Анализ географических различий между стандартом медицинского обслуживания и пациентами в Канзасе в ходе клинических испытаний.

Introduction

Chimeric Antigen Receptor (CAR)‐T is a promising and rapidly evolving cellular therapy treatment for relapsed and refractory B cell malignancies and multiple myeloma. However, delivering this therapy requires specialized care before and after the CAR‐T infusion by an authorized treatment center (ATC). Furthermore, only a small portion (20%) of eligible patients do receive approved CAR‐T [ 1 ]. Access to CAR‐T clinical trials disproportionately affects underserved populations, such as those living far away from ATCs, minority populations, and those with low socioeconomic status (SES) [ 2 ]. Previous studies show that only 1% of the population in myeloma CAR‐T trials were African American, and only 7.3% were from neighborhoods with a mean income < $40,000 [ 2 , 3 ]. A third of African Americans live in a county with a trial, and 7 out of 10 states with the highest proportion of African American residents (18.6%–41.4%) have no or fewer than 4 trial sites [ 4 ]. Barriers to clinical trial enrollment exist at multiple levels: structurally, including a lack of referrals or late referrals by community oncology providers (COP) [ 5 , 6 ]; institutionally, with complex and lengthy processes, protocols requiring CAR‐T recipients to relocate within 30 min of an ATC, and limited enrollment opportunities at COP sites [ 7 ]; within the clinical team, due to limited COP awareness, engagement, education, and prioritization of CAR‐T trials [ 5 ]; and at the patient level, encompassing financial and physical burdens of monitoring near an ATC, the need for relocation [ 6 , 8 , 9 ], and negative perceptions of clinical trials [ 10 ]. Prior studies have identified substantial barriers to accessing advanced cellular therapies, including structural, geographic, and system‐level factors that disproportionately affect underserved and rural populations [ 11 ].

The University of Kansas Cancer Center (KUCC) is a CAR‐T ATC, and the only National Cancer Institute (NCI) comprehensive cancer center in the region, serving a catchment area of 4.5 million people in Kansas and 18 counties in western Missouri. Serving the Midwest region, it has a distinctly large catchment area as compared to the East and West Coast ATCs. According to the 2023 Rural–Urban Continuum Codes (RUCC) developed by the U.S. Department of Agriculture's Economic Research Service, Kansas has 89 rural counties (RUCC codes 4 through 9) with its remaining 16 counties as urban (RUCC codes 1 through 3) [ 12 ]. When considering the population distribution in Kansas, 71.1% ( n = 859,138) live in rural areas and 29.2% ( n = 2,081,408) live in urban areas [ 13 ].

The KUCC catchment area includes underrepresented communities, rural areas, health professional shortage areas (HPSAs), and ethnic minorities, which account for over 30% of the population served. Across the catchment area, 24% of the population lives in rural areas or extended areas designated as HPSAs [ 14 , 15 ]. Rural populations have been shown to have higher poverty levels [ 16 ]. African Americans and Hispanics represent 8% and 11% of the catchment area population, respectively [ 16 , 17 , 18 ].

This study aims to compare patient demographics from two cohorts: individuals participating in CAR‐T clinical trials offered through the University of Kansas Cancer Center (KUCC) and those receiving CAR‐T commercially through The University of Kansas Health System (TUKHS). The comparison extends beyond demographic categorization to include geographical locations, with the broader goal of addressing healthcare disparities and alleviating the burden on vulnerable populations. This study examines demographic and geographic access patterns among CAR‐T recipients across B‐cell malignancies and multiple myeloma, conditions that share similar referral pathways and delivery requirements at authorized treatment centers.

