Side Effects and Cancer-Related Stress Determine Quality of Life in Long-Term Survivors of Testicular Cancer
http://www.100md.com
《临床肿瘤学》
the Research Centre for Health Promotion and Department of Oncology, Haukeland University Hospital, University of Bergen, Bergen
Department of Clinical Cancer Research, The Norwegian Radium Hospital
Departments of Psychiatry and Oncology, Ullevaal University Hospital, and Medical Faculty, University of Oslo, Oslo
Department of Oncology, University Hospital of Northern Norway, University of Troms, Troms
Department of Oncology, Saint Olav's Hospital, National University for Science and Technology, Trondheim, Norway
ABSTRACT
PURPOSE: The prevalence of long-term survivors after treatment for testicular cancer (TC) is increasing, and most studies display normal or only slightly reduced quality of life (QOL) in TC survivors (TCSs). Impaired QOL is claimed to be associated with treatment modality and its side effects, although most studies in this field can be criticized for various methodologic shortcomings. We wanted to examine variation in long-term QOL in TCSs in relation to TC treatment modality, side effects, and TC-related stress in a large population.
PATIENTS AND METHODS: QOL, side effects, and TC-related stress were self-rated by a questionnaire at a mean of 11 years of follow-up in 1,409 TCSs treated from 1980 to 1994. Norm data was obtained from 2,678 males who were representative of the general population. QOL was measured with the Short Form-36 (SF-36), and TC-related stress was measured with the Impact of Event Scale.
RESULTS: There were no clinically relevant differences in QOL between TCSs and age-adjusted norm data, although there was a slightly lowered SF-36 Physical Component Summary Score in TCSs. Variation of QOL in TCSs was related to self-reported side effects and TC-related stress but not to TC treatment modality. A significant association was found between side effects and TC-related stress.
CONCLUSION: TCSs do not suffer long term from reduced QOL, and only minor differences in QOL were found between different treatment modalities. TCSs who report more side effects or TC-related stress have increased risk for reduced QOL, but these associations are not explained by TC treatment modalities. Further QOL research in this area should explore vulnerability factors for side effects and TC-related stress.
INTRODUCTION
The incidence of testicular cancer (TC) is increasing in most Western countries,1 and TC is a curable malignancy in more than 95% of the patients. Thus, an increasing number of patients with TC become long-term survivors (TCSs). Different postorchiectomy risk-adapted treatment options are available (surveillance, retroperitoneal lymph node dissection [RPLND], radiation, and/or chemotherapy), leading to almost equal survival rates in patients with comparable extent of TC. Therefore, in patients with low-risk disease, other parameters determine the choice of treatment, such as side effects, and presumed health-related quality of life (QOL) in the long term.
Previous research has indicated that QOL in TCSs is almost identical to the QOL of the normal population.2-4 However, reduced QOL has been reported in subgroups of patients (particularly after intensive chemotherapy),3 such as those who are single, who have a low education, and who have sexual problems. The findings of only marginal reduction in QOL and minimal influence of TC treatment modality on long-term QOL in TCSs contradict clinical intuition, although memory bias may lead to selective remembrance by clinicians of those TCSs who present with major problems. Furthermore, several previous studies are based on relatively small samples, and low statistical power could explain the lack of differences reported. Many studies have short follow-up time, and therefore, the influence of long-term treatment sequelae on QOL could have been missed. In addition, psychometrically less valid measures of QOL have frequently been used.
The mechanisms presumed to be involved in an eventual association between TC treatment modality and QOL are physical and psychological sequelae in the long term, but no studies have examined the explanatory power of these factors on the hypothesized relationship between treatment modality and QOL.
In the present study, a large, representative population of TCSs from Norway is used to examine whether self-reported QOL in the long term by TCSs is a function of having a history of TC or of TC treatment modality. In addition, we use path analysis to explore the influence on QOL of self-reported long-term side effects and TC-related psychological distress. Finally, we identify predictors of decreased QOL in TCSs.
PATIENTS AND METHODS
Study Design and Patient Recruitment
We conducted a long-term follow-up assessment of all TCSs treated in Norway between 1980 and 1994. TC patients were treated at the oncologic departments of the five university hospitals of Norway. In 1998, these departments started a collaborative cross-sectional study of unilaterally orchiectomized TCSs, addressing their long-term somatic and psychosocial health as well as QOL.
Patients treated for TC between 1980 and 1994 were identified by the Cancer Registry of Norway and cross checked with the registries of the five hospitals. Eligibility criteria for invitation to the study were age between 18 and 75 years at the time of invitation and no evidence of TC disease activity. Patients with extragonadal germ cell malignancy, bilateral TC, or a second non–germ cell malignancy (except skin cancer) were excluded; patients in whom the nonaffected testicle had been removed previously because of a benign condition were also excluded.
Treatment had preferably been administered according to the specified protocols defined by the Swedish-Norwegian testicular cancer project, the European Organisation for Research and Treatment of Cancer Genito-Urinary Group, or the Medical Research Council Testicular Cancer Working Party, as described previously.5 With few exceptions, patients with early seminoma received infradiaphragmatic radiotherapy (mean target dose, 33 Gy). Patients with advanced metastatic seminoma received cisplatin-based chemotherapy, which, up until approximately 1988, was often followed by radiotherapy or surgery. During the 1980s, patients with early nonseminoma underwent RPLND followed by adjuvant chemotherapy (three cycles) in case of metastases. Early in the 1980s, RPLND was performed as unilateral or bilateral template RPLND, whereas nerve-sparing techniques were introduced around 1990. At the end of the 1980s, the policy of primary RPLND was abandoned, and a surveillance policy was introduced for nonmetastatic low-risk patients, whereas high-risk patients received adjuvant chemotherapy (two cycles). Patients with advanced metastatic stages of nonseminoma received chemotherapy (four to six cycles) and underwent resection of residual tumor masses. The principal treatment of the 90 patients with relapse consisted of chemotherapy followed by surgery or radiotherapy, although exceptional patients had surgery only.
