Slideflow: deep learning for digital histopathology with real-time whole-slide visualization.

Deep learning methods have emerged as powerful tools for analyzing histopathological images, but current methods are often specialized for specific domains and software environments, and few open-source options exist for deploying models in an interactive interface. Experimenting with different deep learning approaches typically requires switching software libraries and reprocessing data, reducing the feasibility and practicality of experimenting with new architectures. We developed a flexible deep learning library for histopathology called Slideflow, a package which supports a broad array of deep learning methods for digital pathology and includes a fast whole-slide interface for deploying trained models. Slideflow includes unique tools for whole-slide image data processing, efficient stain normalization and augmentation, weakly-supervised whole-slide classification, uncertainty quantification, feature generation, feature space analysis, and explainability. Whole-slide image processing is highly optimized, enabling whole-slide tile extraction at 40x magnification in 2.5 s per slide. The framework-agnostic data processing pipeline enables rapid experimentation with new methods built with either Tensorflow or PyTorch, and the graphical user interface supports real-time visualization of slides, predictions, heatmaps, and feature space characteristics on a variety of hardware devices, including ARM-based devices such as the Raspberry Pi.

Validation of the RSClin risk calculator in the National Cancer Data Base.

BACKGROUND: Guidelines recommend the use of genomic assays such as OncotypeDx to aid in decisions regarding the use of chemotherapy for hormone receptor-positive, HER2-negative (HR+/HER2-) breast cancer. The RSClin prognostic tool integrates OncotypeDx and clinicopathologic features to predict distant recurrence and chemotherapy benefit, but further validation is needed before broad clinical adoption. METHODS: This study included patients from the National Cancer Data Base (NCDB) who were diagnosed with stage I-III HR+/HER2- breast cancer from 2010 to 2020 and received adjuvant endocrine therapy with or without chemotherapy. RSClin-predicted chemotherapy benefit was stratified into low (<3% reduction in distant recurrence), intermediate (3%-5%), and high (>5%). Cox models were used to model mortality adjusted for age, comorbidity index, insurance, and race/ethnicity. RESULTS: A total of 285,441 patients were identified for inclusion from the NCDB, with an average age of 60 years and a median follow-up of 58 months. Chemotherapy was associated with improved overall survival only for those predicted to have intermediate (adjusted hazard ratio [aHR], 0.68; 95% confidence interval [CI], 0.60-0.79) and high benefit per RSClin (aHR, 0.66; 95% CI, 0.61-0.72). Consistent benefit was seen in the subset with a low OncotypeDx score (<26) and intermediate (aHR, 0.66; 95% CI, 0.53-0.82) or high (aHR, 0.71; 95% CI, 0.58-0.86) RSClin-predicted benefit. No survival benefit with chemotherapy was seen in patients with a high OncotypeDx score (≥26) and low benefit per RSClin (aHR, 1.70; 95% CI, 0.41-6.99). CONCLUSIONS: RSClin may identify high-risk patients who benefit from treatment intensification more accurately than OncotypeDx, and further prospective study is needed.

Quantitative Biomarkers, Genomic Assays, and Demographics Associated with Breast-Conserving Surgery Following Neoadjuvant Therapy in Early-Stage, Hormone Receptor-Positive, HER-Negative Breast Cancer.

BACKGROUND: Given increased neoadjuvant therapy use in early-stage, hormone receptor (HR)-positive/HER2-negative breast cancer, we sought to quantify likelihood of breast-conserving surgery (BCS) after neoadjuvant chemotherapy (NACT) or endocrine therapy (NET) as a function of ER%/PR%/Ki-67%, 21-gene recurrence scores (RS), or 70-gene risk groups. METHODS: We analyzed the 2010-2020 National Cancer Database. Surgery was categorized as "mastectomy/BCS." Logistic regression was performed. Adjusted odds ratios (AOR) were per 10-unit increase in ER%/PR%/Ki-67%. RESULTS: Overall, 42.3% underwent BCS after NACT, whereas 64.0% did after NET. Increasing ER% (AOR = 0.96, 95% confidence interval [CI] 0.94-0.97) or PR% (AOR=0.98, 95% CI 0.96-0.99) was associated with lower odds of BCS after NACT. Increasing Ki-67% was associated with greater odds of BCS (AOR = 1.07, 95% CI 1.04-1.10). Breast-conserving surgery rates increased by ~20 percentage points, with Ki-67% ≥15 or RS >20. Patients with a low (43.0%, AOR = 0.50, 95% CI 0.29-0.88) or intermediate (46.4%, AOR = 0.58, 95% CI 0.41-0.81) RS were less likely than patients with a high RS (65.0%) to undergo BCS after NACT. Increasing ER% was associated with higher odds of BCS after NET (AOR = 1.09, 95% CI 1.01-1.17). Breast-conserving surgery rates increased by ~20 percentage points between ER <50% and >80%. In both cohorts, the odds of BCS were similar between 70-gene low-risk and high-risk groups. Asian or uninsured patients had lower odds of BCS. CONCLUSIONS: Neoadjuvant chemotherapy is unlikely to downstage tumors with a low-intermediate RS, higher ER%/PR%, or lower Ki-67%. Breast-conserving surgery after NET was most dependent on ER%. Findings could facilitate treatment decision-making based on tumor biology and racial/socioeconomic disparities and improve patient counseling on the likelihood of successful BCS.

