Meta-Analysis of Pharmacological, Nutraceutical and Phytopharmaceutical Interventions for the Treatment of Cancer Related Fatigue

Simple Summary In our study we found that the overall meta-analysis of all cancer related fatigue (CRF) treatment studies showed significant reduction of CRF. The meta-analysis did not show significant reduction of CRF with treatment ginseng, guarana, megestrol, mistletoe, psychostimulants, selective serotonin reuptake inhibitors/antidepressants. Metanalysis of Corticosteroids studies showed significant reduction in CRF. Further studies are needed. Abstract Purpose: In this study we aimed to estimate the effectiveness of pharmacological, nutraceutical, and phytopharmaceutical treatments on CRF. Methods: Ovid MEDLINE, Ovid Embase, Ovid Psych info, CINHAHL and Cochrane Library databases were searched up to 30 September 2021. Randomized controlled trials of pharmacological, nutraceutical and phytopharmaceutical interventions for treatment of CRF for at least one week duration and have used valid tool to assess severity of CRF as a primary or secondary outcome were considered. Results: 32 eligible studies (4896 patients) were reviewed. For the overall meta-analysis, the random effect models yielded the treatment effect (95% CI) of −0.29 (−0.48,−0.09), p < 0.001. The meta-analysis did not show significant reduction of CRF with treatment with ginseng (n = 6), guarana (n = 3), megestrol (n = 2), mistletoe (n = 3), psychostimulants (n = 14), SSRI/antidepressants (n = 2). Corticosteroids (n = 2) showed significant reduction in CRF with treatment effects of 0.94 (−1.21, −0.67), p <0.0001, respectively. Conclusions: In this study, overall meta-analysis of all studies demonstrates significant reduction of CRF using Pharmacological, Nutraceutical and Phytopharmaceutical interventions with a pooled standardized treatment effect of −0.29. Metanalysis of Corticosteroids studies showed significant reduction in CRF. Further studies are needed.


Introduction
Cancer-related fatigue (CRF) is one of the most common and distressing symptoms associated with cancer and its treatment [1][2][3][4][5]. The frequency of CRF in cancer patients varies from 60% to 90% [1 -6]. CRF negatively affects quality of life (QOL), interferes with daily activity, has potentially devastating social and economic consequences, affects the patient's ability to receive cancer therapy, and may potentially negatively impact public health [2,[6][7][8][9]. The National Comprehensive Cancer Network defines CRF as a "distressing, persistent, subjective sense of physical, emotional, and/or cognitive tiredness or exhaustion related to cancer or cancer treatment that is not proportional to activity and that interferes with usual functioning" [2].
proposed by which ginseng may improves CRF is by its action on (a) CNS, including cognition/memory, sleep disturbance, anxiety/depression, (b) by Neuroprotection via potentiation of nerve growth factor activity [45,46], (c) inhibition of the activity of N-methyld-aspartate(NMDA) receptor activity [47], (d) inhibition of excitatotoxicity and calcium overflux into neurons, increase dopamine and norepinephrine in the cerebral cortex [48] and modulate the activity of presynaptic and postsynaptic receptors, and (e) modulation of inflammatory cytokines [49,50].
Mistletoe (Viscum album L.) total plant extracts have been used for the treatment of CRF [56]. Possible role in improvement in CRF may be related to its immunomodulatory effect [57].
However, there are very limited recent meta-analysis conducted using rigorous methods including all pharmacological and nutraceuticals medications used for the treatment of CRF despite several full systematic reviews [2,33,[58][59][60][61][62][63]. Due to which the current guidelines from various professional organizations such as American Society of Clinical Oncology (ASCO), NCCN, and European Society of Medical Oncology (ESMO) provide limited indications for the use of pharmacological or nutraceuticals [64,65]. Hence, to date there is no FDA approved pharmaceutical or nutraceutical treatment for cancer related fatigue. Therefore, in this study we aimed to conduct a metanalysis of pharmacological, nutraceutical, and phytopharmaceutical treatments on CRF to estimate their effectiveness to treat CRF.