Until approximately 1985, cisplatin-based chemotherapy consisted of the cisplatin-vinblastine-bleomycin regimen.6 From the mid-1980s, vinblastine was substituted by etoposide.6 On the basis of these postorchiectomy treatment principles, the TCSs were allotted to one of the following four groups: (1) surveillance without subsequent relapse; (2) RPLND only; (3) radiotherapy only; and (4) chemotherapy with or without RPLND and/or RAD. Age-adjusted norm data on QOL were obtained from the Survey of Level of Living in Norway 1998,7 which comprised 2,678 male participants aged 23 to 75 years. The norm data were weighted by age to match the age distribution of the TCSs.
Instruments
Health-related QOL was assessed by the Medical Outcome Study 36-item Short Form Survey (SF-36).8,9 SF-36 assesses eight dimensions of physical and mental health, which can be summarized as the Physical Component Summary (PCS) and Mental Component Summary (MCS) scales. The data of these summary scales are presented as T scores, with a normal healthy population mean score set at 50 and a score of 60 and 40 representing 1 standard deviation above or below the mean, respectively. SF-36 has well-established general population norm data in many countries including Norway,10,11 but in this study, we used a more recent and larger norm material.7
The Impact of Event Scale (IES) measures mental symptoms caused by traumatic stress,12,13 and in this study, stress was specifically related to the experience of TC. IES-Intrusion refers to intrusively experienced images, thoughts, feelings, and dreams, and IES-Avoidance covers consciously recognized avoidance of certain feelings, ideas, or situations. Intrusion and avoidance are core symptoms of posttraumatic stress disorder, which has shown considerable relevance to cancer.14 In this article, IES scores are referred to as TC-related stress, except when referring specifically to the intrusion and avoidance subscales. IES is a distinctly different construct from MCS, which does not cover mental symptoms related to specified traumas.
Variables indicating four common TC treatment-related side effects were computed for impaired sexual function, Raynaud's phenomena and peripheral neuropathy, tinnitus and impaired hearing, and gastrointestinal complaints. The Brief Male Sexual Function Inventory (BMSFI)15 was used to measure sexual function. BMSFI covers the following four functional domains: sexual drive, erectile function, ejaculation, and sexual problem assessment. BMSFI consists of 11 items that are scored on 5-point Likert scales and has previously shown satisfactory internal consistency and test-retest reliability. In our sample of TCSs, Cronbach's coefficient was .85 across the four domains, and principal component analysis identified only one factor with an eigenvalue 1, explaining 69% of the variance. These results support a one-dimensional index for sexual dysfunctions, and therefore, a cutoff level at the 80th percentile of the total BMSFI score was identified as the most clinically relevant cutoff for impaired sexual function. By this definition, 315 TCSs (22%) with sexual impairment were identified. Two questions on Raynaud's phenomena and two on peripheral neuropathy were scored with 4-point Likert scales.16 Principal component analyses on these four questions gave a one-factor solution, with 63% explained variance ( = .75). By dichotomization of none versus one or more side effect, TCSs with at least one of these side effects were identified. The two questions on tinnitus and impaired hearing had an internal consistency of = .68. A dichotomy (yes v no) was computed and identified TCSs with ototoxicity. Three questions on gastrointestinal problems covered nausea, dyspepsia, and diarrhea ( = .58), were rated on a 3-point Likert-scale (none v mild v very much), and had also been used in a Norwegian population study.17 Patients reporting moderate or severe problems on any of the three questions on gastrointestinal problems were defined as clinical cases in the dichotomy included in the analyses.
These four side effects were included simultaneously as dichotomies for modeling side effects in the path analysis. In addition, a single-variable operationalization of two or more side effects was used.
Statistical Analysis
Multivariate linear and logistic regression analyses were used. Age was included in all models in addition to other explanatory variables. Effects are given as mean group differences of the dependent variable in each analysis, with explained variance (R2) and significance levels. The explained variances (R2) do not comprise variables previously included in the model for adjustment. For example, the explained variances in Figure 2 show the independent contribution of self-reported side effects to the explanation of the variance in QOL and do not comprise the variance explained by age, TC treatments, and TC-related stress (IES-Avoidance and IES-Intrusion). Accordingly, significance is based on F statistics of change in explained variance by adding the independent variable (for example, side effects) to other variables included for adjustment. t tests for tests of equality of means were used, and the two-sided level of significance was set at P = .05.
Ethics
The Committee for Medical Research Ethics of the Southern Health Region of Norway approved the study. The patients gave written informed consent when they returned the questionnaire.
RESULTS
Attendance
On the basis of the eligibility criteria, 1,814 TCSs were invited to take part in the survey, of whom 1,438 returned questionnaires. Nonresponders were emigrated, without permanent address, deceased, did not respond, refused to participate (n = 376), or returned incomplete questionnaires (n = 29), giving a study sample of 1,409 TCSs (compliance rate, 78%). No significant demographic or TC-related differences between noncompliers and compliers were observed (results not shown).
Demographics and Treatment Group Characteristics
There were 117, 151, 598, and 543 TCSs in the surveillance without subsequent relapse, RPLND only, radiotherapy only, and chemotherapy with or without RPLND and/or RAD groups, respectively (Table 1). The median cumulative cisplatin dose was 780 mg (range, 185 to 2,455 mg), and 105 patients received cisplatin 850 mg. A total of 358 patients had postchemotherapy RPLND, and 59 patients had radiotherapy after their cytostatic treatment.
Patients from the radiation group were significantly older than those belonging to the other treatment groups, which was as expected for seminomatous patients. Clinical characteristics and the crude prevalence of side effects are listed in Table 1; the prevalence odds ratios (ORs) and explained variances after adjustment for age are also listed in Table 1. With surveillance as reference category, the chemotherapy group suffered from the most side effects. They had significantly higher ORs for peripheral neuropathy and Raynaud's phenomena (OR = 3.8), impaired sexual function (OR = 2.5), and two or more of the side effects (OR = 3.0). The three other TC treatment modalities did not show significant differences between them. TC treatment modality explained 4.7% of the total variance in the global side effects score (Table 1).
In TCSs, MCS and PCS were associated with common sociodemographic variables. Slightly deteriorated MCS was associated with living alone and smoking, and slightly lowered PCS was associated with smoking and low educational level (data not shown).
