Supportive Care in Cancer

, Volume 21, Issue 12, pp 3261–3270

Handgrip strength predicts survival and is associated with markers of clinical and functional outcomes in advanced cancer patients


    • McGill Nutrition and Performance LaboratoryMcGill University Health Centre (MUHC)
    • Department of Exercise Science, The Richard J. Renaud Science ComplexConcordia University
  • A. Vigano
    • McGill Nutrition and Performance LaboratoryMcGill University Health Centre (MUHC)
    • Supportive and Palliative Care ProgramMUHC
  • B. Trutschnigg
    • McGill Nutrition and Performance LaboratoryMcGill University Health Centre (MUHC)
  • E. Lucar
    • McGill Nutrition and Performance LaboratoryMcGill University Health Centre (MUHC)
  • M. Borod
    • Supportive and Palliative Care ProgramMUHC
  • J. A. Morais
    • McGill Nutrition and Performance LaboratoryMcGill University Health Centre (MUHC)
    • Division of GeriatricsMUHC
Original Article

DOI: 10.1007/s00520-013-1894-4

Cite this article as:
Kilgour, R.D., Vigano, A., Trutschnigg, B. et al. Support Care Cancer (2013) 21: 3261. doi:10.1007/s00520-013-1894-4



Handgrip strength (HGS) has been shown to predict survival and is associated with changes in body composition, nutritional status, inflammation, and functional ability in several chronic disease conditions. Whether similar relationships exist between HGS and clinical outcomes in patients with advanced cancer are currently unknown. We evaluated the association between HGS and survival as well as several key markers of body composition (e.g., sarcopenia), subjective performance measures (e.g., quality of life), and muscle strength (e.g., isokinetic torque of the quadriceps) in patients with advanced forms of non-small cell lung and gastrointestinal cancers.


A consecutive cohort of 203 patients with advanced cancer was enrolled and categorized into three HGS percentiles (e.g., ≥50th, 25th, and ≤10th) according to published normative values. Multivariate regression analyses were used to test for independent associations between HGS and survival, sarcopenia, quality of life (QoL), and lower extremity muscle strength as well as key biological markers (e.g., hemoglobin, albumin, and C-reactive protein) while controlling for age, gender, cancer diagnosis, treatment (chemotherapy/radiotherapy), medications, and time from diagnosis to assessment.


When compared to HGS ≥50th, patients in the HGS ≤10th percentile had lower BMI (B, −2.5 kg/m2; 95% CI, −4.5 to −0.45), shorter survival (hazard ratio, 3.2; 2.0–5.1), lower hemoglobin (−19.70 g/L; −27.28 to −12.13) and albumin (−4.99 g/L; −7.85 to −2.13), greater occurrence of sarcopenia (odds ratio, 9.53; 1.95–46.55), lower isokinetic torque of the quadriceps at both 60°/s (−30.6 Nm; −57.9 to −3.3) and 120°/s (−25.1 Nm; −46.4 to −3.7), lower QoL (−1.6 on McGill Quality of Life Questionnaire scale; −2.5 to −0.6), higher levels of fatigue (18.8 on Brief Fatigue Inventory scale; 4.7 –32.9), poorer performance status (0.75 on Eastern Cooperative Oncology Group Performance Status scale; 0.34–1.15), lower fat mass (−7.4 kg; −14.4 to −0.5), and lower lean body mass (−6.5 kg; −10.3 to −2.8).


HGS is independently associated with survival and important biological, functional, and quality of life characteristics in advanced cancer patients. Patients presenting with very low percentiles with respect to their handgrip assessment may require timely referral to supportive and/or palliative care services.


HandgripSurvivalSarcopeniaQuality of lifeStrengthAdvanced cancer


Advanced cancer is typically associated with a multitude of inter-related clinical and functional abnormalities that are likely to have a significant impact on the quality of life (QoL) and survival of the patient. Conditions such as anorexia and anemia coupled with abnormal biochemical (e.g., increased pro-inflammatory markers) and hormonal profiles (e.g., insulin resistance and hypogonadism) all share a complex interaction that promotes fat loss and muscle wasting leading to significant weight loss, overall weakness, and chronic fatigue [1]. These conditions make up the cachexia syndrome; a collection of signs and symptoms that is difficult to define and is proving to be, thus far, extremely challenging to treat effectively. Although not every patient with advanced cancer develops cachexia, there is still a concentrated effort on behalf of researchers and clinicians to assess, evaluate, and to categorize patients so that those who are at particular high risk of developing cachexia can be identified early, in order to benefit from more timely and proactive treatments [2]. Weight loss has been a hallmark feature of cachexia; however, the decrease in body weight is not an absolute diagnostic criterion of cancer cachexia as evidenced by those patients with sarcopenic obesity [3]. Thus, complementary means of identifying and classifying the sequelae of cachexia such as sarcopenia (muscle wasting) and dynapenia (muscle weakness) should be more vigorously pursued.

