Cancer Risk Assessment Models Research Articles

Deep learning vs traditional breast cancer risk assessment models: Are we offering supplemental services to those who benefit the most?

1556 Background: Mammography-based deep learning (DL) risk stratification models improve discriminatory accuracy in predicting future breast cancer compared to traditional risk models; however, decisions for and insurance payment of supplemental services such as MRI are driven by traditional scores. The purpose of this study was to compare risk scores and cancer detection rates in patients identified as increased risk by DL vs traditional models in a large screening mammography cohort. Methods: This multisite study included consecutive patients ≥40y undergoing routine bilateral screening mammography from 9/2017 to 1/2022 at five facilities. Tyrer-Cuzick version 8 (TC) and National Cancer Institute Breast Cancer Risk Assessment Tool (BCRAT) 5-year and lifetime models and a DL 5-year model were used to assess risk. The following thresholds were used to define intermediate risk: ≥1.67% for TC and BCRAT 5-year, 2.2 for DL model and high risk: ≥20% for TC and BCRAT lifetime, ≥6.0 for DL model. Patients were included if all risk scores were available. Patient demographics were retrieved from electronic medical records and cancer outcomes through regional tumor registry linkage. The proportion of increased-risk patients and cancer detection rates (CDR [cancers per 1000 women screened]) across models were compared using Pearson’s Chi-square test. Results: 148476 exams in 69464 patients (mean age 58.0y [IQR: 50.0-67.0]) were performed. 81.7% (119386/146075) of exams were in White patients and 18.3% (26689/146075) in races other than White. CDR of those classified as intermediate risk by DL was 7.1 (450/63109) vs 4.7 (320/68545; P<0.001) by TC and 4.2 (300/70831; P<0.001) by BCRAT, and as high risk was 20.6 (305/14772) by DL vs 3.9 (64/16442; P<0.001) by TC and 5.0 (29/5782; P<0.001) by BCRAT. 13.0% (19248/148476) were classified as intermediate risk by DL but not by TC or BCRAT, with a CDR of 6.0 in this cohort. 32.0% (47435/148476) were classified as intermediate risk by TC or BCRAT but not by DL with a CDR of 1.1. 8.5% (12577/148476) were classified as high risk by DL but not by TC or BCRAT with a CDR of 20.8. 11.1% (16451/148476) were classified as high risk by TC or BCRAT but not by DL with a CDR of 1.9. Conclusions: A significant proportion of patients identified as increased risk by DL model are not assessed as increased risk by commonly used risk models, effectively excluding them from accessing potentially life-saving supplemental services. DL scores met the American College of Radiology's acceptable CDR on screening mammography (≥2.5) regardless of traditional scores, however, elevated traditional scores in the absence of elevated DL scores did not meet acceptable standards. Risk assessment guidelines incorporating DL models are necessary to capture patients who benefit the most.

Personalized survivorship care: Routine breast cancer risk assessment in the gynecologic oncology clinic

IntroductionGynecologic and breast cancers share several risk factors. Breast cancer risk assessment tools can identify those at elevated risk and allow for enhanced breast surveillance and chemoprevention, however such tools are underutilized. We aim to evaluate the use of routine breast cancer risk assessment in a gynecologic oncology clinic. MethodsA patient-facing web-based tool was used to collect personal and family history and run four validated breast cancer risk assessment models (Tyrer-Cuzick (TC), Gail, BRCAPRO, and Claus) in a gynecologic oncology clinic. We evaluated completion of the tools and identification of patients at elevated risk for breast cancer using the four validated models. ResultsA total of 99 patients were included in this analysis. The BRCAPRO model had the highest completion rate (84.8%), followed by the TC model (74.7%), Gail model (74.7%), and the Claus model (52.1%). The TC model identified 21.6% of patients completing the model as having ≥20% lifetime risk of breast cancer, compared to 6.8% by the Gail model, and 0% for both the BRCAPRO and Claus models. The Gail model identified 52.5% of patients as having ≥1.67% 5-year risk of breast cancer. Among patients identified as high-risk for breast cancer and eligible for screening, 9/9 (100%) were referred to a high-risk breast clinic. ConclusionAmong patients that completed the TC breast cancer risk assessment in a gynecologic oncology clinic, approximately 1 in 5 were identified to be at significantly elevated lifetime risk for breast cancer. The gynecologic oncologist's office might offer a convenient and feasible setting to incorporate this risk assessment into routine patient care, as gynecologic oncologists often have long-term patient relationships and participate in survivorship care.