QOL in TCSs Compared With the Norm Sample
Having adjusted for age, TCSs had significantly more bodily pain, less vitality, and poorer social functioning than the norm sample on the SF-36, but they had better mental health than the norm sample (Fig 1). These statistically significant differences were small and did not fulfill criteria for clinical significance.18 A lower mean PCS, but not MCS, was observed in the TCSs compared with the norm sample, corresponding to 0.2% explained variance. Being a TCS had no effect on MCS. Being a TCS explained only between 0.3% and 0.7% of the variance in MCS and PCS (details not shown). Thus, variation in QOL could only marginally be explained by having had TC when compared with the norm sample.
QOL in Relation to TC Treatment Modality
After age adjustment, there were no statistically significant effects of TC treatment modality on PCS and MCS. The largest, but nonsignificant, group difference of 1.3 points on PCS was found between the chemotherapy and surveillance groups (Table 2). Adjustment for TC-related stress (IES) did not change this finding, but additional adjustment for self-reported side effects resulted in a significant effect of TC treatment modality with better QOL for the chemotherapy group compared with the surveillance group. This finding indicated that TCSs who had received chemotherapy but who did not suffer from side effects reported a better QOL than TCSs from the surveillance group. No effects were observed for the burden of chemotherapy (comparing TCSs whose cumulative dose of cisplatin was 850 mg with those who had received < 850 mg; data not shown). Also, no age-adjusted differences were found between surviving patients with recurrence (n = 90) and the rest of the patient population (standardized regression coefficients were –0.021, P = .442; and –0.015, P = .57, respectively).
Effect of Side Effects and TC-Related Stress on QOL
Side effects showed strong associations with age-adjusted PCS and MCS, with 18% explained variance in PCS and 8.9% explained variance in MCS (P < .001; Table 3). Adjustment for TC treatment modalities did not contribute to the explanation of these associations, but TC-related stress reduced the explained variance by side effects on QOL from 18% to 11% in PCS and from 8.9% to 3.0% in MCS. Therefore, side effects definitely had a clinically significant effect on QOL.
TC-related stress (IES) explained 11% of the age-adjusted variance in PCS and 16% of the age-adjusted variance in MCS (Table 4). The effect of intrusion was stronger than that of avoidance (data not shown). Side effects, but not treatment modality, reduced the association between TC-related stress and QOL, and side effects reduced this association more in PCS than in MCS. Adjustment for TC treatment modality did not explain any of the associations between TC-related stress and QOL because there were no changes of group differences. Despite the substantial effects of intrusion and avoidance on QOL, a similar analysis showed that TC treatment modalities explained only 1.1% of the variance in intrusion (P < .001) and had no significant effect on avoidance (data not shown). The marginal effect of TC treatment options on intrusion was entirely explained by self-reported side effects. However, side effects contributed considerably to the explanation of variance in both intrusion (12%) and avoidance (10%), and these variances were not explained by TC treatments (details not shown).
Figure 2 summarizes the findings of our analyses. QOL was mainly influenced by self-reported side effects and TC-related stress, but none of these effects could be attributed to TC treatment modality. Nevertheless, TC treatment modalities had significant influences on both self-reported side effects and TC-related stress, but these effects were weak relative to the other effects in the path analyses.
DISCUSSION
Compared with the age-adjusted norm sample, being a TCS comprised only a weak negative effect on PCS and no effect on MCS. No statistically significant differences between TC treatment modalities were observed in relation to QOL. Side effects and TC-related stress were strongly associated with QOL, but these effects could not be explained by the selected TC treatment modality. The effects of TC treatment modality on side effects were moderate, and the effect on TC-related stress was weak.
The few studies that have analyzed the effect of different TC treatment options have had relatively small samples3,4 and have been unable to show any effect of treatment modalities, which could be a result of low statistical power. Only Rudberg et al3 showed that extraordinarily intensive treatment could lead to reduced QOL. Our study had adequate statistical power, and we were not able to show any influence of TC treatment modality on QOL. Furthermore, patients without side effects after chemotherapy displayed slightly better scores on MCS than patients belonging to other treatment groups. However, the experience of side effects reduced QOL regardless of TC treatment modality, and the number of TCSs with side effects was highest after chemotherapy. Our results indicate that associations between side effects and QOL, as well as between TC-related stress and QOL, explain a considerable proportion of the variance in QOL, which is not attributable to TC treatment modality. This surprising finding may be explained by the fact that the symptoms described in this article as side effects, such as hearing deficits, tinnitus, and slight gastrointestinal distress, are frequently reported by males in the normal population. Haug et al,17 for example, described that 48% of the general Norwegian population reported nausea, heartburn, diarrhea, or constipation during the last year. Foss et al19 observed that patients with metastatic TC scored their symptoms of perceived peripheral Raynaud's-like symptoms and perceived neuropathy at a mean of about 5 on a scale ranging from 0 to 100, even before any chemotherapy was applied.
The findings that being a TCS and TC treatment modalities hardly influence QOL seem to contradict clinical intuition. Similar findings from the literature have commonly been debated,2-4 with reference to selection biases and lack of statistical power. Our large sample makes such objections less relevant. These findings raise the following questions: why is the impact of being a TCS on QOL so weak, and why doesn't TC treatment modality influence QOL We suggest two possible answers. SF-36, as a generic QOL measure, may lack coverage of specific parameters with high significance for QOL in TCSs, such as changes in body image and masculinity.18 Alternatively, the lack of difference in QOL in long-term TCSs compared with norm data and between TC treatment groups might be a result of response shift,20,21 which could occur when a TCS with impaired functions or reduced QOL has become used to his situation and has regained a mainly positive outlook because of the cure of a life-threatening disease.
We found that self-reported side effects were strongly associated with QOL, but the association between side effects and QOL could not be attributed to treatment modality. These results do not undermine the impact of long-term self-reported side effects on QOL or the consequences for TCSs suffering from such morbidities. However, our findings cast doubts on the often claimed strong relationship between patient-perceived side effects and TC treatment modality. Objectively assessed side effects would probably ensure stronger association to TC treatment modality but would probably also have a weaker association with QOL. Therefore, we doubt that TC treatment modality would explain more of the eventual effects of objectively assessed side effects on QOL. However, this is an empirical question of clinical relevance that should be addressed in future research on this topic.