At the present time, it is unknown if the loss of overall muscle strength occurs early in the disease progression and significantly contributes to QoL in advanced cancer. The ability to properly assess muscle strength and its changes using handgrip dynamometry in clinic and hospital settings may assist health professionals in determining the likelihood of increased morbidity and mortality. Although there are a variety of ways to measure and to assess muscle strength, the use of handgrip dynamometry has emerged as a valid and reliable tool that is known to represent overall body strength [4]. Handgrip strength (HGS) offers the benefits of being relative simple to perform and can be measured in either laboratory or hospital settings [5]. This technique has allowed for large cohorts of HGS data to be obtained and to make predictions concerning key health indictors, such as functional capacity and integrity, especially with aging [6, 7] and clinical populations where malnutrition and cachexia are prevalent [810].

Furthermore, HGS was shown to have predictive power with respect to morbidity and mortality in clinical cohorts. For example, there is a growing body of longitudinal evidence from several large population studies indicating that poor HGS predicts mortality from all-causes, independent of age [6, 7, 11, 12]. There is also evidence that HGS is an independent prognostic indicator in prospective cross-sectional follow-up studies of patients with chronic kidney disease undergoing peritoneal dialysis [10].

The commonality among several clinical studies, including patients with chronic renal failure [13] on hemodialysis [1416] and those with rheumatoid arthritis [17], describes links among the state of malnutrition and general muscle weakness. These associations are also seen in the elderly whose nutritional intake, body composition, and strength status are frequently compromised due to age, frailty, and multiple comorbidities [18]. Taking into account these associations, it is reasonable to assume that other clinical conditions that are defined by chronic malnutrition, cachexia, and reductions in lean and fat mass could also use HGS as a surrogate predictor of survival and sarcopenia. These relationships are particularly prevalent in patients with cancer and especially those patients with more advanced primaries with or without the presence of cachexia [5, 8, 9]. This study focuses on whether categorizing patients according to HGS percentiles reliably identifies survival, sarcopenia and other relatively common biological (e.g., hemoglobin, albumin, and C-reactive protein), functional (e.g., sit-to-stand and quadriceps muscle strength), and subjective (e.g., fatigue and quality of life) characteristics in a cohort of advanced cancer patients.

Patients and methods

This prospective study received ethical approval from the McGill University Institutional Review Board and is in accordance with the Declaration of Helsinki. Patient recruitment took place between March 2006 and November 2007. Patients diagnosed within the previous 6 months with either locally advanced, metastatic or recurrent non-small cell lung, or gastrointestinal (GI) cancers were eligible for recruitment. A consecutive cohort of 245 patients (≥18 years old) who were either admitted or attending the oncology clinics at the McGill University Health Centre (Montreal General Hospital and Royal Victoria Hospital) were approached by their oncologists about the study. Two hundred three patients provided written informed consent and were then assessed in two separate settings: first at the hospital bedside (n = 203) and later at the McGill Nutrition and Performance Laboratory (MNUPAL; n = 94), an outpatient facility designed for evaluating human nutritional and functional status (see All data and information such as age, survival, gender, treatment regime, medication profile, and hospitalization follow up were stored in the Human Cancer Cachexia Database (HCCD) at MNUPAL [2]. All patients in the HCCD were followed from baseline (initial assessment) until their death or study termination. The median survival for all participants was 31.8 weeks (95 % CI; 23.2–40.5). Forty-nine patients (23.4 %) were still alive at study termination.

Hospital bedside evaluations

Each patient completed the Edmonton Symptom Assessment System (ESAS) for symptom profiling. Patients also completed an Eastern Cooperative Oncology Group Performance Status (ECOG). Handgrip strength was assessed using Jamar® dynamometry (Sammons Preston, Bolingbrook, IL, USA). A single intravenous blood sample was obtained for analysis of the following routine laboratory markers: complete blood count and differential count, albumin, hemoglobin, cholesterol as measured by apolipoproteins (Apo) A and B, and C-reactive protein (CRP). The presence of anemia, elevated white blood cells (WBC) and CRP, low albumin, and low cholesterol have been suggested as biological markers for the identification and classification of cancer cachexia [1].

The ESAS is a valid and reliable test comprised of nine questions with a 0–10 point scale for each question [19]. This test assesses the severity of symptoms experienced at the time of assessment on a scale of “0” (no symptoms) to “10” (worst symptoms). Symptoms include pain, tiredness, drowsiness, nausea, lack of appetite, shortness of breath, depression, anxiety, and wellbeing.

