Precision oncology extends beyond medical oncology alone. In this multidisciplinary session from the Precision Oncology Summit, Stanford Medicine experts in radiology, surgery, and pathology examine how imaging phenotypes, molecular subtypes, and surgical decision-making shape real-world personalized breast cancer care.

Pictured from left to right: Eric Rosen, MD, Uzma Waheed, MD, Irene Wapnir, MD, Carlos Jose Suarez, MD, and Christopher Chen, MD

This video from the Precision Oncology Summit shows insights from radiologists, surgeons, and molecular pathologists, sharing their unique perspectives on personalizing cancer treatment. It addresses the practical clinical challenges of implementing precision oncology beyond the traditional realms of medical oncology and molecular pathology. The goal is to examine precision oncology through a wider multidisciplinary lens, incorporating perspectives from radiology, surgery, and molecular pathology to demonstrate how personalized therapy, molecular or otherwise, functions in real-world clinical care contexts.

The session was chaired by Dr. Christopher Chen (Stanford Medicine), featuring presentations and panel discussions by Dr. Eric Rosen (Stanford Healthcare), Dr. Uzma Waheed (Stanford Medicine), Dr. Irene Wapnir (Stanford Medicine), and Dr. Carlos Jose (Stanford University School of Medicine).

The transcript below has not been reviewed by the speakers and may contain errors.

Dr. Rosen on Radiologic Imaging Characteristics of Breast Cancer Molecular Subtypes

Breast cancer molecular subtypes demonstrate distinct radiologic appearances that reflect their underlying biology. The three main categories—luminal cancers, HER2-enriched cancers, and basal-like or triple-negative cancers—are commonly discussed in terms of immunohistochemistry (IHC) markers, which serve as surrogates for their DNA phenotype. Understanding these imaging patterns helps inform clinical decision-making and treatment planning.

Luminal A and Luminal B Breast Cancers: Imaging Phenotypes

Luminal cancers, the most common subtype, divide into luminal A and luminal B categories, all characterized by genes that activate estrogen receptor (ER) expression. These subtypes display specific radiologic appearances reflecting their biological behavior.

Luminal A cancers, being slow-growing, typically present as unifocal, irregular, spiculated masses. On mammography, this appears as the classic spiculated lesion. In contrast, luminal B cancers, while retaining some luminal A characteristics, grow more rapidly and usually manifest as more oval masses. They can have spiculated margins, and depending on HER2 overexpression status, may or may not be associated with calcifications.

Ultrasound imaging reveals similar patterns. Luminal A tumors tend to be shadowing masses without significant volume, whereas luminal B tumors, due to rapid growth, demonstrate more volume and ovoid shapes. These differences persist on vascular-based imaging like MRI. Luminal A tumors appear as small, irregular, spiculated masses, while luminal B tumors present as more oval masses with potentially aggressive features like rim enhancement and often associated adenopathy.

HER2-Enriched Breast Cancers: Calcification-Predominant Patterns

HER2-enriched cancers are radiographically characterized by predominant calcifications, usually associated with a high-grade in situ component. A typical presentation involves palpable masses with segmental or near-quadrantal calcifications, along with dense, abnormal-appearing axillary lymph nodes.

Ultrasound demonstrates large masses containing calcifications, with multiple calcifications in surrounding tissue and grossly abnormal axillary lymph nodes. MRI, the optimal imaging modality for breast cancer, often shows diffuse abnormal enhancement throughout the affected quadrant.

HER2-overexpressed cancers demonstrate relative similarity in presentation patterns. Common features include diffuse calcifications involving specific breast quadrants, abnormal lymph nodes on mammography, and spot magnification imaging revealing calcifications associated with multiple irregular masses. These cancers tend to be multifocal or multicentric with extensive mammographic calcifications, frequently associated with axillary disease.

Ultrasound findings mirror mammography, showing multiple small masses containing calcifications scattered throughout the affected quadrant, with axillary lymph nodes lacking normal appearance—exhibiting excessively thick cortex and compressed hilum. MRI provides the most comprehensive assessment, revealing multiple masses involving entire quadrants with non-mass components and linear enhancement associated with high-grade in situ disease.

Triple-Negative Breast Cancer: Deceptively Benign Imaging Appearance

Triple-negative breast cancer can be confounding radiologically, primarily due to rapid growth patterns. These tumors predominantly present in younger women, and their imaging appearance often appears relatively benign on both mammography and ultrasound, the typical first-line imaging modalities.

