Use of comprehensive geriatric assessment (CGA) to define frailty in geriatric oncology: Searching for the best threshold. Cross-sectional study of 418 old patients with cancer evaluated in the Geriatric Frailty Clinic (G.F.C.) of Toulouse (France)

Journal of Geriatric Oncology 10 (2019) 944–950

Contents lists available at ScienceDirect

Journal of Geriatric Oncology

Use of comprehensive geriatric assessment (CGA) to define frailty in geriatric oncology: Searching for the best threshold. Cross-sectional study of 418 old patients with cancer evaluated in the Geriatric Frailty Clinic (G.F.C.) of Toulouse (France) Delphine Bréchemier a,⁎, Sandrine Sourdet a,b, Philippe Girard a, Zara Steinmeyer a, Loic Mourey c, Stéphane Gérard a, Laurent Balardy a a b c

Geriatric Department, Internal Medicine and Oncogeriatry Unit, University Hospital, Place du Docteur Baylac, TSA 40031, 31059 Toulouse Cedex 9, France Medical faculty, INSERM UMR 1027, 37 allées Jules Guesde, 31000 Toulouse, France Medical oncology department, Claudius Régaud Institute-Oncopole-Toulouse Cancer University Institute (IUCT-O), 1 avenue Irène Joliot-Curie, 31100 Toulouse, France

a r t i c l e

i n f o

Article history: Received 26 November 2018 Received in revised form 15 March 2019 Accepted 17 March 2019 Available online 4 April 2019

a b s t r a c t Objectives: A consensual and operational definition of frailty is necessary in geriatric oncology. While many studies have focused on geriatric syndromes evaluated in the comprehensive geriatric assessment (CGA) to select patients at higher risk of poor outcomes, few have compared CGA data with Fried's phenotype of frailty, the most consensual measurement of frailty in geriatrics. Our objective was to determine a threshold of impaired domains evaluated in CGA associated with Frailty status. Methods: A cross-sectional study including all patients with cancer, evaluated from January 2011 to February 2016 at the Geriatric Frailty Clinic, Toulouse. A CGA was performed evaluating seven geriatric domains. Frailty was measured by Fried's phenotype to classify patients into three groups (robust/pre-frail/frail). We plotted a ROC curve to determine the threshold of impaired domains associated with frailty according to Fried. Results: We included 418 patients aged 82.8 years (range 66–100 years). Thirty-three patients (7.9%) were robust, 155 (37.1%) pre-frail and 230 (55%) frail. There was a significant difference in ADL, IADL, nutrition, cognition and polypharmacy between the three groups (p b .001 for each domain). Frail patients had more impaired geriatric domains on CGA than pre-frail and robust patients (respectively 4.5 ± 1.5, 2.8 ± 1.6 and 2.1 ± 1.2; p b .001). The threshold of 4 impaired geriatric domains associated with Fried's Frailty status was identified (Se 77.39%, Sp 67.55%). Area under the curve was 79.6%. Conclusion: The phenotype of frailty is associated with more impaired geriatric domains and a threshold of 4 altered domains could be used to detect frailty from CGA data. © 2019 Elsevier Ltd. All rights reserved.

1. Introduction The increase in the proportion of older patients with cancer raises the question of selecting patients who can tolerate cancer treatments. Nearly 50% of patients over 70 years will experience severe chemotherapy-related toxicity [1,2], and 60% of older patients will have post-operative complications after colorectal cancer surgery (78% of which will have severe ones) [3]. One of the major challenges of geriatric oncology is to estimate for each patient the risk/benefit balance of any therapeutic

⁎ Corresponding author. E-mail address: [email protected] (D. Bréchemier).

https://doi.org/10.1016/j.jgo.2019.03.011 1879-4068/© 2019 Elsevier Ltd. All rights reserved.

proposal. Since ageing is a highly heterogeneous process, therapeutic decision must take into account the individual quality of patient's ageing, physiological reserves and the potential risks of the planned treatment. A great number of tools and scales have been developed in the literature to identify so called “frail” patients at high risk of complications and whose care should be individually tailored [4,5]. The definition of frailty in geriatric Oncology remains highly debated [6]. It is most often associated with the presence of impairments of domains in comprehensive geriatric assessment (CGA). CGA is a multidimensional evaluation of different geriatric domains (nutrition, comorbidities, cognition, functional autonomy, physical performance…) [7,8]. Impairment of geriatric parameters is associated with poor treatment tolerance [9]. The geriatric tools and thresholds to