Next Top Model: An Overview of Breast Cancer Risk Assessment Models

Next Top Model: An Overview of Breast Cancer Risk Assessment Models

Breast cancer: genetic personal risk factors: A review

Determination of cancer risk factors allow us to develop diagnostics tests that improved identification and reduced the rate of mortality of most frequent cancer diseases including breast cancer, prostate cancer, gastrointestinal tumors. Today individual risk of breast cancer considers personal genetics, medical history of patient, lifestyle, and a number of additional factors. Calculation of the first mathematical models for breast cancer risk assessment included anthropometric data, hormonal status, and family history of cancer. The discovery of BRCA1 and BRCA2 genes role in the development of breast cancer and the accumulation of data from population studies contributed to the introduction of the genetic component into mathematical models. The trend of the last decade is the integration of the polygenic component into the scheme for calculating the individual risk of breast cancer. In this review, we have analyzed existing models, assessed their relevance for certain groups of patients, studied the trends in the development of methods for molecular genetic diagnosis of breast cancer and determining the personal risk of developing the disease.

Deep Learning Approaches with Digital Mammography for Evaluating Breast Cancer Risk, a Narrative Review.

Breast cancer remains the leading cause of cancer-related deaths in women worldwide. Current screening regimens and clinical breast cancer risk assessment models use risk factors such as demographics and patient history to guide policy and assess risk. Applications of artificial intelligence methods (AI) such as deep learning (DL) and convolutional neural networks (CNNs) to evaluate individual patient information and imaging showed promise as personalized risk models. We reviewed the current literature for studies related to deep learning and convolutional neural networks with digital mammography for assessing breast cancer risk. We discussed the literature and examined the ongoing and future applications of deep learning techniques in breast cancer risk modeling.

Unlocking learner potential: A course director's experience building an Educational Escape Room for Cancer Risk Assessment.

Using a traditional lecture format alone often limits students' opportunities to engage with and apply the concepts taught in a classroom. This report describes the design, curricular implementation, and application of an innovative educational escape room as an active learning tool for genetic counselor educators to teach cancer risk assessment skills. A Cancer Risk Assessment Escape Room is a novel way to offer experiential learning and to assess a learner's critical thinking and ability to apply cancer risk assessment models to evolving clinical scenarios. This successful implementation of an interactive escape room can be adapted to other genetic counseling competencies. The use of an escape room allows learners to actively engage in and apply content taught in lectures before entering clinical internships.

The History of the Linear No-Threshold Model and Recommendations for a Path Forward