Methods
This study was approved by the institutional review board (IRB) of the University of Texas, MD Anderson Cancer Center, Houston Texas. Our meta-analysis has been registered with PRISMA Transparent reporting of systematic reviews and meta-analyses (http://www.prismastatement.org/Protocols/Registration accessed on 1 November 2022). The project ID in the PRISMA/PROSPERO website is CRD42021203093.
We reviewed pharmacological, nutraceutical, and phytopharmaceutical Intervention studies in adult cancer patients on CRF receiving treatment or during post-treatment. We had five reviewers (SY, ARM, ZL, NN, and MK), to assess the inclusion and exclusion of the studies via reviewing the titles, abstracts, and full texts. The reviewers were randomly assigned to two groups and each group received half of the papers identified by the librarian (GY). Individually, each member assessed the inclusion and exclusion criteria and then cross checked the results. In case of conflicts, the study principal investigator (SY) made the final decision whether include or not a given study.

Study Selection
To be eligible to be included in the study, the articles had to meet the following eligibility criteria: (a) English language publications, (b) available in Ovid MEDLINE, Ovid Embase, Ovid Psych info, CINHAHL and Cochrane Library databases, (c) pharmacological, nutraceutical, or phytopharmaceutical interventions in human subjects with cancer, (d) Use randomized clinical trial phase II or III design, (e) Timeline-from 1 January 1967 to 30 September 2021, (f) Use a valid and acceptable fatigue outcome measures [66] such as Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F), Brief Fatigue Inventory (BFI), Piper Fatigue Scale (PFS), Fatigue Symptom Inventory (FSI), Multidimensional Fatigue Symptom Inventory (MFSI), Multidimensional Assessment of Fatigue (MAF), European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire (EORTC-QLQ-30) fatigue, The MD Anderson Symptom Inventory(MDASI), that assess both presence of fatigue and severity of fatigue, (g) cancer related fatigue should be a primary or secondary outcome, (h) The duration of the selected intervention should be at least 1 week, (i) The selected intervention should be compared to a placebo or another intervention or standard care. The articles were excluded if there was: (a) assessment of CRF using treatment toxicity only using the Common Terminology Criteria for Adverse Events (CTCAE) or equivalent, (b) combinations treatments, (e.g., combination treatment of Coenzyme-Q, N acetyl Carnitine), (c) hemopoietin growth factors, (d) less than 30 patients enrolled in the study.
Based on the above strategies, we have included 32 eligible peer-reviewed manuscripts (4896 total patients) that assessed the effects of pharmacological, nutraceutical, and phytopharmaceutical interventions on cancer induced fatigue. These selected peer-reviewed manuscripts involved the interventions of psychostimulants (Methylphenidate, Dexamphetamine, Modafinil and armodafinil); corticosteroids (Dexamethasone and Methylprednisolone); selective serotonin reuptake inhibitors (SSRI's); Ginseng; Guarana; Melatonin; Minocycline; Mistletoe; and Megestrol, and they were assessed using the meta-analysis approach.
Subject heading searches were explored to include narrower terms in the Medical Subject Headings (MeSH) or EMTREE (subject headings unique to Embase) hierarchy as needed. The search terms were combined by "or" if they represented the similar concept, and by "and" if they represented different concepts. Searches were restricted to randomized clinical trials of phase II or Phase III design in adult cancer patients. Conference abstracts, systematic reviews and case reports were excluded.
We searched the abstracts of the identified articles from the databases for inclusion in the review. The primary author and the medical librarian independently carried out a study selection to determine that the articles meet the inclusion criteria. Any disagreement about a particular study was resolved by discussion.

