Systems approach to enhance Lynch syndrome diagnosis through tumour testing.

BACKGROUND: Guidelines recommend universal mismatch repair (MMR) tumour testing of colorectal adenocarcinomas (CRCs) to screen for Lynch syndrome (LS). However, its implementation remains disjointed and referral for genetic testing dismal, particularly among minorities. We aimed to increase referral, cancer genetic testing and eventually LS diagnosis by developing the CLEAR LS (Closed Loop Enhanced Assessment and Referral for Lynch Syndrome) intervention, a systems approach which in the second phase was automated. METHODS: This is a cohort study of all patients diagnosed with CRC at an academic centre between 1 January 2012, when implementation of universal CRC testing began, and 31 January 2021. The original cohort spanned through 31 May 2015. Tumour testing included MMR immunohistochemistry, followed by BRAF V600E/MLH1 promoter methylation testing when indicated. The intervention included a manual phase (1 June 2015 through 31 July 2018), which systematised pathology screening and cancer genetics (CG) referral mechanisms, and an automated phase (1 August 2018 through 31 January 2021) using computer programming. RESULTS: A total of 249/1541 CRC (17.38%) had MMR loss of expression and 129 (8.37%) qualified for CG evaluation. Referral was 27.58% in the original cohort and 92.1% in the intervention (p<0.001). Patients seen by CG among referred were 27.58% in the original cohort and 74.3% in the intervention (p two-sided<0.001). The distribution of race/ethnicity among patients qualifying and referred for CG evaluation was not significantly different across cohorts. LS diagnosis increased from 0.56% (original cohort) to 1.43% (intervention). Cost per new diagnosis of LS decreased from US$173 675 to $87 960 from original cohort to intervention. CONCLUSION: Implementation of systematic case identification and referral support mechanisms significantly increased the proportion of patients undergoing genetic testing and doubled the percentage of patients diagnosed with LS with no referral differences across racial/ethnic groups.

An independent assessment of an artificial intelligence system for prostate cancer detection shows strong diagnostic accuracy.

Prostate cancer is a leading cause of morbidity and mortality for adult males in the US. The diagnosis of prostate carcinoma is usually made on prostate core needle biopsies obtained through a transrectal approach. These biopsies may account for a significant portion of the pathologists' workload, yet variability in the experience and expertise, as well as fatigue of the pathologist may adversely affect the reliability of cancer detection. Machine-learning algorithms are increasingly being developed as tools to aid and improve diagnostic accuracy in anatomic pathology. The Paige Prostate AI-based digital diagnostic is one such tool trained on the digital slide archive of New York's Memorial Sloan Kettering Cancer Center (MSKCC) that categorizes a prostate biopsy whole-slide image as either "Suspicious" or "Not Suspicious" for prostatic adenocarcinoma. To evaluate the performance of this program on prostate biopsies secured, processed, and independently diagnosed at an unrelated institution, we used Paige Prostate to review 1876 prostate core biopsy whole-slide images (WSIs) from our practice at Yale Medicine. Paige Prostate categorizations were compared to the pathology diagnosis originally rendered on the glass slides for each core biopsy. Discrepancies between the rendered diagnosis and categorization by Paige Prostate were each manually reviewed by pathologists with specialized genitourinary pathology expertise. Paige Prostate showed a sensitivity of 97.7% and positive predictive value of 97.9%, and a specificity of 99.3% and negative predictive value of 99.2% in identifying core biopsies with cancer in a data set derived from an independent institution. Areas for improvement were identified in Paige Prostate's handling of poor quality scans. Overall, these results demonstrate the feasibility of porting a machine-learning algorithm to an institution remote from its training set, and highlight the potential of such algorithms as a powerful workflow tool for the evaluation of prostate core biopsies in surgical pathology practices.

Customizing Laboratory Information Systems: Closing the Functionality Gap.

Highly customizable laboratory information systems help to address great variations in laboratory workflows, typical in Pathology. Often, however, built-in customization tools are not sufficient to add all of the desired functionality and improve systems interoperability. Emerging technologies and advances in medicine often create a void in functionality that we call a functionality gap. These gaps have distinct characteristics­a persuasive need to change the way a pathology group operates, the general availability of technology to address the missing functionality, the absence of this technology from your laboratory information system, and inability of built-in customization tools to address it. We emphasize the pervasive nature of these gaps, the role of pathology informatics in closing them, and suggest methods on how to achieve that. We found that a large number of the papers in the Journal of Pathology Informatics are concerned with these functionality gaps, and an even larger proportion of electronic posters and abstracts presented at the Pathology Informatics Summit conference each year deal directly with these unmet needs in pathology practice. A rapid, continuous, and sustainable approach to closing these gaps is critical for Pathology to provide the highest quality of care, adopt new technologies, and meet regulatory and financial challenges. The key element of successfully addressing functionality gaps is gap ownership­the ability to control the entire pathology information infrastructure with access to complementary systems and components. In addition, software developers with detailed domain expertise, equipped with right tools and methodology can effectively address these needs as they emerge.

