Faculty Bibliography

Treatment selection in breast cancer is guided by risk assessment using molecular subtypes and clinicopathological characteristics. However, current approaches lack the precision required for optimal clinical decision-making. To address this, we use data from 8161 patients to develop and evaluate an AI test integrating digital pathology with clinical data. The AI test provides a robust method for predicting disease-free interval (C-index: 0.71 [0.68-0.75], HR: 3.63 [3.02-4.37, p < 0.001]). In a direct comparison, the AI test displays numerically higher discrimination (C-index: 0.67 [0.61-0.74]) than the standard-of-care 21-gene assay (C-index: 0.61 [0.49-0.73]). Across molecular subtypes, the AI test demonstrates robust prognostic performance, including in triple negative breast cancer (C-index: 0.71 [0.62-0.81], HR: 3.81 [2.35-6.17, p=0.02]), where no guideline-recommended assays currently exist. These findings highlight the potential of AI-based pathology tests as a promising tool for improved risk stratification across all major subtypes, with implications for clinical decision-making.

INTRODUCTION/BACKGROUND:Oncoplastic breast surgery (OBS) combines plastic surgery techniques with conventional breast-conserving surgery (BCS) and expands BCS eligibility. Limited data are available on patient-reported outcomes (PROs) after OBS. Here we compare long-term PROs after OBS and BCS utilizing the BREAST-Q. PATIENTS AND METHODS/METHODS:Women undergoing OBS or BCS between 2006 and 2019 who completed ≥ 1 long-term BREAST-Q survey 3-5 years postoperatively were identified. Baseline characteristics were compared between women who underwent OBS/BCS. Women who underwent OBS were paired with those who underwent BCS using 1:2 propensity matching [by age, body mass index (BMI), race, T stage, and multifocality]. BREAST-Q scores were compared preoperatively and 3-5 years postoperatively. RESULTS:A total of 297 patients were included for analysis (99 OBS/198 BCS). Women who underwent OBS were younger (p < 0.001) and had higher BMI (p = 0.005) and multifocal disease incidence (p = 0.004). There was no difference between groups in nodal stage, re-excision rates, axillary surgery, chemotherapy, endocrine therapy, or radiotherapy. After propensity matching preoperatively, women who underwent OBS reported lower psychosocial well-being (63 versus 100, p = 0.039) but similar breast satisfaction and sexual well-being compared with women who underwent BCS; however, only three patients who underwent BCS had preoperative BREAST-Q scores available for review. In long-term follow-up, women who underwent OBS reported lower psychosocial scores (74 versus 93, p = 0.011) 4 years postoperatively, but not at 5 years (76 versus 77, p = 0.83). There was no difference in long-term breast satisfaction or sexual well-being. CONCLUSIONS:Women who undergo OBS present with a larger disease burden and may represent a group of non-traditional BCS candidates; they reported similar long-term breast satisfaction and sexual well-being compared with women who undergo BCS. While women who underwent OBS reported lower psychosocial well-being scores preoperatively and during a portion of the follow-up period, this difference was no longer seen at 5 years postoperatively.

Pathology reports are considered the gold standard in medical research due to their comprehensive and accurate diagnostic information. Natural language processing (NLP) techniques have been developed to automate information extraction from pathology reports. However, existing studies suffer from two significant limitations. First, they typically frame their tasks as report classification, which restricts the granularity of extracted information. Second, they often fail to generalize to unseen reports due to variations in language, negation, and human error. To overcome these challenges, we propose a BERT (bidirectional encoder representations from transformers) named entity recognition (NER) system to extract key diagnostic elements from pathology reports. We also introduce four data augmentation methods to improve the robustness of our model. Trained and evaluated on 1438 annotated breast pathology reports, acquired from a large medical center in the United States, our BERT model trained with data augmentation achieves an entity F1-score of 0.916 on an internal test set, surpassing the BERT baseline (0.843). We further assessed the model's generalizability using an external validation dataset from the United Arab Emirates, where our model maintained satisfactory performance (F1-score 0.860). Our findings demonstrate that our NER systems can effectively extract fine-grained information from widely diverse medical reports, offering the potential for large-scale information extraction in a wide range of medical and AI research. We publish our code at https://github.com/nyukat/pathology_extraction.