The use of the Jamar® handgrip dynamometer to measure grip strength has been shown to be valid and highly reliable [20] and has been recently validated in advanced cancer patients [5]. This measure of upper limb strength is a good surrogate of generalized muscle strength, as it correlates well with lower limb strength in this patient population and in healthy adults [21].

Handgrip dynamometry protocol

Each participant was seated in a chair with both feet touching the ground and the test arm placed comfortably at 90° on an armrest. The nondominant arm rested in a neutral position by the side of the participant. All participants underwent a familiarization period that consisted of one to two trial attempts. Grip settings were set at position 3 of the dynamometer that corresponds to 5.1 cm. This setting is a standard testing position approved by the American Society of Hand Therapists [22]. Using the dominant hand, three maximal performances were measured, each with a 3-s contraction duration and a 1-min rest interval between each test. The participant was instructed to start and stop the contractions; however, no other verbal encouragement or cues were given.

Laboratory assessments at MNUPAL

Patients who were willing and able to commute to MNUPAL completed the following tests within 1 week after their bedside assessment: the Brief Fatigue Inventory (BFI), the McGill Quality of Life Questionnaire (MQoL), body composition measurements using dual-energy X-ray absorptiometry (DXA), the sit-to-stand test (STS), and quadriceps strength using Biodex® isokinetic dynamometry. Even though these tests provide clinically useful information regarding patients' functional and psychological status, they are impractical to complete on a routine basis in a clinic or a hospital setting.


The BFI assesses level of fatigue and its impact on activities of daily living [23, 24]. The test has nine questions. Three questions are designed to assess patients' fatigue during the immediate waking hours. The other six questions address how fatigue has interfered in the patients' lives over the previous 24 h. Each question uses a scale rating from “0” (no fatigue) to “10” (unimaginable fatigue) for a total of 90 points. The MQoL assesses psychological, physical, support, and existential status on a scale of “0” (worst problem) to “10” (no problem) [25].

DXA measurements

Fat mass and fat-free mass were measured using the DXA Lunar Prodigy Advance Direct-Digital Densitometer (GE Healthcare, Madison, WI, USA). The use of DXA for measuring body composition in advanced cancer patients was recently validated showing that it is a precise instrument in the clinical assessment for sarcopenia in these patients [5]. Sarcopenia was calculated using the male and female cutoffs as determined by Baumgartner et al. [26]; ≤7.26 kg/m2 of appendicular (arm + leg) skeletal muscle mass/height2 and ≤ 5.45 kg/m2 of appendicular (arm + leg) skeletal muscle mass/height2, respectively.

Leg strength

Knee extensor (quadriceps) muscle strength was assessed using isokinetic dynamometry and the STS, a simple and reliable functional test that can be performed in almost any setting [27]. The peak isokinetic strength of the quadriceps muscle group was measured via a Biodex® System 3 (Biodex Medical Systems, Shirley, NY, USA) device using a standardized protocol. This protocol has been shown to be valid and reliable in measuring and eliciting peak isokinetic strength in healthy subjects [28, 29] and obtaining lower limb strength measures in patients with advanced cancer [30].

Isokinetic quadriceps strength protocol

All subjects were seated in an inclined position (approximately 80–90°) on the Biodex chair with stabilizing straps secured firmly across the shoulders and hips. The chair was adjusted so that the knee joint was aligned with the rotational axis of Biodex leg attachment. The leg was firmly secured to the extension attachment just above the ankle. Pre-assessment range of motion was determined by maximally extending followed by lowering the leg to approximately 90° of flexion. Prior to the practice repetitions, each subject was instructed on the protocol. During the assessments, subjects were permitted to use handrails on either side of the Biodex. One to two practice repetitions at the testing velocities were permitted. The protocol consisted of two sets of five repetitions at angular velocities of 60° and 120° with 1 min rest between sets. Subjects were instructed to perform the knee extensions as fast as possible and to begin when ready. The laboratory evaluator counted down the repetitions while reminding the subject to move as fast as possible. During the rest interval, subjects extended and then flexed their leg to reconfirm their preset range of motion. The greatest peak torque from the five repetitions of both sets and velocities were recorded and used for subsequent analysis.

