Chronic fatigue in 812 testicular cancer survivors during longterm follow up: increasing prevalence and risk factors

Transcription

1 Annals of Oncology Advance Access published August 11, Chronic fatigue in 812 testicular cancer survivors during longterm follow up: increasing prevalence and risk factors M. Sprauten 1, H. S. Haugnes 2,3, M. Brydøy 4, C. Kiserud 1, T. Tandstad 5, T. Bjøro 6,7, J. Bjerner 8, M. Cvancarova 1, S. D. Fosså 1, J. Oldenburg 9 1 National Advisory Unit on Late Effects after Cancer Treatment, Oslo University Hospital, Oslo, Norway 2 Institute of Clinical Medicine, Oncology, University of Tromsø, Tromsø, Norway 3 Department of Oncology, University Hospital of North Norway, Tromsø, Norway 4 Department of Oncology, Haukeland University Hospital, Bergen, Norway 5 The Cancer Clinic, St Olavs University Hospital, Trondheim, Norway 6 Department of Medical Biochemistry, Oslo University Hospital, Norway 7 Institute of Clinical Medicine, Faculty of Medicine, University of Oslo, Norway 8 Fürst Medical Laboratory, Oslo, Norway 9 Department of Oncology, Oslo University Hospital, Oslo, Norway Corresponding author: Dr. Jan Oldenburg, Department of Oncology, Akershus University Hospital, Lørenskog, Norway, & University of Oslo, Norway, Telephone: , E- mail: jan.oldenburg@medisin.uio.no Key Message: "This longitudinal study reveals a substantial increase in CF during 12 to 19 years after treatment, with higher levels of anxiety, depression, and neurotoxicity, and lower levels of testosterone, being associated with CF at the second assessment. Moderate and high physical activity indicated a protective effect. These findings might indicate means to prevent or treat CF. Health care professionals should have early prevention of CF through life style interventions and early detection, and treatment and follow-up of comorbid conditions in mind when caring for TCSs and probably other cancer survivors as well." The Author Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please journals.permissions@oup.com.

2 2 Abstract Background: Chronic fatigue (CF) has been reported to be slightly more prevalent in Testicular Cancer Survivors (TCSs) than in the general population. In this study, we wanted to explore possible determinants of CF in TCSs median 12 (Survey I) and 19 years (Survey II) after treatment, in particular the relation to late effects after treatment. Patients & Methods: Overall, 812 TCSs treated between provided blood samples (Testosterone and Luteinizing Hormone) and completed questionnaires at Survey I ( ), and Survey II ( ). Hormone levels were categorized according to quartile thresholds for decadal age groups of controls. Associations between CF and possible risk factors including the Hospital Anxiety and Depression Scale (HADS), treatment, physical activity, hormone levels, neurotoxicity, and comorbidity, were analyzed by logistic regression. Results: Prevalence of CF increased from 15% at Survey I to 27% at Survey II (p<0.001). At Survey II, risk for CF was increased 3-4-fold for high levels of neuropathy compared with no neuropathy, and 2-3-fold for high levels of Raynaud like phenomena and having testosterone levels in the lowest quartile, while being moderately and highly physically active had a protective effect. Risk for CF in TCSs with higher levels of HADS-Anxiety and HADS- Depression was increased 2-5-fold, respectively. Conclusion: The increasing prevalence of CF in TCSs is a novel finding. Life style interventions, early detection and treatment of depression and anxiety and possibly testosterone substitution might reduce the risk of CF. Extended long-term follow-up seems important. Keywords: Fatigue, survivorship, testicular cancer, physical activity, testosterone, hypogonadism

