Association of Race With Receipt of Proton Beam Therapy for Patients With Newly Diagnosed Cancer in the US, 2004-2018

Key Points

Question  Are there racial disparities in the use of proton beam therapy (PBT) among individuals newly diagnosed with cancer in the US?

Findings  In this cross-sectional study, Black patients were less likely to be treated with PBT than White patients, especially for cancers for which PBT is recommended over photon-based radiation therapy. The racial disparities in receipt of PBT increased over time despite increases in the number of facilities offering PBT in the US.

Meaning  These findings indicate that racial disparities in the use of PBT are substantial, especially for cancers for which PBT is the recommended radiation therapy modality, and that increased PBT availability did not help eliminate racial disparities for these cancers.

Importance  Black patients are less likely than White patients to receive guideline-concordant cancer care in the US. Proton beam therapy (PBT) is a potentially superior technology to photon radiotherapy for tumors with complex anatomy, tumors surrounded by sensitive tissues, and childhood cancers.

Objective  To evaluate whether there are racial disparities in the receipt of PBT among Black and White individuals diagnosed with all PBT-eligible cancers in the US.

Design, Setting, and Participants  This cross-sectional study evaluated Black and White individuals diagnosed with PBT-eligible cancers between January 1, 2004, and December 31, 2018, in the National Cancer Database, a nationwide hospital-based cancer registry that collects data on radiation treatment, even when it is received outside the reporting facility. American Society of Radiation Oncology model policies were used to classify patients into those for whom PBT is the recommended radiation therapy modality (group 1) and those for whom evidence of PBT efficacy is still under investigation (group 2). Propensity score matching was used to ensure comparability of Black and White patients’ clinical characteristics and regional availability of PBT according to the National Academy of Medicine’s definition of disparities. Data analysis was performed from October 4, 2021, to February 22, 2022.

Exposure  Patients’ self-identified race was ascertained from medical records.

Main Outcomes and Measures  The main outcome was receipt of PBT, with disparities in this therapy’s use evaluated with logistic regression analysis.

Results  Of the 5 225 929 patients who were eligible to receive PBT and included in the study, 13.6% were Black, 86.4% were White, and 54.3% were female. The mean (SD) age at diagnosis was 63.2 (12.4) years. Black patients were less likely to be treated with PBT than their White counterparts (0.3% vs 0.5%; odds ratio [OR], 0.67; 95% CI, 0.64-0.71). Racial disparities were greater for group 1 cancers (0.4% vs 0.8%; OR, 0.49; 95% CI, 0.44-0.55) than group 2 cancers (0.3% vs 0.4%; OR, 0.75; 95% CI, 0.70-0.80). Racial disparities in PBT receipt among group 1 cancers increased over time (annual percent change = 0.09, P < .001) and were greatest in 2018, the most recent year of available data.

Conclusions and Relevance  In this cross-sectional study, Black patients were less likely to receive PBT than their White counterparts, and disparities were greatest for cancers for which PBT was the recommended radiation therapy modality. These findings suggest that efforts other than increasing the number of facilities that provide PBT will be needed to eliminate disparities.

Proton beam therapy (PBT) is potentially superior to photon radiation therapy (RT) for tumors with complex anatomy surrounded by sensitive tissues and for childhood cancers, for which decreasing late effects of RT is a major concern.^1,2 Black patients are less likely to receive any RT,³ including use of advanced technologies.^4-6 Previous studies investigating disparities in receipt of PBT evaluated only a single cancer site,^7-15 age group,¹⁶ or geographic region¹⁷ and, importantly, only included patients receiving RT rather than all patients for whom RT is recommended. Because Black patients are less likely than White patients to receive any type of RT,^3,6,18-24 these studies^{7,10-13,16,25-30} might have underestimated the racial disparity in receipt of PBT. Furthermore, differences in referral patterns and regional availability of cancer therapy modalities can influence cancer care receipt^31-36 and racial disparities in access to care.^37-39 The aim of this study was to conduct a comprehensive evaluation of racial disparities in receipt of PBT among individuals diagnosed with all PBT-eligible cancers in the US using recent nationwide data.

