Association of plasma leptin, pro‐inflammatory adipokines and cancer‐related fatigue in early‐stage breast cancer patients: A prospective cohort study

Abstract Cancer‐related fatigue (CRF) is subjective and has wide inter‐individual variability. Given that leptin is commonly associated with fatigue syndrome, its use as a potential biomarker for CRF is being investigated. The primary objective of this study was to evaluate the association between leptin and CRF in early‐stage breast cancer patients receiving chemotherapy. In a prospective cohort study, patients completed assessments at baseline (T1), during chemotherapy (T2) and after chemotherapy (T3). Levels of plasma leptin and adipokines were measured using a Luminex bead‐immunoassay and CRF was measured using the Multi‐Dimensional Fatigue Symptom Inventory‐Short Form (MFSI‐SF). Data were analysed longitudinally using a generalised estimating equation incorporating clinically relevant parameters and pro‐inflammatory adipokines. The analysis included 136 patients (mean age ± SD = 51.5 ± 8.8 years; 69.1% receiving anthracycline‐based chemotherapy). More patients experienced CRF at T3 (23.8%) than at T2 (13.8%) compared to baseline. An increase was observed in the median plasma leptin level at T2, followed by a decrease at T3 (T1: 4.07 ng/mL, T2: 4.95 ng/mL and T3: 3.96 ng/mL). In the multivariate model, the change in leptin levels over time was significantly associated with the total MFSI‐SF score (β = −0.15, P = 0.003) after adjusting for the tumour necrosis factor‐α (TNF‐α) level, anxiety, depression, insomnia, age, menopausal status and type of chemotherapy. This is the first study to report leptin as a biomarker that predicts the onset of CRF over time. Future studies are required to validate the findings.


| INTRODUC TI ON
The estimated incidence of cancer-related fatigue (CRF) has been reported to range from 28% to 91% among breast cancer patients, 1 depending on the type of cancer, treatment modality and method of assessment. CRF is characterised by a persistent sense of tiredness that may be related to cancer and/or cancer treatment, is disproportionate to recent activity and is not alleviated by rest. 2 Left unresolved, CRF adds on to patients' distress and has a negative impact on their quality of life. 3 Given the extent of the impact of CRF on daily functioning, 4 addressing CRF should be an integral part of cancer-supportive care. Nevertheless, this has proved challenging due to the lack of an objective biomarker to monitor these symptoms. The experience of fatigue is subjective in nature and has wide inter-individual variability because some patients may have poor physical functioning and impaired performance status yet will not complain of fatigue, and vice versa. One study has shown that pro-inflammatory cytokines could be potentially employed as biomarkers of CRF in patients with advanced cancer, 5 and that changes from baseline of selected biomarkers were associated with changes in patient-reported outcome measures of appetite and fatigue.
However, the circadian variability and fluctuation of cytokine levels could influence the outcomes associated with the biomarkers used and hinder their utility. 6 Thus, further research is warranted to determine a suitable biomarker for monitoring CRF and related symptoms in cancer patients.
Leptin is an endocrine hormone that is more commonly known for its metabolic effects, which range from appetite suppression to regulating body weight. 7 Leptin has shown moderate correlations to varying extents with fatigue in cohorts of patients with cardiovascular risk factors (r = 0.22), 8 chronic fatigue syndrome (r = −0.51 to 0.73), 6 chronic hepatitis C infection (r = 0.25; r = 0.30), 9,10 and irritable bowel syndrome (r = 0.60). 11 In another study, researchers found that plasma leptin levels could be induced by hydrocortisone, and were markedly increased in patients with chronic fatigue syndrome. 12 Hence, we hypothesized that leptin levels may be positively correlated with increases in fatigue in cancer patients, as their pro-inflammatory markers are typically elevated.
This study was designed to investigate the association between plasma leptin levels and CRF in a cohort of early-stage breast cancer patients, in relation with other relevant clinical factors and adipokines. Our primary objective was to assess leptin's potential to function as a biomarker that could predict the onset of CRF.