D. Bréchemier et al. / Journal of Geriatric Oncology 10 (2019) 944–950

risk-stratify a patient are very varied in the literature and are actually chosen randomly [5,10,11]. Thus so called “frailty” is defined by the impairment of at least 1 to 4 geriatric domains evaluated in the CGA with different tools. The prevalence of “frailty” is therefore very variable from 7% to 68% depending on the threshold chosen [11]. A more standardized definition of frailty with cutoff points is needed in geriatric oncology [10]. In geriatric medicine one of the most consensual assessment of frailty is based on Fried's phenotype of frailty [12]. In this concept, the ability to withstand stress depends on the level of functional performance of the individual (walking speed, grip strength). This concept has shown its capacity to predict pejorative geriatric events (functional decline, hospitalization, institutionalization, risk of mortality). While cancer and its specific treatments represent great stressors, the frailty phenotype has still been little studied in geriatric oncology and has rarely been compared with CGA data [11]. In our study, we first describe the functional and geriatric parameters of a population of older adults with solid tumors or malignant hemopathies, evaluated at the Geriatric Frailty Clinic of the University Hospital of Toulouse, and classified according to their frail, pre-frail or robust phenotype. Our objective is to determine the threshold of impairment of geriatric domains evaluated by CGA associated with Fried's phenotype of frailty in order to better stratify patients based on CGA data. 2. Patients and Methods 2.1. Patients We carried out a cross-sectional study including all the patients referred to the Geriatric Frailty Clinic (GFC) of the University Hospital of Toulouse, France, for a solid tumor or malignant hemopathy from January 2011 to February 2016 before specific cancer treatment. This clinic is aimed at performing a Comprehensive Geriatric Assessment on patients older than 65 years through a multidimensional approach, in order to identify frail patients, and to guide geriatric interventions [13]. Patients were referred to the Geriatric frailty clinic before specific treatment for cancer. They were sent in order to benefit from a comprehensive assessment to guide therapeutic decision, according to the judgment of the geriatricians. Screening tools such as the G8 have not been systematically performed to select patients to refer to the clinic. Ethics approval has been delivered by the Ethics committee of the Toulouse University Hospital. 2.2. Data Collection A standardized geriatric assessment was performed and collected by a multidisciplinary team: - Socio-demographic data - Comorbidities using Charlson comorbidity index (CCI) [14], drug exposures. We used a threshold of CCI ≥ 2 to define a patient with significant comorbidities. Polypharmacy was defined by 5 or more ongoing medications. - Cancer stage and characteristics, therapeutic plan. - Frailty, as defined by Fried's phenotype [12], using the 5 original criteria: unintentional weight loss, slow walking speed (as measured by 4-m usual gait speed stratified by height and sex), exhaustion, weakness (as measured by hand grip stratified by BMI and sex), low physical activity. Patients were divided into three groups: robust (no criteria), pre-frail (1–2 criteria), and frail (3–5 criteria). - Cognition using the Mini Mental State Examination (MMSE) [15]. Impairment was defined for scores ≤24/30.

945

- Functional Autonomy using Activity of daily Life (ADL) [16] and Instrumental ADL (IADL) scales [17]. Disability in ADL or IADL was defined for scores of ADL ≤ 5/6 and/or IADL ≤7/8. - Functional performances scored with the Short Physical Performance Battery (SPPB) [18]. The score comprises of 3 tests for gait speed, balance and chair stand. Performances are sorted into 3 classes: high performance (score 10–12), medium performance (score 7–9), low performance (score 0–6). Impairment was defined for score ≤9. Gait speed was calculated for a 4 m distance at usual walking speed. Physical strength was defined by a grip test with a dynamometer [19]. - Nutritional status using the Mini Nutritional Assessment (MNA) [20]. Patients were considered malnourished for scores b17/30, at risk of malnutrition for scores between 17 and 23.5/30, and with a normal nutritional status for score ≥24/30. - Mood using the GDS 15 scale [21]. Depression was supposed for scores ≥6. - Domain impairment was defined in case of one of the following scores considered abnormal (ADL ≤ 5, IADL≤7, CCI ≥ 2, medications ≥5, MMSE ≤24/30, SPPB ≤9, MNA ≤ 23.5).

2.3. Data Analysis Continuous variables are described by mean and standard deviation, qualitative variables by frequency and percentage. Comparison between groups for qualitative variables was performed using the chi-square test or Fisher's exact test for small numbers. The threshold of significance is retained for a p-value b 5%. Continuous quantitative variables were compared using Student's ttest in case of normal distribution and variances equality. For the comparison of mean values with more than two observation groups, an analysis of variance (ANOVA) was used. A receiver operating characteristic (ROC) curve was plotted to evaluate the ability of different cut-off points of impaired domains in CGA to predict frailty, and to calculate sensitivity and specificity values. The optimum cut-off point was chosen using the Youden index [maximum (sensitivity + specificity – 1)] [22]. The Youden index is a commonly used index to determine an appropriate cutoff -value on the ROC curve. We chose this index with the objective of maximizing overall correct classification rates and minimizing misclassification rates. The association of impaired domains with frailty was tested in a multivariate logistic regression model adjusted for age, gender, educational level, marital status and living arrangements. The goodness of fit of the logistic regression model was assessed using Hosmer-Lemeshow test (non-significant p-values indicated that the fit of the model was good). All statistical analysis was carried out using statistical software STATAv11. 3. Results 3.1. Patients Features Between 10/01/2011 and 01/30/2016, 452 patients with malignant disease were admitted to the GFC of the Toulouse university hospital for geriatric evaluation. Thirty-four patients were excluded from analysis (seventeen patients due to significant alteration of general condition rendering their assessment impossible, six patients due to an uncertain cancer diagnosis, and eleven due to a diagnosis of non-malignant hemopathy). The analysis concerns the remaining 418 patients. Mean age was 82.8 years, ranging from 66 to 100 years. The sexratio was 1.08 (Table 1). Most patients were community-dwellers (n

946

D. Bréchemier et al. / Journal of Geriatric Oncology 10 (2019) 944–950

3.3. Fried's Criteria

Table 1 Comprehensive geriatric assessment data. All patients

Robust

Pre-frail

Frail

n = 418

n = 33

n = 155

n = 230

Age Age (±SD)

82.8 (5.5)

80.6 (5.5) 82.3 (5.2) 83.4 (5.7) b0.001

Sexe Male/Female

217/201

21/12

84/71

112/118

0.21

Dependency IADL (±SD) ADL (±SD)

5.1 (2.4) 5.4 (0.9)

6.2 (2.1) 5.9 (0.3)

6.3 (1.9) 5.8 (0.3)

4.1 (2.4) 5.1 (1.1)

b0.001 b0.001

N = 411 N = 33 169 (41.1) 28 (84.9)

