Potential gonadotoxicity of treatment in relation to quality of life and mental well-being of male survivors of childhood acute lymphoblastic leukemia
- First Online:
- Cite this article as:
- Gunn, M.E., Lähteenmäki, P.M., Puukko-Viertomies, L. et al. J Cancer Surviv (2013) 7: 404. doi:10.1007/s11764-013-0285-7
Results of earlier studies concerning quality of life (QOL) and psychosocial coping of childhood acute lymphoblastic leukemia (ALL) survivors have been inconsistent. Some treatments for ALL affect testicular function and we hypothesized that this may influence the QOL and psychosocial coping of male survivors. Our aims were to assess the QOL and psychosocial coping of male long-term ALL survivors and to evaluate the effect of both testosterone level and the potential gonadotoxicity of various treatment modalities on them.
Fifty-two male long-term survivors treated for childhood ALL at Helsinki University Hospital between 1970 and 1995, and 56 age- and gender-matched controls were studied. The participants completed a self-report questionnaire including questions on sociodemographics, RAND-36 to assess QOL, General Health Questionnaire and Beck Depression Inventory to assess mental well-being, and CAGE to assess alcohol abuse/dependence. Testosterone levels were measured, and treatment details were reviewed.
ALL survivors in general had QOL close to that of controls or population norms. Decreased QOL was seen in physical health-related subscales, and vitality and emotional well-being were lowered in survivors with more gonadotoxic treatment modalities. No single independent factor in the treatment or the level of testosterone could, however, be found to clearly explain the variation in QOL scores of the survivors. Mental well-being of most of the survivors was good, but a subgroup with previous cyclophosphamide treatment or testicular irradiation showed increased risk of psychiatric morbidity.
The male ALL survivors generally cope well, but increased focus on specific risk groups seems to be necessary. Further studies using patient interviews would probably point out issues concerning the QOL and psychosocial coping of ALL survivors, which may not emerge in these screening studies.
Implications for Cancer Survivors
In general, more attention should be paid for physical functioning of childhood ALL survivors. Increased focus should also be on QOL and mental well-being of survivors with more gonadotoxic treatment modalities and those whose diagnosis was made in their adolescence.
KeywordsAcute lymphoblastic leukemiaQuality of lifePsychosocialAdolescent/young adultSurvivorshipTesticular function
During the past years, there has been a wide interest in studying quality of life (QOL) of childhood cancer survivors. Because of the large scale of childhood cancer diagnoses and the variety of treatment protocols, the results of most of the publications studying all cancer survivors together, may, however, not necessarily be generalizable to any specific survivor group. The difficulties survivors face at different ages may also differ. Young adults meet new challenges like higher education, work life, cohabitation and marriage, and fertility issues, which may strongly influence their QOL, and to which previous cancer experience may impact. Some studies have been conducted focusing on young adult survivors of acute lymphoblastic leukemia (ALL) specifically [1–4]. However, the results of those studies have been somewhat inconsistent.
The main objective of this study was to evaluate the health-related quality of life (HRQOL), and psychological (including alcohol abuse), social and vocational outcomes of young adult male survivors of childhood ALL using valid screening instruments: RAND-36-Item Health Survey, General Health Questionnaire (GHQ-12), The Beck Depression Inventory (BDI), (CAGE) [5–16]. The studied patient cohort is a part of a larger survey where gonadal function and fertility outcome of male long-term survivors of ALL was studied. In our earlier study of this same patient cohort, we showed that treatment of childhood leukemia with high cumulative dose of cyclophosphamide (>20 g/m2) and testicular irradiation (10–24 Gy) led to permanent infertility, while use of 0–10 g/m2 of cyclophosphamide and cranial irradiation caused long-term Leydig cell function without fertility impairment . We therefore hypothesized that testicular dysfunction could also impact the QOL and psychosocial coping of male ALL survivors. Thus, our aim was to evaluate if gonadotoxic treatment modalities of childhood acute lymphoblastic leukemia and low serum testosterone level affect on the QOL and psychological outcome in adulthood.
The entire cohort of 164 boys diagnosed with ALL below the age of 16 years at the Hospital for Children and Adolescents, Helsinki University Hospital during 1970–1995 was included in this study. Altogether, 77 patients (47 %) became long-term survivors. An invitation letter was sent to 75 long-term survivors living in Finland. Fifty-two of these men (69 %) responded. Based on hospital records, the clinical characteristics of the 23 nonresponders were comparable to those of the responders.
