Essentialism and Exclusion: Racism in Cancer Risk Prediction Models

Can evidence drive health equity in the COVID-19 pandemic and beyond?

Using scoping review methods, we systematically searched multiple online databases for publications in the first year of the pandemic that proposed pragmatic population or health system-level solutions to health inequities. We found 77 publications with proposed solutions to pandemic-related health inequities. Most were commentaries, letters, or editorials from the USA, offering untested solutions, and no robust evidence on effectiveness. Some of the proposed solutions could unintentionally exacerbate health inequities. We call on health policymakers to co-create, co-design, and co-produce equity-focussed, evidence-based interventions with communities, focussing on those most at risk to protect the population as a whole. Epidemiologists collaborating with people from other relevant disciplines may provide methodological expertise for these processes. As epidemiologists, we must interrogate our own methods to avoid propagating any unscientific biases we may hold. Epidemiology must be used to address, and never exacerbate, health inequities-in the pandemic and beyond.

Artificial Intelligence-Driven Mammography-Based Future Breast Cancer Risk Prediction: A Systematic Review

PURPOSE

To summarize the literature regarding the performance of mammography-image based artificial intelligence (AI) algorithms, with and without additional clinical data, for future breast cancer risk prediction.

MATERIALS AND METHODS

A systematic literature review was performed using six databases (medRixiv, bioRxiv, Embase, Engineer Village, IEEE Xplore, and PubMed) from 2012 through September 30, 2022. Studies were included if they used real-world screening mammography examinations to validate AI algorithms for future risk prediction based on images alone or in combination with clinical risk factors. The quality of studies was assessed, and predictive accuracy was recorded as the area under the receiver operating characteristic curve (AUC).

RESULTS

Sixteen studies met inclusion and exclusion criteria, of which 14 studies provided AUC values. The median AUC performance of AI image-only models was 0.72 (range 0.62-0.90) compared with 0.61 for breast density or clinical risk factor-based tools (range 0.54-0.69). Of the seven studies that compared AI image-only performance directly to combined image + clinical risk factor performance, six demonstrated no significant improvement, and one study demonstrated increased improvement.

CONCLUSIONS

Early efforts for predicting future breast cancer risk based on mammography images alone demonstrate comparable or better accuracy to traditional risk tools with little or no improvement when adding clinical risk factor data. Transitioning from clinical risk factor-based to AI image-based risk models may lead to more accurate, personalized risk-based screening approaches.

Climate change and cancer: the Environmental Justice perspective

Despite advances in cancer control-prevention, screening, diagnosis, treatment, and survivorship-racial disparities in cancer incidence and survival persist and, in some cases, are widening in the United States. Since 2020, there's been growing recognition of the role of structural racism, including structurally racist policies and practices, as the main factor contributing to historical and contemporary disparities. Structurally racist policies and practices have been present since the genesis of the United States and are also at the root of environmental injustices, which result in disproportionately high exposure to environmental hazards among communities targeted for marginalization, increased cancer risk, disruptions in access to care, and worsening health outcomes. In addition to widening cancer disparities, environmental injustices enable the development of polluting infrastructure, which contribute to detrimental health outcomes in the entire population, and to climate change, the most pressing public health challenge of our time. In this commentary, we describe the connections between climate change and cancer through an Environmental Justice perspective (defined as the fair treatment and meaningful involvement of people of all racialized groups, nationalities, or income, in all aspects, including development, implementation, and enforcement, of policies and practices that affect the environment and public health), highlighting how the expertise developed in communities targeted for marginalization is crucial for addressing health disparities, tackling climate change, and advancing cancer control efforts for the entire population.

Identifying Frequently Endorsed Benefits and Barriers to Breast Cancer Screening for African-Born Women in the NYC Metropolitan Area: a Pilot Study

Most cancer screening data report on Black participants without distinguishing nativity, limiting our understanding of the needs of distinct groups within the African diaspora. The purpose of this pilot study was to assess demographic characteristics and perceptions of the benefits of and barriers to mammography among African immigrant women in New York City (NYC). Forty-two women who were 40 years or older, born in Africa, and English and/or French-speaking were recruited from African immigrant communities in NYC to complete a survey. Eighty percent of our sample aged 50 to 73 was adherent to the 2016 USPSTF mammography screening guideline. The most frequently endorsed benefits were that mammography will help find breast cancer early, could help find a breast lump before it is big enough to feel, and that if found early, breast cancer could be successfully treated. The most endorsed barriers were that having a mammogram is painful and that lack of insurance or being treated rudely at the mammogram center would keep participants from having a mammogram. Chi-square analyses assessed relationships between demographic characteristics and perceptions about mammography and revealed that endorsement of barriers to screening (e.g., health issues, transportation problems, pain, and time associated with mammography) varied by educational attainment. Findings suggest that future interventions should be multi-level and (1) support patients in accessing screening via resource sharing, (2) address other commonly cited barriers such as fear of pain during the procedure, and (3) support anti-racist healthcare environments especially in terms of treatment by providers.

