Impact of pain and adverse health outcomes on long-term US testicular cancer survivors

Abstract

Background

No study has quantified the impact of pain and other adverse health outcomes on global physical and mental health in long-term US testicular cancer survivors or evaluated patient-reported functional impairment due to pain.

Methods

Testicular cancer survivors given cisplatin-based chemotherapy completed validated surveys, including Patient-Reported Outcomes Measurement Information System v1.2 global physical and mental health, Patient-Reported Outcomes Measurement Information System pain questionnaires, and others. Multivariable linear regression examined relationships between 25 adverse health outcomes with global physical and mental health and pain-interference scores. Adverse health outcomes with a $\widehat{\mathrm{\beta }}$  of more than 2 are clinically important and reported below.

Results

Among 358 testicular cancer survivors (median age = 46 years, interquartile range [IQR] = 38–53 years; median time since chemotherapy = 10.7 years, IQR = 7.2–16.0 years), median adverse health outcomes number was 5 (IQR = 3–7). A total of 12% testicular cancer survivors had 10 or more adverse health outcomes, and 19% reported chemotherapy-induced neuropathic pain. Increasing adverse health outcome numbers were associated with decreases in physical and mental health (P < .0001 each). In multivariable analyses, chemotherapy-induced neuropathic pain ($\widehat{\mathrm{\beta }}$ = −3.72; P = .001), diabetes ($\widehat{\mathrm{\beta }}$ = −4.41; P = .037), obesity ($\widehat{\mathrm{\beta }}$ = −2.01; P = .036), and fatigue ($\widehat{\mathrm{\beta }}$ = −8.58; P < .0001) were associated with worse global mental health, while being married or living as married benefited global mental health ($\widehat{\mathrm{\beta }}$ = 3.63; P = .0006). Risk factors for pain-related functional impairment included lower extremity location ($\widehat{\mathrm{\beta }}$ = 2.15; P = .04) and concomitant peripheral artery disease ($\widehat{\mathrm{\beta }}$ = 4.68; P < .001). Global physical health score reductions were associated with diabetes ($\widehat{\mathrm{\beta }}$ = −3.81; P = .012), balance or equilibrium problems ($\widehat{\mathrm{\beta }}$ = −3.82; P = .003), cognitive dysfunction ($\widehat{\mathrm{\beta }}$ = −4.43; P < .0001), obesity ($\widehat{\mathrm{\beta }}$ = −3.09; P < .0001), peripheral neuropathy score ($\widehat{\mathrm{\beta }}$ = −2.12; P < .0001), and depression ($\widehat{\mathrm{\beta }}$ = −3.17; P < .0001).

Conclusions

Testicular cancer survivors suffer adverse health outcomes that negatively impact long-term global mental health, global physical health, and pain-related functional status. Clinically important factors associated with worse physical and mental health identify testicular cancer survivors requiring closer monitoring, counseling, and interventions. Chemotherapy-induced neuropathic pain must be addressed, given its detrimental impact on patient-reported functional status and mental health 10 or more years after treatment.

Testicular cancer is the leading cancer in men aged 20–39 years (1). Because of the effectiveness of cisplatin-based chemotherapy, overall 10-year relative survival rates exceed 95% (2). Given the young age at diagnosis, excellent survival, and increasing incidence of testicular cancer in the United States (3-5), a growing population is at risk for the late effects of cisplatin-based chemotherapy (6). However, no study has comprehensively evaluated the patient-reported impact of treatment-related pain and other adverse health outcomes on US-specific measures of global physical and mental health or examined pain-related functional impairment. Patient-reported outcomes are increasingly recognized as clinically important to enable risk-adapted follow-up (7-10). The National Institutes of Health Patient-Reported Outcomes Measurement Information System (PROMIS) (11-16) provides reliable, precise outcome measurement, with results normed to the US population through validation studies of more than 20 000 participants (14,17).

The PROMIS Global Health survey addresses mental and physical functional status (17), integrating results into 2 overall scores (physical and mental) for comparison with US population norms. Scores can be used clinically to risk-stratify patients for follow-up and targeted interventions and to quantify adverse health outcomes impact on mental and physical functional status. Other PROMIS questions quantify pain intensity and, importantly, patient-reported functional impairment due to pain (16,18,19). Chronic neuropathic pain is one of the most incapacitating neurologic conditions (20-22) and often accompanied by disability, dysfunction, and depression (23). Although Kerns et al. (24) linked moderate-to-severe pain in testicular cancer survivors to 4- to 11-fold increased risks of disability or unemployment, detailed pain information was not collected, although it appeared related to chemotherapy-induced peripheral neuropathy (P < .001). A comprehensive analysis of chemotherapy-induced neuropathic pain and related functional impairment has not been undertaken to date in any testicular cancer survivor population (25-27). Further, quality-of-life studies in testicular cancer survivors have not distinguished neuropathic pain (26-30) and/or included nonspecific bodily pain as just one component of quality of life (27-34). Understanding pain’s effect on functional status is critical to facilitate risk-adapted follow-up and management strategies, including referral to specialists as needed.

To address these important gaps, we administered PROMIS questionnaires and other validated surveys (15-17,35-46) to a subset of testicular cancer survivors in a large, multicenter investigation (Platinum Study). We sought to determine which adverse health outcomes had the greatest deleterious effect on global physical and mental health and quantify for the first time chemotherapy-induced neuropathic pain and its functional impact, using specific PROMIS instruments.

Methods

Cisplatin-treated testicular cancer survivors were enrolled at 8 cancer centers in the Platinum Study (2012−2018) (47-49), completing questionnaires and undergoing physical examinations (47-51). Entry criteria were histologic and/or serological germ cell tumor diagnosis as a first primary cancer (age 55 years and younger), completion of first-line cisplatin-based chemotherapy at least 1 year previously, and routine follow-up at participating sites. Informed consent was obtained from all participants (48). For participants at Indiana University, Memorial Sloan Kettering, Dana-Farber Cancer Institute, University of Rochester, and University of Pennsylvania, institutional review boards approved the administration of a subsequent survey; entry criteria for this investigation included enrollment into the original study at these sites. This report includes all patients with complete surveys through September 13, 2022. Demographic and clinical information, including medical history, lifestyle, and comorbidities, were gathered with standardized questionnaires. Information on adverse health outcomes (identified through comprehensive literature reviews) was collected with validated instruments (15-17,35-45). Supplementary Appendix A1 (available online) describes demographic variables and all 25 adverse health outcomes, questions, and scoring; these included platinum-related adverse health outcomes (hearing damage or impairment, tinnitus, kidney disease, neuropathy), cardiovascular adverse health outcomes (eg, hypertension, coronary artery disease), and others (eg, diabetes, problems with balance, equilibrium, or ability to reach). See Table 1 for a list of all adverse health outcomes.