3 3 Introduction Chronic Fatigue (CF) has been described as one of the most common and distressing adverse effects of cancer and its treatment.[1] Research on the development of CF, possible riskfactors and long-term prevalence in cancer survivors represents an area of particular clinical interest. Population-based studies and longitudinal assessments might be prerequisite to answer some of these conundrums, however, such studies are exceedingly rare. Testicular cancer (TC) is regarded a model for a curable neoplasm with 5-year relative survival rates exceeding 97% in Norway.[2] Thereby the increasing cohort of young TC survivors (TCSs) is regarded as optimal for cancer survivorship research, including late effects like cardiovascular disease (CVD), secondary cancer, neurotoxicity, fatigue and hypogonadism.[3,4] Premature hormonal ageing might represent a particular threat for the well-being of cancer survivors and this long-term complication in TCSs was recently reported by our group.[5] Primary endocrine hypogonadism, i.e. a decreased testosterone production, accompanied by increased levels of Luteinizing Hormone (LH), may result in reduced energy levels, diminished libido, erectile dysfunction, depressed mood, reduced muscle mass and strength, anaemia, osteoporosis, CVD and metabolic syndrome. [6-12] CF, defined as fatigue above a certain level with duration 6 months, is more common in TCSs (16%) than in the general male population (10%).[13-15] The prevalence of CF increases with age in the general male population (from 9.6% to 12.2% for the decadal age groups and 50-59, respectively).[15] Currently, little is known about the longitudinal development of CF in cancer survivors and a relevant question is whether the observed increase of CF in the general population will be more or possibly less pronounced in TCSs. Anxiety, depression and comorbidity have been shown to be associated with CF, whereas other factors associated with fatigue have rarely been assessed in TCSs, and to our knowledge not with follow-up data up to 20 years after treatment.[14] The number of cancer survivors referred due to CF to the Norwegian Advisory Unit on Late Effects after Cancer Treatment has been increasing and includes TCSs with debut of CF several years after treatment. Intrigued by this observation, we wanted to assess the prevalence of CF and its possible determinants in TCSs after longer time of observation than previous reports,[14] and in particular assess CF in relation to the comorbidity and toxicities commonly observed in TCSs like CVD, anxiety, depression, hypogonadism, and neurotoxicity.[4]

4 4 Aims of this explorative study were: 1) to describe the prevalence of CF in a population-based cohort of TCSs at two late time points after treatment, 2) to assess the relationship between CF and possible risk factors including serum levels of testosterone and LH, treatment type, long-term toxicities, depression, anxiety, and comorbidities. Methods Design and Study Population This longitudinal study originates from two national population-based surveys conducted during (SI) and (SII), assessing somatic and psychosocial health in long-term TCSs. All men aged years treated for unilateral germ cell TC in Norway from 1980 through 1994 were invited (n=1,814, SI). Exclusion criteria were: Extragonadal germ cell cancer or secondary malignancies except skin cancer, bilateral orchiectomy, or mental retardation. Treatment details for TCSs during the study period have been described previously.[16] A total of 1,463 males (81%) participated at SI. However, exclusion of those with missing hormone samples, questionnaire data, and other reasons, resulted in 812 men eligible at both surveys (Figure 1). At SI, median 12 years (range 5-21 years) after TC treatment, participants underwent a clinical examination. Serum for hormone analyses was generally collected before 11 am. SII was performed median 19 years (range years) after treatment, and blood samples were collected before 12 noon at the primary care physician s office, and all samples were sent to central analysis. At SI and SII, the participants completed a questionnaire including the Fatigue Questionnaire (FQ), the Scale for Chemotherapy-Induced Neurotoxicity (SCIN), and questions addressing comorbidities, medication, tobacco use, and marital status.[17,18] Both surveys were approved by the Regional Committee for medical and health research ethics of the South Eastern Health Region in Norway.

5 5 Questionnaires The FQ includes 11 items (7 items covering physical fatigue, 4 items covering mental fatigue).[18,19] Each item has four response alternatives (0-3). Summation of the 11 item scores yields a total fatigue score between 0 and 33, with higher scores indicating higher levels of fatigue. Two additional questions evaluate the extent and duration of fatigue (less than a week = 0, less than three months = 1, between three and six months =2, and more than six months =3). For case definition, the 11 item-scores were dichotomized (0 and 1 = 0, and 2 and 3 = 1) and summarized, with a sum score of 4 and duration of symptoms 6 months defining CF.[13,20] The Hospital Anxiety and Depression Scale (HADS) consists of 7 items covering anxiety (HADS-A) and 7 items covering depression (HADS-D), with item-scores of 0-3. Total sum scores (0-21) are calculated by summation of 7 item-scores with higher scores indicating higher symptom levels, and with scores of 8 or above defining caseness of both HADS-A and HADS-D.[21] Comorbidity was recorded based on self-reported diabetes, cardiac disease (incident angina or heart attack) or pulmonary disease (yes/no). Levels of physical activity were categorized into three groups (inactive, moderately active, highly active) as previously published.[22] Neurotoxicity was assessed by the SCIN.[17] The TCSs responded to how bothered they were (not at all=0, a little=1, quite a bit=2, very much=3), by each of the six symptoms (neuropathy in hands/fingers and/or feet/toes, Raynaud s phenomenon in hands/fingers and/or feet/toes, tinnitus and/or reduced hearing). To reduce the number of variables in the multivariate analyses, the two questions regarding neuropathy (hands/fingers and feet/toes) were combined into one item, recording the worst score for any of these two symptoms with corresponding procedures for Raynaud s phenomenon and ototoxicity. Laboratory analyses Testosterone and LH levels were measured at both surveys by routine immunoassays. During the study period the laboratory changed the analytic method. However, for both testosterone and LH the correlations between assays were within biological variations. Cross reactivity between LH and Human Chorionic Gonadotropin was minimal.

