Patients Will Benefit if We Expand Cancer Registries to Capture Method of Detection

Introduction

Controversies related to screening for breast cancer persist in the United States despite cutting-edge tools for diagnosis and treatment. Patients continue to receive conflicting recommendations from providers and specialty societies who rely on models and historic clinical trial data that may no longer accurately represent the diversity of the screening-eligible population or advances in screening technologies. One challenge to conducting population-based breast cancer research is the lack of a direct link between cancer outcomes and the initial method of detection (MOD) for each registrant. Inclusion of this valuable data element would facilitate assessment of linkage between screening and cancer stage, treatment received, patient outcomes, sociodemographics, geography, access to health care, and molecular signatures, for example.

The results of randomized controlled trials (RCTs) conducted between 1963 and 1990 from multiple international sources provide robust evidence that screening mammography significantly reduces breast cancer mortality.¹ The American College of Radiology (ACR), US Preventive Services Task Force (USPSTF), and American Cancer Society (ACS) agree that annual screening mammography beginning at age 40 years will save the most lives.^2-5 However, the same organizations disagree over the balance of risks and benefits of screening mammography and recommend different frequencies and ages to initiate early detection of breast cancer in the United States.¹ In the decades since the RCTs concluded, the technology, health care systems, and racial diversity in the United States have drastically changed. The ACR, USPSTF, ACS, and other organizations advocate for new research of technologic efficacy inclusive of Black, Hispanic, Latina, Asian, Pacific Islander, Native American, and Alaska Native women in the United States but lack the national population data to do so.

Shortly after the data from RCTs confirmed the benefits of early detection, many nations instituted population-wide breast cancer screening programs.⁶ Administrators of those programs had the foresight to include the initial MOD, such as mammography screening or clinical examination, in cancer registries for every patient with a new diagnosis of breast cancer. Using these data, which were acquired annually for decades, administrators and physicians can actively review contemporary patient-specific links between MOD and outcomes to understand and continuously adapt breast cancer care to the local and evolving populations in those countries.⁶ The United States, lacking a centralized screening program or an ability to link MOD to individual patients, cannot. Our understanding of the impact of screening on minority and vulnerable and underinsured populations in screening remains low while barriers to participation remain high.

Knowledge Gaps

The Surveillance, Epidemiology, and End Results (SEER) program, the National Program of Cancer Registries (NPCR), the National Cancer Database (NCDB), and the ACR's National Mammography Database collect specific data for every patient with a new diagnosis of breast cancer, but MOD has never been included. The North American Association of Central Cancer Registries (NAACCR) maintains the data dictionary for all state and regional registries. Neither NAACCR nor NPCR nor NCDB nor SEER require registries to assign or collect MOD.

Thus, among nations with high rates of breast cancer and state-of-the-art screening facilities and cutting-edge technology, the United States lacks the fundamental ability to directly link breast cancer outcomes to MOD and scientifically address ongoing controversies over screening. This is compounded by continuously evolving population demographics of the United States. National organizations such as the USPSTF, ACS, and the American College of Physicians must rely on observational data from the Breast Cancer Surveillance Consortium, SEER, and other sources and simulation models that employ nonuniform assumptions. These assumptions may be subject to bias when examining the impact of screening.^2,5,7,8 The lack of a consistent, patient-specific link between MOD and outcomes has permitted ongoing speculation and fostered disagreement about the risks and benefits of screening in the United States.⁹ Ultimately, the conflicting recommendations confuse patients and providers and disrupt clear and critical opportunities to save lives.

Defining and Determining MOD

The initial MOD of breast cancer is defined as the first test or clinical event to trigger the work-up leading to the histologic diagnosis of breast cancer. When national service-screening programs and registries were built in the 1980s and 1990s, the choices for initial MOD were limited. Originally, film-screen mammography was the only image-based test. Today, initial MOD can include multiple other image-based screening modalities (Table 1). Screening with 2-dimensional digital mammography, digital breast tomosynthesis (DBT), ultrasound, magnetic resonance imaging (MRI), and other tests can now provide the earliest evidence of breast cancer. Self-examination and clinical breast examination, which detect lumps, thickening, or tenderness, can also be the initial MOD leading to a diagnosis. Patients may also trigger detection of breast cancer when they seek care for nipple discharge, erythema, pain, dimpling, or skin ulceration. In addition, other imaging or laboratory tests not designed to evaluate the breast, such as abdominal computed tomography or brain MRI, may detect metastases that lead to a diagnosis of breast cancer.

Table 1.

The Proposed Categories of Initial MOD Designed to Capture Information Relevant to Contemporary Technology and the Impact of Self Breast Examination and Clinical Breast Examination

Category S: Initially detected with image-based screening.

