Frailty in Patients With Hematologic Malignancies and Patients Undergoing Hematopoietic Stem Cell Transplantation: A Systematic Review.

1
Introduction
The aging population of patients with hematologic malignancies, eligible for hematopoietic stem cell transplantation (HSCT), has a high risk of frailty [1, 2]. Frailty is characterized by a decline in age‐related multiple physiological functioning, resulting in increased vulnerability and reduced ability to withstand acute stressors, such as hematological malignancies, chemotherapy, and HSCT [2, 3]. The conditioning regimen administered before stem cell infusion varies on a continuum from high‐dose (myeloablative) to reduced‐intensity to non‐myeloablative. The higher intensity of the conditioning regimen is often associated with increased toxicity [4]. The introduction of reduced‐intensity conditioning has led to a rise in HSCT, specifically in older patients and patients with comorbidity [5]. The major adverse outcomes of HSCT are relapse of the underlying disease, toxicity of the conditioning regimen, infections, and graft‐versus‐host disease (GvHD) [6]. Over the past three decades, treatment‐related mortality has been reduced; however, survival decreases with age [7, 8].
Screening patients with hematological malignancies for frailty pre‐HSCT may greatly enhance health personnel's ability to identify those at risk of frailty, allowing for timely intervention before treatment begins and reducing the risk of adverse outcomes [2]. Traditionally, decisions regarding patient selection and the optimal time point to perform HSCT depend on the risk stratification of the underlying disease, chronological age, and comorbidities. Tools such as the hematopoietic cell transplant comorbidity index (HCT‐CI) [9] and the European Group for Blood and Marrow Transplantation risk score (EBMT‐score) are commonly used to aid in such a decision‐making process [10]. However, chronological age, comorbidity, and performance status may have limited utility in capturing the heterogeneity of older patients with hematological malignancies. In addition, comorbidity indices, such as the HSCT‐CI and EBMT scores, may not adequately identify frailty or physical and psychological vulnerability and impairments, factors that could significantly impact morbidity and mortality after HSCT [1]. Thus, there is a growing need to supplement standard risk assessment with measures of physical and psychological functioning [1]. Multiple assessment tools are used to identify frailty, impairments, or vulnerabilities in patients with hematological malignancies and those undergoing HSCT [2]. Frailty is multidimensional, including physical and psychosocial factors. Individuals may fluctuate between states of severity of frailty, such as vulnerable, unfit, prefit, or frail [1]. Frailty can be evaluated by generic geriatric assessment (GA) tools, such as the clinical frailty scale (CFS), the vulnerable elders survey‐13 vulnerability score (VES‐13), or by more disease‐specific tools, such as the revised myeloma comorbidity index (R‐MCI), the International Myeloma Working Group frailty score (IMWG frailty score), or a combination of these [1]. This additional layer of assessment can aid in determining which patients may benefit from HSCT and who is most likely to have an adverse outcome [2].
Three systematic reviews reported the results of GA in patients with hematological malignancies and in patients undergoing HSCT [11, 12, 13]. Typically, GA tools contain domains such as cognition, physical function, comorbidities, polypharmacy, social support, mental health, and nutritional status. Based on GA scores, patients were classified into groups of the extent of frailty or vulnerability (e.g., fit, prefrail/unfit/intermediate‐fit, or frail) [12]. One systematic review, including patients with hematological malignancies, found that several geriatric impairments and frailty were predictive of shorter overall survival rates. This review suggests that GA assessment, even in patients with a good performance status, may detect impaired geriatric domains, which may be predictive of mortality. Moreover, these geriatric impairments propose a higher risk of treatment‐related toxicity, treatment non‐completion, and increased utilization of healthcare services [13]. This is in line with a systematic review of patients with hematological malignancies. Geriatric impairments are associated with shorter overall survival, poor physical activities, nutritional status, and cognitive capacities [11]. Comorbidity, physical capacity, and nutritional status retained their significance in multivariate analyses, whereas age and performance status lost their predictive value in most studies [11]. A systematic review and meta‐analysis of patients with myeloma found that the risks of hematologic adverse events were similar in intermediate‐fit and frail patients. However, a significantly increased risk of non‐hematologic adverse events was found in frail patients compared to fit patients. Patients classified as frail showed a higher risk of death than fit patients [12]. A narrative review of GA and the management of cancer patients in general emphasized that GA may not cover unspecific complications such as delirium, falls, and functional decline, and that older patients are a heterogenous group [14]. This suggests that older adults undergoing HSCT might be fitter and younger than the traditional older adult population. Moreover, intensive chemotherapy before HSCT in middle‐aged patients has been associated with frailty. Previous systematic reviews assessing frailty screening in patients with hematological malignancies and in patients undergoing HSCT have mainly focused on GA tools [7, 14, 15, 16]. Thus, there is a need to include multiple tools when assessing frailty and clinical outcomes, and the interplay between frailty and age in these patients [14, 17]. This broader approach can provide a more comprehensive understanding of frailty in patients with hematological malignancies and patients undergoing HSCT. This systematic review aimed to analyze the extent of frailty in patients with hematologic malignancies and those undergoing HSCT, and to explore the associations between frailty and age, and clinical outcomes.

