Importance of Complete Pathology Reporting for Neuroendocrine Carcinoma: WHO Guidelines Are a Good Start but Not Enough.

BACKGROUND: Neuroendocrine carcinomas (NECs) are diagnosed through a combination of immunohistochemistry (IHC) and morphology according to WHO guidelines. The presence of these crucial components for classification in the pathology report is critical for appropriate understanding of the report especially since terminology and definitions of NEC have been changing a lot lately. OBJECTIVES: The aim of this study is to assess the effect of WHO 2010 on the quality of pathology reporting for NEC and to assess the relevance of the criteria demanded. METHODS: Patients registered with a NEC (gastrointestinal or unknown origin) in the Netherlands Cancer Registry (NCR) between 2008 and 2012 were included. Local pathology reports were reviewed for reporting of morphology and IHC comparing 2008-2010 (baseline) with 2011-2012. The diagnosis of NEC was confirmed according to WHO 2010, if synaptophysin or chromogranin were positive in a majority of cells and Ki-67 or mitotic count confirmed a grade 3 tumour. RESULTS: 591 patients were registered with a NEC in the NCR. 436 pathology reports were reviewed. 62.2% of reports described morphology, IHC and grading in accordance with WHO 2010. Reporting of these parameters increased from 50.0% in 2008 to 69.2% in 2012. Large-cell NEC could be confirmed in 45.0% of patients, increasing from 31.7% in 2008 to 56.7% in 2012 (p = 0.02). Other diagnoses included neuroendocrine tumour (NET) G1/2 13.3%, small-cell carcinoma 2.8%, no neuroendocrine neoplasm (NEN) 17.7%, NEN grade unknown 21.3%. Mean survival was 1.1 years in large cell NEC versus 2.2 years in NET G1/2 (p = 0.005). CONCLUSION: Implementation of the WHO 2010 guideline is associated with a significant increase in reporting parameters needed for classification. Stratification of patients is more reliable based on reports containing all parameters. Guidelines alone however are not enough to warrant complete reporting; synoptic reports might be needed.

Disparities in the management of cutaneous malignant melanoma. A population-based high-resolution study.

Population-based cancer registry data from three Spanish areas were used to assess the patterns of care and adherence to guidelines for cutaneous malignant melanoma. We included 934 cases diagnosed in 2009-2013. Completeness of the pathology reports, imaging for detecting distant metastasis and the use of sentinel lymph node biopsy (SLNB) were analysed. The proportion of pathology reports that mentioned the essential pathological features required for T staging was 93%, ranging across geographic areas from 81% to 98% (p < 0.001). The percentage of low-risk patients who underwent no imaging studies, as proposed by guidelines, or only chest imaging ranged among areas from 0.6% to 84% (p < 0.001). Of the patients with clinically node-negative melanoma >1 mm thick and no distant metastases, 68% underwent SLNB, varying by area from 61% to 78% (p = 0.017). This study revealed wide geographic variation in different aspects of melanoma care. The use of a standardised structured pathology report could strengthen the completeness of reporting. Improvement strategies should also include efforts to reduce overuse of imaging in low-risk patients and to increase the adherence to guidelines recommendations on the use of SLNB.

Report formatting in laboratory medicine - a call for harmony.

The results of medical laboratory testing are only useful if they lead to appropriate actions by medical practitioners and/or patients. An underappreciated component of the medical testing process is the transfer of the information from the laboratory report into the reader's brain. The format of laboratory reports can be determined by the testing laboratory, which may issue a formatted report, or by electronic systems receiving information from laboratories and controlling the report format. As doctors can receive information from many laboratories, interpreting information from reports in a safe and rapid manner is facilitated by having similar report layouts and formats. Using Australia as an example, there is a wide variation in report formats in spite of a body of work to define standards for reporting. In addition to standardising of report formats, consideration needs to be given to optimisation of report formatting to facilitate rapid and unambiguous reading of the report and also interpretation of the data. Innovative report formats have been developed by some laboratories; however, wide adoption has not followed. The need to balance uniformity of reporting with appropriate innovation is a challenge for safe reporting of laboratory results. This paper discusses the current status and opportunity for improvement in safety and efficiency of the reading of laboratory reports, using current practise and developments in Australia as examples.

