Cores for Geriatric Oncology Infrastructure in the Cancer and Aging Research Group: Biostatistics, Epidemiology, and Research Design (The Analytics Core)

Introduction

The Cancer and Aging Research Group (CARG) has received support from a National Institute of Health/National Institute on Aging (NIH/NIA) R21/R33 grant to develop a national infrastructure for geriatric oncology research through dedicated Cores (Table 1). Here, we summarize recommendations from the inaugural R21/R33 conference, held in October 2018 at City of Hope for the Biostatistics, Epidemiology, and Research Design (“Analytics”) Core. The mission of the Analytics Core is to serve as a central resource of expertise and counselling in epidemiology, biostatistics, and research design to advance cancer and aging research nationally. Similar to statistical analysis units embedded in existing aging-focused, local infrastructures, such as the Claude D. Pepper Older American Independence Centers or the Nathan Shock Centers of Excellence in the Basic Biology of Aging, the Analytics Core will provide national statistical design support to accelerate high-quality research at the aging and cancer interface.^1,2 The Analytics Core will provide these services to investigators at all stages of the research process. This infrastructure will enable the continuation and advancement of the late Dr. Arti Hurria’s impactful work in geriatric oncology and support the currently geographically disparate but deeply committed group of interdisciplinary CARG investigators. To our knowledge, the R21/R33 is the first organized effort in the US that will bring together aging and cancer experts, technical analytical expertise, stakeholder engagement, and a coordinated approach to foster and accelerate interdisciplinary, collaborative research in geriatric oncology.

Table 1.

Dedicated Cores to develop a national infrastructure for geriatric oncology research as supported by a National Institute of Health/National Institute on Aging (NIH/NIA) R21/R33 grant

Core 1	Leadership Core	Leadership, Mentorship, and Training
Core 2	Aging Measures Core	Clinical and Biological Measures of Aging
Core 3	Supportive Care Core	Behavioral, Psychological, & Supportive Care Intervention
Core 4	Health Services Core	Health Services Research
Core 5	Analytics Core	Epidemiology, Biostatistics, & Research Design
Core 6	Communication Core	Dissemination and Communication

Rationale for Analytics Core

To facilitate the conduct of high quality, high-impact, clinically meaningful research that propels the field of aging and cancer, the Analytics Core will be a critical resource for investigators. In particular, this Core will serve to help investigators generate scientifically sound hypotheses, identify the right datasets (if existing) or alternatives, use the appropriate methods, collect accurate exposure and outcome data, and interpret/analyze the results. Given the newness of the field of aging and cancer research, investigators are often early in their career trajectories, dispersed across the country, isolated within their institutions, and under-supported by local resources and mentorship.³ There is often limited availability in their institutions of statistical resources, expertise, and data-analytics specific to aging-related cancer research. Thus, the field of geriatric oncology requires a nationally coordinated effort to grow.¹ As it evolves and grows, both in size and sophistication, there is a clear unmet need for nationally organized efforts to ensure that studies of cancer and aging are scientifically rigorous.⁴ The Analytics Core will fill a gap by providing a central resource of statistical and trial design support to investigators studying older adults with cancer.

Core Functions

The Analytics Core will provide several services, including: the provision of expert consultations, identification of appropriate datasets, direction to analytic resources, and the fostering of appropriate coordination with the other Cores.

First, as a consultant to researchers, the Analytics Core will make aging-specific recommendations on study design, dataset selection, sample size/power calculations, and statistical analysis plans. This is vital given the complexity related to designing studies of older and/or frailer adults with cancer. For example, this Core will provide guidance on novel design approaches, as identified during a U13 conference,⁵ including: single-arm trials which incorporate novel end points such as functional changes and age-related pharmacokinetic changes; randomized controlled trials specific for older patients with cancer or including all ages but stratifying recruitment according to the age distribution of the disease; extended trials to obtain data regarding new gold standards within the older population; embedded study measures, geriatric assessment measures, and biomarkers of aging placed within the infrastructure of the parent study; and prospective cohort studies to assess treatment received as standard of care to evaluate geriatric oncology outcomes. Additionally, there are a number of important, and often neglected, statistical and design issues in studying older adults with cancer that should be kept in mind. These include outcome choices most relevant to older adults (e.g. toxicity, function, quality of life),^5–11 inclusion of those most likely to have the disease (e.g. pragmatic trial designs),⁶ and sample size/power issues for inclusion of sufficient numbers of older adults.⁵