A total of 56 age- and gender-matched controls without a history of cancer participated in the study. The controls were recruited by invitation letter from occupational health services of the Helsinki municipality area and the Helsinki University Hospital.
All participants provided written informed consent according to the study protocol approved by The Ethics Committee of Helsinki University Hospital and underwent the same examinations.
All controls and survivors underwent physical examination and provided blood and semen samples. Adult semen quality, attained fertility, and androgen insufficiency were correlated to various treatment modalities of leukemia treatment. These results are analyzed and published separately . The subgrouping of the present material is based on the results of this study . Group 1 had had no cyclophosphamide treatment or testicular irradiation (group 1, N = 16, 7/16 with central nervous system radiotherapy (CNSRT)). Group 2 had had ≤10 g/m2 cyclophosphamide but no testicular irradiation (group 2, N = 12, 11/12 with CNSRT). Group 3 had had testicular irradiation ± cyclophosphamide treatment (group 3, N = 18; 17/18 with CNSRT). The survivor group with cyclophosphamide dose exceeding 20 g/m2 and no testicular irradiation was not analyzed in the present study because of the small sample size (N = 5).
The participants filled a 200-item questionnaire including data on sociodemographics, QOL, and psychological outcome. Each participant was asked to provide a blood sample. Data on medical history of ALL survivors were reviewed, including the date of diagnosis, type and doses of ALL treatment, and data on possible relapses.
RAND-36  is a self-report questionnaire assessing HRQOL. It has been widely used and its validity and reliability have also been studied in the Finnish population. Aalto et al. studied in 1999 the reliability of RAND-36 in Finnish population and estimated Finnish population norm values for it. They estimated normative values separately for different gender- and age groups. The information had been collected as postal questionnaires and it had a response rate of 64 % (N = 2,175). The population norm values determined for 25–34-year-old males were used in the present analysis to represent normative population values . RAND-36 is composed of 36 items that form eight different subscales: physical functioning, bodily pain, role limitations due to physical health problems, role limitations due to personal or emotional problems, mental health, social functioning, vitality (energy/fatigue), and general health perceptions. All raw scores of the survey are coded to a 0–100 scale in which higher scores indicate better HRQOL.
GHQ-12 is a 12-item self-report instrument intended to screen for current nonpsychotic psychiatric morbidity . The validity and reliability of GHQ-12 have been extensively evaluated [6, 9–11]. We scored the GHQ-12 using the standard GHQ method in which each item is scored as 0 (first two answer categories) or 1 (latter two answer categories), giving a maximum total score of 12. A 3/4 threshold has given the best estimate of minor psychiatric morbidity  and is recommended for screening studies . Individuals whose scores are above the threshold are considered as probable cases. This threshold (caseness) was used while comparing patients and control group.
BDI is a 21-item screening questionnaire used to detect current depressive symptoms . It has been indicated to have a good reliability and validity . It is scored by adding up the ratings given to each of the 21 items. Each item can have scores from 0 to 3 with total scores ranging from 0 to 63. A cutoff score of 9/10 was used in this study, and the participants with total scores of 10 or greater were considered depressed.
CAGE  is a widely used questionnaire for detecting alcohol abuse/dependence. It consists of four questions. CAGE questions have been proven to be superior to other screening instruments for detecting patients’ lifetime and current alcohol abuse and/or dependence in primary care settings when the most commonly used cutoff of 1/2 was used .
Determination of serum hormone levels
Blood samples were collected between 0800 and 1000 hours. Samples were centrifuged and serum testosterone was measured in the laboratory of Helsinki University Hospital . The entire testicular endocrine and sperm quality analysis of this same patient cohort has been published separately . None of the controls, one survivor after treatment without cyclophosphamide or irradiation, and 15 survivors after testicular irradiation reported current testosterone substitution.