N = 153 86 (56.2)

N = 225 55 (24.4)

b0.001

201 (48.9)

4 (12.1)

63 (41.2)

41 (10)

1 (3)

4 (2.6)

134 (59.6) 36 (16)

Cognition MMSE (±SD)

24.8 (5.0)

26.5 (4.7)

25.7 (4.5)

23.9 (5.2)

b0.001

Comorbidities Charlson Index (±SD)

1.4 (1.6)

1.2 (1.5)

1.3 (1.4)

1.5 (1.7)

0.25

5.7 (3.2)

4.7 (2.6)

4.9 (3.0)

6.5 (3.3)

b0.001

Impairment in CGA domains Numbers of impaired 3.7(1.8) domains (±SD)

2.1(1.2)

2.8(1.6)

4.5(1.5)

b0.001

Nutritional status MNA No malnutrition(%) (MNA = 24–30) At risk (%) (MNA = 17–23.5) Malnutrition (%) (MNA b17)

Daily medications Number of medications (±SD)

p

Abbreviations SD (Standard Deviation); ADL (Activity of Daily Living); IADL (Instrumental Activity of Daily Living); MNA (Mini Nutritional Assessment); MMSE (Mini Mental State Examination); CGA (Comprehensive Geriatric Assessment).

= 400, 96%), more than half had help from professional caregivers (n = 244, 59%). Among the patients, 203 patients (49%) were married and 215 patients (51%) were widower, single or divorced. Concerning educational level, 232 patients (56%) had low educational level (primary school) and 183 patients (44%) had higher educational level (3 missing data). Frail patients were statistically older than non-frail patients but the difference is clinically low. There was no difference in sex, educational level or marital status between frail and non-frail patients (p = .145 and 0.736 and 0.194 respectively).

3.2. Cancers Features Solid tumors and malignant hemopathies represented respectively 81.1% and 18.9% of all malignant diseases. Digestive tumors were the most represented with 29.7%, including more than two thirds of colorectal cancer. Urinary tract tumors represented 18.4%, half of them were bladder tumors. Only 16.7% of tumors were breast and gynecological cancers. Most of the cancers were non metastatic: 155 patients (37.1%) had a localized tumor, 75 patients (17.9%) a loco regionally advanced disease and 109 patients (26.1%) a metastatic disease. Most patients were evaluated before beginning specific treatment. Only 34 patients (8.1%) were evaluated for palliative care. After assessment, initial therapeutic plans were recommended to be unchanged for 323 patients (77.3%), with adaptation for 35 patients (8.4%), and changed for 60 patients (14.3%).

Only 33 patients were robust (7.9%) according to Fried's criteria. Prefrail patients represented 37.1% (n = 155), and frail patients 55% (n = 230) of all patients included (Table 2). Mean gait speed was b0.8 m/s. In the group of frail patients, mean gait speed was significantly slower than pre-frail or robust patients (p b .005). Grip strength was also statistically different between these groups (p b .005). Among Fried's criteria, 214/415 patients (51.6%) described a subjective exhaustion, 179/418 patients (42.8%) had a loss of weight, and 200/ 418 patients (47.8%) a low physical activity levels. While, none of the robust patients presented with exhaustion or weight loss as expected, about one quarter of pre-frail patients presented with exhaustion, and/or loss of weight and almost the half presented with low physical activity. These proportions are higher in the frail patients group with 78.1% describing a subjective exhaustion, 59.6% a weight loss, and 48.7% a low physical activity. 3.4. Geriatric Assessment Most patients had preserved autonomy in ADL but had disability in IADL score. Frail patients were significantly more dependent for ADL and for IADL than pre-frail and robust patients (p b .005)(Table 1). More than half of all the patients and most of frail patients had nutritional problems and frail patients had a significantly lower MNA than robust and pre-frail patients (p b .005) (Table 1). About functional and physical performances, more than half of frail patients had low physical performances and one third had medium performances (Table 2). SPPB scores were significantly lower in frail patients as compared to the robust or pre-frail group (p b .005). About cognition, about one third of the patients had a score ≤ 24. Frail patients scored lower than their robust and pre-frail counterparts but the difference was not clinically significant. About mood, GDS 15 was collected for only 167 patients. Forty-one patients (24.6% of the evaluated patients) had a score ≥ 6. Regarding comorbidity and polypharmacy, there was no significant difference between groups (p = .25) in mean CCI. However, frail patients had significantly more ongoing medications than pre-frail and robust patients (p b .005). 3.5. Impairment of Domains in CGA Among all the evaluated patients, only 14 patients (3.3%) had no impaired geriatric domain in CGA, and 45 patients (10.8%) had a single impaired geriatric domain. Three hundred and fifty-nine patients (85.9%) had at least 2 impaired domains, and 239 patients (57.2%) had at least 4 impaired domains (Fig. 1). Frail patients had more impaired geriatric domains than robust and pre-frail patients (p b .005) (Table 1). Only 7 frail patients (3% of frail patients) had 0 or 1 impaired domain in CGA and 178 frail patients (77.4% of frail patients) had 4 or more impaired domains. Thirteen robust patients (39.4% of robust patients) and 39 pre-frail patients (25.2%) had 0 or 1 impaired domain. Only 5 robust patients (15.1%) and 56 pre-frail patients (36.1%) had 4 or more impaired domains (Fig. 1). Frail patients were more dependent in ADL (respectively 44.8%, 8.4% and 6% in frail, pre-frail and robust patients, p b .001), in IADL (91.3%, 61.9%, and 60.6%, p b .001), were more malnourished (75.6%, 43.8% and 15.2%, p b .001), presented with more cognitive dysfunction (44.5%, 29.6% and 18.2%, p b .001), had more polypharmacy (71.3%, 47.1%, and 54.6%; p b .001) and had more frequently poor or

D. Bréchemier et al. / Journal of Geriatric Oncology 10 (2019) 944–950

947

Table 2 Functional performances and fried frailty criteria.