Reporting of mitotic rate in cutaneous melanoma: A study using the national cancer data base.

BACKGROUND: The seventh edition of the American Joint Commission on Cancer staging manual (AJCC7, published 2009), updated thin cutaneous melanoma staging protocols with the incorporation of mitotic rate (MR). In these patients, higher MR is associated with decreased survival. This study utilizes the National Cancer Data Base (NCDB) to evaluate MR reporting since AJCC7. METHODS: The NCDB was queried for patients with primary cutaneous melanoma from 1998 to 2013. Because MR reporting was infrequent prior to implementing AJCC7, records from 2010 to 2013 were analyzed. Categorical variables were compared with chi-square tests; univariate and multivariate logistic regression models were constructed to determine the effects of covariates on MR reporting. RESULTS: A total of 107,134 patients met inclusion criteria. From 2010 to 2013, MR reporting increased dramatically (64.3-80.9%). On multivariate analysis, factors significantly related to increased MR reporting include later diagnosis year, T-classification (T1a and b vs. T1), facility type (academic vs. other specified types of cancer programs), facility volume, patient income, level of education, and county population (metropolitan vs. urban and rural). CONCLUSIONS: MR reporting increased dramatically after the introduction of AJCC7; however, disparities in reporting remain across facility types. Further investigation of procedures performed in academic settings that may influence reporting of MR is warranted. J. Surg. Oncol. 2017;115:281-286. © 2017 Wiley Periodicals, Inc.

Animal and cellular models of human disease.

In this eighteenth (2016) Annual Review Issue of The Journal of Pathology, we present a collection of 19 invited review articles that cover different aspects of cellular and animal models of disease. These include genetically-engineered models, chemically-induced models, naturally-occurring models, and combinations thereof, with the focus on recent methodological and conceptual developments across a wide range of human diseases.

Completeness of pathology reports in stage II colorectal cancer.

INTRODUCTION: The completeness of the pathological examination of resected colon cancer specimens is important for further clinical management. We reviewed the pathological reports of 356 patients regarding the five factors (pT-stage, tumor differentiation grade, lymphovascular invasion, tumor perforation and lymph node metastasis status) that are used to identify high-risk stage II colon cancers, as well as their impact on overall survival (OS). METHODS: All patients with stage II colon cancer who were included in the first five years of the MATCH study (1 July 2007 to 1 July 2012) were selected (n = 356). The hazard ratios of relevant risk factors were calculated using Cox Proportional Hazards analyses. RESULTS: In as many as 69.1% of the pathology reports, the desired information on one or more risk factors was considered incomplete. In multivariable analysis, age (HR: 1.07, 95%CI 1.04-1.10, p < .001), moderately- (HR: 0.35, 95%CI 0.18-0.70, p = .003) and well (HR 0.11, 95%CI 0.01-0.89, p = .038) differentiated tumors were significantly associated with OS. CONCLUSIONS: Pathology reports should better describe the five high-risk factors, in order to enable proper patient selection for further treatment. Chemotherapy may be offered to stage II patients only in select instances, yet a definitive indication is still unavailable.

Diagnosing lung cancer in the 21st century: are we ready to meet the challenge of individualized care?