Second, this Coreserves as a valuable resource linking investigators with datasets (e.g., Cancer and Aging Group [CARG] Chemotoxicity,^12–13 Improving Communication in Older Cancer Patients and Their Caregivers [COACH],¹⁴ Specialized Oncology Care & Research in the Elderly [SOCARE],¹⁵ Women’s Health Initiative’s Life and Longevity After Cancer [LILAC]) for secondary data analysis,¹⁶ study design for pilot data, a toolbox of methods and analytical plans for utilizing such data, and informatics guidance to best use the data. These national datasets can serve as preliminary data for analyses by junior investigators applying for larger grants.^17,18 Secondary analyses of existing datasets are often a cost-effective means to explore new hypotheses on issues including biological and clinical changes affecting health across the life span, on cancer and toxicity at older ages, and on outcomes from specific interventions. These studies can then inform the design of tuture studies in aging and cancer research (Table 2).

Table 2.

Examples of Existing Data for Geriatric Oncology Studies Available for Preliminary Studies

Dataset	Studies	Data	Examples of Previous Collaborative Efforts
Cancer and Aging Group [CARG] Chemotoxicity	Developing and validating a chemotherapy toxicity prediction tool (n=750 patients accmed)	Geriatric assessment data; toxicity and health care utilization outcomes; chemotherapy decisions	• Has led to 11 manuscripts by CARG investigators • Data used to help power Dr. Mohile’s intervention studies below
Improving Communication in Older Cancer Patients and Their Caregivers [COACH]	Evaluating if geriatric assessment improves outcomes of older patients receiving cancer treatment (n>1000 patients, 400 caregivers, and 300 oncologists)	Geriatric assessment data; toxicity outcomes; communication outcomes; audio recordings of clinical encounters; caregiver and oncologists characteristics	• Plan for data to be shared with CARG investigators when mature
Specialized Oncology Care & Research in the Elderly [SOCARE]	Registry (n>1000 patients who underwent geriatric oncology evaluation at 2 institutions)	Geriatric assessment data linked to medical records	• Data has supported Dr. Mohile’s studies • Several CARG investigators and mentees have used data to support grant applications

Third, the Analytics Core provides an opportunity for collaborations, working closely with other Cores to coordinate design. For example, this Core can collaborate closely with the Aging Measures Core to provide resources for geriatric assessment scoring and analysis; or the Supportive Care Core to provide expertise in design of behavioral, psychological, and supportive care interventions.

Finally, to ensure that these functions are successful, the Analytics Core is organized and led with diverse expertise. Specifically, the Core is led by Can-Lan Sun, PhD, Associate Research Professor at City of Hope. Dr. Sun, has over 20 years of experience conducting and managing epidemiological and clinical studies. She has worked closely with Dr. Hurria and investigators from CARG since 2015, providing study design and statistical expertise for the group. She is the Statistical Lead for the R21/R33 grant and the lead biostatistician at the Center for Cancer and Aging at City of Hope, currently directed by aging researcher and co-PI of the grant, Dr. William Dale. Additional members bring diverse knowledge, expertise, and experience in fields such as, analysis of large secondary databases; survey development and qualitative methods; behavioral and supportive care intervention research methods; and design of therapeutic clinical trials for older and/or frail adults with cancer.