Data Extraction
We extracted data from each included study into the following areas of focus: (I) General information: article identification (author, year), full citation, geographic location, setting (e.g., hospital-based, clinic-based, community-based, referral criteria/process, other), declared conflict of interest, and source(s) of funding. (II) Study characteristics including design (Randomized controlled trials (RCT's), clinical trial, survey, other), duration of study, number of centers, sample size, follow-up assessments, primary outcome (definition, instrument used, scoring), and secondary outcomes (definition, instrument used, scoring) (III) Patient Population details included in the study such as inclusion criteria, exclusion criteria, baseline characteristics (targeted to the specific topic-e.g., mean age, gender (male/female %), description of the therapy (time, intensity, frequency, etc.), and descrip-tion of the comparator (intensity, time, frequency). (IV) Quality appraisal: Physiotherapy Evidence Database (PEDro) scoring systems was used to evaluate the quality of the articles [67]. PEDro scoring is a checklist on 10 scored yes or no questions. It evaluates internal validity and statistical information. At least two blinded evaluators checked independently each include paper to score it and an average result was calculated. Any major discrepancies were evaluated and discussed with principal investigator. The Cochrane risk assessment tool [68] was used to assess the methodological quality of each study including the risk of selection, performance, detection, attrition, and reporting biases, Each reviewer scored the Cochrane assessment tool individually analyzing the study method of randomization, allocation concealment, blinding of patients, investigator and assessor, incomplete outcome data (participation rate, sampling procedure/sample size calculation, analysis, completion per study design, handling of missing outcome data), selective outcome reporting, and the possibility of other biases. (V) Outcomes: Continuous (fatigue scales such as FACIT-F, BFI, etc.), Group mean (final value or mean change from baseline), Group standard deviation (can be calculated from SE, T-test, p values), and total number of patients per group were assessed.

Statistical Analysis
Results were reviewed from 43 studies which assessed the effect of drug treatment on fatigue using a variety of scales. Drug treatments included ginseng, guarana, megestrol, mistletoe, psychostimulants, SSRI/antidepressants, and steroids. For studies which reported results on multiple scales, single scale results were selected, prioritized in order of FACIT-F, then BFI, followed by other scales. To allow meta-analysis of the pooled results from these different scales, reported score differences were converted to standardized mean difference (Z-scores), such that an increase in score corresponds to increase in fatigue, with 95% confidence intervals. Nine studies were excluded because they did not have sufficient information to estimate a standardized difference with confidence intervals, which left 32 studies for the meta-analysis ( Figure 1). We excluded the metanalysis of two pharmacological or nutraceutical agents (melatonin and minocycline) as there was only one study for each of these agents which were eligible as per the eligibility criteria of the study [69,70]. These studies were grouped by treatment sub-group analyses. Meta-analyses, both fixedeffect and mixed effect, were performed for each of the treatments as well as overall, using the inverse variance method, with the DerSimonian-Laird estimator for tauˆ2 and the Jackson method for confidence intervals of tauˆ2. Cochrane's Q and Iˆ2 were used to assess heterogeneity among studies. Given the heterogeneous nature of drug treatments and the design and conduct of these studies, the random effects model estimates of effect size, confidence interval, and p-values were prioritized over those of the fixed effects models. Therefore, in the results we reported only the random effects model estimates of treatment effects. Corresponding forest and funnel plots were produced. All statistical analyses were performed using R Core Team (2020) [71], with meta-analyses performed using the "meta" package [72].

Results
Among the 32 studies (4896 patients) reviewed ( Figure 1, Table 1), 10 studies reported significantly lower fatigue due to treatment. Results of the meta-analyses on the effect of treatments on fatigue are illustrated in the forest plot ( Figure 2). Fatigue scales were transformed to standardized mean difference (Z-scores), such that an increase in score corresponds to increase in fatigue. The funnel plot ( Figure 3) shows that most treatment effects are clustered near zero and with some evidence of bias towards smaller effect sizes and some heterogeneity. Table 2 shows the Risk Bias for studies included in the Meta-Analysis. Figures 4 and 5 show the Cochrane risk of bias assessment using traffic-light plot, and Cochrane risk of bias assessment summary plot, respectively.