Functional performance Gait speed (GS) Median GS ± SD (m/s) GS b 1 m/s (N, %) SPPB SPPB median ± SD SPPB 0–6 (%) SPPB 7–9 (%) SPPB 10–12 (%) Grip strength Grip strength (kg+/-SD) Weight loss (fried criteria) Weight loss (fried criteria) (N, %) Exhaustion Exhaustion (N, %) Weak physical activity Weak physical activity (N, %)

All patients

Robusts

Pre-frail

Frail

p

N = 418 0.77 (0.27) 322 (77.0) N = 414 7.5 (3.1) 149 (36.0) 139 (33.6) 126 (30.4) N = 415 21.6 (8.4) N = 418 179 (42.8) N = 415 214 (51.6) N = 418 200 (47.8)

N = 33 1.0 (0.21) 14 (42.4) N = 33 10.4 (2.2) 2 (6.0) 6 (18.2) 25 (75.8) N = 33 31.9 (7.3) N = 33 0 (0) N = 33 0 (0) N = 33 14 (42.4)

N = 155 0.91 (0.18) 100 (64.5) N = 154 9.1 (2.0) 20 (13.0) 63 (40.9) 71 (46.1) N = 155 23.4 (8.3) N = 155 42 (27.1) N = 154 36 (23.4) N = 155 74 (47.7)

N = 230 0.65 (0.27) 208 (90.4) N = 227 5.9 (3.0) 127 (56.0) 70 (30.8) 30 (13.2) N = 227 18.9 (7.1) N = 230 137 (59.6) N = 228 178 (78.1) N = 230 112 (48.7)

b0.001 b0.001 b0.001 b0.001

b0.001 b0.001 b0.001 0.8

Abbreviations: GS (Gait speed); SPPB (Short Performance Physical Battery); SD (standard deviation).

Table 3 Association of frailty status with several thresholds of impaired domains on comprehensive geriatric assessment.

1.00

0.75

0.50

0.25

0.00

0.25

Area under ROC curve = 0.7960

0.50 1 - Specificity

0.75

1.00

Number of impaired domains

Sensibility, %

Specificity, %

Youden index

≥0 ≥1 ≥2 ≥3 ≥4 ≥5 ≥6 ≥7 N7

100 99.57 96.96 90.43 77.39 53.04 26.09 6.96 0

0 6.91 27.66 47.34 67.55 88.30 95.21 98.94 100

0 0.06 0.25 0.38 0.45 0.41 0.21 0.06 0

a

Fig. 1. ROC curve assessing the link between the number of impaired domains on CGA and frailty status. Area Under the curve:79.6%.

Multivariate analysesa Odds ratio

95% CI

p

16.9 11.8 8.3 7.05 8.0 6.3 5.9

2.2–131.5 5.2–27.1 4.8–14.2 4.47–11.13 4.7–13.6 3.0–13.3 1.3–27.2

0.007 b0.001 b0.001 b0.001 b0.001 b0.001 0.023

Adjusted for age, gender, educational level, marital status, institutionalization.

frailty status with several thresholds of impaired domains in Comprehensive Geriatric Assessment is shown in Table 3.

Number of Impaired domains 180

4. Discussion

Number of patients

160 140 120 100

All patients Robust Pre-frail Frail

80 60 40 20 0

N=0

N=1 N=2-3 N=4-5 Number of Impaired domains

N=6-7

Fig. 2. Distribution of number of impaired domains.

intermediate SPPB performance (86.8%, 53.9% and 24.2%. p b .001). The distribution of altered geriatric domain profiles is shown in Fig. 2. The ROC curve dealing with the association between impairment of geriatric domains and frailty is shown in Fig. 3. Area under the curve was 79.6% (95% CI: 75.4–83.8). The more efficient threshold of number of impaired domains associated with frailty was 4, with a Youden index of 0.45. A logistic regression adjusted for age, gender, educational level, marital status, and institutionalization showed a significant association between having 4 or more impaired CGA domains and Frailty (OR = 7.05, 95% CI:4.47–11.13, p b .001). The goodness of fit of the model was good (Hosmer-Lemeshow test: p:0.775). The association of

The International Society of Geriatric Oncology (SIOG) recommends an approach based on a comprehensive geriatric assessment (CGA) to help oncologists in selecting patients able to tolerate cancer treatment [8,23]. Many studies have largely shown that impairment of geriatric domains is predictive of many poor outcomes in geriatric oncology [9]: toxicity of chemotherapy [1,2,24,25], feasibility of treatment [26– 28], postoperative complications [29], functional decline [30], unplanned hospitalization, overall survival [31–35]. Frailty is one of the main topics in geriatric oncology literature but however, there is no consensus on an operational definition of frailty which differs among authors and studies [10]. Most authors use a definition of frailty based on CGA but the threshold is actually chosen arbitrarily from 1 to 4 impaired geriatric domains in CGA [11]. None have studied the most discriminant threshold associated with frailty. In our study we used Fried's phenotype as a reference to define frailty. First, we evaluated CGA data according to Fried's phenotype (frail, pre-frail or robust). Then we determined the threshold of impaired geriatric domains evaluated in the CGA associated with Fried's frailty phenotype. At our knowledge our study is one of the largest to have used Fried's phenotype of Frailty in comparison with CGA in an oncogeriatric population and the first one to have described geriatric features of each of Fried's Frailty groups. We chose to define frailty according to the