Table 1.

Clinical features, sociodemographic characteristics, health behaviors, and adverse health outcomes for 358 survivors of cisplatin-treated germ cell tumors^a

Variables	Total No. (%) (n = 358)
Age at testicular cancer diagnosis, median (1st quartile, 3rd quartile), y	31 (25, 40)
Age at survey, median (1st quartile, 3rd quartile), y	46 (38, 53)
Time since chemotherapy, median (1st quartile, 3rd quartile), y	10.7 (7.2, 16.0)
Cumulative cisplatin dose, median (1st quartile, 3rd quartile), mg/m2b	400 (300, 400)
Retroperitoneal lymph node dissectionc	149 (42.0)
Raced
Asian	3 (0.8)
Black or African American	4 (1.1)
Multiple races	9 (2.5)
Other race	3 (0.8)
White	338 (94.7)
Ethnicitye
Hispanic or Latino	11 (3.1)
Not Hispanic or Latino	347 (96.9)
Married or living as marriedf	283 (80.4)
Educationg
High school or less	20 (5.7)
Training after high school or some college	54 (15.3)
College graduate or postgraduate	278 (79.0)
Current employmenth
Employed	312 (88.4)
Unemployed	12 (3.4)
On disability	8 (2.3)
Retired	21 (5.9)
Alcohol intake
Rarely, never	93 (26.0)
1-3 per month	43 (12.0)
1-6 per week	166 (46.4)
≥1 per day	56 (15.6)
Smoking statusi
Never	238 (73.5)
Former or current	86 (26.5)
Pack-yearsj
0 pack-years	247 (76.2)
>0 to <5 pack-years	38 (11.7)
≥5 pack-years	39 (12.0)
Body mass index, median (1st quartile, 3rd quartile), kg/m2k	27 (24, 30)
Sedentary time, median (1st quartile, 3rd quartile), h/wkl	42 (20, 65)
Physical activity: vigorous, ≥6 METs	202 (56.4)
Physical activity: moderate, 3 to <6 METs	341 (95.3)
No. and types of adverse health outcomesm
Adverse health outcomes, all: median (1st quartile, 3rd quartile)	5 (3, 7)
Adverse health outcomes, all: number
0	8 (2.2)
1-2	63 (17.6)
3-4	97 (27.1)
5-9	146 (40.8)
≥10	44 (12.3)
Platinum-related adverse health outcomesm
Peripheral sensory neuropathy, source: EORTC-CIPN20	206 (57.5)
Hearing damage or impairment	202 (56.4)
Tinnitus	218 (60.9)
Kidney disease	4 (1.1)
Cardiovascular disease adverse health outcomesm,n
Hypertension	103 (28.8)
Thromboembolic event (deep vein thrombosis or pulmonary embolism)	27 (7.5)
Peripheral artery disease	11 (3.1)
Coronary artery disease	11 (3.1)
Other adverse health outcomesm,o
Painp	69 (19.3)
Fatigueq	76 (21.2)
Cognitive dysfunctionq	50 (14.0)
Diabetes	16 (4.5)
Obesity	102 (28.5)
Hypercholesterolemia	132 (36.9)
Problems with balance, equilibrium, or ability to reach	24 (6.7)
Erectile dysfunction	132 (36.9)
Infertility	89 (24.9)
Raynaud phenomenon	154 (43.0)
Thyroid disease	16 (4.5)
Hypogonadism	79 (22.1)
Anxietyq	93 (26.0)
Depressionq	95 (26.5)
PROMIS global physical healthr
Excellent	48 (13.4)
Very good	187 (52.2)
Good	95 (26.5)
Fair, poor	28 (7.8)
PROMIS global mental healths
Excellent	63 (17.6)
Very good	140 (39.1)
Good	111 (31.0)
Fair, poor	44 (12.3)

^aData reported herein are from an ongoing study. Patients with complete surveys through September 13, 2022, are included, representing all 5 study sites. EORTC-CIPN20 = European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-Chemotherapy-Induced Peripheral Neuropathy 20-Item Scale (42); MET = Metabolic equivalent of task; N/A = not available; PROMIS = Patient-Reported Outcomes Measurement Information System.

^bThree patients did not have this data available.

^cThree patients did not have this data available.

^dThe epidemiology of testicular cancer is such that it largely affects White males, as reflected in this study and reviewed by McGlynn et al. (112). One patient preferred not to designate race and was of Hispanic or Latino ethnicity. Three additional patients did not designate race, and data were obtained from information provided at initial enrollment: 2 multiple race patients and 1 white patient. Three patients were categorized as “Other race” from selecting “Other, please specify” on the questionnaire; 2 of these further specified their races as Persian, and 1 declined to specify.

^eSix patients preferred not to designate ethnicity, and data were obtained from information provided at initial enrollment: all 6 were not Hispanic or Latino.

^fSix patients did not answer or preferred not to designate marital status.

^gTwo patients did not answer or preferred not to specify educational level, and 4 patients answered “Other” for the highest level of education.

^hFive patients did not answer or preferred not to designate current employment status.

ⁱSix patients are current smokers (1.9%), and 80 patients are former smokers (24.7%); 34 patients did not answer the tobacco use questions.

^jThirty-eight patients did not answer the pack-year question: 34 patients did not answer any tobacco-related questions, and 4 patients did not answer the pack-year question. For 4 of these 38 patients, we obtained pack-year data from information provided at the time of initial study enrollment. The remaining 34 patients were placed in a category of N/A for analyses.

