexicon models, our study and the Fukui et al. study both revealed that high TILs were associated with lobulated margins and posterior acoustic enhancement (22). Our results were also similar to those of the Celebi et al. study, which analyzed both ultrasound and MRI features using the BI-RADS lexicon and found that high sTILs were associated with round tumor shape and circumscribed tumor margins (21). We know of only one study with a cohort comprised of only TNBC patients: Ku et al. (2016), which evaluated 112 patients using the BI-RADS MRI lexicon (23). Although it was an MRI study rather than an ultrasound study, our results were similar in showing that high sTILs have a strong association with a round shape and circumscribed margins (23). Indeed, the results of our study—the largest of its kind (n = 284)—validated that TNBC with high sTILs are more likely than TNBC with low sTILs to manifest with a shape and margin that are predominantly observed in benign breast lesions (24,25). Indeed, TNBC tumors with high sTILs may be misinterpreted as benign tumors, such as fibroadenomas. The classic TNBC imaging appearance of an oval mass with circumscribed margins are TNBC tumors with high sTILs (26,27). The shape and margin of the tumor may be factors that induce its microenvironment, or it may be the tumor microenvironment that drives the appearance of the tumor. Indeed, the underlying molecular and biological mechanisms that guide the relationship between a tumor and its surrounding stroma remain to be elucidated.

The results of our study showed a different distribution in the T stage (p = 0.01) and N stage (p = 0.05) between high sTIL and low sTIL levels. Our study had a higher frequency of T3-T4 disease and a higher frequency of N0 disease in patients with low sTIL tumors. Although Denkert et al. used a different categorization of sTILs (low [0–10%], intermediate [11–59%], high [60–100%]), they reported similar results with low sTILs having both a higher incidence of T3-T4 disease and N0 disease (4). In contrast, Loi et al. noted that lower sTILs had larger tumor burden and higher nodal burden (2). Notably, Loi et al. evaluated sTIL as a continuous variable compared to our dichotomous cutoff approach; additionally, the Loi et al. study was a pooled analysis of a mixture of randomized control trials and retrospective studies (2). These mixed results in T stage and N stage with their relationships with sTIL levels underscore the need for further investigations.

The findings from our imaging study have clinical implications. Because TNBC is considered the most immunogenic breast cancer, there has been continued interest in further exploring the use of sTILs as a biomarker of immunogenicity (1,8). Multiple clinical trials have shown that there is a strong linear association between increased TIL levels and improved disease-free survival in early-stage TNBC after adjuvant chemotherapy (14). Several studies have also shown that higher TILs are associated with higher rates of pCR in the neoadjuvant treatment of TNBC (3–7). A recent retrospective study demonstrated that patients with stage I TNBC with high sTILs had excellent survival outcomes in the absence of chemotherapy (28). If the pretreatment imaging appearance of tumors can help identify TNBC patients with a favorable prognosis, then there is the potential for de-escalating therapy in these patients. A resulting advantage of de-escalation is avoiding the potential toxicities of chemotherapy. The information obtained during pretreatment ultrasound imaging could address the shortcoming of determining sTILs from a core specimen that represents only a limited sample of a tumor, which may have inherent heterogeneity (9). Imaging with ultrasound could provide a relatively inexpensive, more complete assessment of the tumor and its surrounding environment. Nonetheless, ultrasound alone cannot and should not replace the pathology results from biopsy. Perhaps the combination of sTILs with pretreatment ultrasound imaging characteristics would provide an even more reliable predictive and prognostic signature that can be used to stratify patients for appropriate therapy.

There are some limitations to our study. This retrospective study was performed at a single institution. Interobserver variability was not performed; instead, we used consensus decision making. Since all patients had TNBC, the findings from our study may not be applicable to all breast cancer subtypes. Additionally, the assessment of sTIL for each case was from a core needle biopsy specimen, which provides an intrinsically limited evaluation of the tumor and peritumoral stroma. We further recognize that our study findings may be most pertinent to comparable referral centers with similar availability in advanced equipment and expertise in breast imaging and breast pathology. Lastly, both a larger internal study and multi-institutional studies are necessary for validation.

CONCLUSION

Increased levels of sTILs have been shown to be a favorable prognostic and predictive biomarker in TNBC but have inherent limitations. Ultrasound may provide a more complete evaluation of both the tumor and surrounding stroma than the core needle biopsy sample from which sTIL is determined. Our study demonstrates that BI-RADS ultrasound descriptors of the pretreatment appearance of a TNBC tumor can be useful in discriminating between tumors with low and high sTIL levels. Therefore, there is a potential use of ultrasound tumor characteristics to complement sTILs when used as stratification factors in treatment algorithms for TNBC.

Abbreviations:

ACR

American College of Radiology

BI-RADS

Breast Imaging Reporting and Data System

ER

estrogen receptor

HER2

human epidermal growth factor receptor 2

PCR

pathologic complete response

PR

progesterone receptor

sTIL

stromal tumor-infiltrating lymphocyte

TNBC

triple-negative breast cancer