BACKGROUND: Histologic and molecular subtyping have become increasingly important as predictors of treatment benefit in lung cancer. The objective of the present study was to determine whether current diagnostic approaches provide adequate tissue to allow for individualized treatment decisions. METHODS: Our retrospective cohort study of new lung cancer patients seen at an academic centre between July 2007 and June 2008 collected baseline demographic and diagnostic information, including mode of diagnosis, type of diagnostic material, and pathology diagnosis. RESULTS: Of the 431 study patients, 20% had stage i or ii non-small-cell lung cancer (nsclc), 24% stage iii disease, and 39% stage iv nsclc. Three quarters of the small-cell lung cancer (sclc) cases were extensive stage. Diagnostically, 18% of patients had sclc; 30%, adenocarcinoma; 27%, squamous-cell cancer; 2%, large-cell carcinoma; 1%, bronchoalveolar carcinoma; 1%, mixed histology; 18%, nsclc not otherwise specified; 4%, other; and 2%, no pathology diagnosis. Surgical pathology material was available in 80% of cases, and cytology material alone in 20%. Surgical pathology material was more common in patients with early-stage than with advanced disease (89% for stages i and ii vs. 74% for stages iii and iv, p < 0.0001). The pathology report included ambiguous terms in 24% of cases: "consistent" (12%), "suspicious" (3%), "favour" (2%), "suggestive" (2%), "likely" (1%), "compatible" with malignancy (1%), "at least" (1%), "atypical" (0.5%), and "no pathology" (1.5%). CONCLUSIONS: Current diagnostic approaches in most lung cancer patients appear adequate, but complete histopathologic identification is missing in nearly 20% of cases, and some uncertainty as to the final diagnosis is expressed in 24% of pathology reports. Some improvement in diagnostic sampling and pathology reporting are required to allow for implementation of current treatment approaches.

A comparison of WHO-5 and ICC classifications in a series of myeloid neoplasms, considerations for hematopathologists and molecular pathologists.

OBJECTIVES: The International Consensus Classification (ICC) and 5th Edition of the World Health Organization Classification (WHO-5) made substantive updates to the classification of myeloid neoplasms. This study compares the systems in a series of myeloid neoplasms with increased blasts, analyzing implications for diagnostic workflow and reporting. METHODS: Bone marrow biopsies categorized as myelodysplastic syndrome with excess blasts (MDS-EB) or acute myeloid leukemia (AML) by WHO-R4 were identified. Results of morphology review, karyotype, fluorescence in situ hybridization, and next-generation sequencing were compiled. Cases were retrospectively re-classified by WHO-5 and ICC. RESULTS: 46 cases were reviewed. 28 cases (61 %) had ≥20 % blasts, with the remaining cases having 5-19.5 % blasts. The most common differences in classification were 1) the designation of MDS versus MDS/AML (10/46, 22 %) for cases with 10-19 % blasts and 2) the ICC's designation of TP53 variants as a separate classifier for AML (8/46, 17 %). Bi-allelic/multi-hit TP53 alterations were identified in 15 cases (33 %). Variants of potential germline significance were identified in 29 (63 %) cases. CONCLUSIONS: While terminology differences between WHO-5 and ICC exist, both systems invoke similar opportunities for improved reporting: standardized classification of pathogenic variants (notably TP53), streamlined systems to evaluate for potential germline variants, and integrated reporting of morphologic and genetic data.

Ex vivo 3D scanning and specimen mapping in anatomic pathology.

Structured light three-dimensional (3D) scanning is a ubiquitous mainstay of object inspection and quality control in industrial manufacturing, and has recently been integrated into various medical disciplines. Photorealistic 3D scans can readily be acquired from fresh or formalin-fixed tissue and have potential for use within anatomic pathology (AP) in a variety of scenarios, ranging from direct clinical care to documentation and education. Methods for scanning and post-processing of fresh surgical specimens rely on relatively low-cost and technically simple procedures. Here, we demonstrate potential use of 3D scanning in surgical pathology in the form of a mixed media pathology report with a novel post-scan virtual inking and marking technique to precisely demarcate areas of tissue sectioning and details of final tumor and margin status. We display a sample mixed-media pathology report (3D specimen map) which integrates 3D and conventional pathology reporting methods. Finally, we describe the potential utility of 3D specimen modeling in both didactic and experiential teaching of gross pathology lab procedures.

Missing parameters in malignant polyp histology reports: can appropriate decisions be made?