6 6 The reference intervals reported by Bjerner et al were used for both study periods.[23] Testosterone and LH levels from TCSs were categorized into quartiles according to the cutoff values for the 25, 50, and 75 percentiles of the reference intervals for each decadal age group, thereby adjusting for age-related variations, as previously published.[5] This categorization yields more information than a dichotomization according to cut-off levels. Statistical Methods Crude associations between continuous variables were assessed with paired t-tests and with χ 2 -test for trend for ordered categorical variables. Additional univariate logistic regression analyses were performed to evaluate associations between CF and testosterone levels, LH levels, and treatment. Age was not linear in the logit, moreover, when entered as decades it was not statistically significant and therefore not included into the final model in order to save statistical power. When selecting the best final model, we used both the objective criteria, the Akaiki information criterion (AIC) and our clinical knowledge. Our modeling process focused on variables that were both of clinical interest and statistically significantly associated with the dependent variable. We have tried several models and chose to present the one with the smallest AIC.[24] We did not use any automated variable selection methods like backward or forward selection because the number of possible covariates and models was rather limited. Multivariate logistic regression analyses were performed to investigate the associations between CF at SI and SII separately, and hormone levels, type of treatment, neurotoxicity and comorbid conditions. Variables considered of high clinical relevance (e.g. treatment) and variables associated with CF (p 0.05) in the univariate analyses, were included in the multivariate analyses. All tests were two-sided, and p-values 0.05 were considered statistically significant. Our analyses were considered exploratory, so adjustment for multiple testing was not performed. All analyses were performed using SPSS version 19 0 (SPSS, Chicago, IL).

7 7 Results Patient characteristics Of the 812 TCSs included, 29% had metastatic disease at diagnosis (Table 1). In total 164 men (20%) had been treated with surgery alone, while 343 men (42%) received radiotherapy and 305 (38%) received chemotherapy. In the latter group, 32 males had also received radiotherapy. Median age at SII was 50 years in the surgery and chemotherapy group, and 55 years in the radiotherapy group. Testosterone substitution was initiated in 19 men between the surveys, due to contralateral testicular cancer in seven. These 19 men were excluded from all analyses at SII. Participants and non-participants did not differ with regard to age, stage and treatment type. Prevalence of CF and comparisons between TCSs with CF at SII only, or at both surveys At SI, 126 males (15%) had CF, whereas at SII, 215 TCSs (27%) had CF. Two thirds (n=85) of the 126 TCSs with CF at SI were also classified with CF at SII, while there were 130 new cases of CF at SII. Univariate analyses comparing those with CF at SII only, and those with CF at both surveys for all variables did not reveal any significant differences except for a significantly higher mean score at SII for total fatigue (21 0, p=0 005) and HADS-A (7 4, p=0 012), for TCSs having CF at both surveys. Variables associated with CF, univariate analyses At both surveys, CF was statistically significantly associated with increased levels of anxiety (HADS-A), depression (HADS-D), higher levels of neuropathy, Raynaud s phenomenon and tinnitus and/or reduced hearing (all p-values <0.001) (Table 2). At SII, a low educational level, ever smoking, physical inactivity, diabetes, cardiac disease and pulmonary disease, were also significantly associated with CF. In total, 21%, 30%, and 28% of those treated by surgery, radiotherapy, or chemotherapy, respectively, reported CF at SII. Logistic regression analyses assessing associations between treatment groups or age-adjusted

