Comparing deep learning and pathologist quantification of cell-level PD-L1 expression in non-small cell lung cancer whole-slide images.

Programmed death-ligand 1 (PD-L1) expression is currently used in the clinic to assess eligibility for immune-checkpoint inhibitors via the tumor proportion score (TPS), but its efficacy is limited by high interobserver variability. Multiple papers have presented systems for the automatic quantification of TPS, but none report on the task of determining cell-level PD-L1 expression and often reserve their evaluation to a single PD-L1 monoclonal antibody or clinical center. In this paper, we report on a deep learning algorithm for detecting PD-L1 negative and positive tumor cells at a cellular level and evaluate it on a cell-level reference standard established by six readers on a multi-centric, multi PD-L1 assay dataset. This reference standard also provides for the first time a benchmark for computer vision algorithms. In addition, in line with other papers, we also evaluate our algorithm at slide-level by measuring the agreement between the algorithm and six pathologists on TPS quantification. We find a moderately low interobserver agreement at cell-level level (mean reader-reader F1 score = 0.68) which our algorithm sits slightly under (mean reader-AI F1 score = 0.55), especially for cases from the clinical center not included in the training set. Despite this, we find good AI-pathologist agreement on quantifying TPS compared to the interobserver agreement (mean reader-reader Cohen's kappa = 0.54, 95% CI 0.26-0.81, mean reader-AI kappa = 0.49, 95% CI 0.27-0.72). In conclusion, our deep learning algorithm demonstrates promise in detecting PD-L1 expression at a cellular level and exhibits favorable agreement with pathologists in quantifying the tumor proportion score (TPS). We publicly release our models for use via the Grand-Challenge platform.

Rapid on-site evaluation of touch imprint cytology in navigation bronchoscopy for small peripheral pulmonary nodules.

BACKGROUND: Rapid on-site evaluation (ROSE) of cytopathology plays an important role in determining whether representative samples have been taken during navigation bronchoscopy. With touch imprint cytology (TIC), histologic samples can be assessed using ROSE. Although advised by guidelines, there have been almost no studies on the performance of TIC during navigation bronchoscopy. The objective of this study was to evaluate the value of TIC-ROSE (forceps/cryobiopsy) in combination with conventional ROSE (cytology needle/brush). METHODS: In this single-center, prospective cohort study, patients who had pulmonary nodules with an indication for navigation bronchoscopy were consecutively included. The primary outcome of the study was the concordance of ROSE and the procedural outcome. The concordance rates of TIC-ROSE and the combination of TIC-ROSE plus conventional ROSE were compared. RESULTS: Fifty-eight patients with 66 nodules were included. Conventional ROSE and TIC-ROSE were assessable in 61 nodules (90.9%) each. By combining both ROSE techniques, all sampled lesions were assessable. Combining conventional ROSE with TIC-ROSE showed concordant results in 51 of 66 cases (77.3%) versus 44 of 66 (66.7%) and 48 of 66 (72.8%) concordant results for conventional ROSE and TIC-ROSE alone, respectively, compared with the procedural outcome. There was no indication of tissue depletion as a result of TIC. The combined ROSE approach had a statistically significant higher concordance rate compared with conventional ROSE alone. CONCLUSIONS: TIC-ROSE is a cheap, easily implementable technique that can result in higher concordant ROSE outcomes. This could lead to more efficient procedures and possibly higher diagnostic results. In a monomodality sampling setting with only histologic samples, TIC can provide ROSE.

Severe Immune-Related Enteritis after In Utero Exposure to Pembrolizumab.

Immune checkpoint blockade has become standard treatment for many types of cancer. Such therapy is indicated most often in patients with advanced or metastatic disease but has been increasingly used as adjuvant therapy in those with early-stage disease. Adverse events include immune-related organ inflammation resembling autoimmune diseases. We describe a case of severe immune-related gastroenterocolitis in a 4-month-old infant who presented with intractable diarrhea and failure to thrive after in utero exposure to pembrolizumab. Known causes of the symptoms were ruled out, and the diagnosis of pembrolizumab-induced immune-related gastroenterocolitis was supported by the results of histopathological assays, immunophenotyping, and analysis of the level of antibodies against programmed cell death protein 1 (PD-1). The infant's condition was successfully treated with prednisolone and infliximab.