Molecular profiling of a real-world breast cancer cohort with genetically inferred ancestries reveals actionable tumor biology differences between European ancestry and African ancestry patient populations.

BACKGROUND: Endocrine-resistant HR+/HER2- breast cancer (BC) and triple-negative BC (TNBC) are of interest for molecularly informed treatment due to their aggressive natures and limited treatment profiles. Patients of African Ancestry (AA) experience higher rates of TNBC and mortality than European Ancestry (EA) patients, despite lower overall BC incidence. Here, we compare the molecular landscapes of AA and EA patients with HR+/HER2- BC and TNBC in a real-world cohort to promote equity in precision oncology by illuminating the heterogeneity of potentially druggable genomic and transcriptomic pathways. METHODS: De-identified records from patients with TNBC or HR+/HER2- BC in the Tempus Database were randomly selected (N = 5000), with most having stage IV disease. Mutations, gene expression, and transcriptional signatures were evaluated from next-generation sequencing data. Genetic ancestry was estimated from DNA-seq. Differences in mutational prevalence, gene expression, and transcriptional signatures between AA and EA were compared. EA patients were used as the reference population for log fold-changes (logFC) in expression. RESULTS: After applying inclusion criteria, 3433 samples were evaluated (n = 623 AA and n = 2810 EA). Observed patterns of dysregulated pathways demonstrated significant heterogeneity among the two groups. Notably, PIK3CA mutations were significantly lower in AA HR+/HER2- tumors (AA = 34% vs. EA = 42%, P < 0.05) and the overall cohort (AA = 28% vs. EA = 37%, P = 2.08e-05). Conversely, KMT2C mutation was significantly more frequent in AA than EA TNBC (23% vs. 12%, P < 0.05) and HR+/HER2- (24% vs. 15%, P = 3e-03) tumors. Across all subtypes and stages, over 8000 genes were differentially expressed between the two ancestral groups including RPL10 (logFC = 2.26, P = 1.70e-162), HSPA1A (logFC = - 2.73, P = 2.43e-49), ATRX (logFC = - 1.93, P = 5.89e-83), and NUTM2F (logFC = 2.28, P = 3.22e-196). Ten differentially expressed gene sets were identified among stage IV HR+/HER2- tumors, of which four were considered relevant to BC treatment and were significantly enriched in EA: ERBB2_UP.V1_UP (P = 3.95e-06), LTE2_UP.V1_UP (P = 2.90e-05), HALLMARK_FATTY_ACID_METABOLISM (P = 0.0073), and HALLMARK_ANDROGEN_RESPONSE (P = 0.0074). CONCLUSIONS: We observed significant differences in mutational spectra, gene expression, and relevant transcriptional signatures between patients with genetically determined African and European ancestries, particularly within the HR+/HER2- BC and TNBC subtypes. These findings could guide future development of treatment strategies by providing opportunities for biomarker-informed research and, ultimately, clinical decisions for precision oncology care in diverse populations.

The impact of race and age on response to neoadjuvant therapy and long-term outcomes in Black and White women with early-stage breast cancer.

PURPOSE: There are a paucity of data and a pressing need to evaluate response to neoadjuvant chemotherapy (NACT) and determine long-term outcomes in young Black women with early-stage breast cancer (EBC). METHODS: We analyzed data from 2196 Black and White women with EBC treated at the University of Chicago over the last 2 decades. Patients were divided into groups based on race and age at diagnosis: Black women [Formula: see text] 40 years, White women [Formula: see text] 40 years, Black women [Formula: see text] 55 years, and White women [Formula: see text] 55 years. Pathological complete response rate (pCR) was analyzed using logistic regression. Overall survival (OS) and disease-free survival (DFS) were analyzed using Cox proportional hazard and piecewise Cox models. RESULTS: Young Black women had the highest risk of recurrence, which was 22% higher than young White women (p = 0.434) and 76% higher than older Black women (p = 0.008). These age/racial differences in recurrence rates were not statistically significant after adjusting for subtype, stage, and grade. In terms of OS, older Black women had the worst outcome. In the 397 women receiving NACT, 47.5% of young White women achieved pCR, compared to 26.8% of young Black women (p = 0.012). CONCLUSIONS: Black women with EBC had significantly worse outcomes compared to White women in our cohort study. There is an urgent need to understand the disparities in outcomes between Black and White breast cancer patients, particularly in young women where the disparity in outcome is the greatest.

