Ethical and Regulatory Perspectives on Generative Artificial Intelligence in Pathology.

CONTEXT.­: Technology companies and research groups are increasingly exploring applications of generative artificial intelligence (GenAI) in pathology and laboratory medicine. Although GenAI holds considerable promise, it also introduces novel risks for patients, communities, professionals, and the scientific process. OBJECTIVE.­: To summarize the current frameworks for the ethical development and management of GenAI within health care settings. DATA SOURCES.­: The analysis draws from scientific journals, organizational websites, and recent guidelines on artificial intelligence ethics and regulation. CONCLUSIONS.­: The literature on the ethical management of artificial intelligence in medicine is extensive but is still in its nascent stages because of the evolving nature of the technology. Effective and ethical integration of GenAI requires robust processes and shared accountability among technology vendors, health care organizations, regulatory bodies, medical professionals, and professional societies. As the technology continues to develop, a multifaceted ecosystem of safety mechanisms and ethical oversight is crucial to maximize benefits and mitigate risks.

A vision to the future: value-based laboratory medicine.

The ultimate goal of value-based laboratory medicine is maximizing the effectiveness of laboratory tests in improving patient outcomes, optimizing resources and minimizing unnecessary costs. This approach abandons the oversimplified notion of test volume and cost, in favor of emphasizing the clinical utility and quality of diagnostic tests in the clinical decision-making. Several key elements characterize value-based laboratory medicine, which can be summarized in some basic concepts, such as organization of in vitro diagnostics (including appropriateness, integrated diagnostics, networking, remote patient monitoring, disruptive innovations), translation of laboratory data into clinical information and measurable outcomes, sustainability, reimbursement, ethics (e.g., patient empowerment and safety, data protection, analysis of big data, scientific publishing). Education and training are also crucial, along with considerations for the future of the profession, which will be largely influenced by advances in automation, information technology, artificial intelligence, and regulations concerning in vitro diagnostics. This collective opinion paper, composed of summaries from presentations given at the two-day European Federation of Laboratory Medicine (EFLM) Strategic Conference "A vision to the future: value-based laboratory medicine" (Padova, Italy; September 23-24, 2024), aims to provide a comprehensive overview of value-based laboratory medicine, projecting the profession into a more clinically effective and sustainable future.

Clinical outcomes and ctDNA correlates for CAPOX BETR: a phase II trial of capecitabine, oxaliplatin, bevacizumab, trastuzumab in previously untreated advanced HER2+ gastroesophageal adenocarcinoma.

Preclinical studies suggest that simultaneous HER2/VEGF blockade may have cooperative effects in gastroesophageal adenocarcinomas. In a single-arm investigator initiated clinical trial for patients with untreated advanced HER2+ gastroesophageal adenocarcinoma, bevacizumab was added to standard of care capecitabine, oxaliplatin, and trastuzumab in 36 patients (NCT01191697). Primary endpoint was objective response rate and secondary endpoints included safety, duration of response, progression free survival, and overall survival. The study met its primary endpoint with an objective response rate of 81% (95% CI 65-92%). Median progression free and overall survival were 14.0 (95% CI, 11.3-36.4) and 23.2 months (95% CI, 16.6-36.4), respectively. The median duration of response was 14.9 months. The regimen was well tolerated without unexpected or severe toxicities. In post-hoc ctDNA analysis, baseline ctDNA features were prognostic: Higher tumor fraction and alternative MAPK drivers portended worse outcomes. ctDNA at resistance identified oncogenic mutations and these were detectable 2-8 cycles prior to radiographic progression. Capecitabine, oxaliplatin, trastuzumab and bevacizumab shows robust clinical activity in HER2+ gastroesophageal adenocarcinoma. Combination of VEGF inhibitors with chemoimmunotherapy and anti-PD1 regimens is warranted.

FGFR Alterations in Thyroid Carcinoma: A Novel Class of Primary Drivers with Significant Therapeutic Implications and Secondary Molecular Events Potentially Mediating Resistance in Thyroid Malignancy.

