Tutorial Series on the Instruments of the Merkel Cell Carcinoma Patient Registry: Presentation and Initial Staging

Objectives for Tutorial Series

• The Merkel Cell Carcinoma Patient Registry (MCC Registry) is an ongoing effort to develop a multi-institution database of real world data (RWD) on clinicopathologic features of disease, genomics, management decisions, and disease responses of patients with Merkel Cell Carcinoma for hypothesis generation and testing.
• The database is being developed and optimized on the REDCap (research electronic data capture) platform created by Vanderbilt University in 2004 and currently in use by thousands of research groups worldwide. 1,2
• For additional details on the process of optimizing the user experience and interface (UX/UI) of data entry forms on REDCap, please refer to our series of tutorial posts on Optimizing Real World Data Collection.
• In this series of tutorials on the Instruments of the MCC Registry, we will discuss the development and deployment of select data entry forms from the registry.
• Lessons gleaned from our experiences in development are transferrable to similar efforts to create and deploy databases for other tumors.

Presentation and Initial Staging Instrument: Tutorial Overview

• In this post we discuss the developmemt of the “Presentation and Initial Staging” instrument of the MCC registry, including the general concepts behind its deployment and helpful tips on REDCap code.
• After this lesson, you will:
• Understand how the “Presentation and Initial Staging” instrument was constructed.
• Understand how our approach might be adapted to create staging instruments for other cancers.
• Skill Level: Intermediate

Challenges of Capturing Real World Staging Data

• Cancer stage is used in prognostication of patient courses and is an important data point in the capture of real world evidence on rare tumors. Unaided, collection of disease stage is a challenging endeavor for the Data Abstractor, even with disease specific training

1. Because of the diversity and complexity of disease presentation and subsequent diagnostic and therapeutic decision-making, the American Joint Committe on Cancer (AJCC) staging system for MCC is fairly sophisticated.
2. The Data Abstractor often cannot directly rely on provider assigned stage from clinician notes.
• Clinician assigned disease stage may be unreliable due to provider entry error, or incomplete availablity of staging information at the time of documentation.
• Moreover, staging systems are updated over time, and it may not be immediately obvious to the Data Abstractor which edition of staging criteria is in use. For example, the introduction of AJCC 8th edition criteria on MCC staging in 2018 introduced a distinct tumor category for Merkel Cell Carcinoma presenting without a primary cutaneous lesion (T0, for MCC of unknown primary).
• Rare tumor registries, therefore, must comprehensively collect the component data used to arrive at clinical and pathological staging decisions, allowing for generation of consistent staging classifications on the back-end by the Data Scientist.

• There is still utility in having staging classification of records upfront within a registry. For example you might want to quickly view all patients with AJCC 8th Edition Pathological Stage IIIB disease within a dashboard in REDCap.

• Therefore, it is also beneficial to incorporate front-end collection of staging assignments within the data entry form.

The task, then, is to create a staging form that guides the Data Abstractor in assigning stage consistently and accurately.

Clinical and Pathological Staging of MCC according to AJCC Eigth Edition Staging Criteria

• The Eighth Edition of the American Joint Committee on Cancer (AJCC) staging manual was published in 2018, and provides standardized criteria for the staging of Merkel Cell Carcinoma.1
• Clinical staging is based on clinical and radiologic information acquired from work-up prior to definitive surgical management of the primary cutaneous lesion and/or metastatic disease, versus pathological staging, which incorporates histopathological data obtained from definitive surgical therapy.

TNM classification of MCC determines clinical and pathological stage.

• Both clinical and pathological staging rely on T (primary tumor), N (nodal), and M (metastasis) categories to generate an overall stage.

• Whereas clinical staging uses clinical nodal and metastasis (cN and cM, respectively) categories, pathological staging relies on pathological nodal category (pN) and, if distant metastasis is microscopically confirmed, a pathological metastasis category (pM).

• Tumor (T) category is based on clinical/radiologic evaluation of the primary tumor as well as histopathologic microstaging. T category is used both for clinical and pathological staging.

• Definitions of primary tumor category according to AJCC 8th edition criteria are given below. Comprehensive details on clinical and pathological staging may be found in the AJCC Staging Manual, 8th edition.

T (Primary Tumor) Category AJCC 8th Edition Definition
TX Primary tumor was not able to be assessed (ex. was destroyed by prior procedure)
T0 No primary tumor found on clinical evaluation
Tis Primary tumor confined to the epidermis (i.e. an in situ tumor)
T1 Primary tumor clinically sized <= 2 cm
T2 Primary tumor clinically sized >2 cm, but <= 5 cm
T3 Primary tumor clinically sized > 5 cm
T4 Primary tumor shows extracutaneous invasion of deeper structures (fascia, muscle, cartilage, bone)

Structure and Function of the Presentation and Initial Staging Instrument

Our design of the staging instrument emphasizes clustered collection of all relevant component data for determining T, N, and M categories.