Proposed Workflows

The proposed work flow, both within and across the Cores, was discussed at the inaugural R21/R33 meeting. One possible flow is that new investigators will bring new concepts to the Leadership Core (Figure 1). Members of the Leadership Core will send an email linking members from the aging assessment (Aging Measure) Core and/or an interventions (Health Services or Supportive Care) Core with the new investigator if there is a need for aging assessment and/or intervention expertise, depending on the specific aims of the proposed research concept. Study proposals are then reviewed for input on research methodology by the Analytics Core, which may provide recommendations about study design, sample size/power calculations, and statistical analysis. Members of this Core may refer the investigator back to an intervention core, such as the Health Services Core, which may have more extensive experience in study design and analysis specific to health services research. Alternatively, if the study proposal utilizes the geriatric assessment, ^19–21 members from the Analytics Core may re-refer the investigator back to the Aging Measure Core to refine statistical analysis plan for aging assessment as an endpoint, provide more guidance on analysis of these assessment scores, and assist with interpretation of results. Finally, the Communications Core can serve as a resource for effective communication and dissemination of research findings. The ultimate goal is to utilize the R21/R33 infrastructure and expertise, in forms of the dedicated Cores, to facilitate the design of high-quality, high-priority cancer and aging research. We envision an increase in the number of research projects that lays the foundation for competitive multi-site studies and peer-reviewed grants, propelling the national research agenda and leading to clinically meaningful changes for patients.

Fig. 1.

Example of a proposed workflow that utilizes the R21/R33 infrastructure and expertise, in forms of the Cores, to facilitate the design of impactful cancer and aging research.

Proposed guidelines, policies and procedures

Proposed guidelines, policies and procedures to facilitate the effectiveness of the Analytics Core include: 1) having a target audience of junior and senior investigators; 2) requiring institutions to provide funding matches for services (hourly charges); 3) identifying resources that can be utilized at the investigator’s institution; 4) and requiring that the initial consultation by junior investigators for Core services involve their mentors. The Analytics Core is primarily designed to help junior investigators and those new to aging research. This Core will also support senior investigators with specific aging-related questions, such as parameter estimates, power calculations, and statistical support for grant applications. We have proposed a policy to require matching funds from the investigator’s institution or individual grant (e.g., a K-award) to help sustain the analytic expertise. This Core is designed to provide support that is not available to the investigator at their own institution. We will perform an early exploration of their needs with investigators in order to identify and engage available institutional resources (such as biostatistical support) prior to utilizing the Analytics Core. For example, investigators who do not have access to statistical services at their institution will be provided with initial number of hours of support from the Analytics Core. Members of Analytics will help investigators complete a template requesting analytic support to ensure that it is accurate and appropriate.

Interactions with other cores

The Analytics Core will engage and collaborate closely with other Coregroups to support investigators’ external projects. We will be providing high-quality statistical services and resources. In particular, this Core will connect closely with the Aging Core, to provide recommendations incorporating geriatric assessment in study designs and analyses.

Sustainability

Specific strategies for sustainability after the R21/R33 grant of the Analytics Core within the infrastructure will be informed by experiences during the grant period and ideas such as development of a “charge-back” system, rotating staff, recruitment of volunteers and future grants. The Analytics Core can be supported in part by matching funds from home institutions as investigators apply for pilot grant funding under the R21/R33. Because of limited funding, the current Analytics Core can vet and provide recommendations to investigators for developing statistical plans in grant applications and manuscripts, but not help investigators write statistical analysis plans in their entirety. For sustainability, we propose establishing a charge-back system with hourly charges on services. Investigators will be charged based on the amount of time the Core invests in supporting biostatistical design and analyses. Additionally, this Core can recruit volunteers, or have the staff on rotating services.

Conclusion

The Analytics Core will provide a hub for investigators across the nation to receive statistical and research design support while conducting their research at the interface of aging and cancer. It is designed to fill the gaps at a national level for investigators who are isolated and under-supported at their local institutions. This is the first infrastructure project, to our knowledge, that will grow the geriatric oncology field by fostering high-impact, innovative aging and cancer research nationally and, ultimately, carry forward Dr. Hurria’s legacy to improve the lives of older adults with cancer.