The typical triple-negative tumor appears as an obscured but oval-appearing mass in young women, mimicking a benign fibroadenoma appearance. On ultrasound, these masses also demonstrate relatively benign characteristics with more lobulations than desired, but appear almost anechoic. Due to rapid growth and a tendency toward necrosis, these masses can actually appear cystic on ultrasound.

Only MRI reveals the truly aggressive nature of these tumors. While adjacent tissue changes may result from biopsy, the mass typically demonstrates peripheral enhancement, a highly suspicious MRI finding. However, the mass itself remains relatively benign in shape and morphology. Additional cases consistently demonstrate this pattern: oval obscured masses on mammography, complex solid-cystic masses on ultrasound, and irregular rim-enhancing masses on MRI.

Breast cancer is now defined by molecular subtype rather than histology alone, and these subtypes all possess unique imaging phenotypes valuable for understanding disease characteristics and behavior.

Dr. Waheed on Imaging Assessment of Treatment Response and Correlation with Pathologic Complete Response

Building upon molecular subtype imaging characteristics, the assessment of response to systemic therapy varies significantly based on molecular subtype, with important discrepancies between imaging assessment and pathologic complete response (PCR).

Molecular Subtype Classifications and Treatment Responsiveness

The surrogate definitions of intrinsic subtypes include luminal A, luminal B, HER2-enriched (or ERBB2 overexpression), and basal subtypes. These categories are not homogeneous. The basal subtype encompasses a heterogeneous group including malignancy categories with better prognosis compared to other triple-negative subtypes, such as adenoid cystic and medullary subtypes.

Treatment approaches differ by molecular subtype. Luminal A tumors respond better to hormone therapy and less to chemotherapy. Luminal B tumors, being more aggressive, show greater chemotherapy responsiveness. Some luminal B tumors that are hormone receptor-positive can also be HER2-positive, qualifying for targeted therapy. HER2-enriched tumors receive Herceptin targeted therapy plus chemotherapy, while basal subtypes currently lack targeted therapies.

MRI Assessment of Complete Response: Limitations and Predictive Values

MRI can demonstrate complete response to neoadjuvant systemic therapy, but imaging assessment varies from the pathologic response. One study showed radiologic PCR in approximately 63% of cases, yielding a positive predictive value of about 63% and a negative predictive value of about 84%. Overall predictive value was inadequate to eliminate surgical excision. This small study included only 129 tumors in 128 patients, requiring cautious interpretation.

A meta-analysis of MRI prediction of complete response showed similar results with a sensitivity of around 60% and a specificity of approximately 90%. Multiple contributing factors affect MRI assessment of neoadjuvant treatment response, including tumor biology, initial tumor size (larger tumors showing less imaging response), chemotherapeutic agent type (particularly angiogenic therapeutic agents demonstrating lower positive predictive values with higher false positives), and morphologic response patterns.

Tumors demonstrating concentric response are more likely to correlate well with pathologic complete response on MRI, whereas non-concentric treatment patterns do not. Confounding factors include necrosis, fibrosis, and inflammation, all producing different enhancement patterns on MRI. MRI technique and magnet strength also impact accuracy, including timing of imaging performance and contrast administration.

Subtype-Specific Imaging Response Patterns

For luminal A and B subtypes, which can also be HER2-positive and tend to be approximately 50-50 unifocal versus multifocal, the highest discordance rates exist between MRI assessment of neoadjuvant treatment response and pathologic complete response. In approximately 30% of cases, MRI underestimates residual disease extent.

HER2-enriched tumors with targeted therapy more commonly demonstrate non-mass enhancement, coinciding with visible calcifications. Systemic response is variable but higher, representing the highest concordance between MRI and pathology at approximately 55-75%. Discordance on surgical pathology is rare, attributable to targeted therapy effectiveness.

MRI accuracy in predicting PCR shows that HER2-positive tumors have very few false negatives, whereas HER2-negative tumors demonstrate much higher false negative rates and lower true negative rates.

The basal subtype, comprising multiple heterogeneous subgroups, typically receives neoadjuvant systemic therapy consideration when tumors exceed five millimeters or demonstrate nodal involvement. Following neoadjuvant treatment, high negative predictive values reach up to 60%, though these studies remain small (most ranging from 50-150 patients).

