Use of Medications for Treating Anxiety or Depression among Testicular Cancer Survivors: A Multi-Institutional Study

Testicular cancer, the most common cancer affecting young men (18–39 years), is highly curable with a 10-year survival rate of 95% (1). The introduction of cisplatin-based chemotherapy (CBCT) in the 1970s largely contributed to this increased survival (1). However, the high cure rate is offset by long-term morbidities such as hearing loss (HL), tinnitus, peripheral neuropathy, and nephrotoxicity as well as psychosocial effects (anxiety or depression), cognitive impairment, and poorer health-related quality of life (HRQOL; ref. 2).

Testicular cancer survivors (TCS) can experience anxiety or depression due to cancer diagnosis or treatment, with prevalence estimates ranging from 10% to 20%, which when compared with normative populations are modestly greater for anxiety, but inconclusive regarding depression (3). Numerous sociobehavioral and psychologic factors contribute to anxiety or depression. For example, anxiety is associated with younger age (4), lower educational levels (5–7), unemployment (5), living alone (7), economic concerns (4), alcohol abuse (4), sexual dysfunction (4), fear of relapse (4), prior psychological distress (4, 7), and peripheral neuropathy (4). Depression in TCS is related to older age (7–9), lower educational level (6, 7), living alone (7, 10), previous psychologic distress (7), low physical activity (11), and smoking (12).

A higher number of coexisting conditions also constitute risk factors for anxiety and/or depression in cancer survivors (13). The most prevalent adverse health outcomes (AHO) related to CBCT in TCS are tinnitus, hearing impairment, and peripheral neuropathy (14, 15). Our previous study showed that 16.4% TCS experience peripheral neuropathy plus tinnitus and/or hearing impairment (14). Both hearing impairment (16, 17) and tinnitus (18) are associated with anxiety and depression in the general population, and peripheral neuropathy (4, 19) is related to both conditions in cancer survivors. However, to our knowledge, no large study to date has investigated the association between the number and severity of cisplatin-related AHOs and anxiety and depression in TCS.

To provide new information with regard to the association of sociodemographic and cisplatin-related AHOs (hearing impairment, tinnitus, peripheral sensory neuropathy, or kidney disease; refs. 1, 2, 20) with the use of medications to treat anxiety and depression, we studied 1,802 TCS enrolled in the Platinum (Pt) Study, a large, multicenter clinical investigation. We also evaluated the association between the cumulative burden of morbidity (CBM_Pt) score for cisplatin-related AHOs (1, 2, 20) and the use of these medications.

Patients were enrolled in the Platinum Study at nine sites in the United States, Canada, and United Kingdom. Cohort methods were described previously (14, 15). Briefly, eligible patients had histologic/serologic diagnosis of germ cell tumor (GCT), were ≥18 years at consent, completed first-line CBCT for either initial GCT or recurrence after active surveillance, had no subsequent salvage chemotherapy, no radiotherapy, or no antecedent chemotherapy for another primary cancer, and were followed up routinely at enrolling sites. All participants were disease-free at clinical evaluation and referred to as TCS. Each enrolling site's Institutional Review Board approved the investigation, and all participants provided written informed consent. This study included 1,802 TCS ≥12 months post-CBCT not treated with carboplatin.

Detailed clinical characteristics were abstracted from medical records by trained personnel. Collected data included cancer diagnosis and histology as well as treatment information (dates of administration, number of cycles, and cumulative cisplatin dose) as described previously (15).

Sociodemographic and health behavior information were collected through self-reporting at study enrollment using validated questionnaires (21–24). Sociodemographic characteristics included education, marital status, employment status, and health insurance coverage. Health behaviors included smoking status (never, former, or current), alcohol consumption, and physical activity in the past year. Participants were asked about types of physical activity and average time per week spent on each one. A metabolic equivalent task (MET) score was assigned to each physical activity based on its energy cost. One MET is defined as the amount of energy expended while sitting quietly (24). Vigorous physical activity was defined based on activities with MET value ≥6 (23).

AHOs including patient-reported HL, tinnitus, peripheral sensory neuropathy, and kidney disease were collected through validated questionnaires (25–27). AHOs were mapped to individual AHOs using a modified version of the CTCAE version 4.03 (28) and graded 0 to 4 based on severity (Supplementary Table S1; ref. 15).