The treatment of colorectal malignant polyps is dependent upon quality reporting of the histopathological features known to predict the risk of residual disease or lymph node metastasis. The Royal College of Pathologists of Australasia (RCPA) has produced protocols covering mandatory and recommended pathological parameters to be included in the pathology reporting of malignant polyps. This paper aimed to assess the quality of the pathological reporting in a population-wide analysis from 2011-2019 in Queensland, Australia. A retrospective population-wide cohort study was performed using the Queensland Oncology Repository as a data source. The number of missing pathological parameters (assessed against the RCPA protocol standards and guidelines) for each patient was then summed. Demographic and other patient details were collated. The number of patients whose recommended treatment could theoretically be altered by the full reporting of missing parameters was calculated. A total of 1,646 histopathological reports of malignant polyps were reviewed. From this, 30.8% of all reports had a sufficient number of missing parameters that may have seen an alternate management strategy chosen. The most commonly under-reported parameter from the standards was either a Haggitt or Kikuchi level with 48.6% missing. Synoptic reporting significantly reduced the mean number of missing pathological parameters (p<0.001) There was a significant improvement in the number of missing pathological details over time (p<0.001). Accurate and complete pathology reports are essential to formulate appropriate surgical recommendations after the resection of malignant polyps. In this population-based study, pathology reports remain incomplete for the established parameters despite the introduction of an RCPA structured reporting protocol. Fortunately, the quality of pathological reporting has improved since the introduction of the first RCPA protocol covering reporting of malignant polyps.

The Driver Role of Pathologists in Endocrine Oncology: What Clinicians Seek in Pathology Reports.

Endocrine neoplasia represents an increasingly broad spectrum of disorders. Endocrine neoplasms range from incidental findings to potentially lethal malignancies. In this paper, we cover the impact of pathology in the interpretation of the clinic-pathological, genetic, and radiographic features underpinning these neoplasms. We highlight the critical role of multidisciplinary interactions in structuring a rational diagnostic and efficient therapeutic plan and emphasize the role of histopathological input in decision-making. In this context, standardized pathology reporting and second opinion endocrine pathology review represent relevant tools to improve the overall diagnostic workup of patients affected by endocrine tumors in every specific scenario. In fact, although a relevant proportion of cases may be correctly identified based on clinical presentation and biochemical/imaging investigations, a subset of cases presents with atypical findings that may lead to an inappropriate diagnosis and treatment plan based on a wrong pathological diagnosis if all pieces of the puzzle are not correctly considered. Pathologists have a responsibility to actively guide clinicians before and during surgical procedures to prevent unnecessary interventions. In all areas of endocrine pathology, pathologists must understand the complexity of tissue preservation and assay sensitivities and specificities to ensure the optimal quality and interpretation of diagnostic material. Finally, pathologists are central actors in tumor tissue biobanking, which is an expanding field in oncology that should be promoted while adhering to strict ethical and methodological standards.

Quality of pathology reporting and adherence to guidelines in rectal neuroendocrine neoplasms: a Belgian national study.

The incidence of neuroendocrine neoplasms (NEN) in the rectum is rising since the introduction of colonoscopy screening programs. Guidelines, such as the European NeuroEndocrine Tumor Society (ENETS) algorithm, are mainly based on expert opinion. The goal of this nationwide study is to gain a better insight into the evolution in pathology reporting and adherence to the ENETS guidelines in Belgium. In Belgium, all NENs have to be reported to the Belgian Cancer Registry. We thoroughly reviewed all available pathology reports, coded as rectal NEN between 2004 and 2015, and reclassified according to World Health Organisation (WHO) classification 2019. To evaluate the adherence to the ENETS guidelines, population-based cancer registry data were linked with the medical procedures of the Belgian Health Insurance database. A total of 670 rectal NEN were retained and 16% of the cases needed reclassification. Annual incidence between 2004 and 2015 tripled from 0,20 to 0,61 per 100.000 inhabitants. Reporting of Ki67 proliferation index ameliorated most, while reporting of tumor size, lymphovascular and perineural invasion remained disappointing. Endoscopic ultrasound was performed in only 36.6% of the cases, while the mostly recommended mode of treatment (endoscopic/surgical/no resection) was followed in the majority of the cases. Incidence of rectal NEN in Belgium increased throughout the years and quality of pathology reporting improved especially after the WHO classification update in 2010. The growing awareness and knowledge among clinicians and pathologists in the community counters the need for centralization.