Statistical analysis

All data were analyzed using SPSS (version 14.0, 2005, SPSS Inc., Chicago, IL, USA) [31]. In the paper published by Mathiowetz et al. [32], normative HGS data have been provided for healthy men and women ranging in age from 20 to 75+ years old. From this data, five handgrip strength percentile categories (e.g., ≤10th, 25th, 50th, 75th, and ≥90th) were generated. Upon visual examination of our handgrip data scores from the advanced cancer patients, we observed that all scores could be included in the following three percentile categories: (1) ≤10th, (2) 25th, and (3) ≥50th. Since there are no pre-established HGS categories for advanced cancer patients, we made the decision to group our patients into these three percentiles that best reflected their HGS performance scores according to the average norms. We determined the relationships between these HGS categories (independent variables) with clinical and laboratory characteristics (Table 2), subjective assessments from questionnaires (Table 3), and measures of body composition and muscle strength (Table 4), which were treated as dependent variables in all analyses. A one-way ANOVA was used to compare means for all continuous measures. Chi-square and Kaplan–Meier with log-rank tests were used to analyze any differences in percentages and survival curves, respectively. In order to test for independent relationships between HGS (independent variable) with the above dependent outcome variables, we performed several multivariable linear, logistic, Cox regression models (one for each dependent variable), controlling for gender, age, time between diagnosis and baseline assessment, survival time from baseline assessments to death/loss to follow-up (variable not included in the survival analyses), diagnosis of lung vs. gastrointestinal malignancies, presence/absence of concurrent oncological treatment (radio/chemotherapy), and concurrent medications. For medications potentially impacting on the relationships under investigation, the presence or absence of at least one of the following medications was considered: statins, anti-inflammatories, steroids, angiotensin-converting enzyme inhibitors, antihormonal agents, anti-oxidants, essential amino acids, anabolic hormones, and metformin.

In Table 2, there are 12 dependent variables listed (BMI, WBC, lymphocytes, Hgb, CRP, Apo A, Apo B, number of administration/30 day follow up, number of days admitted/follow-up days, chemotherapy cycle dose reduction, and survival). Hence, 12 multivariate regression models were created for Table 2. In Table 3, seven regression models were run for each of the seven dependent variables (three from the ESAS (QoL, strength, and appetite), ECOG, BFI, and two from the MQoL (physical score and total score)). In Table 4, a total of nine regression models were created for nine dependent variables (appendicular fat, total fat mass, percent body fat, appendicular lean, total lean mass), sarcopenia, and three variables representing quadriceps muscle strength performance (sit-to-stand, isokinetic peak torque at 60 and at 120°/s)). Because HGS was categorized in three groups (HGS ≥ 50th percentile, HGS = 25th percentile, and HGS ≤10th percentile), we created two “dummy” variables (X1 and X2) and inserted them in each of our regression equations. We decided on our reference category to be the HGS ≥ 50th percentile. This category was represented by the other two “dummy” variables, both being equal to zero. We did make X1 = 1 for HGS 25th percentile and X2 = 1 for HGS ≤10th percentile. For each of these, we compared the category in question to the HGS ≥ 50th percentile category (our reference). So, in each of our models, we tested the regression coefficient (B) associated with X1 and X2. If the regression coefficient was determined to be “significant” according to its 95 % confident interval, then we interpreted this to mean that X1 (HGS 25th percentile vs. HGS ≥ 50th percentile) and/or X2 (HGS ≤10th percentile vs. HGS ≥ 50th percentile) contributed to the model's explanatory power and the values in the HGS categories were independently related to the dependent variable being compared. A more detailed explanation about the use of dummy variables in our regression models can be found in the legend for Table 2.


Two hundred three patients from the Human Cancer Cachexia Database at MNUPAL were included. All patients had advanced disease, with 64 % showing evidence of distant metastases (Table 1). Two thirds of the participants had GI cancers with pancreatic cancer being the most common primary. Slightly more than 70 % of the patients demonstrated HGS scores below the 50th percentile with 27 % of the patients in the lowest percentile (HGS ≤10th). On average, the HGS ≤10th group had significantly lower BMI and ECOG performance scores than the reference group (HGS ≥50th). Both the HGS 25th and ≤10th percentiles were associated with significant survival differences in the Kaplan–Meier curves (Fig. 1).
Table 1

Patient demographics and clinical characteristics



HGS ≥50th

HGS 25th

HGS ≤10th



Mean ± SD

Mean ± SD

Mean ± SD

Mean ± SD


N = 57

N = 91

N = 55

N = 203

Age (years)


66.1 ± 10.2

63.9 ± 12.7

63.1 ± 15.2

64.3 ± 12.8

BMI (kg/m2)


25.5 ± 5.4

24.6 ± 5.7

22.1 ± 3.7*

24.2 ± 5.3

Time from diagnosis to baseline assessment (days)

42.7 ± 61.7

91.8 ± 135.0

62.1 ± 113.7

69.3 ± 113.6

Survival (weeks)

51.9 ± 36.6

33.9 ± 28.2

23.3 ± 22.5*

36.1 ± 31.3


Percent (n)

Percent (n)

Percent (n)

Percent (n)



14.8 (30)

30.0 (61)

12.3 (25)

57.1 (116)


13.3 (27)

14.8 (30)

14.8 (30)

42.9 (87)

Cancer staging

Locally advanced

13.8 (28)

14.8 (30)

6.9 (14)

35.5 (72)


14.3 (29)