Clinical trials of immunotherapy in triple-negative breast cancer.

PURPOSE: Immunotherapy has started to transform the treatment of triple-negative breast cancer (TNBC), in part due to the unique immunogenicity of this breast cancer subtype. This review summarizes clinical studies of immunotherapy in advanced and early-stage TNBC. FINDINGS: Initial studies of checkpoint blockade monotherapy demonstrated occasional responses, especially in patients with untreated programmed death-ligand 1 (PD-L1) positive advanced TNBC, but failed to confirm a survival advantage over chemotherapy. Nonetheless, pembrolizumab monotherapy has tumor agnostic approval for microsatellite instability-high or high tumor mutational burden cancers, and thus can be considered for select patients with advanced TNBC. Combination chemoimmunotherapy approaches have been more successful, and pembrolizumab is approved for PD-L1 positive advanced TNBC in combination with chemotherapy. This success has been translated to the curative setting, where pembrolizumab is now approved in combination with neoadjuvant chemotherapy for high-risk early-stage TNBC. CONCLUSION: Immunotherapy has been a welcome addition to the growing armamentarium for TNBC, but responses remain limited to a subset of patients. Innovative strategies are under investigation in an attempt to induce immune responses in resistant tumors-with regimens incorporating small-molecule inhibitors, novel immune checkpoint targets, and intratumoral injections that directly alter the tumor microenvironment. As the focus shifts toward the use of immunotherapy for early-stage TNBC, it will be critical to identify those who derive the most benefit from treatment, given the potential for irreversible autoimmune toxicity and the lack of predictive accuracy of PD-L1 expression in the early-stage setting.

Highly accurate response prediction in high-risk early breast cancer patients using a biophysical simulation platform.

PURPOSE: Pathologic complete response (pCR) to neoadjuvant chemotherapy (NAC) in early breast cancer (EBC) is largely dependent on breast cancer subtype, but no clinical-grade model exists to predict response and guide selection of treatment. A biophysical simulation of response to NAC has the potential to address this unmet need. METHODS: We conducted a retrospective evaluation of a biophysical simulation model as a predictor of pCR. Patients who received standard NAC at the University of Chicago for EBC between January 1st, 2010 and March 31st, 2020 were included. Response was predicted using baseline breast MRI, clinicopathologic features, and treatment regimen by investigators who were blinded to patient outcomes. RESULTS: A total of 144 tumors from 141 patients were included; 59 were triple-negative, 49 HER2-positive, and 36 hormone-receptor positive/HER2 negative. Lymph node disease was present in half of patients, and most were treated with an anthracycline-based regimen (58.3%). Sensitivity and specificity of the biophysical simulation for pCR were 88.0% (95% confidence interval [CI] 75.7 - 95.5) and 89.4% (95% CI 81.3 - 94.8), respectively, with robust results regardless of subtype. In patients with predicted pCR, 5-year event-free survival was 98%, versus 79% with predicted residual disease (log-rank p = 0.01, HR 4.57, 95% CI 1.36 - 15.34). At a median follow-up of 5.4 years, no patients with predicted pCR experienced disease recurrence. CONCLUSION: A biophysical simulation model accurately predicts pCR and long-term outcomes from baseline MRI and clinical data, and is a promising tool to guide escalation/de-escalation of NAC.

Characterizing the Increase in Artificial Intelligence Content Detection in Oncology Scientific Abstracts From 2021 to 2023.

PURPOSE: Artificial intelligence (AI) models can generate scientific abstracts that are difficult to distinguish from the work of human authors. The use of AI in scientific writing and performance of AI detection tools are poorly characterized. METHODS: We extracted text from published scientific abstracts from the ASCO 2021-2023 Annual Meetings. Likelihood of AI content was evaluated by three detectors: GPTZero, Originality.ai, and Sapling. Optimal thresholds for AI content detection were selected using 100 abstracts from before 2020 as negative controls, and 100 produced by OpenAI's GPT-3 and GPT-4 models as positive controls. Logistic regression was used to evaluate the association of predicted AI content with submission year and abstract characteristics, and adjusted odds ratios (aORs) were computed. RESULTS: Fifteen thousand five hundred and fifty-three abstracts met inclusion criteria. Across detectors, abstracts submitted in 2023 were significantly more likely to contain AI content than those in 2021 (aOR range from 1.79 with Originality to 2.37 with Sapling). Online-only publication and lack of clinical trial number were consistently associated with AI content. With optimal thresholds, 99.5%, 96%, and 97% of GPT-3/4-generated abstracts were identified by GPTZero, Originality, and Sapling respectively, and no sampled abstracts from before 2020 were classified as AI generated by the GPTZero and Originality detectors. Correlation between detectors was low to moderate, with Spearman correlation coefficient ranging from 0.14 for Originality and Sapling to 0.47 for Sapling and GPTZero. CONCLUSION: There is an increasing signal of AI content in ASCO abstracts, coinciding with the growing popularity of generative AI models.