Diagnosis and Pathologic Reporting of Prostate Cancer in the Era of MRI-Targeted Prostate Biopsy.

Historically, the detection of prostate cancer relied upon a systematic yet random sampling of the prostate by transrectal ultrasound guided biopsy. This approach was a nontargeted technique that led to the under detection of cancers at biopsy and the upgrading of cancers at radical prostatectomy. Multiparametric MRI-targeted prostate biopsy allows for an image-directed approach to the identification of prostate cancer. MRI-targeted biopsy of the prostate is superior for the detection of clinically significant prostate cancer. As this technique has become more prevalent among urologists, pathologists need to recognize how this development impacts cancer diagnosis and reporting.

Automated Generation of Synoptic Reports from Narrative Pathology Reports in University Malaya Medical Centre Using Natural Language Processing.

Pathology reports represent a primary source of information for cancer registries. University Malaya Medical Centre (UMMC) is a tertiary hospital responsible for training pathologists; thus narrative reporting becomes important. However, the unstructured free-text reports made the information extraction process tedious for clinical audits and data analysis-related research. This study aims to develop an automated natural language processing (NLP) algorithm to summarize the existing narrative breast pathology report from UMMC to a narrower structured synoptic pathology report with a checklist-style report template to ease the creation of pathology reports. The development of the rule-based NLP algorithm was based on the R programming language by using 593 pathology specimens from 174 patients provided by the Department of Pathology, UMMC. The pathologist provides specific keywords for data elements to define the semantic rules of the NLP. The system was evaluated by calculating the precision, recall, and F1-score. The proposed NLP algorithm achieved a micro-F1 score of 99.50% and a macro-F1 score of 98.97% on 178 specimens with 25 data elements. This achievement correlated to clinicians' needs, which could improve communication between pathologists and clinicians. The study presented here is significant, as structured data is easily minable and could generate important insights.

Current Pathology Model of Pancreatic Cancer.

Pancreatic cancer (PC) is one of the most aggressive and lethal malignant neoplasms, ranking in seventh place in the world in terms of the incidence of death, with overall 5-year survival rates still below 10%. The knowledge about PC pathomechanisms is rapidly expanding. Daily reports reveal new aspects of tumor biology, including its molecular and morphological heterogeneity, explain complicated "cross-talk" that happens between the cancer cells and tumor stroma, or the nature of the PC-associated neural remodeling (PANR). Staying up-to-date is hard and crucial at the same time. In this review, we are focusing on a comprehensive summary of PC aspects that are important in pathologic reporting, impact patients' outcomes, and bring meaningful information for clinicians. Finally, we show promising new trends in diagnostic technologies that might bring a difference in PC early diagnosis.

Pathology reporting of pelvic exenteration specimens for locally recurrent rectal cancer.

INTRODUCTION: Reporting of pelvic exenteration specimens for locally recurrent rectal cancer (LRRC) can be challenging for structured pathological analysis and currently, there is a lack of specific guidelines. The aim of this study was to assess the quality of pathology reporting in a cohort of patients who underwent pelvic exenteration for LRRC in a high-volume tertiary unit. MATERIALS AND METHODS: In a retrospective analysis of histopathology reports of consecutive patients who underwent pelvic exenteration for LRRC from 1996 to 2018, the quality of pathology reporting was assessed using the Structure Reporting Protocol for Colorectal Cancer. The primary endpoint was the completeness of pathology reporting, secondary endpoints were the association between the reporting style (narrative versus synoptic), reporting period (the first half versus the second half), as well as the activity of the pathologists with the completeness of pathology reporting. RESULTS: 221 patients who underwent pelvic exenteration for LRRC were included into the study. There was a high variability in completeness of pathology reporting within the cohort, ranging from 9.5% to 100%. Notably, microscopic clearance was reported in only 92.4% of the reports. Overall, a significantly higher rate of completeness was observed in synoptic reports when compared to narrative reports and in more recent compared to earlier reports. There was no significant association between the activity of pathologists and the completeness of reporting. CONCLUSIONS: This study shows a significant variability in the quality of reporting in pelvic exenteration for LRRC. The use of synoptic reporting clearly resulted in more complete reports.