The National Cancer Database (NCDB) is a nationwide hospital-based cancer registry jointly sponsored by the American College of Surgeons and the American Cancer Society that captures approximately 72% of all cancer cases in the US from more than 1500 facilities accredited by the American College of Surgeons’ Commission on Cancer and collects RT information even when it is received outside the reporting facility.^40,41 Therefore, the NCDB captures PBT received both at NCDB facilities (59.5% of PBT patients in this study) and at facilities outside the NCDB (40.5% of PBT patients in this study). This cross-sectional study was granted exemption from review by the institutional review board of the Morehouse School of Medicine in Atlanta, Georgia because the study involves secondary data analysis only; therefore, informed consent was not required. All data were deidentified. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Because Black patients are less likely than White patients to receive any type of cancer treatment,⁴² including RT,^3,6,18-24 the study sample was not restricted to patients who received RT to avoid biasing the disparity estimates toward the null. To account for PBT availability and patient opportunity for referral, only patients diagnosed or treated at facilities that reported at least 5 patients receiving PBT between January 1, 2004, and December 31, 2018, or patients who were treated by a radiation oncologist who treated at least 5 patients with PBT between January 1, 2004, and December 31, 2018 (n = 3 961 245), were included. The following patients were excluded from the analysis: patients with missing race information or with race or ethnicity other than non-Hispanic Black and non-Hispanic White (n = 765 912) and patients diagnosed with cancer sites for which fewer than 10 Black patients received PBT throughout the study period because of sparse data to generate stable estimates (n = 1 138 057).

We used the American Society of Radiation Oncology (ASTRO) model policies published in 2017 to classify patients into group 1 and group 2 according to cancer type and RT anatomical target.⁴³ Group 1 (those for whom PBT is the recommended radiation therapy modality) included patients diagnosed with ocular, head and neck (mouth, parotid gland, tonsil, oropharynx, nasopharynx, pyriform sinus, hypopharynx, and paranasal sinuses), central nervous system (CNS) (including cerebral meninges, brain, spinal cord, and other CNS), hepatocellular, skull and spine, and rhabdomyosarcoma cancers. Group 2 (those for whom evidence of PBT efficacy is still under investigation) included patients diagnosed with cancers of prostate, lung, breast, esophagus, colorectum, anus, uterus, cervix, and pancreas and Hodgkin lymphoma.

Patients’ self-identified race was ascertained from medical records. Patient comorbidities were identified according to the modified Charlson-Deyo Comorbidity Index for patients with cancer and categorized into 0, 1, or 2 or greater.⁴⁴

We used propensity score matching to ensure that Black and White patients’ clinical characteristics and regional availability of PBT were comparable. We generated a propensity score for each patient that predicted the probability of being Black. Variables that were deemed relevant a priori using the National Academy of Medicine definition of disparity as “differences in health care services received by the two groups that are not due to underlying health care needs”^45,46 were included in the propensity score model, and no further filtering or selection was conducted. Because structural racism is a primary cause of racial differences in socioeconomic status (SES) by limiting access to education, employment opportunities, and intergenerational transfer of wealth,⁴⁷ we chose not to match on SES variables when estimating racial disparities in receipt of PBT. Similarly, because the Social Security Act of 1935 created a system of employment-based health insurance coverage that interacts with discriminatory hiring practices⁴⁸ to restrict access to health care for racialized groups, we chose not to match on health insurance coverage when estimating racial disparities in receipt of PBT. For each cancer site, the propensity score model included age at diagnosis, sex, cancer stage, comorbidities, year of diagnosis, and geographic region. Patients were matched (1:1) on propensity scores using a greedy match,⁴⁹ wherein Black patients were matched to the nearest White patient, starting with the best match.⁵⁰ To estimate the contribution of modifiable factors to the racial disparity in receipt of PBT,⁴² patients’ zip code of residence median income quintile and health insurance coverage type were further added to separate propensity score models in addition to matching on PBT eligibility and availability.

Data analyses were conducted from October 4, 2021, to February 22, 2022. We used the standardized difference to compare the balance between variables,⁵¹ with no imbalance after propensity score matching (eTables 1-3 in the Supplement). Therefore, unadjusted odds ratios (ORs) and 95% CIs are presented.⁵² We used χ² statistics to compare patients’ characteristics and logistic regression to compare disparities in receipt of PBT by racialized group. To characterize trends in racial disparities in PBT use through time, annual percent change (APC) of the absolute difference in PBT receipt between Black and White patients was calculated by fitting a least-squares regression using year of diagnosis as the independent variable. Changes in trends (structural breaks) were identified by using the additive outliers method.⁵³ In sensitivity analysis, we excluded stage IV cancers, for which PBT is often not recommended as first-course treatment, and stratified the breast cancer analysis by laterality. All analyses were performed using SAS software, version 9.4 (SAS Institute Inc). Statistical significance was set at 2-sided α = .05.