^kOne patient did not report height and weight.

^lTwo patients did not answer the sedentary time questions. Two patients reported more than 270 hours of sedentary time per week and were thus set to missing.

^m Supplementary Appendix A1 (available online) describes all 25 adverse health outcome, questions, and scoring. Unless otherwise noted, adverse health outcomes are presented above as the No. (%) of patients within the indicated adverse health outcomes category, where % is percentage of the total patients (n = 358).

ⁿAn adverse health outcomes of cerebrovascular disease, consisting of stroke and transient ischemic attack, is not shown because of small numbers (n = 4 patients). Two of these patients had strokes, and the other 2 had transient ischemic attacks. An adverse health outcome of congestive heart failure is not shown because of small numbers (n = 1 patient).

^oAn adverse health outcome of cytopenia is not shown because of small numbers (n = 4 patients).

^pAmong 69 patients reporting chemotherapy-induced neuropathic pain, 15 reported the use of prescription pain medication. Ten patients reported the use of a single medication (6 gabapentin, 2 oxycodone, 1 oxycodone-acetaminophen, and 1 buprenorphine). Three patients reported 2 medications (gabapentin and celecoxib; gabapentin and duloxetine; and gabapentin and amitriptyline). One patient reported 3 medications (gabapentin, oxycodone, and fentanyl patch), and 1 patient reported 4 medications (gabapentin, buprenorphine, oxcarbazepine, and pregabalin).

^qCase numbers for fatigue, cognitive dysfunction, anxiety, and depression are rigorously defined by thresholds, employing validated threshold recommendations from validated instruments, as described in detail in Supplementary Appendix A1 (available online). For these variables, means (SD), medians, interquartile ranges (1st quartiles, 3rd quartiles), and ranges (minimums, maximums) are presented in Supplementary Table 1 (available online).

^rGlobal physical health T-score cut-points for excellent, very good, good, fair, and poor were more than 58, more than 50, more than 42, and more than 35, respectively (113). The fair group consisted of 16 (4.5%) patients, and the poor group consisted of 12 (3.4%) patients.

^sGlobal mental health T-score cut-points for excellent, very good, good, fair, and poor were more than 56, more than 48, more than 40, more than 29, respectively (113). The fair group consisted of 33 (9.2%) patients, and the poor group consisted of 11 (3.1%) patients.

Global physical and mental health and pain

Overall health status was quantified using the PROMIS Global Health (v1.2) instrument (17) (see Supplementary Appendix A2 and Supplementary Methods, available online, for instrument validity, questions, and scoring). This comprehensive survey uses 10 global health items to quantify 5 primary domains (physical function, fatigue, pain, emotional distress, social health) and general health perceptions. These health domain scores contribute to global scores for physical and mental health (17). Resultant T-scores are standardized with 50 equivalent to the US general population mean (10). Higher scores denote better global physical and mental health.

Chemotherapy-induced neuropathic pain consisted of affirmative responses to the question, “Do you currently have any degree of pain in your fingers/hands or toes/feet as a result of your chemotherapy?” followed by further questions to ensure diagnostic specificity (Supplementary Appendix A3, available online). To quantify functional impairment due to pain’s impact on important domains (including social, cognitive, emotional, physical, and recreational), we administered PROMIS Pain Interference Short Form 4a (16,18,19) (see Supplementary Appendix A3, available online, for instrument validity, questions, and scoring). Higher T-scores indicate worse impairment.

Other variables

As recommended by a National Cancer Institute Clinical Trials planning meeting (52), the validated European Organization for Research and Treatment of Cancer (EORTC) Chemotherapy-Induced Peripheral Neuropathy 20 (CIPN20) (42) instrument was used to assess chemotherapy-induced peripheral neuropathy; peripheral sensory neuropathy was coded “yes” if any responses to sensory items were non-zero (Supplementary Appendix A1, available online). For multivariate analyses, an EORTC-CIPN20 sensory score to quantify severity was derived using methods adapted from Dolan et al. (53).

Statistical analyses

Descriptive statistics are provided as frequencies (proportions) for categorical variables or medians (interquartile range [IQR]) for continuous variables. PROMIS global physical health and global mental health T-scores were compared with US male population norms (54) via t tests. Bivariable associations of individual adverse health outcomes with physical and mental health were assessed using linear and logistic regression for continuous and dichotomous PROMIS scores, respectively. All bivariable models were adjusted for age, time since chemotherapy, cumulative cisplatin dose, college education, and smoking pack-years. Multivariable backward-selected linear regression models were developed for simultaneously testing associations of multiple adverse health outcomes with physical or mental health T-scores. Age at survey, time since chemotherapy, cumulative cisplatin dose, and college education were required for retention in all multivariable models. Adverse health outcome variables were initially entered into multivariable models based on adjusted bivariable results (P < .20), plausibility, and nonsparsity (“yes” counts > 10) for model stability. Final variable retention required a P value less than .05. Similar bivariable and multivariable modeling approaches were used for comparisons between adverse health outcomes and pain interference T-scores (Supplementary Methods, available online). All tests were 2-sided (α = 0.05). Because 2- to 3-point PROMIS T-score differences are considered clinically meaningful (55), a $\widehat{\mathrm{\beta }}$ of more than 2 denotes potentially important T-score differences in linear regression models between presence and absence of dichotomous adverse health outcomes and per 1 standard deviation increase in standardized (SD = 1) continuous variables (EORTC-CIPN20 sensory score; sedentary time).

Results

Table 1 presents clinical and sociodemographic characteristics, health behaviors, and adverse health outcomes for all 358 testicular cancer survivors (median age at diagnosis = 31 years; median survey age = 46 years). Median time since testicular cancer diagnosis was 11.9 years (IQR = 8.0−17.1 years); 349 (98%) were 5-year survivors. Most (79.0%) testicular cancer survivors were college educated or greater. Median adverse health outcomes number was 5 (IQR = 3−7); 12.3% of testicular cancer survivors had at least 10 adverse health outcomes. The most prevalent adverse health outcomes were platinum-related: tinnitus (60.9%), peripheral sensory neuropathy (57.5%), and hearing damage or impairment (56.4%). Other highly prevalent adverse health outcomes included Raynaud phenomena (43.0%), hypercholesterolemia (36.9%), erectile dysfunction (36.9%), hypertension (28.8%), and obesity (28.5%). Anxiety and depression affected 26.0% and 26.5%, respectively (also see Supplementary Table 1, available online).