8 8 hormone levels with CF revealed that radiotherapy, but not chemotherapy, was associated with an increased risk of CF at SII when compared to surgery only (p=0.028 and p=0.106, respectively). Furthermore, having testosterone in the lowest versus the highest quartile was associated with increased risk of CF (p=0.023), while no significant associations emerged for LH levels, (data not shown), Age was not significantly associated with CF at any of the surveys (Table 2). There were no significant associations between CF and hormone levels at SI. Multivariate analyses At SI, TCSs with higher levels of anxiety and depression had a roughly 3-fold increased risk of CF compared to those with lower levels of anxiety and depression, while those with the highest level of neuropathy, and the second highest level of Tinnitus/decreased hearing had roughly 2-3-fold increased risks of CF compared to those with none of these symptoms (Table 3). No other significant associations emerged at SI. At SII, TCSs being moderately or highly physically active had a protective effect against CF compared with those being inactive (reference), with ORs of 0.47 and 0.31, (95%CI; and , respectively). TCSs with testosterone in the lowest and second highest quartile had significantly elevated risks of CF when compared to TCSs with testosterone in the highest quartile (ORs of 3.2 and 2.6, respectively), testosterone levels in the second quartile, had a similar OR of 2.2 but did not reach statistical significance (Table 3). There was a stepwise increase in the risk of CF with increasing levels of neuropathy with OR of 3.4 (95%CI; ) for the highest level compared with no neuropathy. The highest level of Raynaud s phenomenon was significantly associated with a higher risk of CF with an OR of 2.4. No significant associations emerged for CF and tinnitus/reduced hearing. Prevalent cardiac disease displayed borderline significant risk for CF with OR of 1.9 (95%CI; ), compared with no cardiac disease (p-value 0.08). CF was not significantly associated with level of education, smoking, diabetes, pulmonary disease, age, or treatment. Sensitivity analyses when excluding the 32 males who received both radiotherapy and chemotherapy did not change our results, indicating robust associations and no major impact from the combination of radiotherapy and chemotherapy on CF.

9 9 Discussion On the background of a 10% prevalence of CF in the Norwegian general male population,[15] the increasing prevalence of CF in TCSs from 15% after 12 years to 27% 19 years after treatment, is remarkable and might have ramifications for long-term cancer survivors in general. Two thirds of those with CF at SI also had CF at SII. Those with CF at both surveys had significantly higher total fatigue scores and anxiety levels than those only having CF at SII. In univariate analyses at SI, only few of the examined covariates were associated with CF (all three symptoms assessed by the SCIN, the score of HADS-A and HADS-D). At SII, however, most of the 17 variables examined were significantly associated with CF. Also in the multivariate analysis an increase of variables associated with CF from SI to SII became apparent. Results from SI should, however, be interpreted with caution due to the low number of disease events at this time point. Higher levels of anxiety, depression, and neurotoxicity (neuropathy and Raynaud s phenomenon) and lower quartiles of testosterone were all significantly associated with CF at SII, while moderate and high physical activity had a protective effect. The increase in factors associated with CF accompanying an increase in the prevalence of fatigue might represent a shared development: reduced stress coping. Presumably, TCSs being fatigued will be less able to filter out the stress from e.g. neuropathy in fingers or toes. On the other hand, chronic neuropathy might tire out cancer survivors, finally resulting in fatigue. Apparently the development of CF is a slow process in the relatively young TCSs, as it took up to 2 decades until about one of four TCSs reported CF. Prior radiotherapy was significantly associated with CF at SII in univariate- (logistic regression), but not in the multivariate analysis. Possibly, late-effects known to be associated with radiotherapy (e.g. hypogonadism, diabetes, CVD, neuropathy), and ageing competed with radiotherapy itself as risk factors for CF.[4] This may to some extent have contributed to the different proportions of CF after radiotherapy and surgery of 30% versus 21%, respectively. Lack of significant associations between CF and type of treatment, but with treatment-related toxicities and self-reported comorbidity is in accordance with the majority of studies on CF in cancer survivors.[25-27] Fewer TC patients are currently treated with RT than during the study period, and modern radiation techniques, reduced volumes and target doses will probably render radiotherapy-induced side-effects a less prominent problem among future TCSs.