HL, tinnitus, peripheral sensory neuropathy, and kidney disease were included in a cisplatin-related CBM_Pt score because each has been previously shown to be related to CBCT (1, 2, 20). A CBM_Pt score was calculated for each TCS based on the number and severity of AHOs following methods adapted from Geenen and colleagues (29) using a modified version described by Kerns and colleagues (30) that removed autonomic neuropathy (Supplementary Tables S1 and S2).

Patient-reported prescription drug use included current medications at study enrollment if taken consistently for >1 month. TCS were also asked to provide the indication for each medication. Medications were categorized as used for anxiety and/or depression according to pharmacologic classes defined by the National Center for Health Statistics (NCHS; refs. 31, 32), and included benzodiazepines, tricyclic antidepressants, selective serotonin reuptake inhibitors, monoamine oxidase inhibitors, phenylpiperazine antidepressants, tetracyclic antidepressants, serotonin-norepinephrine reuptake inhibitors, and miscellaneous antidepressants or anxiolytics (e.g., buspirone). Anxiety and/or depressive disorders were defined as anxiety, depression, obsessive compulsive disorder, panic attacks, and posttraumatic stress disorder. If patients reported that they used one of the listed medications for a disorder other than the ones specified, they were not considered users of medications for anxiety and/or depression.

Percentages were used for categorical variables and medians (ranges) for discrete and continuous variables. Binomial logistic regression models tested the univariate association of sociodemographic characteristics, health behaviors, individual AHOs, and CBM_Pt with medication use for anxiety and/or depression (yes/no) as the dependent variable. Variables with Wald χ² test P ≤ 0.25 were included in the multivariable model following methods of Hosmer and colleagues (33). Two multivariable models were fitted: (i) variables selected from univariate models in addition to individual AHOs; (ii) variables selected from univariate models in addition to CBM_Pt. Final multivariable models were developed by backward elimination of variables with Wald χ² test P > 0.10. ORs and 95% confidence intervals (CI) were reported for both univariate and multivariable (adjusted) models.

Multicollinearity between independent variables was assessed for final multivariable models using the variance inflation factor (VIF; ref. 34). All tests were two-sided, with statistical significance P < 0.05. All analyses were performed using SAS version 9.4 (SAS Institute).

The use of medications for anxiety and/or depression in our population was compared with that reported to the National Health and Nutrition Examination Survey (NHANES) for survey years 2017 and 2018 (35). NHANES is a program of the NCHS designed to obtain demographic, health, dietary, and laboratory data from a representative sample of U.S. residents. Utilization of medications for anxiety and/or depression was determined from self-reported medication use and self-reported medication indication. NCHS has classified medications to pharmacologic classes including anxiolytics and antidepressants (31, 32). Medication indication was classified using up to three International Classification of Disease-10 codes [F32.9: major depressive disorder, single episode, unspecified; F33.9: major depressive disorder, recurrent, unspecified; F41.0: panic disorder (episodic paroxysmal anxiety) without agoraphobia; F41.9: anxiety disorder, unspecified; F42: obsessive-compulsive disorder; F43.1: posttraumatic stress disorder; and F43.9: reaction to severe stress, unspecified]. Those who reported pain or other reasons as the indication (M79.2: neuralgia and neuritis, unspecified; R53: malaise and fatigue; M79.7: fibromyalgia; M62.83: muscle spasm; M54: dorsalgia; M19.9: osteoarthritis, unspecified site; M79.1: myalgia; R52: pain, unspecified) for anxiety and/or depression medication use (N = 9) were excluded.

To account for the complex sample design, the entire NHANEs population was used to estimate the SEs, with subpopulation analysis to compare those who were male, ≥18 and ≤65 years, and without a history of cancer between the Platinum Study population and NHANES. For comparisons, published NHANES sampling weights, primary sampling units, and strata were used, whereas a strata, a primary sampling unit, and a weight of 1 were applied to our population. The logistic regression model for anxiety and/or depression medication use was adjusted for age, education, and race/ethnicity. Marginal standardization (36) was used to estimate predicted probabilities for anxiety and/or depression medication use in the Platinum Study and in NHANES, reflecting weighted averages over the confounder distributions of age, education, and race/ethnicity in NHANES. Estimates were performed in Stata v16.1.

A total of 1,802 TCS were studied (Table 1). Median age at clinical evaluation and median time since chemotherapy completion were 37 years and 3.8 years, respectively. Over 90% of TCS were treated with etoposide and cisplatin with/without bleomycin (BEP/EP). Most participants were white (87.7%), college-educated (64.2%), employed (89%), and had health insurance (91%). Participants were largely never-smokers (57.3%) and had high engagement in vigorous physical activity (67.9%).

