If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding author at: Department of Radiation Oncology, West Cancer Center and Research Institute, 7945 Wolf River Blvd, Germantown, TN 38138, United States of America.
Background: Older adults are under-represented in cancer clinical trials. However, it remains unclear which types of trials under-enroll aging patients. We aimed to identify associations between trial characteristics and disparate enrollment of older adults onto trials sponsored by the Alliance for Clinical Trials in Oncology (Alliance).
Methods: Actual age ≥ 65 percentage and trial data were extracted from the Alliance closed study list. Each trial, based on its cancer type and years of enrollment, was assigned an expected age ≥ 65 percentage extracted from the Surveillance, Epidemiology, and End Results (SEER) US population-based database. Enrollment disparity difference (EDD), the difference between the expected age ≥ 65 percentage and the actual age ≥ 65 percentage, was calculated for each trial. Linear regression determined trial variables associated with larger EDDs and variables with an overall association p-value <0.20 were included in a multivariable fixed-effects linear model.
Results: The median age of 66,708 patients across 237 trials was 60 years (range 18–102). The average actual age ≥ 65 percentage enrolled per trial was lower than each trial's expected age ≥ 65 percentage average (39% vs. 58%; EDD 19, 95% CI 17.1–21.3%, p < 0.0001). In multivariable analyses, non-genitourinary (GU) cancer types (p < 0.001), trimodality+ trials (estimate 8.78, 95%CI 2.21–15.34, p = 0.009), and phase 2 trials (estimate 4.43 95% CI -0.06-8.91; p = 0.05) were all associated with larger EDDs.
Conclusions: Disparate enrollment of older adults is not equal across cancer trials. Future strategies to improve older adult inclusion should focus on trial types associated with the highest disparate enrollment.
Patients age 65 years and older represent the majority of patients with cancer in the United States (US). As the US population continues to age, the proportion of cancers occurring in older adults is increasing [
Racial/ethnic differences in clinical trial enrollment, refusal rates, ineligibility, and reasons for decline among patients at sites in the National Cancer Institute’s Community Cancer Centers Program.
]. The lack of inclusion of older adults in standard-setting trials has led to a dearth of information and evidence gaps regarding toxicity and important clinical endpoints for older adults with cancer [
Understanding why some studies under-accrue older adults could allow researchers to better design future studies to include patients that are more generalizable to real-world clinics. Previous secondary analysis of >160 consecutive non-age restricted Southwest Oncology Group (SWOG) cooperative group studies from 1993 to 1996 demonstrated that although 63% of the estimated US cancer population were ≥ 65 years of age, only 25% of the patients accrued to SWOG studies were 65 and older. [
] One of the potential reasons for the low relative accrual of older patients was thought to be related to physician concern for higher toxicity rates among experimental chemotherapeutic agents [
Comparative age-based prospective multi-institutional observations of 12,367 patients enrolled to the American college of surgeons oncology group (ACOSOG) Z901101 trials (alliance).
Several recent site-specific (breast, lung) analyses of National Clinical Trials Network (NCTN) studies have demonstrated lower rates of accrual of older patients despite higher incidences of these cancers in older patients in the general population, whereas studies in other disease types (i.e., genitourinary) have not shown the same disparity in accrual by age [
]. Due to the limitations of these previous analyses, it is unclear whether there are certain types of trials that may be at higher risk for under-enrolling older adults.
Because previous analyses have focused on a particular disease site, phase 3 randomized trials, or specific treatment modalities, it has been difficult to model whether certain types of studies are at higher risk for disparate older adult enrollment. Additionally, we felt it was important to evaluate cooperative group data from the Alliance for Clinical Trials in Oncology (‘Alliance’) specifically to help increase older adult accrual in future Alliance trials. Therefore, we examined the differences between the expected accrual percentage of patients ≥ 65 years (as defined by the population-based incidence percentage ≥ 65 years in the Surveillance, Epidemiology, and End Results (SEER) database of the cancer type and enrollment period of each trial) and the actual accrual percentage ≥ 65 in clinical trials sponsored by the Alliance legacy groups (American College of Surgeons Oncology Group [ACOSOG], Cancer and Leukemia Group B [CALGB], and North Central Cancer Treatment Group [NCCTG]). This difference was used to determine the level of disparity of older adult enrollment for each trial. We then analyzed trial-level variables to determine whether certain types of trials were at higher risk of disparate accrual.