10 10 The risk of CF was significantly decreased for moderately and highly active men when compared with those reporting being inactive at SII. Whether CF results in inactivity, and/or inactivity contributes to the development of fatigue, is still open for discussion. However, fatigue after cancer treatment may be reduced by initiating or maintaining adequate levels of physical activity according to meta-analyses, systematic reviews, and randomized trials.[28] Contributing factors (e.g. medical and substance-induced) should be assessed and treated in cancer survivors with moderate to severe fatigue, before intervening through physical activity programs. Low testosterone levels were significantly associated with CF at SII. Importantly, the majority of TCSs (428 of the 793) had testosterone levels in the lowest quartile, rendering this the strength of this association more important than the borderline association between CF and more normal testosterone levels. This finding contrasts with Huddart et al, who found no significant relationship between testosterone and fatigue in TCSs, possibly due to a shorter follow-up time. Different categorizations of testosterone in the two studies, (levels below 10 nmol/l versus ordinal categorization according to quartile thresholds for decadal age groups), might also have contributed to the varying results. Importantly, our observations do not imply testosterone substitution as a means to treat chronic fatigued in general, but rather that high testosterone levels are associated with the absence of CF. Nevertheless, as testosterone replacement in a study of severely hypogonadal males yielded increased energy levels, such treatment may be considered in fatigued TCSs with consistently low testosterone levels.[29] Strengths of our study comprise the large population-based and well-characterized cohort of TCSs, and the longitudinal study design. Blood samples were collected at relatively consistent diurnal time points, with available hormone levels and FQ for 812 men. Detailed information regarding previous cancer treatment was available through collaboration between the five University hospitals in Norway, with all TCSs treated according to international guidelines. High participation rates of roughly 80% at both surveys, implies validity of our findings and renders selection bias less likely, although the rigorous selection including only those with complete blood samples and questionnaires at both surveys reduced the number of eligible men to 812. Hormone levels represent the only objective measure in this study, and reliance on subjective factors might be considered a limitation. However, CF is not objectively measurable, and self-reported comorbidity might actually better represent the impact on CF than physiological assessments of cardiac or pulmonary function.[17,30] The

11 11 lack of assessment of CF before treatment poses a limitation as we cannot assess a potential impact of TC treatment on CF before SI. In conclusion, this longitudinal study reveals a substantial increase in CF during 12 to 19 years after treatment, with higher levels of anxiety, depression, and neurotoxicity, and lower levels of testosterone, being associated with CF at the second assessment. Moderate and high physical activity indicated a protective effect. These findings might indicate means to prevent or treat CF. Health care professionals should have early prevention of CF through life style interventions and early detection, and treatment and follow-up of comorbid conditions in mind when caring for TCSs and probably other cancer survivors as well. Testosterone substitution may be considered in order to ameliorate or treat CF in hypogonadal males after repeated testosterone evaluations. CF may dramatically impair quality of life, and its disturbing increase in TCSs and the association with partly treatable side effects, underlines the importance of continued long-term assessments of cancer survivors. Funding: The Norwegian Cancer Society (salary HSH) and the South-Eastern Norway Regional Health Authority (Grant No , MS). Acknowledgement; Professor Jon Håvard Loge for valuable comments. His generous contribution of time and expertise were greatly appreciated. The authors declare to have no conflict of interest.

12 12 Reference List 1. Bower JE. Cancer-related fatigue-mechanisms, risk factors, and treatments. Nat Rev Clin Oncol 2014; 11: Cancer in Norway Cancer registry of Norway. 3. Oldenburg J, Fosså SD, Nuver J et al. Testicular seminoma and non-seminoma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2013; 24: vi125-vi Haugnes HS, Bosl GJ, Boer H et al. Long-Term and Late Effects of Germ Cell Testicular Cancer Treatment and Implications for Follow-Up. J Clin Oncol 2012; 30: Sprauten M, Brydøy M, Haugnes HS et al. Longitudinal Serum Testosterone, Luteinizing Hormone, and Follicle-Stimulating Hormone Levels in a Population-Based Sample of Long- Term Testicular Cancer Survivors. J Clin Oncol 2014; 32: Huddart RA, Norman A, Moynihan C et al. Fertility, gonadal and sexual function in survivors of testicular cancer. Br J Cancer 2005; 93: Dahl AA, Haaland CF, Mykletun A et al. Study of Anxiety Disorder and Depression in Long- Term Survivors of Testicular Cancer. J Clin Oncol 2005; 23: Eberhard J, Ståhl O, Cohn-Cedermark G et al. Sexual Function in Men Treated for Testicular Cancer. The Journal of Sexual Medicine 2009; 6: Dahl AA, Bremnes R, Dahl O et al. Is the Sexual Function Compromised in Long-Term Testicular Cancer Survivors? European Urology 2007; 52: Haugnes HS, Wethal T, Aass N et al. Cardiovascular Risk Factors and Morbidity in Long-Term Survivors of Testicular Cancer: A 20-Year Follow-Up Study. J Clin Oncol 2010; 28: Barrett-Connor E. Lower Endogenous Androgen Levels and Dyslipidemia in Men with Non- Insulin-Dependent Diabetes Mellitus. Annals of Internal Medicine 1992; 117:

13 Greenfield DM, Walters SJ, Coleman RE et al. Quality of life, self-esteem, fatigue, and sexual function in young men after cancer. Cancer 2010; 116: Fukuda K, Straus SE, Hickie I et al. The Chronic Fatigue Syndrome: A Comprehensive Approach to Its Definition and Study. Ann Intern Med 1994; 121: Fosså SD, Dahl AA, Loge JH. Fatigue, Anxiety, and Depression in Long-Term Survivors of Testicular Cancer. J Clin Oncol 2003; 21: Loge JH, Ekeberg O, Kaasa S. Fatigue in the general Norwegian population: normative data and associations. J Psychosom Res 1998; 45: Sprauten M, Darrah TH, Peterson DR et al. Impact of Long-Term Serum Platinum Concentrations on Neuro- and Ototoxicity in Cisplatin-Treated Survivors of Testicular Cancer. J Clin Oncol 2012; 30: Oldenburg J, Fosså SD, Dahl AA. Scale for chemotherapy-induced long-term neurotoxicity (SCIN): Psychometrics, validation, and findings in a large sample of testicular cancer survivors. Quality of Life Research 2006; 15: Chalder T, Berelowitz G, Pawlikowska T et al. Development of a fatigue scale. Journal of Psychosomatic Research 1993; Minton O, Stone P. A systematic review of the scales used for the measurement of cancerrelated fatigue (CRF). Ann Oncol 2009; 20: Pawlikowska T, Chalder T, Hirsch SR et al. Population Based Study Of Fatigue And Psychological Distress. BMJ: British Medical Journal 1994; 308: Bjelland I, Dahl AA, Haug TT et al. The validity of the Hospital Anxiety and Depression Scale: An updated literature review. J Psychosom Res 2002; 52:

14 Thorsen L, Nystad W, Dahl O et al. The level of physical activity in long-term survivors of testicular cancer. Eur J Cancer 2003; 39: Bjerner J, Biernat D, Fosså SD et al. Reference intervals for serum testosterone, SHBG, LH and FSH in males from the NORIP project. Scand J Clin Lab Invest 2009; 69: Collett D. Modelling Survival Data in Medical Research, Third Edition, Taylor & Francis, Oerlemans S, Mols F, Issa DE et al. A high level of fatigue among long-term survivors of non- Hodgkin's lymphoma: results from the longitudinal population-based PROFILES registry in the south of the Netherlands. Haematologica 2013; 98: Daniels L, Oerlemans S, Krol A et al. Persisting fatigue in Hodgkin lymphoma survivors: a systematic review. Ann Hematol 2013; 92: Bower JE, Bak K, Berger A et al. Screening, Assessment, and Management of Fatigue in Adult Survivors of Cancer: An American Society of Clinical Oncology Clinical Practice Guideline Adaptation. J Clin Oncol Ganz PA, Bower JE. Cancer related fatigue: A focus on breast cancer and Hodgkin's disease survivors. Acta Oncol 2007; 46: Snyder PJ, Peachey H, Berlin JA et al. Effects of Testosterone Replacement in Hypogonadal Men. J Clin Endocr Metab 2000; 85: Hansen SW, Helweg-Larsen S, Trojaborg W. Long-term neurotoxicity in patients treated with cisplatin, vinblastine, and bleomycin for metastatic germ cell cancer. J Clin Oncol 1989; 7:

15 15 Table and figure legends Figure 1. Overview of included TCSs with available hormone samples and Fatigue Questionnaires at both surveys Table 1. Patient characteristics Table 2. Univariate associations with Chronic Fatigue Table 3. Presence of Chronic Fatigue, multivariate logistic regression