The intent of this paper and the accompanying video series is to inform the scientific community about the historical foundations that underpin the linear no-threshold (LNT) model's use for cancer risk assessment. There is a clear distinction here: this effort is about the history of how LNT came to be the regulatory paradigm and model for cancer risk assessment that it is today and not a discussion of the pros and cons of the LNT model. The overarching goal of this effort is to reframe the conversation around low-dose response models in light of this history and to determine how this history influences the scientific understanding of low-dose radiation responses. The timing of this series is intentional, as the International Commission on Radiological Protection (ICRP) has embarked on a mission to review the entire system of radiation protection. This effort necessarily requires rigorous scientific debate that must be based in fact. The history of the LNT model is paramount to this discussion, and it warrants consideration. Unfortunately, rather than engendering respectful debate, the topic of cancer risks associated with low dose radiation exposures has forged two disparate and sometimes contentious camps: (1) low doses, no matter how low, present some form of health risk and (2) an alternative model better represents the actual risks. The video series, conceived by John Cardarelli II, current President of the Health Physics Society (HPS), features Edward Calabrese, professor of toxicology in the School of Public Health and Health Sciences at the University of Massachusetts at Amherst, being interviewed by HPS Past-President Barbara Hamrick, CHP, JD, with support from Daniel Sowers, the Chair of the HPS Public Information Committee, and HPS Executive Director Brett Burk. Emily Caffrey, the Chief Editor of our Ask-the-Experts website (https://hps.org/publicinformation/ate/), was invited to watch the completed series as an independent peer reviewer. Further, an email address, factcheck@hps.org, was created to allow for peer-review by the scientific community to facilitate ongoing discussion and allow for corrections to the record as necessary. It is the sincere hope of this team that this work inspires new discussions about the system of radiological protection. We encourage everyone in this field to watch all 22 episodes to be informed about the underpinnings of current regulatory policy in the US.

Optimization of Cancer Risk Assessment Models for PM2.5-Bound PAHs: Application in Jingzhong, Shanxi, China.

The accurate evaluation of the carcinogenic risk of PM2.5-bound polycyclic aromatic hydrocarbons (PAHs) is crucial because of the teratogenic, carcinogenic, and mutagenic effects of PAHs. The best model out of six models was selected across three highly used categories in recent years, including the USEPA-recommended inhalation risk (Model I), inhalation carcinogen unit risk (Models IIA-IID), and three exposure pathways (inhalation, dermal, and oral) (Model III). Model I was found to be superior to the other models, and its predicted risk values were in accordance with the thresholds of PM2.5 and benzo[a]pyrene in ambient-air-quality standards. Models IIA and III overestimated the risk of cancer compared to the actual cancer incidence in the local population. Model IID can replace Models IIB and IIC as these models exhibited no statistically significant differences between each other. Furthermore, the exposure parameters were optimized for Model I and significant differences were observed with respect to country and age. However, the gender difference was not statistically significant. In conclusion, Model I is recommended as the more suitable model, but in assessing cancer risk in the future, the exposure parameters must be appropriate for each country.

Risk prediction models for head and neck cancer: A rapid review.

Cancer risk assessment models are used to support prevention and early detection. However, few models have been developed for head and neck cancer (HNC). A rapid review of Embase and MEDLINE identified n=3045 articles. Following dual screening, n=14 studies were included. Quality appraisal using the PROBAST (risk of bias) instrument was conducted, and a narrative synthesis was performed to identify the best performing models in terms of risk factors and designs. Six of the 14 models were assessed as "high" quality. Of these, three had high predictive performance achieving area under curve values over 0.8 (0.87-0.89). The common features of these models were their inclusion of predictors carefully tailored to the target population/anatomical subsite and development with external validation. Some existing models do possess the potential to identify and stratify those at risk of HNC but there is scope for improvement.

22 (PB-022) Poster Spotlight - An individualized breast cancer risk assessment model for women attending screening in BreastScreen Norway

Background: The results of the effectiveness of population-based screening are controversial in terms of the balance between mortality reduction and adverse effects. In order to improve it, studies have proposed personalized screening strategies based on woman’s individual breast cancer (BC) risk. There is, therefore, a need to create individual risk prediction models through the analysis of large population-based databases. We developed a model that could be used to classify women targeted for mammography screening according to individual BC risk.

Can National Registries Contribute to Predict the Risk of Cancer? The Cancer Risk Assessment Model (CRAM).