Intrinsic pulmonary sealing, its mechanisms and impact on validity and translational value of lung sealant studies: a pooled analysis of animal studies.

Background: No validated and standardized animal models of pulmonary air leakage (PAL) exist for testing aerostatic efficacy of lung sealants. Lack of negative control groups in published studies and intrinsic sealing mechanisms of healthy animal lungs might contribute to a translational gap, leading to poor clinical results. This study aims to address the impact of intrinsic sealing mechanisms on the validity of PAL models, and investigate the conditions required for an ovine model of PAL for lung sealant testing. Methods: An ovine acute aerostasis model was developed, consisting of a bilateral thoracotomy with lesion creation, chest tube insertion and monitoring of air leaks using digital drains (≥80 minutes), under spontaneous respiration. Healthy mixed-breed adult female sheep were used and all in vivo procedures were performed under terminal anesthesia. Superficial parenchymal lesions were tested post-mortem and in vivo, extended lesions including bronchioles (deep bowl-shaped and sequential lung amputation lesions) were tested in vivo. Experiment outcomes include air leakage (AL), minimal leaking pressure (MLP) and histology. Results: Two post-mortem (N=4 superficial parenchymal lesions) and 10 in vivo experiments (N=5 superficial parenchymal and N=16 lesions involving bronchioles) were performed. In contrast to the post-mortem model, superficial parenchymal lesions in vivo showed less air leak [mean flow ± standard deviation (SD): 760±693 vs. 42±33 mL/min, P=0.055]. All superficial parenchymal lesions in vivo sealed intrinsically within a median time of 20 minutes [interquartile range (IQR), 10-75 minutes]. Histology of the intrinsic sealing layer revealed an extended area of alveolar collapse below the incision with intra-alveolar hemorrhage. Compared to superficial parenchymal lesions in vivo, lesions involving bronchioles induced significantly higher air leak post-operatively (normalized mean flow ± SD: 459±221 mL/min, P=0.003). At termination, 5/9 (55.6%) were still leaking (median drain time: 273 minutes, IQR, 207-435 minutes), and intrinsic sealing for the remaining lungs occurred within a median of 115 minutes (IQR, 52-245 minutes). Conclusions: Lung parenchyma of healthy sheep shows a strong intrinsic sealing mechanism, explained pathologically by an extended area of alveolar collapse, which may contribute to a translational gap in lung sealant research. A meaningful ovine model has to consist of deep lesions involving bronchioles of >⌀1.5 mm. Further research is needed to develop a standardized PAL model, to improve clinical effectiveness of lung sealants.

Integrating artificial intelligence in pathology: a qualitative interview study of users' experiences and expectations.

Recent progress in the development of artificial intelligence (AI) has sparked enthusiasm for its potential use in pathology. As pathology labs are currently starting to shift their focus towards AI implementation, a better understanding how AI tools can be optimally aligned with the medical and social context of pathology daily practice is urgently needed. Strikingly, studies often fail to mention the ways in which AI tools should be integrated in the decision-making processes of pathologists, nor do they address how this can be achieved in an ethically sound way. Moreover, the perspectives of pathologists and other professionals within pathology concerning the integration of AI within pathology remains an underreported topic. This article aims to fill this gap in the literature and presents the first in-depth interview study in which professionals' perspectives on the possibilities, conditions and prerequisites of AI integration in pathology are explicated. The results of this study have led to the formulation of three concrete recommendations to support AI integration, namely: (1) foster a pragmatic attitude toward AI development, (2) provide task-sensitive information and training to health care professionals working in pathology departments and (3) take time to reflect upon users' changing roles and responsibilities.

Increased Colorectal Neoplasia Risk in Patients with Inflammatory Bowel Disease and Serrated Polyps with Dysplasia.