Background: Diagnostic classification of thyroid malignancy is primarily accomplished through examination of histomorphological features and may be substantiated and clarified by molecular data. Individual molecular drivers show relatively robust and specific associations with histological subtypes of thyroid malignancy, including BRAF sequence variants and kinase gene fusions in papillary thyroid carcinoma, predominantly RAS variants in follicular-patterned neoplasia, and additional "late" mutations affecting TERT promoter, TP53, and the PI3K/AKT/PTEN pathway in high-grade malignancies. Given the oncogenic role of FGFR, particularly FGFR1-3, the goal of this study was to explore the role of FGFR in thyroid carcinoma biology. Methods: We completed a multicenter retrospective observational study for thyroid carcinomas with pathogenic alterations in the FGFR gene family. We performed this study by querying the molecular data accumulated for thyroid carcinomas from each center. Results: Overall, 5030 sequenced thyroid malignancies were reviewed, yielding 17 tumors with FGFR alterations, including 11 where FGFR was the primary molecular driver and 6 where FGFR was a secondary pathogenic alteration, with a subset for which there was available clinical follow-up data. Of the 11 carcinomas with an FGFR driver, 9 were gene fusions involving FGFR2:VCL (4 tumors), TG::FGFR1 (3 tumors), FGFR2::CIT, and FGFR2::SHTN1, and the remaining 2 were driven by FGFR1 amplification. In the 6 tumors where a canonical driver of thyroid neoplasia was present (5 cases) or no clear primary driver was detected (1 case), sequencing detected secondary FGFR2 p.W290C, p.Y375C, and p.N549K, as well as FGFR1 p.N546K in the respective tyrosine kinase domains, some at subclonal variant allele frequencies. Conclusions: This study presents the first description of a collection of thyroid carcinomas grouped by primary driver alterations in FGFR, as well as a cohort of thyroid tumors with secondary alterations that potentially lead to tumor progression or resistance to targeted therapy. Given the availability of small molecular inhibitors targeting oncogenic FGFR, this study emphasizes the significant implications for patients from identification of FGFR alterations as they are currently under-recognized in the literature and, most importantly, have potential novel treatment options.

Technology and Future of Multi-Cancer Early Detection.

Cancer remains a significant global health challenge due to its high morbidity and mortality rates. Early detection is essential for improving patient outcomes, yet current diagnostic methods lack the sensitivity and specificity needed for identifying early-stage cancers. Here, we explore the potential of multi-omics approaches, which integrate genomic, transcriptomic, proteomic, and metabolomic data, to enhance early cancer detection. We highlight the challenges and benefits of data integration from these diverse sources and discuss successful examples of multi-omics applications in other fields. By leveraging these advanced technologies, multi-omics can significantly improve the sensitivity and specificity of early cancer diagnostics, leading to better patient outcomes and more personalized cancer care. We underscore the transformative potential of multi-omics approaches in revolutionizing early cancer detection and the need for continued research and clinical integration.

Harmonized cross-species cell atlases of trigeminal and dorsal root ganglia.

Sensory neurons in the dorsal root ganglion (DRG) and trigeminal ganglion (TG) are specialized to detect and transduce diverse environmental stimuli to the central nervous system. Single-cell RNA sequencing has provided insights into the diversity of sensory ganglia cell types in rodents, nonhuman primates, and humans, but it remains difficult to compare cell types across studies and species. We thus constructed harmonized atlases of the DRG and TG that describe and facilitate comparison of 18 neuronal and 11 non-neuronal cell types across six species and 31 datasets. We then performed single-cell/nucleus RNA sequencing of DRG from both human and the highly regenerative axolotl and found that the harmonized atlas also improves cell type annotation, particularly of sparse neuronal subtypes. We observed that the transcriptomes of sensory neuron subtypes are broadly similar across vertebrates, but the expression of functionally important neuropeptides and channels can vary notably. The resources presented here can guide future studies in comparative transcriptomics, simplify cell-type nomenclature differences across studies, and help prioritize targets for future analgesic development.

Bridging the Clinical-Computational Transparency Gap in Digital Pathology.

CONTEXT.­: Computational pathology combines clinical pathology with computational analysis, aiming to enhance diagnostic capabilities and improve clinical productivity. However, communication barriers between pathologists and developers often hinder the full realization of this potential. OBJECTIVE.­: To propose a standardized framework that improves mutual understanding of clinical objectives and computational methodologies. The goal is to enhance the development and application of computer-aided diagnostic (CAD) tools. DESIGN.­: The article suggests pivotal roles for pathologists and computer scientists in the CAD development process. It calls for increased understanding of computational terminologies, processes, and limitations among pathologists. Similarly, it argues that computer scientists should better comprehend the true use cases of the developed algorithms to avoid clinically meaningless metrics. RESULTS.­: CAD tools improve pathology practice significantly. Some tools have even received US Food and Drug Administration approval. However, improved understanding of machine learning models among pathologists is essential to prevent misuse and misinterpretation. There is also a need for a more accurate representation of the algorithms' performance compared to that of pathologists. CONCLUSIONS.­: A comprehensive understanding of computational and clinical paradigms is crucial for overcoming the translational gap in computational pathology. This mutual comprehension will improve patient care through more accurate and efficient disease diagnosis.