Current Evidence and Future Directions for Surgical De-escalation

Currently, insufficient evidence exists for MRI to predict pathologic complete response, necessitating continued surgical excision. However, one interesting study examined selective elimination of surgery in a small patient cohort across seven medical centers. Women aged 40 or older with triple-negative or HER2-positive tumors underwent assessment after neoadjuvant systemic therapy.

Patients with no residual disease or less than two centimeters of disease on imaging underwent vacuum-assisted biopsy using nine-gauge devices with at least 12 cores. Patients with negative cores proceeded to radiation without axillary surgery if clinically node-negative at initial treatment, or underwent axillary surgery but not breast surgery if initially node-positive. Patients with residual disease on vacuum-assisted biopsy received standard-of-care breast and axillary surgery.

The study identified 31 of 50 patients with PCR on vacuum-assisted biopsy. Of eight initially node-negative patients with PCR who underwent targeted axillary dissection, none showed axillary lymphadenopathy. At median follow-up of approximately 55 months, no patients experienced recurrence. However, this time period is quite short, requiring more longitudinal analysis. If only removing part of the imaging-identified disease response, new methods for identifying recurrent disease will be necessary, particularly in HER2-positive patients often presenting with extensive calcifications.

Clinical Case Examples Demonstrating Imaging-Pathology Variance

A 55-year-old woman presented with a palpable right breast mass, diagnosed as hormone receptor-positive, HER2-negative with high Ki-67 (luminal B). Imaging revealed a palpable subareolar mass on mammography with a corresponding hypoechoic mass on ultrasound. MRI showed a unifocal mass with fast initial and fast washout kinetics. Following neoadjuvant systemic therapy, the mass appeared smaller but persisted on mammography and ultrasound, also smaller on MRI. The notable finding was altered tumor kinetics post-chemotherapy, reflecting treatment response and antiangiogenic effects visible on MRI.

A 52-year-old woman presented with a palpable right breast lump, diagnosed as HER2-positive. Imaging showed a partially obscured outer right breast mass on mammography, confirmed on ultrasound with a positive lymph node. MRI revealed multifocal disease, including the palpable primary mass, non-mass enhancement extending anteriorly, another non-mass enhancement area in a different quadrant, and multiple smaller masses. Nodal involvement included levels one and two with borderline level three disease. Following neoadjuvant treatment, the patient demonstrated an apparent complete treatment response in the breast and axilla, confirmed as a pathologic complete response.

A 47-year-old woman presented for asymptomatic screening, diagnosed with a HER2-positive left breast tumor. MRI showed a unifocal tumor. Post-neoadjuvant systemic therapy ultrasound revealed a residual area near the biopsy marker, likely fibrosis, though distinguishing from residual disease proved difficult. MRI provided greater certainty, showing no residual enhancement, with the patient achieving a pathologic complete response.

A 39-year-old woman with self-detected triple-negative cancer involving the majority of the upper breast demonstrated a vague hypoechoic mass on ultrasound with a difficult-to-assess true extent. MRI revealed extensive enhancement throughout the entire upper breast with marker and level one adenopathy, confirmed on PET-CT. Post-neoadjuvant systemic therapy, the patient appeared to have a complete response on maximum intensity projection imaging without residual enhancement. However, delayed post-contrast sagittal imaging showed slight enhancement, difficult to distinguish as fibrosis versus residual disease. Unfortunately, pathology revealed multiple tumor cell clusters measuring less than 0.1 centimeters in a fibrotic bed with high RCB (residual cancer burden) of RCB2 and positive lymph nodes at surgery.

In summary, MRI lacks sufficient negative predictive value to eliminate surgical excision, but demonstrates greater accuracy in predicting PCR in HER2-positive and basal subtypes. Neoadjuvant systemic therapeutic agents and initial tumor size are among the factors affecting imaging assessment accuracy compared to the pathologic complete response. Future opportunities may exist to de-escalate therapy and eliminate surgery for selected patients in specific imaging settings.

Dr. Wapnir on Surgical De-escalation in Breast Cancer Treatment

Examining treatment de-escalation proves important and may offer greater applicability to low-resource situations. While not providing a comprehensive review of all de-escalation advances, this discussion aims to provoke thought and conversation about precision surgery applications.

