A single source for all syndicated, published materials on oncology healthcare disparities.

Artificial Intelligence: a tool overcoming systemic causes of oncologic disparities

May 12, 2024

April 12, 2024

Irby B. Hunter Jr., Medical Director Oncology Independent Education, Inc & Executive Publisher Oncology Disparities www.oncologydisparities.com

Medical research has been conducted in a manner that has not been fair to members of racial and ethnic minority groups. This has led to underrepresentation of many of these groups in clinical trials, which in turn has limited the generalizability of research results. Additionally, underrepresentation of certain groups has hindered opportunities to examine how candidate treatments may affect them differently. One of the reasons for this underrepresentation is that clinical trial entry requirements may disadvantage underrepresented patients, which leads to lower participation in clinical trials. This month This month, the editors of wwwoncodisparities.com highlight the systemic nature of certain practices that contribute to the ongoing oncologic disparities experienced by underrepresented patients. The editors of wwwoncodisparities.com highlight the impact of such practices which are viewed by many experts to systemic nature on the ongoing oncologic disparities experienced by underrepresented patients.

Lung cancer remains the second most diagnosed cancer among American Blacks, despite the diminishing incidence and mortality of lung cancer in recent years. (1). Robinson-Oghogho et al quantified the relationship between structural racism and this alarming disparity by examining the association between a multidimensional measure of county-level structural racism and county levels of environmental quality. (1) The research group merged 2016-2020 data from the United States Cancer Statistics Data Visualization Tool, a pre-existing county-level structural racism index, the Environmental Protection Agency’s 2006-2010 Environmental Quality Index (EQI), 2023 County Health Rankings, and the 2021 United States Census American Survey. (1) Robinson-Oghogho conducted a multivariable linear regression to examine associations between county-level structural racism and county-level lung cancer incidence and mortality rates. (1) The study demonstrated that structural racism contributes to both the number of new lung cancer cases and the number of deaths caused by lung cancer among Black populations. (1) Among Black males and females, each standard deviation increase in county-level structural racism score was associated with an increase in county-level lung cancer incidence of 6.4 (95% Confidence Interval: 4.4, 8.5) cases per 100,000 and an increase of 3.3 (95% Confidence Interval: 2.0, 4.6). (1)

Steinman recently identified that women MD-PhD physician scientists are less likely to serve as principal investigators on mid and late career awards. (2)) Steinman touts the institutional and systemic changes required to overcome the causes of gender disparities in academic medicine. (2) Additionally, Minarim et al, studied a cohort of 3,129 patients with prostate cancer. (3) The group identified that a significant number of patient reported experiencing discrimination in healthcare settings, with Black patients being significantly more likely to perceive discrimination compared with other racial/ethnic groups. (3)

Wolf et al concluded that clinical trial eligibility may contribute to the underrepresentation of racial groups in clinical trials. (4) Patients with recurrent or progressive endometrial cancer diagnosed from January 2010 to December 2021 who received care at a single institution were identified. (4) Clinical trial eligibility was based on 14 criteria from the KEYNOTE-775 trial. (4) The research group evaluated each ineligibility criterion by race. (4) Wolf’s research highlighted Black patients had 33% lower odds of being eligible (95% Confidence Interval: 0.33-1.34) and were more likely to meeting the exclusion criterion for going onto endometrial cancer clinical trials. (4)

The narrow eligibility criteria may have contributed to the underrepresentation of racial and ethnic subgroups in multiple myeloma clinical trials.(5) Kanapuro et al. conducted a retrospective pooled analysis of multicenter global clinical trials submitted to the US Food and Drug Administration between 2006 and 2019 to support the approval of the use of multiple myeloma therapies that analyze the rates and reasons for trial ineligibility based on race and ethnicity in multiple myeloma clinical trials. (5) Black patients were 17% more likely to be ineligibility than White patients. (5) Failure to meet the hematologic laboratory criteria and treatment related criteria. (5) The group concluded that specific eligibility criteria may contribute to enrollment disparities for Black patients in multiple myeloma clinical trials. (5)

In 2021, Raman et al. examined racial and ethnic differences in recruitment methods and trail eligibility in multisite preclinical Alzheimer Disease trial.(6) The research group conducted a cross-sectional study analyzed screening data from the Anti-Amyloid in Asymptomatic AD study, collected from April 2014 to December 2017.(6) Participants were categorized into 5 mutually exclusive ethnic/racial groups.(6) Data were analyzed from May through December 2020 and included 5945 cognitively unimpaired older adults between the ages of 65 and 85 years screened in North American study sites.(6) Black participants were more frequently excluded for failure to meet cognitive inclusion criteria (eg, screen failures by specific inclusion criteria: 45.5% Black participants vs 26.2% White participants.(6) Racial/ethnic groups differed in sources of recruitment, reasons for screen failure, and probability of eligibility in a preclinical Alzheimer Disease trial.(6)

Flores investigated whether racial/ethnic minority groups and female and older adults are underrepresented among patients in vaccine clinical trials (7) A total of 230 US based trials with 219,555 participants were included in the study. (7) Overall, among adult study participants, White individuals were overrepresented while Black and American Indian participants were underrepresented compared with US census data. (7)

Dianne Pulte et al. analyzed the rates and reasons for trial ineligibility by race and ethnicity in trials of acute myeloid leukemia (AML) submitted to the U.S. Food and Drug Administration (FDA) between 2016 and 2019.(8) The research team examined the rate of ineligibility among participants screened for studies of AML therapies submitted to the FDA from 2016 to 2019.(8) Data were extracted from 13 trials used in approval evaluations, including race, screen status, and reason for ineligibility. The results concluded that patients in historically underrepresented racial and ethnic groups were less likely to meet entry criteria for studies compared to White patients. (8) The lack of relevant disease mutation was the cited as the reason for ineligibility among Black patients.