Methods

We collected data from two sources: the electronic medical records of The University of Kansas Health System (TUKHS) and the Clinical Trial Management System of The University of Kansas Cancer Center (KUCC). These datasets contained patients' information, including Race (African American, White, Other), Gender (Female, Male), Age (18 & under, 19 to 64, 65 & above), and Zip Code. Patients who received CAR‐T treatment for B cell malignancies or multiple myeloma were divided into two cohorts: those who received FDA‐approved CAR‐T between May 2021 and May 2023 were classified as the Standard of Care (SOC) cohort, while those participating in CAR‐T clinical trials between January 2015 and February 2023 were classified as the Clinical Trial cohort. As a requirement for treatment, any patient who resided more than 30 min from the clinic needed to relocate closer for 4 weeks with a 24/7 caregiver, and no driving was allowed for 8 weeks from the day of infusion. Patients' counties were determined by linking their zip code to a reference county database from the R package zipcodeR [ 19 ].

Patients included in this study received commercially available CD19‐directed CAR‐T therapies approved for the treatment of B‐cell malignancies, including products such as axicabtagene ciloleucel, tisagenlecleucel, and lisocabtagene maraleucel, administered according to institutional protocols.

To conduct our analysis, we created three additional variables: Rurality (Rural, Urban), Health Professional Shortage Area (HPSA) (HPSA shortage, non‐HPSA shortage), and KUMC 50 Mile Radius (> 50 Miles, ≤ 50 Miles). We chose a 50‐mile cutoff based on research showing that patients living more than 50 miles from a hospital often present at a later stage, have lower treatment adherence, and experience worse prognosis and quality of life [ 20 ]. The Rurality variable was generated by merging the patients' county information with the 2013 Rural–Urban Continuum Codes (RUCC) [ 12 ], where RUCC codes of 1–3 were classified as Urban and codes 4–9 as Rural. Similarly, we identified patients from counties with a Health Professional Shortage Area (HPSA) using data from the Health Resources & Services Administration HPSA website. Counties were classified as HPSA shortage if the HPSA variable indicated a “Yes” response and as non‐HPSA shortage if “No.” Lastly, ArcGIS was used to create a list of all zip codes that fell within a 50‐mile radius centered on The University of Kansas Medical Center (KUMC). The dataset was then merged with patient zip codes to identify which patients resided within 50 miles of KUMC. Figure 1 presents a visual representation of geographic regions within the catchment area, classified by their RUCC rural–urban status. Detailed clinical variables, including comorbidities and disease stage, were not consistently available for patients enrolled in clinical trials, limiting the ability to perform fully adjusted comparisons across cohorts.

In our analysis, we first summarized the data to present counts and percentage values. To test whether there was a significant association between categorical variables, we conducted a Pearson Chi‐Squared test. When any expected cell count was less than 5, we used Fisher's Exact test to examine associations between categorical variables, as the Pearson Chi‐Squared test may not be reliable in such cases. For this study we consider the 5% significance level.

To assess associations between binary response variables such as treatment type, rurality, and residence within a 50‐mile radius while adjusting for other patient demographics, we fitted multivariable logistic regression models. Then we performed model reduction through stepwise selection based on the Akaike Information Criterion (AIC) to identify the best‐fitting variables. Odds ratios with 95% confidence intervals were then calculated to interpret the associations.

Results

Table 1 shows the frequency and percentage of different categories of the variables of interest in this study. We found that nearly 80% of the participants are from urban areas, and most of the patients live within a 50‐mile radius of KUMC (54.1%). About 57.4% of the participants do not reside in a Health Professional Shortage Area (HPSA). Female and male represent around 41.6% and 58.4% of the sample, respectively. Most patients fall within the age group of 19 to 64 years (60.40%), with only 1% of patients being 18 years old or younger. Most of the patients are white (87%), followed by African American patients (7%), and 6.3% classified under other races.

