Functional Assessment of Synoptic Pathology Reporting for Ovarian Cancer

A cancer pathology report is the final, written product of a surgical pathology laboratory with important information crucial for patient care and cancer surveillance. Traditional narrative cancer pathology reports contain text describing the information in relevant headings like: final diagnosis, macroscopic description and microscopic description. These descriptive free text reports show significant variability in style and content; do not contain all clinically important data, and might cause an erroneous decision regarding the therapeutic protocols. A large audit of pathology reports of 479 cases from 40 institutions in 11 countries checked for quality report indicators in advanced ovarian, Fallopian tube and peritoneal cancers revealed that basic pathological data were missing in a substantial number of reports with possible adverse consequences on the quality of cancer care. The quality of reports differs significantly by country [1]. Some clinicians at referring institutions were faced with therapeutic dilemmas due to incomplete pathology reports. Data from the National Cancer Data Base in the late 1990s showed that up to one fourth of colon cancer pathology reports lacked information on lymph node status and as many as half of the sarcomas were without assignment of the grade [2].

Contrary to the traditional narrative reports, synoptic reporting (replacing part or all of the free text component) is characterized by uniformity with standardized data elements in the form of checklists. A synoptic standardized pathology report contains an accurate and consistent diagnosis and staging information, and constitutes a reliable basis for therapy recommendations and prognosis. The amount of information pathologists provided in the cancer pathology report has increased recently due to the expansion of scientific knowledge about cancer and advances in molecular diagnostics and personalized medicine. The Cancer Committee of the College of American Pathologists (CAP) recognized the need for more standardized and complete reporting of cancer diagnoses and published the cancer protocols as guidelines and as a resource tool for pathologists, encouraging a synoptic pathology report for completeness and consistency [2]. The changes in pathology reporting started from the point of transition from narrative to synoptic-like reporting and then using a discrete data field format facilitated a level of true synoptic reporting. Subsequently, progressive implementation of CAP electronic cancer checklists helped to create content and formatting standards and to implement electronic reporting solutions. The entire collection of the CAP Cancer Protocols is available online at www.cap.org [3].

The ‘structured', ‘synoptic' or ‘templated' reporting is a type of medical document/report in which the underlying structure is fairly predictable. The synoptic cancer pathology report combines discrete, standardized, human-readable data items instead of free text transcription. Each question and answer in the synoptic report is associated with specific data identifiers and terminology codes, e.g. ICD-O3 (International Classification of Diseases for Oncology) and SNOMED CT (Systematized Nomenclature of Medicine - Clinical Terms) [4,5].

Ovarian cancer is not the most frequent gynecologic malignancy, but it is the most lethal with 65,538 new cases in Europe in 2012 and 42,704 deaths [6]. Epithelial ovarian cancer, accounting for more than 90% of the diagnosed ovarian cancer, has been considered a homogenous disease to be treated with one therapeutic approach (combination of surgery and platinum-based chemotherapy) for years [6]. Over the past quarter of the century, several scientific developments have recognized that ovarian cancer is not a homogeneous disease, but rather a group of diseases presenting with different morphology and biological behavior. Approximately 90% of ovarian cancers are carcinomas with (at least) 5 main types distinguished on the basis of histopathology, immunohistochemistry and molecular genetic analysis [7]. The most frequent type is high-grade serous carcinoma (HGSC) - 70%; followed by endometrioid carcinoma (EC) - 10%, clear-cell carcinoma (CCC) - 10%, mucinous carcinoma (MC) - 3%, and low-grade serous carcinoma (LGSC) - <5% [8]. Different tumor histotypes respond differently to chemotherapy. In the era of personalized cancer medicine, a histopathological diagnosis alone may not be sufficient for treatment to be successful. The recent FIGO (International Federation of Gynecology and Obstetrics) staging classification considered the most relevant prognostic parameters shared by all tumor types and proposed that the histological type should be designated by staging [3]. The new criteria for staging of ovarian and/or Fallopian tube cancers were presented to the FIGO Board and approved by the American Joint Commission on Cancer and the International Union Against Cancer in 2013 [9].

The pathology information systems and digital pathology image acquisition modalities (gross photography, microphotography, telepathology and virtual microscopy) are the two main components of digital pathology. The concept of digital pathology refers to the use of information technology that supports the creation, sharing or exchange of information, including data and images, in order to support the complex workflow from clinician requests to the pathomorphological reporting [10,11,12].