Simple SummaryEarly identification of individuals with an increased risk of cancer is an important challenge. Danish administrative registers may be useful in this respect because they cover the entire population and include comprehensive and consistently coded long-term data. We aimed to develop a predictive model based on Danish administrative registers to facilitate the automated identification of individuals at risk of any type of cancer. In addition to age, almost all the included factors contributed statistically significantly, but also only marginally, to the prediction models, which means that we have not overlooked obvious information available in the register. Future prediction studies should focus on specific cancer types where more precise risk estimations might be expected. It is our ultimate ambition that an effective model can be used at the point of care, integrated into electronic patient record systems to alert physicians of patients at a high risk of cancer.Purpose: To develop a predictive model based on Danish administrative registers to facilitate automated identification of individuals at risk of any type of cancer. Methods: A nationwide register-based cohort study covering all individuals in Denmark aged +20 years. The outcome was all-type cancer during 2017 excluding nonmelanoma skin cancer. Diagnoses, medication, and contact with general practitioners in the exposure period (2007–2016) were considered for the predictive model. We applied backward selection to all variables by logistic regression to develop a risk model for cancer. We applied the models to the validation cohort, calculated the receiver operating characteristic curves, and estimated the corresponding areas under the curve (AUC). Results: The study population consisted of 4.2 million persons; 32,447 (0.76%) were diagnosed with cancer in 2017. We identified 39 predictive risk factors in women and 42 in men, with age above 30 as the strongest predictor for cancer. Testing the model for cancer risk showed modest accuracy, with an AUC of 0.82 (95% CI 0.81–0.82) for men and 0.75 (95% CI 0.74–0.75) for women. Conclusion: We have developed and tested a model for identifying the individual risk of cancer through the use of administrative data. The models need to be further investigated before being applied to clinical practice.

Epidemiological characteristics and risk factors of lung adenocarcinoma: A retrospective observational study from North China.

BackgroundThe main aim of the study was to determine the risk factors of lung adenocarcinoma and to analyze the variations in the incidence of lung adenocarcinoma according to time, sex, and smoking status in North China.MethodsPatients with lung cancer in local household registries diagnosed and treated for the first time in the investigating hospital were enrolled from 11 cities in North China between 2010 and 2017. Baseline characteristics and tumor-related information were extracted from the patients’ hospital medical record, clinical course records, and clinical examination. Some of the variables, such as smoking, alcohol consumption, medical history, and family history of cancer, were obtained from interviews with the enrolled patients. The statistical method used were the chi-square test and multi-factor logistic regression analysis. The time trend was statistically analyzed using Joinpoint regression models, and p values were calculated.ResultsA total of 23,674 lung cancer cases were enrolled. People in severely polluted cities were at higher risk for lung adenocarcinoma (p < 0.001). Most patients with lung adenocarcinoma had no history of lung-related diseases (p = 0.001). Anatomically, lung adenocarcinoma was more likely to occur in the right lung (p < 0.001). Non-manual labor workers were more likely to develop from lung adenocarcinoma than manual workers (p = 0.015). Notably, non-smokers were more likely to develop lung adenocarcinoma than smokers (p < 0.001). The proportion of lung adenocarcinoma increased significantly in Hebei Province (p < 0.001). Among non-smokers, the proportion of lung adenocarcinoma showed a higher rise than in smokers (p < 0.001).ConclusionsLung adenocarcinoma is the most common histological type of lung cancer in North China (Hebei Province), and the proportion of lung adenocarcinoma is increasing, especially among non-smokers. Lung adenocarcinoma is more common in women, severely polluted cities, individuals with no history of lung-related diseases, in the right lung, and in non-smokers. These can serve as a great guide in determining the accuracy of lung adenocarcinoma high-risk groups and lung cancer risk assessment models.

Cancer risk assessment models versus National Comprehensive Cancer Network guidelines: What is the most comprehensive method to assess cancer risk and triage for genetic testing? (373)

Cancer risk assessment models versus National Comprehensive Cancer Network guidelines: What is the most comprehensive method to assess cancer risk and triage for genetic testing? (373)

Clinical tools and counseling considerations for breast cancer risk assessment and evaluation for hereditary cancer risk.