A total of 151 participants (8.4%) used medications for anxiety and/or depression in the Platinum Study. After adjustment for race/ethnicity, age, and education level to the NHANES distribution, the predicted probability for anxiety and/or depression medication use was 6.3% (95% CI, 4.61–7.96) in the Platinum Study compared with 7.4% (95% CI, 5.20–9.59%) in NHANES (P value for difference = 0.51). The OR for anxiety and/or depression medication use in the Platinum Study was not significantly different than NHANES (OR, 0.83; 95% CI, 0.47–1.48).

Table 2 shows frequencies of CBM_Pt scores and individual AHOs among all TCS and stratified by medication use for anxiety or depression. Overall, only 511 (28.4%) participants reported no cisplatin-related AHO. CBM_Pt scores of very low, low, medium, and high were noted in 622 (34.5%), 334 (18.5%), 287 (15.9%), and 48 (2.7%) TCS, respectively. Among 1,651 TCS not taking medications for anxiety and/or depression, any grade of HL, tinnitus, peripheral sensory neuropathy, and kidney disease were reported by 36.3%, 37.8%, 53.7%, and 2.1% TCS, respectively. In contrast, corresponding percentages were greater (56.3%, 57.7%, 72.8%, and 4.7%, respectively) for 151 TCS using these medications. Similarly, the prevalence of higher scores for CBM_Pt (low, medium, and high) was greater in survivors using medications for anxiety and/or depression versus those who did not take these medications.

Supplementary Table S3 shows univariate associations of various factors and medication use for anxiety and/or depression. However, we emphasize here the multivariable results. In the final multivariable model (including tinnitus, peripheral sensory neuropathy, employment status, health insurance coverage, and engagement in vigorous physical activity), being employed (OR, 0.39; 95% CI, 0.23–0.66) and engaging in vigorous physical activity (OR, 0.63; 95% CI, 0.44–0.89) were associated with significantly less use of medications for anxiety and/or depression, whereas having health insurance was associated with greater medication use (OR, 2.15; 95% CI, 1.01–4.56; Table 3). The likelihood of medication use for anxiety and/or depression was significantly increased among TCS with grade 2 tinnitus (OR, 2.2; 95% CI, 1.25–3.88), grade 3 tinnitus (OR, 2.74; 95% CI, 1.89–4.75), and grade 2 peripheral sensory neuropathy (OR, 2.2; 95% CI, 1.33–3.63), with results for grade 3 neuropathy of borderline significance (OR, 1.68; 95% CI, 0.99–2.87).

The second multivariable model (including CBM_Pt score, employment status, health insurance coverage, and engagement in vigorous physical activity) indicated similar results. Being employed (OR = 0.38, P < 0.05) and engaging in vigorous physical activity (OR = 0.63, P < 0.05) were associated with a significantly reduced use of medications for anxiety and/or depression, whereas having health insurance was associated with greater use (OR = 2.12, P < 0.05). Significantly increased odds for medication use for anxiety and/or depression were also observed among TCS with CBM_Pt scores of low (OR, 2.96; 95% CI, 1.67–5.24), medium (OR, 3.47; 95% CI, 1.95–6.18), and high (OR, 3.18; 95% CI, 1.22–8.3; Table 3). In addition, we also included country of enrollment in both multivariable models and found that study region (the United States, UK, and Canada) did not alter the association between employment status and use of medications for anxiety and depression. No multicollinearity was detected between variables in either model (VIF < 5).

In a large cohort of TCS (N = 1,802), we investigated for the first time four cisplatin-related AHOs (HL, tinnitus, peripheral sensory neuropathy, and kidney disease) as well as CBM_Pt scores and sociodemographic characteristics in relation to the use of medications for anxiety and/or depression. New findings include that higher grades of tinnitus, peripheral sensory neuropathy, and a higher CBM_Pt score were significantly associated with increased use of these medications as was having health insurance. Conversely, being employed and engaging in vigorous physical activity were associated with a significantly lower use of these medications.