Our intention was to share the experience with structured synoptic reporting provided as a research study on ovarian cancer. The aims of our project were to: (1) develop the e-template for synoptic reporting of epithelial ovarian neoplasms [based on the CAP checklist, recent histological classification of tumors of the World Health Organization (WHO) and FIGO classification], to introduce a uniform and improved quality of cancer pathology reports; (2) recognize the functional aspects of the e-synoptic reporting in the Hospital Information System (HIS) and its potential implementation into the routine workflow, and (3) evaluate quality requirements of synoptic reporting (by content analysis of the traditional and synoptic cancer reports).

Research has been conducted on ovarian cancers and borderline tumors selected from the pathology report catalog or from the current workflow of the Department of Pathology of the Military Institute of Medicine (MIM) in Warsaw during 1 year. According to the standard clinical protocol, staging laparotomy for ovarian/tubal carcinoma usually includes a hysterectomy and bilateral salpingo-oophorectomy, omentectomy (sometimes splenectomy) and lymphadenectomy together with peritoneal biopsies, washing and appendicectomy, and diaphragmatic scrapes in certain cases [13]. For more advanced stages of cancer, some additional tissue samples are included for the microscopic evaluation of distant metastases. Usually, the diagnostic clinical procedure started with preoperative incision or excision biopsy of adnexa (as frozen and paraffin-embedded sections). Such a multistep (overlong) clinical protocol for staging and surgical treatment of the patients with ovarian tumors at our Institute resulted in a few partial (separate) pathology reports referring to the surgical samples of each ovarian tumor case (on average 3-4 partial reports per case).

Altogether, 106 pathology reports were evaluated for the final standardized synoptic cancer reports together with hundreds of corresponding glass slides. Two to four pathologists signed out the checklists (never less than 2).

Routine pathology reports were arranged as text narrative or narrative synoptic-like cancer reports, and were based on the previous WHO histological classification of tumors (edited in 2003) [8] and 7th TNM Classification (UICC, 2009) [14]. The synoptic template or ‘worksheet' for pathology ovarian cancer reporting was prepared (fig. 1) based on the latest edition of the WHO histological classification of ovarian tumors with SNOMED ‘M' coding of surface epithelial neoplasms and TNM classification [7], recent FIGO staging [9] and the CAP cancer checklist [3]. The modified CAP checklist included diagnostic information desired by the clinicians and essential for the pathologists. For analysis of the retrospective pathology reports, the templates for the synoptic reporting were completed by two pathologists after revision of the partial pathology reports and relevant glass slides. In the prospective study, the worksheet was attached to each case as part of the grossing protocols and filled out by the pathologist. One of the pathologists was responsible for supervising the reporting and imaging process of all ovarian cancer cases.

Our HIS (CompuGroupMedical CliniNet) comprises a special pathomorphology module for the pathology laboratory workflow. For the research study, this indispensable module was designed for e-synoptic pathology reporting and was inserted in the pathomorphology module. The data entry sheets were created based on the local form designed for the HIS system and stored on the SQL servers. The final report form presented only selected data built on the basis of RTF templates. The applied technology of the final report enables to connect the RTF template with the XML data source based on the input data. The template was designed using discrete data fields with predefined values and a few free text fields according to the local needs. The discrete elements from the report were captured and stored on the relational database and were included in the generated final report. The cancer reports were in synoptic format but not at the level of automatic encoding in SNOMED. The SNOMED CT coding could be added manually [15]. The Polish pathology laboratories do not use SNOMED CT coding in routine reporting.

Each e-record of an ovarian tumor was designed to contain the text documentation (traditional and standardized synoptic reports) and the representative digital macro- and micro-images (HE with immunohistochemical staining) as well as whole-slide images (WSIs) prepared with a 3DHistech scanner at ×20 magnification. The WSIs have been stored on the server of the Department of Pathology until being transferred to the referred main server of MIM: the Case Center platform.

The series of 42 primary ovarian tumors comprised 29 carcinomas and 13 borderline tumors. The most common tumors were serous carcinomas 62% (18/29) with preponderance of HGSCs over LGSCs: 72% (13/18) vs. 28% (5/18), respectively. Much less common were the following histological types of carcinoma: 3 MC, 3 EC; 2 CCC and 1 undifferentiated carcinoma. The borderline tumors were mainly of the serous type (54%; 7 of 13), 4 were MC and 2 were EC; 67% of carcinomas presented advanced stage of disease (FIGO III/IV). The patients' age ranged from 37 to 83 years, but 60% of the patients were aged 50-70 years.