Assessing the Relationship Between Neighborhood Socioeconomic Disadvantage and Telemedicine Use Among Patients With Breast Cancer and Examining Differential Provisions of Oncology Services Between Telehealth and In-Person Visits: Quantitative Study.

BACKGROUND: Since the COVID-19 pandemic began, we have seen rapid growth in telemedicine use. However, telehealth care and services are not equally distributed, and not all patients with breast cancer have equal access across US regions. There are notable gaps in existing literature regarding the influence of neighborhood-level socioeconomic status on telemedicine use in patients with breast cancer and oncology services offered through telehealth versus in-person visits. OBJECTIVE: We assessed the relationship between neighborhood socioeconomic disadvantage and telemedicine use among patients with breast cancer and examined differential provisions of oncology services between telehealth and in-person visits. METHODS: Neighborhood socioeconomic disadvantage was measured using the Area Deprivation Index (ADI), with higher scores indicating greater disadvantages. Telemedicine and in-person visits were defined as having had a telehealth and in-person visit with a provider, respectively, in the past 12 months. Multivariable logistic regression was performed to examine the association between ADI and telemedicine use. The McNemar test was used to assess match-paired data on types of oncology services comparing telehealth and in-person visits. RESULTS: The mean age of the patients with breast cancer (n=1163) was 61.8 (SD 12.0) years; 4.58% (52/1161) identified as Asian, 19.72% (229/1161) as Black, 3.01% (35/1161) as Hispanic, and 72.78% (845/1161) as White. Overall, 35.96% (416/1157) had a telemedicine visit in the past 12 months. Of these patients, 65% (266/409) had a videoconference visit only, 22.7% (93/409) had a telephone visit only, and 12.2% (50/409) had visits by both videoconference and telephone. Higher ADI scores were associated with a lower likelihood of telemedicine use (adjusted odds ratio [AOR] 0.89, 95% CI 0.82-0.97). Black (AOR 2.38, 95% CI 1.41-4.00) and Hispanic (AOR 2.65, 95% CI 1.07-6.58) patients had greater odds of telemedicine use than White patients. Compared to patients with high school or less education, those with an associate's degree (AOR 2.67, 95% CI 1.33-5.35), a bachelor's degree (AOR 2.75, 95% CI 1.38-5.48), or a graduate or professional degree (AOR 2.57, 95% CI 1.31-5.04) had higher odds of telemedicine use in the past 12 months. There were no significant differences in providing treatment consultation (45/405, 11.1% vs 55/405, 13.6%; P=.32) or cancer genetic counseling (11/405, 2.7% vs 19/405, 4.7%; P=.14) between telehealth and in-person visits. Of the telemedicine users, 95.8% (390/407) reported being somewhat to extremely satisfied, and 61.8% (254/411) were likely or very likely to continue using telemedicine. CONCLUSIONS: In this study of a multiethnic cohort of patients with breast cancer, our findings suggest that neighborhood-level socioeconomic disparities exist in telemedicine use and that telehealth visits could be used to provide treatment consultation and cancer genetic counseling. Oncology programs should address these disparities and needs to improve care delivery and achieve telehealth equity for their patient populations.

Evaluation of large language models as a diagnostic aid for complex medical cases.

Background: The use of large language models (LLM) has recently gained popularity in diverse areas, including answering questions posted by patients as well as medical professionals. Objective: To evaluate the performance and limitations of LLMs in providing the correct diagnosis for a complex clinical case. Design: Seventy-five consecutive clinical cases were selected from the Massachusetts General Hospital Case Records, and differential diagnoses were generated by OpenAI's GPT3.5 and 4 models. Results: The mean number of diagnoses provided by the Massachusetts General Hospital case discussants was 16.77, by GPT3.5 30 and by GPT4 15.45 (p < 0.0001). GPT4 was more frequently able to list the correct diagnosis as first (22% versus 20% with GPT3.5, p = 0.86), provide the correct diagnosis among the top three generated diagnoses (42% versus 24%, p = 0.075). GPT4 was better at providing the correct diagnosis, when the different diagnoses were classified into groups according to the medical specialty and include the correct diagnosis at any point in the differential list (68% versus 48%, p = 0.0063). GPT4 provided a differential list that was more similar to the list provided by the case discussants than GPT3.5 (Jaccard Similarity Index 0.22 versus 0.12, p = 0.001). Inclusion of the correct diagnosis in the generated differential was correlated with PubMed articles matching the diagnosis (OR 1.40, 95% CI 1.25-1.56 for GPT3.5, OR 1.25, 95% CI 1.13-1.40 for GPT4), but not with disease incidence. Conclusions and relevance: The GPT4 model was able to generate a differential diagnosis list with the correct diagnosis in approximately two thirds of cases, but the most likely diagnosis was often incorrect for both models. In its current state, this tool can at most be used as an aid to expand on potential diagnostic considerations for a case, and future LLMs should be trained which account for the discrepancy between disease incidence and availability in the literature.