Proceedings of the Association for Pathology Informatics Bootcamp 2022.

The Pathology Informatics Bootcamp, held annually at the Pathology Informatics Summit, provides pathology trainees with essential knowledge in the rapidly evolving field of Pathology Informatics. With a focus on data analytics, data science, and data management in 2022, the bootcamp addressed the growing importance of data analysis in pathology and laboratory medicine practice. The expansion of data-related subjects in Pathology Informatics Essentials for Residents (PIER) and the Clinical Informatics fellowship examinations highlights the increasing significance of these skills in pathology practice in particular and medicine in general. The curriculum included lectures on databases, programming, analytics, machine learning basics, and specialized topics like anatomic pathology data analysis and dashboarding.

Organizational preparedness for the use of large language models in pathology informatics.

In this paper, we consider the current and potential role of the latest generation of Large Language Models (LLMs) in medical informatics, particularly within the realms of clinical and anatomic pathology. We aim to provide a thorough understanding of the considerations that arise when employing LLMs in healthcare settings, such as determining appropriate use cases and evaluating the advantages and limitations of these models. Furthermore, this paper will consider the infrastructural and organizational requirements necessary for the successful implementation and utilization of LLMs in healthcare environments. We will discuss the importance of addressing education, security, bias, and privacy concerns associated with LLMs in clinical informatics, as well as the need for a robust framework to overcome regulatory, compliance, and legal challenges.

Measuring Faculty Effort: A Quantitative Approach That Aligns Personal and Institutional Goals in Pathology at Yale.

US medical schools increasingly seek ways to reduce costs and improve productivity. One aspect of this effort has been the development of performance-based incentives for individual faculty. A myriad of such plans exist. Typically, they incentivize clinical revenue generation but vary widely in how teaching, investigation, and administrative contributions are recognized. In Pathology at Yale, we have developed a transparent metrically driven approach that recognizes all missions and allows faculty significant control over their career path. Although some metrics derive from traditional measures such as workload relative value units and one's level of grant support, the key concept underpinning our approach is to define one's contributions not in terms of the revenue generated, but rather on the effort devoted to each of our missions, benchmarked against national or local standards. Full-time faculty are paid a competitive rank-based salary and are expected to contribute at least 100% effort in support of the school's missions: clinical, research, education, administration, and professional service. Metrics define the effort assigned to each activity. Faculty achieving greater than 100% effort receive bonus compensation in proportion to their excess effort. By codifying explicitly how such effort is recognized into a single metric (% effort), we achieve a process that better aligns the professional and personal goals of faculty with the aims of the school. To facilitate its implementation, we have developed a web-based software platform called SWAY (Standardized Workload Analysis at Yale) that enables faculty to monitor their progress and record their activities in real time.

TQuest, A Web-Based Platform to Enable Precision Medicine by Linking a Tumor's Genetic Defects to Therapeutic Options.

PURPOSE: Currently, there are only a few software tools designed to assist physicians to translate molecular abnormalities in the cancer genome into potential treatment options. There is a pressing need to develop software to reliably identify known targeted therapies and experimental treatments for patients on the basis of the results of tumor DNA sequencing. METHODS: The TQuest platform includes a data layer, data acquisition layer, search engine, and user interface. It identifies associations between one or more molecular targets and therapeutic options. The data layer consists of indexed interventional clinical trials and an expert-curated database of clinically or experimentally validated associations between mutations and drug response. The data acquisition layer includes an information-harvesting module that keeps an up-to-date full-text index of clinical trials by crawling clinicaltrials.gov and combining it with US Food and Drug Administration label data. The user interface is a Web-based module that allows users to upload genomic variants, tumor morphology, and diagnosis. The search results are qualified and ranked by a relevance score. RESULTS: We have manually curated information for 368 distinct genomic variants of 162 gene targets corresponding to 863 drug and target interactions. The platform currently contains a full-text index of approximately 80,000 interventional clinical trials. We applied TQuest to molecular data from 73 metastatic breast cancers. TQuest identified a total of 276 drugs as potential therapeutic options, ranging from one to 103 per patient. CONCLUSION: TQuest correctly identified all US Food and Drug Administration-approved drugs and routine indications for all cases and also identified many additional drugs that were used in the context of a given molecular abnormality in various clinical trials. The prototype Web application is available at www.tquest.us , and the source code is open and available on GitHub.