BACKGROUND:Screening MRI as an adjunct to mammography is recommended by the ACS for patients with a lifetime risk for breast cancer > 20%. While the benefits are clear, MRI screening is associated with an increase in false-positive results. The purpose of this study was to analyze our institutional database of high-risk patients and assess the uptake of screening MRI examinations and the results of those screenings. METHODS:Our institutional review board-approved High-Risk Breast Cancer Database was queried for patients enrolled from January 2017 to January 2023 who were at high risk for breast cancer in a comparative analysis between those who were screened versus not screened with MRIs. Variables of interest included risk factor, background, MRI screening uptake, and frequency and results of image-guided breast biopsies. RESULTS:A total of 254 of 1106 high-risk patients (23%) had MRI screening. Forty-six of 852 (5.3%) patients in the non-MRI-screened cohort and nine of 254 (3.5%) patients in the MRI-screened cohort were diagnosed with a malignant lesion after image-guided biopsy (p = 0.6). There was no significant difference between MRI and non-MRI guided biopsies in detecting breast cancer. All malignant lesions were T1 or in situ disease. The 254 patients in the MRI-screened group underwent 185 biopsies. Fifty-seven percent of MRI-guided biopsies yielded benign results. CONCLUSIONS:Although the addition of MRI screening in our high-risk cohort did not produce a significant number of additional cancer diagnoses, patients monitored in our high-risk cohort who developed breast cancer were diagnosed at very early stages of disease, underscoring the benefit of participation in the program.

INTRODUCTION/BACKGROUND:Acute postoperative pain affects time to opioid cessation and quality of life, and is associated with chronic pain. Effective screening tools are needed to identify patients at increased risk of experiencing more severe acute postoperative pain, and who may benefit from multimodal analgesia and early pain management referral. In this study, we develop a nomogram to preoperatively identify patients at high risk of moderate-severe pain following mastectomy. METHODS:Demographic, psychosocial, and clinical variables were retrospectively assessed in 1195 consecutive patients who underwent mastectomy from January 2019 to December 2020 and had pain scores available from a post-discharge questionnaire. We examined pain severity on postoperative days 1-5, with moderate-severe pain as the outcome of interest. Multivariable logistic regression was performed to identify variables associated with moderate-severe pain in a training cohort of 956 patients. The final model was determined using the Akaike information criterion. A nomogram was constructed using this model, which also included a priori selected clinically relevant variables. Internal validation was performed in the remaining cohort of 239 patients. RESULTS:In the training cohort, 297 patients reported no-mild pain and 659 reported moderate-severe pain. High body mass index (p = 0.042), preoperative Distress Thermometer score ≥4 (p = 0.012), and bilateral surgery (p = 0.003) predicted moderate-severe pain. The resulting nomogram accurately predicted moderate-severe pain in the validation cohort (AUC =  0.735). CONCLUSIONS:This nomogram incorporates eight preoperative variables to provide a risk estimate of acute moderate-severe pain following mastectomy. Preoperative risk stratification can identify patients who may benefit from individually tailored perioperative pain management strategies and early postoperative interventions to treat pain and assist with opioid tapering.

OBJECTIVE:We aimed to investigate eligibility for breast-conserving surgery (BCS) pre- and post-neoadjuvant systemic therapy (NST), and trends in the surgical treatment of young breast cancer patients. BACKGROUND:Young women with breast cancer are more likely to present with larger tumors and aggressive phenotypes, and may benefit from NST. Little is known about how response to NAC influences surgical decisions in young women. METHODS:The Young Women's Breast Cancer Study (YWS), a multicenter prospective cohort of women diagnosed with breast cancer at age ≤40, enrolled 1302 patients from 2006 to 2016. Disease characteristics, surgical recommendations, and reasons for choosing mastectomy among BCS-eligible patients were obtained through the medical record. Trends in use of NST, rate of clinical and pathologic complete response (cCR and pCR), and surgery were also assessed. RESULTS:Of 1117 women with unilateral stage I-III breast cancer, 315 (28%) received NST. Pre-NST, 26% were BCS eligible, 17% were borderline eligible, and 55% were ineligible. After NST, BCS eligibility increased from 26% to 42% (p < 0.0001). Among BCS-eligible patients after NST (n = 133), 41% chose mastectomy with reasons being patient preference (53%), BRCA or TP53 mutation (35%) and family history (5%). From 2006 to 2016, the rates of NST (p = 0.0012), cCR (p < 0.0001) and bilateral mastectomy (p < 0.0001) increased, but the rate of BCS did not increase (p = 0.34). CONCLUSION/CONCLUSIONS:While the proportion of young women eligible for BCS increased after NST, many patients choose mastectomy, suggesting that surgical decisions are often driven by factors beyond extent of disease and treatment response.