Descriptive statistics for demographic variables and caseness in BDI, GHQ-12, and CAGE with appointed cutoff scores were calculated and compared between survivors and controls. These categorical variables were analyzed using the Chi-square test. Differences in mean scores of subscales of RAND-36 between the survivors, population norm values, and the controls were compared using one-sided Student’s t test. Differences in RAND-36 between three different treatment groups (groups 1–3) were analyzed using ANOVA, and differences in GHQ-12, BDI, and CAGE using Fisher’s exact test. Pairwise comparison between different treatment modalities (groups 1–3) was done using one-sided Student’s t test for RAND-36 subscales. Those main outcome variables with which we found statistical difference either between the survivors and the control group or between the different treatment modality groups (RAND-36 and GHQ-12) were analyzed more closely in a hierarchical regression model. The association between RAND-36 subscales and GHQ-12, and the following variables were first analyzed with univariate analyses: relationship (married/cohabiting or single/divorced), children in the household (yes/no), education (high school graduation yes/no), employment (employed/student or unemployed), age at the diagnosis, time since diagnosis, relapse, central nervous system (CNS) irradiation, treatment modality (groups 1–3), and testosterone level (<10 or ≥10 nmol/l). Those variables with p < 0.1 were then included to a hierarchical regression model in the order of significance in univariate analyses. A statistical significance level of p < 0.05 was used in all of the analyses unless otherwise mentioned. All statistical analyses were performed using SURVO MM Version 3.17.
Characteristics of the 52 survivors. Nine out of 52 (17%) had been treated also for a relapse. Two patients had got bone marrow transplantation
Age at study (years)
Age at ALL diagnosis (years)
Follow-up time (years)
Cranial irradiation (Gy)
Spinal irradiation (Gy)
Testicular irradiation (Gy)
Sociodemographic variables of all participants
ALL survivors, n = 52
Control group, n = 56
Children in the household
High school graduatea
Health-related quality of life and mental well-being outcomes among male long-term survivors of childhood acute lymphoblastic leukemia (ALL), a control group, and population norms measured by RAND-36, GHQ-12 (General Health Questionnaire), The Beck Depression Inventory (BDI), and (CAGE, a test for screening alcoholism)
12/52 (9.6 %)
8/56 (14.3 %)
4/52 (7.7 %)
2/55 (3.6 %)
22/51 (43.1 %)
19/56 (33.9 %)
Health-related quality of life and mental well-being outcomes RAND-36, GHQ-12 (General Health Questionnaire), The Beck Depression Inventory (BDI), and (CAGE, a test for screening alcoholism) among male long-term survivors of childhood acute lymphoblastic leukemia (ALL) in three different treatment groups defined by the differences in potential gonadotoxicity
Treatment group 1
Treatment group 2
Treatment group 3
Three group comparison
0/16 (0.0 %)
4/12 (33.3 %)
6/18 (33.5 %)
1/16 (6.3 %)
1/12 (8.3 %)
1/18 (5.6 %)
6/16 (37.5 %)
4/12 (33.3 %)
10/18 (55.6 %)
There was no significant difference between ALL survivors and the control group on psychological outcome measured by GHQ-12 (p = 0.240), BDI (p = 0.362), or CAGE (p = 0.328). Significantly fewer probable cases in GHQ-12 was found with the survivor group with no cyclophosphamide nor testicular irradiation (group 1) than with the groups with more intensive treatment (groups 2 and 3) (p = 0.016). There was no difference in the results of BDI and CAGE between different treatment groups (Table 4).
The results of the hierarchical regression analysis of the variables explaining the variation in the quality of life and mental well-being scores of the cancer survivors. Variables are included into hierarchical regression model based on the order of significance in univariate analyses. Only those variables with p value <0.1 in univariate analyses were taken in to the model. R2 (coefficient of determination) defines the percentage by which the added variable contributes to the explanation of the variation. β (beta standardized regression coefficient) defines the direction of the effect
Variables explaining the variation in the test scores
Age at diagnosis
Time since diagnosis
Total R2 (%)
On RAND-36, only 8 % of the variation in physical functioning scores of the survivors was explained by all the studied variables, and none of them showed to be statistically significant. Eleven percent of the variation in the role/physical scores was explained. The level of education had the strongest effect on the studied scores although none of the studied variables were statistically significant explanators. In role/emotional subscale, 23 % of the variation of scores was explained by the model. More gonadotoxic treatment modality had a negative effect, and longer time since diagnosis and a higher testosterone level had a positive effect, but none of the variables reached statistically significant explanatory level. On the social functioning subscale, the whole model explained 16 % of the variation in QOL scores, and major positive effect was shown by higher testosterone level (p = 0.064) and negative effect by longer time since diagnosis. Of variation in emotional well-being scores, the model explained 30 %. Most of this was due to age at diagnosis (older age causing more negative effect), even though it did not quite reach the level of significance (p = 0.05). Employment for its part had a positive effect on this subscale. Almost 50 % of the variation on the vitality subscale scores was explained by the studied variables. Here, the key effectors were age at diagnosis (p = 0.003) and time since diagnosis. The latter, however, did not reach statistical significance contrary to treatment modality (p = 0.040). Higher age at diagnosis and more intensive treatment were in association with worse QOL scores. Sixteen percent of the variation in the scores for bodily pain was explained by the model and a major factor associated with better scores was higher education (p = 0.049). On the general health subscale, the model explained 28 % of the variation, and the positive effect of higher education reached statistical significance (p = 0.024). The variation in the GHQ-12 scores was explained by up to 18 % with the studied variables, and the effect of treatment modality was statistically significant (p = 0.035). The coefficients of determination and β-standardized regression coefficients are shown in Table 5.