Comparative analysis of report completeness against the CAP checklist standard between the synoptic and traditional forms showed differences regarding adequate information between both forms. Examples of synoptic and traditional reports are presented in figures 2 and 3. Traditional cancer reports were usually long with descriptive text paragraphs and with many numeric or alphabetic references; some of them were very brief without a final conclusion (surgical staging appeared rarely on the request for histopathological evaluation, too). Many reports lacked or presented a scanty final conclusive diagnosis (confined to pTNM staging). In the traditional reports, lack of pTNM staging was noticed in 80% of partial and/or final reports; morphological ICD-O coding was missing in 83% of case reports. In the group of 13 borderline tumors, the pTNM stage was mentioned in only 2 tumors staged higher than pT0 (pT1c and pT2c).

Using the new criteria of the recent WHO histological classification, 3 of 29 carcinomas were recategorized from ‘mixed type' in the traditional reports to HGSC (2 cases) and CCC (1 case) in the synoptic standardized reports. One borderline tumor was rediagnosed as endometrioid instead of serous based on additional immunostaining. In 1 serous borderline tumor, a new subcategory was added as micropapillary variant of serous-type tumors.

The recent WHO histological classification recommends three-tiered histological grading (G1-G3) for the four types of epithelial ovarian tumors except for the serous type of carcinoma. Instead of G grading, the serous ovarian carcinoma was classified into two types: low- (LGSC) and high-grade (HGSC). The distinction between both subtypes was based on tumor morphology (cytological not architectural features) [16]. In the series of 29 ovarian carcinomas analyzed, histological grading was reported in 90% of the cases. For the 18 serous carcinomas, a three-tiered system was used in the traditional reports (G1-G3) and a two-tiered system (LGSC and HGSC) for the synoptic structured reporting. Almost all serous carcinomas that have previously been classified as moderately (G2) or poorly differentiated (G3) were reclassified as HGSCs. Two serous carcinomas previously classified as G2 were categorized as LGSCs and 1 G1 serous carcinoma was reported as HGSC on the basis of the high nuclear grade present.

In addition, other relevant elements missing in the descriptive text reports were: lack of data on carcinomatous infiltration of the ovarian capsule (about 20% of reports) and shortage of information regarding angioinvasion (34% reports). The size of the primary tumor was correctly determined in all case reports. The underestimated elements within the pathological reports were implants (invasive vs. noninvasive) concerning the serous carcinoma and the serous borderline tumors; only two traditional reports contained sufficient descriptions based on proper sampling.

With respect to proper histological typing of ovarian tumors, discrepancies between traditional and synoptic reports were seen in 3 cases, and in 2 of the 3 cases they were related to the new criteria for histological typing of ovarian epithelial tumors. The recent histological classification is based on new evidence showing that epithelial ovarian carcinomas arise from distinct (often nonovarian) precursor lesions (peritoneal HGSC, for example, arises from serous tubal precursor lesions), and ovarian EC and CCC are associated with endometriosis. Carcinomas diagnosed in the past as poorly differentiated, high-grade endometrioid or mixed cancers had p53 mutations and are best classified as HGSCs.

Recognition of unique types of epithelial ovarian carcinomas has an impact on clinical trials. Failure to achieve an accurate diagnosis can decrease treatment outcome or expose the patients to unnecessary toxicity. The present five epithelial ovarian carcinoma cell types (HGSC, EC with or without CCC component, pure CCC, MC and LGSC) differ regarding risk factors, precursor lesions, patterns of spreading, underlying molecular abnormalities and response to chemotherapy, and are considered as separate disease entities.

The discrepancies noticed in the grading (G) system between the traditional and synoptic standardized reports for serous carcinomas were associated with the new clear criteria for the two-tiered division (HGSCs and LGSCs) instead of the old three-tiered grading. The stratification of the histological subtypes and grades of tumors (not only ovarian tumors) is very important to significantly improve treatment outcome in clinical trials of new agents for the defined patient subgroups.

Apart from the exact recognition of the histological type and grade of carcinoma, the proper TNM staging of the cancer disease is the third most important element in the cancer pathology report. The contribution of the pathologist in the staging classification appears as a major factor improving clinical stratification of the patients to their optimal chemotherapy and molecular-targeted therapy. Underestimation or lack of essential information within the cancer pathology reports might mislead the clinicians in the selection of the adequate treatment regimen for the patient. The general approach to ovarian cancer treatment (usually given after surgery) includes: treatment by stage; treatment of patients with implants; consolidation therapy and treatment of recurrent disease. One possible reason for the incomplete data in TNM staging noticed in so many traditional pathology reports could be the system of partial reporting (in time series) related to the multistep staging procedure, which might decrease the pathologist's vigilance to the final TNM conclusion in every separate report.

