Quality of Life in Long-Term Survivors of CNS Tumors of Childhood and Adolescence By Eliot N. Mostow, Julianne Byrne, Roger R.Connelly, and John J. Mulvihill Clinical reports of small numbers of pediatric brain tumor patients observed for brief periods suggest that long-term survivors continue to have major handicaps into adulthood. To quantify these late effects we interviewed 342 adults (or their proxies) who had CNS tumors diagnosed before the age of 20 between 1945 and 1974, survived at least 5 years, and reached 21 years of age. Survivors were 32 years old on average at follow-up. When compared with 479 matched siblings as controls, CNS tumor survivors were more likely to have died or to have become mentally incompetent sometime during the follow-up period. They were more likely to be at risk for such adverse outcomes as unemployment (odds ratio [OR], 10.8; 95%
A
confidence interval [CI], 4.6 to 25.7), to have a health condition that affected their ability to work (OR, 5.9; CI, 3.7 to 9.4), to be unable to drive (OR, 28.8; CI, 6.9 to 119.9), or to describe their current health as poor (OR, 7.8; CI, 1.7 to 35.7). Unfavorable outcomes were more frequent in male survivors than in females, in those with supratentorial tumors compared with infratentorial ones, and in those who received radiation therapy. As clinicians consider improving therapies, they should anticipate late effects, such as those we observed, and attempt to target subgroups for interventions that may improve subsequent quality of life. J Clin Oncol 9:592-599, 1991. This is a US government work. There are no restrictions on its use.
FTER LEUKEMIA, brain and CNS tumors are the leading cause of cancer mortality in children younger than 15 years old.' Survival rates after diagnosis of pediatric CNS tumors have increased over the last 2 decades: 5-year relative survival in white patients younger than age 15 improved from 35% from 1960 to 1963 to 56%
increasing numbers of cancer survivors. As part of a multifaceted investigation of late effects of childhood and adolescent cancer, we studied what may be the largest cohort of long-term survivors of pediatric CNS tumors yet assembled to evaluate quality of life into adulthood.
from 1979 to 1984.2 But survival alone does not indicate the quality of the life that is lengthened by successful therapy. Although quality of life has a clear intuitive
MATERIALS AND METHODS
meaning, it is defined differently in many studies that attempt its assessment."
Nevertheless, clini-
cal reports based on relatively small numbers of patients suggest that quality of life is impaired in survivors of pediatric and adolescent CNS tumors.6- 11 Few studies followed these patients into adulthood. Further description of the adult lives of
these long-term survivors may permit identification of subgroups for early intervention and ulti-
mately improve the quality of life for the ever-
From the Cancer ControlApplications, ClinicalEpidemiology and Biometry Branches, National Cancer Institute, National Institutesof Health, Bethesda, MD. SubmittedApril 4, 1990; accepted October 16, 1990. Presentedin part at the Meeting of the American Association for CancerResearch, May 1989. No reprintsavailable. This is a US government work. There are no restrictionson its use. 0732-183X/91/0904-0007$0.00/0
592
The National Cancer Institute collaborated with five cancer registries, and as part of a large historical cohort study,"•" we interviewed 342 adults (or their proxies) who had CNS tumors diagnosed before the age of 20 years between 1945 and 1974, survived at least 5 years, and reached 21 years of age by December 31, 1979. Eligible survivors were told that they were contacted because they had a serious childhood illness that required hospitalization. Up to two sibling controls were selected (N = 479) among those born before 1961 and at least 19 years old by December 31, 1979; preference was given to full siblings, same sex, and closest dates of birth. Medical records were abstracted for details of diagnosis, treatment, physical examination, medical history, and death certificate information. Subjects were interviewed in person when possible; the interview covered medical and reproductive experiences and also included questions on functioning. Proxy interviews were conducted for 81 (24%) survivors and 26 (5%) controls. The potential bias of proxy compared with direct interviews was addressed with separate analyses. Quality of life was defined in several different ways. Among the questions asked were "Have you ever been unable to take a job, had to change jobs, or had to stop working for at least 3 months because of any health condition (not including pregnancy)?"; "Could you tell me if any of the following conditions ever affected your ability to work: sight problems, hearing problems, or emotional problems?"; "How much help do you need to drive a car?";