Historical Evolution from Radical Mastectomy to Breast-Conserving Surgery

Breast cancer treatment history has pioneered approaches influencing other cancer treatments. At the end of the 19th century, Halstead and other New York surgeons developed the radical mastectomy, the first operation achieving efficient local disease control. This approach remained predominant for over 50 years.

Radiation therapy was introduced in the 1940s, with the mid-20th century bringing the modified radical mastectomy, a less disfiguring technique modification still used today. However, Bernie Fisher's work fundamentally changed breast cancer treatment by introducing the concept that breast cancer represents systemic disease from early stages. He also transformed research methodology, replacing institutional case series with prospective randomized clinical trials, now the benchmark for medical oncology research.

Fisher's landmark NSABP-B06 trial compared modified radical mastectomy to lumpectomy with axillary node dissection, demonstrating no survival difference and establishing lumpectomy as a viable treatment option. By the 1990s, the NCI declared lumpectomy the preferred approach for early breast cancer.

Lumpectomy application extended through additional studies to stage zero (ductal carcinoma in situ, DCIS) and later to patients presenting with advanced disease receiving upfront neoadjuvant chemotherapy and downstaging. This exemplifies precision surgery, tailoring surgery to tumor response in patients who initially qualified only for mastectomy—a novel concept over 30 years ago.

Evolving Standards for Tumor-Free Lumpectomy Margins

Defining tumor-free lumpectomy margins has occupied research arenas, tumor board discussions, and daily surgical practice. Current recommendations evolved from a systematic review of published series. Margin recommendations for invasive cancer specify no ink on the tumor, while DCIS requires two millimeters.

Recent work has challenged these standards through analysis of an NSABP DCIS study where margin width was prospectively collected on over 3,000 cases. This study examined DCIS treatment in post-menopausal women, requiring estrogen receptor-positive status for entry. The study compared tamoxifen versus anastrozole, completing accrual in 2006 and reporting approximately 10 years later that no outcome differences existed between drugs for DCIS.

This homogeneous patient population enabled retrospective analysis of margin width on local recurrence. Examining different cutoffs (one millimeter and two millimeters), while differences in local recurrence between less than one millimeter versus greater than one millimeter (or using two millimeter cutoffs) were statistically significant, absolute local recurrence differences were 1.6% and 1.5%—differences lacking clinical meaningfulness. These results support treatment de-escalation, suggesting patients should not necessarily undergo a second re-excision lumpectomy based on margin width alone.

Axillary Lymph Node Management De-escalation

Significant de-escalation over recent decades occurred in lymph node management. Historically, all B06 trial patients underwent at least level one and two axillary dissection regardless of clinical node status. Approximately 70% of United States patients were node-negative, prompting the development of sentinel lymphatic mapping and sentinel node biopsy procedures using blue dye and radioisotopes.

A large NSABP study demonstrated superimposed survival curves, whether stopping at negative sentinel nodes or proceeding to axillary dissection, with local recurrence rates below 1% in both groups.

Recent studies expanded the understanding of negative node status. Radiologists began axillary ultrasound interrogation in clinically node-negative patients, hypothesizing that biopsy-confirmed node-positive status would eliminate intraoperative lymphatic mapping necessity. While this approach hasn't widely adopted, preoperative axillary ultrasound evaluation has increased.

The SOUND trial, including over 1,400 patients of any age with tumors less than two centimeters, compared sentinel node biopsy to no sentinel node biopsy in patients deemed clinically node-negative by ultrasound. Among patients undergoing surgery, only 85% were truly node-negative. However, disease-free survival curves showed no difference between groups.

Summary: Practical Implementation Challenges in Precision Oncology

This multidisciplinary session demonstrates that precision oncology implementation requires coordinated approaches across radiology, surgery, and molecular pathology. Key challenges include:

• Discrepancies between imaging assessment and pathologic response across molecular subtypes
• Balancing treatment de-escalation opportunities against ensuring complete disease eradication
• Evolving evidence-based guidelines for surgical intervention based on molecular profiles
• Need for continued validation studies before eliminating established surgical standards
• Integration of imaging, molecular, and clinical data to optimize personalized treatment decisions

The presentations highlight both progress made in precision oncology clinical adoption and ongoing barriers requiring collaborative multidisciplinary solutions to advance personalized cancer care in resource-limited settings.