Global physical health T-scores were classified as excellent, very good, good, and fair or poor in 13.4%, 52.2%, 26.5%, and 7.8% of testicular cancer survivors, respectively, with corresponding global mental health T-score groups of 17.6%, 39.1%, 31.0%, and 12.3%. Global physical health and global mental health T-scores were statistically significantly correlated (r = 0.64; P < .0001). Overall, testicular cancer survivors had lower mental (mean global mental health T-score = 49.2 vs 50.8; P = .0005) but similar physical health (mean global physical health T-score = 51.5 vs 51.2; P = .62) than US male norms.

Adverse health outcomes effect on PROMIS-quantified global physical health and global mental health

Supplementary Table 2 (available online) presents adjusted bivariable associations of adverse health outcomes with PROMIS global physical health and global mental health; multiple adverse health outcomes were statistically significant. In 24 testicular cancer survivors reporting problems with balance, equilibrium, or ability to reach, degree of difficulty in walking was correlated with EORTC-CIPN20 sensory score (r = 0.748; P<.0001). Increasing adverse health outcome numbers were associated with decreases in physical and mental health (P < .0001 each). Multivariable models identified which adverse health outcomes and modifiable risk factors were most strongly associated with global physical health and global mental health (Table 2; Figures 1 and 2); $\widehat{\mathrm{\beta }}$ coefficients represent PROMIS T-score mean differences between testicular cancer survivors with vs without adverse health outcomes. Diabetes ($\widehat{\mathrm{\beta }}$ = −3.81), cognitive dysfunction ($\widehat{\mathrm{\beta }}$ = −4.43), problems with balance, equilibrium, or ability to reach ($\widehat{\mathrm{\beta }}$=−3.82), EORTC-CIPN20 sensory score ($\widehat{\mathrm{\beta }}$=−2.12), depression ($\widehat{\mathrm{\beta }}$ = −3.17), hypercholesterolemia ($\widehat{\mathrm{\beta }}$ = −1.85), and hearing damage or impairment ($\widehat{\mathrm{\beta }}$ = −1.38) were independently associated with lower global physical health (P < .05 each). Worse global physical health was also statistically significantly associated with increased sedentary time ($\widehat{\mathrm{\beta }}$ = −0.73) and obesity ($\widehat{\mathrm{\beta }}$ = −3.09), while college education ($\widehat{\mathrm{\beta }}$ = 1.97) was associated with better global physical health. Chemotherapy-induced neuropathic pain ($\widehat{\mathrm{\beta }}$ = −3.72), diabetes ($\widehat{\mathrm{\beta }}$ = −4.41), obesity ($\widehat{\mathrm{\beta }}$ = −2.01), and fatigue ($\widehat{\mathrm{\beta }}$ = −8.58) were statistically significantly associated with worse global mental health, while married or living as married was related to better global mental health ($\widehat{\mathrm{\beta }}$ = 3.63).

Table 2.

Multivariable associations of adverse health outcomes with PROMIS global physical and mental health among 358 survivors of cisplatin-treated germ cell tumors^a

Select model variables	$\widehat{\beta }$ b (95% CI)	SE	P
Global physical health T-scorec
Diabetes	−3.81 (−6.80 to −0.82)	1.52	.012
Problems with balance, equilibrium, or ability to reach	−3.82 (−6.33 to −1.32)	1.27	.003
EORTC-CIPN20 sensory scored	−2.12 (−2.80 to −1.45)	0.34	<.0001
Obesity	−3.09 (−4.43 to −1.74)	0.68	<.0001
Cognitive dysfunction	−4.43 (−6.27 to −2.58)	0.94	<.0001
Depression	−3.17 (−4.59 to −1.76)	0.72	<.0001
Hearing damage or impairment	−1.38 (−2.62 to −0.15)	0.63	.027
Hypercholesterolemia	−1.85 (−3.23 to −0.46)	0.70	.009
Sedentary time, h/wkd	−0.73 (−1.32 to −0.14)	0.30	.015
College education	1.97 (0.46 to 3.48)	0.77	.010
Global mental health T-scoree
Chemotherapy-induced neuropathic pain	−3.72 (−5.95 to −1.49)	1.13	.001
Diabetes	−4.41 (−8.56 to −0.26)	2.11	.037
Obesity	−2.01 (−3.89 to −0.13)	0.96	.036
Fatigue	−8.58 (−10.72 to −6.44)	1.09	<.0001
Married or living as married	3.63 (1.55 to 5.71)	1.06	.0006

^aBackward selection linear regression models were used with age at survey, time since chemotherapy, cumulative cisplatin dose, and college education required to be in both models (see Methods). Only college education was statistically significant (α = 0.05) in the global physical health model. $\widehat{\beta }$ coefficients and P values for the nonsignificant required variables (not shown in the table) are as follows: global physical health model: age at survey, per 5 years ($\widehat{\beta }$ = 0.07, 95% CI = −0.27 to 0.41; P = .681); time since chemotherapy, per 5 years ($\widehat{\beta }$ = 0.01, 95% CI = −0.49 to 0.50; P = .976); cumulative cisplatin dose, 100 mg/m² ($\widehat{\beta }$ = 0.50, 95% CI = −0.16 to 1.16; P = .136); global mental health model: age at survey, per 5 years ($\widehat{\beta }$ = 0.07, 95% CI = −0.37 to 0.52; P = .743); time since chemotherapy, per 5 years ($\widehat{\beta }$ = −0.33, 95% CI = −1.02 to 0.36; P = .343); cumulative cisplatin dose, 100 mg/m² ($\widehat{\beta }$ = −0.16, 95% CI = −1.09 to 0.76; P = .734); and college education ($\widehat{\beta }$ = 0.41, 95% CI = −1.76 to 2.58; P = .709). CI = confidence interval; EORTC-CIPN20 = European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-Chemotherapy-Induced Peripheral Neuropathy 20-Item Scale (42); PROMIS = Patient-reported Outcomes Measurement Information System.