Breast cancer is a complex and heterogeneous disease. Unfortunately, it is the most common malignancy diagnosed in women in the USA, with 281,550 new cases of invasive breast cancer and 49,290 new cases of noninvasive breast cancer are diagnosed per year. In England, it is currently estimated that approximately 1 in 7 (14%) women will be diagnosed with breast cancer in their lifetime. In the UK in 2017, 54,700 women and 390 men were diagnosed with breast cancer. The risk of breast cancer is influenced by many factors, including but not limited to age, family history, reproductive history, hormonal exposure, proliferative breast lesions, physical activity, alcohol use, tobacco use, breast density, and environmental exposures. Breast cancer risk assessment is a critical part of public health. By identifying women at high risk for breast cancer, personalized recommendations can be deployed with regards to modes of screening, the age to initiate breast screening, and the frequency for completing such screens. In addition, breast cancer risk assessment can assist in determining a woman's eligibility for interventions to reduce risk, either through the use of chemoprevention medications or through surgical means with risk-reducing bilateral mastectomy. This chapter summarizes breast cancer risk assessment models and discusses interventions to reduce breast cancer risk to aid in reducing morbidity and mortality from breast cancer.

Beyond the AJR: Validation of an Artificial Intelligence Breast Cancer Risk Assessment Model Across Diverse International Cohorts

Beyond the <i>AJR</i>: Validation of an Artificial Intelligence Breast Cancer Risk Assessment Model Across Diverse International Cohorts

Laboratory-related outcomes from integrating an accessible delivery model for hereditary cancer risk assessment and genetic testing in populations with barriers to access

PurposeThis study aimed to evaluate the laboratory-related outcomes of participants who were offered genomic testing based on cancer family history risk assessment tools. MethodsPatients from clinics that serve populations with access barriers, who are screened at risk for a hereditary cancer syndrome based on adapted family history collection tools (the Breast Cancer Genetics Referral Screening Tool and PREMM5), were offered exome-based panel testing for cancer risk and medically actionable secondary findings. We used descriptive statistics, electronic health record review, and inferential statistics to explore participant characteristics and results, consultations and actions related to pathogenic/likely pathogenic variants identified, and variables predicting category of findings, respectively. ResultsOf all the participants, 87% successfully returned a saliva kit. Overall, 5% had a pathogenic/likely pathogenic cancer risk variant and 1% had a secondary finding. Almost all (14/15, 93%) participants completed recommended consultations with nongenetics providers after an average of 17 months. The recommended actions (eg, breast magnetic resonance imaging) were completed by 17 of 25 participants. Participant personal history of cancer and PREMM5 score were each associated with the category of findings (history and colon cancer finding, Fisher’s exact P = .02; history and breast cancer finding, Fisher’s exact P = .01; PREMM5TM score; and colon cancer finding, Fisher’s exact P < .001). ConclusionThis accessible model of hereditary cancer risk assessment and genetic testing yielded results that were often acted upon by patients and physicians.

Prediction of Short-Term Breast Cancer Risk with Fusion of CC- and MLO-Based Risk Models in Four-View Mammograms.

This study performed and assessed a novel program to improve the accuracy of short-term breast cancer risk prediction by using information from craniocaudal (CC) and mediolateral-oblique (MLO) views of two breasts. An age-matched dataset of 556 patients with at least two sequential full-field digital mammography examinations was applied. In the second examination, 278 cases were diagnosed and pathologically verified as cancer, and 278 were negative, while all cases in the first examination were negative (not recalled). Two generalized linear-model-based risk prediction models were established with global- and local-based bilateral asymmetry features for CC and MLO views first. Then, a new fusion risk model was developed by fusing prediction results of the CC- and MLO-based risk models with an adaptive alpha-integration-based fusion method. The AUC of the fusion risk model was 0.72 ± 0.02, which was significantly higher than the AUC of CC- or MLO-based risk model (P < 0.05). The maximum odds ratio for CC- and MLO-based risk models were 8.09 and 5.25, respectively, and increased to 11.99 for the fusion risk model. For subgroups of patients aged 37-49years, 50-65years, and 66-87years, the AUCs of 0.73, 0.71, and 0.75 for the fusion risk model were higher than AUC for CC- and MLO-based risk models. For the BIRADS 2 and 3 subgroups, the AUC values were 0.72 and 0.71 respectively for the fusion risk model which were higher than the AUC for the CC- and MLO-based risk models. This study demonstrated that the fusion risk model we established could effectively derive and integrate supplementary and useful information extracted from both CC and MLO view images and adaptively fuse them to increase the predictive power of the short-term breast cancer risk assessment model.