Developing a low-cost, open-source, locally manufactured workstation and computational pipeline for automated histopathology evaluation using deep learning.

BACKGROUND: Deployment and access to state-of-the-art precision medicine technologies remains a fundamental challenge in providing equitable global cancer care in low-resource settings. The expansion of digital pathology in recent years and its potential interface with diagnostic artificial intelligence algorithms provides an opportunity to democratize access to personalized medicine. Current digital pathology workstations, however, cost thousands to hundreds of thousands of dollars. As cancer incidence rises in many low- and middle-income countries, the validation and implementation of low-cost automated diagnostic tools will be crucial to helping healthcare providers manage the growing burden of cancer. METHODS: Here we describe a low-cost ($230) workstation for digital slide capture and computational analysis composed of open-source components. We analyze the predictive performance of deep learning models when they are used to evaluate pathology images captured using this open-source workstation versus images captured using common, significantly more expensive hardware. Validation studies assessed model performance on three distinct datasets and predictive models: head and neck squamous cell carcinoma (HPV positive versus HPV negative), lung cancer (adenocarcinoma versus squamous cell carcinoma), and breast cancer (invasive ductal carcinoma versus invasive lobular carcinoma). FINDINGS: When compared to traditional pathology image capture methods, low-cost digital slide capture and analysis with the open-source workstation, including the low-cost microscope device, was associated with model performance of comparable accuracy for breast, lung, and HNSCC classification. At the patient level of analysis, AUROC was 0.84 for HNSCC HPV status prediction, 1.0 for lung cancer subtype prediction, and 0.80 for breast cancer classification. INTERPRETATION: Our ability to maintain model performance despite decreased image quality and low-power computational hardware demonstrates that it is feasible to massively reduce costs associated with deploying deep learning models for digital pathology applications. Improving access to cutting-edge diagnostic tools may provide an avenue for reducing disparities in cancer care between high- and low-income regions. FUNDING: Funding for this project including personnel support was provided via grants from NIH/NCIR25-CA240134, NIH/NCIU01-CA243075, NIH/NIDCRR56-DE030958, NIH/NCIR01-CA276652, NIH/NCIK08-CA283261, NIH/NCI-SOAR25CA240134, SU2C (Stand Up to Cancer) Fanconi Anemia Research Fund - Farrah Fawcett Foundation Head and Neck Cancer Research Team Grant, and the European UnionHorizon Program (I3LUNG).

Development and validation of a clinical breast cancer tool for accurate prediction of recurrence.

Given high costs of Oncotype DX (ODX) testing, widely used in recurrence risk assessment for early-stage breast cancer, studies have predicted ODX using quantitative clinicopathologic variables. However, such models have incorporated only small cohorts. Using a cohort of patients from the National Cancer Database (NCDB, n = 53,346), we trained machine learning models to predict low-risk (0-25) or high-risk (26-100) ODX using quantitative estrogen receptor (ER)/progesterone receptor (PR)/Ki-67 status, quantitative ER/PR status alone, and no quantitative features. Models were externally validated on a diverse cohort of 970 patients (median follow-up 55 months) for accuracy in ODX prediction and recurrence. Comparing the area under the receiver operating characteristic curve (AUROC) in a held-out set from NCDB, models incorporating quantitative ER/PR (AUROC 0.78, 95% CI 0.77-0.80) and ER/PR/Ki-67 (AUROC 0.81, 95% CI 0.80-0.83) outperformed the non-quantitative model (AUROC 0.70, 95% CI 0.68-0.72). These results were preserved in the validation cohort, where the ER/PR/Ki-67 model (AUROC 0.87, 95% CI 0.81-0.93, p = 0.009) and the ER/PR model (AUROC 0.86, 95% CI 0.80-0.92, p = 0.031) significantly outperformed the non-quantitative model (AUROC 0.80, 95% CI 0.73-0.87). Using a high-sensitivity rule-out threshold, the non-quantitative, quantitative ER/PR and ER/PR/Ki-67 models identified 35%, 30% and 43% of patients as low-risk in the validation cohort. Of these low-risk patients, fewer than 3% had a recurrence at 5 years. These models may help identify patients who can forgo genomic testing and initiate endocrine therapy alone. An online calculator is provided for further study.

Generative Adversarial Networks Accurately Reconstruct Pan-Cancer Histology from Pathologic, Genomic, and Radiographic Latent Features.