2. Materials and methods
2.1 Overview
Patients had completed trial-specific informed consents that were in accordance with local institutional review boards (IRB) as well as federal and local guidelines. For the current analysis, no further IRB review or consent was deemed necessary as the data was de-identified before analysis. This study was approved by the Alliance for Clinical Trials in Oncology Cancer in the Older Adult Committee (Alliance A151736).
2.2 Data source
Actual enrollment percentage data were obtained from the Alliance closed study list, which includes information on >1200 individual studies conducted by the Alliance legacy groups. We included prospective intervention and observational studies without adult age restrictions (i.e. excluded neither older adults [>60] nor younger adults [<60]), that had >50 patients enrolled, and that completed accrual between 01/01/1995 and 12/31/2015. The trials also have to have tumor sites evaluable according to SEER categories (for example trials targeting patients with brain metastases from multiple primary cancer types had no corresponding SEER category and therefore were excluded).
Variables available for analysis in this database included: study lead group (CALGB, ACOSOG, NCCTG); disease group (Breast, Central Nervous System [CNS], Gastrointestinal [GI], Heme/Leukemia/Lymphoma/Transplant, Lung/respiratory, Melanoma, Genitourinary [GU], and Others [Head and Neck Cancers (HNC), Sarcoma, and Multiple]); study phase (2, 3, and Other [observational/pilot/phase 1/phase 1 and 2/pilot and phase 2]); time of enrollment as defined by years open and closed, the era of the close of enrollment (1995–2000, 2001–2005, 2006–2010, 2011–2015); and treatment type (e.g. any radiation, targeted systemic or endocrine therapy, chemotherapy alone). Additional variables included the number of modalities (single vs dual vs trimodality or more); whether the study compared one modality against another (yes or no) (e.g. surgery vs radiotherapy); randomized trial (yes or no); and protocol-dictated stage of eligible patients (metastatic or non-metastatic).
The expected enrollment percentage per trial was obtained using the SEER 18 cancer registries (Surveillance Research Program, National Cancer Institute SEER*Stat software [seer.cancer.gov/seerstat) version 8.3.5. These data, collected by the NCI SEER program from 18 registries throughout the US, represent roughly 28% of the US cancer population [
]. The expected enrollment percentage was obtained by collecting the number of patients ≥65 years and dividing by the total number of patients with the specific cancer type of each trial during the years in which each trial enrolled patients.
2.3 Study endpoint
Our primary endpoint was the enrollment disparity difference (EDD) for each trial [
]. This was calculated by taking the actual enrollment percentage of patients ≥ 65 (from the Alliance database) subtracted from the expected enrollment percentage ≥ 65 (from the SEER database). For example, if a trial performed in breast cancer patients between 1999 and 2005 had an actual enrollment percentage ≥ 65 of 3% (i.e. only 3% of participants in the trial were age 65 or older), but an expected enrollment percentage ≥ 65 in the SEER database of 45%, the EDD would be 42%) Thus, a positive EDD is indicative of higher disparate-accrual of older adults compared to what would be expected from the SEER population of cancer patients.
2.4 Statistical analysis
All analyses were conducted at the trial level (each trial is an observation). Trial factors were summarized by frequency and percentage. Association between trial factors and EDD were evaluated using fixed-effects linear regression. Variables included in the final linear regression model were selected using simple linear regression with EDD as the dependent variable and each trial factor as the only covariate. Variables with an overall association p-value <0.20 were included in the final multivariable linear model. In the final model, all trial factors with overall p-values <0.05 were considered statistically significant. Least square means were estimated from the final multivariable model to better understand each variables' disparity level (comparison EDD to 0). Data collection and statistical analyses were conducted by the Alliance Statistics and Data Center. All statistical analyses were performed using SAS statistical software version 9.4 (Cary, NC).