Abstract P2-10-06: Validation of abridged breast cancer risk assessment model for the general population

Abstract The ability to efficiently and accurately risk stratify women can lead to improved risk-reduction measures. Actionable risk thresholds for both screening and risk-reduction guidance exist; however, despite several clinical risk models currently in existence, the utilization of such models is underwhelming. We have developed and validated an abridged clinical risk model, COSMO (Clinical pOlygenic denSity faMily mOdel), incorporating three of the most relevant epidemiological risk factors associated with sporadic and familial breast cancer in a format that is designed for “ease of use”. Polygenic risk is recognized as a promising epidemiological risk factor however, its value is greater when integrated with other phenotypic risk factors including 1st and 2nd family history, breast density menopausal status, BMI and age. We used 200,009 active UK Biobank (UKB) members who were female, genetically Caucasian, had SNP data for 313 breast cancer associated variants and aged 40 to 69 years at baseline assessment date. We used age and date of baseline assessment to calculate censoring age for those who had died before completing 5 years of follow-up and to determine the date at which 5 years of follow-up was completed for the others. We used both self-reported and linked cancer registry data (ICD9 code 174 or ICD10 code C50) to determine the earliest diagnosis of invasive breast cancer for each affected woman. In terms of discrimination, COSMO (AUC=0.651; 95% CI 0.642, 0.661) performed better than the Gail model (AUC=0.567; 95% CI=0.557, 0.567); P value for difference &amp;lt;0.0001. Overall calibration showed the number of breast cancers expected using COSMO was slightly higher than the number observed (SIR=0.94; 95% CI=0.91, 0.97), while for the Gail model, the number of breast cancers was lower than the number observed (SIR=1.07; 95% CI=1.04, 1.11). The modest over estimation may be explained by the “healthy volunteer” selection bias of the UKB. Importantly, COSMO significantly improved predicted risk classification of women compared to the Gail when stratified according to actionable thresholds. (Table) Limitations to the validation include the lack of mammographic breast density data in the UKB—a critical factor in COSMO. To this end, we have an ongoing validation in a Nurses’ Health Study population (1,269 cases and 1,812 controls) that includes breast density data. Ongoing model comparison studies will confirm the equivalence of COSMO to standard clinical models. If performance is equivalent, COSMO has an advantage in clinical implementation because the abridged model can be less time consuming to implement in a general practice or screening clinic setting due to the truncated pedigree. Furthermore, as multiethnic cohort data access continues to improve, the epidemiological components can be recalibrated for application across any ethnicity. COSMO integrates the most impactful epidemiological risk factors, is well calibrated for the general population and provides improved discrimination over a standard risk model. General population risk assessment like COSMO can lead to the identification of more women who would benefit from clinical intervention based on current risk-reduction guidelines. Beyond that, incidental by products of a general population risk assessment include promoting breast cancer awareness, empowering women to understand their own risk, and encouraging all women to consider their lifestyle habits. Citation Format: Erika Spaeth, Gillian Dite, Richard Allman. Validation of abridged breast cancer risk assessment model for the general population [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P2-10-06.

Are English-language online patient education materials related to breast cancer risk assessment understandable, readable, and actionable?

