Psychological and behavioral symptoms in patients with melanoma: A systematic review and meta‐analysis

Improved survival rates have made it increasingly important for clinicians to focus on cancer survivorship issues affecting the quality of life of melanoma patients. To provide a comprehensive overview of the disease and treatment‐related issues affecting such patients, we conducted a systematic review and meta‐analysis of the literature to estimate the prevalence of symptoms of depression, anxiety, fatigue, sleep disturbance, and cognitive problems among melanoma patients, both uveal and cutaneous, before, during and after treatment.


| INTRODUCTION
Melanoma is an aggressive form of cancer, especially in its later stages, that originates from malignant transformed melanocytes. 1 In 2020, the incidence of melanoma was approximately 325,000 cases worldwide. [2][3][4] This number is estimated to increase to 510,000 newly diagnosed cases by 2040. 4 Melanoma affects people of all ages, but occurs most often during middle adulthood between the ages of 40 and 60 years. 5 The estimated lifetime prevalence rate is one in 34 males and one in 53 women, 6 with melanoma accounting for six out of seven skin cancer deaths. 2 Historically melanoma patients have been treated with chemotherapy, radiotherapy, and cytokine-based immune therapies (e.g., interferon-α, interleukin-2). 2,3,7 The development of modern immunotherapy with checkpoint inhibitors together with targeted therapies with BRAF and MEK inhibitors in patients with BRAF mutated melanoma has led to new treatments with substantially improved survival rates over the past 10 years. 3,7-10 The treatment of metastatic melanoma has particularly been improved by the introduction of CTLA4 and PD1 inhibitors that have beneficially impacted the prognosis for many patients. [11][12][13][14] CTLA4 and PD1 antibodies are immune checkpoint inhibitors (ICIs) that enhance activation of the immune system and tumor cell targeting. 15,16 ICIs differ from classical cytokine-based immunotherapy (e.g., interferon-α, interleukin-2) and are associated with autoimmune like side effects. However, surgery is still the primary treatment for most melanoma patients (stage I-III), and currently adjuvant treatment is indicated for radically resected stage III-IV disease. ICIs and targeted therapy are also used in nonoperable stage III-IV disease. 2,3 As a consequence of improved survival, there has been an increased focus on short-and longer-term side and late effects experienced by cancer patients and survivors in general. 17 Across different cancers and treatments, a significant proportion of patients experience a wide range of co-occurring symptoms, including depression, anxiety, fatigue, sleep disturbances, and cognitive problems. [18][19][20] While the etiology of these symptoms, that is, whether they are due to the cancer or its treatment, is not always clear, these symptoms are associated with lower quality of life, poor daily functioning, slower recovery, and increased morbidity. [21][22][23][24][25][26] A systematic review of 44 studies published in 2009 indicated that approximately one third of patients with melanoma report high levels of psychological distress requiring clinical intervention. 27 A more recent systematic review of 18 studies showed that patients with uveal melanoma experienced impaired physical functioning and well-being after treatment. 28 Several studies included in these reviews have focused on psychological distress and quality of life in melanoma patients, 27,29-31 but have typically not examined other symptoms that often co-occur in cancer patients. Indeed, in cancer patients, such symptoms may contribute to or even exacerbate each other over time 19,32 with important treatment implications. For example, successful treatment of one symptom may have a beneficial impact on other symptoms in the cancer symptom cluster. 33 There is thus a need for a comprehensive systematic review with metaanalysis of important symptoms in uveal and cutaneous melanoma patients beyond psychological symptoms, and which includes studies of other psychological and behavioral symptoms typically present in the cancer symptom cluster.
In view of the recent evolution of treatments for melanoma patients, we aimed to conduct a comprehensive systematic review to investigate the nature and prevalence of psychological and behavioral symptoms in the cancer-related symptom cluster, that is, depression, anxiety, fatigue, sleep disturbances, cognitive problems, before, during, and after treatment in patients and survivors with melanoma. Such information could have important implications for the development of targeted interventions to ameliorate such symptoms in this population.