TC-related stress was strongly associated with QOL. Furthermore, the association between TC treatment modality and TC-related stress was only marginal, but TC treatments did not explain any of this association. TC-related stress many years after treatment is probably more related to the psychological characteristics of the patient before TC than to the treatment itself.
Studies of QOL in TCSs frequently include identification of patient subgroups that show high risk for reduced QOL, as shown in Table 3. Typically, associations between QOL and educational level, employment, or civil status within a group of TCSs are reported. These parameters are then used to describe subgroups at risk for low QOL. Although the selected sociodemographic variables, such as educational level and smoking, most probably also are relevant for the normal population, they might assist the clinicians in their attempts to understand factors influencing the QOL in TCSs beyond TC treatment strategies, side effects, and TC-related stress. However, a major theoretical problem of this approach is that important clinical parameters, such as being a TCS, are not included as a mediator or moderator in such models. To test whether a demographic parameter assumed to be specific in TCSs, for example being involuntarily childless, has any specific relevance for QOL in TCSs as opposed to the general population, the model must include TC, childlessness, and the interaction term TC by childlessness as independent variables. Childlessness has a TC-related effect on QOL only if the interaction term TC by childlessness is statistically significant.
Our study has the following three clinical implications: (1) TCSs, in general, do not suffer from reduced QOL, and interventions to improve long-term QOL in TCSs should only be offered to identified subgroups with low QOL; (2) TC treatment modality does not predict long-term QOL in TCSs; therefore, as a general rule, QOL considerations should not have a major impact on choice of treatment in TC patients; (3) TCSs who report more side effects or more TC-related stress are more likely to show reduced QOL, but these associations are not attributable to TC treatment strategies.
Further research of QOL in TCSs (and other groups of cancer patients) should explore the mechanisms underlying the associations between QOL, self-reported side effects, and cancer-related stress. The patient's premorbid psychological vulnerability, which influences perception of side effects and of the experience of cancer-related psychological distress, may, to a certain degree, explain the associations between side effects, cancer-related stress, and QOL and the lack of an explanatory relationship between TC modality and QOL. Such vulnerability could be related to personality, coping strategies, cognitive processes, ability to benefit from social support, or intelligence.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Supported by Cancer Registry of Norway, The Norwegian Cancer Society, The Norwegian Foundation for Health and Rehabilitation, and Bothner's Foundation in Norway.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Cancer Registry of Norway: Cancer in Norway 2000. Oslo, Norway, Cancer Registry of Norway, 2002
Kaasa S, Aass N, Mastekaasa A, et al: Psychosocial well-being in testicular cancer patients. Eur J Cancer 27:1091-1095, 1991
Rudberg L, Carlsson M, Nilsson S, et al: Self-perceived physical, psychologic, and general symptoms in survivors of testicular cancer 3 to 13 years after treatment. Cancer Nurs 25:187-195, 2002
Joly F, Heron JF, Kalusinski L, et al: Quality of life in long-term survivors of testicular cancer: A population-based case-control study. J Clin Oncol 20:73-80, 2002
Fossa SD, Dahl AA, Loge JH: Fatigue, anxiety, and depression in long-term survivors of testicular cancer. J Clin Oncol 21:1249-1254, 2003
Williams SD, Birch R, Einhorn LH, et al: Treatment of disseminated germ-cell tumors with cisplatin, bleomycin, and either vinblastine or etoposide. N Engl J Med 316:1435-1440, 1987
Heine Strand B, Dalgard OS, Tambs K, et al: Measuring the mental health status of the Norwegian population: A comparison of the instruments SCL-25, SCL-10, SCL-5 and MHI-5 (SF-36). Nord J Psychiatry 57:113-118, 2003
Ware JE, Kosinski M, Gandek B, et al: The factor structure of the SF-36 Health Survey in 10 countries: Results from the IQOLA Project—International Quality of Life Assessment. J Clin Epidemiol 51:1159-1165, 1998
Ware JE, Snow KK, Kosinski M: SF-36 Health Survey: Manual and Interpretation Guide. Lincoln, RI, QualityMetric Incorporated, 2000
Loge JH, Kaasa S, Hjermstad MJ, et al: Translation and performance of the Norwegian SF-36 Health Survey in patients with rheumatoid arthritis: I. Data quality, scaling assumptions, reliability, and construct validity. J Clin Epidemiol 51:1069-1076, 1998
Loge JH, Kaasa S: Short form 36 (SF-36) health survey: Normative data from the general Norwegian population. Scand J Soc Med 26:250-258, 1998
Sundin EC, Horowitz MJ: Impact of Event Scale: Psychometric properties. Br J Psychiatry 180:205-209, 2002
Thewes B, Meiser B, Hickie IB: Psychometric properties of the impact of event scale among women at increased risk for hereditary breast cancer. Psychooncology 10:459-468, 2001
Gurevich M, Devins GM, Rodin GM: Stress response syndromes and cancer: Conceptual and assessment issues. Psychosomatics 43:259-281, 2002
O'Leary MP, Fowler FJ, Lenderking WR, et al: A brief male sexual function inventory for urology. Urology 46:697-706, 1995
Fossa SD, Moynihan C, Serbouti S: Patients' and doctors' perception of long-term morbidity in patients with testicular cancer clinical stage I: A descriptive pilot study. Support Care Cancer 4:118-128, 1996
Haug T, Mykletun A, Dahl A: Are anxiety and depression related to gastrointestinal symptoms in the general population Scand J Gastroenterol 37:294-298, 2002
Jacobsen NS, Truax P: Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. J Consult Clin Psychol 59:12-19, 1991
Foss SD, de Wit R, Roberts JT, et al: Quality of life in good prognosis patients with metastatic germ cell cancer: A prospective study of the European Organization for Research and Treatment of Cancer Genitourinary Group/Medical Research Council Testicular Cancer Study Group (30941/TE20). J Clin Oncol 21:1107-1118, 2003
Basten JP, Jonker-Pool G, van Driel MF: Fantasies and facts of the testes. Br J Urol 78:756-762, 1996
Norman G: Hi! How are you Response shift, implicit theories and differing epistemologies. Qual Life Res 12:239-249, 2003(Arnstein Mykletun, Alv A.)