Results
Among the 32 studies (4896 patients) reviewed ( Figure 1, Table 1), 10 studies reported significantly lower fatigue due to treatment. Results of the meta-analyses on the effect of treatments on fatigue are illustrated in the forest plot ( Figure 2). Fatigue scales were transformed to standardized mean difference (Z-scores), such that an increase in score corresponds to increase in fatigue. The funnel plot ( Figure 3) shows that most treatment effects are clustered near zero and with some evidence of bias towards smaller effect sizes and some heterogeneity. Table 2 shows the Risk Bias for studies included in the Meta-Analysis. Figures 4 and 5 show the Cochrane risk of bias assessment using traffic-light plot, and Cochrane risk of bias assessment summary plot, respectively.        For the overall meta-analysis, the random effect models yielded the estimates of treatment effect and 95% confidence interval of −0.29 (−0.48, −0.09), p < 0.001. Cochrane's Q supports heterogeneity among studies (Q = 305.8 with 33 degrees of freedom, p < 0.0001), as did Iˆ2, estimated as 89% (86%, 92%). Tauˆ2 was estimated as 0.27 (0.16, 0.50), suggesting a low variance of the estimated treatment effect size.
The SSRI/antidepressant meta-analysis of 2 studies [101,102], one of which had a significant negative-trending effect size, while the other was slightly positive-trending, yielded random effects model treatment effects of −0.25 (−0.88, 0.38), p = 0.44. Cochrane's Q supports heterogeneity among studies (Q = 6.84 with 1 degrees of freedom, p = 0.009), as did Iˆ2, estimated as 85% (41%, 96%). Tauˆ2 was estimated as 0.18, suggesting a low variance of the estimated treatment effect size.
The steroid meta-analysis of 2 studies [103,104], both of which had significant negativetrending effect sizes, yielded the random effects model treatment effects of −0.94 (−1.21, −0.67), p <0.0001. Cochrane's Q does not support heterogeneity among studies (Q = 0.06 with 1 degrees of freedom, p = 0.80), nor did Iˆ2, estimated as 0%. Tauˆ2 was estimated as 0.0, suggesting a low variance of the estimated treatment effect size.