Mutation based treatment recommendations from next generation sequencing data: a comparison of web tools.

Interpretation of complex cancer genome data, generated by tumor target profiling platforms, is key for the success of personalized cancer therapy. How to draw therapeutic conclusions from tumor profiling results is not standardized and may vary among commercial and academically-affiliated recommendation tools. We performed targeted sequencing of 315 genes from 75 metastatic breast cancer biopsies using the FoundationOne assay. Results were run through 4 different web tools including the Drug-Gene Interaction Database (DGidb), My Cancer Genome (MCG), Personalized Cancer Therapy (PCT), and cBioPortal, for drug and clinical trial recommendations. These recommendations were compared amongst each other and to those provided by FoundationOne. The identification of a gene as targetable varied across the different recommendation sources. Only 33% of cases had 4 or more sources recommend the same drug for at least one of the usually several altered genes found in tumor biopsies. These results indicate further development and standardization of broadly applicable software tools that assist in our therapeutic interpretation of genomic data is needed. Existing algorithms for data acquisition, integration and interpretation will likely need to incorporate artificial intelligence tools to improve both content and real-time status.

A retrospective population-based comparison of HER2 immunohistochemistry and fluorescence in situ hybridization in breast carcinomas: impact of 2007 American Society of Clinical Oncology/College of American Pathologists criteria.

CONTEXT: In 2007 the American Society of Clinical Oncology/College of American Pathologists made new recommendations for HER2 testing and redefined HER2 positivity. OBJECTIVE: To analyze results from simultaneous HER2 testing with immunohistochemistry and fluorescence in situ hybridization (FISH) in 2590 invasive breast carcinomas between 2002 and 2010, using 2 scoring systems. DESIGN: Cases from between 2002 and 2006 were scored by using original US Food and Drug Administration criteria (N = 1138) and those from between 2007 and 2010 were evaluated according to American Society of Clinical Oncology/College of American Pathologists criteria (N = 1452). Concordance between testing methods and clinicopathologic associations were determined. RESULTS: Overall concordance between immunohistochemistry/FISH in the 9-year period was 96.2% (κ = 0.82), and positive concordance was lower. After 2007, the proportion of HER2/neu-positive and HER2/neu-negative cases was not significantly changed when using immunohistochemistry (10.5% versus 8.9%, P = .22 and 69.4% versus 63%, P = .13, respectively), but the number of equivocal cases was higher (19.9% versus 28%, P < .001). While the proportion of negative cases by FISH remained unchanged after 2007 (86.5% versus 88.2%, P = .76), the number of positive cases was lower (13.4% versus 9.2%, P < .001). In addition, 38 cases (2.6%) were FISH equivocal, 16 of which were also equivocal by immunohistochemistry. Overall, immunohistochemistry/FISH concordance was 95.9% between 2002 and 2006 (κ = 0.82) and 96.4% after 2007 (κ = 0.82). However, an approximately 13% lower positive assay concordance was noted in the last period. CONCLUSIONS: Application of American Society of Clinical Oncology/College of American Pathologists recommendations is associated with comparable overall immunohistochemistry/FISH concordance, reduced positive concordance, and increased equivocal results.

Are amended surgical pathology reports getting to the correct responsible care provider?

OBJECTIVES: Amended reports (AmRs) need to follow patients to treating physicians, to avoid erroneous management based on the original diagnosis. This study was undertaken to determine if AmRs followed the patient appropriately. METHODS: AmRs with diagnostic changes and discrepancies between ordering and treating physicians were tracked. Chart reviews, electronic medical report (EMR) reviews, and interviews were conducted to establish receipt of the AmR by the correct physician. RESULTS: Seven of 60 AmRs had discrepancies between the ordering and treating physicians, all with malignant diagnoses. The AmR was present in the treating physician's chart in only one case. Ordering physicians indicated that AmRs were not forwarded to treating physicians when corrected results arrived after patient referral, under the assumption that the new physician was automatically forwarded pathology updates. No harm was documented in any of our cases. In one case with a significant amendment, the correct information was entered in the patient chart based on a tumor board discussion. A review of two electronic health record systems uncovered significant shortcomings in each delivery system. CONCLUSIONS: AmRs fail to follow the patient's chain of referrals to the correct care provider, and EMR systems lack the functionality to address this failure and alert clinical teams of amendments.