Journalof Clinical Oncology, Vol 9, No 4 (April), 1991: pp 592-599
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
593
LONG-TERM SURVIVORS OF PEDIATRIC CNS TUMORS "Compared to other persons your age would you say that your health is excellent, good, fair, or poor?" Interview responses were used to determine the following: employment status (never employed, considered as unemployed), the need to change jobs or stop working because of a health condition (including specific questions about emotional, sight, and hearing problems), the ability to drive a car without assistance, income level in most recent job, selfdescriptions of health with positive or negative adjectives, marital status, highest level of school attended, and the need for proxy interview because of mental incompetence. Histologic confirmation of a second primary cancer diagnosed at least 5 years after the first was obtained from medical records. The risk of death was assessed with a hazards ratio (HR) generated from a proportional hazards linear regression model stratified by family (PROC PHGLM14; data did not violate the proportional hazards assumptions) and KaplanMeier survival estimates. The risk of a second tumor in survivors or a first tumor in controls was assessed similarly. Risk for other outcomes was measured with odds ratios (OR) generated from a conditional logistic regression model (PROC MCSTRAT)15 to compare survivors with their matched sibling controls. Sex differences were assessed by comparing male or female survivors with their matched controls, regardless of sex, and similar stratified analyses were performed to assess tumor site (supratentorial v infratentorial) and a history of radiation therapy compared with no radiotherapy.16 When matched analyses were impractical (because of small numbers) or inappropriate (age at diagnosis has no meaning in controls), odds ratios, X2 analyses, and t-tests were calculated."7 The accepted level of statistical significance was alpha less than .05 for two-tailed comparisons of means or 95% confidence intervals (CIs). RESULTS The average age at CNS tumor diagnosis was 11.3 years; sibling controls were an average 12.2 years old at the survivor's diagnosis. Survivors were 32.0 years old on average at last follow-up (interview or death) and controls were 33.2. Survivors were about two inches shorter (P < .01) and weighed 9 lb less (P < .01) than controls. Only 4% of subjects were nonwhite. The 342 tumors were astrocytomas (56%); other gliomas (8%); ependymoma (8%); medulloblastomas (7%); and oligodendrogliomas, meningiomas, hemangioblastomas, and others (21%). Tumor sites were supratentorial (18%); infratentorial (39%); cranial, not otherwise specified (36%); and extracranial (7%). Medical record abstracts did not permit reliable assessment of the degree of histologic malignancy, size of tumors, or the severity of symptoms at diagnosis. Surgery was all or part of their treatment for 94% of survivors; other treatments were radiother-
apy (54%), radiation plus chemotherapy (2%), and chemotherapy only (1%). Of those who received radiotherapy, the mean total dose was 4,940 cGy. Use of radiotherapy was not related to the age at diagnosis, but was more likely in cases diagnosed in recent years (x 2 for trend; P = .01). Deaths and Second Cancers Death occurred before the interview in 47 survivors and seven controls. As adults, survivors were more likely to die than controls at any given age (HR, 14.0; CI, 4.9 to 39.6) (Fig 1). Among survivors, those who received radiation therapy were more likely to die than those who did not (HR, 5.0; CI, 2.3 to 10.7). After review of death certificates, only three deaths seemed unrelated to the history of cancer (one death due to a car accident with alcoholism listed as "other condition," one to being hit by a car as a pedestrian, and one to a stab wound not considered a suicide); the remaining 44 deaths were directly due to the primary tumor, its treatment, or a recurrence. Of the seven dead controls, one died from a brain tumor (Table 1 and text below); other causes of death in controls included accidental fall, skull fractures, pulmonary edema, and suicide (amobarbital poisoning). Second primary tumors were diagnosed more than 5 years after the first tumor diagnosis in five survivors, and four controls had cancer in adulthood (Table 1). As adults, survivors were no more likely to develop a second tumor than controls were to develop a first tumor at any given age (HR, 2.6; CI, 0.5 to 14.2). The boy with a hemangioblas-
e a
iE mi U.
cn
Age at Death (yr) Fig 1. Survival after CNS tumors (Kaplan-Meier curves). To be included in the study, subjects had to have survived at least 5 years and have reached age 21 years. Average follow-up time was 13 years after entry. (---) Sibling controls, (-) CNS tumor survivors (I, 95% confidence interval).
