Longitudinal associations of bioelectrical phase angle and fatigue in breast cancer patients

Cancer‐related fatigue is commonly treated in an undifferentiated manner, because its pathophysiology is still not well understood. Therefore, we investigated if bioelectrical phase angle (PhA), a non‐invasive marker of cell integrity, could help to single out specific fatigue subtypes. In a randomized controlled strength training intervention trial, PhA was measured by bioelectrical impedance analysis in 158 breast cancer patients. Fatigue was assessed with the multidimensional 20‐item Fatigue Assessment Questionnaire. Multiple regression analyses considering changes in PhA and fatigue from baseline to post‐intervention and ANCOVA models investigating the strength training effect on PhA were conducted. Further, explorative mediation and moderation analyses were performed. Decrease (=worsening) in PhA was significantly associated with increase in physical (P = .010) and emotional (P = .019) fatigue. These associations were markedly stronger in patients with normal BMI (interaction P = .059 and .097) and with low pre‐diagnosis exercise level (interaction P = .058 and .19). Among patients with normal BMI strength training was associated with an increase in PhA (ANCOVA P = .059), but not among overweight/obese patients (interaction P = .035). Chemotherapy was a major determinant for low PhA, but PhA did not mediate the effect of chemotherapy on fatigue. In conclusion, PhA has a significant inverse association with physical and emotional fatigue. This association is moderated by BMI and previous exercise. Significant relationships of PhA were also observed with chemotherapy and strength training. Thus, PhA might be a marker that could help in the classification of subtypes of fatigue with different pathophysiology, which may require specifically tailored treatment. Further research on this is warranted.


What's new?
Cancer-related fatigue is commonly treated in an undifferentiated manner as its pathophysiology remains poorly understood. In this randomized controlled strength training intervention trial, the authors measured the bioelectrical phase angle (PhA), a non-invasive, inexpensive marker of cell integrity, in breast cancer patients. The results show a significant inverse association of PhA with physical and emotional cancer-related fatigue, depending on the patient's body mass index and pre-diagnosis exercise level, as well as significant relationships with chemotherapy and strength training. PhA could be a helpful marker in the classification of fatigue subtypes with different pathophysiologies that may require tailored treatments.

| INTRODUCTION
Cancer-related fatigue is commonly reported as the most distressing symptom by patients during the cancer trajectory, and is considered a dose-limiting toxicity for some cancer treatments. 1,2 It is estimated that the prevalence of fatigue among cancer patients can reach up to 80% to 90% over the course of the disease and treatment, with varying degrees of intensity. 3,4 Fatigue exerts in different facets such as physical, emotional or cognitive fatigue. There is wide agreement that it is likely multi-causal. However, up-to-date fatigue management is commonly provided in an undifferentiated manner without further evaluation of a patient's individual symptomatology. One reason may be that the pathophysiology is still not well understood and there are no established hallmarks for different fatigue types. Singling out specific subtypes of fatigue might enable a more adapted and individual patient treatment. 5 Bioelectrical impedance analysis (BIA) could possibly be useful in this context. BIA is a widely used non-invasive, portable, quick and inexpensive method to assess body composition. Electrodes transmit a low alternating current through the patient's body while simultaneously capturing the total physical resistance (= impedance) which occurs in the tissue by measuring the voltage. The impedance is dependent on the measured resistance from body water that contains electrolytes (= resistance, R), as well as on the capacitive resistance which is generated by cell membranes (=reactance, Xc). From these values parameters such as fat-mass, fat-free mass and total body water are estimated, which however highly depend on appropriate predictive age-, sex-and population-specific algorithms. 6 The phase angle (PhA) is a BIA-derived parameter that is less dependent on such algorithms. 7 The PhA is based on the sinusoidal waveform of the alternating current and alternating voltage. In body water both curves are in parallel. At the cell membranes, however, the voltage curve lags behind the current. This delay can be explained by the fact that the membranes of the body's cells charge and discharge under alternating current-like a capacitor. The time delay causes a phase shift between current and voltage. This shift can be measured in degrees and is then referred to as the "phase angle." Cell membranes in good condition, but also large amounts of muscular mass, lead to a larger shift, that is, larger PhA. A low PhA, on the other hand, can be a sign of cell death or decreased cell integrity, muscle wasting, sarcopenia, as well as malnutrition. [8][9][10][11] A recent study among patients with head and neck cancer, where malnutrition is common, found PhA as the most crucial predictor of survival, 12 and a systematic review considering advanced cancer patients also identified the PhA as a predictive factor of survival. 13 However, the meaning of the PhA, which is a biomarker rather than a biological pathway, is still not fully understood.
As PhA appears to be a biomarker of decreased cell integrity, poor muscle function or malnutrition, which are conditions that may contribute to fatigue, it is conceivable that PhA may act as a useful marker that could help to single out specific subtypes of fatigue. Sofar, only few studies examined the association between the PhA and fatigue in cancer patients. [14][15][16][17][18] Most but not all of these studies have suggested an inverse association. However, they mostly considered fatigue and PhA only cross-sectionally and did not distinguish the different fatigue dimensions.
Therefore, the aims of this explorative study were (1) to investigate longitudinal associations between PhA and fatigue in breast cancer patients, considering physical, affective and cognitive fatigue dimensions; (2) to explore if these associations might be moderated by other factors and (3) to investigate the effect of strength training on PhA. In addition, for a better understanding of PhA, we explored its determinants as well as its potential mediating role in previously shown effects of strength training and chemotherapy on physical fatigue.

