Simplified frailty assessment tools: are we really capturing frailty or something else?

To the Editor:

In their recent paper “A simplified frailty scale predicts outcomes in transplant-ineligible patients with newly diagnosed multiple myeloma treated in the FIRST (MM-020) trial” [1], Facon et al. propose a simple, dichotomous, easy-to-use frailty score to predict clinical outcomes among transplant-ineligible older adults with multiple myeloma (MM) [1]. Using data from the FIRST trial, the authors classified patients into frail/nonfrail based on age, comorbidities, and Eastern Cooperative Oncology Group performance status (ECOG PS). Patients categorized as frail had significantly worse progression-free survival and overall survival (OS), as well as higher rates of grade 3/4 treatment-related toxicities [1]. We congratulate the authors for development of a simple, practical tool that uses routinely measured variables to improve treatment decision-making in older adults with myeloma. However, we caution readers against abandoning more established frailty assessment tools, providing a brief overview of relevant literature to date.

MM is a disease of aging. Because older patients are at an increased risk of treatment-related toxicity, there is a need to personalize treatment strategies based on their anticipated tolerance of treatment [2]. However, chronological age does not adequately capture this vulnerability due to the heterogeneity of patients in physical and psychological functioning.

Frailty is an aging-associated state of cumulative decline in many physiological systems, resulting in diminished resistance to stressors, such as cancer treatment [3]. Despite being regarded as relevant to the care of older adults, operationalizing a precise definition of frailty remains challenging [4]. A recent review identified 67 separate frailty measures in the literature, yet no consensus has been reached on how to best measure frailty. Two of the most commonly cited approaches are the Fried phenotypic frailty approach and the Rockwood accumulation of deficits approach [5]. In their seminal paper, Fried et al. [2] defined frailty as a clinical syndrome in which three or more of the following criteria was present: unintentional weight loss (10 lbs in past year), self-reported exhaustion, weakness (grip strength), slow walking speed, and low physical activity. Frailty independently predicted falls, disability, hospitalization, and death over a 3-year period in a cohort of older adults [2]. Meanwhile, Rockwood et al. operationalized frailty as an accumulation of deficits, where an individual’s health status can be quantified as a proportion of aging-associated deficits, using measures from comprehensive geriatric assessment (GA) [3].

With increasing recognition of the risk of cancer treatment in older adults, there has been interest in measuring frailty among older adults with cancer. Hurria et al. first proposed a cancer-specific GA tool to predict the risk of chemotherapy toxicity among older adults with cancer [6].Additional studies led to development of 51-item deficit accumulation frailty index and 36-item Carolina frailty index, to predict survival among older adults with cancer[6]. However, most of these studies included patients with solid tumors receiving chemotherapy and a similar tool among patients with MM (often treated with novel drugs rather conventional chemotherapy) was notably lacking.

To address the above gaps, in 2014, the International Myeloma Working Group (IMWG) introduced a frailty score based on age, comorbidities, and functional status and showed that frail patients (defined by a frailty score of 2 or higher) had a significantly inferior survival, greater treatment-related toxicity, and discontinuation of therapy[7]. The German group proposed an alternative risk assessment tool, the revised Myeloma Comorbidity Index (R-MCI), incorporating organ function (renal, lung), performance status, age, and Fried frailty, and showed that the R-MCI outperformed other commonly used comorbidity indices to predict OS [8].

However, perceived complexity or concerns about the time required for gathering the data for these frailty models (e.g., 14 questions for ADLs/IADLs for IMWG and the challenges of defining frailty as required by R-MCI) have led to limited utilization of these above tools and proposals for newer clinical prediction models using readily available information. In this regard, the current study by Facon et al. reflects an important effort to risk stratify older adults with newly diagnosed MM [1]. Since age, comorbidity burden, and ECOG performance status are routinely measured in clinical practice, the proposed tool is also highly feasible, easy to implement, and potentially retrospectively applicable as comorbidities and PS are typically available in medical records, while activities of daily living/instrumental activities of daily living are not.

Despite being a strong predictor of outcomes, we would like to stress that this score does not necessarily reflect frailty. Frailty is a multidimensional concept including physical, psychological, and social constructs, best captured in the context of a GA [3]. ECOG performance status is highly subjective with a marked inter-observer variability, influenced by disease associated conditions, and does not reliably capture frailty [9]. GA can identify deficits in more than 50% of older patients with cancer deemed to have a normal PS [9]. Furthermore, objective measures such as grip strength and gait speed, sarcopenia, and aging biomarkers such as p16 may predict outcomes among older adults with hematologic malignancies independent of ECOG PS [10]. Adding ECOG PS to existing frailty scores also does not improve their predictive performance [7]. Clinical trials tend to select patients with lower ECOG scores; almost 80% of patients in the FIRST trial had an ECOG score of 0 or 1. The external validity of this frailty scale and simultaneous comparison to other frailty measures (e.g., IMWG frailty score, R-MCI) should be measured using real-life patient-level data with varying age and performance status. Lastly, operationalizing frailty using a one-time dichotomous measure ignores the dynamic and heterogeneous nature of this construct.

To summarize, although routinely obtained clinical measures such as age and Karnofsky or ECOG PS can predict outcome, it does not adequately encompass the heterogeneous vulnerabilities in older adults compared with GA and other frailty measures in older adults with MM. We worry that the proposed tool may endorse a step back toward arguably oversimplified measures, and encourage continued exclusion of older adults in MM clinical trials. Although we appreciate efforts to promote personalization of treatment decision-making based on aging-associated vulnerability while increasing the efficiency of such assessments, researchers and clinicians must appreciate that indeed frailty comprises more than just age and PS; more comprehensive assessments such as GA will remain vital in optimizing management for this extremely heterogeneous patient population. Additional studies are needed to further refine and simplify the existing frailty assessment tools so as to promote their widespread implementation in routine clinical care.