Older Adult Participation in Early-Phase Lung Cancer Clinical Trials, 1998 to 2020

Abstract

Objectives

Despite advances, lung cancer treatment remains associated with substantial toxicity. Early-phase clinical trials inform the safety and efficacy of novel lung cancer treatments. Although older adults represent most patients with lung cancer, and they are underrepresented in phase 3 trials, age disparity in early-phase lung cancer trials is ill-defined.

Methods

We queried ClinicalTrials.gov to identify early-phase interventional clinical trials conducted in adults with lung cancer since database conception. We calculated the difference in age (DA) between the clinical trial populations and U.S. populations, using t test and one-sided analysis of variance to evaluate trial characteristics associated with DA.

Results

We identified 141 clinical trials enrolling 7723 participants from 1998 to 2020. Early-phase lung cancer trial participants were, on average, 7.6 years younger than patients with lung cancer in the U.S. population (mean DA: −7.6 y, SD 4.2). Age disparities were magnified among clinical trials that were industry-sponsored (mean DA −8.5 versus −6.1, p = 0.001) and those limiting eligibility to Eastern Cooperative Oncology Group performance status less than or equal to 1 (mean DA –8.0 versus −6.0, p = 0.040). There was no association between the median age of trial participants and the proportion of patients with serious adverse events.

Conclusions

Older adults remain underrepresented in early-phase lung cancer clinical trials. With the rapid expansion of novel cancer therapies, focused efforts in the design of early-phase trials are warranted to reflect real-world populations. Otherwise, limitations in the generalizability of treatment safety and efficacy may increase in the future.

Introduction

Lung cancer is the leading cause of cancer mortality in the United States and worldwide.^1,2 The past two decades are marked by rapid therapeutic advances spanning systemic therapies, including targeted therapies, immune checkpoint inhibitors, antibody drug conjugates, and novel radiation and procedural techniques.^3,4 Despite these advances, older adults, who represent most patients diagnosed with lung cancer, experience excess morbidity and mortality during lung cancer treatment.5, 6, 7, 8 Several explanations have been postulated for the disparate rate of adverse outcomes among older adults with cancer, including differential rates of drug response and toxicity because of age-related physiological changes, such as the pharmacokinetics of the drug and pharmacodynamic responses to the drug.^9,10 Furthermore, older adults may have comorbid conditions requiring additional medications, which could impact the absorption, efficacy, and toxicity of cancer treatment.^11,12 Clinical trials are critical to establish the efficacy, safety, regulatory approval, and thus, the standard of care of cancer treatments. However, patients in phase 3 lung cancer clinical trials are, on average, nine years younger than patients diagnosed with lung cancer in the general population.¹³ This underrepresentation of older adults in lung cancer trials limits the generalizability of clinical trial results to the real-world setting.

Although several studies have extensively described older adult enrollment in phase 3 cancer clinical trials, the representation of older adults among early-phase (i.e., phase 1 and 2) lung cancer trials is not known.13, 14, 15 Early-phase clinical trials establish the safety, exposure, and response to inform the study design and dose selection of later-phase studies.⁹ If novel cancer therapeutics are developed in highly selected populations, the selected dose and concomitant medication interactions may not reflect the real-world setting, therefore, compounding the problem of generalizability as drugs advance through the development process.¹⁶ In recognition of the public health importance of including older adults in cancer clinical trials, the U.S. Food and Drug Administration (FDA) published the “Inclusion of Older Adults in Cancer Clinical Trials: Guidance for Industry” in March 2022, including specific guidance to include older adults in early-phase cancer clinical trials.⁹ In this study, we sought to identify the trial characteristics associated with the greatest age disparities among early-phase lung cancer clinical trials and characterize the trends in trial enrollment and eligibility restrictions over time. In addition, using trial-level aggregate data, we explored the association between median age at enrollment and serious adverse event reporting.

Methods

Data Source

We queried the ClinicalTrials.gov website on September 25, 2023 to identify early-phase interventional clinical trials conducted in adults with lung cancer (NSCLC, SCLC, and lung cancer not further specified) since database conception. The full search terms are available in Supplementary Appendix 1. Briefly, search terms included: “cancer” (condition or disease), “early-phase” and “phase 1” (study phase), “interventional” (study type), and “with results” (study results). This initial search yielded 2739 trials, which we then screened to identify cancer-specific, early-phase (i.e., phase 1 and phase 2), therapeutic trials in NSCLC or SCLC (Fig. 1). Trials done only in lung cancer were included. Clinical trials were excluded if they were restricted to the pediatric population or if they did not report the median age of included adult participants. Clinical trials in cancer-related symptoms, diagnosis, screening, prevention, or biomarker-only trials were all excluded. Two physicians independently reviewed the search results to identify eligible clinical trials. Discrepancies were adjudicated by means of group discussion.