Of the 5 225 929 patients eligible for PBT and included the study, 4 515 679 (86.4%) were White and 710 250 (13.6%) were Black; 2 837 066 (54.3%) were female and 2 388 863 (45.7%) were male; and mean (SD) age at diagnosis was 63.2 (12.4) years. At baseline, Black patients were younger and more likely to be diagnosed with hepatocellular, prostate, and cervical cancers; be uninsured or covered by Medicaid; have comorbidities; be treated at teaching hospitals; live in lower-income and metropolitan areas; and be diagnosed more recently. White patients were more likely to be diagnosed with stage I cancer (Table 1). Less than 1% of patients received PBT for most cancer sites included in the study (eTable 4 in the Supplement).

Black patients were significantly less likely (OR, 0.67; 95% CI, 0.64-0.71) to receive PBT overall and by each ASTRO indication group than their White counterparts (Table 2). Racial disparity in receipt of PBT was higher in group 1 cancers (0.4% vs 0.8%; OR, 0.49; 95% CI, 0.44-0.55) than in group 2 cancers (0.3% vs 0.4%; OR, 0.75; 95% CI, 0.70-0.80) and was statistically significant for rhabdomyosarcoma (OR, 0.50 95% CI, 0.34-0.75), CNS (OR, 0.46; 95% CI, 0.39-0.53), head and neck (OR, 0.48; 95% CI, 0.39-0.59), and hepatocellular (OR, 0.47; 95% CI, 0.28-0.78) cancer (group 1) and prostate (OR, 0.83; 95% CI, 0.76-0.91), breast (OR, 0.60; 95% CI, 0.52-0.68), lung (OR, 0.62; 95% CI, 0.51-0.75), and esophagus (OR, 0.57; 95% CI, 0.36-0.90) cancer (group 2).

The overall disparity measured as absolute difference in receipt of PBT between Black and White patients was statistically significant between 2010 and 2018 (APC = 0.07, P < .001) (Figure 1A). Racial disparities increased over time for group 1 (APC = 0.09, P < .001) and group 2 (APC = 0.06, P = .004) cancers. In group 1 cancers, disparities were greatest in 2018 (Figure 1B), whereas disparities for group 2 cancers decreased in 2018 (Figure 1C), mainly because of an increase in receipt of PBT among Black patients with prostate cancer (Figure 2).

Racial disparities narrowed but remained statistically significant after further matching on health insurance (OR, 0.72; 95% CI, 0.68-0.76) or income (OR, 0.73; 95% CI, 0.69-0.78) overall, by ASTRO group, and by all but 2 (hepatocellular and breast when matching on income) cancer sites (Table 2).

In sensitivity analyses, excluding stage IV cancers did not change the disparity estimates (eTable 5 in the Supplement), and disparity estimates were similar by breast cancer laterality (eTable 6 in the Supplement).

In this large, comprehensive, national evaluation of racial disparities in PBT receipt, Black patients were less likely to be treated with PBT than White patients with similar PBT eligibility and availability at diagnosis. Racial disparities were greater for group 1 cancers, for which PBT is the recommended RT modality, than for group 2 cancers.^43,54,55 In addition, increase in availability of PBT during the study period coincided with increases rather than decreases in the racial disparity in PBT receipt for group 1 cancers, which was greatest in 2018. Further matching on health insurance or income narrowed but did not eliminate the racial disparity in receipt of PBT. These findings underscore the importance of identifying modifiable determinants of access to care other than regional availability to eliminate disparities in PBT receipt.

It is noteworthy that racial disparities in receipt of PBT were highest in group 1 cancers (which are rare and therefore require more frequent interactions with the health care system,^56,57 thus increasing the cumulative burden of exposure to racism⁵⁸), especially for rhabdomyosarcoma, the most common pediatric soft-tissue sarcoma, and CNS cancer, the next most commonly diagnosed cancer in children.⁵⁹ Because PBT reduces the integral dose to surrounding healthy tissue, reducing the risk of secondary malignant neoplasms and other long-term consequences of RT, PBT is especially beneficial in children, making the racial disparities especially concerning.^60-73 These results align with and extend findings from previous studies^11,27 that reported racial disparities in receipt of PBT among pediatric patients with CNS cancer, with Black pediatric patients with cancer being less likely to receive PBT than White patients in previous studies, even when both racialized groups resided in high-income neighborhoods¹⁶ or when both were enrolled in clinical trials.⁷⁴

Black patients were also less likely to receive PBT for hepatocellular and head and neck cancers, for which PBT is the recommended RT modality (group 1).⁴³ One previous study¹³ used older data and restricted the analysis to patients receiving RT (which can underestimate the disparity because Black patients are less likely to receive any type of RT^6,75,76) and did not find significant racial disparities in receipt of PBT for head and neck cancer. Another study³⁰ used older data, restricted the analysis to nonsurgical patients receiving PBT or stereotactic body RT, and found that Black patients were less likely to receive PBT for hepatocellular cancer treatment than White patients.