Department of Clinical Cancer Research, The Norwegian Radium Hospital
Departments of Psychiatry and Oncology, Ullevaal University Hospital, and Medical Faculty, University of Oslo, Oslo
Department of Oncology, University Hospital of Northern Norway, University of Troms, Troms
Department of Oncology, Saint Olav's Hospital, National University for Science and Technology, Trondheim, Norway
ABSTRACT
PURPOSE: The prevalence of long-term survivors after treatment for testicular cancer (TC) is increasing, and most studies display normal or only slightly reduced quality of life (QOL) in TC survivors (TCSs). Impaired QOL is claimed to be associated with treatment modality and its side effects, although most studies in this field can be criticized for various methodologic shortcomings. We wanted to examine variation in long-term QOL in TCSs in relation to TC treatment modality, side effects, and TC-related stress in a large population.
PATIENTS AND METHODS: QOL, side effects, and TC-related stress were self-rated by a questionnaire at a mean of 11 years of follow-up in 1,409 TCSs treated from 1980 to 1994. Norm data was obtained from 2,678 males who were representative of the general population. QOL was measured with the Short Form-36 (SF-36), and TC-related stress was measured with the Impact of Event Scale.
RESULTS: There were no clinically relevant differences in QOL between TCSs and age-adjusted norm data, although there was a slightly lowered SF-36 Physical Component Summary Score in TCSs. Variation of QOL in TCSs was related to self-reported side effects and TC-related stress but not to TC treatment modality. A significant association was found between side effects and TC-related stress.
CONCLUSION: TCSs do not suffer long term from reduced QOL, and only minor differences in QOL were found between different treatment modalities. TCSs who report more side effects or TC-related stress have increased risk for reduced QOL, but these associations are not explained by TC treatment modalities. Further QOL research in this area should explore vulnerability factors for side effects and TC-related stress.
INTRODUCTION
The incidence of testicular cancer (TC) is increasing in most Western countries,1 and TC is a curable malignancy in more than 95% of the patients. Thus, an increasing number of patients with TC become long-term survivors (TCSs). Different postorchiectomy risk-adapted treatment options are available (surveillance, retroperitoneal lymph node dissection [RPLND], radiation, and/or chemotherapy), leading to almost equal survival rates in patients with comparable extent of TC. Therefore, in patients with low-risk disease, other parameters determine the choice of treatment, such as side effects, and presumed health-related quality of life (QOL) in the long term.
Previous research has indicated that QOL in TCSs is almost identical to the QOL of the normal population.2-4 However, reduced QOL has been reported in subgroups of patients (particularly after intensive chemotherapy),3 such as those who are single, who have a low education, and who have sexual problems. The findings of only marginal reduction in QOL and minimal influence of TC treatment modality on long-term QOL in TCSs contradict clinical intuition, although memory bias may lead to selective remembrance by clinicians of those TCSs who present with major problems. Furthermore, several previous studies are based on relatively small samples, and low statistical power could explain the lack of differences reported. Many studies have short follow-up time, and therefore, the influence of long-term treatment sequelae on QOL could have been missed. In addition, psychometrically less valid measures of QOL have frequently been used.
The mechanisms presumed to be involved in an eventual association between TC treatment modality and QOL are physical and psychological sequelae in the long term, but no studies have examined the explanatory power of these factors on the hypothesized relationship between treatment modality and QOL.
In the present study, a large, representative population of TCSs from Norway is used to examine whether self-reported QOL in the long term by TCSs is a function of having a history of TC or of TC treatment modality. In addition, we use path analysis to explore the influence on QOL of self-reported long-term side effects and TC-related psychological distress. Finally, we identify predictors of decreased QOL in TCSs.
PATIENTS AND METHODS
Study Design and Patient Recruitment
We conducted a long-term follow-up assessment of all TCSs treated in Norway between 1980 and 1994. TC patients were treated at the oncologic departments of the five university hospitals of Norway. In 1998, these departments started a collaborative cross-sectional study of unilaterally orchiectomized TCSs, addressing their long-term somatic and psychosocial health as well as QOL.
Patients treated for TC between 1980 and 1994 were identified by the Cancer Registry of Norway and cross checked with the registries of the five hospitals. Eligibility criteria for invitation to the study were age between 18 and 75 years at the time of invitation and no evidence of TC disease activity. Patients with extragonadal germ cell malignancy, bilateral TC, or a second non–germ cell malignancy (except skin cancer) were excluded; patients in whom the nonaffected testicle had been removed previously because of a benign condition were also excluded.
Treatment had preferably been administered according to the specified protocols defined by the Swedish-Norwegian testicular cancer project, the European Organisation for Research and Treatment of Cancer Genito-Urinary Group, or the Medical Research Council Testicular Cancer Working Party, as described previously.5 With few exceptions, patients with early seminoma received infradiaphragmatic radiotherapy (mean target dose, 33 Gy). Patients with advanced metastatic seminoma received cisplatin-based chemotherapy, which, up until approximately 1988, was often followed by radiotherapy or surgery. During the 1980s, patients with early nonseminoma underwent RPLND followed by adjuvant chemotherapy (three cycles) in case of metastases. Early in the 1980s, RPLND was performed as unilateral or bilateral template RPLND, whereas nerve-sparing techniques were introduced around 1990. At the end of the 1980s, the policy of primary RPLND was abandoned, and a surveillance policy was introduced for nonmetastatic low-risk patients, whereas high-risk patients received adjuvant chemotherapy (two cycles). Patients with advanced metastatic stages of nonseminoma received chemotherapy (four to six cycles) and underwent resection of residual tumor masses. The principal treatment of the 90 patients with relapse consisted of chemotherapy followed by surgery or radiotherapy, although exceptional patients had surgery only.