BACKGROUND:Most patients recur after resection of intrahepatic cholangiocarcinoma (IHC). We studied whether machine-learning incorporating radiomics and tumor size could predict intrahepatic recurrence within 1-year. METHODS:This was a retrospective analysis of patients with IHC resected between 2000 and 2017 who had evaluable computed tomography imaging. Texture features (TFs) were extracted from the liver, tumor, and future liver remnant (FLR). Random forest classification using training (70.3%) and validation cohorts (29.7%) was used to design a predictive model. RESULTS:138 patients were included for analysis. Patients with early recurrence had a larger tumor size (7.25 cm [IQR 5.2-8.9] vs. 5.3 cm [IQR 4.0-7.2], P = 0.011) and a higher rate of lymph node metastasis (28.6% vs. 11.6%, P = 0.041), but were not more likely to have multifocal disease (21.4% vs. 17.4%, P = 0.643). Three TFs from the tumor, FD1, FD30, and IH4 and one from the FLR, ACM15, were identified by feature selection. Incorporation of TFs and tumor size achieved the highest AUC of 0.84 (95% CI 0.73-0.95) in predicting recurrence in the validation cohort. CONCLUSION/CONCLUSIONS:This study demonstrates that radiomics and machine-learning can reliably predict patients at risk for early intrahepatic recurrence with good discrimination accuracy.

INTRODUCTION/UNASSIGNED:As breast cancer treatment options have multiplied and biologic diversity within breast cancer has been recognized, the use of the same treatment strategies for patients with early-stage and favorable disease, and for those with biologically aggressive disease, has been questioned. In addition, as patient-reported outcome measures have called attention to the morbidity of many common treatments, and as the cost of breast cancer care has continued to increase, reduction in the overtreatment of breast cancer has assumed increasing importance. AREAS COVERED/UNASSIGNED:Here we review selected aspects of surgery, radiation oncology, and medical oncology for which scientific evidence supports de-escalation for invasive carcinoma and ductal carcinoma in situ, and assess strategies to address overtreatment. EXPERT OPINION/UNASSIGNED:The problems of breast cancer overtreatment we face today are based on improved understanding of the biology of breast cancer and abandonment of the 'one-size-fits-all' approach. As breast cancer care becomes increasingly complex, and as our knowledge base continues to increase exponentially, these problems will only be magnified in the future. To continue progress, the move must be made from advocating the maximum-tolerated treatment to advocating the minimum-effective one.

PURPOSE/OBJECTIVE:Synchronous bilateral breast cancer is uncommon, and its pattern and incidence among younger women is unknown. Here we report the incidence, phenotypes, and long-term oncologic outcomes of bilateral breast cancer in women enrolled in the Young Women's Breast Cancer Study (YWS). METHODS:The YWS is a multi-center, prospective cohort study of women with breast cancer diagnosed at age ≤ 40 years. Those with synchronous bilateral breast cancer formed our study cohort. Tumor phenotypes were categorized as luminal A (hormone receptor (HR)+/HER2-/grade 1/2), luminal B (HR+ /HER2+ or HER2- and grade 3), HER2-enriched (HR-/HER2+), or basal-like (HR-/HER2-). Descriptive statistics were used to evaluate tumor phenotypes of bilateral cancers for concordance. RESULTS:Among 1302 patients enrolled in the YWS, 21 (1.6%) patients had synchronous bilateral disease. The median age of diagnosis was 38 years (range 18-40 years). Seventeen (81.0%) underwent genetic testing with 6 found to have pathogenic germline mutations in BRCA1, BRCA2, or TP53. The majority of patients (76.2%) underwent bilateral mastectomy. On pathology, 2 patients had bilateral in-situ disease, 6 had unilateral invasive and contralateral in-situ disease, and 13 had bilateral invasive disease. Of those with bilateral invasive disease, 10 (76.9%) had bilateral luminal tumors and, when fully characterized, 6 were of the same luminal subtype. Only 1 patient had bilateral basal-like breast cancer. At median follow-up of 8.2 years, 14 patients are alive with no recurrent disease. CONCLUSIONS:Bilateral breast cancer is uncommon among young women diagnosed with breast cancer at age ≤ 40. In our cohort, the majority of invasive tumors were of the luminal phenotype, though some differed by grade or HER2 status. These findings support the need for thorough pathologic workup of bilateral disease when it is found in young women with breast cancer to determine risk and tailor treatment.