PurposeTo evaluate the readability, understandability, and actionability of online patient education materials (OPEM) related to breast cancer risk assessment. Material and methodsWe queried seven English-language search terms related to breast cancer risk assessment: breast cancer high-risk, breast cancer risk factors, breast cancer family history, BRCA, breast cancer risk assessment, Tyrer-Cuzick, and Gail model. Websites were categorized as: academic/hospital-based, commercial, government, non-profit or academic based on the organization hosting the site. Grade-level readability of qualifying websites and categories was determined using readability metrics and generalized estimating equations based on written content only. Readability scores were compared to the recommended parameters set by the American Medical Association (AMA). Understandability and actionability of OPEM related to breast cancer high-risk were evaluated using the Patient Education Materials Assessment Tool (PEMAT) and compared to criteria set at ≥70%. Descriptive statistics and inter-rater reliability analysis were utilized. Results343 websites were identified, of which 162 met study inclusion criteria. The average grade readability score was 12.1 across all websites (range 10.8–13.4). No website met the AMA recommendation. Commercial websites demonstrated the highest overall average readability of 13.1. Of the 26 websites related to the search term breast cancer high-risk, the average understandability and actionability scores were 62% and 34% respectively, both below criteria. ConclusionsOPEM on breast cancer risk assessment available to the general public do not meet criteria for readability, understandability, or actionability. To ensure patient comprehension of medical information online, future information should be published in simpler, more appropriate terms.

Review of cohort diversity in the development and validation of cancer risk assessment models.

e18570 Background: Multiple validated models have been proposed to quantify an individual's lifetime risk of breast and colon cancer. However, an evaluation of the race and ethnicity of the cohorts studied in the development and validation of each of these models has not been reviewed. Predicting cancer risks accurately in Black, Indigenous and People of Color (BIPOC) can be crucial in helping to reduce cancer mortality rates and improving access to preventative care for these individuals. Methods: A literature search was conducted to identify published development and validation studies for the following cancer risk assessment models: Breast Cancer Surveillance Consortium (BCSC) Risk Calculator, Tyrer-Cuzick, Gail, Claus, CanRisk/BOADICEA, BRCAPRO and MMRPRO. Articles included were identified through review of a number of electronic databases and websites for the cancer risk prediction models. Authors were contacted for data not readily available through literature search. Results: A total of 15 development studies and 19 validation studies of the cancer risk prediction models were reviewed for the seven models listed above. Out of the 19 validation studies, seven were internal and twelve were external validation studies. 80% (12/15) of development studies and 68% (13/19) of validation studies did not include information on racial and ethnic composition of the cohorts. After obtaining additional information from authors, 53% (8/15) of the development studies were conducted solely in non-Hispanic White (NHW) cohort. The development cohorts ranged from 50%-100% NHW, 0%-7% non-Hispanic Black (NHB), 0%-8% Hispanic/Latinx, 0%-3% Asian and 0%-1% Indigenous participants. 58% (7/12) of external validation studies included ethnically and racially diverse populations compared to 14% (1/7) of internal validation studies. The BCBS, Gail, BRCAPRO and MMRPRO models were the only models with external validation studies conducted in ethnically or racially diverse populations. Overall, the model that had the most diverse cohort for its development and internal validation studies was the BCBS with 70% NHW, 6.7% NHB, 7.5% Hispanic/Latinx, 2.7% Asian, 0.8% Indigenous and 11.5% mixed/other ethnicities. Conclusions: The majority of the models reviewed did not have ethnically or racially diverse populations in their development and validation cohorts. Awareness of the under-representation of ethnically and racially diverse populations in these models is an important precaution for extrapolating data when using these models in medical decision making for BIPOC individuals. Although several barriers exist for participation of BIPOC individuals in clinical studies, these findings highlight the critical, yet unmet need for the development and use of appropriate cancer risk models in racially and ethnically diverse populations as a means to reduce health-related disparities.