Anxiety and depression are associated with a reduced HRQOL in TCS (37). Pharmacologic or nonpharmacologic interventions (cognitive behavioral therapy, stress reduction therapy, etc.) can be utilized for treatment (13), with evidence showing that antidepressant and anxiolytic medications are safe and effective in patients with cancer (38–40). They are typically recommended for use in severe cases of anxiety and depression (41, 42), but most cancer survivors do not discuss cancer-related psychologic effects with healthcare providers (43) due to lack of awareness or identification of symptoms as stigma (43, 44). Only Hawkins and colleagues (45) investigated the characteristics of U.S. cancer survivors (80% aged ≥50 years) at higher risk of prescription use for anxiety and depression. Patients with various cancer types (e.g., breast, prostate, melanoma, cervix, colorectal, hematologic, ovary, uterus, and others) were studied, but it was not clear whether TCS were included. Hawkins and colleagues (45) analyzed only the number of comorbidities, not the related severity, associated with anxiolytic and antidepressant use. These investigators concluded that, in general, U.S. cancer survivors took medications for anxiety and depression about twice as frequently as the general population, likely due to cancer treatment-related physical and emotional distress (45). However, little is known with regard to the characteristics of U.S. TCS (46% age <50 years; ref. 46) prescribed medications for anxiety and depression. To our knowledge, this large study is the first to identify characteristics of TCS associated with prescription medication use for anxiety and/or depression in addition to assessing the association with the number and severity of cisplatin-relation AHOs.

To our knowledge, this is the first study to report an association between a greater CBM_Pt score and an increased use of medications for anxiety and/or depression in adult-onset cancer survivors. Our findings mirror those of Hawkins and colleagues (45) who showed an increased use of these medications in cancer survivors with larger numbers of chronic health conditions. Previous investigations of risk factors for anxiety and depression among adult-onset cancer survivors also confirm associations of number of comorbid conditions with anxiety/depression (47, 48). A higher number of comorbidities can result in greater complexities in disease management, the use of multiple prescription medications, less physical activity, and less social interaction (48). SEER-Medicare linkage studies of colorectal cancer survivors show a significant relationship between a higher number of comorbidities and poorer HRQOL, physical health, and anxiety and depression (48).

Higher grades of tinnitus were associated with a 2- to 3-fold greater use of medications for anxiety and depression. Although Dahl and colleagues (4) found no significant association between anxiety disorders in TCS and a combined variable of “tinnitus/hearing problems,” tinnitus was not evaluated independently. Moreover, Dahl and colleagues used a four-item Likert scale, with responses analyzed as binary variables, e.g., by collapsing responses of “quite a bit” and “very much” as cases and responses of “none” and “a little bit” as controls. In contrast, we used a considerably finer grading of AHOs based on CTCAE criteria (15, 28). In the general population, the association between tinnitus and anxiety and depression is well-established (49–52) and has been attributed to poorer HRQOL, sleep disorders, and reduced work productivity (49).

Higher grades of peripheral sensory neuropathy were associated with a 2-fold greater use of medications for anxiety and/or depression. Our finding confirms previous reports in either TCS or other adult-onset cancer survivors of significant associations between peripheral neuropathy and anxiety (4, 53–55) or depression (54–56). Because our study is cross-sectional, causal inferences cannot be made; moreover, the relation between peripheral neuropathy and anxiety/depression may be bidirectional. The reciprocal adverse effects of pain and depression/anxiety on one another are well-established (57, 58). The pain and reduced physical activity associated with peripheral neuropathy could increase anxiety/depression. On the other hand, patients with anxiety might experience and report a greater degree of neuropathy, perhaps due to shared biological pathways involved in the production of inflammatory cytokines in both anxiety and peripheral neuropathy (54). Similarly, patients with depression may experience more severe and persistent pain as the result of deficiencies in neurotransmitters (e.g., serotonin, norepinephrine) that may occur in depressed states (59). Thus, it has been suggested that antidepressants can decrease pain signals by increasing serotonin and norepinephrine levels in the brain (59).

Having health insurance was associated with an increased use of medications for anxiety and/or depression, which likely reflects readier access to the health care system, including paid prescription medications (45).

Our TCS engaged in vigorous physical activity were about 37% less likely to use medications for anxiety and/or depression. Physical activity can improve symptom management and HRQOL (physical, psychologic, social, and spiritual) among cancer survivors (60). Moreover, physical activity is recommended in cancer survivorship guidelines published by the National Comprehensive Cancer Network, in part due to its effectiveness in reducing symptoms of both anxiety and depression (61).