Until approximately 1985, cisplatin-based chemotherapy consisted of the cisplatin-vinblastine-bleomycin regimen.6 From the mid-1980s, vinblastine was substituted by etoposide.6 On the basis of these postorchiectomy treatment principles, the TCSs were allotted to one of the following four groups: (1) surveillance without subsequent relapse; (2) RPLND only; (3) radiotherapy only; and (4) chemotherapy with or without RPLND and/or RAD. Age-adjusted norm data on QOL were obtained from the Survey of Level of Living in Norway 1998,7 which comprised 2,678 male participants aged 23 to 75 years. The norm data were weighted by age to match the age distribution of the TCSs.
Instruments
Health-related QOL was assessed by the Medical Outcome Study 36-item Short Form Survey (SF-36).8,9 SF-36 assesses eight dimensions of physical and mental health, which can be summarized as the Physical Component Summary (PCS) and Mental Component Summary (MCS) scales. The data of these summary scales are presented as T scores, with a normal healthy population mean score set at 50 and a score of 60 and 40 representing 1 standard deviation above or below the mean, respectively. SF-36 has well-established general population norm data in many countries including Norway,10,11 but in this study, we used a more recent and larger norm material.7
The Impact of Event Scale (IES) measures mental symptoms caused by traumatic stress,12,13 and in this study, stress was specifically related to the experience of TC. IES-Intrusion refers to intrusively experienced images, thoughts, feelings, and dreams, and IES-Avoidance covers consciously recognized avoidance of certain feelings, ideas, or situations. Intrusion and avoidance are core symptoms of posttraumatic stress disorder, which has shown considerable relevance to cancer.14 In this article, IES scores are referred to as TC-related stress, except when referring specifically to the intrusion and avoidance subscales. IES is a distinctly different construct from MCS, which does not cover mental symptoms related to specified traumas.
Variables indicating four common TC treatment-related side effects were computed for impaired sexual function, Raynaud's phenomena and peripheral neuropathy, tinnitus and impaired hearing, and gastrointestinal complaints. The Brief Male Sexual Function Inventory (BMSFI)15 was used to measure sexual function. BMSFI covers the following four functional domains: sexual drive, erectile function, ejaculation, and sexual problem assessment. BMSFI consists of 11 items that are scored on 5-point Likert scales and has previously shown satisfactory internal consistency and test-retest reliability. In our sample of TCSs, Cronbach's coefficient was .85 across the four domains, and principal component analysis identified only one factor with an eigenvalue 1, explaining 69% of the variance. These results support a one-dimensional index for sexual dysfunctions, and therefore, a cutoff level at the 80th percentile of the total BMSFI score was identified as the most clinically relevant cutoff for impaired sexual function. By this definition, 315 TCSs (22%) with sexual impairment were identified. Two questions on Raynaud's phenomena and two on peripheral neuropathy were scored with 4-point Likert scales.16 Principal component analyses on these four questions gave a one-factor solution, with 63% explained variance ( = .75). By dichotomization of none versus one or more side effect, TCSs with at least one of these side effects were identified. The two questions on tinnitus and impaired hearing had an internal consistency of = .68. A dichotomy (yes v no) was computed and identified TCSs with ototoxicity. Three questions on gastrointestinal problems covered nausea, dyspepsia, and diarrhea ( = .58), were rated on a 3-point Likert-scale (none v mild v very much), and had also been used in a Norwegian population study.17 Patients reporting moderate or severe problems on any of the three questions on gastrointestinal problems were defined as clinical cases in the dichotomy included in the analyses.
These four side effects were included simultaneously as dichotomies for modeling side effects in the path analysis. In addition, a single-variable operationalization of two or more side effects was used.
Statistical Analysis
Multivariate linear and logistic regression analyses were used. Age was included in all models in addition to other explanatory variables. Effects are given as mean group differences of the dependent variable in each analysis, with explained variance (R2) and significance levels. The explained variances (R2) do not comprise variables previously included in the model for adjustment. For example, the explained variances in Figure 2 show the independent contribution of self-reported side effects to the explanation of the variance in QOL and do not comprise the variance explained by age, TC treatments, and TC-related stress (IES-Avoidance and IES-Intrusion). Accordingly, significance is based on F statistics of change in explained variance by adding the independent variable (for example, side effects) to other variables included for adjustment. t tests for tests of equality of means were used, and the two-sided level of significance was set at P = .05.
Ethics
The Committee for Medical Research Ethics of the Southern Health Region of Norway approved the study. The patients gave written informed consent when they returned the questionnaire.
RESULTS
Attendance
On the basis of the eligibility criteria, 1,814 TCSs were invited to take part in the survey, of whom 1,438 returned questionnaires. Nonresponders were emigrated, without permanent address, deceased, did not respond, refused to participate (n = 376), or returned incomplete questionnaires (n = 29), giving a study sample of 1,409 TCSs (compliance rate, 78%). No significant demographic or TC-related differences between noncompliers and compliers were observed (results not shown).
Demographics and Treatment Group Characteristics
There were 117, 151, 598, and 543 TCSs in the surveillance without subsequent relapse, RPLND only, radiotherapy only, and chemotherapy with or without RPLND and/or RAD groups, respectively (Table 1). The median cumulative cisplatin dose was 780 mg (range, 185 to 2,455 mg), and 105 patients received cisplatin 850 mg. A total of 358 patients had postchemotherapy RPLND, and 59 patients had radiotherapy after their cytostatic treatment.
Patients from the radiation group were significantly older than those belonging to the other treatment groups, which was as expected for seminomatous patients. Clinical characteristics and the crude prevalence of side effects are listed in Table 1; the prevalence odds ratios (ORs) and explained variances after adjustment for age are also listed in Table 1. With surveillance as reference category, the chemotherapy group suffered from the most side effects. They had significantly higher ORs for peripheral neuropathy and Raynaud's phenomena (OR = 3.8), impaired sexual function (OR = 2.5), and two or more of the side effects (OR = 3.0). The three other TC treatment modalities did not show significant differences between them. TC treatment modality explained 4.7% of the total variance in the global side effects score (Table 1).
In TCSs, MCS and PCS were associated with common sociodemographic variables. Slightly deteriorated MCS was associated with living alone and smoking, and slightly lowered PCS was associated with smoking and low educational level (data not shown).