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
594
MOSTOW ET AL Table 1. Second Tumors in Survivors of Childhood and Adolescent CNS Tumors and Any Tumor in Their Sibling Controls Diagnosis First Tumor Diagnosis Type (site)
Year
Age (years)
Hemangioblastomat
1969
15
Female
Optic glioma
1953
3
Female Female
Astrocytoma (temporal lobe) Astrocytoma (spinal cord)
1954 1956
5 19
Female
Ependymoma
1956
6
Seminoma Renal cell carcinoma
1981 1976
35 21
Astrocytoma Hemangioblastoma
1976 1965
31 20
Sex Survivors Male
Sibling controls Male Female Male Male
Interval in Years*
Second Tumor
Renal cell carcinoma Pheochromocytoma Von Hippel-Lindau syndrome Neurofibrosarcoma (thigh) Neurofibromatosis Papillary thyroid carcinoma Breast (infiltrating ductal) Li-Fraumeni syndrome Oligodendroglioma
9 9 10 20 24 9
Survivor Oligodendroglioma Astrocytoma Von Hippel-Lindau syndrome Glioma Hemangioblastomat Von Hippel-Lindau syndrome
*Years from initial diagnosis to diagnosis of second cancer. tSame family as survivor noted above. Von Hippel-Lindau syndrome (see text).
toma as his first tumor had von Hippel-Lindau syndrome, which may include angiomas, cysts, pheochromocytomas, and tumors of the liver, pancreas, and kidneys (Table 1). The survivor with the optic glioma had neurofibromatosis type 1 with cafe au lait spots since birth and multiple neurofibrosarcomas on the left thigh as her second tumor. None of these five survivors received chemotherapy for their primary tumor. Only the survivor of the optic glioma received radiotherapy for her primary tumor, and she was the only one of the five who was dead at the time of interview. Among the four controls with confirmed tumors, the control with a hemangioblastoma was from the family with von Hippel-Lindau syndrome, which is transmitted with autosomal dominant inheritance. Interview Responses
Survivors differed from their siblings in many measures of adult functioning (Table 2). The greatest relative difference was in hearing problems that affected the survivors' ability to work or caused them to stop working. Employment difficulties were considerable; survivors were 11 times more likely never to have had a job and were at increased risk for a health condition that affected their ability to work or caused them to stop working. Although only 13% of the survivors could
not drive, inability to drive was almost 30 times
more likely for survivors than controls. Survivors rated their health as fair or poor more often than controls; they were also more likely never to have married. The mean highest grade of education Table 2. Quality-of-Life Measures in CNS Tumor Survivors Compared With Sibling Controls
Quality-of-Life Measure
Employment Never employed Health condition led to job change/stop working Conditions affecting work Emotional problems Sight problems Hearing problems Cannot drive a car Most recent income under $15,000 per year Health perception Excellent Good Fair Poor Never married
Crude Percentage of Survivors
Odds* Ratio
95% Confidence Interval*
15
10.8
4.6-25.7
37
5.9
12 24 8 13
2.4 7.7 45.2 28.8
76
3.3
2.3-4.9
38 42 15 5 42
0.6 1.0 3.8 7.8 3.6
0.4-0.8 0.7-35.7 1.9-7.5 1.7-35.7 2.5-5.3
3.7-9.4 1.4-4.0 4.2-14.0 6.1-333 6.9-119
*Derived from a stratified analysis that controlled for family; compares survivors with controls.
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
595
LONG-TERM SURVIVORS OF PEDIATRIC CNS TUMORS completed was 12 for survivors and 13 for siblings
(P < .01).