948

D. Bréchemier et al. / Journal of Geriatric Oncology 10 (2019) 944–950

l ai

l ai

l

il

ai Fr

fra

Al

Pr

e-

ob R

Fr

t

-fr

us

Pr e

us

s nt ie at lp

Pr

Al

t

R

pa Al l l ai

fra e-

ob R

Fr

t us

nt ie at

il

450 400 350 300 250 200 150 100 50 0

lp

ob

tie

l

l ai Fr

ai -fr

450 400 350 300 250 200 150 100 50 0 s

Al l

R

Pr e

ob

us

t

nt s tie pa

nt s

450 400 350 300 250 200 150 100 50 0

450 400 350 300 250 200 150 100 50 0

l ai Fr

il fra e-

Pr

R

ie at Al

lp

ob

nt

us

t

s

l ai Fr

Pr

e-

fra

t us

R

ob

nt ie at Al

lp

il

450 400 350 300 250 200 150 100 50 0 s

450 400 350 300 250 200 150 100 50 0

Fig. 3. Distribution of impaired geriatric domain according to each subgroup.

phenotype of frailty, since it is the most widely studied in geriatric medicine to predict poor outcomes, even if it has not been yet validated in elderly patients with cancer. It has been developed in a large cohort of 5317 community-dwelling older adults from the Cardiovascular Heart Study [12]. This concept has shown in many studies in geriatric medicine its capacity to predict adverse geriatric outcomes. Only a few studies in geriatric oncology have focused specifically on Fried's phenotype [11]. Most of them were small cohorts and most often involved surgical outcomes. These studies suggest that Fried's phenotype might help predict post-operative complication risk in surgery for digestive [3,29,36– 39] and gynecological cancers [40]. A recent study suggests Fried's phenotype could be a predictive factor of severe toxicity, occurring during the first two cycles of chemotherapy, in 50 patients with lung cancer [41]. Another study showed frailty markers could influence therapeutic decision [42]. Physical performance parameters that are central to define the phenotype of frailty are associated with adverse events in older adults with cancer. Thus, low get up and go score (GUG test) [43] is associated with significantly higher risk of severe chemotherapy toxicity [2], early death in patients undergoing chemotherapy [31], 30 days postoperative morbidity [44] and loss of autonomy [30]. Gait speed is independently associated with early death occurring within 6 months [32]. Cesari et al. showed that SPPB scores and gait speed were independent prognosis factors in women treated for gynecological

cancers [45]. Last but not least grip strength could predict chemotherapy-related toxicity [19,46]. Our study shows that Fried's criteria accurately stratified patients into three categories with distinct geriatric features. The 3 groups identified by these criteria not only have significantly different functional performances but are also distinguished by other geriatric domains such as nutrition, autonomy, cognition, and polypharmacy. Frail patients were more dependent in ADL (p b .001), in IADL (p b .001), had a worse nutritional status evaluated by MNA (p b .001), presented with more cognitive impairment (p b .001), and had more polypharmacy (p b .001) than pre-frail and robust patients. Furthermore the number of impaired geriatric domains in CGA was more important in frail patients than in pre-frail and robust patients (respectively 4.5 ± 1.5, 2.8 ± 1.6 and 2.1 ± 1.2; p b .001). Finally we found that a threshold of 4 geriatric impaired domain could be associated with frailty status (Se 77.39%, Sp 67.55%). Only a few studies tried to compare CGA data with phenotype of frailty. Kristjanson et al. found a weak concordance between frailty as defined by Fried and as defined by CGA with a kappa coefficient of 0.05 in a population of 176 old patients undergoing surgery for colorectal cancer [36]. But CGA frailty was defined arbitrarily by the presence of one or more impaired geriatric domain. Furthermore the prevalence of frailty was low in this study: 43% of patients had one or more geriatric