Our study showed that compared to population norms, male long-term ALL survivors have slightly decreased QOL, which could not be explained with psychiatric morbidity. Mental well-being of the survivors was in keeping with those of the control group in general, but the survivors with more gonadotoxic treatment modalities including alkylating agent cyclophosphamide and/or testicular irradiation (groups 2 and 3) seemed to have more nonpsychotic psychiatric morbidity than other survivors. In order to avoid the risk that the bigger employment rate in our control group could bias the QOL results, also another control group was used which comprised normative values for the Finnish standard population . A difference in the QOL scores of our control group and the standard population was actually found. However, the entire survivor group showed lowered QOL in physical health-related subscales compared with both of the reference groups. In other subscales of the QOL, the entire survivor group did not differ from a standard population.
When we compared the effects of different treatment modalities on the QOL scores, treatment with cyclophosphamide with or without testicular irradiation (groups 2 and 3) was identified as a factor associated with lowered QOL. We have previously shown in this same patient cohort that survivors treated with ≤10 g/m2 cyclophosphamide without testicular irradiation had decreased testosterone levels, but their fertility remained normal, while survivors with testicular irradiation were sterile and required testosterone substitution. The survivors treated without cyclophosphamide or testicular irradiation did not differ from controls and showed normal testicular endocrine function and fertility . Thus, we hypothesized that the increased risk for lowered QOL with the survivors who had received cyclophosphamide treatment or testicular irradiation could be related to a decreased testicular endocrine function. In hierarchical regression analysis, we analyzed the effect of gonadotoxicity on QOL and mental well-being not only by testosterone level but also by defining the effect of treatment group. This way, we analyzed the effects of cranial irradiation, testicular irradiation, and cumulative dose of cyclophosphamide all together. Patients without cyclophosphamide and testicular irradiation (group 1) were, however, also less likely to have received cranial irradiation than the other survivor groups (44 vs. >90 %), which could at least partly explain the difference between these treatment groups. In the earlier work of this same cohort, cranial irradiation did not have any significant effect on gonadotropin and testosterone levels or on sperm counts . The association between poorer QOL and CNSRT has previously been shown in other studies [3, 4]. However, in those analyses, the direct gonadotoxicity has not been taken into the model. Consequently, we did regression analyses to evaluate independent effects of different variables, including CNSRT, as well as to find out how much the variation in the QOL or GHQ scores actually was explained by the variables of our interest.
In regression analysis, we found that the younger age at the time of the diagnosis seemed to protect the survivor from the negative effects of the cancer and its treatment, and the older the patients were at the time of the diagnosis the more their vitality was impaired. There was also a trend for younger age at diagnosis to improve the emotional well-being of the survivors. As expected, more gonadotoxic treatment modality had some effect on the variability of the QOL and mental well-being. A significant effect was seen in vitality scores of QOL and in the nonpsychotic psychiatric morbidity scores. Higher educational level seemed to improve general health and bodily pain scores in regression analysis, but this can be interpreted also the other way around: the better the general health of the survivors and less pain they suffer, the more they have a capability of educating themselves.