Table 4. Quality-of-Life Measures in CNS Tumor Survivors Compared With Sibling Controls by Tumor Site Crude
Differences by Sex, Tumor Type, and Treatment Male survivors differed from females on several measures of adult functioning (Table 3). Males
were at much greater risk of unemployment, having a health condition that affected employability, having emotional problems, or perceiving their health as fair or poor rather than excellent or good. Females were more likely to be earning less than $15,000 at their most recent job. Hearing problems and difficulty driving a car seemed to occur with equal frequency in both sexes. For instance, 9% of the male survivors versus 7% of the female survivors said that hearing problems had led to job difficulties; the corresponding percentages for driving a car were 16% and 19%, respectively. Sex-specific rates could not be estimated for hearing problems because none of the 11 female survivors had a sibling with hearing problems; a similar situation existed for male Table 3. Quality-of-Life Measures in CNS Tumor Survivors Compared to Sibling Controls by Sex
Quality-of-Life Measure
Never employed Males Females Health condition led to job change/stop working Males Females Emotional problems Males Females Sight problems Males Females Most recent income < $15,000 Males Females Health perceived as fair or poor compared to excellent or good Males Females Never married Males Females
Crude Percentage of Survivors
Odds* Ratio
95% Confidence Interval*
14 17
31.6 6.7
4.2-237.7 2.5-18.0
42 30
11.6 3.2
5.3-25.7 1.8-6.0
13 11
3.2 1.7
1.5-6.8 0.8-3.7
28 18
19.4 4.2
6.0-63.3 2.0-8.7
65 90
2.1 9.3
1.3-3.3 4.0-21.9
21 17
15.5 2.8
3.6-66.8 1.3-6.1
56 24
6.4 1.5
3.8-10.9 0.8-2.7
*Derived from a stratified analysis that controlled for family; compares survivors with controls.
Quality-of-Life Measure
Never employed Supratentorial Infratentorial Health condition led to job change/stop working Supratentorial Infratentorial Emotional problems Supratentorial Infratentorial Sight problems Supratentorial Infratentorial Most recent income < $15,000 Suprotentorial Infratentorial Health perceived as fair or poor compared to excellent or good Supratentorial Infratentorial Never married Supratentorial Infratentorial
Percentage of Survivors
Odds* Ratio
95% Confidence Interval*
25 8
17.7 5.0
2.3-138 1.3-18.7
53 28
17.8 4.3
4.2-75.9 2.0-9.3
23 10
19.8 2.5
2.6-153 0.9-7.1
33 18
25.7 9.3
3.4-195 2.7-31.8
79 71
7.1 2.3
2.4-21.1 1.3-3.9
11 21
6.0 4.8
0.6-57.7 1.9-12.2
34 40
5.8 3.0
1.8-17.8 1.7-5.5
*Derived from a stratified analysis that controlled for family; compares survivors with controls.
survivors and the ability to drive a car. Male survivors were more likely never to have married than female survivors (P < .001). Both tumor site and treatment were associated with altered quality-of-life measures (Tables 4 and 5). Survivors of supratentorial tumors and those who received radiation therapy were generally at greater risk for adverse outcomes than survivors of infratentorial tumors and those who received no radiation when each was compared with controls. Although sight problems were most likely among supratentorial tumor survivors compared with controls, the greatest relative difference between supratentorial and infratentorial tumor survivors was for emotional problems. Survivors who were treated with radiotherapy seemed to have poorer life experiences in most of the measures we looked at (Table 5), unemployment was nearly four times as frequent, and ratings of poor health status were very much more frequent in these survivors compared with those who did not receive radiotherapy. Hearing problems occurred in 12% of survi-
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
596
MOSTOW ET AL
Table 5. Quality-of-Life Measures in CNS Tumor Survivors Compared with Sibling Controls by Treatment
Quality-of-Life Measure
Never employed Radiotherapy No radiotherapy Health condition led to job change/stop working Radiotherapy No radiotherapy Emotional problems Radiotherapy No radiotherapy Sight problems Radiotherapy No radiotherapy Most recent income < $15,000 Radiotherapy No radiotherapy Health perceived as fair or poor compared to excellent or good Radiotherapy No radiotherapy Never married Radiotherapy No radiotherapy
Crude Percentage of Survivors
Odds* Ratio
95% Confidence Interval*
21 8
16.0 4.5
4.8-52.8 1.2-17.2
47 23
6.8 4.7
3.7-12.5 2.0-10.9
16 8
3.0 1.6
1.5-5.9 0.7-3.8
30 15
9.3 5.1
4.4-19.8 1.9-13.9
83 69
5.6 2.1
3.0-10.3 1.3-3.6
30 10
13.1 1.0
4.6-37.2 0.3-2.8
47 35
4.9 2.5
2.8-8.5 1.4-4.3
*Derived from a stratified analysis controlled for family; compares survivors with controls.