BACKGROUND: The impact of serrated polyps on the advanced colorectal neoplasia (CRN) risk in inflammatory bowel disease (IBD) patients is unknown. Serrated polyps are histologically categorized as hyperplastic polyps (HPs), sessile serrated lesions (SSLs), and traditional serrated adenomas (TSAs). AIMS: We aimed (1) to characterize the serrated polyps in IBD patients, (2) to identify factors associated with the presence of serrated polyps in IBD, and (3) to assess the CRN risk in IBD patients with serrated polyps. METHODS: We established a retrospective cohort of IBD patients with and without colonic serrated polyps. Cox-regression analysis with time-dependent variables was used to compare advanced CRN risk in IBD patients with and without serrated polyps. RESULTS: Of the 621 enrolled IBD patients, 198 had a serrated polyp (92 HPs, 88 SSLs without dysplasia, 13 SSLs with dysplasia, and 5 TSAs). Independent factors associated with serrated polyps were ulcerative colitis (UC) (odds ratio (OR) 1.77, 95% confidence interval (CI) 1.19-2.62, p = 0.005), male gender (OR 1.63, 95% CI 1.11-2.40, p = 0.013), and older age (per year increase, OR 1.06, 95%CI 1.05-1.08, p < 0.001). TSAs and SSLs with dysplasia were risk factors for subsequent advanced CRN (HR 13.51, 95% CI 3.11-58.68, p < 0.001), while HPs (HR 1.98, 95% CI 0.46-8.60, p = 0.36) and SSLs without dysplasia (HR 0.87, 95% CI 0.11-6.88, p-0.89) did not impact the subsequent advanced CRN risk. CONCLUSIONS: UC, male gender and older age were associated with the presence of serrated polyps. The majority of serrated polyps (91%) were HPs and SSL without dysplasia and did not affect the CRC risk. However TSAs and SSLs with dysplasia, representing a small subgroup of serrated polyps (9%), were associated with subsequent advanced CRN.

Multi-modal tissue sampling in cone beam CT guided navigation bronchoscopy: comparative accuracy of different sampling tools and rapid on-site evaluation of cytopathology.

BACKGROUND: Advanced technological aids are frequently used to improve outcome of transbronchial diagnostics for peripheral pulmonary lesions. Even when lesion access has been confirmed by 3D imaging, obtaining an accurate tissue sample however remains difficult. In this single institution study, we evaluate the comparative accuracy of different sampling methodologies and the accuracy of rapid on-site evaluation of cytopathology (ROSE) in navigation bronchoscopy cases where imaging has confirmed the catheter to have accurately accessed the lesion. METHODS: All consecutive navigation bronchoscopies in between December 2017- June 2020 performed in a room with a cone beam CT (CBCT) system where catheter position was intra-procedurally confirmed to be within or adjacent to the lesion by cone beam CT and augmented fluoroscopy were included. Individual tool outcomes were compared against one another and follow-up outcome. RESULTS: A mean of 11.39 samples using 2.93 tools were obtained in 225 lesions (median diameter 15 mm, 195 patients). A correct diagnosis was most often obtained by forceps (accuracy 70.6%), followed by 1.1 mm cryoprobe (68.4%), needle aspiration (46.7%), 1.9 mm cryoprobe (41.2%), brush (30.3%) and lavage (23.7%). Procedural outcome corresponded to follow-up outcome in 75.1% of lesions (80.5% of patients). Accurately diagnosed lesions were sampled significantly more often (11.91 vs. 9.72 samples, P=0.014). In cases where procedural outcome proved malignant, ROSE had also detected this in 47.5%. CONCLUSIONS: Of all clinically available biopsy tools, the forceps showed most often accurate. However, extensive multi-modal sampling resulted in highest diagnostic accuracy. A hypothetical multi-modal approach of only using forceps and needle aspiration provided eventual diagnostic outcome in 91.7% of successfully diagnosed lesions. In the circumstances of our study, confirmation of malignancy on ROSE did not reduce number of biopsies taken nor biopsy time. Future research on how to improve the accuracy and effectivity of tissue sampling is needed.