Table 2 summarizes the results of the Pearson Chi‐Squared and Fisher's Exact tests for associations between treatment type and various demographic and geographical variables. Specifically, Fisher's Exact test was used to assess the association between treatment type and Age due to the small sample sizes of the 18 and under groups, while Pearson's Chi‐Squared test was used for Gender, Race, Rurality, HPSA, and residence within a 50‐mile radius of KUMC. At the 5% significance level, we fail to reject the null hypothesis for all bivariate analyses involving the response variable, treatment type. Therefore, we do not have sufficient evidence to conclude that an association exists between CAR‐T patients at KUCC and TUKHS enrolled in Standard of Care (SOC) or clinical trials and the variables of gender, age, race, rurality, HPSA status, or residence within a 50‐mile radius of KUMC. Additionally, a multivariable logistic regression model was fitted with CAR‐T therapy as the response variable, using all patient characteristics as predictors. The final multivariable logistic regression model with the lowest AIC value (513.56) was selected using the stepwise procedure, and the results are summarized in Table 3 . However, the analysis revealed no significant associations between CAR‐T therapy and the demographic variables at the 5% significance level.

Table 4 presents the results of the Pearson Chi‐Squared and Fisher's Exact tests for the response variable rurality and the explanatory variables gender, age, race, HPSA status, and residence within a 50‐mile radius of KUMC. The Pearson Chi‐Squared test was used to assess whether there is a significant association between rurality and gender, as well as between rurality and HPSA status. For the remaining explanatory variables—age, race, and residence within a 50‐mile radius of KUMC—the Fisher's Exact test was applied due to the small sample sizes or expected cell counts. Based on the analysis, we concluded with 95% confidence that there are significant associations between rurality and race, age, HPSA status, and residence within a 50‐mile radius of KUMC among CAR‐T patients. However, no significant association was found between patients' rural/urban status and gender. Additionally, there were no rural patients in the study who were African American, and only four were from other racial groups, suggesting that most rural patients were White.

The multivariable logistic regression model, with rurality as the response variable and other patient characteristics as predictors, is summarized in Table 5 . The reduced model with the lowest AIC (291.64) identified age 65 and above and residence within the KUMC 50‐mile radius as significant factors associated with rurality. The odds of a patient being rural in the age group 65 and above are 1.80 (95% CI: 1.00–3.24) times that of the reference age group 19 to 64. Additionally, the odds of rurality for the patients residing outside of the 50‐mile radius of KUMC are 25.76 times (95% CI: 10.86–76.84) that of patients residing within the radius.

Table 6 summarizes the Pearson Chi‐Squared and Fisher's Exact test results for the association between residence within the KUMC 50‐mile radius and other patient characteristics. Significant associations were found between residence within a 50‐mile radius of KUMC and rurality and HPSA status.

The multivariable logistic regression model was fitted using residence within the 50‐mile radius of KUMC as the response and patient characteristics as predictors. Based on the results of the reduced model with the lowest AIC (436.78) summarized in Table 7 , the odds of residing outside the 50‐mile radius of KUMC for patients from a rural area are 27.01 (95% CI: 11.57–79.09) times that of patients from an urban area. Similarly, patients from an HPSA were found to have 2.29 times the odds (95% CI: 1.44–3.66) of residing outside the 50‐mile radius of KUMC compared to those from a non‐HPSA.

Discussion

This study is among the first to characterize demographic and geographic features of CAR T recipients treated at a large Midwestern center, with an emphasis on structural and geographic correlations of access. In our cohort, only a minority of patients reside in rural counties. This pattern is consistent with the underlying population distribution in urban versus rural regions and prior reports suggesting urban–rural differences in lymphoma incidence [ 21 , 22 , 23 , 24 ].

Importantly, our analysis focused on measurable geographic and access‐related indicators: rurality, distance from the treatment center, and Health Professional Shortage Area (HPSA) designation. We observed that rural residence was significantly associated with older age, race, HPSA designation, and living farther from the University of Kansas Medical Center (KUMC) [ 25 ]. In addition, rurality and HPSA designation were strongly associated with residing outside the 50‐mile radius from KUMC. These findings highlight the structural realities of CAR T delivery, which requires intensive monitoring and proximity to specialized care, and therefore may pose greater logistical burdens for patients living farther from an authorized treatment center [ 26 ].