Riner et al sought to determine the impact of eligibility criteria on disparities in pancreatic ductal adenocarcinoma clinical trials. (9) Patients with PDAC seeking care at the Virginia Commonwealth University Health from 2010 to 2019 were included. (9) Chi-squared tests and unconditional maximum likelihood-based odds rations. Black patients were more likely to ineligible for participation compared with White patient: Black patients were more likely to be ineligible because of renal dysfunction, recent coronary stenting, and uncontrolled diabetes mellitus. (9) Common eligibility criteria differentially exclude Black patients from participating in pancreatic duct adenocarcinoma clinical trials. (9)

Artificial intelligence (AI) systems may be used to overcome the burden and inaccuracy of abstracting data elements from multiple components of the longitudinal health record and matching them to inclusion and exclusion for cancer clinical trials. Thereby diminishing the probability of eligible patients from underrepresented populations are erroneously missed due to laborious task of abstracting data. Recently, the Watson for Clinical Trial Matching Clinical Decision Support System (CDSS), an AI CDSS, was used to identify eligible patients for breast cancer clinical trials. (10) Haddad et al. study included the deidentified data from a cohort of patients with breast cancer seen at the medical oncology clinic of an academic medical center between May and July 2017 and assessed patient eligibility for four breast cancer clinical trials. (10) The AI CDSS tool eligibility screening performance was validated against manual screening. (10) The group evaluated the tools accuracy, sensitivity, specificity, positive predictive value, and negative predictive value for eligibility determinations were calculated. (10) Disagreements between manual screeners and the CDSS were examined to identify sources of discrepancies. (10)

Haddad et al. used Cohen (pairwise) and Fleiss (three-way) to analyze the interrater reliability between manual reviewers and the interrater reliability between manual reviewer at the AI CDSS tool. (10) In total, 318 patients with breast cancer were included. Interrater reliability for manual screening ranged from 0.60 -0.77, indicating substantial agreement. (10) The overall accuracy of breast cancer trial eligibility determinations by the CDSS was 87.6%. AI CDSS tool sensitivity was 81.1% and specificity was 89%. (10)


  • Robinson-Oghogjo, J., Alcarz, K., Thorpe, R., Structural Racism as a contributor to Lung Cancer Incidence and Mortality Rates Among Black Populations in the United States. 2024. Cancer Control. 2024 Jan-Dece 31.
  • Steinman, R., Gandy, L., Qi, H., Fertig, E., Blackbord, A., Grandis, J. Career trajectories of MD-PhD physician scientists: This loss of women investigators. 2024. Cancer Cell. 2024. April.
  • Minarim D, et al. The impact of perceived healthcare discrimination on health outcomes among patients with prostate cancer across racial and ethnic groups. Presented at ASCO Genitourinary Cancers Symposium 2024. Accessed April 21, 2024.
  • Wolf, J., Hamilton, A., An, A., Leonard, J., Kanis, M. Racial Disparities in Endometrial Cancer Clinical Trial Representation: Exploring the Role of Eligibility Criteria. 2024. American Journal of Clinical Oncology. 2024. May 3.
  • Kanapuru, B., Fernandes, L., Baines, A., Ershler, R., Bhatnagar, V., Pulte, E., Gwise, T., Theoret, M. Pazdur, R., Fashoyin, L. Gormley, N. Eligibility criteria and enrollment of a diverse racial and ethnic population in multiple myeloma clinical trials.
  • Raman, R., Quiroz, Y., Langford, O., Choi, J., Ritche, M., Baumgartner, M., Rentz, D., Aggarwal, N., Aisen, P., Sperling, R., Grill, J. Disparities by Race and Ethnicity Among Adults Recruited for a Preclinical Alzheimer Disease Trial. 2021. JAMA Network Open. 2021 July 1;4(7)
  • Flores, L., Frontera, W., Andraski, M., Del Rio, C., Mondriguez-Gonzalez, A., Price, S., Krantz, E., Pergam, S., Silver, J. Assessment of the Inclusion of Racial/Ethnic Minority, Female, and Older Individuals in Vaccine Clinical Trials. 2021. JAMA Network Open. 2021 Feb 1:4(2)
  • Pulte, D., Fernandes L., Wei, G., Woods, A., Norsworthy, N., Kanapuru, B., Gwise, T., Pazdur, R., Schneider, J., Theoret, M., and Fashoyin-Aje, L., Angelo de Claro, R. FDA Analysis of Ineligibility for Acute Myeloid Leukemia Clinical Trails by Race and Ethnicity. 2023. Clinical Lymphoma Myeloma LeukemiaJun;23(6):463-407
  • Riner, A., Girma, S., Vudathia, V., Mukhopadhyay, N., Skorro, N., Gal, T., Freudenberger, D., Herremans, K., George, T., Trevino, J. Eligibility Criteria Perpetuate Disparities in Enrollment and Participation of Black in Pancreatic Cancer Trials. 2022. Journal of Clinical Oncology. 2022 Jul 10;40(20): 2193-2202
  • Haddad, T., Helgeson, J., Pomerleau K., Preininger, A., Roebuck, C., Dankwa-Mullah, I., Jackson, G. Goetz, M. Accuracy of Artificial Intelligence System for Cancer Clinical Trial Eligibility Screening: Retrospective Pilot Study. 2021. JMIR Med Inform 2021;9(3)



Leave the first comment