Approaches to developing de novo cancer population models to examine questions about cancer and race in bladder, gastric, and endometrial cancer and multiple myeloma: the Cancer Intervention and Surveillance Modeling Network incubator program

BACKGROUND

We are developing 10 de novo population-level mathematical models in 4 malignancies (multiple myeloma and bladder, gastric, and uterine cancers). Each of these sites has documented disparities in outcome that are believed to be downstream effects of systemic racism.

METHODS

Ten models are being independently developed as part of the Cancer Intervention and Surveillance Modeling Network incubator program. These models simulate trends in cancer incidence, early diagnosis, treatment, and mortality for the general population and are stratified by racial subgroup. Model inputs are based on large population datasets, clinical trials, and observational studies. Some core parameters are shared, and other parameters are model specific. All models are microsimulation models that use self-reported race to stratify model inputs. They can simulate the distribution of relevant risk factors (eg, smoking, obesity) and insurance status (for multiple myeloma and uterine cancer) in US birth cohorts and population.

DISCUSSION

The models aim to refine approaches in prevention, detection, and management of 4 cancers given uncertainties and constraints. They will help explore whether the observed racial disparities are explainable by inequities, assess the effects of existing and potential cancer prevention and control policies on health equity and disparities, and identify policies that balance efficiency and fairness in decreasing cancer mortality.

A catchment and location-allocation analysis of mammography access in Delaware, US: implications for disparities in geographic access to breast cancer screening

BACKGROUND

Despite a 40% reduction in breast cancer mortality over the last 30 years, not all groups have benefited equally from these gains. A consistent link between later stage of diagnosis and disparities in breast cancer mortality has been observed by race, socioeconomic status, and rurality. Therefore, ensuring equitable geographic access to screening mammography represents an important priority for reducing breast cancer disparities. Access to breast cancer screening was evaluated in Delaware, a state that experiences an elevated burden from breast cancer but is otherwise representative of the US in terms of race and urban-rural characteristics. We first conducted a catchment analysis of mammography facilities. Finding evidence of disparities by race and rurality, we next conducted a location-allocation analysis to identify candidate locations for the establishment of new mammography facilities to optimize equitable access.

METHODS

A catchment analysis using the ArcGIS Pro Service Area analytic tool characterized the geographic distribution of mammography sites and Breast Imaging Centers of Excellence (BICOEs). Poisson regression analyses identified census tract-level correlates of access. Next, the ArcGIS Pro Location-Allocation analytic tool identified candidate locations for the placement of additional mammography sites in Delaware according to several sets of breast cancer screening guidelines.

RESULTS

The catchment analysis showed that for each standard deviation increase in the number of Black women in a census tract, there were 68% (95% CI 38-85%) fewer mammography units and 89% (95% CI 60-98%) fewer BICOEs. The more rural counties in the state accounted for 41% of the population but only 22% of the BICOEs. The results of the location-allocation analysis depended on which set of screening guidelines were adopted, which included increasing mammography sites in communities with a greater proportion of younger Black women and in rural areas.

CONCLUSIONS

The results of this study illustrate how catchment and location-allocation analytic tools can be leveraged to guide the equitable selection of new mammography facility locations as part of a larger strategy to close breast cancer disparities.

Disentangling age, gender, and racial/ethnic disparities in multiple myeloma burden: a modeling study

Multiple myeloma (MM) is a hematological malignancy that is consistently preceded by an asymptomatic condition, monoclonal gammopathy of undetermined significance (MGUS). Disparities by age, gender, and race/ethnicity in both MGUS and MM are well-established. However, it remains unclear whether these disparities can be explained by increased incidence of MGUS and/or accelerated progression from MGUS to MM. Here, we fit a mathematical model to nationally representative data from the United States and showed that the difference in MM incidence can be explained by an increased incidence of MGUS among male and non-Hispanic Black populations. We did not find evidence showing differences in the rate of progression from MGUS to MM by either gender or race/ethnicity. Our results suggest that screening for MGUS among high-risk groups (e.g., non-Hispanic Black men) may hold promise as a strategy to reduce the burden and MM health disparities.