^b $\widehat{\beta }$ coefficients represent the PROMIS T-score mean differences between testicular cancer survivors with vs without the adverse health outcomes for dichotomous variables and per 1 standard deviation increase in the standardized (SD = 1) continuous variables (EORTC-CIPN20 sensory score and sedentary time). Because 2- to 3-point PROMIS T-score differences are clinically meaningful, a $\widehat{\beta }$ of more than 2 is noteworthy for adverse health outcomes. Negative $\widehat{\beta }$ coefficients indicate worse PROMIS global mental or physical health for the respective adverse health outcomes.

^cInitial variables for the global physical health backward selection linear regression model included the EORTC-CIPN20 sensory score, diabetes, obesity, cognitive dysfunction, hearing damage or impairment, problems with balance, equilibrium, or ability to reach, tinnitus, Raynaud phenomena, anxiety, depression, erectile dysfunction, hypercholesterolemia, hypertension, peripheral artery disease, hypogonadism, married or living as married, smoking pack-years categories, and sedentary time. Because the PROMIS global physical health scales include questions regarding pain and fatigue (see Supplementary Appendix A2, available online), pain and fatigue were not included in the pool of initial variables for the global physical health model.

^dThe $\widehat{\beta }$ coefficients for the continuous variables, EORTC-CIPN20 sensory score and sedentary time, are reported for 1 standard deviation unit.

^eInitial variables for the global mental health backward selection linear regression model included the EORTC-CIPN20 sensory score, diabetes, obesity, fatigue, hearing damage or impairment, problems with balance, equilibrium, or ability to reach, tinnitus, Raynaud phenomena, chemotherapy-induced neuropathic pain, erectile dysfunction, hypercholesterolemia, hypertension, hypogonadism, vigorous physical activity, married or living as married, smoking pack-year categories, and sedentary time. Because the PROMIS global mental health scales include questions regarding cognitive function and emotional problems (eg, depression) (see Supplementary Appendix A2, available online), cognitive function, anxiety, and depression were not included in the pool of initial variables for the global mental health model.

Figure 1.

Multivariable associations of adverse health outcomes with Patient-Reported Outcomes Measurement Information System (PROMIS) global physical health T-score. Forest plot shows the $\widehat{\mathrm{\beta }}$ coefficients (center points) and 95% confidence intervals (horizontal lines) of the backward selection linear regression model for PROMIS global physical health (refer to Methods). Age, time since chemotherapy, cumulative cisplatin dose, and college education were required to be in the model, with only college education statistically significant (α = 0.05). $\widehat{\mathrm{\beta }}$ coefficients represent the PROMIS T-score mean differences between testicular cancer survivors with vs without the adverse health outcomes for dichotomous variables and per 1 standard deviation increase in the standardized (SD = 1) continuous variables, EORTC-CIPN20 sensory score and sedentary time. Because 2- to 3-point PROMIS T-score differences are clinically meaningful, a $\widehat{\mathrm{\beta }}$ of more than 2 is noteworthy for adverse health outcomes. Negative $\widehat{\mathrm{\beta }}$ coefficients indicate worse PROMIS global physical health for the respective adverse health outcomes. ^aThe $\widehat{\mathrm{\beta }}$ coefficients for the continuous variables, EORTC-CIPN20 sensory score and sedentary time, are reported for 1 standard deviation unit. CI = confidence interval; EORTC-CIPN20 = European Organization for Research and Treatment of Cancer Chemotherapy-Induced Peripheral Neuropathy survey.

Figure 2.

Multivariable associations of adverse health outcomes with Patient-Reported Outcomes Measurement Information System (PROMIS) global mental health T-score. Forest plot shows the $\widehat{\mathrm{\beta }}$ coefficients (center points) and 95% confidence intervals (horizontal lines) of the backward selection linear regression model for PROMIS global mental health (refer to Methods). Age, time since chemotherapy, cumulative cisplatin dose, and college education were required to be in the model, with each statistically nonsignificant at α of 0.05. $\widehat{\mathrm{\beta }}$ coefficients represent the PROMIS T-score mean differences between testicular cancer survivors with vs without the adverse health outcomes for dichotomous variables. Because 2- to 3-point PROMIS T-score differences are clinically meaningful, a $\widehat{\mathrm{\beta }}$ of more than 2 is noteworthy for adverse health outcomes. Negative $\widehat{\mathrm{\beta }}$ coefficients indicate worse PROMIS global mental health for the respective adverse health outcomes. CI = confidence interval.

Pain and functional impairment

We also examined which adverse health outcomes and modifiable risk factors were associated with pain-related functional impairment; Supplementary Table 3 (available online) presents adjusted bivariable associations, with multiple adverse health outcomes statistically significant. Degree of functional impairment was highly correlated with PROMIS-quantified pain intensity (r = 0.825; P < .0001). In a multivariable model assessing pain-related functional impairment, concomitant peripheral artery disease ($\widehat{\mathrm{\beta }}$ = 4.68), EORTC-CIPN20 sensory score ($\widehat{\mathrm{\beta }}$ = 1.90), and depression ($\widehat{\mathrm{\beta }}$ = 0.99) were statistically significantly associated with worsening impairment (Table 3 and Figure 3). Patients with their most severe pain in lower extremities had statistically significantly worse impairment than those with upper extremity locations ($\widehat{\mathrm{\beta }}$ = 2.15). Higher education was associated with less functional impairment ($\widehat{\mathrm{\beta }}$ = −1.59).

Table 3.