Sdbt: Screening DBT, with synthetic or full field 2D

Sma: Screening 2D mammogram without any DBT

Sus: Screening ultrasound (automated or handheld)

Smri: Screening MRI

Scem: Screening contrast enhanced mammogram

Snuc: Screening nuclear medicine breast examination

So: Other screening modality such as screening breast CT, etc

Category P: initially detected by patient or provider.

Pat: Patient first detected by self-examination or symptom such as pain

Pro: Provider first detected by clinical examination

Ppp: Not possible to determine if patient or provider detected

Category N: Not detected by patient or provider or with image-based screening. Example: liver metastasis detected by abdominal ultrasound prompted by abdominal pain and abnormal liver function tests.

DBT, digital breast tomosynthesis; MRI, magnetic resonance imaging.

Benefits of Collecting MOD

If MOD could be assigned and collected accurately and without bias for each patient, providers from multiple specialties could access new primary data that capture the diversity of our screening-eligible population and advances in screening technologies. Concrete, patient-specific data could bring the ACR, USPSTF, and ACS to consensus recommendations for screening. MOD-inclusive data can contribute to collaborative, multispecialty assessment of efficacy, equity, treatment, and outcomes for breast cancer, such as:

1. What are the relative contributions of screening mammography and treatment to reducing breast cancer mortality?

2. Are the outcomes for patients with stage 1 breast cancers different for different initial methods of detection?

3. Do patients with image-detected cancers have different treatment or mortality outcomes compared to patients with clinically detected cancers?

4. Do image-detected tumors have different molecular signatures compared to clinically detected tumors?

5. What are the relationships between race, demographics, social determinants of health and MOD and outcomes?

6. What percentage of breast cancers are not initially screen-detected, and how does this vary by personal risk, breast density, age, or other factors?

7. Are there differences in initial staging for breast cancers initially detected with image-based screening vs clinical or self-examination?

8. Which specific geographic locations have below average outcomes and below average fractions of image-detected cancer, and can we use that information to improve access to screening at the local level?

9. Are supplemental image-based screening modalities (MRI, ultrasound, etc) providing the same reductions in morbidity and mortality as screening mammography?

Barriers To National Collection of MOD

The US health care system provides cutting-edge care with comparatively brief wait times and less regard to societal cost. However, data collection is a patchwork of public and private entities funded by numerous private and public payors competing at the local and regional levels. This data is stitched together with different electronic medical records of heterogeneous patient populations. Every state has a tumor registry responsible for tracking valuable information such as incidence, stage, race, and mortality for every case of cancer diagnosed. However, there is no state or nationally standardized process for collecting MOD.

Assigning MOD Accurately

Abstractors employed by state, local and hospital registries currently gather information related to a new cancer diagnosis from clinical reports. Most of the information regarding cancer type, size, grade, and receptor status is quickly abstracted from succinct and standardized pathology reports. However, abstractors frequently revert to the tedious and time-consuming strategy of sifting through other clinical notes for details regarding the treatment plan, for example. Abstractors could attempt to assign MOD through retrospective review, but if abstractors already know the patient has breast cancer, the assignment of MOD could be skewed by unconscious bias. It is imperative that the assignment of initial MOD be accurate, prospective, unbiased, easily discoverable, and correctly transferred to registries for future scientific investigation.

The Screening and Emerging Technology Committee of the ACR Breast Commission recommends that radiologists prospectively assign the MOD for each patient with suspicious imaging findings at the time of the diagnostic work-up prior to a new diagnosis of breast cancer. We cannot expect abstractors to retrospectively read multiple radiology and pathology reports to recreate the clinical history and accurately determine the MOD. The diagnostic radiologist has the necessary expertise to understand the subtle nuances of the imaging, history, and clinical information and is uniquely suited to accurately assign the initial MOD. Indeed, in the standard course of imaging work-up when recommending tissue sampling to a patient, the diagnostic radiologist will have the most complete understanding of the clinical scenario and an unbiased prospective opportunity to assign a single, initial, highly accurate MOD.

Summary

Controversies related to screening for breast cancer persist in the United States despite cutting-edge tools for diagnosis and treatment. This is because outcomes cannot be directly related to the initial MOD at the state or national levels. Inclusion of MOD in state and regional cancer registries and national databases is long overdue. In addition to requiring registries to capture MOD, methods to increase MOD reliability and abstraction ease, such as mandating inclusion of MOD in radiology reporting, are necessary. Inclusion of MOD for breast cancer registries may serve as a model for image-detected cancers such as lung and colon cancers and their respective registries. Radiologists have an opportunity to directly facilitate the capture of MOD and contribute to a new critical linkage in our national registries to dramatically improve our understanding of breast cancer and screening and reduce the burden of disease on our patients.