29.1 (59)

19.7 (40)

63.1 (128)

Tumor type


13.3 (27)

15.8 (32)

6.4 (13)

35.5 (72)


14.8 (30)

29.1 (59)

20.6 (42)

64.5 (131)

Concurrent treatments


45.6 (26)

34.1 (31)

21.8 (12) *

34.0 (69)


15.8 (9)

18.7 (17)

10.9 (6)

15.8 (32)


47.4 (27)

46.6 (41)

43.6 (24)

46.0 (92)

BMI body mass index, NSCLC nonsmall cell lung carcinoma

*p ≤ 0.05, HGS ≤10th percentile significantly different from HGS 25th and HGS ≥50th percentiles

aPatients taking at least one of the following medications: statins, anti-inflammatories (both steroidal and nonsteroidal), angiotensin-converting enzyme inhibitors, antihormonal agents, anti-oxidants, essential amino acids, anabolic hormones, and metformin
Fig. 1

Kaplan–Meier survival curves for advanced cancer patients grouped according to handgrip strength percentiles. Blue line (1.00) patients in the HGS ≥50th percentile, green line (2.00) patients in the HGS 25th percentile, yellow line (3.00) patients in the HGS ≤10th percentile

In the multivariate analyses, patients in the HGS ≤10th percentile category had hemoglobin and albumin values that were 20 and 5 g/L lower than the reference group, respectively (Table 2). When compared to the reference group, patients in the HGS ≤10th percentile category expressed greater weakness, less appetite, poorer functional performance on the ECOG, greater fatigue, and poorer QoL (Table 3).
Table 2

Multivariable regression models describing the association of handgrip strength with clinical, laboratory, and hospital follow-up characteristics of advanced cancer patients

Dependent variables

Handgrip 25th percentile

Handgrip ≤10th percentile


95 % CI


95 % CI




−1.99 to 1.62


−4.53 to −0.42

Laboratory parameters (n = 203)

WBC (×109/L)


−1.27 to 2.29


−1.39 to 2.63

Lymphocytes (×109/L)


−0.35 to 0.19


−0.43 to 0.19

Hgb (g/L)

−9.76 **

−16.48 to −3.03


−27.28 to −12.13

CRP (mg/L)


4.69 to 38.51


−6.73 to 31.41

Albumin (g/L)


−6.76 to −1.69


−7.85 to −2.13

Apo A (g/L)


−0.26 to 0.05


−0.28 to 0.08

Apo B (g/L)


−0.18 to 0.08


−0.12 to 0.19

Hospitalization (n = 203)

No. of administration/30 follow-up days


−0.21 to 0.77


−0.36 to 0.74

No. of days admitted/follow-up days (%)


−2.83 to 15.25


−0.62 to 19.75



95 % CI


95 % CI

Chemotherapy toxicity (n = 203)

Any cycle dose reduction (Y/N)


0.37 to 8.71


0.17 to 7.49



95 % CI


95 % CI

Survival (n = 203)



1.26 to 3.04


2.0 to 5.1

Reference category is HGS ≥50th percentile. Each multivariate model was created to test for the independent relationship between HGS percentiles (independent variable) with each clinical, laboratory, and hospital characteristics (dependent variables) while controlling for gender, age, cancer diagnosis, treatment (radio/chemo), survival (in weeks), medications (list reported in text), and time from diagnosis to assessment as potential confounding factors. Sample regression with “dummy” variables: Y = A + B(X1) + B(X2) + B(X3) + B(X4) + B(Xx), where Y is the dependent variable of interest, A is the intercept, B is the regression coefficient, X1is the dummy variable for HGS 25th percentile (X1 = 1 for the 25th percentile and X1 = 0 for the HGS ≤ 10th percentile), X2 is the dummy variable for HGS ≤ 10th percentile (X2 = 1 for the HGS ≤ 10th percentile and X2 = 0 for the HGS 25th percentile), X3 is age as a continuous variable, and X4 is gender where “1” = male and “0” = female, and Xx represents the remainder of the continuous or dichotomous independent variables. The HGS ≥ 50th percentile has been omitted from the regression equation because this becomes our reference category against which the effects of the other categories are assessed. Thus, we can interpret the results or the difference between each category and the omitted or reference category

B unstandardized regression coefficient for linear regression models; OR odds ratio from multivariate logistic regression; HR hazard ratio; 95 % CI confidence interval, WBC white blood cell count, CRP C-reactive protein, Hgb hemoglobin

*p < 0.05; **p < 0.01; ***p < 0.001

Table 3

Multivariable regression models describing the association of handgrip strength with the questionnaires administered to the advanced cancer patients

Dependent variables

Handgrip 25th percentile

Handgrip ≤10th percentile


95 % CI


95 % CI

Symptom severity ESAS (0–10; 10 worst) (n = 203)