Among group 2 cancers, racial disparities in PBT receipt were significant for prostate and breast cancers, the group 2 cancers most frequently treated with PBT,⁵⁷ as well as lung and esophagus cancers. Our results are similar to those of older studies that reported racial disparities in receipt of PBT for prostate^{7,10,17,25,28} and breast cancer.^17,26 In prostate cancer, previous studies have shown that White men who reside in more affluent regions and are diagnosed with lower-risk prostate cancer (for which RT provides no survival advantage over active monitoring)⁷⁷ are more likely to receive PBT,⁷ raising concerns about overtreatment. In breast cancer, PBT is thought to have a potentially lower risk of cardiac toxic effects compared with photon therapy,^78-80 which is especially important in avoiding late adverse effects among younger patients treated with RT targeted to the left side after mastectomy.²⁶ We found that Black women, who are more frequently diagnosed with breast cancer at younger ages than White women,^81,82 were half as likely to receive PBT targeted to the left breast as their White counterparts. The racial disparity in receipt of PBT for group 2 cancers seems to have decreased in more recent years, mainly because of an increase in PBT receipt among Black patients with prostate or breast cancer.

As the number of facilities offering PBT in the US increased, improving regional availability of this novel technology, racial disparities in receipt of PBT also increased, especially among patients diagnosed with group 1 cancers, for which PBT is the recommended treatment modality. This result suggests that developing and increasing regional availability of new cancer treatment technologies without addressing structural determinants of access to care can exacerbate instead of ameliorate racial disparities in receipt of quality cancer care.

Health insurance coverage type,^83-85 including inconsistent coverage of PBT among different insurance providers,^86-89 is as an important factor that contributes to racial disparities in receipt of PBT. Because the US system of employment-based health insurance coverage interacts with discriminatory hiring practices,⁴⁸ health insurance is an especially important factor that contributes to racial disparities in receipt of PBT among patients with group 1 cancers, who are more likely to be diagnosed with cancer before 65 years of age,⁵⁷ when US residents become age-eligible for universal health insurance coverage through Medicare. Therefore, policies such as the Patient Protection and Affordable Care Act, with multiple provisions to expand health insurance coverage options, can potentially help address disparities in access to care.^90-92

Another important factor that contributes to racial disparities in receipt of PBT is SES. Individuals who reside in high-income areas are more likely to be treated with PBT than individuals who reside in low-income areas in the US.⁵⁷ Structural racism is a primary cause of racial differences in SES by limiting access to education, employment opportunities, and intergenerational transfer of wealth.⁴⁷ Living in socioeconomic disadvantage creates barriers in access to quality cancer care because of multiple factors, including but not limited to inability to afford out-of-pocket costs of cancer treatment; transportation insecurity; and lack of paid sick leave, job security, and work schedule flexibility to attend numerous cancer treatment appointments.⁹³

After further matching on health insurance coverage type or income, racial disparities in receipt of PBT narrowed but were not eliminated, suggesting that other factors may contribute to these disparities, such as practitioner referral patterns^10,25,28; practitioner implicit bias,^10,28 whereby practitioner treatment recommendations may be influenced by a patient’s race^94-96; and patient experiences of discrimination while interacting with the health care system.⁹⁷ Increased diversity and training among health care professionals could improve sensitivity to cultural contexts⁹⁸ and increase support to strategies that foster systemic changes necessary for equitable access to care.⁹⁹

Because the study period encompasses the time when several PBT trials were being conducted, access to clinical trial enrollment might also have contributed to racial disparities in receipt of PBT. Black patients are less likely to be enrolled in clinical trials,^100-102 including pediatric oncology clinical trials.¹⁰³ A previous study³⁸ found that, in addition to socioeconomic barriers to participation, bias and stereotyping among health care professionals influence recruitment of participants for oncology trials. In fact, an earlier study⁷⁴ demonstrated that Black pediatric patients were less likely to receive PBT than White patients even while enrolled in clinical trials, in which treatment is highly standardized, further undermining the ability of the health care system to demonstrate trustworthiness to individuals from communities targeted for marginalization. Therefore, racial disparities in access to PBT could be diminished by policies and incentives aimed at developing a more diverse and culturally competent oncology workforce.³