| Study design and population
Here we present exploratory analyses of data from the BEST study, which is a prospective randomized controlled strength training intervention trial in 160 breast cancer patients under adjuvant radiotherapy at Heidelberg University Hospital. Inclusion criteria were histologically confirmed primary breast cancer; stage 0-III; scheduled for adjuvant radiotherapy; age ≥ 18 years; body mass index (BMI) ≥18 kg/m 2 ; ability to understand and follow the study protocol; and willingness to come to the exercise facilities. Exclusion criteria comprised contraindications for strength training (eg, acute infectious disease, severe cardiac disease, severe respiratory insufficiency), other concomitant malignant diseases (except carcinoma in situ of skin or cervix), and currently participating in regular exercise training (at least 1 h twice/week). Patients were 1:1 randomized to a progressive strength training or muscle relaxation (both for 12 weeks, 2 Â 1 h/ week). All patients had received breast surgery and some had received neo-adjuvant or adjuvant chemotherapy before recruitment. Patientreported outcome assessments and BIA measurements were conducted before randomization and start of radiotherapy (baseline, T0), after the end of radiotherapy (week 7, T1), and after the end of intervention (week 13, T2). Further details on intervention procedures and adherence, as well as primary and secondary results are published elsewhere. [19][20][21][22][23][24]

| Outcome measures
The body composition of participants was measured using AKERN BIA-101 (Akern Srl, Pontassieve, Italy). In healthy volunteers, BIA has proven to give reliable measurements of body composition, with only minimal intra-and interobserver variability. 25 For the measurements, patients were requested to lay down, with an angle of approximately 45 between both legs, and 30 between arms and legs. Afterwards, the electrodes were attached to one hand and foot on the same side of the body. Patients were asked to go to the toilet before the measurements, but had not been asked for dietary restrictions. For PhA, the following formula was used based on the variables obtained by BIA: Fatigue was assessed with the 20-item Fatigue Assessment Questionnaire (FAQ), which has been validated for a German-speaking population of cancer patients. 28 The questionnaire covers the physical, affective and cognitive fatigue dimensions. Scores were linearly transformed to a 0 to 100 scale, with higher scores indicating higher fatigue.
Socio-demographic factors were recorded by questionnaires. Exploration for other potential confounders yielded no substantial differences. Data was analyzed on an intention-to-treat basis. Effect modification was explored as described above.
Mediation of the intervention effect as well as the chemotherapy effect on fatigue by PhA were explored based on the approach of Baron and Kenny. 31,32 All models were adjusted for potential confounders, that is, age, BMI and depressive symptoms.
As all analyses were considered to be explorative, no power analysis was performed and no adjustments were made for multiple comparisons. All statistical tests were two-sided, and P < .05 was considered as statistically significant. SAS Version 9.3 was used for all analyses.

| Baseline characteristics of the study population
In the BEST study, 160 patients were randomized, 80 to the strength training intervention group and 80 to the muscle relaxation control group. PhA assessments at baseline were available for 158 participants. Baseline characteristics are presented in Table 1

| Determinants of phase angle
Baseline PhA values ranged from 2.9 to 7.1 with a median (Q1, Q3) No further baseline variables were significantly associated with PhA, nor did substantially change the estimates presented in Table 2.
T A B L E 1 Baseline characteristics of the study population.

| Association between phase angle and cancerrelated fatigue
Multiple regression analysis showed that change in PhA from T0 to T2 was significantly associated with change in physical fatigue (β = À10.7, P = .010) and affective fatigue (β = À9.1, P = .019) but not with cognitive fatigue (P = .39, Table 3). There were indications of moderations of these associations by baseline BMI (P inter = .059, .097 and .030 regarding physical, emotional and cognitive fatigue, respectively) and by pre-diagnostic exercise level (P inter = .058 regarding physical fatigue). Subgroup analysis by patients with normal BMI vs overweight/obese patients and by patients with no/little vs moderateto-high exercise level before diagnosis are also presented in Table 3.
Results showed significant associations between change in PhA and change in fatigue regarding all three fatigue dimensions among patients with normal BMI but not among overweight or obese patients. Regarding moderation by pre-diagnosis exercise level, the association between PhA and physical fatigue was significant among those who had been little active, but not apparent among the patients who had been physically more active. There was no significant moderation by previous chemotherapy (P inter = .64, .12 and .35 regarding physical, emotional and cognitive fatigue, respectively) nor by baseline depression (all P inter > .70).

| Effect of the strength training intervention on phase angle
Overall, ANCOVA revealed no significant intervention effect on PhA, that is, no significant differences in PhA change scores between the strength exercise training and the relaxation group (P = .79). However, when considering subgroup analyses, there was a borderline significant between-group difference (P = .059) for patients with normal BMI but not among overweight/obese patients (P = .29, P inter = .035) ( Table 4).
There were no significant effect modifications nor significant effects in subgroups based on pre-diagnosis exercise level (P inter = .14) nor other subgroups based on chemotherapy (P inter = .13) or baseline depression (P inter = . 33). The results of an explorative mediation analysis among patients with normal BMI are in accordance with a possible weak role for PhA as a partial mediator of a previously shown effect of the strength training intervention on fatigue 24 (Table S1).

| Investigation of phase angle as potential mediator of chemotherapy effects on fatigue
In our data (as known from many previous studies) chemotherapy was associated with higher physical fatigue levels. Thus, the significant association of chemotherapy with a decreased PhA (as shown in   Note: All models were adjusted for baseline fatigue, baseline phase angle, age, baseline BMI category, previous chemotherapy, depressive symptoms and intervention group.   Chong performed the statistical analysis and drafted the manuscript, all authors reviewed the manuscript and contributed important intellectual content. The work reported in the paper has been performed by the authors, unless clearly specified in the text.