Individualized threshold of the involuntary weight loss in prognostic assessment of cancer

Abstract

Background

Involuntary weight loss (WL) is a common symptom in cancer patients and is associated with poor outcomes. However, there is no standardized definition of WL, and it is unclear what magnitude of weight loss should be considered significant for prognostic purposes. This study aimed to determine an individualized threshold for WL that can be used for prognostic assessment in cancer patients.

Methods

Univariate and multivariate analyses of overall survival (OS) were performed using Cox proportional hazard models. The Kaplan–Meier method was performed to estimate the survival distribution of different WL levels. Logistic regression analysis was used to determine the relationship between WL and 90‐day outcomes. Restricted cubic splines with three knots were used to examine the effects of WL on survival under different body mass index (BMI) conditions.

Results

Among the 8806 enrolled patients with cancer, median survival time declined as WL increased, from 25.1 to 20.1, 17.8 and 16.4 months at <2%, 2–5%, 5–10% and ≥10% WL, respectively (P < 0.001). Multivariate adjusted Cox regression analysis showed that the risk of adverse prognosis increased by 18.1% based on the SD of WL (5.45 U) (HR: 1.181, 95% CI: 1.144–1.219, P < 0.001). Similarly, categorical WL was independently associated with OS in patients with cancer. With the worsening of WL, the risk of a poor prognosis in patients increases stepwise. Compared with <2% WL, all‐cause mortalities were 15.1%, 37% and 64.2% higher in 2–5%, 5–10%, and ≥10% WL, respectively. WL can effectively stratify the prognosis of both overall and site‐specific cancers. The clinical prognostic thresholds for WL based on different BMI levels were 4.21% (underweight), 5.03% (normal), 6.33% (overweight), and 7.60% (obese). Multivariate logistic regression analysis showed that WL was independently associated with 90‐day outcomes in patients with cancer. Compared with patients with <2% WL, those with ≥10% WL had more than twice the risk of 90‐day outcomes (OR: 3.277, 95% CI: 2.287–4.694, P < 0.001). Systemic inflammation was a cause of WL deterioration. WL mediates 6.3–10.3% of the overall association between systemic inflammation and poor prognoses in patients with cancer.

Conclusions

An individualized threshold for WL based on baseline BMI can be used for prognostic assessment in cancer patients. WL and BMI should be evaluated simultaneously in treatment decision‐making, nutritional intervention, and prognosis discussions of patients with cancer.

Introduction

Involuntary weight loss (WL) is a common manifestation of nutrition disorder and an important symptom of malignancy. ¹ , ² It has been reported that as many as 80% of patients with advanced malignancy have involuntary WL. Involuntary WL is also a major cause of adverse outcomes in patients with cancer, ranging from 30% to 70% of cases. ³ , ⁴ , ⁵ Patients with cancer are often found to have a high prevalence of nutrition disorder due to the high consumption metabolism of the tumour and its eating disorders. Cancer‐related nutrition disorder, also known as cachexia, is a common complication of a variety of advanced malignancies. About 60–80% of patients with advanced malignancies may develop cachexia, and about 20% of patients with cancer die from cachexia. ⁶ As the most obvious manifestation of cachexia, involuntary WL is often used as a key qualitative factor in the diagnosis of cachexia. ⁷

Involuntary WL affects tolerance to chemoradiotherapy and is strongly associated with severe chemotherapy‐related toxicity. ⁷ , ⁸ , ⁹ Involuntary WL is also strongly associated with reduced physical function, quality of life, and shortened survival. ¹⁰ , ¹¹ Thus, involuntary WL is an important prognostic feature in patients with cancer. ¹² , ¹³ In addition, involuntary WL is an important criterion among several commonly used nutritional screening tools in clinics. ⁷ , ¹⁴ Involuntary WL can be a prognostic marker for cancer patients, predicting poor survival and increased morbidity.

Recently, it was proposed that WL severity should be assessed based on the %WL from the initial body reserve. Martin et al. ⁵ developed a WL grading system by combining involuntary WL and body mass index (BMI), confirming that WL has different prognostic values in patients with cancer with different BMI levels. Subsequently, many studies have proved the effectiveness of the WL grading system in the prognostic evaluation of patients with cancer. ⁸ , ¹⁵ , ¹⁶ However, the clinical thresholds of WL for different BMI reserves remain unclear. In addition, the degree of WL varies greatly among patients with different tumour types. At present, there is a lack of reliable evidence to comprehensively identify the degree of WL in different BMI levels and tumour types.