16 1,814 TCSs invited to participate in SI Non-responders: (n=351) 1,463 TCSs accepted invitation to SI (81%) 1,207 TCSs eligible for SI Excluded for current study: (n=256) Irradiated scrotum (n= 1) Hyperprolactinoma at SI (n= 1) Substitution with T before SI (n=43) Removed retained testicle before SI (n= 2) Removed other testicle after trauma (n= 1) DHEA use at SI (n= 1) Missing hormone samples or FQ (n=207) Noneligible men at SII (n=395) Deceased between SI and SII (n= 49) Emigrated or untraceable (n= 18) Non-responders: (n=125) Recurrence of TC at time for SII (n= 2) Methadon user (n= 1) Missing hormone samples or FQ (n=200) 812 had complete FQ and hormone analyses from both SI Abbreviations: and SII and were included FQ= Fatigue in the Questionnaire, SI= Survey I, SII= Survey II, T= Testosterone, study TC= Testicular Cancer

17 Patients participating in Study, N=812 Age at TC diagnosis, years, median (range) 31.9 ( ) Age at SI, years, median (range) 44.3 ( ) Age at SII, years, median (range) 51.7 ( ) Time between TC diagnosis and SI, years, median (range) 11.5 ( ) Time between TC diagnosis and SII, years, median (range) 19.0 ( ) Histology, number (%) Seminoma Nonseminoma Royal Marsden Hospital Stage, number (%) I IM II III IV Surgery, number (%) Orchiectomy only RPLND Chemotherapy, number (%) Standard dose regimens BEP CVB Others Dose intensive regimens Radiotherapy, number (%) Para aortic field Dog-leg field Others Median dose (Gy) (range) 404 (49.8) 408 (50.2) 577 (71.1) 5 (0.6) 160 (19.7) 18 (6.4) 52 (6.4) 164 (20.2) 77 (9.5) 87 (10.7) 305 (37.6) 285 (35.1) 148 (18.2) 115 (14.2) 22 (2.7) 20 (2.5) 343 (42.2) 21 (2.6) 310 (38.2) 12 (1.5) 27.0 ( ) Chemo- and Radiotherapy, number (%) 32 (3.9) Abbreviations: BEP= Chemotherapy regimen with cisplatin, etoposide, and bleomycin CVB= Chemotherapy regimen with cisplatin, vinblastine, and bleomycin SI= Survey I SII= Survey II At Survey II, 19 men commencing testosterone substitution between surveys were removed from these analyses. As the males were treated in the time period 1980 until 1994, they were classified according to the Royal Marsden Hospital classification system in their medical journals. In the Chemotherapy group, 32 men also received RT. A total of 284 men received 3 cycles of chemotherapy or more.

18 Characteristics No CF N=686 (84%) Presence of Chronic Fatigue Survey I CF N=126 (16%) p-value No CF N=578 (73%) Survey II Ⱡ CF N=215 (27%) p-value Treatment Surgery β 142 (87) 22 (13) (80) 33 (21) Radiotherapy 288 (84) 55 (16) 234 (70) 99 (30) Chemotherapy 256 (84) 49 (16) 216 (72) 83 (28) Partnered relation No 143 (82) 32 (18) (69) 49 (31) Yes 539 (85) 94 (15) 470 (74) 166 (26) Years of education >12 years 257 (87) 40 (14) (82) 48(19) < years 315 (84) 60 (16) 280 (71) 116 (29) 7-10 years 109 (81) 26 (19) 69 (64) 39 (36) Ever smoker No 283 (87) 41 (13) (76) 71 (23) Yes 403 (83) 85 (17) 333 (70) 144 (30) Level of physical activity Inactive 96 (84) 19 (17) (55) 33 (45) <0.001 Minimally active 293 (82) 64 (18) 219 (69) 97 (31) Highly active 297 (87) 43 (13) 318 (79) 85 (21) Diabetes Ῡ No 667 (84) 123 (16) (74) 189 (26) Yes 19 (86) 3 (14) 37 (59) 26 (41) Cardiac disease No 668 (85) 120 (15) (74) 189 (26) Yes 18 (75) 6 (25) 34 (57) 26 (43) Pulmonary disease # No 679 (85) 122 (15) (74) 183 (26)) Yes 7 (64) 4 (36) 49 (61) 32 (40) HADS-D 2.2 (2.5) 5.2 (4.0) < (2.3) 5.6 (3.7) <0.001 HADS-A 4.0 (3.1) 7.2 (4.3) < (3.1) 6.5 (4.2) <0.001 Testosterone level 75 percentile 67 (86) 11 (14) (84) 12 (16) >50-75 percentile 96 (91) 10 (9) 75 (71) 31(29) >25-50 percentile 177 (86) 30 (15) 140 (75) 46 (25) 25 percentile 346 (82) 75 (18) 302 (71) 126 (29) LH level 25 percentile 135 (85) 23 (15) (75) 15 (25) >25-50 percentile 112 (86) 18 (14) 87 (78) 25 (22) >50-75 percentile 149 (86) 25 (14) 109 (68) 52 (32) 75 percentile 290 (83) 60 (17) 337 (73) 123 (27) Neuropathy hands/feet No 384 (89) 47 (11) < (87) 40 (13) <0.001 A little 175 (87) 26 (13) 181 (76) 56 (24) Quite a bit 93 (75) 31 (25) 95 (58) 69 (42)