Development and Validation of the VA Lung Cancer Mortality (VALCAN-M) Score for 90-Day Mortality Following Surgical Treatment of Clinical Stage I Lung Cancer

OBJECTIVE

The aim was to develop and validate the Veterans Administration (VA) Lung Cancer Mortality (VALCAN-M) score, a risk prediction model for 90-day mortality following surgical treatment of clinical stage I nonsmall-cell lung cancer (NSCLC).

BACKGROUND

While surgery remains the preferred treatment for functionally fit patients with early-stage NSCLC, less invasive, nonsurgical treatments have emerged for high-risk patients. Accurate risk prediction models for postoperative mortality may aid surgeons and other providers in optimizing patient-centered treatment plans.

METHODS

We performed a retrospective cohort study using a uniquely compiled VA data set including all Veterans with clinical stage I NSCLC undergoing surgical treatment between 2006 and 2016. Patients were randomly split into derivation and validation cohorts. We derived the VALCAN-M score based on multivariable logistic regression modeling of patient and treatment variables and 90-day mortality.

RESULTS

A total of 9749 patients were included (derivation cohort: n=6825, 70.0%; validation cohort: n=2924, 30.0%). The 90-day mortality rate was 4.0% (n=390). The final multivariable model included 11 factors that were associated with 90-day mortality: age, body mass index, history of heart failure, forced expiratory volume (% predicted), history of peripheral vascular disease, functional status, delayed surgery, American Society of Anesthesiology performance status, tumor histology, extent of resection (lobectomy, wedge, segmentectomy, or pneumonectomy), and surgical approach (minimally invasive or open). The c statistic was 0.739 (95% CI=0.708-0.771) in the derivation cohort.

CONCLUSIONS

The VALCAN-M score uses readily available treatment-related variables to reliably predict 90-day operative mortality. This score can aid surgeons and other providers in objectively discussing operative risk among high-risk patients with clinical stage I NSCLC considering surgery versus other definitive therapies.

Inequities in referrals to a breast cancer risk assessment and prevention clinic: a mixed methods study

BACKGROUND

Inequitable access to personalized breast cancer screening and prevention may compound racial and ethnic disparities in outcomes. The Breast Cancer Personalized Risk Assessment, Education and Prevention (B-PREP) program, located within the Brigham and Women's Hospital (BWH) Comprehensive Breast Health Center (BHC), provides care to patients at high risk for developing breast cancer. We sought to characterize the differences between BWH primary care patients referred specifically to B-PREP for risk evaluation and those referred to the BHC for benign breast conditions. Through interviews with primary care clinicians, we sought to explore contributors to potentially inequitable B-PREP referral patterns.

METHODS

We used electronic health record data and the B-PREP clinical database to identify patients referred by primary care clinicians to the BHC or B-PREP between 2017 and 2020. We examined associations with likelihood of referral to B-PREP for risk assessment. Semi-structured interviews were conducted with nine primary care clinicians from six clinics to explore referral patterns.

RESULTS

Of 1789 patients, 78.0% were referred for benign breast conditions, and 21.5% for risk assessment. In multivariable analyses, Black individuals were less likely to be referred for risk than for benign conditions (OR 0.38, 95% CI:0.23-0.63) as were those with Medicaid/Medicare (OR 0.72, 95% CI:0.53-0.98; OR 0.52, 95% CI:0.27-0.99) and those whose preferred language was not English (OR 0.26, 95% CI:0.12-0.57). Interviewed clinicians described inconsistent approaches to risk assessment and variable B-PREP awareness.

CONCLUSIONS

In this single-site evaluation, among individuals referred by primary care clinicians for specialized breast care, Black, publicly-insured patients, and those whose preferred language was not English were less likely to be referred for risk assessment. Larger studies are needed to confirm these findings. Interventions to standardize breast cancer risk assessment in primary care may improve equity.

Performance of Statistical and Machine Learning Risk Prediction Models for Surveillance Benefits and Failures in Breast Cancer Survivors

BACKGROUND

Machine learning (ML) approaches facilitate risk prediction model development using high-dimensional predictors and higher-order interactions at the cost of model interpretability and transparency. We compared the relative predictive performance of statistical and ML models to guide modeling strategy selection for surveillance mammography outcomes in women with a personal history of breast cancer (PHBC).