All elements required for the standardized pathology cancer template are significant even though some of them seem to be ‘minute' or less important. Those desirable features characterizing ovarian cancer in our study were referred to: lymphovascular invasion of carcinomatous cells, ovarian capsule rupture type or carcinomatous invasion and peritoneal tumor dissemination (determined by adequate sampling of noninvasive or invasive implants in the borderline tumors or serous-type carcinomas). Although there are controversial opinions to what extent some of those features can affect the distinct prognosis of the patients with early-stage ovarian carcinoma, those features were emphasized by various studies and were included in the latest TNM staging and FIGO classification. Ovarian carcinomas, for example, spread both intraperitoneally and retroperitoneally, and, therefore, tumor spreading mainly through lymphatic vessels without intraperitoneal dissemination suggests a more favorable biological tumor behavior. Considering the relatively favorable prognosis associated with lymphatic tumor spread compared with peritoneal tumor spread, stage IIIC was substaged to IIIA2 in the latest FIGO classification [17].

The problems of interpretation errors and missing data in cancer pathology reports have been analyzed and presented in several international studies [1,2,18,19]. Reasons responsible for the lack of complete reporting could be various levels of knowledge of the reporting pathologists, e.g. insufficient familiarity with the recommended guidelines of reporting (or omission of elements), and experience. These defects could be avoided or corrected by the strategy of the e-synoptic standardized protocol for cancer reports. As it was shown in our study, the e-module for synoptic pathology reports was easy to develop on the basis of the CAP checklist and include into the HIS. Navigation through the drop-down lists of the defined elements renders the diagnostic work easier by providing the required data and automatic arrangement of clear structured reports (saving the pathologists from mistakes). The synoptic templates eliminate the problem of missing TNM and morphological coding by the automatic TNM calculation and automatic extraction of the ICD-O number. The e-form of the synoptic reports allowed for electronic recording, storing and direct processing of the data to the recipients as well as to secondary users.

The electronic structured synoptic reports offer the clinicians a lot of important information to choose the optimal option for cancer therapy. The electronic synoptic record is very efficient in collecting important data, eliminates the costs of the transcription service, improves consistency in documentation and enables open access to other health care providers as well as to central cancer registries and oncologists [20].

In North America as well as in Europe, there is a spectrum of pathology reporting practices in use today denoting differing levels of detail, structure and encoding capabilities [21,22,23]. According to Srigley et al. [22], six types (levels) of cancer pathology reports could be recognized in regard to the tools used and description contained. Level 1 corresponds to the traditional narrative description; level 2 is also narrative but with CAP content; level 2+ is a synoptic-like structured format; level 3+ has an electronic reporting tool using drop-down menus; level 4+ uses standardized reporting language and data elements are stored in discrete data fields; level 5+ is sophisticated synoptic reporting with automated ICD-O and SNOMED CT encoding with standardized reporting language; data elements are stored in discrete fields, using drop-down menus. In practice, most cancer pathology reports are somewhere in the middle.

In Poland, there is a mixture of pathology reports in use; in the majority, the pathology laboratories release the narrative traditional forms and/or synoptic-like forms (narrative based on the CAP cancer checklists). Recently, full cancer checklists of CAP have been adapted and approved by the Polish Society of Pathologists; however, there is high discretion in the implementation of CAP checklists among the Polish laboratories. Our present evaluation of the synoptic structured reporting in cancer pathology is a pioneering Polish trial.

At present, one of the best developed structured reporting domains is cancer diagnostic reporting in pathology. There are many checklists (templates) developed and adapted by national associations of pathologists, published online and updated [12,21,23,24]. The most commonly used fully encoded synoptic reports are the CAP electronic cancer checklists (eCC), a fully machine-readable format of the CAP cancer checklists, designed to make the implementation of capture and reporting of fully coded information easier. New and improved versions of the CAP cancer checklists and the eCC are being released on a periodic basis by the CAP eCC Team, with oversight by the CAP Cancer Committee and CAP's Pathology Electronic Reporting Committee [3]. Some effort has been undertaken to develop a uniform international evidence-based reporting data set for each cancer site by the International Collaboration on Cancer Reporting (which includes The Royal College of Pathologists of Australasia, The Royal College of Pathologists of the United Kingdom, CAP, the Canadian Partnership against Cancer and the European Society of Pathology). The first result of this cooperation was a cancer data set for the reporting of primary ovarian, Fallopian tube and peritoneal carcinoma (the ‘required' and ‘recommended' elements with an explanatory commentary) [25].