Bias Introduced by Relying on Incomplete Electronic Pathology Reporting for Rapid Case Ascertainment in Patient Contact Studies.

BACKGROUND: Relying on electronic pathology (ePath) reporting to state cancer registries for rapid ascertainment of cases for patient contact research studies may introduce bias if the patient populations differ for reporting facilities with vs without ePath. We examined changes between 2014-2019 in the percent of cases reported to the New York State Cancer Registry by ePath within 3 months of diagnosis and characteristics of the most recent cases by ePath status. Our goal was to assess the potential bias introduced by relying on incomplete ePath reporting for patient recruitment. METHODS: We restricted our analysis to first malignant cancers diagnosed in New York State residents aged 18 years and older. We examined patient characteristics and used χ2 tests to examine differences in the distribution of each characteristic by ePath status for diagnosis years 2017-2019, and used multivariable-adjusted logistic regression to calculate odds ratios and 95% CIs for the association between each patient characteristic and ePath status for all 2017-2019 cancers combined and common cancer sites. All analyses were conducted using SAS 9.4. RESULTS: The percent of cases reported by ePath increased over time from 15.7% in 2014 to 44.8% in 2019. Among 264,607 cancers diagnosed in 2017-2019 and reported through July 2021, there were statistically significant differences in all variables examined by ePath status (all P < .0001). For all cancers combined, cases reported by ePath were more likely to be younger, female, non-Hispanic White, married, live outside of New York City/ Long Island, still be alive, and have received treatment. We observed statistically significant odds ratios for the associations between all variables examined and ePath status for all cancers combined, but the strength and statistical significance of the associations varied by cancer site. CONCLUSIONS: Our results indicate that relying on incomplete ePath reporting for rapid case ascertainment will introduce selection bias in the study sample for patient contact studies. This bias should decrease as additional facilities acquire ePath reporting capability.

Informatics Methods and Infrastructure Needed to Study Factors Associated with High Incidence of Pediatric Brain and Central Nervous System Tumors in Kentucky.

Pediatric brain and central nervous system tumors (PBCNSTs) are the most common solid tumors and are the leading cause of disease-related death in US children. PBCNST incidence rates in Kentucky are significantly higher than in the United States as a whole, and are even higher among Kentucky's Appalachian children. To understand and eventually eliminate such disparities, population-based research is needed to gain a thorough understanding of the epidemiology and etiology of the disease. This multi-institutional population-based retrospective cohort study is designed to identify factors associated with the high incidence of PBCNST in Kentucky, leveraging the infrastructure provided by the Kentucky Cancer Registry, its Virtual Tissue Repository (VTR), and the National Institutes of Health Gabriella Miller Kids First Data Resource Center (DRC). Spatiotemporal scan statistics have been used to explore geographic patterns of risk measured by standardized incidence ratios (SIRs) with 95% confidence intervals. The VTR is being used to collect biospecimens for the population-based cohort of PBCNST tissues that are being sequenced by Center for Data Driven Discovery in Biomedicine (D3b) at the Children's Hospital of Philadelphia (CHOP) with support from the Kids First DRC. After adjusting for demographic factors, we assess their potential relationship to environmental factors. We have identified regions in north-central and eastern Appalachian Kentucky where children experienced a significant increased risk of developing PBCNST from 1995-2017 (SIR, 1.48; 95% CI, 1.34-1.62). The VTR has been successful in the collection of a population-based cohort of 215 PBCNST specimens. Timely establishment of legal agreements for data sharing and tissue acquisition proved to be challenging which has been somewhat mitigated by the adoption of national agreement templates. Coronavirus disease 2019 (COVID-19) severely limited the generation of sequencing results due to laboratory shutdowns. However, tissue specimens processed before the shutdown indicated that punches were inferior to scrolls for generating enough quality material for DNA and RNA extraction. Informatics infrastructures that were developed have demonstrated the feasibility of our approach to generate and retrieve molecular results. Our study shows that population-based studies using historical tissue specimens are feasible and practical, but require significant investments in technical infrastructures.