METHODS

We cross-validated seven risk prediction models for two surveillance outcomes, failure (breast cancer within 12 months of a negative surveillance mammogram) and benefit (surveillance-detected breast cancer). We included 9,447 mammograms (495 failures, 1,414 benefits, and 7,538 nonevents) from years 1996 to 2017 using a 1:4 matched case-control samples of women with PHBC in the Breast Cancer Surveillance Consortium. We assessed model performance of conventional regression, regularized regressions (LASSO and elastic-net), and ML methods (random forests and gradient boosting machines) by evaluating their calibration and, among well-calibrated models, comparing the area under the receiver operating characteristic curve (AUC) and 95% confidence intervals (CI).

RESULTS

LASSO and elastic-net consistently provided well-calibrated predicted risks for surveillance failure and benefit. The AUCs of LASSO and elastic-net were both 0.63 (95% CI, 0.60-0.66) for surveillance failure and 0.66 (95% CI, 0.64-0.68) for surveillance benefit, the highest among well-calibrated models.

CONCLUSIONS

For predicting breast cancer surveillance mammography outcomes, regularized regression outperformed other modeling approaches and balanced the trade-off between model flexibility and interpretability.

IMPACT

Regularized regression may be preferred for developing risk prediction models in other contexts with rare outcomes, similar training sample sizes, and low-dimensional features.

Development and validation of the lung donor (LUNDON) acceptability score for pulmonary transplantation

There is a chronic shortage of donor lungs for pulmonary transplantation due, in part, to low lung utilization rates in the United States. We performed a retrospective cohort study using data from the Scientific Registry of Transplant Recipients database (2006-2019) and developed the lung donor (LUNDON) acceptability score. A total of 83 219 brain-dead donors were included and were randomly divided into derivation (n = 58 314, 70%) and validation (n = 24 905, 30%) cohorts. The overall lung acceptance was 27.3% (n = 22 767). Donor factors associated with the lung acceptance were age, maximum creatinine, ratio of arterial partial pressure of oxygen to fraction of inspired oxygen, mechanism of death by asphyxiation or drowning, history of cigarette use (≥20 pack-years), history of myocardial infarction, chest x-ray appearance, bloodstream infection, and the occurrence of cardiac arrest after brain death. The prediction model had high discriminatory power (C statistic, 0.891; 95% confidence interval, 0.886-0.895) in the validation cohort. We developed a web-based, user-friendly tool (available at https://sites.wustl.edu/lundon) that provides the predicted probability of donor lung acceptance. LUNDON score was also associated with recipient survival in patients with high lung allocation scores. In conclusion, the multivariable LUNDON score uses readily available donor characteristics to reliably predict lung acceptability. Widespread adoption of this model may standardize lung donor evaluation and improve lung utilization rates.

A Catchment and Location-Allocation Analysis of Mammography Access in Delaware, US: Implications for disparities in geographic access to breast cancer screening

Background

Despite a 40% reduction in breast cancer mortality over the last 30 years, not all groups have benefited equally from these gains. A consistent link between later stage of diagnosis and disparities in breast cancer mortality has been observed by race, socioeconomic status, and rurality. Therefore, ensuring equitable geographic access to screening mammography represents an important priority for reducing breast cancer disparities. This study conducted a catchment and location-allocation analysis of mammography access in Delaware, a state that is representative of the US in terms of race and urban-rural characteristics and experiences an elevated burden from breast cancer.

Methods

A catchment analysis using the ArcGIS Pro Service Area analytic tool characterized the geographic distribution of mammography sites and Breast Imaging Centers of Excellence (BICOEs). Poisson regression analyses identified census tract-level correlates of access. Next, the ArcGIS Pro Location-Allocation analytic tool identified candidate locations for the placement of additional mammography sites in Delaware according to several sets of breast cancer screening guidelines.

Results

The catchment analysis showed that for each standard deviation increase in the number of Black women in a census tract, there were 64% (95% CI, 0.18-0.66) fewer mammography units and 85% (95% CI, 0.04-0.48) fewer BICOEs. The more rural counties in the state accounted for 41 % of the population but only 22% of the BICOEs. The results of the location-allocation analysis depended on which set of screening guidelines were adopted, which included increasing mammography sites in communities with a greater proportion of younger Black women and in rural areas.

Conclusions

The results of this study illustrate how catchment and location-allocation analytic tools can be leveraged to guide the equitable selection of new mammography facility locations as part of a larger strategy to close breast cancer disparities.