19 Very much 33 (61) 21 (39) 34 (41) 50 (60) Raynaud s hands/feet No 375 (87) 54 (13) < (84) 53 (16) <0.001 A little 140 (86) 23 (14) 159 (75) 53 (25) Quite a bit 106 (84) 20 (16) 99 (64) 56 (36) Very much 61 (9) 28 (22) 43 (45) 53 (55) Tinnitus/reduced hearing No 352 (89) 43 (11) < (80) 54 (20) <0.001 A little 201 (84) 39 (16) 201 (73) 75 (27) Quite a bit 85 (75) 28 (25) 105 (67) 52 (33) Very much 46 (75) 15 (25) 53 (61) 34 (39) Years since orchiectomy 11.8 (4.1) 12.4 (4.5) (4.0) 20.0 (4.2) Age at survey, years 45.3 (9.9) 46.6 (9.2) (9.9) 53.7 (9.9) NOTE. Chi test for trend was applied if not stated otherwise. All P < 0.05 (2-sided) were considered statistically significant and are depicted in bold. Numbers are presented with number (%) unless for years since orchiectomy and age at survey, which are presented as mean (SD). β Surgery applies for orchiectomy alone or orchiectomy combined with RPLND. Chemotherapy applies for the 305 men receiving chemotherapy, including 32 men who received both chemotherapy and radiotherapy. Ῡ Diabetes applies for those reporting diabetes and/or the use of diabetes medication. Cardiac disease applies for those reporting having angina or having had a heart attack. # Pulmonary disease at SI applies for the use of asthma medication. At SII it applies for those reporting chronic obstructive pulmonary disease, chronic bronchitis, asthma, emphysema or the use of asthma medication. Chi square, linear trend. T-test. Mean (SD). Ⱡ At Survey II, 19 men commencing testosterone substitution between surveys were removed from the analyses.

20 Explanatory variable Odds Ratio Survey I Survey II Ⱡ 95% CI p-value Odds Ratio 95% CI p-value Treatment Surgery Radiotherapy Chemotherapy Years of education >12 years years years Ever smoker No Yes Level of physical activity Inactive Moderately active Highly active <0.001 Diabetes Ῡ No Yes Cardiac disease No Yes Pulmonary disease # No Yes HADS-A < HADS-D < <0.001 Testosterone levels 75 percentile >50-75 percentile >25-50 percentile percentile Neuropathy Not at all A little Quite a bit <0.001 Very much Raynaud s phenomenon Not at all A little Quite a bit Very much Tinnitus/reduced hearing Not at all A little Quite a bit Very much Abbreviations:

21 CI=Confidence Interval Reference category. Ῡ Diabetes applies for those reporting diabetes or using medication for diabetes. Cardiac disease applies for those reporting having angina or having had a heart attack. # Pulmonary disease at SI applies for the use of asthma medication. At SII it applies for those reporting chronic obstructive pulmonary disease, chronic bronchitis, asthma, emphysema or the use of asthma medication. Ⱡ At Survey II, 19 men commencing testosterone substitution between surveys were removed from the analyses. ORs and p-values considered statistically significant are depicted in bold.