QoL score (0–10)


0.06 to 1.98


−0.05 to 2.10

Strength score (0–10)


0.50 to 2.40


0.58 to 2.71

Appetite score (0–10)


0.33 to 2.59


0.12 to 2.67

ECOG PS (0–4; 4 worst) (n = 203)

Functional score (0–4)


0.22 to 0.94


0.34 to 1.15

BFI (0–90; 90 worst) (n = 94)

Total score (0–90)


−7.43 to 17.13


4.71 to 32.89

MQoL (10–0, 0 worst) (n = 94)

Physical score (10–0)


−1.76 to 1.00


−4.12 to −0.85

Total score (10–0)


−1.13 to 0.47


−2.54 to −0.64

Reference category is HGS ≥50th percentile. Each multivariate model was created to test for the independent relationship between HGS percentiles (independent variable) with each questionnaire (dependent variables) while controlling for gender, age, cancer diagnosis, treatment (radio/chemo), survival (in weeks), medications (list reported in text), and time from diagnosis to assessment as potential confounding factors

B unstandardized regression coefficient for linear regression models, 95 % CI confidence interval, ESAS Edmonton Symptom Assessment System, ECOG PS Eastern Cooperative Oncology Group Performance Status, BFI Brief Fatigue Inventory, MQoL McGill Quality of Life Questionnaire

*p < 0.05; **p < 0.01; ***p < 0.001

Total lean and fat mass were significantly lower in the HGS ≤10th when compared to the reference group (Table 4), whereas the incidence of sarcopenia, as calculated from the appendicular lean mass via DXA, was nearly tenfold higher. Regarding lower limb functional performance in the HGS ≤10th vs. ≥50th percentile, the isokinetic peak torque of the quadriceps at 60 and at 120°/s was significantly reduced in the weakest group by 31 and 25 Nm, respectively. Finally, patients grouped in the HGS ≤10th percentile had the highest probability of dying during follow-up.
Table 4

Multivariable regression models describing the association of handgrip strength with body composition, sarcopenia, and lower limb strength of advanced cancer patients

Dependent variables

Handgrip 25th percentile

Handgrip ≤10th percentile


95 % CI


95 % CI






DXA (n = 94)

Appendicular fat (kg)


−3.58 to 0.89


−5.25 to 0.15

Total fat mass (kg)


−8.75 to 2.64


−14.35 to −0.54

% Fat


−7.71 to 3.75


−11.26 to 2.64

Appendicular lean (kg)

−1.72 *

−3.25 to −0.19


−5.89 to −2.21

Total lean mass (kg)


−6.02 to 0.18


−10.29 to −2.78






Sarcopenia (n = 94)



0.55 to 6.95


1.95 to 46.95






Quadriceps muscle strength (n = 94)

Sit-to-stand (s)


−0.76 to 1.57


−0.58 to 2.21

Isokinetic peak torque (60°/s; Nm)


−44.21 to 2.73


−57.87 to −3.32

Isokinetic peak torque (120°/s; Nm)


−39.04 to −2.36


−46.36 to −3.73

Reference category is HGS ≥50th percentile. Each multivariate model was created to test for the independent relationship between HGS percentiles (independent variable) with each measure of body composition, sarcopenia, and lower limb strength (dependent variables) while controlling for gender, age, cancer diagnosis, treatment (radio/chemo), survival (in weeks), medications (list reported in text), and time from diagnosis to assessment as potential confounding factors

B unstandardized regression coefficient for linear regression models, OR odds ratio from multivariate logistic regression, 95 % CI confidence interval, DXA dual-energy X-ray absorptiometry, Nm Newton meters

*p < 0.05; **p < 0.01; ***p < 0.001


This study is the first to link HGS with survival in advanced cancer patients. By categorizing patients into HGS percentiles (e.g., ≤10th, 25th, and ≥50th), it was found that HGS predicted survival independent of gender, age, time between diagnosis and baseline assessment, survival time from baseline assessment to death/loss to follow-up, presence of tumors known to present more frequently with cachexia (i.e., lung, pancreatic and gastric cancers), concurrent oncological treatment (radio/chemotherapy), and concurrent medications. We have also shown that as HGS worsens from the ≥50th to the ≤10th percentiles, there was a significant lowering of BMI, hemoglobin, and albumin, as well as increased subjective scores on the ESAS, ECOG, and BFI. The greatest losses of lean and fat mass along with the highest incidence of sarcopenia and lowest isokinetic quadriceps strength were observed in the weakest HGS percentile group. Taking into account the directional trends in clinical laboratory markers, body composition, and the patients' responses to the questionnaires, the selected HGS percentiles effectively discriminate among those patients according to the severity of their overall clinical and functional profiles.