D. Bréchemier et al. / Journal of Geriatric Oncology 10 (2019) 944–950

domain impairments while phenotype of frailty was identified in only 13%. In our study the prevalence of geriatric syndromes and of the phenotype of frailty were much higher: 96.7% of patients had one or more geriatric domain impairments and 55% were classified as frail according to Fried's criteria. Biganzoli et al. tried to evaluate the ability of Fried's criteria to predict, as a screening tool, CGA abnormality [47] and found a good sensitivity but poor specificity. CGA abnormality was also defined by one or more impairment. In another study [48], Biganzoli found a poor concordance between Fried's frailty and Balducci's frailty criteria based on impairment of one or more geriatric domains. All of these studies used an arbitrarily chosen cut-off to define CGA-frailty. Further studies are needed to determine if frailty defined by 4 or more impaired CGA domains could predict poor outcomes in geriatric oncology since only too few studies found a correlation between the number of abnormal domains and oncologic outcomes. However our study is original, this is the first attempt to determine a threshold of alteration of parameters of CGA associated with the frailty phenotype as defined by Fried. In the majority of geriatric oncology studies, the frailty criteria are defined in a purely arbitrarily manner. Clough-Gorr et al. found an association between the number of CGA domain deficits and poor treatment tolerance and mortality at 7 years of follow-up in a large population of women with localized breast cancer [49]. Deficits in ≥3 CGA domains were significantly associated with poor treatment tolerance (OR 4.86, 95% CI 2.19–10.77) and mortality (HR 2.31, 95% CI 1.4–2.94) in a multivariate analysis. The main limitation of that study was that CGA data was collected directly from patients by telephone through self-rated scales and did not assess all geriatric domains such as cognitive function, nutrition, polypharmacy and physical performance. Another study showed that the number of impaired CGA domains was correlated with severe chemotherapy-related toxicity in older patients with advanced breast cancer undergoing first-line chemotherapy [34]. In this study, 80% of patients with 3 or more geriatric conditions had grade 3–4 toxicity. The correlation between the number of abnormal domains and oncologic outcomes must therefore be further studied in more and larger studies. In our cohort, the prevalence of frailty according to Fried's criteria represented more than half of the patients (55%), while it is estimated at around 15% in the elderly French population [50]. This difference could be explained by patient selection since our patients were specifically referred to the GFC in order to benefit from a geriatric evaluation before treatment for cancer. Prostate, breast and lung cancers were underrepresented in our cohort, despite their usual prevalence in older adults. This might be due to our privileged relations with certain oncology departments in the university hospital. Our study does have some limits. We were not able to evaluate certain parameters such as mood or weight change. Comorbidities were assessed by the Charlson Comorbidity Index rather than the Cumulative Illness Rating Scale (CIRSG) [51,52], which is known to be more comprehensive in assessing comorbidities and their severity in the elderly. However, our cohort of patients has been homogeneously assessed with standardized tools in our frailty platform with very little missing data for other geriatric parameters. One of the strengths of our study is the comprehensive assessment of physical performances, and the use of Fried's original criteria. Lastly, we did not evaluate the impact of each geriatric domain on the risk of frailty, which could be different from one domain to another, as a previous study showed the different weights of respective geriatric domains on the risk of mortality [53]. Finally, the definition of frailty and the impact of each geriatric domain could be different depending on the endpoint, the type of cancer, the cancer treatment planned and the population assessed and needs to be studied in longitudinal manner. The contribution of comprehensive geriatric assessment in clinical practice has yet to be demonstrated [54]. Corre et al. showed that CGA-based allocation of chemotherapy failed to improve survival outcomes in the elderly treated for advanced lung cancer, and questioned the contribution

949

of CGA in therapeutic decision. However in this study, patients from the standard strategy treatment based on age and performance status (PS) may have been undertreated since all the patients aged 75 or more and/or with a performance status >1 received attenuated treatment, regardless of the robustness. There was no difference in Overall Survival between the standard strategy and the CGA-based allocation of chemotherapy but in the CGA arm a quarter of the patients received best supportive care. This study highlights the importance of defining the most relevant endpoints in geriatric oncology: quality of life, functional decline, toxicity of treatment… However we believe using a threshold of 4 or more altered domains on CGA to define frailty could be useful in future studies even if the practical implication of this threshold needs to be more assessed. The results of our study could be used in everyday clinical practice to identify older adults at increased risk for adverse outcome, to guide the geriatric and oncologic management and care planning. 5. Conclusion A consensual and operational definition of Frailty was necessary in geriatric oncology. We suggest that a threshold of 4 and more altered domains on CGA should be used to define CGA-frailty in future studies. Further studies are needed to determine if this threshold could equally predict poor outcomes in elderly patients with cancer. Conflict of Interest All the authors do not declare any conflict of interest. Authors Contributions DB was responsible for writing the manuscript and preparing the final draft. SS was responsible for collecting the data and statistical analyses, and review. PG and ZS were responsible for review and spelling check in English. LM was responsible for review. SG was responsible for writing the first draft, and review. LB was responsible was responsible for writing the first draft, review and corrections on the final draft. References [1] Extermann M, Boler I, Reich RR, Lyman GH, Brown RH, DeFelice J, et al. Predicting the risk of chemotherapy toxicity in older patients: the chemotherapy risk assessment scale for high-age patients (CRASH) score. Cancer 2012;118(13):3377–86. [2] Hurria A, Togawa K, Mohile SG, Owusu C, Klepin HD, Gross CP, et al. Predicting chemotherapy toxicity in older adults with cancer: a prospective multicenter study. J Clin Oncol 2011;29(25):3457–65. [3] Kristjansson SR, Nesbakken A, Jordhøy MS, Skovlund E, Audisio RA, Johannessen HO, et al. Comprehensive geriatric assessment can predict complications in elderly patients after elective surgery for colorectal cancer: a prospective observational cohort study. Crit Rev Oncol Hematol 2010;76(3):208–17. [4] Ethun CG, Bilen MA, Jani AB, Maithel SK, Ogan K, Master VA. Frailty and cancer: implications for oncology surgery, medical oncology, and radiation oncology. CA Cancer J Clin 2017;67(5):362–77. [5] Hamaker ME, Jonker JM, de Rooij SE, Vos AG, Smorenburg CH, van Munster BC. Frailty screening methods for predicting outcome of a comprehensive geriatric assessment in elderly patients with cancer: a systematic review. Lancet Oncol 2012; 13(10):e437–44. [6] Pal SK, Katheria V, Hurria A. Evaluating the older patient with cancer: understanding frailty and the geriatric assessment. CA Cancer J Clin 2010;60(2):120–32. [7] Extermann M, Hurria A. Comprehensive geriatric assessment for older patients with cancer. J Clin Oncol 2007;25(14):1824–31. [8] Wildiers H, Heeren P, Puts M, Topinkova E, Janssen-Heijnen MLG, Extermann M, et al. International Society of Geriatric Oncology Consensus on geriatric assessment in older patients with Cancer. J Clin Oncol 2014;32(24):2595–603. [9] van Abbema DL, van den Akker M, Janssen-Heijnen ML, van den Berkmortel F, Hoeben A, de Vos-Geelen J, et al. Patient- and tumor-related predictors of chemotherapy intolerance in older patients with cancer: a systematic review. J Geriatr Oncol 2019;10(1):31–41.