Rapid tumor DNA analysis of cerebrospinal fluid accelerates treatment of central nervous system lymphoma.

ABSTRACT: Delays and risks associated with neurosurgical biopsies preclude timely diagnosis and treatment of central nervous system (CNS) lymphoma and other CNS neoplasms. We prospectively integrated targeted rapid genotyping of cerebrospinal fluid (CSF) into the evaluation of 70 patients with CNS lesions of unknown cause. Participants underwent genotyping of CSF-derived DNA using a quantitative polymerase chain reaction-based approach for parallel detection of single-nucleotide variants in the MYD88, TERT promoter, IDH1, IDH2, BRAF, and H3F3A genes within 80 minutes of sample acquisition. Canonical mutations were detected in 42% of patients with neoplasms, including cases of primary and secondary CNS lymphoma, glioblastoma, IDH-mutant brainstem glioma, and H3K27M-mutant diffuse midline glioma. Genotyping results eliminated the need for surgical biopsies in 7 of 33 cases (21.2%) of newly diagnosed neoplasms, resulting in significantly accelerated initiation of disease-directed treatment (median, 3 vs 12 days; P = .027). This assay was then implemented in a Clinical Laboratory Improvement Amendments environment, with 2-day median turnaround for diagnosis of CNS lymphoma from 66 patients across 4 clinical sites. Our study prospectively demonstrates that targeted rapid CSF genotyping influences oncologic management for suspected CNS tumors.

Radical surgical resection with molecular margins is associated with improved survival in IDH wild-type glioblastoma.

BACKGROUND: Survival is variable in patients with glioblastoma IDH wild-type (GBM), even after comparable surgical resection of radiographically detectable disease, highlighting the limitations of radiographic assessment of infiltrative tumor anatomy. The majority of postsurgical progressive events are failures within 2 cm of the resection margin, motivating supramaximal resection strategies to improve local control. However, which patients benefit from such radical resections remains unknown. METHODS: We developed a predictive model to identify which IDH wild-type GBMs are amenable to radiographic gross-total resection (GTR). We then investigated whether GBM survival heterogeneity following GTR is correlated with microscopic tumor burden by analyzing tumor cell content at the surgical margin with a rapid qPCR-based method for detection of TERT promoter mutation. RESULTS: Our predictive model for achievable GTR, developed on retrospective radiographic and molecular data of GBM patients undergoing resection, had an area under the curve of 0.83, sensitivity of 62%, and specificity of 90%. Prospective analysis of this model in 44 patients found that 89% of patients were correctly predicted to achieve a residual volume (RV) < 4.9cc. Of the 44 prospective patients undergoing rapid qPCR TERT promoter mutation analysis at the surgical margin, 7 had undetectable TERT mutation, of which 5 also had a GTR (RV < 1cc). In these 5 patients at 30 months follow-up, 75% showed no progression, compared to 0% in the group with TERT mutations detected at the surgical margin (P = .02). CONCLUSIONS: These findings identify a subset of patients with GBM that may derive local control benefits from radical resection to undetectable molecular margins.

Training pathologists to assess stromal tumour-infiltrating lymphocytes in breast cancer synergises efforts in clinical care and scientific research.

A growing body of research supports stromal tumour-infiltrating lymphocyte (TIL) density in breast cancer to be a robust prognostic and predicive biomarker. The gold standard for stromal TIL density quantitation in breast cancer is pathologist visual assessment using haematoxylin and eosin-stained slides. Artificial intelligence/machine-learning algorithms are in development to automate the stromal TIL scoring process, and must be validated against a reference standard such as pathologist visual assessment. Visual TIL assessment may suffer from significant interobserver variability. To improve interobserver agreement, regulatory science experts at the US Food and Drug Administration partnered with academic pathologists internationally to create a freely available online continuing medical education (CME) course to train pathologists in assessing breast cancer stromal TILs using an interactive format with expert commentary. Here we describe and provide a user guide to this CME course, whose content was designed to improve pathologist accuracy in scoring breast cancer TILs. We also suggest subsequent steps to translate knowledge into clinical practice with proficiency testing.

Virus-Induced Acute Respiratory Distress Syndrome Causes Cardiomyopathy Through Eliciting Inflammatory Responses in the Heart.