Systemic inflammation is one of the most typical interactions between tumours and hosts that affect the prognosis of cancer patients, and it is also an important criterion for diagnosing cachexia in cancer patients. ¹⁷ , ¹⁸ On the one hand, vigorous tumour growth consumes a large amount of nutrients, leading to weight loss in cancer patients. On the other hand, chronic inflammatory responses caused by tumours can result in excessive breakdown of fat and protein, leading to progressive wasting of the body. Systemic inflammation is closely related to WL and may be a cause of WL exacerbation. However, the relationship between systemic inflammation and WL in cancer patients has not been widely clarified.

Therefore, the purpose of this study was to evaluate the prognostic effect of involuntary WL in patients with a range of tumour types, and to determine the individualized threshold for WL based on different BMI levels and tumour types that can be used for prognostic assessment of overall survival (OS) and 90‐day prognosis in cancer patients. Furthermore, we also investigated the relationship between systemic inflammation and the progression of WL, and explored the relationship between systemic inflammation, WL and survival through mediation analysis.

Methods

Data source

The data in this study were obtained from the INSCOC cohort. ¹⁹ Patients who met the following criteria were included: (i) patients diagnosed with solid tumour at any stage; (ii) patients over 18 years old; (iii) patients who provide informed consent to participate in this study. Patients who met the following criteria were excluded: (i) patients with missing BMI data; (ii) patients with missing weight loss data; (iii) patients with incomplete baseline serological data (Figure S1).

Demographic characteristics included data on age, sex, smoking, alcohol consumption, hypertension and diabetes, tumour types, TNM stage, anti‐cancer therapy, serological test and family history. Systemic inflammation markers were assessed by serological test, including C‐reactive protein (CRP) level, neutrophil–lymphocyte ratio (NLR) [neutrophil (/μL)/lymphocyte (/μL)], inflammatory burden index (IBI) [=CRP (mg/dL) × NLR], and albumin level. Complete CRP data were available for 3490 patients. Hospitalization information included hospitalization length and hospitalization expenses. Nutritional screening tools included the Patient‐Generated Subjective Global Assessment (PG‐SGA), NRS2002 and cachexia. Cachexia was diagnosed according to Fearon's criteria in the 2011 International Consensus on cachexia. ⁷ Physical activity was assessed using the Karnofsky Performance Status (KPS). According to the demographic characteristics, we constructed three adjustment models: Model a (no adjusted); Model b (adjusted for age, sex, BMI, tumour types and TNM stage); and Model c (adjusted for age, sex, BMI, tumour types, TNM stage, surgery, radiotherapy, chemotherapy, hypertension, diabetes, smoking, alcohol consumption and family history).

Participants are followed up annually by telephone or outpatient visits. The last follow‐up was on 30 October 2020. The primary study endpoint was OS, defined as the period between registration and death from any cause (in months) or the last follow‐up of the patients under review (in months). The secondary outcome was the 90‐day outcome, defined as the survival outcome 90 days after receiving anti‐cancer therapy.

Measurement

The anthropometric measurements included height, current weight, previous weight, and WL. The patient's reported history of WL during the previous 6 months was assessed and recorded using PG‐SGA questionnaire and verified using the patient's medical records if possible. The percentage of WL was calculated as [(current weight in kg − previous weight in kg)/previous weight in kg] × 100%. Based on previous research, ²⁰ , ²¹ we categorized WL categories as follows: <2% (slight), 2–5% (mild), 5–10% (moderate) and ≥10% (severe). Baseline reserve levels were estimated using BMI, which was defined as the current weight/height‐squared (kg/m²). Baseline BMI categories of this study were defined based on the Chinese BMI classification reported by the World Health Organization and previous literature: underweight = BMI < 18.5, normal weight = BMI 18.5–24, overweight = BMI 24–28 and obese = BMI ≥ 28. ²²

Statistical analysis

The Kruskal–Wallis test was performed for quantitative variables and the χ2 or Fisher exact test for categorical variables. When exploring the distribution of weight loss (WL) among different tumour types, we employed the logarithmic transformation method (log) to make the WL data more closely approximate a normal distribution. This approach aimed to facilitate statistical inference and mitigate the impact of extreme WL values. The Kaplan–Meier method was performed to estimate the survival distribution of different WL levels. Univariate and multivariate survival analyses of OS were performed using Cox proportional hazards models. Logistic regression analysis was performed to determine the relationship between WL and 90‐day outcomes in patients with cancer.