950

D. Bréchemier et al. / Journal of Geriatric Oncology 10 (2019) 944–950

[10] Rouge Bugat M-E, Balardy L, Chicoulaa B, Cesari M, Gérard S, Nourashemi F. “Frailty” in geriatry and oncology: one term for two widely differing concepts. J Am Med Dir Assoc 2014;15(7):528–30. [11] Handforth C, Clegg A, Young C, Simpkins S, Seymour MT, Selby PJ, et al. The prevalence and outcomes of frailty in older cancer patients: a systematic review. Ann Oncol 2015;26(6):1091–101. [12] Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56(3): M146–56. [13] Tavassoli N, Guyonnet S, Abellan Van Kan G, Sourdet S, Krams T, Soto ME, et al. Description of 1,108 older patients referred by their physician to the “geriatric frailty clinic (G.F.C) for assessment of frailty and prevention of disability” at the gerontopole. J Nutr Health Aging 2014;18(5):457–64. [14] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40(5):373–83. [15] Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12(3): 189–98. [16] Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW. Studies of illness in the aged. The index of ADL: a standardized measure of biological and psychosocial function. JAMA 1963;185:914–9. [17] Lawton MP, Brody EM. Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist 1969;9(3):179–86. [18] Guralnik JM, Ferrucci L, Pieper CF, Leveille SG, Markides KS, Ostir GV, et al. Lower extremity function and subsequent disability: consistency across studies, predictive models, and value of gait speed alone compared with the short physical performance battery. J Gerontol A Biol Sci Med Sci 2000;55(4):M221–31. [19] Guo CB, Zhang W, Ma DQ, Zhang KH, Huang JQ. Hand grip strength: an indicator of nutritional state and the mix of postoperative complications in patients with oral and maxillofacial cancers. Br J Oral Maxillofac Surg 1996;34(4):325–7. [20] Vellas B, Guigoz Y, Garry PJ, Nourhashemi F, Bennahum D, Lauque S, et al. The mini nutritional assessment (MNA) and its use in grading the nutritional state of elderly patients. Nutrition 1999;15(2):116–22. [21] Sheikh JI, Yesavage JA, Brooks 3rd JO, Friedman L, Gratzinger P, Hill RD, et al. Proposed factor structure of the geriatric depression scale. Int Psychogeriatr 1991;3 (1):23–8. [22] Youden WJ. Index for rating diagnostic tests. Cancer 1950;3(1):32–5. [23] Mohile SG, Dale W, Somerfield MR, Schonberg MA, Boyd CM, Burhenn PS, et al. Practical assessment and management of vulnerabilities in older patients receiving chemotherapy: ASCO guideline for geriatric oncology. J Clin Oncol 2018;36(22): 2326–47. [24] Marinello R, Marenco D, Roglia D, Stasi MF, Ferrando A, Ceccarelli M, et al. Predictors of treatment failures during chemotherapy: a prospective study on 110 older cancer patients. Arch Gerontol Geriatr 2009;48(2):222–6. [25] Wildes TM, Ruwe AP, Fournier C, Gao F, Carson KR, Piccirillo JF, et al. Geriatric assessment is associated with completion of chemotherapy, toxicity, and survival in older adults with cancer. J Geriatr Oncol 2013;4(3):227–34. [26] Aparicio T, Jouve J-L, Teillet L, Gargot D, Subtil F, Le Brun-Ly V, et al. Geriatric factors predict chemotherapy feasibility: ancillary results of FFCD 2001-02 phase III study in first-line chemotherapy for metastatic colorectal cancer in elderly patients. J Clin Oncol 2013;31(11):1464–70. [27] Klute KA, Brouwer J, Jhawer M, Sachs H, Gangadin A, Ocean A, et al. Chemotherapy dose intensity predicted by baseline nutrition assessment in gastrointestinal malignancies: a multicentre analysis. Eur J Cancer 2016;63:189–200. [28] Woopen H, Richter R, Ismaeel F, Chekerov R, Roots I, Siepmann T, et al. The influence of polypharmacy on grade III/IV toxicity, prior discontinuation of chemotherapy and overall survival in ovarian cancer. Gynecol Oncol 2016;140(3):554–8. [29] Kristjansson SR, Jordhøy MS, Nesbakken A, Skovlund E, Bakka A, Johannessen H-O, et al. Which elements of a comprehensive geriatric assessment (CGA) predict post-operative complications and early mortality after colorectal cancer surgery? J Geriatr Oncol 2010;1(2):57–65. [30] Hoppe S, Rainfray M, Fonck M, Hoppenreys L, Blanc J-F, Ceccaldi J, et al. Functional decline in older patients with cancer receiving first-line chemotherapy. J Clin Oncol 2013;31(31):3877–82. [31] Soubeyran P, Fonck M, Blanc-Bisson C, Blanc J-F, Ceccaldi J, Mertens C, et al. Predictors of early death risk in older patients treated with first-line chemotherapy for cancer. J Clin Oncol 2012;30(15):1829–34. [32] Pamoukdjian F, Lévy V, Sebbane G, Boubaya M, Landre T, Bloch-Queyrat C, et al. Slow gait speed is an independent predictor of early death in older Cancer outpatients: results from a prospective cohort study. J Nutr Health Aging 2017;21(2):202–6.