BACKGROUND: Viral infections can cause acute respiratory distress syndrome (ARDS), systemic inflammation, and secondary cardiovascular complications. Lung macrophage subsets change during ARDS, but the role of heart macrophages in cardiac injury during viral ARDS remains unknown. Here we investigate how immune signals typical for viral ARDS affect cardiac macrophage subsets, cardiovascular health, and systemic inflammation. METHODS: We assessed cardiac macrophage subsets using immunofluorescence histology of autopsy specimens from 21 patients with COVID-19 with SARS-CoV-2-associated ARDS and 33 patients who died from other causes. In mice, we compared cardiac immune cell dynamics after SARS-CoV-2 infection with ARDS induced by intratracheal instillation of Toll-like receptor ligands and an ACE2 (angiotensin-converting enzyme 2) inhibitor. RESULTS: In humans, SARS-CoV-2 increased total cardiac macrophage counts and led to a higher proportion of CCR2+ (C-C chemokine receptor type 2 positive) macrophages. In mice, SARS-CoV-2 and virus-free lung injury triggered profound remodeling of cardiac resident macrophages, recapitulating the clinical expansion of CCR2+ macrophages. Treating mice exposed to virus-like ARDS with a tumor necrosis factor α-neutralizing antibody reduced cardiac monocytes and inflammatory MHCIIlo CCR2+ macrophages while also preserving cardiac function. Virus-like ARDS elevated mortality in mice with pre-existing heart failure. CONCLUSIONS: Our data suggest that viral ARDS promotes cardiac inflammation by expanding the CCR2+ macrophage subset, and the associated cardiac phenotypes in mice can be elicited by activating the host immune system even without viral presence in the heart.

Dimensions of health data Integrity.

Health data integrity, as an emergent concept, stands to reshape the lifecycle of data-driven healthcare and research, ensuring a shared commitment to ethical practices and improved patient care.

Reproducible Reporting of the Collection and Evaluation of Annotations for Artificial Intelligence Models.

This work puts forth and demonstrates the utility of a reporting framework for collecting and evaluating annotations of medical images used for training and testing artificial intelligence (AI) models in assisting detection and diagnosis. AI has unique reporting requirements, as shown by the AI extensions to the Consolidated Standards of Reporting Trials (CONSORT) and Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) checklists and the proposed AI extensions to the Standards for Reporting Diagnostic Accuracy (STARD) and Transparent Reporting of a Multivariable Prediction model for Individual Prognosis or Diagnosis (TRIPOD) checklists. AI for detection and/or diagnostic image analysis requires complete, reproducible, and transparent reporting of the annotations and metadata used in training and testing data sets. In an earlier work by other researchers, an annotation workflow and quality checklist for computational pathology annotations were proposed. In this manuscript, we operationalize this workflow into an evaluable quality checklist that applies to any reader-interpreted medical images, and we demonstrate its use for an annotation effort in digital pathology. We refer to this quality framework as the Collection and Evaluation of Annotations for Reproducible Reporting of Artificial Intelligence (CLEARR-AI).

Diagnostic Test Utilization Management Strategies as an Opportunity for Equitable Access to Molecularly Informed Clinical Care.

BACKGROUND: Companion diagnostics are an essential component of oncology. Timing, cost, and adaptability to new drug/biomarker approvals represent challenges in assuring value-based care. Overcoming these challenges requires strategies for equitable access and efficient integration. METHODS: Based on prior laboratory improvements and payor policy implementations, we define equitable access in laboratory testing and conceptualized a framework for initiatives that optimize diagnostic performance. RESULTS: We define equitable access as an imperative goal seeking to remove disparities that may arise due to financial hardships, geographical isolation, cultural differences, or other social determinants of health. We distinguish (a) utilization, as the practice pattern of ordered tests, (b) utilization management, as the evidence-based guidance of the utilization decisions, and (c) utilization management strategies, defined as the tools and techniques used to influence decision-making. These 3 dimensions establish a standardized vocabulary to clarify equitable alignment of strategies in specific care pathways. Alignment of logistic, administrative, and financial incentive structures is paramount when creating sustainable personalized care pathway programs. CONCLUSIONS: Strategies to accomplish equitable and meaningful use of diagnostic tests can help enhance access to timely and accurate diagnoses, ultimately leading to improved patient outcomes.

Compromised Outcomes in Stage IV Non-Small-Cell Lung Cancer With Actionable Mutations Initially Treated Without Tyrosine Kinase Inhibitors: A Retrospective Analysis of Real-World Data.