A surprising finding in the regression analyses was that none of the evaluated variables explained the variability of the scores in physical functioning or role/physical subscale which were the areas on which the entire survivor group did worse than the normative population. Furthermore, unlike we had expected, the CNS irradiation of the survivors had no independent explanatory effect on the variability of the QOL or psychiatric morbidity scores. Also, the higher testosterone level showed only a minor trend for positive effect on the social functioning subscale, explaining 6.8 % of the variability of this subscale. Its effect was even more modest in other subscales. The amount of CNS irradiation and survivors’ testosterone levels are of course not equally distributed between the specific treatment groups, and thus, the combination of treatments could partly explain the significant effect observed between these groups. However, we could conclude that no single factor in the treatment of ALL had a major role in determining the variability of the later QOL or psychiatric morbidity scores of the survivors. An important player seemed to be the whole treatment protocol and the several adverse effects of all the treatments together.
Most of the studies evaluating the QOL of childhood cancer survivors have found survivors to have better or clinically comparable QOL compared with controls [1–3, 18]. Only few studies have evaluated the QOL of ALL survivors separately [1–4, 19]. In our study, the QOL of the survivors was slightly declined, but, however, six out of eight RAND subscales showed no statistical difference compared with population norms illustrating that the QOL of the survivors is fairly good.
Risk for depressive symptoms, nonpsychotic psychiatric morbidity, or alcohol abuse/dependence was not increased among the whole population of the ALL survivors. However, survivors with cyclophosphamide treatment or testicular irradiation had an increased risk for nonpsychotic psychiatric morbidity. In previous studies concerning mental well-being of leukemia survivors, results have been somewhat inconsistent [20–22]. In all the studies, the majority of the survivors were, however, psychologically healthy. Our positive findings regarding low alcohol consumption among survivors was in line with the study of Frobisher et al. .
In the present study, the proportion of high school graduates or the employment rate of survivors did not differ from those of the normal population. This of course does not exclude the possibility that the survivors still have some cognitive problems but reflects that they have a considerably good academic performance. In earlier studies, CNSRT has been shown to have an adverse effect on cognitive functioning  and scholastic achievement [25–28], and the proportion of employed cancer survivors has been lower than in the comparison group. However, in studies where employment of male and female survivors was evaluated separately, the differences between male survivors and the controls were smaller than the difference for female survivors and their controls [26–28]. The present observation further supports the finding that there is a gender difference in the effect of ALL treatment on scholastic achievements. Consistent with earlier studies [1, 26, 27], we found the ALL survivors to be less likely to be married/cohabiting.
We used the entire cohort of all male long-term survivors, who were diagnosed with ALL in childhood and treated at Helsinki University Central Hospital during a 25-year period, with a good response rate. Thus, our survivor group was very unselected and homogenous considering the diagnosis and gender aspects. Treatment modalities, however, have changed during the time period evaluated, and as we could see, the treatment modality affects the QOL scores of the survivors. The cohort in the study was small considering the heterogeneous treatment the survivors had received and multiple variables studied. It is possible that some variables did not reach the statistical significance because of the small sample size. Thus, in the future, it would be interesting to confirm the results with wider survivor cohort. Our control group was recruited from occupational health services which could lead to bias. Mild decrease of QOL of the survivors was seen also when compared with the normative values for the age- and gender-relevant Finnish population, but on general, the survivor group seemed to have close to normal QOL when scrutinizing the actual scores of RAND-36. The screening tests we used to evaluate QOL and psychological outcome are all globally used and their psychometric properties have been widely tested [5–16].
In conclusion, the psychological well-being of survivors was proven to be good, although more gonadotoxic treatment modalities (groups 2 and 3) led to a greater risk for psychiatric morbidity and to a decrease in the scores for vitality in RAND-36. The studied variables explained poorly the variability in QOL scores for physical and social subscales. However, one third of the variability in scores for emotional well-being and vitality was covered by these variables. In the subscales, older age at diagnosis as well as more gonadotoxic treatment modality and lower educational level best explained the impairment of the scores. Further studies using patient interviews would probably point out issues concerning the QOL and psychosocial coping of ALL survivors, which may not emerge in these screening studies
We thank Marjut Grainger for statistical assistance. This work was funded by the Foundation of Nona and Kullervo Väre, Children’s Cancer Trust in memoriam of Emilia Heinonen, The National Graduate School of Clinical Investigation, and The Foundation of Wilho Kyttä. Study sponsors had no involvement in the study.
Conflict of interest
The authors have nothing to disclose.