2
Methods
This systematic review was reported according to the synthesis without meta‐analysis (SWiM) guidelines [18] (Supporting Information). A systematic search was conducted using the databases CINAHL (EBSCO), Embase (Ovid), and Medline from inception until January 31, 2024. The search strategy was built in Medline by the first authors based on keywords and subject headings used in previous publications, and on the advice from a research librarian. The search strategy consisted of two elements: (1) hematological malignancy and stem cell transplantation and (2) frailty and vulnerability. The search strategy was then piloted in medline, and the search strategy was applied to the other databases (Supporting Information). Moreover, we hand‐searched the reference lists of the included studies to ensure that we did not miss any relevant studies. The inclusion and exclusion criteria are listed in Table 1.
First, the results of our search were transferred to Endnote20, and duplicates were eliminated. Second, the results were transferred to Rayyan [19], a web‐based collaboration software platform that helps organize and blind the screening. In Rayyan, an automatic duplicate check identified more duplicates, which were subsequently removed.
The first authors (M.B. and M.R.K.) independently screened the title and abstract for eligibility. The same two authors then independently assessed the full‐text publications. Any disagreements between these authors (M.B. and M.R.K.) were resolved through discussion until an agreement was reached. The first authors (M.B. and M.R.K.) independently used Joanna Briggs Institute (JBI) critical appraisal tools to evaluate the methodological quality of the included studies. Cohort studies were assessed using the JBI checklist for cohort studies, and cross‐sectional studies were assessed using the JBI checklist for analytical cross‐sectional studies [20]. Disagreements were resolved by discussion until agreement was ascertained. No articles were excluded based on the quality appraisal (Supporting Information). Data from the included articles were independently extracted by the first authors (M.B. or M.R.K.) using multiple standardized data collection forms. Data included author, year, country, study methods, patients' characteristics, description of the assessment tools, and main findings related to the aim of this review. Subsequently, the data were ascertained for accuracy against the original papers. No disagreements in data extraction occurred during this process. We used narrative synthesis to analyze and summarize what is known about the study aim [21]. The narrative synthesis was guided by the European Social Research Council Guidance on the Conduct of Narrative Synthesis in systematic reviews [22]. We analyzed and summarized data from multiple studies by organizing data according to standardized data collection forms and used text and words to summarize and explain the findings of the included studies [22, 23]. We compared the data from the data collection forms to identify similarities and differences across the included articles.

3
Results
The initial search yielded 6764 publications. After the removal of duplicates and the screening of titles and abstracts (n = 5190), 151 full‐text publications were reviewed. One hundred and thirty‐four studies were excluded. The reasons for exclusion are described in Figure 1. No relevant studies were found by hand search. Seventeen articles based on 16 studies were included in this review.
3.1
Article Characteristics
The articles were published between 2013 and 2024 and were conducted in the United States (n = 11), Germany (n = 2), Italy (n = 2), Brazil (n = 1), and Canada (n = 1). Fifteen articles were cohort studies [24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38] and two were cross‐sectional studies [39, 40]. Patients were recruited from single‐center or multicenter sites, and the number of participants varied between 50 and 801. The characteristics of the included articles are presented in Table 2.
3.1.1
Participant Characteristics
The results of the quality assessment are provided in Supporting Information. The participants were hospitalized for a variety of hematologic malignancies. Patients with acute myelogenous leukemia and myelodysplastic syndrome were represented in most articles (n = 12) [24, 26, 29, 30, 31, 32, 34, 36, 37, 38, 39, 40]. Fore articles reported cases of leukemia but did not specify the leukemia subtype [30, 35, 37, 38]. Seven articles included patients with lymphoma (n = 7) [26, 29, 30, 31, 32, 35, 40] and multiple myeloma (n = 5) [25, 27, 28, 33, 35].
Most of the articles (n = 15) described patients' eligibility for HSCT, while two articles did not clearly describe whether patients were eligible for HSCT [27, 38]. Two articles reported that all included patients underwent HSCT [29, 37]. The majority of the articles included patients treated with allo‐HSCT (n = 11) [24, 26, 29, 30, 31, 32, 34, 36, 37, 39, 40], while four articles included patients undergoing auto‐HSCT [25, 27, 28, 33]. Derman et al. [26] included patients who underwent either allo‐HSCT or auto‐HSCT, while in the remaining two articles, the number of patients treated with HSCT was not clear [35, 38].
The ages of the participants ranged from 18 to 93 years, and eight articles presented their results based on the categorization of different age groups [26, 30, 31, 32, 36, 37, 38, 40]. The median age varied between 54.7 and 72.3 years. Four studies did not report median age [30, 31, 34, 38]. One article presented results based on age groups and the prevalence of frailty [25]. The patient characteristics are described in Table 3.