| Study design
This prospective cohort study was conducted between 2014 and 2017 in Singapore. The study was approved by the Singhealth Institutional Review Board (CIRB 2014/754/B), and written informed consent was obtained from all study participants.

| Inclusion and exclusion criteria
The inclusion criteria were: (a) diagnosis with early stage (I-III) breast cancer; (b) no prior history of chemotherapy or radiotherapy; (c) scheduled to receive standard adjuvant chemotherapy; (d) ambulatory status with an Eastern Cooperative Oncology Group score of 0 or 1; and (e) the ability to understand either English or Chinese. Patients were excluded if they were diagnosed with metastatic cancer, had another medical condition that precipitates fatigue (such as severe anemia or thyroid dysfunction), or were on medications such as beta-blockers that might precipitate fatigue as a side effect.

| Study procedures
Patients were assessed at three time points: baseline before treatment initiation (T1), at least 6 weeks after baseline during chemotherapy (T2), and at least 12 weeks after baseline after the completion of chemotherapy (T3). Upon recruitment, patients' demographic information and medication information were collected through patient interviews and electronic databases.

The Beck Depression Inventory (BDI) is a validated questionnaire
that assesses the severity of depression 17 and had been previously used by our team on breast cancer patients. 18 There are 21 items that patients rate on a scale of 0-3. The summation of the scores gives a total score ranging from 0 to 63, with a higher score indicating greater severity of depression.

| Statistical analysis
Patients' demographic information, clinical characteristics at baseline and the proportions of patients with CRF were summarised using descriptive statistics. The Friedman test was used to analyse the changes in plasma adipokines and leptin over the various time points, and post hoc Wilcoxon signed-rank test was used to identify the time points at which a change was observed. Spearman's rank correlation coefficient was used to examine the relationship between each variable and leptin at each time point. Spearman's rank correlation was also used to determine the association between continuous variables of interest and total MFSI-SF score. The Mann-Whitney U test was used to compare fatigue levels between groups for categorical variables.
The longitudinal association of CRF with plasma leptin and adipokine levels was assessed using the generalised estimating equation model. The correlation structure with the smallest criterion was chosen using the quasi-likelihood under the independence model criterion. Leptin was first analysed as a single variable. Next, the following known confounders that affect CRF were analysed individually against total MFSI-score across the time points: anxiety, 25 depression, 25 insomnia, 26 menopausal status, 27 anaemia status, 28 body mass index (BMI), 29 age, 30 type of chemotherapy, 31 and pro-inflammatory adipokines such as IL-6, IL-8 and TNF-α. [21][22][23] Confounders with P-values <0.05 were considered to be statistically significant and included in the final model as covariates to examine plasma leptin levels with respect to fatigue levels. All statistical analyses were conducted with STATA version 15 (Statacorp, TX, 2017), and two-sided P-values <0.05 were considered statistically significant.

| Patient demographics
A total of 136 early-stage breast cancer patients were included in this analysis, and their demographics are summarised in

| Prevalence of cancer-related fatigue
Using the MCID as a cut-off, patients who experienced a deterioration of ≥11 points from baseline were considered as fatigue cases. A total of 23.8% of patients experienced CRF at T3 compared to 13.8% of patients at T2, using the baseline as a reference. The overall incidence of CRF from T1 to T3 was 24.6%.

| Plasma levels of adipokines and leptin across time points
The plasma levels of the individual adipokines are summarised as the medians with the interquartile ranges ( Abbreviation: TNF-α, tumour necrosis factor-α. Bolded are P-values < 0.05 and P-values for post-hoc analysis cut-off = 0.0167. In cases for patients whose plasma levels of adipokine were below the detection limit, the laboratory values are treated as missing values in the statistical analysis.

| Correlation of plasma adipokines levels and confounders against plasma leptin levels at individual time points
Considering relationships between leptin and each individual adipokine, only IL-8 had a statistically significant correlation at T1 (r = 0.20, P = 0.02). At T1, there were no significant correlations between leptin and IL-6 or TNF-α. At T2 and T3, none of the adipokines showed statistically significant correlations with leptin (Table 3).