COVID-19-associated Aspergillus tracheobronchitis: the interplay between viral tropism, host defence, and fungal invasion.

Invasive pulmonary aspergillosis is emerging as a secondary infection in patients with COVID-19, which can present as alveolar disease, airway disease (ie, invasive Aspergillus tracheobronchitis), or both. Histopathology of invasive Aspergillus tracheobronchitis in patients with severe COVID-19 confirms tracheal ulcers with tissue invasion of Aspergillus hyphae but without angioinvasion, which differs from patients with severe influenza, where early angioinvasion is observed. We argue that aggregation of predisposing factors (eg, factors that are defined by the European Organisation for Research and Treatment of Cancer and Mycoses Study Group Education and Research Consortium or genetic polymorphisms), viral factors (eg, tropism and lytic effects), immune defence factors, and effects of concomitant therapies will determine whether and when the angioinvasion threshold is reached. Management of invasive Aspergillus tracheobronchitis should include reducing viral lytic effects, rebalancing immune dysregulation, and systemic and local antifungal therapy. Future study designs should involve approaches that aim to develop improved diagnostics for tissue invasion and airways involvement and identify the immune status of the patient to guide personalised immunotherapy.

Gastric Epithelial Polyps: When to Ponder, When to Panic.

This review provides an overview of different types of gastric epithelial polyps. The polyps are classified based on their cell or epithelial compartment of origin. Some of these polyps can be considered reactive or nonneoplastic, whereas others are neoplastic in origin, are sometimes associated with a hereditary polyposis/cancer syndrome, and may have malignant potential. The aim of this review is to provide a pragmatic overview for the practicing pathologist about how to correctly diagnose and deal with gastric epithelial polyps and when (not) to ponder, and when (not) to panic.

Pathology Image Exchange: The Dutch Digital Pathology Platform for Exchange of Whole-Slide Images for Efficient Teleconsultation, Telerevision, and Virtual Expert Panels.

Among the many uses of digital pathology, remote consultation, remote revision, and virtual slide panels may be the most important ones. This requires basic slide scanner infrastructure in participating laboratories to produce whole-slide images. More importantly, a software platform is needed for exchange of these images and functionality to support the processes around discussing and reporting on these images without breaching patient privacy. This poses high demands on the setup of such a platform, given the inherent complexity of the handling of digital pathology images. In this article, we describe the setup and validation of the Pathology Image Exchange project, which aimed to create a vendor-independent platform for exchange of whole-slide images between Dutch pathology laboratories to facilitate efficient teleconsultation, telerevision, and virtual slide panels. Pathology Image Exchange was released in April 2018 after technical validation, and a first successful validation in real life has been performed for hematopathology cases.

Ethical considerations for modern molecular pathology.

Molecular pathology is becoming an increasingly important discipline in oncology as molecular tumor characteristics will increasingly determine targeted clinical cancer care. In recent years, many technological advances have taken place that contributed to the development of molecular pathology. However, attention to ethical aspects has been lagging behind as illustrated by the lack of publications or professional guidelines. Existing guidelines or publications on ethical aspects of DNA sequencing are mostly aimed at germline or tumor sequencing in clinical genetics or biomedical research settings. As a result, large differences have been demonstrated in the process of tumor sequencing analysis between laboratories. In this perspective we discuss the ethical issues to consider in molecular pathology by following the process of tumor DNA sequencing analysis from the preanalytical to postanalytical phase. For the successful and responsible use of DNA sequencing in clinical cancer care, several moral requirements must be met, for example, those related to the interpretation and returning of genetic results, informed consent, and the retrospective as well as future use of genetic data for biomedical research. Many ethical issues are new to pathology or more stringent than in current practice because DNA sequencing could yield sensitive and potentially relevant data, such as clinically significant unsolicited findings. The context of molecular pathology is unique and complex, but many issues are similar to those applicable to clinical genetics. As such, existing scholarship in this discipline may be translated to molecular pathology with some adaptations and could serve as a basis for guideline development. For responsible use and further development of clinical cancer care, we recommend that pathologists take responsibility for the adequate use of molecular analyses and be fully aware and capable of dealing with the diverse, complex, and challenging aspects of tumor DNA sequencing, including its ethical issues. Copyright © 2018 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Comprehensive Proteomic Profiling-derived Immunohistochemistry-based Prediction Models for BRCA1 and BRCA2 Germline Mutation-related Breast Carcinomas.