3.1.2
Frailty Assessment Tools and Their Domains
Across the articles, 17 different tools were used to describe frailty, vulnerability, or impairments. Eleven tools identified frailty: acute myelocytic leukemia score (AML‐score), CFS, comprehensive geriatric assessment (CGA), deficit accumulation frailty index (DAFI), fried frailty index (FFI), Geriatric‐8 score (G8‐score), GA, hematopoietic cell transplantation‐specific comorbidity index (HCT‐CI), IMWG frailty score, R‐MCI, and short physical performance battery (SPPB). Five tools identified impairments: blessed orientation‐memory‐concentration test (BOMC‐test), cancer‐specific GA (cGA), clinical pretransplantation optimization program (C‐POP), modified cancer‐specific GA (mGA), and montreal cognitive assessment (MoCA), and one tool identified vulnerability: VES‐13. The tool used by most studies was the FFI (n = 8), followed by the IMWG frailty score (n = 3). One article reported a significant association between worse MoCA score and worse SPPB score in multivariable analyses [37].
Abnormal nutrition status, comorbidities, impact on social support, cognitive impairments, and activities of daily living, as well as low physical activity, low grip strength, low gait speed, weight loss, and exhaustion were described as characteristics of frailty in 10 articles [26, 29, 30, 31, 32, 34, 36, 37, 39, 40]. The two domains of low grip strength [26, 27, 31, 32, 34, 35, 38, 39, 40] and weight loss [24, 27, 31, 32, 34, 35, 38, 39, 40] were each reported in nine articles. The domains of physical activity, exhaustion, and low gait speed were reported in eight articles [27, 31, 32, 34, 35, 38, 39, 40]. Seven articles reported comorbidities [25, 26, 28, 29, 33, 36, 40], five articles reported abnormal nutrition status [26, 29, 30, 35, 40], and five articles reported activities of daily living [25, 28, 29, 33, 40] as domains of frailty. In four articles, cognitive impairments were reported [29, 30, 37, 40], while three articles reported the domain impact on social support [26, 29, 35]. The characteristics of each tool are described in Table 4.

3.2
Frailty Score and Age
Six articles showed an association between higher age and the increased occurrence of impairments and comorbidities [25, 27, 28, 30, 31, 32, 34]. Sung et al. observed that the older age cohort showed a higher number of prefrail and frail patients compared with the younger cohort. Another article showed the presence of frailty among younger patients and that frailty increased with age. These results were confirmed in multivariable analysis [27]. Two articles found an association between older age and cognitive decline [30, 37]. Lew et al. also reported that patients aged 60 years and older had a higher risk for nutritional impairment compared with younger patients, and that nutritional impairment and older age could lead to impaired overall survival. Two articles showed no association between older age, impairments, and comorbidities [35, 36]. Moreover, three articles showed no association between patients considered eligible for HSCT and patients classified as fit or between patients considered ineligible for HSCT and patients classified as frail [25, 33, 37]. Two articles show that frailty and geriatric vulnerabilities may be present in patients already considered eligible for HSCT [38, 40].