| Correlation of plasma adipokines levels, known confounders and leptin against fatigue levels across time points (T1-T3)
There were consistent strong correlations with fatigue levels across all time points for anxiety, depression and insomnia. Age was negatively correlated with CRF across time points and CRF level also differed for the type of chemotherapy. There was only a significant negative correlation at T1 for leptin (r = −0.25, P = 0.005) ( Table 4).
Across T1 to T3, leptin was negatively correlated with the total MFSI-SF score (β = −0.56, P < 0.001), with every 1 unit decrease of leptin being associated with a 0.56 increase in total MFSI-SF score ( Table 2). TNF-α was also inversely correlated with the total

| Adjusted model showing an association between leptin and CRF
Using P-values <0.05 as selection criteria, the variables with statistically significant associations with CRF were TNF-α, anxiety,  unit decrease of leptin being associated with a 0.14 increase in total MFSI-SF score across time points (Table 5).

| D ISCUSS I ON
In our cohort of early-stage breast cancer patients, plasma leptin levels were found to be significantly and negatively correlated with Based on our findings, it would be interesting to explore the relationships between leptin and other pro-inflammatory biomarkers.
Various cytokines such as TNF-α, IL-6 and IL-1 can induce metabolic changes by mimicking the action of neuropeptides such as leptin because of their structural similarity. 35
Bolded are P-values < 0.05. metabolism, which regulated leptin levels. 29 In another cohort of lung cancer patients, 37 the research team found significant positive associations among plasma levels of TNF-α, IL-1 and CRF. Together, these studies and our results suggest that leptin should be included in studies in which relationships between fatigue and cytokines are explored.
One strength of this study was the use of a longitudinal study design with repeated measurements of outcomes at various time points, which allowed us to determine whether a decrease in leptin levels signalled a rise in the fatigue experienced by breast cancer patients at a separate time point. Most of the current fatigue studies have been cross-sectional so far, 8,9 and are not able to track the trajectories between fatigue and other variables, given how some of the other symptoms may be transient. A limitation of this study lies in that there were no cancer control patients who did not undergo chemotherapy to provide a comparison of CRF experience.
Understanding the role of leptin in relation to CRF may help us to devise more tailored interventions such as exercise to mitigate CRF in the future. In a randomised controlled trial examining the effects of exercise in overweight or obese breast cancer survivors, 38 circulating biomarkers (insulin, IGF-1, adiponectin and leptin) were significantly improved post-intervention, compared to usual care. In another trial, a decrease in leptin concentrations in breast cancer survivors who exercised compared to controls was reported. 39 Several concerns need to be addressed before any potential fatigue biomarker can have clinical utility. The observed difference in direction for effect of leptin on fatigue may be partly attributed to the non-specific nature of fatigue, which can be challenging to quantify despite use of validated questionnaires and is dependent on the study tools used. There is no consistent definition of CRF and a general lack of agreement about which biomarkers to study or how to collect and test them. Research aimed at resolving these issues would help to advance the understanding of the mechanisms that underpin CRF and could improve the quality of life of individuals who experience it.
In conclusion, our data show that there is an inverse correlation between plasma leptin levels and fatigue levels over time in early-stage breast cancer patients undergoing chemotherapy. The association with total score remained statistically significant after adjusting for known confounders of CRF, enabling leptin to function as a biomarker that could predict onset of CRF.

ACK N OWLED G EM ENTS
We would also like to acknowledge the following doctors from the

CO N FLI C T S O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

AUTH O R S CO NTR I B UTI O N S
HHK, KO and AC: Conceptualisation, funding acquisition, supervision and writing-review and editing. FKM and PC: investigation, project administration and resources. TCJ, AHLY, MS and GYX: data curation, investigation and project administration. TYL: Formal analysis, investigation, writing-original draft, writing-review and editing.
All authors read and approved the final manuscript.