The Cancer Synoptic Reporting Project Group followed a template-based, subject matter expert-driven process to develop the content produced in this project. Content developed is based on the content represented in published reporting protocols using the Observable entity concept model. The overall approach and assumptions are described in the following pages.

Historical SNOMED CT content authored for use in cancer synoptic reporting can be found in the Observable entity and Clinical finding hierarchies.

Observable Entities vs Clinical Findings

The vast majority of these concepts are primitive and have effective dates of 2001-01-31, which is the beginning of SNOMED CT time. These concepts were deemed insufficient to unambiguously represent pathology observations and findings for use in cancer registries. An early design decision in the Cancer Synoptic Reporting Project Group project was to use the Observable entity hierarchy instead of the Clinical finding hierarchy. Regardless of approach, substantial concept modeling would be required in either hierarchy. Ultimately, the decision was made to provide a tangible, needed, and practical use case upon which to demonstrate the efficacy of the newly remodeled Observable entity concept model. Apart from the novelty of the approach, the question, or Observable entity, in each synoptic report must carry sufficient context to unambiguously interpret the observation or finding.

A practical consideration for this modelling approach pertained to the amount of new content that would be necessary to create to meet the needs of the cancer synoptic use case.

Content Development Approach

A significant number of concepts have been created as part of this effort, encompassing all hierarchies. In retrospect, the decision to utilize observable entities to encompass the complete clinical context for each observable entity/observation pair proved beneficial in managing the concept volume effectively.

The histologic type of "malignant neoplasm of organ X" is a good example of this.

Every organ system with a reporting protocol has an average of 10-20 morphologic abnormalities that could be recorded. Many of these morphologies may be observed in multiple organ systems. If clinical findings were used to represent the protocol data, a new concept would be necessary for every organ/morphology pair for example:

• adenocarcinoma of organ X

• mucinous adenocarcinoma of organ X

• serrated carcinoma of organ X

Using observable entities, however, required only a single concept to be created for each organ system The observable entity (e.g., Histologic type of malignant neoplasm of organ X) could be paired with any number of valid morphologies. Ultimately, the decision required less new content to be developed and maintained.

It should be noted that the observable entity/observation pairs in the Cancer Synoptic Reporting Project Group product do NOT follow the clinical finding MRCM specific that uses the defining attributes = << AND = << . The types of information solicited in the cancer pathology protocols extend beyond the content represented in the Qualifier value hierarchy and include values from the Procedure hierarchy, Body structure hierarchy, and concepts in the qualifier hierarchy NOT subsumed by << and discrete numerical values. Current MRCM rules for clinical findings do not include these concept areas in the range of possible concepts for . Pathology synoptic data elements simply record a series of individual observations in a structured fashion.

The Cancer Synoptic Reporting Project Group operated under the data modeling paradigm that the |Observable entity (observable entity)| concept provides the context to correctly and unambiguously interpret the observation. Therefore, understanding the observable entity concept model is fundamental to use of the cancer synoptic content. Authored content for observable entities of this project are found <<1145211006 |Proliferative mass observable (observable entity)|

The table below provides an overview of the defining attributes used to author cancer synoptic reporting concepts, including a description of the target values for these attributes.

• << ;

• << ;

• < ;

| This attribute is used to assert the entity that carries the property being measured. In most cases, the target values are << | The inherent location attribute is used to describe the anatomical location of the entity that carries the property being assessed. In most cases, the inherent location indicates the anatomical location of the primary malignant neoplasm, that is the primary organ affected by the malignancy. | This attribute is used to indicate an entity that is being assessed for presence such as necrosis within a neoplasm. It is also used to represent the numerator in a percent observation. | This attributed is used for the denominator in a percent or number fraction observable. | Direct site is specifically used to define the specimen in which the observation is being made. | Technique is used to define the method by which the observation is being made. This attribute is used to specific methods of tumor staging, histologic grading methods, direct vision (gross) evaluation, microscopy and immunohistochemistry methods. | The time aspect for all cancer pathology observable entities is

| This attribute is used to describe the evaluation scale used for the observation.

• • is used to describe observable entities assessing morphologies, body structures, and procedures.

| Characterizes is used to represent the underlying processes of the neoplasm. These include << and | This attribute is used to define the "end point" of the process indicated by the characterized attribute/value pair. In most concepts, this associated value of this attribute is << to indicate where the neoplasm has grown or metastasized.

Knowledge Representation Model

The underlying principle of the Cancer Synoptic Reporting Project Group is that the "question" (represented by a SNOMED CT observable entity) should include all the context needed to clearly understand the "answer" (or observation) that is recorded. The criteria for deciding these results depend on additional information and knowledge that can't be conveyed by SNOMED CT alone, such as clinical guidelines.

For example, the differentiation between an adenocarcinoma and a mucinous adenocarcinoma is based on the amount of mucin measured in the cells and the organ system involved. In breast tissue, the amount of mucin in the cells to be considered a mucinous adenocarcinoma is > 80% but in the colon is > 50%. The pathologist is responsible for this knowledge, not SNOMED CT. So, this type of explicit knowledge of the pathologist is outside of the scope of the SNOMED CT concepts and will not be discussed further in this document.

The observable entity/observation pairs below each state that the histologic type of the neoplasm assessed is a mucinous adenocarcinoma in the breast and in the colon. It is the SNOMED CT observable entity that provides the context for the observation, specifically the organ system of concern. It does not directly state the amount of mucin in the cells as observed by the pathologist. The pathologist exercised domain-specific knowledge to reach such a conclusion.

Mucinous adenocarcinoma of the breast: = reflects the pathologist's interpretation of the microscopically evaluated slides where the percent tumor cells containing mucin as a proportion of the total number of tumor cells is > 80%.

Mucinous adenocarcinoma of the colon: = __ reflects the pathologist's interpretation of the microscopically evaluated slides that the percent tumor cells containing mucin as a proportion of the total number of tumor cells is > 50%.

Therefore, it is important to understand that the observable entity/observation pairs used throughout the cancer pathology synoptic reporting use cases reflect point-in-time observations as ultimately assessed and interpreted by the pathologist. Domain knowledge specific to the practice of pathology and oncology is NOT intended to be represented by the terminology, but rather, the terminology represents what the observation was and is based on specific domain expertise.

As noted, the synoptic pathology report is comprised of a list of characteristics of the neoplasm that are required to be observed and reported by the pathologist. Each characteristic is modelled using the** 363787002 |Observable entity (observable entity)|** hierarchy and concept model.

Major categories of neoplasm characteristics required in each report are listed below.

• Procedure used to collect the specimen(s)

• Tumor site

• Tumor dimensions

• Histologic type

The list of possible observations that can be made for each characteristic is comprised of a constrained list of acceptable observations (i.e., value sets). For example, a list of acceptable histologic types (morphologic abnormalities) is provided to the pathologist to select when reporting the histologic type of the neoplasm. Semantic types of these value sets are dependent upon the |Property| target value.

Semantic types in a response value set are all of the same semantic type, with the exception of the use of |Qualifier value| concepts employed for pathologist observations, such as 385432009 |Not applicable (qualifier value)| or 1156316003 |Cannot be determined (qualifier value)| .

For example, the concept, , is modeled with = . The range of possible observations must be << .

In addition to the template constraints, the protocol publishers further constrain the acceptable value sets to include only those values that are possible for a particular malignant neoplasm. For example, the possible values for histologic types in the lung protocol would never contain values for Germ cell neoplasms (a condition only possible in reproductive organs).