3.3
The Extent of Frailty and Clinical Outcomes
The extent of frailty was categorized within the three subgroups of fit, prefrail/unfit/intermediate‐fit, and frail across the included articles. Thirteen articles presented their results of frailty based upon these subgroups, and frailty ranged from 7% to 38% [25, 27, 28, 29, 31, 32, 33, 34, 35, 36, 37, 38, 40]. Three articles reported the extent of frailty for two of the three subgroups: fit and frail/unfit/intermediate‐fit [24] and fit and frail [34, 35]. Patients classified as frail or frail/unfit/intermediate‐fit ranged from 21% to 46% [24, 34, 35]. Eleven articles described their patients as prefrail/unfit/intermediate‐fit, ranging from 25% to 60% [24, 25, 27, 28, 29, 31, 32, 33, 37, 38, 40]. Three articles did not report on the extent of frailty [26, 30, 39]. The prevalence of frailty is described in Table 5.
Most articles (n = 9) aimed to assess whether frailty could predict morbidity and mortality. Six of these articles showed a significant association between fit patients and longer overall survival [24, 27, 28, 32, 38]. These results remained significant for AML‐score and G8‐score (not HCT‐CI score) in multivariable analysis in one study [24]. A study showed worse overall survival in patients with one or more impairments [30]. Another study found that non‐relapse mortality was the most common cause of death in relation to frailty, which significantly led to a decrease in deaths due to relapse of the underlying disease [38]. Two articles showed an association between frail patients and an increased risk of infection and serious adverse events of cardiac, pulmonary, and renal impairments [28, 35]. Engelhart et al. found that frailty, impaired renal function, and lung function were significant risk factors of overall survival in multivariable analysis [27]. Holler et al. showed an association between frailty and hematological complications, such as anemia and thrombocytopenia. However, they found no association between frailty and leukocytopenia. Two articles showed a significant association between patients being fit and progression‐free survival [25, 28]. These results were confirmed in multivariable analysis [25]. Three articles presented an association between patients being prefrail/unfit/intermediate‐fit and outcome [28, 29, 38]. Furthermore, two articles showed that patients classified as intermediate fit had better survival and progression‐free survival compared with frail patients [28, 29]. One article showed significantly worse overall survival for patients classified as prefrail compared to patients classified as fit, and higher rates of pneumonia, acute GvHD, and sepsis among prefrail patients compared to fit patients [38].