In terms of clinical and research implications, reduced muscle strength is one of the many phenomena associated with chronic disease states characterized by wasting. In advanced cancer, muscle weakness along with chronic fatigue, low hemoglobin levels, elevated pro-inflammatory cytokines, and significant decreases in muscle and fat mass resulting in weight loss have all been implicated with cancer cachexia [1]. Thus, based upon the findings of this study, handgrip strength can be seen as a suitable measure for the prognostic determination of morbidity and mortality in an advanced cancer population.

Although some studies in patients on hemodialysis [16] and those with rheumatoid arthritis [17] used HGS normalized data for age- and gender-matched healthy individuals, HGS has never been used previously to predict survival in advanced cancer patients using a population-based set of reference values of healthy North American adults [32]. The use of HGS dynamometry has become a ubiquitous functional test to predict future complications arising from a wide variety of disease states and conditions, especially those that involve cachexia and malnutrition due to chronic kidney disease [13], Crohn's disease and ulcerative colitis [33], cancer [8, 9], cirrhosis [34], peritoneal dialysis [10], hemodialysis [14, 15], as well as survival in several longitudinal studies [4, 7, 11, 12, 35, 36]. We are frequently reminded that HGS is a preferred performance measure because it is simple and rapid to implement, inexpensive, non-invasive, objective, and significantly correlated with overall body strength [4, 37]. From an outcomes measure perspective, it has been shown to provide essential prognostic information independent of other possible covariates [12]. Thus, it is not surprising that HGS is the preferred strength measure for both cross-sectional and epidemiological studies.

A possible reason why HGS has become such an important component of patient assessment may be the fact that muscle strength is typically a more relevant determinant of survival than muscle size [7]. Dynamic leg strength (e.g., power output) is significantly reduced when comparing advanced cancer patients with cachexia to healthy controls [30]. Differential leg power outputs have also been observed between different cancer cohorts (e.g., cachectic vs. noncachectic) within the same study [38]. In the present study, we have also demonstrated a decrease in peak torque (power) associated with decreased HGS in advanced cancer. In fact, those with the poorest survival (e.g., HGS ≤10th percentile group) demonstrated the lowest isokinetic power of the quadriceps as evidenced by the 30 and 25 Nm reductions in peak torque at 60 and 120°/s, respectively. Norman et al. [8] also described a similar correlational relationship between upper (HGS) and lower (isokinetic peak torque) extremity strength, whereby those in the weakest HGS group appear to have the poorest quadriceps strength. Thus, there appears to be a uniform weakness in the upper and lower extremities which is not surprising considering that HGS correlates with lower extremity knee extension strength in healthy elderly persons [4]. However, it should be noted that this has not always been a consistent finding in advanced cancer patients [38].

Although we have emphasized the importance of the relationship between strength and function, we cannot ignore the impact that muscle mass, or lack thereof (e.g., sarcopenia), has on survival. Sarcopenia is classically defined as a progressive loss of muscle mass, which is separate from dynapenia or the loss of strength that typically becomes more prevalent and accelerated with advancing age [26]. In several disease states, body composition can change dramatically with significant losses of both lean and fat mass, as observed in this and other studies involving advanced cancer patients [5, 30]. Decreases in appendicular lean mass could signal the strength decrements in extremities that have been observed on several occasions in healthy aging [6, 7] and advanced cancer [30, 39]. An overall reduction in HGS has been observed in advanced cancer; however, it has not been previously shown in this population that HGS measures are sensitive to the appearance of sarcopenia. In the present study, sarcopenia was almost 10 times more prevalent in the weakest group (e.g., HGS ≤10th percentile), whereas no differences were observed when comparing the 25th to the ≥50th percentile group. This finding highlights the importance of categorizing HGS in identifying the possible presence of sarcopenia.

In the present study, the McGill quality of life (MQoL) scores that reflect physical functioning and overall QoL were lowest in the weakest group (HGS ≤10 %) when compared to the 25th and ≥50th percentile groups. Both clinical and research evidence is accumulating with regards to the relationship between HGS and QoL. Jakobsen et al. [40] demonstrated that HGS is a valid measurement of QoL in healthy individuals and hospitalized patients. In other patient populations, Norman et al. [41] assessed nutritionally-deficient individuals with noncancerous GI disease and showed improved muscle strength and QoL following supplementation. With reference to a cancer population, muscle function, as assessed by HGS and limb strength, correlated with QoL functional status [8]. Similarly, in male patients with GI tract cancer cachexia (WL >10 %), poorer global QoL was related to concomitant decreases in physical functioning, quadriceps strength and power, muscle quality as well as increased levels of chronic fatigue [38]. However, in the latter two studies, cancer-related malnutrition is clearly a confounding factor in the relationship between HGS and QoL.