Restricted cubic splines (RCS) with three knots were used to examine the effects of WL on the survival of patients with cancer having different BMI reserves, different tumour types or sex. In the RCS plot, the horizontal axis represents WL, and the vertical axis represents the survival log hazard ratio (HR) with 95% confidence intervals (CI). The clinical prognostic threshold of WL is determined based on the value of the horizontal axis when the survival log HR is equal to 0. It represents that beyond this value, there is a change in the risk of survival.

In addition, the Kruskal–Wallis test was used to examine the distribution of the systemic inflammation levels in different WL groups (using CRP, NLR, albumin and IBI). To investigate whether WL mediates systemic inflammation and cancer survival risk, mediating effect was used to assess mediation.

The statistical significance threshold was set to a P‐value of <0.05. All tests were two‐tailed, and all statistical analyses were performed using R version 4.0.5 (http://www.rproject.org).

Results

Study population characteristics and demographics

A total of 8806 cancer patients were evaluated based on the characteristics shown in Table S1; 57.4% of the patients were men, and the mean age was 58.46 (11.40) years. Lung cancer was the most common cancer (23.0%), followed by colorectal cancer (22.1%), and gastric cancer (15.7%). Most patients were at stage III (26.9%) or stage IV (42.7%). Of the patients, 14.2% had <2% WL, 29.5% had 2–5% WL, 33.1% had 5–10% WL and 23.2% had ≥10% WL. High WL was significantly associated with male sex, advanced age, advanced pathological stage, high inflammatory status, malnutrition, low physical status and adverse outcomes. Moreover, we found that hospitalization expenses were significantly increased in patients with ≥2% WL compared with patients with <2% WL.

Among all tumour types, the WL level in lung cancer was moderate. Compared with lung cancer, the level of WL in pancreatic, gastric, oesophageal, colorectal, and hepatic‐biliary cancers was up‐regulated (P < 0.001). There was no statistically significant difference in the level of WL among lung, gynaecological, urologic and nasopharynx cancers. In addition, compared with lung cancer, the level of WL in breast cancer was down‐regulated (P < 0.001). Therefore, with lung cancer as a reference, common tumour types were divided into three different types of WL: severe WL cancers (pancreatic, gastric, oesophageal, colorectal and hepatic–biliary cancers), moderate WL cancers (lung, gynaecological, urologic and nasopharynx cancers) and mild WL cancers (breast cancer) (Figure 1).

Figure 1.

Distribution of weight loss in different tumour types.

Survival analyses of weight loss categories

The median survival time of patients was 19.0 months. A total of 3698 patients experienced all‐cause death during the follow‐up period. As WL worsened, the patients' median survival times gradually declined, from 25.1 months of <2% WL to 16.4 months of ≥10% WL (25.1 months vs 20.1 months vs 17.8 months vs 16.4 months, P < 0.001) (Figure 2). Multivariate adjusted Cox regression analysis showed that the risk of adverse prognosis increased by 18.1% based on the SD of WL (5.45 U) (HR: 1.181, 95% CI: 1.144–1.219, P < 0.001). Similarly, categorical WL was independently associated with OS in patients with cancer. With the worsening of WL, the risk of a poor prognosis in patients increases stepwise. Compared with <2% WL, all‐cause mortalities were 15.1%, 37% and 64.2% higher in 2–5%, 5–10% and ≥10% WL, respectively (Table 1). Sensitivity analysis that excluded the first 3‐, 6‐, and 12‐month mortalities during follow‐up showed that WL is independently associated with all‐cause mortality. Compared with <2% WL, patients with ≥10% WL had a 47.8% higher risk of all‐cause mortality (Table S2).

Figure 2.

Kaplan–Meier curve of weight loss in patients with cancer.

Table 1.