Figure 1.

Schema of included trials.

Measures

We extracted trial characteristics from the ClinicalTrials.gov website. Our key trial characteristic of interest was the median age of enrolled participants. For each clinical trial that reported the median age of enrolled participants, we calculated the difference in age (DA) between patients in the clinical trial population and the U.S. population diagnosed with lung and bronchus cancer, as reported by the Surveillance, Epidemiology, and End Results (SEER) program. For example, in 2006, the median age of lung cancer diagnosis in the United States was 70 years.^1,17 If a clinical trial opened for enrollment in 2006, and the reported median age of enrolled participants in the trial was 60 years, then the DA was calculated to be –10. Although the median age at lung cancer diagnosis varies globally,¹⁸ we chose to use the SEER-reported median age of lung cancer diagnosis as most of the trials were conducted in the United States or Canada, and several trials were conducted across multiple continents (Supplementary Appendix 2).

Additional trial characteristics were collected, including the year the clinical trial opened for enrollment, the size of the trial according to the number of participants enrolled, lung cancer disease type, treatment modality, industry sponsorship, and proportion of reported serious adverse events.

We also captured eligibility requirements for trial enrollment, including upper age (i.e., $\le$ 90 y versus no upper age limit), molecular status (e.g., EGFR-mutated disease only), and performance status (i.e., Eastern Cooperative Oncology Group [ECOG] $\le$ 1 versus ECOG 2–3). The thresholds for upper age and ECOG performance status categorization were selected a priori on the basis of the U.S. FDA 2024 Draft Guidance for cancer clinical trial eligibility¹⁹ and were supported by previous studies using the same definitions.20, 21, 22 Two physicians independently performed the trial characteristic extraction, and discrepancies were adjudicated by means of group discussion.

Statistical Analyses

We calculated the DA as a continuous variable between the trial population and the U.S. population for each clinical trial as defined above. Descriptive statistics summarized trial characteristics and inclusion criteria. We used t test for characteristics with two variables (e.g., industry-sponsored yes or no) and one-sided analysis of variance for characteristics with at least three variables (e.g., disease type) to evaluate trial characteristics associated with DA. Because the median age of patients diagnosed with lung cancer harboring a potentially targetable molecular alteration is lower than patients with tumors without a potentially targetable molecular alteration,²³ we performed a sensitivity analysis in which we repeated the evaluation of lung cancer trial characteristics associated with age disparities among only those trials with no molecular restriction.

To depict chronological trends in lung cancer clinical trial enrollment according to age, we constructed scatter plots of the median age of trial participants and the size of the trial (i.e., number of participants enrolled) according to the year of trial enrollment. In addition, we plotted the percentage of trials with performance status and upper age eligibility requirements over time to depict chronological trends in trial enrollment restriction. To exploratively correlate the median age of enrolled participants with the proportion of reported serious adverse events per patient reported, we calculated the Pearson correlation coefficient and depicted the association with scatterplots. Given concerns regarding possible skew in proportion of enrolled participants with serious adverse events reported among the smaller trials, we a priori chose to separately calculate the Pearson correlation coefficient and construct scatterplots for trials for medium to large trials (i.e., $\ge$ 20 enrolled participants).

All statistical analyses were conducted using STATA/SE 17 (Stata/SE [Special Edition], StataCorp, College Station, TX).²⁴ A p value of less than 0.05 was considered statistically significant. Because all data were aggregate trial-level and publicly available, informed consent was not obtained, and institutional review board approval was not required.

Results

Trial Characteristics and Age Disparities

Of the 2739 clinical trials identified by our search parameters, 141 early-phase lung cancer clinical trials met our inclusion criteria and were included in the final analysis (Fig. 1). In total, 7723 participants were enrolled across the 141 trials with enrollment start dates between 1998 and 2020. The trial characteristics are detailed in Table 1, and the countries of enrollment are in Supplementary Appendix 2. Most trials included participants with NSCLC (n = 114, 81%), were trials that involved systemic treatment only (n = 113, 80%), and were industry-sponsored (n = 86, 61%). Only 5% of trials (n = 7) included an upper age eligibility requirement ($\le$ 90 y), but most trials (n = 90, 64%) limited eligibility to ECOG performance status less than or equal to 1. A minority of trials (n = 29, 21%) included molecular restriction.

Table 1.