• We present fields for collection of data on the primary tumor (if applicable) and potential regional, in-transit, and distant metastases in 3 clusters, according to the typical flow of MCC diagnosis, work-up, and surgical management:
1. Physical Exam
3. Pathological Evaluation
• Data from all three sources of evaluation are used in in-form calculations to determine the T, cN, and cM category, and, if applicable, the pN and pM category. The calculated T, cN, pN, cM, and pM categories in turn feed into calculations for overall clinical and pathological staging.
• The Data Abstractor is prompted to confirm these calculations. If the calculated categories and staging are incorrect according to the Data Abstractor’s understanding, this should prompt them to review the data they provided for outcomes of clinical and pathological evaluations.
Above, colored squares represent data points that the Data Abstractor enters into the registry, or confirms via checkbox. The colored arrows represent calculations that occur on the REDCap form, but are hidden to the Abstractor.

Calculating TNM Categories and Overall Staging

• All calculations are carried out via the if() function, with a numeric output corresponding to TNM category or stage.
• We outline below how a calculation assigns the T0 category based on entered clinical data.
• The Data Abstractor indicates that no primary cutaneous lesion was found via the pres_prim field.

• A separate calculated field for tumor stage is named t_stg. Below is the appearance of a typical calculated field within the Online Designer.

• The calculation for the field starts as follows:

if([pres_prim] = "0", 2, if (... where 2 is the given numeric output associated with the T0 category.

• If disease indeed presented as a primary cutaneous lesion, the calculation would proceed to a series of nested conditional statements (beginning with if(...) that survey data entered into fields on primary tumor features obtained from physical exam, imaging, and pathological evaluation, with numeric outputs corresponding to the remaining tumor categories.

• Similar calculated fields are created for cN, cM, pN, and pM categories. All these calculations can be hidden from the Data Abstractor through the use of the @HIDDEN action tag, and are only visible in the Online Designer.
• Separate fields for clinical and pathological staging calculations will then use the numeric outputs associated with TNM category calculations to calculate overall stage.

Confirming TNM Category and Overall Stage Assignment

• Note the calculations above are not retained during data exports!

• However, as discussed, we can deploy these calculations for real-time quality control of entered staging data by branching confirmatory fields on the outputs of the TNM category and overall clinical and pathological stage calculations.

• These confirmatory fields are a series of checkbox fields each associated with an invidual TNM category or clinical/pathological stage.

• When a particular category or clinical/pathological stage is checked off as confirmed, the associated field variable takes on a value of 1.

• For example, for a T1cN1pN1bcM0 tumor with TNM categories and overall staging fully confirmed by the Data Abstractor, the t1, cn1, pn1b, and cm0 variables would equal 1, along with the clinical_stage_iii and path_stage_iiib variables.

• All other category and stage variables (ex. t4, cN3, pN2, clinical_stage_IV, etc.) would remain valueless.
• In contrast to the numeric outputs of the calculated fields, data from these confirmatory fields (manually entered as checkboxes) is exported, just as any other data entry field.

Filtering Records by TNM Category and/or Overall Stage

• The confirmatory category and stage fields allow for easy filtering of records by any combination of TNM categories and clinical/pathological stage using the REDCap Custom Dashboard feature.

Date Rendered No Evidence of Disease

• At the conclusion of the instrument, we also include a field for entry of Date Rendered “No Evidence of Disease” (NED) if all clinically apparent disease was removed post initial treatment strategy (typically surgery, or surgery plus adjuvant radiotherapy) to aid in calculations of time to relapse/recurrence elsewhere in the registry.

Takeaways

• Real world data collection on oncologic staging is a complicated undertaking. Direct abstraction of staging from clinical documentation is limited by shifting staging criteria, limitations in data availability during early stages of presentation, and clinician error.
• Consequently, it is important for a rare tumor registry to collect component data used for staging. This can then be used on the back-end by the Data Scientist to assign staging consistently, according to uniform criteria.
• Front-end assignment of stage is nonetheless important for registry maintenance and accessibility, for example through utilization of REDCap’s dashboard feature.
• The deployment of a guided approach, through which calculated stage based on component data entry is presented to the Data Abstractor for confirmation, minimizes inconsistency while generating reliable staging assignments upfront.

References:

Stay tuned for the next in our series of tutorial posts on the Instruments of the Merkel Cell Carcinoma Patient Registry: Adverse Events

1. Amin MB, Edge S, Greene F, Byrd DR, Brookland RK, Washington MK, Gershenwald JE, Compton CC, Hess KR, et al. (Eds.). AJCC Cancer Staging Manual (8th edition). Springer International Publishing: American Joint Commission on Cancer; 2017↩︎

Farees Saqlain
Medical Student

My research interests include clinical and translational research in advanced skin cancers.