Association between weight loss and overall survival of patients with cancer

Weight loss (%)	Model a	P value	Model b	P value	Model c	P value
Continuous (5.45 U)	1.203 (1.169, 1.239)	<0.001	1.165 (1.129, 1.203)	<0.001	1.181 (1.144, 1.219)	<0.001
Categories
<2%	Ref		Ref		Ref
2–5%	1.165 (1.041, 1.303)	0.008	1.149 (1.028, 1.286)	0.015	1.151 (1.029, 1.288)	0.014
5–10%	1.473 (1.323, 1.641)	<0.001	1.348 (1.21, 1.502)	<0.001	1.370 (1.230, 1.527)	<0.001
≥10%	1.771 (1.584, 1.98)	<0.001	1.587 (1.416, 1.778)	<0.001	1.642 (1.465, 1.840)	<0.001
P for trend		<0.001		<0.001		<0.001

Model a: No adjusted. Model b: Adjusted for age, sex, BMI, tumour types, TNM stage. Model c: Adjusted for age, sex, BMI, tumour types, TNM stage, surgery, radiotherapy, chemotherapy, hypertension, diabetes, smoking, drinking and family history.

In the survival analysis of pathological stage subgroups, WL could still effectively stratify the prognosis of patients with the same pathological stage, especially for those with stage III and stage IV disease (Figure S2). When analysing the association between WL and OS in site‐specific cancers, the results were similar to those found in the entire population, namely, the risk of poor prognosis increased progressively with the worsening of WL (Figure S3). Although the WL grades can distinguish the prognoses of patients with different BMI levels, slight and mild WL are increasingly inseparable as BMI level increases (Figure S4).

A total of 582 patients experienced 90‐day outcome. Multivariate logistic regression analysis showed that WL was independently associated with 90‐day outcomes in patients with cancer. Compared with patients with <2% WL, those with ≥10% WL had more than twice the risk of 90‐day outcomes (OR: 3.277, 95% CI: 2.287–4.694, P < 0.001) (Table 2).

Table 2.

Logistic regression analysis of weight loss associated with 90‐day outcomes

Weight loss (%)	Model a	P value	Model b	P value	Model c	P value
Continuous	1.359 (1.269, 1.456)	<0.001	1.359 (1.257, 1.468)	<0.001	1.381 (1.276, 1.494)	<0.001
Categories
<2%	Ref		Ref		Ref
2–5%	1.644 (1.147, 2.356)	0.007	1.610 (1.118, 2.318)	0.010	1.627 (1.128, 2.346)	0.009
5–10%	2.258 (1.598, 3.191)	<0.001	2.067 (1.456, 2.936)	<0.001	2.129 (1.496, 3.028)	<0.001
≥10%	3.365 (2.379, 4.758)	<0.001	3.133 (2.194, 4.476)	<0.001	3.277 (2.287, 4.694)	<0.001
P for trend		<0.001		<0.001		<0.001

Model a: No adjusted. Model b: Adjusted for age, sex, BMI, tumour types, TNM stage. Model c: Adjusted for age, sex, BMI, tumour types, TNM stage, surgery, radiotherapy, chemotherapy, hypertension, diabetes, smoking, drinking and family history.

Internal validation of the research cohort

Using a ratio of 1:1, the entire population was randomized into validation cohorts A (4404 cases) and B (4402 cases) (Table S3). In validation cohort A, WL still significantly affected the prognosis of patients with cancer (Figure S5A). Multivariate Cox regression analysis indicated that WL remained an independent prognostic factor for cancer patients (HR: 1.200, 95% CI: 1.144–1.219, P < 0.001) (Table S4). In validation cohort B, WL still had a good prognostic value in patients with cancer (Figure S5B). Multivariate Cox regression analysis showed that high WL was an independent risk factor for cancer (HR: 1.189, 95% CI: 1.137–1.243, P < 0.001) (Table S5).

Clinical prognostic threshold of weight loss

In the total population, WL was significantly associated with OS in patients with cancer in an inverted L‐shape. The clinical threshold for WL in the cancer population was 5.76% (Figure S6). Subsequently, multivariate adjusted RCS analysis of WL was performed based on the different BMI categories of the patients with cancer. The results showed a consistent positive correlation between WL and all‐cause mortality. With increasing BMI, the clinical threshold of WL showed an increasing trend. The clinical prognostic thresholds of WL for underweight, normal‐weight, overweight, and obese patients were 4.21%, 5.03%, 6.33% and 7.60%, respectively (Figure 3). Similarly, differences were found in the clinical prognostic thresholds of WL for different tumour types (Figure S7). The WL of upper‐digestive cancers had the highest clinical prognostic threshold (6.93%); breast cancer had the lowest clinical prognostic threshold (3.32%). In the sex subgroups, the clinical prognostic thresholds of WL were 6.35% for men and 5.31% for women (Figure S8).