Multivariable associations of adverse health outcomes with patient-reported functional impairment due to pain^a

Functional impairment due to pain (PROMIS Pain Interference T-score)	$\widehat{\beta }$ b (95% CI)	SE	P
EORTC-CIPN20 sensory scorec	1.90 (1.39 to 2.42)	0.26	<.0001
Location of most severe paind
Fingers and/or hands, both left and right	Referent		Referent
Toes and/or feet, both left and right	2.15 (0.35 to 3.94)	0.91	.019
Peripheral artery disease	4.68 (2.40 to 6.95)	1.16	<.0001
Depression	0.99 (0.11 to 1.87)	0.45	.028
College education	−1.59 (-2.55 to -0.63)	0.49	.001

^aA backward selection linear regression model was used with age at survey, time since chemotherapy, cumulative cisplatin dose, and college education required to be in the model. Only college education was statistically significant (α = 0.05). $\widehat{\beta }$ coefficients and P values for the nonsignificant required variables (not shown in the table) are as follows: age at survey, per 5 years ($\widehat{\beta }$ = -0.10, 95% CI = -0.30 to 0.11; P = .347); time since chemotherapy, per 5 years ($\widehat{\beta }$ = -0.03, 95% CI = -0.35 to 0.30; P = 0.872); and cumulative cisplatin dose, 100 mg/m² ($\widehat{\beta }$ = -0.14, 95% CI = -0.56 to 0.28; P = .520). Initial variables for this model were the EORTC-CIPN20 sensory score, hypertension, diabetes, obesity, Raynaud phenomenon, anxiety, depression, peripheral artery disease, location of most severe pain, vigorous physical activity, married or living as married, smoking pack-year categories, and sedentary time. CI = confidence interval; EORTC-CIPN20 = European Organization for Research and Treatment of Cancer chemotherapy-induced peripheral neuropathy survey (42); PROMIS = Patient-reported Outcomes Measurement Information System.

^b $\widehat{\beta }$ coefficients represent the PROMIS T-score mean differences between testicular cancer survivors with vs without the adverse health outcomes for dichotomous variables and per 1 standard deviation increase in the standardized (SD = 1) continuous variable, EORTC-CIPN20 sensory score. Because 2- to 3-point PROMIS T-score differences are clinically meaningful, a $\widehat{\beta }$ of more than 2 is noteworthy for adverse health outcomes. Positive $\widehat{\beta }$ coefficients indicate more functional impairment due to pain for the respective adverse health outcomes.

^cThe $\widehat{\beta }$ coefficient for the continuous variable, EORTC-CIPN20 sensory score, is reported for 1 standard deviation unit.

^dFor location of most severe pain, additional categories used in the modeling of “other” and “no pain” for those patients who did not report chemotherapy-induced neuropathic pain are not shown.

Figure 3.

Multivariable associations of adverse health outcomes with patient-reported functional impairment due to pain. Forest plot showing the $\widehat{\mathrm{\beta }}$ coefficients (center points) and 95% confidence intervals (horizontal lines) of the backward selection linear regression model for patient-reported functional impairment due to pain, Patient-Reported Outcomes Measurement Information System (PROMIS) Pain Interference (refer to Methods). Age, time since chemotherapy, cumulative cisplatin dose, and college education were required to be in the model, with only college education statistically significant at α of 0.05. $\widehat{\mathrm{\beta }}$ coefficients represent the PROMIS T-score mean differences between testicular cancer survivors with vs without the adverse health outcomes for dichotomous variables and per 1 standard deviation increase in the standardized (SD = 1) continuous variable, EORTC-CIPN20 sensory score. Because 2- to 3-point PROMIS T-score differences are clinically meaningful, a $\widehat{\mathrm{\beta }}$ of more than 2 is noteworthy for adverse health outcomes. Positive $\widehat{\mathrm{\beta }}$ coefficients indicate more functional impairment due to pain for the respective adverse health outcomes. ^aThe $\widehat{\mathrm{\beta }}$ coefficient for the continuous variable, EORTC-CIPN20 sensory score, is reported for 1 standard deviation unit. CI = confidence interval; EORTC-CIPN20 = European Organization for Research and Treatment of Cancer Chemotherapy-Induced Peripheral Neuropathy survey.

Discussion

To our knowledge, this is the first comprehensive study of global physical and mental health in long-term US testicular cancer survivors and the only testicular cancer survivors investigation to examine the impact of a wide range of adverse health outcomes and health behaviors on these endpoints, thus, addressing critical research gaps. Increasing adverse health outcome numbers were associated with decreases in physical and mental health (P < .0001 each). Importantly, chemotherapy-induced neuropathic pain affected 1 in 5 long-term testicular cancer survivors and was associated with statistically significantly worse mental health. Risk factors for statistically significant pain-related functional impairment included lower extremity location, concomitant peripheral artery disease, and depression. Diabetes, obesity, and fatigue were also associated with statistically significantly worse mental health. Adverse health outcomes independently associated with a statistically significant negative impact on physical health included diabetes, problems with balance, equilibrium, or ability to reach, and others; modifiable risk factors included sedentary time, obesity, and hypercholesterolemia.

Chemotherapy-induced neuropathic pain, affecting 1 in 5 long-term testicular cancer survivors, was a strong predictor of global mental health. In general, pain is a driving component of physical health and has been associated with mental health in cancer survivors and the general population (21,22,56,57). Although prior investigations also show associations between pain and quality-of-life measures in survivors of other cancer types (21,22,56-58), we are the first, to our knowledge, to evaluate this association in testicular cancer survivors, an important population of young adults. Brydoy et al. (26) assessed neuropathy in testicular cancer survivors with only the 2-item Scale for Chemotherapy-Induced Long-Term Neuropathy and did not evaluate quality of life or consider pain, whereas Lauritsen et al. (27) did not mention pain at all. Weaknesses of the 2-item Scale for Chemotherapy-Induced Long-Term Neuropathy as a sole instrument to assess chemotherapy-induced peripheral neuropathy itself were reviewed by Cavaletti et al. (59). Although other studies of testicular cancer survivors evaluated neuropathy (without distinguishing pain) (26-30) and/or quality of life (27-34) (Supplementary Table 4, available online), none have examined both chemotherapy-induced peripheral neuropathy–related pain and quality of life (25).