Discussion
In our study, we found that the overall meta-analysis of all CRF treatment studies showed significant reduction of fatigue with treatment effect of −0.29. Our meta-analysis suggests significant reduction of CRF with Corticosteroids. Metanalysis of psychostimulants (Methylphenidate, Modafinil, Armodafinil), Ginseng, Guarana, Megestrol, Mistletoe, and antidepressants did not show significant reduction in CRF. Further studies are needed.
As compared to recent meta-analysis of pharmacological treatment for CRF studies (last 6 years) [62,63,[105][106][107][108][109][110], the results of our study suggests significant reduction of CRF with Corticosteroids, whereas no significant reduction in CRF with psychostimulants such as Methylphenidate, Modafinil, Armodafinil, Ginseng, Guarana, Megestrol, Mistletoe, and SSRI/antidepressants. Our study was unique as it included only RCT's with a control or placebo, publications in English, and included both nutraceuticals and pharmaceuticals used to treat CRF as most often patients prefer to use both the nutraceuticals and pharmaceuticals if they decide to address their CRF with medications [32]. In addition, in this study we employed more rigorous methods including strict eligibility criteria, use of Physiotherapy Evidence Database (PEDro) scale to assess the quality of the included RCT studies, and Cochrane risk assessment tool to assess risk bias of the RCT's. Some of the published metanalysis articles focused on specific stage of disease or type of medication. Roji et al. [106], and Junior et al. [107], focused on placebo. Qu et al. [108], Minton et al. [109], and Gong et al. [110], focused on psychostimulants drugs. Mucke et al. [62] focused mainly on patients receiving palliative care. Our study is unique in contrast to other recent published metanalysis using pharmaceuticals and nutraceuticals in that we included all stages and types of cancer patient populations (early, advanced, cancer survivors as well as cancer types) as well as pharmaceuticals and nutraceuticals. Additionally, we excluded patients using erythropoietic agents as they no longer used to treat CRF due to concerns of increased risk for cardiovascular events and tumor growth [111,112]. Pharmacological or nutraceutical agents (Melatonin, Minocycline) which had only one eligible study were excluded from metanalysis of the agent as the results were more a representative of the single randomized control study rather than a metanalysis of various studies using the agent.
Despite the mixed findings of benefit of various interventions for CRF in our study, it is very early to state any of these interventions are not effective in treatment of CRF due to the limited clinical trials conducted using validated outcomes and in well-defined homogenous cancer patients. Further research is necessary to evaluate which subgroup of cancer patients these interventions will be most likely benefit.
The result from this study regards to the beneficial effects of corticosteroids on CRF is consistent with the results of prior studies using steroids [113][114][115][116] which were excluded as they did not meet all the eligibility criteria of our study but support the use of this agent. However, recent advent of immunotherapy as one of the important cancer treatment agents limits its use of corticosteroids, therefore other agents should be considered among patients on immunotherapy [117].
In contrast to other pharmaceutical and nutraceuticals, psychostimulants specifically Methylphenidate is the most investigated medication for the treatment of CRF. The results of the metanalysis showed non-significant trend towards benefit. These results are interesting as patients in clinical practice often report interest and benefit from the medication. Perhaps a possible reason for mixed results is that we have not targeted the intervention is a specific subgroup of patients (e.g., fatigued cancer patients with anxiety or depression or drowsiness) [90] or we have not used Methylphenidate in combination with other CRF treatments (e.g., exercise) as it may not target all the pathophysiologic mechanisms causing fatigue (especially physical fatigue) [118,119]. In clinical practice in view of the mixed results from metanalysis, as well as side effect profile which includes risk of addiction one should only consider using Methylphenidate in a short term and on a trial basis. In a study by our team [120], we found that patients who showed improvement in CRF (response) after 1 day of treatment of Methylphenidate will most likely have a benefit from methylphenidate treatment for CRF. Therefore, in clinical practice for patients appropriate to use Methylphenidate for CRF, one may consider a trial of Methylphenidate for 1 or 2 days and then extend treatment further on an empirical basis if beneficial with close monitoring.
Various methodological factors can make the interpretation of results of our study challenging: (a) CRF is a subjective symptom and various tools were used in the published clinical trials. (b) Most of the studies used statistical significance rather than clinically relevant improvement as a measure to conclude benefit [121]. Therefore, there were limited details in the published studies which show that improvement in subjective fatigue would result in actual improvement in physical function, activity or interference in daily activity. (c) Finally, during the analysis phase (see Figure 1) we had to exclude several studies due to availability of limited data to conduct the meta-analysis.
Finally, CRF is a complex multidimensional syndrome due to various physical, cognitive, psychosocial factors involving brain, muscle, cognition and effecting various pathophysiological changes including inflammation, neuro-immuno-pituitary adrenal axis, mitochondrial pathways [36]. Using a single agent may not target all the causes of this complex multifactorial syndrome. Hence, future studies targeting various predominant causative mechanisms in specific patient, i.e., multimodal personalized therapy similar to the current management of cancer specific therapy should be considered [122].

Conclusions
In this study the results of metanalysis of published studies for the treatment of CRF showed significant reduction of cancer related fatigue after treatment with Pharmacological, Nutraceutical and Phytopharmaceutical interventions with a pooled standardized treat-ment effect of −0.29. Metanalysis of Corticosteroids studies showed significant reduction