Heredity, mostly due to BRCA germline mutations, is involved in 5% to 10% of all breast cancer cases. Potential BRCA germline mutation carriers may be missed following the current eligibility criteria for BRCA genetic testing. The purpose of this study was to, therefore, develop an immunohistochemistry-based model to predict likelihood of underlying BRCA1 and BRCA2 germline mutations in unselected female breast cancer patients. The study group consisted of 100 BRCA1-related, 46 BRCA2-related, and 94 sporadic breast carcinomas. Tumor expression of 44 proteins involved in (BRCA-related) breast carcinogenesis was assessed by immunohistochemistry. A prediction model for BRCA-related versus non-BRCA-related breast cancer was developed using Lasso logistic regression analysis with cross-validation. The model was assessed for its discriminative value and clinical usefulness. The optimal prediction model included 14 predictors (age, cyclinD1, ERα, ERß, FGFR2, FGFR3, FGFR4, GLUT1, IGFR, Ki67, mitotic activity index, MLH1, p120, and TOP2A), showed excellent discriminative performance (area under the receiving operating characteristic curve=0.943; 95% confidence interval=0.909-0.978), and reasonable calibration. To enhance possible implementation, we developed an alternative model only considering more widely available immunostains. This model included 15 predictors (age, BCL2, CK5/6, CK8/18, cyclinD1, E-cadherin, ERα, HER2, Ki67, mitotic activity index , MLH1, p16, PMS2, PR, and vimentin), and still showed very good discriminative performance (area under the receiving operating characteristic curve=0.853; 95% confidence interval=0.795-0.911). We present a well-applicable and accurate tool to predict which breast cancer patients may have an underlying BRCA germline mutation, largely consisting of immunohistochemical markers independent of clinical characteristics. This may improve identification of potential BRCA germline mutation carriers and optimize referral for germline mutation testing.

A systematic review on the frequency of BRCA promoter methylation in breast and ovarian carcinomas of BRCA germline mutation carriers: Mutually exclusive, or not?

BACKGROUND: A considerable number of breast and ovarian carcinomas are due to underlying BRCA gene aberrations. Of these, BRCA germline mutations and BRCA promoter methylation are thought to be mutually exclusive, which could be exploited in clinical practice. However, this paradigm has not been studied extensively and systematically. OBJECTIVE: To systematically investigate to what extent BRCA promoter methylation has been reported in breast and ovarian carcinomas of BRCA germline mutation carriers. METHODS: A comprehensive search on BRCA promoter methylation was performed in PubMed and Embase databases. Two authors independently selected studies, assessed study quality and extracted data according to PRISMA and QUADAS-2 guidelines. RESULTS: 21 articles met the inclusion criteria. BRCA1 methylation was found in at least 10/276 (3,6%) breast and 2/174 (1,1%) ovarian carcinomas of BRCA germline mutation carriers, and BRCA2 methylation was found in at least 7/131 (5.3%) breast and 0/51 (0.0%) ovarian carcinomas of BRCA germline mutation carriers. Methylation frequencies varied between individual CpG sites. The selected studies showed important differences in methodology and performed in general a limited methylation and incomplete mutation analysis. CONCLUSIONS: BRCA methylation is rare in breast and ovarian carcinomas of BRCA germline mutation carriers, although the frequency of BRCA promoter methylation may be underestimated. This could have major implications for clinical practice, including referral for genetic testing and BRCAness analysis for treatment decision-making.