Early-Phase Lung Cancer Trial Characteristics Associated With Age Disparities

Trial Characteristics	N (%)	Mean Difference in Agea (SD), y	p Value
Total trials, N	141	−7.6 (4.2)	--
Size of trial, number of participants
Small (<20)	49 (33)	−7.3 (4.5)	0.826
Medium (20–99)	74 (52)	−7.6 (3.7)
Large (≥100)	20 (14)	−8.0 (5.1)
Disease type
NSCLC	114 (81)	−7.8 (4.2)	0.174
SCLC	13 (9.2)	−7.9 (2.7)
Lung cancer, not further specified	14 (9.9)	−5.6 (4.7)
Treatment modality
Systemic treatment only	113 (80)	−8.4 (3.5)	<0.001
Radiation ± systemic treatment	25 (18)	−3.5 (4.5)
Procedure ± systemic treatment	3 (2.1)	−10.0 (3.5)
Industry sponsored
Yes	86 (61)	−8.5 (3.6)	0.001
No	55 (39)	−6.1 (4.6)
Age eligibility
Age ≤90 y	7 (5.0)	−9.6 (3.8)	0.183
No upper age limit	134 (95)	−7.4 (4.2)
Molecular eligibility
Molecular restriction	29 (21)	−9.5 (4.0)	0.005
No molecular restriction	112 (79)	−7.1 (4.1)
Performance status eligibility
ECOG $\le$ 1	90 (64)	−8.0 (3.4)	0.040
ECOG 2–3	36 (26)	−6.0 (5.3)
Missing	15 (11)	−8.4 (4.6)

ECOG, Eastern Cooperative Oncology Group; N, number.

^aThe mean difference in age represents the average difference in age between patients in the clinical trial population and the U.S. population, as reported by the Surveillance, Epidemiology, and End Results Program.

Overall, patients enrolled in early-phase lung cancer clinical trials were on average 7.6 years younger than those in the U.S. population diagnosed with lung cancer (mean DA −7.6 years, SD 4.2). Trials that evaluated systemic treatment only or a procedure, were industry-sponsored, included a molecular restriction, or limited performance status to those with ECOG less than or equal to 1 were associated with greater DA (Table 1). On average, clinical trials with ECOG performance status restriction less than or equal to 1 had a mean DA of −8.0 years (SD 3.4), which was significantly greater than trials that allowed ECOG performance status 2 to 3 (DA −6.0 y, SD 5.3, p = 0.040). Of note, we did not detect a significant difference in DA between clinical trials that restricted to age 90 years or younger and those that had no upper age limit.

Given the younger age of lung cancer diagnosis among people with NSCLC that harbors a driver mutation, we performed a sensitivity analysis to evaluate the age disparities after excluding trials that had a molecular restriction (Supplementary Appendix 3). After molecular restriction exclusion, treatment modality (p < 0.001), industry sponsorship (p = 0.001), and eligibility restricting to ECOG performance status restriction of less than or equal to (p = 0.001) remained significantly associated with greater differences in mean DA.

Trends in Trial Enrollment and Eligibility Requirements

From 1998 to 2020, the median age of participants in early-phase lung cancer trials has been lower than the median age of lung cancer diagnosis in the U.S. population (Fig. 2). The DA has numerically widened since 2016, which was when the median age of lung cancer diagnosis in the United States increased. As illustrated in Figure 2, there were five clinical trials (3.5% of the trials included in our analysis) in which the median age of participants was at or above the median age of lung cancer diagnosis in the United States. All five of the trials allowed ECOG performance status of 3 or less, did not have molecular restriction, and were not industry-sponsored. Most (n = 4, 80%) were radiation trials; one clinical trial was systemic therapy only and evaluated a combination of an immune checkpoint inhibitor and an investigational small molecule. On average, the percentage of early-phase lung cancer clinical trials that limited eligibility to patients with ECOG performance status less than or equal to 1 increased from 1998 to 2020 (Fig. 3). Clinical trials that limited eligibility to those with upper limit of age of 90 years or younger were rare, first being noted in 2009, followed by 2017, and 2019.

Figure 2.

Trends in early-phase lung cancer clinical trial enrollment according to age. The median age of lung cancer diagnosis in the United States (dashed orange) is illustrated according to the Surveillance, Epidemiology, and End Results database.

Figure 3.

Early-phase lung cancer clinical trial eligibility requirements for performance status and upper age over time. ECOG, Eastern Cooperative Oncology Group.