3. Results
After applying the pre-defined inclusion/exclusion criteria, we identified 237 trials suitable for analysis (see Fig. 1). These trials included 66,708 patients with a median age of 60 years (range 18–102). The characteristics of these trials are listed in Table 1. Two-hundred and four of the 237 studies (86.1%) had an EDD of >0 (the percentage of patients in SEER ≥ 65 with that cancer type during the time the study was enrolling, was greater than the percentage of patients ≥65 in the study). The average EDD between the trials and their individually assigned SEER expected enrollment percentage was 19% (95% CI 17.1–21.3%, p < 0.0001).
Fig. 1Consort diagram of included and excluded trials.
Other (Observational, Pilot, Phase 1, phase 1 and 2, pilot and phase 2)
36 (19, 4, 6, 5, 2)
15.2%
Phase 2
127
53.6%
Phase 3
74
31.2%
Randomized
Yes
99
41.8%
No
138
58.2%
Target stage
Metastatic
123
51.9%
Non-metastatic
114
48.1%
Era study closed to accrual
1995–2000
82
34.6%
2001–2005
89
37.6%
2006–2010
48
20.3%
2011–2015
18
7.6%
ACOSOG: American College of Surgeons Oncology Group, NCCTG: North Central Cancer Treatment Group, CALGB: Cancer and Leukemia Group B. CNS: Central Nervous System, GI: Gastrointestinal. GU: Genitourinary. HNC: Head and Neck Cancer.
Fig. 22A: Actual older adult accrual percentage versus expected percentage by cancer type. 2B: Actual older adult accrual percentage versus expected percentage by number of treatment modalities.
ACOSOG: American College of Surgeons Oncology Group, NCCTG: North Central Cancer Treatment Group, CALGB: Cancer and Leukemia Group B. CNS: Central Nervous System, GI: Gastrointestinal. GU: Genitourinary.
On univariable analysis (Supplemental Table 1) of the trial level variables, the legacy groups (p = 0.170), disease groups (<0.001), number of treatment modalities (p = 0.012), and phase of the trials (p = 0.179) all met the criteria to be included in the final multivariable model. The other variable groups, including the type of modality (p = 0.212), randomized (p = 0.987), target stage (p = 0.420), and era closed to accrual (p = 0.308) did not meet the criteria for inclusion in multivariable analysis. On multivariable analysis, all of the included variable groups demonstrated statistical significance (legacy groups [p = 0.0016], disease groups [p < 0.0001], number of modalities [p = 0.027], and the type of study [phase] [p = 0.0337]) (Table 2). Within each variable group, certain types of trials appeared to have higher levels of disparity when compared to their reference. Specifically, there appeared to be a wide range of EDDs depending upon which disease group the trial was conducted (Fig. 2a). With the exception of melanoma trials, all of the disease types had significantly more disparity than GU trials (Table 2). Studies that included 3 or more modalities (more than 3 modalities may include a targeted therapy with chemoradiation in a neoadjuvant setting), had higher disparate accrual than the studies with 1 or 2 modalities (Fig. 2b and Table 2). Lastly, phase 2 studies appeared to have more disparity in enrollment of older adults than phase 3 studies. The least-square means of the EDDs of the individual variables are represented in Table 2 and Fig. 3. Although the majority of variables had mean EDDs indicative of disparate enrollment, some of the variables appeared to be associated with significantly greater disparity, including trials in patients with CNS, GI, and breast, studies involving 3 or more cancer modalities, and phase 2 studies.