| METHODS
The present study was conducted in accordance with the guidelines for Meta-Analysis Of Observational Studies in Epidemiology (MOOSE). 34 The protocol was preregistered in The International Prospective Register of Systematic Reviews (PROSPERO) 35 (#CRD42020189847), and the review is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. 36

| Search strategy and selection criteria
A comprehensive literature search with no publication date restrictions was undertaken on September 9, 2021 with an updated search conducted on June 8, 2022 in the PubMed, PsycInfo, Cochrane Library, and CINAHL electronic databases. When possible, relevant MeSH terms or MeSH term equivalents were included in each database. Key words related to the population (melanoma* OR "malignant melanoma*" OR "metastatic melanoma*") were combined with key words related to outcomes (depressi* OR anxiety OR fatigue* OR "cancer-related fatigue*" OR cognition* OR "cognitive function*" OR sleep* OR "sleep disturbances*" OR "cognitive impairment*" OR "cognitive dysfunction*" OR "cognitive deficit*" OR chemobrain OR "chemo-brain*" OR neuropsycholog* OR "cogniti*" OR "sleep disturb*" OR "chemo fog*" OR insomnia*). The first and second author (JTD and LSL) conducted the search independently.
Study eligibility was determined using the Population, Intervention/Exposure, Comparator, Outcome approach 37 referring to population, intervention/exposure, comparator, and outcome. 37 Population: studies of adults in treatment or previously treated for melanoma; Intervention/exposure: any melanoma treatment; Comparator: N/A; Outcome: one or more of the following symptoms: depression, anxiety, fatigue, sleep disturbances, and cognitive problems prior to, during, and/or after treatment assessed independently with a standardized validated measurement/symptom scale.
Only English-language reports published in peer-reviewed journals were considered eligible. We excluded non-human research, reports that were not primary research papers, for example, review DANIELSEN ET AL.
articles, single case studies, and "grey literature," for example, conference abstracts and dissertations. Intervention studies were also excluded.
The literature search and data extraction were conducted using the Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia, www.covidence.org). In the first round of screening, JTD and LSL independently screened the title and abstract of all identified references and excluded ineligible reports. In the second round of screening, the full-text of the remaining reports were evaluated independently by JTD and LSL and reasons for exclusion were registered. Uncertainties and disagreements were discussed with the last author (RZ). After each screening, the two authors discussed discrepancies, and reached a negotiated decision.

| Quality assessment
Methodological quality and risk of bias assessment was undertaken independently by two authors (JTD and SMK) for all included studies, using the National Institutes of Health Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies. 39

| Data extraction and synthesis
Data extraction was performed independently by two authors (JTD and SMK) and included authors, publication year, aim of study, study design, sampling strategy, setting, follow-up (number and assessment time points), number of participants, mean sample age, gender of participants (percent women), mean education level in years, participation rate of eligible persons, loss to follow up (percentage), melanoma diagnoses of participants (cutaneous and/or uveal melanoma), cancer stage (I-IV), treatment type (surgery, radiotherapy, chemotherapy, targeted therapy, ICI and/or cytokine-based immunotherapies), outcomes (cancer symptoms represented; depression, anxiety, fatigue, sleep and cognition), method used for assessing cancer symptoms and results (measurement, mean, standard deviation and prevalence rates of symptoms). A meta-analysis was conducted when a minimum of three studies assessing the same symptom were available. Analyses were conducted separately for each outcome, assessment time point (prior to treatment, during treatment and after treatment) and melanoma subtype (cutaneous or uveal melanoma).