Association Between Median Age of Trial Participants and Reported Serious Adverse Events

Of the 141 trials included in the analysis, 140 trials reported serious adverse events. Across these 140 trials, the mean proportion (SD) of serious adverse events was 0.39 (0.22). We did not identify evidence of correlation between the median age of enrolled trial participants and the proportion of patients with serious adverse events using aggregate trial data (Supplementary Appendices 4–5). The median age of trial participants had a null association with the proportion of serious adverse events among all trials (Pearson r = −0.072) and those with 20 or more patients (Pearson r = −0.213).

Discussion

This study, to our best knowledge, is the first to characterize the age disparity among patients enrolled in early-phase lung cancer clinical trials. The median age of participants in early-phase lung cancer clinical trials was, on average, 7.6 years younger than the median age of lung cancer diagnosis in the United States. Clinical trials that were systemic treatment only, procedural interventions, industry-sponsored, included a molecular status restriction, or limited eligibility to those with ECOG performance status less than or equal to 1 were associated with the greatest age disparities. The age disparity among early-phase lung cancer clinical trials is not improving over time, nor is the proportion of trials limiting eligibility on the basis of performance status less than or equal to 1 or upper age of 90 years or younger. In an exploratory analysis, we did not identify an association between the median age of enrolled participants and the proportion of reported serious adverse events.

The trial characteristics we identified as associated with the greatest age disparities among early-phase lung cancer trials generally align with previously reported factors associated with age disparities among phase 3 cancer trials, namely industry sponsorship and eligibility restriction on the basis of molecular status or performance status.^13,25 However, in contrast to existing literature,¹³ only 5% of the early-phase lung cancer clinical trials we evaluated included an upper age restriction as part of the eligibility criteria, and this upper age restriction was not associated with an age disparity between trial participants and the general population diagnosed with lung cancer. Rather, the ECOG performance status less than or equal to 1 restriction may be inadvertently excluding older adults.^14,21,26 This is partially because of the inherent limitations of performance status assessments among older adults. For example, Broderick et al.²⁷ reported that clinicians assigned older adults a numerically higher ECOG performance status score than their younger counterparts, even when no difference was detected in objectively measured physical activity. Performance status restriction in clinical trial design was first identified in 2003 as a key contributing factor to the underrepresentation of older adults among cancer clinical trials.²⁶ Since then, despite recognition of the need to re-evaluate trial design to better capture patients representative of the real-world setting,²⁸ ECOG performance status restriction of less than or equal to 1 remains pervasive, and restriction rates did not improve over time, as we noted here and as what the others have previously reported.²¹

Our finding that the ECOG performance status of less than or equal to 1 restriction, rather than the upper age limit, is associated with age disparities among early-phase lung cancer trial participants has important implications for clinical trial design. Recently, the U.S. FDA published guidance to increase older adult representation in clinical trials⁹ and a separate draft guidance for broadening clinical trial eligibility on the basis of performance status.²⁹ Particularly for older adults, the U.S. FDA recommends assessment of patients’ overall health status when screening for trials, as existing performance status scales are suboptimal for this population.^8,30 This can be achieved by conducting a geriatric assessment, a multidimensional, holistic evaluation of patients’ medical, psychosocial, and physical functioning,^31,32 which is recommended by several professional societies, including the American Society of Clinical Oncology.³³ A clinical trial that incorporates a performance status criterion on the basis of an available geriatric assessment tool could identify potentially eligible older adults who may have been otherwise excluded if using a performance status scale alone. In addition, including patient advocates and clinicians with geriatric expertise early in the trial design process can help ensure an inclusive trial design.²⁴

Because early-phase clinical trials establish the safety of novel therapeutic approaches, they have unique design considerations that distinguish them from later-phase clinical trials. There is a theoretical concern that including older adults may compromise trial outcomes by increasing adverse events and decreasing efficacy. In the present study, we did not identify evidence of correlation between the median age of enrolled trial participants and the proportion of patients with serious adverse events using aggregate trial data. Likewise, Nicolò et al.¹⁶ evaluated safety outcomes of older versus younger adults enrolled in early-phase clinical trials at their institution and found similar rates of grade 3 or higher adverse events across the age groups (33% versus 31%, p = 0.7). Several studies have evaluated efficacy outcomes among older adults in early-phase clinical trials and reported similar rates of clinical benefit across the age groups.^16,34,35 The U.S. FDA proposed several strategies to design clinical trials that promote enrollment of underrepresented populations without compromising safety or efficacy outcomes. These include designing trials with a prespecified older adult or limited performance status cohort that would be smaller in size and exploratory in nature, with possible different doses, and early safety review.²⁹ Other recommendations include leveraging an adaptive clinical trial design, which would allow for prespecified trial design changes during the trial when data become available, including altering the trial population. For example, a trial could begin with a narrow population to establish safety and later expand to a broader population on the basis of interim safety data.³⁶ Using these approaches when designing early-phase clinical trials could increase enrollment of patients reflective of the real-world setting.