Although this study reflects experience at a single center, the broader delivery model for CAR T therapy is highly centralized and typically requires specialized infrastructure and close follow‐up. Accordingly, geographic distance and proximity requirements are likely to remain important considerations for access in rural and underserved regions across health systems.

We also found that most rural patients in our cohort were White, with no rural patients identified as African American and only a small number from other racial groups. This distribution should be interpreted in the context of the racial and geographic composition of the region.

When comparing clinical trial versus standard‐of‐care (SOC) CAR T recipients, we did not observe statistically significant differences across gender, age, race, rurality, HPSA status, or residence within the 50‐mile radius. However, within the overall cohort, age, race, and HPSA status were significantly associated with rurality, and rurality and HPSA status were significantly associated with residence outside the 50‐mile radius. Together, these results suggest that the primary disparities observed in this study are geographic and structural in nature rather than differences attributable to treatment setting (clinical trial vs. SOC).

Because our dataset did not include individual‐level socioeconomic variables (e.g., income, insurance), referral timing, or patient‐reported barriers (e.g., caregiving capacity, travel resources), we cannot determine the mechanisms driving these observed geographic patterns. Future studies integrating these patient‐level factors and referral pathways will be important to identify actionable targets for improving equitable access to CAR T therapy for rural and HPSA‐designated populations. Although derived from a single Midwestern center, these findings are most applicable to regions with similar centralized CAR‐T delivery models and large rural catchment areas.

Conclusions

Our study uniquely examines the impact of rurality on access to CAR‐T therapy in the Midwest, a region characterized by distinct demographic features and a broad catchment area. While overall demographic patterns between standard‐of‐care and clinical trial CAR‐T recipients appeared comparable, fewer individuals from rural counties—particularly younger patients and minority patients living more than 50 miles from the KU Cancer Center—received CAR‐T therapy, suggesting persistent geographic and access‐related disparities.

To address these challenges, we have established a robust partnership with rural collaborators through the Masonic Cancer Alliance, which enables expanded recruitment and outreach through community sites. This initiative reflects the KU Cancer Center's ongoing commitment to improving equity in cancer care and research participation. Our efforts have meaningfully increased access to clinical trials for rural and historically underrepresented populations, ensuring that our research is more inclusive and representative of the communities we serve.

These findings underscore both the progress made and the critical need for continued investment in innovative, community‐based strategies. Strengthening partnerships with Community‐Oriented Primary Care (COP) teams and rural primary care physicians will be essential to reduce barriers further and advance equitable access to advanced therapies like CAR‐T across all populations.

Although response and prognosis data were not available in this analysis, the observed differences in demographic and clinical characteristics between SOC and CT populations have important implications for the generalizability of clinical trial findings. Underrepresentation of certain groups in clinical trials may limit the applicability of treatment outcomes and contribute to disparities in real‐world effectiveness. Future studies integrating response and survival data are needed to better understand how these differences translate into clinical outcomes.

Future multi‐center studies with larger sample sizes should evaluate disease‐specific and region‐specific access barriers to further refine strategies for improving equitable CAR‐T delivery.

Limitations

This study assumes that data collected at the University of Kansas Medical Center and Health System indicate comprehensive CAR‐T treatment for the catchment area. However, patients could reside in the catchment area and have traveled outside to another hospital for treatment elsewhere. Furthermore, we assume that patients treated at the hospital represent the catchment area population, but this is likely not true as those without medical insurance may not seek treatment at the hospital. Individual‐level socioeconomic data, including income and insurance status, were not consistently available across datasets, particularly for clinical trial participants. As a result, these important determinants of access could not be directly evaluated.

Our second limitation for our study was that it did not evaluate treatment outcomes or therapy‐related toxicities such as cytokine release syndrome or immune effector cell–associated neurotoxicity syndrome. Future studies incorporating detailed clinical and toxicity data may help determine whether geographic or structural factors also influence treatment outcomes and post‐CAR‐T management.

A key limitation of this study is the lack of granular clinical data (such as comorbidities, disease stage), particularly within the clinical trial cohort, which may introduce residual confounding and restrict the interpretation of differences between groups.