PURPOSE: Identification and targeting of actionable oncogenic drivers (AODs) in advanced non-small-cell lung cancer (NSCLC) has dramatically improved outcomes. However, genomic testing uptake is variable and hampered by factors including slow turnaround time, frequently resulting in initial non-tyrosine kinase inhibitor (TKI) treatment. We investigate how this behavior affects outcomes. METHODS: This retrospective analysis of real-world, deidentified data from the Integra Connect Database included adults with stage IV NSCLC newly diagnosed from January 1, 2018, to December 31, 2020, with mutations of EGFR, ALK, ROS1, BRAF, MET, RET, ERBB2, or NTRK. Outcomes were reported as time to next treatment or death (TTNT) and overall survival (OS). RESULTS: Five hundred ten patients harboring AODs were identified and grouped as follows: group A (n = 379) were treated after the AOD was reported and served as the comparator. One hundred thirty-one patients treated before their AOD report were divided into group B (n = 47) who were initially started on chemotherapy and/or checkpoint inhibitor but switched to appropriate TKI within 35 days and group C (n = 84) who were also started empirically on non-TKI and did not switch within 35 days. Survival (OS) was significantly superior in group A compared with group C; TTNT was significantly superior in group A compared with groups B and C. CONCLUSION: For patients harboring AODs in advanced NSCLC, initial treatment before receipt of genomic test results yields significantly inferior outcomes and should be avoided. Molecular profiling panels with rapid turnaround times are essential to optimize patient outcomes and should be standard of care.

Allosteric PI3Kα Inhibition Overcomes On-target Resistance to Orthosteric Inhibitors Mediated by Secondary PIK3CA Mutations.

PIK3CA mutations occur in ∼8% of cancers, including ∼40% of HR-positive breast cancers, where the PI3K-alpha (PI3Kα)-selective inhibitor alpelisib is FDA approved in combination with fulvestrant. Although prior studies have identified resistance mechanisms, such as PTEN loss, clinically acquired resistance to PI3Kα inhibitors remains poorly understood. Through serial liquid biopsies and rapid autopsies in 39 patients with advanced breast cancer developing acquired resistance to PI3Kα inhibitors, we observe that 50% of patients acquire genomic alterations within the PI3K pathway, including PTEN loss and activating AKT1 mutations. Notably, although secondary PIK3CA mutations were previously reported to increase sensitivity to PI3Kα inhibitors, we identified emergent secondary resistance mutations in PIK3CA that alter the inhibitor binding pocket. Some mutations had differential effects on PI3Kα-selective versus pan-PI3K inhibitors, but resistance induced by all mutations could be overcome by the novel allosteric pan-mutant-selective PI3Kα-inhibitor RLY-2608. Together, these findings provide insights to guide strategies to overcome resistance in PIK3CA-mutated cancers. SIGNIFICANCE: In one of the largest patient cohorts analyzed to date, this study defines the clinical landscape of acquired resistance to PI3Kα inhibitors. Genomic alterations within the PI3K pathway represent a major mode of resistance and identify a novel class of secondary PIK3CA resistance mutations that can be overcome by an allosteric PI3Kα inhibitor. See related commentary by Gong and Vanhaesebroeck, p. 204 . See related article by Varkaris et al., p. 240 . This article is featured in Selected Articles from This Issue, p. 201.

Virtual Molecular and Precision Medicine Clinic to Improve Access to Clinical Trials for Patients With Metastatic Breast Cancer: An Academic/Community Collaboration.

PURPOSE: There is a demand for improved care delivery surrounding genomic testing and clinical trial enrollment among patients with metastatic breast cancer (MBC). We sought to improve the current process via real-time informal consultation and prescreening assessment for patients with MBC treated by community and academic medical oncologists by implementing a virtual molecular and precision medicine (vMAP) clinic. METHODS: The vMAP program used a virtual referral system directed to a multidisciplinary team with precision medicine expertise. Providers contacted vMAP regarding patients with MBC, and on receipt of referral, the vMAP team engaged in discussion to identify if further diagnostics were needed (including genomic testing) and to identify potential clinical trials or standard treatment options. Recommendations were then sent to the referring provider within 72 hours. Pre-/postsurveys were issued to network physicians to assess for barriers, clinical trial access, and vMAP referral experience. Program implementation was evaluated with the Squire 2.0 reporting guidelines for quality improvement in health care as a framework. RESULTS: Eighty-one cases from 22 providers were referred to vMAP over a 26-month period. The average response time to the referring provider with a finalized recommendation was 1.90 ± 1.82 days. A total of 86.4% of cases had clinical trial options on vMAP prescreen, with 40.7% initiating formal screening assessments and 27 patients (33.3%) ultimately enrolling on trials. On resurvey, 92% of survey responses across community oncology referring providers said that they were very likely to use vMAP again. CONCLUSION: In the initial 2-year period, vMAP demonstrated an efficient means to offer real-time interpretation of genomic testing and identification of clinical trials for patients with MBC, with effective clinical trial enrollment and high rates of referring provider satisfaction.