4. Discussion
In this large sample of trials coordinated by the Alliance legacy groups, there was a lower actual enrollment of patients ≥65 to cancer clinical trials than expected enrollment based on the population-based percentage of patients ≥65. However, that disparity was not equal across all types of trials. Non-GU trials, trials that tested trimodality therapy, and phase 2 trials all appeared to be associated with significantly higher disparity differences of older adults enrolled. This was true both when comparing these variables to a reference and comparing them to EDD = 0 (least square mean analysis)
The results of this analysis are consistent with several other previous studies demonstrating disparate enrollment of older adults into cancer clinical trials [
]. Many previous studies focused on either disease-specific trials, phase 3 studies alone, or treatment-specific trials, which limited the ability to identify differences in enrollment across disease groups or treatment types. More recently Ludmir et al. identified over 300 randomized controlled trials from the clinicaltrials.gov website and compared the median age of the participants in those trials to the median age of the respective cancer types in SEER. They identified a widening gap in median age over time, and found several variables that were associated with disparate older adult enrollment including industry sponsorship, lung cancer trials, and targeted therapy trials [
]. As opposed to looking at median age, we were able to identify the actual percentage of older adults in each trial, which may be the reason that we did not identify an association between time/era of enrollment and older adult enrollment. Our results are consistent with previous data demonstrating no change in older adult enrollment in National Cancer Institute-sponsored trials over time [
]. Additionally, rather than seeing an association between a specific treatment modality and disparate enrollment, we identified an association between the number of modalities within a study protocol and the latter; suggesting that the intensity of the treatments on a trial may cause increased difficulty with enrolling older adults. Our findings suggest that there may be legitimate reasons that providers do not enroll their older adult patients into clinical trials that include arms with “intensive treatment”. It may be appropriate for certain trials to expect higher enrollment of younger and healthier patients. Additionally, the lack of enrollment of older adults has not changed over time despite the increasing number of older adult patients with cancer and the increasing awareness of the disparate enrollment of these patients. Rather than continuing to try to enroll older adults in all trials, designing age or performance-status appropriate trials for certain cancer types might be a better approach to improve disparities in clinical trial enrollment and ultimately the care of older adults.
The results of this study should inform trial design in the future to improve the generalizability of trial results. For example, from the results of our study, it is clear that age-specific trial design may not be needed for most prostate or renal trials (made up the majority of GU trials in this database), as these trials appear to enroll older adults at similar percentages to their populations in SEER. However, for other disease types such as GI cancers, and especially those that may require trimodality therapy, such as rectal cancer or pancreatic cancer, increased effort to enroll older adults may require age-specific trial design [
] Previous efforts, including a randomized trial of an educational intervention for clinicians, have unfortunately not had a significant impact on improving older adult enrollment [
Improving accrual of older persons to cancer treatment trials: a randomized trial comparing an educational intervention with standard information: CALGB 360001.
]. Yet there continues to be significant interest, and multifaceted efforts are being made to improve trial design and educate researchers/clinicians across the NCTN [
Modernizing clinical trial eligibility criteria: recommendations of the American society of clinical oncology-friends of cancer research organ dysfunction, prior or concurrent malignancy, and comorbidities working group.
] and the US Food and Drug Administration (FDA) (https://www.fda.gov/media/135804/download). Efforts are already underway within the Alliance to prospectively identify proposed trials at risk for older adult enrollment disparities and to offer suggestions for improvement before the trials are initiated. It is hoped that results from this study and others, combined with a multipronged effort to a) improve clinical trial development, within the NCTN and industry (through the efforts of the FDA) and b) educate researchers and patients on barriers to enrollment of older adults, will soon improve older adult enrollment to trials in the US. By enrolling older adults, trials will create more generalizable results that can narrow the evidence gap that currently exists on the best way to treat older adults with cancer.