[33] Bourdel-Marchasson I, Diallo A, Bellera C, Blanc-Bisson C, Durrieu J, Germain C, et al. One-year mortality in older patients with Cancer: development and external validation of an MNA-based prognostic score. PLoS ONE 2016;11(2):e0148523. [34] Hamaker ME, Seynaeve C, Wymenga ANM, van Tinteren H, Nortier JWR, Maartense E, et al. Baseline comprehensive geriatric assessment is associated with toxicity and survival in elderly metastatic breast cancer patients receiving single-agent chemotherapy: results from the OMEGA study of the Dutch breast Cancer Trialists' Group. Breast 2014;23(1):81–7. [35] Hamaker ME, Vos AG, Smorenburg CH, de Rooij SE, van Munster BC. The value of geriatric assessments in predicting treatment tolerance and all-cause mortality in older patients with cancer. Oncologist 2012;17(11):1439–49. [36] Kristjansson SR, Rønning B, Hurria A, Skovlund E, Jordhøy MS, Nesbakken A, et al. A comparison of two pre-operative frailty measures in older surgical cancer patients. J Geriatr Oncol 2012;3(1):1–7. [37] Tan K-Y, Kawamura YJ, Tokomitsu A, Tang T. Assessment for frailty is useful for predicting morbidity in elderly patients undergoing colorectal cancer resection whose comorbidities are already optimized. Am J Surg 2012;204(2):139–43. [38] Shen Y, Hao Q, Zhou J, Dong B. The impact of frailty and sarcopenia on postoperative outcomes in older patients undergoing gastrectomy surgery: a systematic review and meta-analysis. BMC Geriatr 2017;17(1):188. [39] Makary MA, Segev DL, Pronovost PJ, Syin D, Bandeen-Roche K, Patel P, et al. Frailty as a predictor of surgical outcomes in older patients. J Am Coll Surg 2010;210(6): 901–8. [40] Courtney-Brooks M, Tellawi AR, Scalici J, Duska LR, Jazaeri AA, Modesitt SC, et al. Frailty: an outcome predictor for elderly gynecologic oncology patients. Gynecol Oncol 2012;126(1):20–4. [41] Ruiz J, Miller AA, Tooze JA, Crane S, Petty WJ, Gajra A, et al. Frailty assessment predicts toxicity during first cycle chemotherapy for advanced lung cancer regardless of chronologic age. J Geriatr Oncol 2019;10(1):48–54. [42] Farcet A, de Decker L, Pauly V, Rousseau F, Bergman H, Molines C, et al. Frailty markers and treatment decisions in patients seen in Oncogeriatric clinics: results from the ASRO pilot study. PLoS ONE 2016;11(2):e0149732. [43] Mathias S, Nayak US, Isaacs B. Balance in elderly patients: the “get-up and go” test. Arch Phys Med Rehabil 1986;67(6):387–9. [44] Huisman MG, van Leeuwen BL, Ugolini G, Montroni I, Spiliotis J, Stabilini C, et al. “Timed up & go”: a screening tool for predicting 30-day morbidity in onco-geriatric surgical patients? A multicenter cohort study. PLoS ONE 2014;9(1):e86863. [45] Cesari M, Cerullo F, Zamboni V, Di Palma R, Scambia G, Balducci L, et al. Functional status and mortality in older women with gynecological cancer. J Gerontol A Biol Sci Med Sci 2013;68(9):1129–33. [46] Puts MTE, Monette J, Girre V, Pepe C, Monette M, Assouline S, et al. Are frailty markers useful for predicting treatment toxicity and mortality in older newly diagnosed cancer patients? Results from a prospective pilot study. Crit Rev Oncol Hematol 2011;78(2):138–49. [47] Biganzoli L, Boni L, Becheri D, Zafarana E, Biagioni C, Cappadona S, et al. Evaluation of the cardiovascular health study (CHS) instrument and the vulnerable elders Survey13 (VES-13) in elderly cancer patients. Are we still missing the right screening tool? Ann Oncol 2013;24(2):494–500. [48] Biganzoli L, Mislang AR, Di Donato S, Becheri D, Biagioni C, Vitale S, et al. Screening for frailty in older patients with early-stage solid tumors: a prospective longitudinal evaluation of three different geriatric tools. J Gerontol A Biol Sci Med Sci 2017;72(7): 922–8. [49] Clough-Gorr KM, Thwin SS, Stuck AE, Silliman RA. Examining five- and ten-year survival in older women with breast cancer using cancer-specific geriatric assessment. Eur J Cancer 2012;48(6):805–12. [50] Santos-Eggimann B, Cuénoud P, Spagnoli J, Junod J. Prevalence of frailty in middleaged and older community-dwelling Europeans living in 10 countries. J Gerontol A Biol Sci Med Sci 2009;64A(6):675–81. [51] Miller MD, Paradis CF, Houck PR, Mazumdar S, Stack JA, Rifai AH, et al. Rating chronic medical illness burden in geropsychiatric practice and research: application of the cumulative illness rating scale. Psychiatry Res 1992;41(3):237–48. [52] Parmelee PA, Thuras PD, Katz IR, Lawton MP. Validation of the cumulative illness rating scale in a geriatric residential population. J Am Geriatr Soc 1995;43(2):130–7. [53] Frasca M, Soubeyran P, Bellera C, Rainfray M, Leffondre K, Mathoulin-Pélissier S, et al. Alterations in comprehensive geriatric assessment decrease survival of elderly patients with cancer. Eur J Cancer 2018;90:10–8. [54] Corre R, Greillier L, Le Caër H, Audigier-Valette C, Baize N, Bérard H, et al. Use of a comprehensive geriatric assessment for the Management of Elderly Patients with advanced non-small-cell lung cancer: the phase III randomized ESOGIA-GFPCGECP 08-02 study. J Clin Oncol 2016;34(13):1476–83.