Our findings can lead to some important clinical implications. HGS could be used as a proxy by clinicians for the early identification of impairment or issues related to QoL. Also, patients with lower HGS may exhibit higher symptom burden. For these patients, early referrals through supportive or palliative care programs are indicated. For researchers, better identification of cancer cachexia stages is needed as these stages may correlate well with functional status and muscle mass in these patients [42]. Vigano et al. [2] have identified a lower HGS as a critical measure for differentiating patients who are in refractory cachexia versus those who are either in an earlier or less severe stage of cachexia.

There are likely several confounding factors other than decreases in muscle mass that could also explain the decrease in HGS. First, a portion of the participants was undergoing chemotherapy and the potential for peripheral neuropathy with related loss of grip strength cannot be excluded. However, in our multivariate models, we have adjusted our analyses for the presence of concurrent oncological treatments such as chemotherapy or radiotherapy. Nevertheless, future studies could further address the role of particular neurotoxic chemotherapies such as taxanes or the newly targeted treatments. Secondly, both peripheral (e.g., skeletal muscle contractile properties) and central (e.g., brain, spinal cord, α-motoneurons) factors combine to regulate neural drive and activation of the skeletal muscles that could possibly contribute to HGS performance. Thus, the voluntary drive and/or activation of skeletal muscles, especially those responsible for motor control of forearm flexion, may be suppressed. Another possible factor is cancer-related fatigue, which is frequently reported and could very well have a significant negative impact on HGS performance. We have noted previously that fatigue is negatively correlated to muscle mass and performance [39]. Fatigue, albeit peripheral or central, could have a negative impact on generating a maximal voluntary handgrip contraction. Although there are a paucity of studies examining the central and peripheral input regulating maximal voluntary contractions such as HGS, there is emerging evidence to suggest that exhaustion from submaximal efforts (30 % max voluntary contraction of the biceps brachii) occurs earlier in patients with advanced cancer when compared to healthy controls and that this fatigue is related more to central as opposed to peripheral mechanisms or those more distal to the neuromuscular junction [43]. In this study, it is important to highlight the fact that the weakest HGS group also experienced a significantly greater level of fatigue as evidenced by scoring approximately 19 points higher on the BFI when compared to the reference group. Whether or not fatigue is the cause for the decrease in HGS is currently unknown, but it may play a critical role in the functional performance of these patients.


Our study has several limitations. We included a heterogeneous population in regard to type of cancer diagnosis, time of assessment from diagnosis of advanced disease, and survival time from baseline assessment. Patient heterogeneity in terms of time from diagnosis and survival from baseline assessments typically reflects the usual clinical practice. In order to address this heterogeneity, all multivariate models were controlled for both time between diagnosis and baseline assessment, and survival time from baseline assessments to death/loss to follow-up. The latter variable was not included in the survival analyses.

We recruited cancer patients who were treated at two tertiary cancer care centers and most of the patients presented with stage III or IV lung, pancreatic, and gastric cancer, which are more prone to develop the anorexia-cachexia syndrome. This cohort may not be representative of cancer patients followed elsewhere; therefore, our results may not be applicable to patients who are followed within other community hospitals.

For those patients (n = 94) who were measured at MNUPAL, there is a probability of a type II error whereby we may not have found a difference when in fact one existed. However, for all the dependent variables or clinical outcomes that we regressed on the HGS percentiles via multivariate models on the 94 patients who came to MNUPAL, we found statistically significant results.


Early detection and categorization of handgrip performances in advanced cancer patients may provide clinicians with useful prognostic clues that may direct them to the timely implementation of appropriate therapeutic measures. Those patients who present with the lowest HGS (i.e., ≤10th percentile) clearly need timely referral to palliative care. However, patients in the HGS 25th percentile may be an important group of interest, where individual differences could be most predictive of future mortality and ultimately demonstrate a greater potential for stability or improvement in strength. In this group, improving overall muscle strength could be a possible way to reduce the likelihood of mortality. Despite the well-known benefits of progressive resistance training (PRT) towards preventing, delaying, and reversing muscle strength decline in healthy, older adults, it is unknown whether better overall muscle strength translates into a better quality of life in advanced cancer patients. These findings indicate that there are critical points in the decline of muscle strength where targeted interventions would be most effective. Whether PRT or some other regimen of multimodal intervention is effective in improving muscular strength and ultimately survival is the subject of future investigations.


This work was supported by the Canadian Foundation for Innovation (CFI)-New Opportunities Award (AV), Canadian Institute for Health Research (CIHR)-Operating Grant (AV), McGill University Health Centre Research Institute-Pilot project Competition (AV), Canada Graduate Scholarships-CIHR Master's Award (BT), and the Academic Unit Head Research Fund of Concordia University (RK)

Copyright information

© Springer-Verlag Berlin Heidelberg 2013