Being employed was associated with a significantly lower use of medications for anxiety and/or depression. The association between unemployment and depression has been shown in the general population (62–64). We previously showed that peripheral sensory neuropathy (OR = 2.44: grade 3 vs. 0, P = 0.006), patient-reported HL (OR = 1.82: grade 2/3 vs. 0, P = 0.04), and pain (OR = 3.75: grade 2/3 vs. 0, P < 0.001) were associated with TCS unemployment (30). However, reasons for unemployment are mixed, because influences related to characteristics of both cancer survivors (e.g., sociodemographic and socioeconomic factors, and comorbidities) and employers (e.g., accommodation for cancer-related and treatment-related adverse effects) can influence employment status (65). Thus, multidisciplinary approaches, including physical, psychologic, and vocational considerations, are required to understand employment and work outcomes among cancer survivors (65).

Strengths of this study include the large sample size and comprehensive assessment of sociodemographic features, health behaviors, and cisplatin-related AHOs using validated questionnaires. Medications used for anxiety and/or depression were defined according to pharmacologic classes using published metrics (31, 32), and stringent rules classified patients as users of these medications only if they specifically indicated that the reason for use was for one of these conditions. A rigorous approach was used to select independent variables and build multivariable models (45). Limitations include those inherent to all cross-sectional designs in that causal inferences cannot be made, because the temporal sequence of events cannot be ascertained. For example, it is not known whether unemployment preceded the use of medications for anxiety and/or depression or vice versa. In addition, patients with anxiety or depression may not have been prescribed these medications, may not have been compliant in taking these medications, or may not have reported medication use. Therefore, the number of patients taking medications for anxiety or depression could be less than the actual number of patients with these disorders. Moreover, information with regard to medication use for anxiety and or depression, employment status, and individual cisplatin-related AHOs (HL, tinnitus, peripheral neuropathy, and kidney disease) prior to chemotherapy was not available. Thus, it is not known whether these factors were cancer/chemotherapy-related; however, given the young median age of our population, it is unlikely that AHOs were pre-existing (3).

In view of the significant associations we found between vigorous physical activity and less usage of medications for anxiety and/or depression, healthcare providers should encourage TCS to increase physical activity to improve both physical and mental health (66). In order to support employment and work resumption after cancer treatment, rehabilitation programs should incorporate tools that improve physical and psychologic performance, assess work-related skills, and provide career development counseling/training and when possible, counseling for employers (65).

The association of higher grades of tinnitus and peripheral neuropathy with greater use of medications for anxiety and/or depression should prompt healthcare providers to actively address these conditions in TCS. Strategies such as acoustic stimulation, cognitive behavioral therapy, and educational counseling can alleviate tinnitus symptoms (67). Although no medications have been approved to treat tinnitus (18), antidepressants may have modest efficacy as an adjunct treatment to nonpharmacologic treatments (52, 68). To manage chemotherapy-induced peripheral neuropathy, guidelines support the use of duloxetine to treat associated pain, but there is limited scientific evidence for the use of other medications (69). Further research is required to prospectively investigate the efficacy of symptom management interventions for tinnitus and peripheral neuropathy in reducing anxiety and depression.

S. ArdeshirRouhaniFard reports grants from NCI during the conduct of the study and that her spouse is an employee of Eli Lilly company and holds stock shares in the company. P.O. Monahan reports grants from NIH during the conduct of the study. R. Huddart reports grants from NIHR Biomedical Research Council and grants from Josh Carrick Foundation during the conduct of the study. D.R. Feldman reports grants from NIH/NCI during the conduct of the study; and other support from Decibel Inc., Astellas, Seattle Genetics, and Novartis; and personal fees from UpToDate outside the submitted work. L.B. Travis reports grants from NCI during the conduct of the study. No disclosures were reported by the other authors.

S. ArdeshirRouhaniFard: Conceptualization, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. P.C. Dinh: Formal analysis, writing–review and editing. P.O. Monahan: Methodology, writing–review and editing. S.D. Fossa: Conceptualization, writing–review and editing. R. Huddart: Conceptualization, data curation, writing–review and editing. C. Fung: Conceptualization, data curation, writing–review and editing. Y. Song: Methodology, writing–review and editing. D.R. Feldman: Data curation, writing–review and editing. R.J. Hamilton: Data curation, writing–review and editing. D.J. Vaughn: Data curation, writing–review and editing. N.E. Martin: Data curation, writing–review and editing. C. Kollmannsberger: Data curation, writing–review and editing. L. Einhorn: Data curation, writing–review and editing. K. Kroenke: Conceptualization, writing–review and editing. L.B. Travis: Conceptualization, resources, supervision, funding acquisition, methodology, writing–review and editing.

This work was supported by the NCI (1R01 CA157823 to L.B. Travis).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.