Nowadays, the diagnosis and treatment of cancer become more complicated and depend to a great extent on pathological examination. Synoptic reports offer significantly more complete data, are better readable and preferred by physicians. At our Institute, the recipients of pathology reports (surgeons and oncologists) were very satisfied by the format of synoptic ovarian cancer reporting [pers. commun.].

Synoptic reporting has become the accepted practice in most pathology departments and might have various data consumers within and beyond the hospitals, starting from the oncologist and surgeon, clinicians and researchers, national cancer registries, medical organizations related to the guidelines and standards, jurisdictions and insurance companies. Standardized pathology processes are also important in the framework of clinical trials. The clinical trial based on pathology sections with clear guidelines present more reliable results, reduced variability and less errors [1]. However, we should bear in mind that use of a checklist by itself does not guarantee an adequate report, since about 12% of reports in the study by Idowu et al. [19] and up to 165 of the reports in the study by Srigley et al. [22] did not have all the mandated elements despite the routine use of synoptic reports/checklists. The checklists are useful in ensuring adequate reports only if the required elements in the checklists are thoroughly completed and the reports are carefully reviewed before verification.

Besides completeness, the quality of pathology reports also depends on the timeliness and accuracy of the diagnosis. A central review of specimens is a very useful tool to test diagnostic and prognostic accuracy. At our Institute, the pathomorphological e-data are available in the daily diagnostic process with access to all previous reports via HIS. The beneficial result of our study was that synoptic reports were supplemented with digital images of the respective ovarian cancers. The imaging data (macro- and microscopic images) visualize and support the final histopathological diagnosis. The WSI database was programmed for diagnostic purposes (clinicopathological meetings, remote consultations and revisions), research, archiving and education on ovarian cancer. The set of 42 WSIs was (temporarily) stored on the pathology laboratory server for further investigations until being transferred to the Case Center platform.

Since the scanned microscopic slides and WSIs can be automatically digitized and integrated in the clinical workflow, quality assessment of WSIs has become a crucial issue [10,26,27,28]. Standardizing the use of WSIs in diagnostic implementation of WSI platforms for consultation is not an easy task. Many pathology laboratories have questioned if digital pathology should be considered for large-scale implementation in routine clinical practice. The number of institutions that adopt digital pathology is limited because of the high costs and the complexity of digital pathology equipment [11,20,29].

Pathology informatics (informatics for practicing pathomorphology) was created as a response to the overwhelming amount of data and information that are available in patient service. We are moving into a new area of extensive application of pathology informatics. The most commonly used tools of informatics are classified into digital imaging and virtual microscopy, telepathology, the Internet and electronic data mining. The development of digital pathology highly depends on the informatics technology [12]. The complex interorganizational relationship between pathology laboratories, regional providers of information and communication technology services, regional health authorities and vendors of information and communication technology systems makes it crucial to establish a clear organizational instrument for the future development, implementation and maintenance of electronic templates for histopathological cancer reporting [21].

Our results confirm previously published data that synoptic pathology reporting improved the quality of ovarian cancer reports compared with traditional narrative reports. The content of the synoptic cancer reports was improved regarding the essential information for the patient management in comparison to the descriptive unstructured reports; was clearly arranged, conclusive and better readable, and satisfactory for surgeons and oncologists. The synoptic standardized reports generated by the electronic pathology system using a standardized structure with drop-down lists of defined elements ensured completeness and consistency of reporting with the required guidelines, which eliminates the problem of missing pathological staging and morphological coding by the automatic TNM calculation and automatic extraction of the ICD-O number, for example. Standardized reporting and electronic tools facilitate easy, open access and downstream utilization of pathology data for clinicians and other health care providers, and data can be entered in tumor registries. The manual SNOMED coding based on standardized vocabularies has poor reproducibility and should be replaced by automated encoding in cancer pathology reports. The improved quality of reporting, developmental implementation and maintenance of e-templates as well as digitization and remote diagnostics require further complex collaboration of pathology laboratories, IT service providers and clinicians in structured cancer pathology reporting.

The study was supported by the Ministry of Science and Higher Education through statutory grants (project No. 279).