It is important to investigate the functional impact of chemotherapy-induced neuropathic pain, and here greater impairment was independently associated with lower extremity location, concomitant peripheral artery disease, and depression. Worse pain in lower vs upper extremities has been reported for cisplatin and oxaliplatin (25,60-63); however, these were measures of pain intensity and not pain interference as evaluated here. The larger detrimental influence of lower extremity pain on functional status suggests it is more impactful on daily activities, perhaps by limiting mobility. In testicular cancer survivors reporting problems with balance, equilibrium, or ability to reach, severity of difficulty walking was also correlated with pain-related functional impairment (Spearman ρ = 0.67; P = .0003). It was unsurprising that the EORTC-CIPN20 sensory score was associated with functional impairment, as neuropathic pain is one possible component of neuropathy; chemotherapy-induced peripheral neuropathy in general also appears to be more frequent in lower extremities following platinum and/or taxanes (64).

Pain and depression are frequently diagnosed together in cancer patients (65-69). The nature of the relationship remains under investigation (70-73), but multiple studies show that they have reciprocal adverse effects (69,74). Peripheral artery disease as defined here referred to intermittent claudication or arterial surgery in pelvis or legs (Supplementary Appendix A1, available online), signifying arteriosclerotic conditions that could contribute to pain-related functional impairment (75). These may also contribute by limiting mobility. Greater education was associated with less pain-related functional impairment, likely reflecting better health-care access (76) and pain management resources.

Fatigue has been strongly associated with mental health, including depression and anxiety (77), and is a core somatic symptom in depressive disorders (78). Moreover, some treatments may be beneficial for fatigue and depression and anxiety (79,80). Marriage or living as married was associated with better global mental health, highlighting the importance of social support, as shown in European testicular cancer survivors (31,32). Obesity was associated with worse global mental health as in the general population (81,82).

Diabetes is associated with worse quality of life in the general population (83-86), with its negative impact shown here for the first time in testicular cancer survivors. Importantly, diabetes was deleteriously related to global physical and mental health, as was obesity, likely reflecting its chronic nature, requiring lifelong medications and/or monitoring. Diabetic complications can also adversely impact quality of life (83,86) and include angiopathy, retinopathy, neuropathy, cardiovascular and/or renal disease (87), and hearing loss (88). Diabetes and obesity are metabolic syndrome components (89), with obesity modifiable.

Greater sedentary time (measured apart from physical activity level) was associated with poor physical health. In the general population, it is increasingly recognized that this modifiable behavior is related to worse physical health (90), as is obesity (82,91,92). Sedentary time, physical activity, and obesity are interrelated, with all amenable to intervention.

Cognitive dysfunction is linked with different dimensions of quality of life, such as social role(s) and planning of physical activities and daily routines. However, the association with overall quality of life remains inconclusive (93,94). Among patients with solid tumors and chemotherapy-related cognitive impairment (94), the degree or type of cognitive dysfunction was only sporadically associated with quality of life, although we found a statistically significant relationship with global physical health.

Problems with balance, equilibrium, or ability to reach were also associated with negative global physical health, with survey questions querying ability to walk, engage in daily routines, and reach for or manipulate objects (Supplementary Appendix A1, available online). Given its association with EORTC-CIPN20 sensory score (P = .003), it is likely another indicator of the adverse functional consequences of neuropathy.

Prior studies show that cisplatin-treated testicular cancer survivors have increased prevalence of hyperlipidemia (95,96) but have not addressed relationships between hypercholesterolemia and global physical health. Individuals with homozygous familial hypercholesterolemia demonstrate poorer physical functioning than normative population controls (97) but not decreased mental health. Other studies have also not found associations between types of hypercholesterolemia and mental health (98,99).

Overall, testicular cancer survivors reported statistically significantly lower global mental health, but similar global physical health, compared with US men. It is important to note, however, that for global physical health, as adverse health outcome numbers increased, scores decreased statistically significantly (P < .0001). Adverse health outcomes associated with these decreases (Table 2) are discussed above. The lower global mental health approached the 2- to 3-point threshold considered the minimum clinically important difference on PROMIS scales (55). It should be noted that PROMIS validation studies were done in large, representative US population samples from which individuals with health conditions were not excluded (14).

In general, investigations describing the overall mental and physical health of long-term testicular cancer survivors are conflicting and largely based in Europe and include limited numbers by treatment (Supplementary Table 3, available online). Moreover, most European studies examined only a few factors (29-32,100) or combined them into single indicators (28). To our knowledge, this is the first study among testicular cancer survivors to examine in detail multiple factors associated with global physical and mental health while accounting for health behaviors and treatment. It is also the most comprehensive investigation of US testicular cancer survivors. Previously the largest study addressing quality of life in US testicular cancer survivors was among active duty service members (33); it did not consider health behaviors or adverse health outcomes but found statistically significantly lower physical and mental component scores in 75 chemotherapy-treated testicular cancer survivors vs noncancer military population controls (36-item Short Form Health Survey SF-36) (33).

European studies using validated surveys and reporting quality-of-life scores similar or even better for chemotherapy-treated testicular cancer survivors vs normative populations (28-32,101) might be explained in part by response shift (28-32,101), which may also be operational here to some extent. In reviewing European results, sociocultural differences must also be considered, as each nation has different health-care systems and social support structures that likely impact health-related quality of life (102).

As 10-year testicular cancer survival rates now exceed 95% through cisplatin-based treatment successes (2), identification of factors impacting health-related quality of life in long-term survivors assumes heightened importance. We provide the first report of PROMIS-quantified global physical and mental health and patient-reported functional impairment due to pain in a clinical cohort of long-term US testicular cancer survivors. Study strengths include homogenous cisplatin-based treatment, comprehensive assessment of risk factors, health behaviors, adverse health outcome, and use of the National Institutes of Health’s rigorous PROMIS surveys with comparisons to US norms. Using reliable instruments, we quantified the negative impact of adverse health outcomes in US testicular cancer survivors, which may facilitate targeted interventions for those adverse health outcomes that are most impactful and modifiable (eg, pain, diabetes, depression, obesity). It may be that behavioral interventions and medical treatments early in survivorship could have preventive or mitigating impacts for various adverse health outcomes.