Beyond clinical trial design, underrepresentation in clinical trials is highly complex and includes patient, provider, and system factors. Several efforts have been made to improve enrollment of older adults in phase 3 cancer clinical trials, including patient-facing educational materials and decision aids, lay patient navigators, telehealth to reduce the number of in-person visits required for a study, and engagement with caregivers.^9,15 In addition, clinicians play a critical role in offering and referring older adults for clinical trials. Sedrak et al.²⁵ reported that, when screened and offered a clinical trial, enrollment rates among patients with cancer in community settings were similar, independent of age, and older adults were as willing to participate as their younger counterparts. Similar findings were reported for older adult enrollment in early-phase clinical trials in France.³⁷ It is likely that a combined approach of both patient-facing initiatives and intentional design of clinical trials would be needed to improve the inclusion of older adults in early-phase clinical trials.

We acknowledge several limitations of this study. First, the DA was calculated on the basis of U.S. demographic data from SEER. We chose to use SEER data as our standard because some of the trials were missing country data, 18% of the trials were conducted across multiple continents, and most trials (53%) were conducted in the United States or Canada. Of the 141 trials included in our study, 11 (8%) were conducted solely in countries with a median age of lung cancer diagnosis younger than the U.S. median, namely Spain, Germany, Austria, and the People's Republic of China.¹⁸ On the other hand, 13 (9%) trials were conducted solely in countries with a median age of lung cancer diagnosis older than the U.S. median, namely Japan, South Korea, and the United Kingdom.¹⁸ Therefore, we believe the U.S. median age of lung cancer diagnosis represents a reasonable standard to approximate the median age of lung cancer diagnosis in the trials we included. Second, the DA was calculated on the basis of the median age of lung cancer diagnosis, not on the median age of individuals potentially eligible for NSCLC treatment. In most clinical scenarios, early-phase clinical trials are considered after disease progression on standard of care treatment. In real-world settings, there is conflicting data regarding whether older adults are less or as likely as younger patients to receive second-line and beyond NSCLC treatments.^38,39 Therefore, it is possible that the difference in age between participants in early-phase lung cancer clinical trials and those who are eligible for subsequent lines of NSCLC treatment is lower than the 7.6 years DA we observed in this study. Third, we did not capture enrollment data for specific older adult subgroups (i.e., age $\ge$ 75 y) that have been previously recognized as an important underrepresented population.⁹ Fourth, our study was restricted to evaluating trial-level factors associated with DA and does not capture the full complexity of barriers to trial enrollment. Fifth, we did not capture additional eligibility data on comorbid conditions, concomitant medications, or organ function requirements that may otherwise exclude some older adults. Sixth, because our search criteria were restricted to only those clinical trials with results, we were unable to include clinical trials with more recent enrollment start dates (i.e., 2021 and onward). Finally, the limited number of trials that met our inclusion criteria precluded our ability to perform multivariable analyses evaluating the association between trial characteristics and DA.

The results of this study suggest that older adult underrepresentation in early-phase lung cancer clinical trials is widespread and not improving over time. Removal of upper age restriction in clinical trial eligibility is unlikely to improve enrollment of older adults, and rather, clinical trials that include dedicated limited performance status cohorts could help promote older adult enrollment. Older adult representation in early-phase clinical trials is critical to obtain necessary data on safety, exposure, and response to best inform later-phase clinical trials.

Disclosure

Dr. Schokrpur has received consultant/advisory board fees from OncoHost and Navya Network outside the submitted work. Dr. Gandara has received research funding to institution from Amgen, Astex, Genentech, and Razor Genetics; has served as consultant with fees paid to institution for Adagene, AstraZeneca, IO Biotech, Guardant Health, and OncoHost; and has served as consultant/advisory board for Roche Genentech, Merck, Novartis, Boehringer Ingelheim, Regeneron, Revolution Medicine, and Sanofi. Dr. Riess has received research funding to institution from AstraZeneca, Novartis, Merck, Nuvalent, IO Biotech, ArriVent, Revolution Medicines, Summit, and Boehringer Ingelheim; and consultant/advisory board fees from Bristol-Myers Squibb, Daiichi Sankyo, Boehringer Ingelheim, Janssen, Pfizer, OncoHost, GlaxoSmithKline, Catalyst, and Amgen. Dr. Singhal has received consultant/advisory board fees from OncoHost, Bristol-Myers Squibb, and Caris Life Sciences. The remaining authors declare no conflict of interest.