| Meta-analytic strategy
A pooled inverse-variance weighted event rate (ER) was calculated for each outcome as the summary statistic of prevalence. ER is defined as the proportion of patients reporting high/clinical levels of the symptom in question while taking the individual study sample size into consideration, yielding an ER-range from 0 to 1. When available, ERs were calculated for the following four time points: (1) before treatment, (2) during treatment, (3) less than 1 year after treatment, and (4) 1 year or longer after treatment. A random effects model was chosen over a fixed effects model as ERs were assumed to vary according to underlying study characteristics, for example, between study populations of different ages, with different gender distributions, with different diagnoses, and receiving different combinations of treatments. The possibility of publication bias, a widespread problem when conducting meta-analysis, was inspected visually with funnel plots 40 and analyzed with Egger's method. 41 If our analyses indicated possible publication bias, it was planned to use the trim-and-fill procedure 42 to impute "missing studies" and calculate an adjusted pooled ER.
Study heterogeneity was quantified by calculating the I 2 statistic which describes the percentage of the variability in prevalence estimates that is likely to be due to heterogeneity, that is, systematic differences between studies, rather than sampling error. 40 The I 2 statistic tends to overestimate heterogeneity when the number of studies in the meta-analysis is small and should be interpreted with caution. As recommended, we therefore supplement the point estimate I 2 with its confidence interval. 43 Based on the prevalence estimate from the meta-analysis, we calculated two interval estimates. In addition to the confidence interval (i.e., the precision of the estimate), we identified the 95% prediction interval based on Tau as an estimate of the standard deviation of the true effect sizes (μ � Z α √τ 2 ). The prediction interval represents the range of values that the results of 95% of future similar studies are expected to fall within. 44 If three or more studies reported relevant data, we conducted sub-group analyses of ERs for each of the symptoms with respect to assessment time point (before, during, <1 year after, ≥1 year after treatment), melanoma diagnosis (uveal and cutaneous melanoma), and disease stage (early (I-II) and late stage (III, IV)). Finally, we conducted exploratory meta-regression analyses to investigate the role of potential continuous moderators (mean age of the sample, percent women in the sample, time since treatment, publication year) as well categorical moderators (diagnosis and treatment types, i.e., surgery, chemotherapy, radiotherapy, classical immunotherapy, ICIs, and targeted therapy).
Finally, if means and standard deviations or standard errors were reported for a sufficiently large number of studies, we calculated the pooled means for the most commonly used instruments, for example, the Hospital Anxiety and Depression Scale (HADS). 45 All analyses were conducted using Comprehensive Meta-Analysis, version 3 46 and various formulas in Microsoft Excel.

| Search results
A total of 1798 articles were identified through database searches and an additional eight from other sources. After removal of duplicates, a total of 1410 records were eligible for abstract screening. A total of 1108 records were excluded after screening, leaving 302 articles eligible for full text screening. Two hundred and thirty-six articles were then excluded primarily due to reporting a "non-relevant outcome" (52%), leaving 66 papers that were included in the systematic review. Ninety-percent agreement was achieved by reviewers during the abstract review process and 96% agreement during full-text screening. All disagreements were resolved through negotiation. The study selection process is visualized in Figure 1.
Authors of 31 studies were contacted and asked to provide additional data. Ten authors replied and provided the requested data.

| Study characteristics
The participants, treatments, symptoms examined, and other characteristics of the included studies are summarized in Table 1 Approximately one-third of the studies (K = 24 (36%)) had used a longitudinal design, that is, presented prevalence data for more than one time-point.

| Quality rating and risk of bias
See online Table S1 for an overview of the quality ratings of each study. Fifty-three studies were assessed to be of good quality (>9 criteria met) and 13 studies to be of fair quality (5-9 criteria meet).
Overall, studies received high ratings in terms of presenting clear

| Anxiety
The most commonly used scale to assess anxiety was the HADS (K = 33), followed by the State-Trait Anxiety Inventory (K = 6). As seen in Table 2, the highest prevalence of anxiety was found before (19%) and during treatment (19%). Before treatment, more than four times as many uveal melanoma patients (30%) had clinical levels of anxiety than cutaneous melanoma patients (7%). The same pattern, although much less pronounced, was observed after treatment, but, as seen in Table 3, the difference between uveal and cutaneous melanoma only reached statistical significance for anxiety assessed before treatment. After treatment, more patients with late stage melanoma, that is, stage III and IV, had clinically significant levels of anxiety (20%) than those with early stage melanoma, that is, stage I and II (13% respectively), but, as seen in Table 3, the difference did not reached statistical significance (See online supplementary material, Figure S2 for a forest plot of anxiety after treatment T A B L E 3 Exploring sample-, disease-, and treatment-related predictors of the prevalence of clinically significant anxiety and depression in melanoma patients before, during, and after treatment-results of meta-regression analyses. HADS anxiety scores range between 0 and 21 with scores ≥8