BRCA promoter methylation in sporadic versus BRCA germline mutation-related breast cancers.

BACKGROUND: In breast cancer, BRCA promoter hypermethylation and BRCA germline mutations are said to occur together rarely, but this property has not yet been translated into a clinical test. Our aim in this study was to investigate the diagnostic value of BRCA1/2 methylation in distinguishing breast carcinomas of BRCA1 and BRCA2 germline mutation carriers from sporadic breast carcinomas using a recently developed BRCA methylation assay based on methylation-specific multiplex ligation-dependent probe amplification (MS-MLPA). METHODS: MS-MLPAs were performed to assess BRCA1 and BRCA2 methylation in breast carcinoma tissues from 39 BRCA1 and 33 BRCA2 germline mutation carriers, 80 patients with sporadic breast cancer, and normal breast tissues from 5 BRCA1 and 4 BRCA2 mutation carriers and 5 nonmutation carriers. RESULTS: Methylation frequencies varied considerably between CpG sites across the BRCA1 and BRCA2 promoters. Some CpG sites were methylated more frequently in BRCA1/2-related than in sporadic carcinomas, whereas other CpG sites were methylated more frequently in sporadic carcinomas, with large variances in sensitivity and specificity as a consequence. CONCLUSIONS: The diagnostic value of BRCA promoter methylation analysis in distinguishing BRCA1/2-related from sporadic breast carcinomas seems to be considerably dependent on the targeted CpG sites. These findings are important for adequate use of BRCA methylation analysis as a prescreening tool for BRCA germline genetic testing or to identify BRCAness patients who may benefit from targeted therapies such as poly(adenosine diphosphate-ribose) polymerase inhibitors.

Moral Duties of Genomics Researchers: Why Personalized Medicine Requires a Collective Approach.

Advances in genome sequencing together with the introduction of personalized medicine offer promising new avenues for research and precision treatment, particularly in the field of oncology. At the same time, the convergence of genomics, bioinformatics, and the collection of human tissues and patient data creates novel moral duties for researchers. After all, unprecedented amounts of potentially sensitive information are being generated. Over time, traditional research ethics principles aimed at protecting individual participants have become supplemented with social obligations related to the interests of society and the research enterprise at large, illustrating that genomic medicine is also a social endeavor. In this review we provide a comprehensive assembly of moral duties that have been attributed to genomics researchers and offer suggestions for responsible advancement of personalized genomic cancer care.

miRNA expression patterns in normal breast tissue and invasive breast cancers of BRCA1 and BRCA2 germ-line mutation carriers.

miRNA deregulation has been found to promote carcinogenesis. Little is known about miRNA deregulation in hereditary breast tumors as no miRNA expression profiling studies have been performed in normal breast tissue of BRCA1 and BRCA2 mutation carriers. miRNA profiles of 17 BRCA1- and 9 BRCA2-associated breast carcinomas were analyzed using microarrays. Normal breast tissues from BRCA1 and BRCA2 mutation carriers (both n = 5) and non-mutation carriers (n = 10) were also included. Candidate miRNAs were validated by qRT-PCR. Breast carcinomas showed extensive miRNA alteration compared to normal breast tissues in BRCA1 and BRCA2 mutation carriers. Moreover, normal breast tissue from BRCA1 mutation carriers already showed miRNA alterations compared to non-mutation carriers. Chromosomal distribution analysis showed several hotspots containing down- or up-regulated miRNAs. Pathway analysis yielded many similarities between the BRCA1 and BRCA2 axes with miRNAs involved in cell cycle regulation, proliferation and apoptosis. Lesser known pathways were also affected, including cellular movement and protein trafficking. This study provides a comprehensive insight into the potential role of miRNA deregulation in BRCA1/2-associated breast carcinogenesis. The observed extensive miRNA deregulation is likely the result of genome-wide effects of chromosomal instability caused by impaired BRCA1 or BRCA2 function. This study's results also suggest the existence of common pathways driving breast carcinogenesis in both BRCA1 and BRCA2 germ-line mutation carriers.