QOL in TCSs Compared With the Norm Sample
Having adjusted for age, TCSs had significantly more bodily pain, less vitality, and poorer social functioning than the norm sample on the SF-36, but they had better mental health than the norm sample (Fig 1). These statistically significant differences were small and did not fulfill criteria for clinical significance.18 A lower mean PCS, but not MCS, was observed in the TCSs compared with the norm sample, corresponding to 0.2% explained variance. Being a TCS had no effect on MCS. Being a TCS explained only between 0.3% and 0.7% of the variance in MCS and PCS (details not shown). Thus, variation in QOL could only marginally be explained by having had TC when compared with the norm sample.
QOL in Relation to TC Treatment Modality
After age adjustment, there were no statistically significant effects of TC treatment modality on PCS and MCS. The largest, but nonsignificant, group difference of 1.3 points on PCS was found between the chemotherapy and surveillance groups (Table 2). Adjustment for TC-related stress (IES) did not change this finding, but additional adjustment for self-reported side effects resulted in a significant effect of TC treatment modality with better QOL for the chemotherapy group compared with the surveillance group. This finding indicated that TCSs who had received chemotherapy but who did not suffer from side effects reported a better QOL than TCSs from the surveillance group. No effects were observed for the burden of chemotherapy (comparing TCSs whose cumulative dose of cisplatin was 850 mg with those who had received < 850 mg; data not shown). Also, no age-adjusted differences were found between surviving patients with recurrence (n = 90) and the rest of the patient population (standardized regression coefficients were –0.021, P = .442; and –0.015, P = .57, respectively).
Effect of Side Effects and TC-Related Stress on QOL
Side effects showed strong associations with age-adjusted PCS and MCS, with 18% explained variance in PCS and 8.9% explained variance in MCS (P < .001; Table 3). Adjustment for TC treatment modalities did not contribute to the explanation of these associations, but TC-related stress reduced the explained variance by side effects on QOL from 18% to 11% in PCS and from 8.9% to 3.0% in MCS. Therefore, side effects definitely had a clinically significant effect on QOL.
TC-related stress (IES) explained 11% of the age-adjusted variance in PCS and 16% of the age-adjusted variance in MCS (Table 4). The effect of intrusion was stronger than that of avoidance (data not shown). Side effects, but not treatment modality, reduced the association between TC-related stress and QOL, and side effects reduced this association more in PCS than in MCS. Adjustment for TC treatment modality did not explain any of the associations between TC-related stress and QOL because there were no changes of group differences. Despite the substantial effects of intrusion and avoidance on QOL, a similar analysis showed that TC treatment modalities explained only 1.1% of the variance in intrusion (P < .001) and had no significant effect on avoidance (data not shown). The marginal effect of TC treatment options on intrusion was entirely explained by self-reported side effects. However, side effects contributed considerably to the explanation of variance in both intrusion (12%) and avoidance (10%), and these variances were not explained by TC treatments (details not shown).
Figure 2 summarizes the findings of our analyses. QOL was mainly influenced by self-reported side effects and TC-related stress, but none of these effects could be attributed to TC treatment modality. Nevertheless, TC treatment modalities had significant influences on both self-reported side effects and TC-related stress, but these effects were weak relative to the other effects in the path analyses.
DISCUSSION
Compared with the age-adjusted norm sample, being a TCS comprised only a weak negative effect on PCS and no effect on MCS. No statistically significant differences between TC treatment modalities were observed in relation to QOL. Side effects and TC-related stress were strongly associated with QOL, but these effects could not be explained by the selected TC treatment modality. The effects of TC treatment modality on side effects were moderate, and the effect on TC-related stress was weak.
The few studies that have analyzed the effect of different TC treatment options have had relatively small samples3,4 and have been unable to show any effect of treatment modalities, which could be a result of low statistical power. Only Rudberg et al3 showed that extraordinarily intensive treatment could lead to reduced QOL. Our study had adequate statistical power, and we were not able to show any influence of TC treatment modality on QOL. Furthermore, patients without side effects after chemotherapy displayed slightly better scores on MCS than patients belonging to other treatment groups. However, the experience of side effects reduced QOL regardless of TC treatment modality, and the number of TCSs with side effects was highest after chemotherapy. Our results indicate that associations between side effects and QOL, as well as between TC-related stress and QOL, explain a considerable proportion of the variance in QOL, which is not attributable to TC treatment modality. This surprising finding may be explained by the fact that the symptoms described in this article as side effects, such as hearing deficits, tinnitus, and slight gastrointestinal distress, are frequently reported by males in the normal population. Haug et al,17 for example, described that 48% of the general Norwegian population reported nausea, heartburn, diarrhea, or constipation during the last year. Foss et al19 observed that patients with metastatic TC scored their symptoms of perceived peripheral Raynaud's-like symptoms and perceived neuropathy at a mean of about 5 on a scale ranging from 0 to 100, even before any chemotherapy was applied.
The findings that being a TCS and TC treatment modalities hardly influence QOL seem to contradict clinical intuition. Similar findings from the literature have commonly been debated,2-4 with reference to selection biases and lack of statistical power. Our large sample makes such objections less relevant. These findings raise the following questions: why is the impact of being a TCS on QOL so weak, and why doesn't TC treatment modality influence QOL We suggest two possible answers. SF-36, as a generic QOL measure, may lack coverage of specific parameters with high significance for QOL in TCSs, such as changes in body image and masculinity.18 Alternatively, the lack of difference in QOL in long-term TCSs compared with norm data and between TC treatment groups might be a result of response shift,20,21 which could occur when a TCS with impaired functions or reduced QOL has become used to his situation and has regained a mainly positive outlook because of the cure of a life-threatening disease.
We found that self-reported side effects were strongly associated with QOL, but the association between side effects and QOL could not be attributed to treatment modality. These results do not undermine the impact of long-term self-reported side effects on QOL or the consequences for TCSs suffering from such morbidities. However, our findings cast doubts on the often claimed strong relationship between patient-perceived side effects and TC treatment modality. Objectively assessed side effects would probably ensure stronger association to TC treatment modality but would probably also have a weaker association with QOL. Therefore, we doubt that TC treatment modality would explain more of the eventual effects of objectively assessed side effects on QOL. However, this is an empirical question of clinical relevance that should be addressed in future research on this topic.