vors treated with radiotherapy compared with 4% of those who were not, and inability to drive was increased twofold (24% v 12%, respectively). The mean highest grade of education was grade 11 for those survivors who received radiation and 13 for those who did not (P < .01). No histologic diagnosis was consistently associated with increased risk for any of the outcomes when survivors were compared with controls in matched analyses, although matched analyses stratified by histology were less meaningful because most survivors had astrocytomas. Among survivors, however, medulloblastoma survivors were most likely to have had every one of the unfavorable outcomes listed in Table 4. Survivors diagnosed at younger ages were at increased risk for having never been employed, having a lower final level of education achieved, or never marrying (each tested with XZ for trend; P < .01). Age at diagnosis had no bearing on other measures.
Mental Incompetence Mental incompetence was given as the reason for a proxy interview by 28 survivors and two controls. Considering only subjects who were alive at the time of interview, survivors were more likely to be mentally incompetent than matched sibling controls (OR, 16.2; CI, 3.8 to 69.8). Compared with survivors with normal mental functioning, mentally incompetent survivors were younger at diagnosis (mean ages, 8.6 and 11.1 years, respectively; P = .02), more likely to have supratentorial tumors (OR, 8.7; CI, 3.0 to 24.7), and more likely to have been treated with radiotherapy (OR, 5.3; CI, 2.1 to 13.2). Male and female survivors were equally likely to be mentally incompetent (OR, 1.8; CI, 0.8 to 4.1). DISCUSSION Survival for 5 years after diagnosis was a requirement for inclusion in the study. Thus, this cohort was not representative of all incident pediatric and adolescent CNS tumors. Our 342 survivors comprised about 2% of all incident CNS tumors in the United States diagnosed between 1945 and 1975 and at ages younger than 20 years, assuming 1,100 new cases annually and 45% with 5-year survival."8 Astrocytoma was the leading cell type in this cohort of survivors, as it is in incidence series. 19 Survivors diagnosed more recently may already be benefiting from improved imaging techniques that detect tumors earlier, as well as surgical procedures and therapeutic regimens designed to reduce late morbidity. This study may serve as a historical comparison when patients diagnosed today become long-term survivors. In our study, matched analyses tended to be less likely to result in statistically significant risk estimates than unmatched analyses, but matching survivors to their specific siblings was consistent with the original study design. To the extent that any adult's quality of life could have been influenced by a variety of factors in childhood and adolescence (eg, neighborhood, schools attended, and parental influences-their occupations, marital status, education level, and attitudes toward their children), sibling matching assured that these factors were similar in survivors and controls. When proxy interviews were excluded from analyses, risk estimates for adverse outcomes were consistently lower, though conclusions of statistical significance were never changed. In other
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
597
LONG-TERM SURVIVORS OF PEDIATRIC CNS TUMORS
words, survivors interviewed by proxy were more likely to have adverse outcomes. We included proxy interviews to reflect the experience of all pediatric CNS tumor survivors and to minimize the selection bias from analyzing only individuals doing well enough to be interviewed. To the degree that siblings may be affected by the cancer diagnosis in a family member, they may not be the best control group for psychosocial outcomes. However, population-based data are not available that could be matched to our group of cancer survivors. Comparing survivors with their sibling(s) may be less than ideal, but this comparison is often unavoidable among parents, teachers, and close community members. Thus, sibling controls may be more useful for family counseling and similar interventions. Since OR express relative odds, a very low risk in controls can be just as important as high risk in survivors. For example, only two of the 453 sibling controls were unable to drive. Thus, the OR was high even though the majority of survivors were able to drive. We have presented OR that express risks compared with matched sibling controls because these are the comparisons that will be made by parents, survivors, siblings, and teachers. The survival rates beyond age 20 years (Fig 1) suggest that, as adults, long-term survivors of pediatric and adolescent CNS tumors have a sustained excess risk of death, compared with their siblings. Other authors who have followed CNS survivors for long periods 20' 21 noted that death rates seemed to diminish over time. Although we lack details on severity of the disease at diagnosis, the reported causes of death suggest continued complications owing to the original tumor, its treatment, or both. Again, the increased risk of death was associated with radiation therapy. The uncertain future after a tumor diagnosis-termed the Damocles syndrome-is a burden that may be carried forever and includes concerns about recurrence of disease and/or second primary tumors.22 Our survivors showed no excess of adultonset tumors, and two of the five observed were explained by mendelian disorders complicated by multiple neoplasms, namely the von HippelLandau syndrome and Recklinghausen neurofibromatosis. Familial clustering was also represented by the family with Li-Fraumeni syndrome (Table 1). The apparent excess of cancer among controls in our study, four cases in 479 siblings, is explained