4
Discussion
This systematic review aimed to analyze the extent of frailty in patients with hematologic malignancies and those undergoing HSCT, and to explore the associations between frailty and age, and clinical outcomes. A variety of tools were used to predict frailty in these patients, each containing multiple domains, and these domains were mainly similar across the different tools. Inconsistency in how the outcomes of these domains were assessed makes it challenging to compare the various tools and to determine the number of frail patients as well as the association between frailty, morbidity, and mortality.
All patients in the included articles had various hematologic malignancies, and most patients were eligible for HSCT. In contrast, other reviews have included cancer patients in general, with a minority having hematological malignancies, and chemotherapy was the most common treatment [11, 13, 14]. While most of the tools in our review aimed to identify frailty, some were also designed to identify impairments or vulnerabilities, which often included aspects of frailty. Tools such as the Fried Frailty Index, G8 score, IMWG frailty score, and R‐MCI were frequently used in the articles in this review, while previous reviews included articles concerning GA [11, 13]. Domains such as nutritional status, comorbidities, social support, cognitive functioning, and physical activity were frequently included in our review. These domains are also included in GA; thus, systematic reviews limited to GA may per se not inherently diminish the understanding of frailty in recipients of HSCT. However, the operationalization of specific domains (i.e., nutritional status) may differ between the tools and may impact the results. For instance, the evaluation of nutritional status can be based on BMI, weight loss, physical signs of malnutrition (e.g., muscle wasting, edema), or biochemical markers of nutritional status. These variations may not be directly comparable and may impact the frailty score. We found that the categorization of frailty (i.e., frail or frail/unfit/intermediate‐fit) differed between the articles and may impact the proportion of frailty. In line with this, the proportion of patients categorized as frail ranged from 7% to 38%, while 21% to 46% were categorized as frail or frail/unfit/intermediate‐fit.
Our review indicates that the role of age in predicting frailty is conflicting. Some of the included articles revealed that frailty, vulnerability, and impairments increased with age, and that HSCT is particularly challenging among older patients [38]. In contrast, an included article found that frailty at baseline did not differ between the younger and older cohorts of HSCT‐eligible patients [35]. However, the patients included in the latter article were younger than the patients included in another article [38].
Our findings suggest an association between the tools used to identify frailty and outcomes, such as survival and toxicities. This aligns with a systematic review stating that the use of GA can identify vulnerabilities in older adults planned for HSCT that are not captured by traditional measures [15]. Therefore, conducting a GA prior to HSCT may uncover vulnerabilities, and these results could tailor pretransplantation interventions, potentially improving patient outcomes [15]. In line with other reviews [11, 13], we found an association between frailty and adverse treatment events. Frail patients exhibited an increased risk of infection and serious adverse events, including cardiac, pulmonary, and renal impairments. Moreover, the relationship between frailty and hematological toxicity (i.e., a decrease in bone marrow and blood cells may lead to infection, bleeding, or anemia) was explored. We found that anemia and thrombocytopenia were observed in frail patients, whereas no association between frailty and leukocytopenia was found [28]. Symptoms of anemia, such as feeling weak and tired, are captured by frailty assessment, whereas leukocytopenia without infectious complications may be non‐symptomatic. Similarly, a systematic review found that impairments in patients with hematologic malignancies may be associated with toxicity, thereby predicting clinical outcomes [11]. Additionally, we found that frailty was associated with non‐hematologic toxicity grades 3–4 (i.e., adverse events of the treatment affecting organs and systems outside the blood and bone marrow) measured 1 year after HSCT [35]. In contrast, one article reported the same rates of hematological and non‐hematological toxicity in patients considered fit, intermediate fit, or frail [33]. These conflicting results may be due to heterogeneous diagnoses, treatment modalities, age, assessment tools, and follow‐up times, which impact the assessment of toxicity. Furthermore, we found that frailty was associated with reduced overall survival, which aligns with three systematic reviews describing an association between impairments and a decline in overall survival [11, 15, 16].
A methodological strength of this review was that the literature search was conducted in highly relevant databases and that a hand search was performed. The search strategy was established with advice from an experienced research librarian utilizing search terms based on keywords employed in previous articles. Consequently, we captured a wide number of articles. Another strength was that the first authors independently assessed eligibility, appraised the methodological quality, and extracted data. Another strength was the inclusion of multiple tools to assess frailty, providing a broader understanding of frailty. This systematic review was also strengthened by limiting it to hematological malignancies and not cancer patients in general. Thus, the patient population was more homogeneous in relation to diagnosis and treatment modalities impacting frailty.
This systematic review has some limitations. Articles were included regardless of their quality. Several studies had a small sample size, and there was heterogeneity within the populations due to the variation in diagnosis, conditioning regimes for HSCT (i.e., toxicity of the chemotherapy), type of transplantation (i.e., auto‐HSCT and allo‐HSCT), age, frailty screening tools, and study outcome. The latter reflects methodological challenges within studies of adults with hematological malignancies undergoing HSCT. Moreover, multivariable analysis of interactions between covariates potentially affecting the results was limited by the small sample size. Baseline characteristics (e.g., disease stage, time from diagnosis to transplant) that may predict outcomes in frail patients were also limited. The studies used one or several tools to identify frailty, vulnerability, or impairments. However, several studies did not report on which specific clinical team member (i.e., physician, nurse, social worker) used the tool, or what kind of competence was needed for performing the screening.
This systematic review aimed to analyze the extent of frailty in patients with hematologic malignancies and patients undergoing HSCT, and explore the associations between frailty and age, and clinical outcomes. Seventeen tools were identified that were used to assess frailty, vulnerability, or impairment, and some of them were used in several studies. Frailty in recipients of HSCT was characterized by abnormal nutrition status, comorbidities, impact on social support, activities of daily living, low physical activity, low grip strength, low gait speed, weight loss, exhaustion, and cognitive decline. Further, there seemed to be an association between fit patients and longer overall and progression‐free survival and an association between frailty and morbidity and mortality in patients with hematologic malignancies and recipients of HSCT. HSCT among the older population is challenging, and studies suggest a clear connection between frailty and age.
Inclusion of frailty assessment in the traditional pretransplant risk assessment may improve the predictive value of pretransplant risk assessment and identify patients at risk of morbidity and mortality. Thus, identifying patients at risk of frailty pre‐HSCT may identify a window for tailored interventions and reduce HSCT‐associated morbidity and mortality. The use of multiple tools is recommended for frailty assessment in hematological malignancies. Validated disease‐specific tools such as the IMWG frailty score and R‐MCI in patients with myeloma, and the FFI in patients with myelodysplastic syndrome, acute leukemia, and lymphoma should be preferred for use in clinical practice. However, GA tools such as the VES‐13, G8‐score, and CFS combined with a comorbidity index may identify frail patients undergoing HSCT regardless of the underlying disease.

Author Contributions

Marit Bakken: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, project administration, data curation. Marie Roko Kallager: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, project administration, data curation. Marie Hamilton Larsen: metodology, validation, visualization, writing – review and editing, data curation. Simen A. Steindal: methodology, validation, visualization, writing – review and editing, data curation. Kristin J. Skaarud: conceptualization, investigation, writing – original draft, methodology, visualization, writing – review and editing, formal analysis, project administration, data curation, supervision, validation.

Funding
This work was supported by the Oslo University Hospital.

Consent
There is no patient or public contribution, as this is a systematic review.

Conflicts of Interest
The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Information.