TC-related stress was strongly associated with QOL. Furthermore, the association between TC treatment modality and TC-related stress was only marginal, but TC treatments did not explain any of this association. TC-related stress many years after treatment is probably more related to the psychological characteristics of the patient before TC than to the treatment itself.
Studies of QOL in TCSs frequently include identification of patient subgroups that show high risk for reduced QOL, as shown in Table 3. Typically, associations between QOL and educational level, employment, or civil status within a group of TCSs are reported. These parameters are then used to describe subgroups at risk for low QOL. Although the selected sociodemographic variables, such as educational level and smoking, most probably also are relevant for the normal population, they might assist the clinicians in their attempts to understand factors influencing the QOL in TCSs beyond TC treatment strategies, side effects, and TC-related stress. However, a major theoretical problem of this approach is that important clinical parameters, such as being a TCS, are not included as a mediator or moderator in such models. To test whether a demographic parameter assumed to be specific in TCSs, for example being involuntarily childless, has any specific relevance for QOL in TCSs as opposed to the general population, the model must include TC, childlessness, and the interaction term TC by childlessness as independent variables. Childlessness has a TC-related effect on QOL only if the interaction term TC by childlessness is statistically significant.
Our study has the following three clinical implications: (1) TCSs, in general, do not suffer from reduced QOL, and interventions to improve long-term QOL in TCSs should only be offered to identified subgroups with low QOL; (2) TC treatment modality does not predict long-term QOL in TCSs; therefore, as a general rule, QOL considerations should not have a major impact on choice of treatment in TC patients; (3) TCSs who report more side effects or more TC-related stress are more likely to show reduced QOL, but these associations are not attributable to TC treatment strategies.
Further research of QOL in TCSs (and other groups of cancer patients) should explore the mechanisms underlying the associations between QOL, self-reported side effects, and cancer-related stress. The patient's premorbid psychological vulnerability, which influences perception of side effects and of the experience of cancer-related psychological distress, may, to a certain degree, explain the associations between side effects, cancer-related stress, and QOL and the lack of an explanatory relationship between TC modality and QOL. Such vulnerability could be related to personality, coping strategies, cognitive processes, ability to benefit from social support, or intelligence.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Supported by Cancer Registry of Norway, The Norwegian Cancer Society, The Norwegian Foundation for Health and Rehabilitation, and Bothner's Foundation in Norway.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Cancer Registry of Norway: Cancer in Norway 2000. Oslo, Norway, Cancer Registry of Norway, 2002
Kaasa S, Aass N, Mastekaasa A, et al: Psychosocial well-being in testicular cancer patients. Eur J Cancer 27:1091-1095, 1991
Rudberg L, Carlsson M, Nilsson S, et al: Self-perceived physical, psychologic, and general symptoms in survivors of testicular cancer 3 to 13 years after treatment. Cancer Nurs 25:187-195, 2002
Joly F, Heron JF, Kalusinski L, et al: Quality of life in long-term survivors of testicular cancer: A population-based case-control study. J Clin Oncol 20:73-80, 2002
Fossa SD, Dahl AA, Loge JH: Fatigue, anxiety, and depression in long-term survivors of testicular cancer. J Clin Oncol 21:1249-1254, 2003
Williams SD, Birch R, Einhorn LH, et al: Treatment of disseminated germ-cell tumors with cisplatin, bleomycin, and either vinblastine or etoposide. N Engl J Med 316:1435-1440, 1987
Heine Strand B, Dalgard OS, Tambs K, et al: Measuring the mental health status of the Norwegian population: A comparison of the instruments SCL-25, SCL-10, SCL-5 and MHI-5 (SF-36). Nord J Psychiatry 57:113-118, 2003
Ware JE, Kosinski M, Gandek B, et al: The factor structure of the SF-36 Health Survey in 10 countries: Results from the IQOLA Project—International Quality of Life Assessment. J Clin Epidemiol 51:1159-1165, 1998
Ware JE, Snow KK, Kosinski M: SF-36 Health Survey: Manual and Interpretation Guide. Lincoln, RI, QualityMetric Incorporated, 2000
Loge JH, Kaasa S, Hjermstad MJ, et al: Translation and performance of the Norwegian SF-36 Health Survey in patients with rheumatoid arthritis: I. Data quality, scaling assumptions, reliability, and construct validity. J Clin Epidemiol 51:1069-1076, 1998
Loge JH, Kaasa S: Short form 36 (SF-36) health survey: Normative data from the general Norwegian population. Scand J Soc Med 26:250-258, 1998
Sundin EC, Horowitz MJ: Impact of Event Scale: Psychometric properties. Br J Psychiatry 180:205-209, 2002
Thewes B, Meiser B, Hickie IB: Psychometric properties of the impact of event scale among women at increased risk for hereditary breast cancer. Psychooncology 10:459-468, 2001
Gurevich M, Devins GM, Rodin GM: Stress response syndromes and cancer: Conceptual and assessment issues. Psychosomatics 43:259-281, 2002
O'Leary MP, Fowler FJ, Lenderking WR, et al: A brief male sexual function inventory for urology. Urology 46:697-706, 1995
Fossa SD, Moynihan C, Serbouti S: Patients' and doctors' perception of long-term morbidity in patients with testicular cancer clinical stage I: A descriptive pilot study. Support Care Cancer 4:118-128, 1996
Haug T, Mykletun A, Dahl A: Are anxiety and depression related to gastrointestinal symptoms in the general population Scand J Gastroenterol 37:294-298, 2002
Jacobsen NS, Truax P: Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. J Consult Clin Psychol 59:12-19, 1991
Foss SD, de Wit R, Roberts JT, et al: Quality of life in good prognosis patients with metastatic germ cell cancer: A prospective study of the European Organization for Research and Treatment of Cancer Genitourinary Group/Medical Research Council Testicular Cancer Study Group (30941/TE20). J Clin Oncol 21:1107-1118, 2003
Basten JP, Jonker-Pool G, van Driel MF: Fantasies and facts of the testes. Br J Urol 78:756-762, 1996
Norman G: Hi! How are you Response shift, implicit theories and differing epistemologies. Qual Life Res 12:239-249, 2003(Arnstein Mykletun, Alv A.)