Our study has several strengths. It has the largest sample of PBT recipients to date, representing 70% of all patients with newly diagnosed cancer in the US,⁴¹ including information on treatment received outside the reporting facilities, which improves generalizability of our findings. In addition, to our knowledge, this is the first study to include all patients eligible for PBT and the first study to evaluate disparities in access to PBT regardless of receipt of RT (which can lead to underestimation of the disparity because Black patients are less likely to receive any type of RT). This strategy led to the identification of racial disparities in receipt of PBT among patients diagnosed with rhabdomyosarcoma, lung, and esophageal cancers, which have not been previously reported. Furthermore, we implemented several approaches to account for geographic heterogeneity and increased PBT availability over time. First, we excluded patients whose reporting facility or treating radiation oncologist had not treated at least 5 patients with PBT throughout the 14-year study period. Second, to address changes in regional density of PBT centers and the increase in availability of PBT over time,^57,104 we matched Black and White patients on geographic region and diagnosis year. These approaches allowed us to estimate racial disparities in receipt of PBT that are not due to eligibility or regional availability of PBT and are not influenced by downstream consequences of exposure to structural racism (such as SES and health insurance coverage type). Third, we evaluated the contribution of modifiable factors, such as income and health insurance, to the racial disparities in PBT receipt.

Our study also has several limitations. First, propensity score methods are limited by their inability to control for unmeasured confounders.^105,106 Second, the NCDB is not population based; therefore, patterns of PBT receipt might not be representative of the US population. However, NCDB facilities collect RT information even when provided at another facility. In our study, 40% of the patients who received PBT were treated somewhere other than the reporting facility, strengthening the generalizability of our findings. In addition, the demographic and clinical characteristics of patients with cancer in the NCDB are comparable to those from population-based cancer registries.⁴¹ Third, NCDB only collects information for first-course treatment. Therefore, no information is available on use of PBT for reirradiation or treatment of recurrent tumors.¹⁰⁷ No information is available on social services provided at the facility level that might affect racial disparities in access to care (such as transportation services). Fourth, we were not able to look at disparities among patients of other racialized groups or ethnicities because of the small sample size.

The findings of this cross-sectional study raise concerns regarding racial disparities in access to PBT and have policy importance. Racial disparities were greatest for cancers for which PBT is the recommended RT modality (group 1). Of note, racial disparity in PBT receipt did not decrease as the number of facilities that offer PBT in the US increased. The greatest racial disparity for group 1 was in 2018, the most recent year of available data. Further adjusting for modifiable factors known to contribute to racial disparities in access to quality cancer care (income and health insurance coverage type) narrowed but did not eliminate racial disparities in receipt of PBT. Future research should investigate the contribution of practitioner, facility, and health care system characteristics (such as referral patterns and reimbursement policies) to the racial disparity in receipt of PBT.

Accepted for Publication: March 8, 2022.

Published: April 26, 2022. doi:10.1001/jamanetworkopen.2022.8970

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Nogueira LM et al. JAMA Network Open.

Corresponding Author: Leticia M. Nogueira, PhD, MPH, Department of Surveillance and Health Equity Science, American Cancer Society, 250 Williams St, Atlanta, GA 30067 (leticia.nogueira@cancer.org).

Author Contributions: Dr Nogueira had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Efstathiou and Yabroff contributed equally to the work.

Concept and design: Nogueira, Sineshaw, Jemal, Efstathiou, Yabroff.

Acquisition, analysis, or interpretation of data: Nogueira, Sineshaw, Pollack, Efstathiou, Yabroff.

Drafting of the manuscript: Nogueira, Sineshaw, Efstathiou.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Nogueira, Sineshaw.

Obtained funding: Efstathiou.

Administrative, technical, or material support: Jemal, Efstathiou.

Supervision: Efstathiou.

Conflict of Interest Disclosures: Dr Sineshaw reported receiving an American Association for Cancer Research Scholar-in-Training Award to support meeting attendance outside the submitted work. Dr Pollack reported owning stock in Gilead Pharmaceuticals outside the submitted work; in September 2019, Johns Hopkins University entered into a contract with US Department of Housing and Urban Development (HUD) for Dr Pollack to work part time on a temporary assignment assisting the agency with housing and health issues (the findings and conclusions in this report do not necessarily represent those of HUD). Dr Efstathiou reported receiving personal fees for consulting from Blue Earth Diagnostics, Boston Scientific, and AstraZeneca; an honorarium from Genentech; and serving on advisory boards for Merck, Roivant Pharma, Myovant Sciences, Janssen, and Bayer Healthcare outside the submitted work. Dr Yabroff reported serving on the Flatiron Health Equity Advisory Board and receiving honoraria, which is donated to the American Cancer Society, outside the submitted work. No other disclosures were reported.