in part by familial clustering, which may account for some of the excess cancer risk seen in siblings of children with CNS tumors.23'24 Our findings on societal functioning show that when these survivors of pediatric and adolescent CNS tumors were compared with their siblings, the survivors were more likely to have a number of adverse outcomes as adults. Still, 85% of the survivors had some employment, 83% could drive a car, and 80% described their health as excellent or good (Table 2). Stable employment is one measure of security and social independence. Although jobs differ in physical and intellectual demands, survivors were at significantly greater risk for being permanently unemployed, consistent with prior reports. 25 A recent analysis of 40 adolescent non-CNS cancer survivors showed unemployment to be almost twice as likely in survivors than controls; furthermore, 7% reported discrimination in hiring based on their cancer history.26 Occupational therapists, parents, and others with the opportunity to help develop vocational skills might devote extra attention to recovering pediatric and adolescent CNS tumor patients to attenuate long-term disabilities. Male survivors had consistently greater risk for adverse outcomes than female survivors who were functioning almost as well as their siblings. It is unlikely that males had more severe disease and more aggressive treatment than females because the distribution of major histologic tumor types and treatments was similar in each sex. Psychosocial interactions between parents and affected children may have long-term effects; if female survivors were given more support, attention, or different treatment, this might have reduced their risk for adverse outcomes. Alternatively, sexspecific differences in brain composition and structure 27 could explain the female's superior ability to recover from CNS tumors. Improved attention to psychosocial aspects of care could reduce the risks for adverse outcomes in both sexes, but male survivors seemed to be in greater need of assistance. Supratentorial tumors are more likely to result in higher function deficits (eg, intellect, memory, emotion) than infratentorial tumors (mostly cerebellar), which tend to affect motor coordination.28 More severe disability has been noted for other survivors of pediatric supratentorial compared with infratentorial brain tumors. 29 However, infra-
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
598
MOSTOW ET AL
tentorial tumors can also affect higher cortical function since any intracranial space-occupying lesion can affect the entire brain. This may partially explain the fact that infratentorial survivors were also at significant risk for most of the same adverse outcomes as supratentorial survivors. Still, the greater impact of supratentorial tumors in this study might be because of more severe long-term effects on intellectual and emotional development. Pediatric brain tumor survivors who had motor impairment as their primary long-term deficit generally fared better than those with intellectual and psychologic-emotional sequelae.2 9 We should not attribute reduced quality of life simply to radiation therapy; radiotherapy in these CNS tumor survivors might simply indicate increased severity or reduced operability of the initial tumor. Although it is difficult to separate the effect of the tumor from the treatment, other studies have documented significant intellectual and social deficits in survivors who had received
opment. Physical disfigurement or handicap may also diminish a survivor's desire and ability to interact socially. Although our measures were crude surrogates for the highly subjective assessment of any individual's quality of life, our results suggest that there is considerable room to improve the quality of the future lives of current patients. The ideal goal of designing improvements in therapy should be a life similar to that expected for siblings and peers. Survivors and their families may take solace in knowing that the majority of individuals in our cohort were functioning well as adults. Parent self-help groups are an important part of families' social support systems,34 and they may use the quality-of-life measures presented here as the basis for discussions with their children and their health-care providers. Most importantly, identification of differences may permit interventions (eg, physical and occupational therapy and special education) to minimize possible late effects.
radiotherapy especially early in life.30 3- 3 However,
one recent study suggests that several factors other than radiotherapy contribute to diminished neurodevelopmental function."0 The effect of age at diagnosis may reflect the sensitivity of the developing brain to radiation. It could also represent impaired development of social skills at crucial points of psychosocial devel-
ACKNOWLEDGMENT The authors are grateful to Mary McAdams, IMS, for technical assistance, and acknowledge the assistance of the patients and staff of the following registries: the California State Department of Health Services, Yale University, Connecticut Tumor Registry, University of Kansas, University of Iowa, and The University of Texas MD Anderson Cancer Center.