Custom software development for use in a clinical laboratory.

In-house software development for use in a clinical laboratory is a controversial issue. Many of the objections raised are based on outdated software development practices, an exaggeration of the risks involved, and an underestimation of the benefits that can be realized. Buy versus build analyses typically do not consider total costs of ownership, and unfortunately decisions are often made by people who are not directly affected by the workflow obstacles or benefits that result from those decisions. We have been developing custom software for clinical use for over a decade, and this article presents our perspective on this practice. A complete analysis of the decision to develop or purchase must ultimately examine how the end result will mesh with the departmental workflow, and custom-developed solutions typically can have the greater positive impact on efficiency and productivity, substantially altering the decision balance sheet. Involving the end-users in preparation of the functional specifications is crucial to the success of the process. A large development team is not needed, and even a single programmer can develop significant solutions. Many of the risks associated with custom development can be mitigated by a well-structured development process, use of open-source tools, and embracing an agile development philosophy. In-house solutions have the significant advantage of being adaptable to changing departmental needs, contributing to efficient and higher quality patient care.

Standardization of estrogen receptor measurement in breast cancer suggests false-negative results are a function of threshold intensity rather than percentage of positive cells.

PURPOSE: Recent misclassification (false negative) incidents have raised awareness concerning limitations of immunohistochemistry (IHC) in assessment of estrogen receptor (ER) in breast cancer. Here we define a new method for standardization of ER measurement and then examine both change in percentage and threshold of intensity (immunoreactivity) to assess sources for test discordance. METHODS: An assay was developed to quantify ER by using a control tissue microarray (TMA) and a series of cell lines in which ER immunoreactivity was analyzed by quantitative immunoblotting in parallel with the automated quantitative analysis (AQUA) method of quantitative immunofluorescence (QIF). The assay was used to assess the ER protein expression threshold in two independent retrospective cohorts from Yale and was compared with traditional methods. RESULTS: Two methods of analysis showed that change in percentage of positive cells from 10% to 1% did not significantly affect the overall number of ER-positive patients. The standardized assay for ER on two Yale TMA cohorts showed that 67.9% and 82.5% of the patients were above the 2-pg/µg immunoreactivity threshold. We found 9.1% and 19.7% of the patients to be QIF-positive/IHC-negative, and 4.0% and 0.4% to be QIF-negative/IHC-positive for a total of 13.1% and 20.1% discrepant cases when compared with pathologists' judgment of threshold. Assessment of survival for both cohorts showed that patients who were QIF-positive/pathologist-negative had outcomes similar to those of patients who had positive results for both assays. CONCLUSION: Assessment of intensity threshold by using a quantitative, standardized assay on two independent cohorts suggests discordance in the 10% to 20% range with current IHC methods, in which patients with discrepant results have prognostic outcomes similar to ER-positive patients with concordant results.

An approach to object-relational mapping in bioscience domains.

Object-relational mapping (mapping object-oriented software systems to relational databases) remains a challenging undertaking because of the inherent differences between these two technologies. Effective application of this technique in biomedical informatics applications can provide robust, adaptive solutions but requires careful attention to specific issues in the bioscience domain. Mapping from a single, fixed, stably defined object to a table is straightforward, and several standard patterns have been described for mapping inheritance trees to tables. However, mapping becomes more problematic when it involves complex, evolving inheritance trees, sparse, dynamic attributes, or associations and aggregations whose requirements and definitions change frequently during development. All of these seem to be inherent, inescapable facts in the development of biological information systems. Here we describe modifications of the basic patterns of object-relational mapping that are being applied to two systems in ongoing collaborations between biomedical investigators and informaticians. In this approach a small set of objects for each entity collaborates to provide the usual required services of an object-relational implementation including query capacity, transaction management, object relationships (associations, aggregation, composition, many-many relations) and persistence functions. This set of objects also allows dynamic attributes for a selected set of entities by merging an entity-attribute-value approach for data storage with the object-relational approach to managing relationships among objects. Templates are used for class generation to allow flexibility and synchronization of schema/object during design evolution, facilitating changing persistence strategies across the whole system for testing and development. The entity-attribute-value details as well as all database access are encapsulated in a persistence layer which would allow change of this layer to a conventional storage model, if the design evolution is stabilized or for performance requirements, with no change to the domain or presentation layers.