Artificial intelligence models have been increasingly used in the analysis of tumor histology to perform tasks ranging from routine classification to identification of novel molecular features. These approaches distill cancer histologic images into high-level features which are used in predictions, but understanding the biologic meaning of such features remains challenging. We present and validate a custom generative adversarial network - HistoXGAN - capable of reconstructing representative histology using feature vectors produced by common feature extractors. We evaluate HistoXGAN across 29 cancer subtypes and demonstrate that reconstructed images retain information regarding tumor grade, histologic subtype, and gene expression patterns. We leverage HistoXGAN to illustrate the underlying histologic features for deep learning models for actionable mutations, identify model reliance on histologic batch effect in predictions, and demonstrate accurate reconstruction of tumor histology from radiographic imaging for a 'virtual biopsy'.

Datopotamab-deruxtecan plus durvalumab in early-stage breast cancer: the sequential multiple assignment randomized I-SPY2.2 phase 2 trial.

Sequential adaptive trial designs can help accomplish the goals of personalized medicine, optimizing outcomes and avoiding unnecessary toxicity. Here we describe the results of incorporating a promising antibody-drug conjugate, datopotamab-deruxtecan (Dato-DXd) in combination with programmed cell death-ligand 1 inhibitor, durvalumab, as the first sequence of therapy in the I-SPY2.2 phase 2 neoadjuvant sequential multiple assignment randomization trial for high-risk stage 2/3 breast cancer. The trial includes three blocks of treatment, with initial randomization to different experimental agent(s) (block A), followed by a taxane-based regimen tailored to tumor subtype (block B), followed by doxorubicin-cyclophosphamide (block C). Subtype-specific algorithms based on magnetic resonance imaging volume change and core biopsy guide treatment redirection after each block, including the option of early surgical resection in patients predicted to have a high likelihood of pathologic complete response, which is the primary endpoint assessed when resection occurs. There are two primary efficacy analyses: after block A and across all blocks for six prespecified HER2-negative subtypes (defined by hormone receptor status and/or response-predictive subtypes). In total, 106 patients were treated with Dato-DXd/durvalumab in block A. In the immune-positive subtype, Dato-DXd/durvalumab exceeded the prespecified threshold for success (graduated) after block A; and across all blocks, pathologic complete response rates were equivalent to the rate expected for the standard of care (79%), but 54% achieved that result after Dato-DXd/durvalumab alone (block A) and 92% without doxorubicin-cyclophosphamide (after blocks A + B). The treatment strategy across all blocks graduated in the hormone-negative/immune-negative subtype. No new toxicities were observed. Stomatitis was the most common side effect in block A. No patients receiving block A treatment alone had adrenal insufficiency. Dato-DXd/durvalumab is a promising therapy combination that can eliminate standard chemotherapy in many patients, particularly the immune-positive subtype.ClinicalTrials.gov registration: NCT01042379 .

Response to Treatment, Racial and Ethnic Disparity, and Survival in Patients With Breast Cancer Undergoing Neoadjuvant Chemotherapy in the US.

Importance: With the increasing delivery of neoadjuvant chemotherapy (NACT) for patients with breast cancer in the US, it is important to know whether there is differential response to NACT by race and ethnicity and the potential long-term outcomes. Objective: To examine whether there were any racial and ethnic differences in pathologic complete response (pCR) rate following NACT and, if so, whether they varied by molecular subtype and were associated with survival. Design, Setting, and Participants: A retrospective cohort study was conducted including patients with stage I to III breast cancer diagnosed between January 2010 and December 2017 who underwent surgery and received NACT; median follow-up was 5.8 years, and data analysis was conducted from August 2021 to January 2023. Data were obtained from the National Cancer Data Base, a nationwide, facility-based, oncology data set that captures approximately 70% of all newly diagnosed cases of breast cancer in the US. Main Outcomes and Measures: Pathologic complete response, defined as ypT0/Tis ypN0, was modeled using logistic regression. Racial and ethnic differences in survival were analyzed using a Weibull accelerated failure time model. Mediation analysis was conducted to measure whether racial and ethnic differences in the pCR rate affect survival. Results: The study included 107 207 patients (106 587 [99.4%] women), with a mean (SD) age of 53.4 (12.1) years. A total of 5009 patients were Asian or Pacific Islander, 18 417 were non-Hispanic Black, 9724 were Hispanic, and 74 057 were non-Hispanic White. There were significant racial and ethnic differences in pCR rates, but the differences were subtype-specific. In hormone receptor-negative (HR-)/erb-b2 receptor tyrosine kinase 2 (ERBB2; formerly HER2 or HER2/neu)-positive (ERBB2+) subtype, Asian and Pacific Islander patients achieved the highest pCR rate (56.8%), followed by Hispanic (55.2%) and non-Hispanic White (52.3%) patients with the lowest pCR rate seen in Black patients (44.8%). In triple-negative breast cancer, Black patients had a lower pCR rate (27.3%) than other racial and ethnic groups (all >30%). In HR+/ERBB2- subtype, Black patients had a higher pCR rate (11.3%) than other racial/ethnic groups (all ≤10%). In mediation analysis, racial and ethnic differences in achieving pCR after NACT could explain approximately 20% to 53% of the subtype-specific survival differences across racial and ethnic groups. Conclusions and Relevance: In this cohort study of patients with breast cancer receiving NACT, Black patients had a lower pCR rate for triple-negative and HR-/ERBB2+ breast cancer but a higher pCR rate for HR+/ERBB2- diseases, whereas Asian and Pacific Islander patients had a higher pCR rate for HR-/ERBB2+ diseases. Tumor grade and ERBB2 copy number could account for some of these within-subtype disparities, but further studies are warranted. Inability to achieve a pCR can mediate in part, but not entirely, the worse survival outcomes experienced by Black patients.