Author Contributions

S.P. and D.P.M. conceived and designed the study. S.P., I.R., M.M.R., M.R.I.T., and M.M. conducted data extraction, data curation, and statistical analyses. D.P.M. provided overall study oversight, informatics infrastructure support, and geospatial methodology expertise. A.R., C.T., O.R., J.M., M.H., A.D.S., and N.A. contributed clinical expertise and assisted with interpretation of the findings. S.P. and D.P.M. drafted the manuscript. All authors critically reviewed, revised, and approved the final manuscript and agree to be accountable for all aspects of the work.

Funding

This work was supported by Cancer Center Support Grant, P30CA168524.

Ethics Statement

The study was approved by the Institutional Review Board of the University of Kansas Medical Center, Kansas City, KS (USA).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

References

1. Neelapu S. S., “Targeted Oncology,” 2022, https://www.targetedonc.com/view/many‐challengesopportunities‐for‐car‐t‐cell‐therapies‐in‐lymphoma%202022 .

2. Ahmed N., Shahzad M., Shippey E., et al., “Socioeconomic and Racial Disparity in Chimeric Antigen Receptor T Cell Therapy Access,” Transplantation and Cellular Therapy 28, no. 7 (2022): 358–364.

3. Ahmed N., Sun F., Teigland C., et al., “Chimeric Antigen Receptor T‐Cell Access in Patients With Relapsed/Refractory Large B‐Cell Lymphoma: Association of Access With Social Determinants of Health and Travel Time to Treatment Centers,” Journal of Transplantation and Cellular Therapy 30, no. 7 (2024): 714–725, 10.1016/j.jtct.2024.04.017.

4. Shahzad M., Khalid M. F., Amin M. K., et al., “Geographic and Racial Disparities in Chimeric Antigen Receptor; T Cells and Bispecific Antibodies Trials Access for Diffuse Large B‐Cell Lymphoma,” Clinical Lymphoma, Myeloma & Leukemia 24, no. 5 (2024): 316–322.

5. Burus T., Martinez J., DelNero P., et al., “Interinstitutional Approach to Advancing Geospatial Technologies for US Cancer Centers,” JCO Clinical Cancer Informatics 8 (2024): e2400099.

6. Hoffmann M. S., Hunter B. D., Cobb P. W., Varela J. C., and Munoz J., “Overcoming Barriers to Referral for Chimeric Antigen Receptor T Cell Therapy in Patients With Relapsed/Refractory Diffuse Large B Cell Lymphoma,” Transplantation and Cellular Therapy 29, no. 7 (2023): 440–448.

7. Ahmed N., Wesson W., Lutfi F., et al., “Optimizing the Post‐CAR T Monitoring Period for Axicabtagene Ciloleucel, Tisagenlecleucel, and Lisocabtagene Maraleucel,” Blood Advances 8, no. 20 (2024): 2023012549.

8. Wesson W. and Ahmed N., “Axi‐Cel Outcomes Among Non‐Hispanic Black Patients,” Blood 143, no. 26 (2024): 2681–2682.

9. Ahmed N. and Grover N. S., “Race and Insurance: Real‐World Insights on CAR‐T Outcomes,” Blood Advances 8, no. 10 (2024): 2589–2591.

10. Geana M., Erba J., Krebill H., et al., “Searching for Cures: Inner‐City and Rural Patients' Awareness and Perceptions of Cancer Clinical Trials,” Contemporary Clinical Trials Communications 5 (2017): 72–79.

11. El‐Khazragy N., Ghozy S., Emad P., et al., “Chimeric Antigen Receptor T Cells Immunotherapy: Challenges and Opportunities in Hematological Malignancies,” Immunotherapy 12, no. 18 (2020): 1341–1357, 10.2217/imt-2020-0181.

12. U.S. Department of Agriculture, Economic Research Service , Rural‐Urban Continuum Codes, January 2024, https://www.ers.usda.gov/data‐products/rural‐urban‐continuum‐codes/ .