This study has several limitations. First, because data were not collected on patients who refused enrollment or were not offered enrollment, we utilized population-level data from the SEER database to estimate the percentage of older adults eligible for trials, which is consistent with previous studies [
]. By estimating eligible populations, we may have missed specific details of eligibility including performance status, laboratory values, staging, and other patient details. Without prospective questionnaires of physicians enrolling patients onto specific trials, or of patients offered trial enrollment, it was not possible to identify the specific causes of under-enrollment or if a patient's specific tumor types or stages were directly matched to the trials open. However, previous studies have suggested that when older adults are eligible and are offered trials they often will consent to enrollment. [
] Similarly, trial level variables including exclusion criteria such as performance status, and kidney function were not available in SEER data and may significantly impact the enrollment of older adults. Additionally, although the type of modality of treatment did not meet significance, we did not analyze differences in trials with novel agents (where toxicity is unknown) versus known agents (of which physicians may be more comfortable). Though, that difference may explain why phase II studies had more disparate enrollment than phase III. Lastly, because humans age heterogeneously, dichotomizing age is not ideal in situations in which discovering associations between age and outcomes is the goal. However, when studying disparate enrollment between age groups, using an age cutoff was felt to be appropriate and consistent with previous publications [
Older adults were under-represented in this large sample of trials. Trimodality trials, phase 2 trials, as well as specific disease types were all associated with disparate accrual of older adults onto Alliance clinical trials. To improve the standards of care for older adults with cancer, enrollment into standard-setting trials is needed. Understanding why older adults are under-enrolled in these trials is a necessary first step. The results of this study and others can be used to help design future trials to improve older adult enrollment onto standard-setting trials. Improvement in trial design combined with educational efforts and continued data collection on older adult enrollment should lead to improved enrollment and ultimately improved care for a growing population of older adults with cancer.
The following is the supplementary data related to this article.
Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under the Award Number UG1CA189823 (Alliance for Clinical Trials in Oncology NCORP Grant), U10CA180821, U10CA180882, UG1CA233160, UG1CA233180, UG1CA233253, UG1CA233290, UG1CA233373, P30CA033572; SML is supported by the NCI Cancer Center Support Grant (P30CA008748). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. https://acknowledgments.alliancefound.org.
Author contributions
Study concepts: NAV, JS, RJ, SML. Study design: NAV, TD, DS, JS, JLR. Data Acquisition: NAV, TD, DVW, DS, JML. Quality control of data and algorithms: NAV, TD, DS, JLR, AJ. Data Analysis and Interpretation: NAV, TD, DVW, DS, JLR, Statistical Analysis: TD, DS, JLR. Manuscript Preparation: NAV, TD, DS. Manuscript editing: NAV, TD, DVW, DS, JS, RJ, SML, RAF, JML, HM, HJC, JLR, AJ. Manuscript Review: NAV, TD, DVW, DS, JS, RJ, SML, RAF, JML, HM, HJC, JLR, AJ.
Declaration of Competing Interest
NV reports advisory board member for Concerto Health AI. RJ has stock options as compensation for her advisory board role in Equity Quotient, a company that evaluates culture in health care companies; she has received personal fees from Amgen and Vizient and grants for unrelated work from the National Institutes of Health, the Doris Duke Foundation, the Greenwall Foundation, the Komen Foundation, and Blue Cross Blue Shield of Michigan for the Michigan Radiation Oncology Quality Consortium. She has a contract to conduct an investigator initiated study with Genentech. She has served as an expert witness for Sherinian and Hasso and Dressman Benzinger LaVelle. She is an uncompensated founding member of TIME'S UP Healthcare and a member of the Board of Directors of ASCO.
Acknowledgments
The authors would like to acknowledge the late Dr. Arti Hurria. As co-chair of the Alliance for Clinical Trials in Oncology Older Adult Committee when this study was reviewed and approved, she was significantly involved in the study's planning and execution. She is missed.
References
Smith B.D.
Smith G.L.
Hurria A.
Hortobagyi G.N.
Buchholz T.A.
Future of cancer incidence in the United States: burdens upon an aging, changing nation.
Racial/ethnic differences in clinical trial enrollment, refusal rates, ineligibility, and reasons for decline among patients at sites in the National Cancer Institute’s Community Cancer Centers Program.
Comparative age-based prospective multi-institutional observations of 12,367 patients enrolled to the American college of surgeons oncology group (ACOSOG) Z901101 trials (alliance).
Improving accrual of older persons to cancer treatment trials: a randomized trial comparing an educational intervention with standard information: CALGB 360001.
Modernizing clinical trial eligibility criteria: recommendations of the American society of clinical oncology-friends of cancer research organ dysfunction, prior or concurrent malignancy, and comorbidities working group.
Overall p-value.
00170-3&id=gr1.jpg){kind=link}
00170-3&id=gr2.jpg){kind=link}
00170-3&id=gr3.jpg){kind=link}