Evaluation of multigene assays as predictors for response to neoadjuvant chemotherapy in early-stage breast cancer patients.

OncotypeDX and MammaPrint assays have not been validated to predict pathologic complete response (pCR) to neoadjuvant chemotherapy (NACT) in early-stage breast cancer patients. We analyzed the 2010-2019 National Cancer Database and found that high OncotypeDX recurrence scores or high MammaPrint scores were associated with greater odds of pCR. Our findings suggest that OncotypeDX and MammaPrint testing predict pCR after NACT and could facilitate clinical decision-making between clinicians and patients.

Bilateral asymmetry of quantitative parenchymal kinetics at ultrafast DCE-MRI predict response to neoadjuvant chemotherapy in patients with HER2+ breast cancer.

PURPOSE: To assess whether measurement of the bilateral asymmetry of semiquantitative and quantitative perfusion parameters from ultrafast dynamic contrast-enhanced MRI (DCE-MRI), allows early prediction of pathologic response after neoadjuvant chemotherapy (NAC) in patients with HER2+ breast cancer. MATERIALS AND METHODS: Twenty-eight female patients with HER2+ breast cancer treated with NAC who underwent pre-NAC ultrafast DCE-MRI (3-9 s/phase) were enrolled for this study. Four semiquantitative and two quantitative parenchymal parameters were calculated for each patient. Ipsilateral/contralateral (I/C) ratio (for four parameters) and the difference between (for two parameters) ipsi- and contra-lateral parenchymal kinetic parameters (kBPE) were compared for patients with pathologic complete response (pCR) and those having residual disease. Lasso regression with leave-one-out cross validation was used to determine the optimal combination of parameters for a regression model and multivariable logistic regression was used to identify independent predictors for pCR. Chi-squared test, two-sided t-test and Kruskal-Wallis test were used. RESULTS: The Ktrans I/C ratio cutoff value of 1.11 had a sensitivity of 83.3% and specificity of 75% for pCR. The ve I/C ratio cutoff value of 1.1 had a sensitivity of 75% and specificity of 81.3% for pCR. The area under the receiver operating characteristic curve of the three-kBPE parameter model, including initial area under the enhancement curve (AUC30) I/C ratio, KtransI/C ratio and ve I/C ratio, was 0.89 with sensitivity of 91.7% at specificity of 81.3%. CONCLUSION: Quantitative assessment of bilateral asymmetry kBPE from pre-NAC ultrafast DCE-MRI can predict pCR in patients with HER2+ breast cancer.

Clinicopathologic Characteristics and Prognosis of ERBB2-Low Breast Cancer Among Patients in the National Cancer Database.