Figure 3.

Restricted cubic splines of weight loss in patients with cancer based on different BMI categories.

Cross‐classification analysis of weight loss and body mass index

Cross‐classification analysis was performed between WL and BMI. Survival risk was calculated for 16 groups, using <2% WL with normal weight as a reference. In general, patients had an increased risk of a poor prognosis. It should be noted that the ‘obesity paradox’ was prevalent across the different WL groups. With the worsening of WL, the prognoses of patients with different BMI categories deteriorated gradually; the underweight patients had the largest worsening of prognoses. Patients with the worst prognoses were those who had ≥10% WL and were underweight. Interestingly, compared with patients in the <2% WL and normal weight group, patients in the 2–5% WL and overweight/obese groups had a relatively low risk of an adverse prognosis, suggesting that a high energy reserve has a protective effect (Figure 4). Sex subgroup analysis found that the general trend in the different sex subgroups was consistent with that of the overall population; however, men benefited more from the BMI energy reserve (Figure S9).

Figure 4.

Adjusted HRs of overall mortality according to BMI and weight loss. Note: Adjusted for age, sex, TNM stage, tumour types, surgery, radiotherapy, chemotherapy, hypertension, diabetes, smoking, drinking and family history.

Relationship between systemic inflammation and weight loss

The optimal threshold values were determined to be 3.11 for NLR, 38 for albumin, 3.63 for CRP, and 15.92 or IBI (Figure S10). Multivariate adjusted Cox proportional hazard models showed that these systemic inflammation biomarkers were independent factors affecting OS in patients with cancer (Table S6). The distribution of systemic inflammation at different WL grades indicated that systemic inflammation levels gradually increased with worsening WL. Compared with <2% WL, patients with ≥10% WL had higher levels of NLR, CRP, and IBI and lower levels of albumin (Figure 5). In addition, mediating‐effect analysis showed that WL mediated the overall association between systemic inflammation and OS in patients with cancer by 6.3–10.3% (Figure 6).

Figure 5.

Distribution of the systemic inflammation according to weight loss level. Note: (A) NLR; (B) ALB; (C) CRP; (D) IBI.

Figure 6.

Association systemic inflammation with weight loss in survival of cancer patients. Note: (A) NLR; (B) ALB; (C) CRP; (D) IBI.

Discussion

This research primarily encompasses two aspects of exploration. Firstly, we provided evidence to support that WL acts as an independent risk factor for OS and 90‐day outcomes in patients with cancer. Additionally, it effectively stratifies the prognoses of both overall and site‐specific cancers. Following that, our research based on pre‐defined WL categories revealed that as the degree of WL increases, the adverse prognosis risk in cancer patients gradually rises. In the second part of our study, we found that the extent of WL varies among different tumour types and different baseline BMI levels. Therefore, it is necessary to determine the individualized threshold for WL more accurately based on different BMI levels and tumour types. The RCS analysis demonstrated that the clinical prognostic thresholds of WL varied based on different BMI levels, namely 4.21% for underweight patients, 5.03% for normal weight patients, 6.33% for overweight patients, and 7.60% for obese patients. Furthermore, we also observed differences in the clinical prognostic thresholds of WL for different tumour types.

Because the ‘obesity paradox’ endows high BMI patients with a survival advantage, it is difficult to determine the clinical prognostic threshold of WL in patients with cancer. ²¹ , ²³ Thus, the severity of WL should be evaluated based on the level of WL and depletion of body reserves. ²⁴ In this study, RCS were employed to create fluid statistical models to determine the clinical prognostic thresholds of WL across various BMI classifications. The results revealed a gradual increase in the WL thresholds as BMI increased, consistent with the findings reported by Martin et al. ⁵ In addition, the ‘obesity paradox’ was prevalent in patients with different WL grades, and all‐cause mortality was significantly higher in patients with low BMI and high WL. Notably, men benefited more from BMI energy reserves than women. If people who already have low energy reserves experience significant WL, they may not be able to maintain muscle and fat mass, which can adversely affect their prognosis. Muscle mass is closely related to physical activity. The monotonous negative correlation between WL and physical activity (KPS) in this study further supports these results. Previous studies reported that fat loss is strongly associated with shorter survival times in cancer patients. ²⁵ , ²⁶ Based on these findings, we suggest that WL and BMI should be simultaneously evaluated in clinical practice.