Symptom
denoting possible clinical cases of anxiety. 45

| Depression
As seen in Table 1, the HADS was also the most commonly used scale to assess depression (K = 33), followed by the Beck Depression Inventory (K = 6). (See online supplementary material, Figure S1 for a forest plot of depression after treatment). The highest rate of depressed patients was observed during treatment (12%). Before treatment, almost three times as many uveal melanoma patients had clinically significant levels of depression (16%) than patients with cutaneous melanoma (6%). The same pattern, although much less pronounced, was observed after treatment. When comparing the prevalence with meta-regression (see Table 3), the differences only reached statistical significance for depression assessed before treatment. After treatment, more patients with late stage melanoma, that is, stage III and IV, had clinically significant levels of depression (14%) than those with early stage melanoma, that is, stage I and II (8% respectively) (see Table 2). When comparing the prevalence with meta-regression, the difference reached statistical significance (see Table 3).
We also calculated the pooled mean scores of HADS depression across assessment time-points, diagnostic subtype, and stage (see online supplementary material, Table S2). Mean depression scores were approximately twice as high before and during treatment (HADS depression: 7.9 and 7.6) compared with after treatment (HADS depression: 3.3). After treatment, mean depression scores

| Cognitive problems
Three studies reported prevalence rates of cognitive complaints 52,53,58 varying from 20% to 44% among patients with cutaneous melanoma during and/or after treatment. A total of four studies reported data on cognitive complaints measured by a selfreported scale and a range of neuropsychological tests. Of these, one study reported data before treatment, two reported data during treatment and two reported data more than 1 year after treatment. Bartels  impaired on two or more neuropsychological tests during 53 or more than 1 year after treatment. 52 No studies reporting data on cognitive complaints or impairments were eligible to be included in the meta-analysis.

| Sleep
One study reported data on sleep in melanoma survivors assessed approximately 14 months after diagnosis with the Pittsburgh Sleep Quality Index (PSQI), the Insomnia Severity Index, and actigraphy. 60 The study reported poor subjective sleep quality among melanoma survivors assessed with the PSQI (Mean = 6.6) and slightly lowered actigraphy-assessed sleep efficiency (SE) (Mean = 82.2). However, for self-reported insomnia severity, the mean ISI-score was below cutoffs for subthreshold and clinical insomnia among survivors of melanoma.

| Heterogeneity
As seen in Table 2

| Clinical implications
The present systematic review and meta-analysis has a number of practical and clinical implications. The available data show that a 1218 -DANIELSEN ET AL.
proportion of melanoma patients report clinically elevated levels of depression and anxiety, not only before and during treatment, but also after treatment completion. Our findings highlight vulnerable periods for these patients and the need to screen for distress, that is, anxiety and/or depression, as the "6th vital sign" 72 before, during, and after treatment and for providing relevant interventions for patients in need. Timely identification of clinically significant symptoms of anxiety and depression has the potential to improve overall quality of life, prognosis, and other treatment outcomes. 23,24,26,73 While there currently is no agreed-upon universal screening tool, one of the most used screening tools for depression and anxiety, the HADS, 45 was included in several studies, and our review provides pooled mean HADS depression and anxiety scores at different time points for comparison. Unfortunately, few studies had examined fatigue, cognitive problems, and sleep, preventing us from evaluating the possible factors associated with higher rates of these symptoms.

| Study limitations
Some limitations should be noted. First, the high I 2 -values indicate substantial heterogeneity of results, suggesting considerable underlying systematic differences between the available studies, which are further reflected in the relatively wide prediction intervals. were melanoma survivors. More than two-thirds (69%) had FCR scores above the suggested cut-off. 74,75 Finally, symptoms of depression and anxiety were assessed using a range of different scales, threatening across-study comparability and increasing between-study variability.

| CONCLUSION
This, to our knowledge, first comprehensive systematic review and meta-analysis of its kind in the melanoma patient and survivor population demonstrates that symptoms of depression and anxiety are prevalent in patients with melanoma, not only before and during treatment, but also after completed treatment. Furthermore, our results show that, prior to treatment, depression and anxiety are more prevalent in uveal melanoma than in cutaneous melanoma.
More research is clearly needed in the area of fatigue, cognitive problems, and sleep, not least in light of the considerable improvements in melanoma treatment in recent years.

AUTHOR CONTRIBUTIONS
All authors contributed to the protocol of this systematic review. The