Natural Language Processing for Surveillance of Cervical and Anal Cancer and Precancer: Algorithm Development and Split-Validation Study.

BACKGROUND: Accurate identification of new diagnoses of human papillomavirus-associated cancers and precancers is an important step toward the development of strategies that optimize the use of human papillomavirus vaccines. The diagnosis of human papillomavirus cancers hinges on a histopathologic report, which is typically stored in electronic medical records as free-form, or unstructured, narrative text. Previous efforts to perform surveillance for human papillomavirus cancers have relied on the manual review of pathology reports to extract diagnostic information, a process that is both labor- and resource-intensive. Natural language processing can be used to automate the structuring and extraction of clinical data from unstructured narrative text in medical records and may provide a practical and effective method for identifying patients with vaccine-preventable human papillomavirus disease for surveillance and research. OBJECTIVE: This study's objective was to develop and assess the accuracy of a natural language processing algorithm for the identification of individuals with cancer or precancer of the cervix and anus. METHODS: A pipeline-based natural language processing algorithm was developed, which incorporated machine learning and rule-based methods to extract diagnostic elements from the narrative pathology reports. To test the algorithm's classification accuracy, we used a split-validation study design. Full-length cervical and anal pathology reports were randomly selected from 4 clinical pathology laboratories. Two study team members, blinded to the classifications produced by the natural language processing algorithm, manually and independently reviewed all reports and classified them at the document level according to 2 domains (diagnosis and human papillomavirus testing results). Using the manual review as the gold standard, the algorithm's performance was evaluated using standard measurements of accuracy, recall, precision, and F-measure. RESULTS: The natural language processing algorithm's performance was validated on 949 pathology reports. The algorithm demonstrated accurate identification of abnormal cytology, histology, and positive human papillomavirus tests with accuracies greater than 0.91. Precision was lowest for anal histology reports (0.87, 95% CI 0.59-0.98) and highest for cervical cytology (0.98, 95% CI 0.95-0.99). The natural language processing algorithm missed 2 out of the 15 abnormal anal histology reports, which led to a relatively low recall (0.68, 95% CI 0.43-0.87). CONCLUSIONS: This study outlines the development and validation of a freely available and easily implementable natural language processing algorithm that can automate the extraction and classification of clinical data from cervical and anal cytology and histology.

An international survey-based study on colorectal cancer pathology reporting-guidelines versus local practice.

Different guidelines for colorectal cancer (CRC) pathology reporting have been published. We aimed to identify differences between publicly available CRC reporting guidelines and to survey pathologists from different countries to establish the degree of guideline implementation in local routine practice. We compared all core and non-core items of CRC reporting guidelines to identify discrepancies. We then created a survey, which was sent out to 782 pathologists practicing in 30 different countries. It included questions on the demographics of the reporting pathologist as well as resection specimen handling and microscopic evaluation, grading, staging, and additional techniques, such as immunohistochemistry or molecular pathology. First, core and non-core items of five national CRC reporting guidelines were compared and 12 items were found to differ. Different items are considered core or non-core by different guidelines and more than one TNM staging edition was applied across guidelines. The survey was completed by 143 pathologists from 30 countries. We identified differences between local practice and guidelines with potential clinical impact, e.g., tumor budding was never reported by 28.7% of responders, although it has prognostic value for survival in stage II CRC. This is the first international study comparing CRC pathology reporting guidelines with real-world local practices. There are differences in CRC pathology reporting guidelines and in guideline implementation into local practice, both with potential impact on patient care. Harmonization of datasets, use of templates, and audits of local pathology practice are needed to ensure best possible quality of CRC pathology reporting.