13. U.S. Department of Agriculture, State Fact Sheets , Kansas, https://www.ers.usda.gov/data‐products/state‐fact‐sheets.aspx .

14. Mudaranthakam D. P., Gajewski B., Krebill H., et al., “Barriers to Clinical Trial Participation: Comparative Study Between Rural and Urban Participants,” JMIR Cancer 8, no. 2 (2022): e33240.

15. Mudaranthakam D. P., Harlan‐Williams L. M., Jensen R. A., et al., “OPTIK: A Database for Understanding Catchment Areas to Guide Mobilization of Cancer Center Assets,” Database 2020 (2020): baaa054.

16. Gafford J. A., Gurley‐Calvez T., Krebill H., Lai S. M., Christiadi, and Doolittle G. C., “Expanding Local Cancer Clinical Trial Options: Analysis of the Economic Impact of the Midwest Cancer Alliance in Kansas,” Academic Medicine 92, no. 9 (2017): 1274–1279, 10.1097/ACM.0000000000001612.

17. Ahmed N., Mahmoudjafari Z., Hamadani M., and Hashmi H., “Bridging the Gap: Early Transition and Hybrid Models of Care to Improve Access to Chimeric Antigen Receptor T Cell Therapy,” Transplantation and Cellular Therapy 29, no. 6 (2023): 399–402.

18. Ahmed N., Wesson W., Pal Mudaranthakam D., et al., “Rural Communities and Health Professional Shortage Areas (Hpsa) and Access to Immune Effector Cellular Therapy and Lymphoma/Myeloma Clinical Trials,” American Society of Clinical Oncology 41 (2023): e18633.

19. Rozzi G. C., “zipcodeR: Advancing the Analysis of Spatial Data at the ZIP Code Level in R,” Software Impacts 9 (2021): 100099, 10.1016/j.simpa.2021.100099.

20. Ambroggi M., Biasini C., Del Giovane C., Fornari F., and Cavanna L., “Distance as a Barrier to Cancer Diagnosis and Treatment: Review of the Literature,” Oncologist 20, no. 12 (2015): 1378–1385, 10.1634/theoncologist.2015-0110.

21. Henley S. J., Anderson R. N., Thomas C. C., Massetti G. M., Peaker B., and Richardson L. C., “Invasive Cancer Incidence, 2004–2013, and Deaths, 2006–2015, in Nonmetropolitan and Metropolitan Counties ‐ United States,” MMWR Surveillance Summaries 66, no. 14 (2017): 1–13.

22. Nasca P. C., Burnett W. S., Greenwald P., Brennan K., Wolfgang P., and Carlton K., “Population Density as an Indicator of Urban‐Rural Differences in Cancer Incidence, Upstate New York, 1968–19721,” American Journal of Epidemiology 112, no. 3 (1980): 362–375.

23. Howe H. L., Keller J. E., and Lehnherr M., “Relation Between Population Density and Cancer Incidence, Illinois, 1986–1990,” American Journal of Epidemiology 138, no. 1 (1993): 29–36.

24. Kansas Data Access and Support Center , Kansas Statistical Abstract 2023 (Kansas Data Access and Support Center, 2023), https://ksdata.ku.edu/ksdata/ksah/population/2pop29.pdf .

25. Medina‐Olivares F. J., Gómez‐De León A., and Ghosh N., “Obstacles to Global Implementation of CAR T Cell Therapy in Myeloma and Lymphoma,” Frontiers in Oncology 14 (2024): 1397613, 10.3389/fonc.2024.1397613.

26. Chung A. P., Shafrin J. T., Vadgama S., et al., “Inequalities in CAR T‐Cell Therapy Access for US Patients With Relapsed/Refractory DLBCL: A SEER‐Medicare Data Analysis,” Blood Advances 9, no. 18 (2025): 4727–4735, 10.1182/bloodadvances.2024015634.

Associated Data

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.