The degree of WL also varies greatly among different tumour types. In this study, real‐world data were used for the first time to classify routine clinical tumours into three different WL tiers. We define severe WL cancer as pancreatic cancer, moderate WL cancer as lung cancer, and mild WL cancer as breast cancer. Furthermore, we determined the clinical prognostic thresholds of WL for different tumour types. Systemic inflammation is an important factor that affects the prognosis of patients with cancer and is also considered as one of the most typical host–tumour interactions. ¹⁷ , ¹⁸ WL may be one of the important external manifestations in the fight between the body and the tumour. On one hand, systemic inflammation directly impacts the appetite of cancer patients, leading to reduced food intake. On the other hand, systemic inflammation alters metabolic pathways, resulting in increased breakdown metabolism and decreased synthesis metabolism. ²⁷ In this study, in addition to finding that systemic inflammation may be a cause of WL exacerbation, mediation analysis showed that WL also mediates the effect of systemic inflammation on the adverse prognosis of cancer by 6.3–10.3%.

Our study has some limitations. First, we could not accurately measure lean body mass and fat mass using tools such as computed tomography, dual‐energy X‐ray absorptiometry and bioelectrical impedance analysis, which can provide additional information regarding the role of fat and muscle in survival. Second, given the wide range of tumour sites and stages, there was some heterogeneity in this study. Finally, although we achieved consistent results across the two internal validation cohorts, external validation is still necessary. Despite these shortcomings, the study still has important implications for clinical practice, highlighting the need for individualized assessments of weight loss in cancer patients as opposed to a standardized threshold value. The findings also underscore the importance of monitoring patients for involuntary weight loss throughout their cancer treatment to identify those who may require additional support and interventions.

Conclusions

The study suggests that an individualized threshold value for involuntary WL should be incorporated into prognostic assessments of cancer patients. Clinicians should be aware of the optimal threshold for WL in their patients and use it in clinical practice to improve prognostic accuracy. Systemic inflammation is one of the causes of WL deterioration in patients with cancer. WL mediates the effect of systemic inflammation on poor prognosis in patients with cancer.

Funding

This study was supported by the National Key Research and Development Program (No. 2017YFC1309200 and 2022YFC2009600) and Young Elite Scientists Sponsorship Program by CAST (2022QNRC001).

Conflict of interest

The authors have declared that no competing interest exists.

Supporting information

Figure S1. Flow chart.

Figure S2. Adjusted HRs of overall mortality according to different pathological stage.

Figure S3. The association between weight loss and overall survival in site‐specific cancers.

Figure S4. The association between weight loss and overall survival based on different BMI.

Figure S5. Kaplan–Meier curve of weight loss in patients with cancer at internal validation cohorts according to the ratio of 1:1.

Figure S6. Restricted cubic splines of weight loss in patients with cancer.

Figure S7. Restricted cubic splines of weight loss in patients with cancer based on different tumour types.

Figure S8. Restricted cubic splines of weight loss in patients with cancer based on different sex.

Figure S9. Adjusted HRs of overall mortality according to BMI and weight loss based on sex subgroups.

Figure S10. Cutoff of different systemic inflammation indexes.

Table S1. Characteristics by different level of weight loss in patients with cancer.

Table S2. The association of weight loss with all‐cause mortality after excluding the first 3, 6, 12 months mortalities of the follow‐up.

Table S3. Demographics of patients with cancer between validation cohort A and validation cohort B.

Table S4. Association between weight loss and overall survival of patients with cancer in the validation A.

Table S5. Association between weight loss and overall survival of patients with cancer in the validation B.

Table S6. The association of systemic inflammation indexes with all‐cause mortality.

Xie H., Zhang H., Ruan G., Wei L., Ge Y., Lin S., et al (2023) Individualized threshold of the involuntary weight loss in prognostic assessment of cancer, Journal of Cachexia, Sarcopenia and Muscle, 14, 2948–2958, doi: 10.1002/jcsm.13368

Data availability statement

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.