In viewing our results, it should be recognized that adverse health outcome estimates represent the long-term effects of cisplatin-based chemotherapy as administered at major cancer centers and with a large percentage of college-educated patients; thus, our results may not be generalizable to some community-based settings. An inherent limitation of all cross-sectional studies is the inability to assess causality of adverse health outcomes, demographics, and health behaviors with various endpoints or to pain-related functional impairment. Symptom measurement over time would better enable the identification of trajectory classes and key predictive factors. Adverse health outcomes were self-reported without pretesticular cancer baseline data, similar to several other testicular cancer survivors studies (28,33,34). Given the young median age (31 years) at testicular cancer diagnosis, it is unlikely, but still possible, that some conditions were preexisting. Although additional pain questions specifically queried chemotherapy-induced neuropathic pain (Supplementary Appendix A3, available online), small numbers of patients (4.5%, n = 16) indicated chemotherapy-induced neuropathic pain and Raynaud phenomena but did not respond affirmatively to EORTC-CIPN20 pain questions. Thus, in these cases, there is a possibility that pain could have represented a reported manifestation of Raynaud phenomena (103). However, in separate multivariable sensitivity analyses to examine this, Raynaud phenomena was not associated with global physical health (P = .72), global mental health (P = .79), or pain-related functional impairment (P = .52). In the general population, prevalence estimates for chronic pain with neuropathic characteristics vary widely (3.35%−17.9%) because of numerous conditions included in this term, lack of consensus definitions, and heterogeneous study-dependent methodologies (104). In contrast, almost 1 in 5 of our testicular cancer survivors reported a relatively specific diagnosis of chemotherapy-induced neuropathic pain, which occurs only in cancer survivors and with onset after chemotherapy. Chemotherapy-induced peripheral neuropathy was assessed via self-report as recommended by a National Cancer Institute Clinical Trials planning meeting to prevent and treat chemotherapy-induced peripheral neuropathy (52) and in an American Society of Clinical Oncology Clinical Practice Guideline (105).

In conclusion, because sedentary time and obesity impact physical health, testicular cancer survivors should be encouraged to lead healthy, active lifestyles. Chemotherapy-induced neuropathic pain should be addressed, given its deleterious impact on mental health, even 10 or more years after treatment. No agents are currently approved to prevent and/or treat chemotherapy-induced peripheral neuropathy, and only a moderate recommendation exists for duloxetine for related pain (105,106). Testicular cancer survivors with the most severe pain will require long-term management strategies, with further research needed on optimal approaches (106,107). Meanwhile, potential management strategies include medications proven beneficial for nonchemotherapy-induced peripheral neuropathy neuropathic pain as well as integrative and other nonpharmacological pain treatments, stepped-care approaches tested for varying types of cancer pain, and referral to pain specialists (106-111).

Data availability

De-identified data that support the findings of this study are available upon request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Author contributions

Paul Dinh, Jr, PhD (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Software; Supervision; Validation; Visualization; Writing—original draft; Writing—review & editing), Patrick O. Monahan, PhD (Conceptualization; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Supervision; Validation; Writing—original draft; Writing—review & editing), Sophie D. Fosså, MD (Formal analysis; Writing—review & editing), Howard D. Sesso, ScD (Conceptualization; Funding acquisition; Methodology; Writing—review & editing), Darren R. Feldman, MD (Data curation; Investigation; Project administration; Supervision; Writing—review & editing), M. Eileen Dolan, PhD (Funding acquisition; Writing—review & editing), Kathryn Nevel, MD (Conceptualization; Data curation; Investigation; Methodology; Project administration; Validation; Writing—review & editing), John Kincaid, MD (Conceptualization; Methodology; Resources; Writing—review & editing), David J. Vaughn, MD (Investigation; Methodology; Project administration; Writing—review & editing), Neil E. Martin, MD (Investigation; Methodology; Project administration; Writing—review & editing), Victoria A. Sanchez, AuD, PhD (Data curation; Investigation; Methodology; Writing—review & editing), Lawrence H. Einhorn, MD (Conceptualization; Data curation; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Writing—review & editing), Robert Frisina, PhD (Conceptualization; Funding acquisition; Methodology; Supervision; Writing—review & editing), Chunkit Fung, MD (Conceptualization; Data curation; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Writing—review & editing), Kurt Kroenke, MD (Conceptualization; Formal analysis; Investigation; Methodology; Validation; Writing—original draft; Writing—review & editing), and Lois B. Travis, MD, ScD (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Writing—original draft; Writing—review & editing).

Funding

Drs Frisina, Dolan, Sesso, Feldman, Monahan, and Travis were supported by 2 R01 CA157823 funded by the National Cancer Institute. Dr Sanchez was supported by R01 DC019408-01 funded by the National Institute on Deafness and Other Communication Disorders.

Conflicts of interest

Darren R. Feldman has consulting or advisory roles with Telix Pharmaceuticals, BioNTech, and Renibus. He receives research funding from Telix Pharmaceuticals and Decibel Therapeutics. He receives research funding (Institutional PI) from Astellas Pharma and Exelixis. He also receives royalties for authorship of topic review from UpToDate. Robert D. Frisina was awarded patents (to his institution) for new neural stimulation technique involving laser light and gold nanoparticles, for a new drug combination (hormone, anti-inflammatory) to treat age-related hearing loss, and for self-inflating flotation devices to prevent drowning in children and adults. He was also awarded a U.S. patent on Peristaltic Micropumps and Fluid Delivery Devices that incorporate them. Chunkit Fung discloses his Open Payments Link: https://openpaymentsdata.cms.gov/physician/450635. John Kincaid is a sub-investigator for Ionis Pharmaceuticals in a treatment trial of their compound, which treats familial amyloid neuropathy, and was compensated for performing the clinical exams. Kurt Kroenke discloses research funding. Kathryn Nevel received personal fees from Aptitude Health, LLC, for answering survey on glioma management. Victoria A. Sanchez has consulting or advisory roles with Autifony Therapeutics, Boehringer Ingelheim, and Spiral. She receives research funding (Institutional) from Otonomy Inc., Frequency Therapeutics, Pipeline Therapeutics, Aerin Medical, Oticon Medical, and Helen of Troy Ltd. She receives honoraria from Oticon Medical, Sonova Holding AG, and Phonak USA. She receives non-financial support/donations from Sonova Holding AG and Phonak USA of hearing technology devices donated for educational or research purposes. David J. Vaughn discloses research funding (Institutional) from the NIH. No other potential conflicts of interest are reported.