Importance: Given conflicting results regarding the prognosis of erb-b2 receptor tyrosine kinase 2 (ERBB2; formerly HER2 or HER2/neu)-low breast cancer, a large-scale, nationally applicable comparison of ERBB2-low vs ERBB2-negative breast cancer is needed. Objective: To investigate whether ERBB2-low breast cancer is a clinically distinct subtype in terms of epidemiological characteristics, prognosis, and response to neoadjuvant chemotherapy. Design/Participants/Setting: This retrospective cohort study was conducted using the National Cancer Database, including 1 136 016 patients in the US diagnosed with invasive breast cancer from January 1, 2010, to December 31, 2019, who had ERBB2-negative disease and had immunohistochemistry results available. ERBB2-low tumors were classified as having an immunohistochemistry score of 1+, or 2+ with a negative in situ hybridization test. Data were analyzed from November 1, 2021, through November 30, 2022. Exposures: Standard therapy according to routine clinical practice. Main Outcomes and Measures: The primary outcomes were overall survival (OS), reported as adjusted hazard ratios (aHRs), and pathologic complete response, reported as adjusted odds ratios (aORs), for ERBB2-negative vs ERBB2-low breast cancer, controlling for age, sex, race and ethnicity, Charlson-Deyo Comorbidity Index score, treatment facility type, tumor grade, tumor histology, hormone receptor status, and cancer stage. Results: The study identified 1 136 016 patients (mean [SD] age, 62.4 [13.1] years; 99.1% female; 78.6% non-Hispanic White), of whom 392 246 (34.5%) were diagnosed with ERBB2-negative and 743 770 (65.5%) with ERBB2-low breast cancer. The mean (SD) age of the ERBB2-negative group was 62.1 (13.2) years and 62.5 (13.0) years for the ERBB2-low group. Higher estrogen receptor expression was associated with increased rates of ERBB2-low disease (aOR, 1.15 per 10% increase). Compared with non-Hispanic White patients, of whom 66.1% were diagnosed with ERBB2-low breast cancer, fewer non-Hispanic Black (62.8%) and Hispanic (61.0%) patients had ERBB2-low disease, although in non-Hispanic Black patients this was mediated by differences in rates of triple-negative disease and other confounders. A slightly lower rate of pathologic complete response was seen in patients with ERBB2-low disease vs patients with ERBB2-negative disease on multivariable analysis (aOR, 0.89; 95% CI, 0.86-0.92; P < .001). ERBB2-low status was also associated with small improvements in OS for stage III (aHR, 0.92; 95% CI, 0.89-0.96; P < .001) and stage IV (aHR, 0.91; 95% CI, 0.87-0.96; P < .001) triple-negative breast cancer, although this amounted to only a 2.0% (stage III) and 0.4% (stage IV) increase in 5-year OS. Conclusions and Relevance: This large-scale retrospective cohort analysis found minimal prognostic differences between ERBB2-low and ERBB2-negative breast cancer. These findings suggest that, moving forward, outcomes in ERBB2-low breast cancer will be driven by ERBB2-directed antibody-drug conjugates, rather than intrinsic differences in biological characteristics associated with low-level ERBB2 expression. These findings do not support the classification of ERBB2-low breast cancer as a unique disease entity.

Comparing scientific abstracts generated by ChatGPT to real abstracts with detectors and blinded human reviewers.

Large language models such as ChatGPT can produce increasingly realistic text, with unknown information on the accuracy and integrity of using these models in scientific writing. We gathered fifth research abstracts from five high-impact factor medical journals and asked ChatGPT to generate research abstracts based on their titles and journals. Most generated abstracts were detected using an AI output detector, 'GPT-2 Output Detector', with % 'fake' scores (higher meaning more likely to be generated) of median [interquartile range] of 99.98% 'fake' [12.73%, 99.98%] compared with median 0.02% [IQR 0.02%, 0.09%] for the original abstracts. The AUROC of the AI output detector was 0.94. Generated abstracts scored lower than original abstracts when run through a plagiarism detector website and iThenticate (higher scores meaning more matching text found). When given a mixture of original and general abstracts, blinded human reviewers correctly identified 68% of generated abstracts as being generated by ChatGPT, but incorrectly identified 14% of original abstracts as being generated. Reviewers indicated that it was surprisingly difficult to differentiate between the two, though abstracts they suspected were generated were vaguer and more formulaic. ChatGPT writes believable scientific abstracts, though with completely generated data. Depending on publisher-specific guidelines, AI output detectors may serve as an editorial tool to help maintain scientific standards. The boundaries of ethical and acceptable use of large language models to help scientific writing are still being discussed, and different journals and conferences are adopting varying policies.

The Impact of Race and Age on Response to Neoadjuvant Therapy and Long-Term Outcomes in Black and White Women with Early-Stage Breast Cancer.

PURPOSE: There are a paucity of data and a pressing need to evaluate response to neoadjuvant chemotherapy (NACT) and determine long-term outcomes in young Black women with early-stage breast cancer (EBC). METHODS: We analyzed data from 2,196 Black and White women with EBC treated at the University of Chicago over the last 2 decades. Patients were divided into groups based on race and age at diagnosis: Black women 40 years, White women 40 years, Black women 55 years, and White women 55 years. Pathological complete response rate (pCR) was analyzed using logistic regression. Overall survival (OS) and disease-free survival (DFS) were analyzed using Cox proportional hazard and piecewise Cox models. RESULTS: Young Black women had the highest risk of recurrence, which was 22% higher than young White women (p=0.434) and 76% higher than older Black women (p=0.008). These age/racial differences in recurrence rates were not statistically significant after adjusting for subtype, stage, and grade. In terms of OS, older Black women had the worst outcome. In the 397 women receiving NACT, 47.5% of young White women achieved pCR, compared to 26.8% of young Black women (p=0.012). CONCLUSIONS: Black women with EBC had significantly worse outcomes compared to White women in our cohort study. There is an urgent need to understand the disparities in outcomes between Black and White breast cancer patients, particularly in young women where the disparity in outcome is the greatest.