REFERENCES 1. Young JL, Ries LG, Silverberg E, et al: Cancer incidence, survival, and mortality for children younger than age 15 years. Cancer 58:598-602, 1986 2. US Department of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute: 1987 Annual Cancer Statistics Review. NIH Publication No. 88-2789 3. Mor V, Guadagnoli E: Quality of life measurement: A psychometric tower of Babel. J Clin Epidemiol 41:10551058, 1988 4. Calman KC: Definitions and dimensions of quality of life, in Aronson NK and Beckman J (eds): Quality of Life of Cancer Patients. New York, NY, Raven, 1987, pp 1-9 5. Donovan K, Sanson-Fisher RW, Redman S: Measuring quality of life in cancer patients. J Clin Oncol 7:959-968, 1989 6. Bamford FN, Morris Jones P, Pearson D, et al: Residual disabilities in children treated for intracranial space-occupying lesions. Cancer 37:1149-1151, 1976 7. Bordeaux JD, Dowell RE, Copeland DR, et al: A prospective study of neuropsychological sequelae in children with brain tumors. J Child Neurol 3:63-68, 1988 8. Ellenberg L, McComb JG, Siegel SE, et al: Factors
affecting intellectual outcome in pediatric brain tumor patients. Neurosurg 21:638-644, 1987 9. Lebaron S, Zeltzer PM, Zeltzer LK, et al: Assessment of quality of survival in children with medulloblastoma and cerebellar astrocytoma. Cancer 62:1215-1222, 1988 10. Mulhern RK, Horowitz ME, Kovnar EH, et al: Neurodevelopmental status of infants and young children treated for brain tumors with preirradiation chemotherapy. J Clin Oncol 7:1660-1666, 1989 11. Packer RJ, Meadows AT, Rorke LB, et al: Long-term sequelae of cancer treatment on the central nervous system in childhood. Med Pediatr Oncol 15:241-253, 1987 12. Mulvihill JJ, Myers MH, Connelly RR, et al: Cancer in offspring of long-term survivors of childhood and adolescent cancer. Lancet 2:813-817, 1987 13. Byrne J, Mulvihill JJ, Myers MH, et al: Effects of treatment on fertility in long-term survivors of childhood or adolescent cancer. New Engl J Med 317:1315-1321, 1987 14. Harrell FE: The PHGLM procedure, in Hastings RP (ed): SUGI Supplemental Library User's Guide, Version 5 Edition. Cary, NC, SAS Institute, Inc, 1986, pp 437-466 15. Naessens JM, Offord KP, Scott WF, et al: The MCSTRAT procedure, in Hastings RP (ed): SUGI Supple-
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
599
LONG-TERM SURVIVORS OF PEDIATRIC CNS TUMORS mental Library User's Guide, Version 5 Edition. Cary, NC, SAS Institute, Inc, 1986, pp 307-328 16. Kleinbaum DG, Kupper LL, Morgensten H: Epidemiologic Research. Principles and Quantitative Methods. New York, NY, Van Nostrand Reinhold, 1982 17. SAS Institute Inc: SAS User's Guide: Statistics, Version 5 Edition. Cary, NC, SAS Institute, Inc, 1985, pp 403-432 18. Li FP: Cancers in children, in Schottenfeld D, Fraumeni JF Jr (eds): Cancer Epidemiology and Prevention. Philadelphia, PA, Saunders, 1982, pp 1012-1024 19. Duffner PK, Cohen ME, Freeman AI: Pediatric brain tumors: An overview. CA 35:287-301, 1985 20. Hawkins MM: Long-term survival and cure after childhood cancer. Arch Dis Child 64:789-807, 1989 21. Spunberg JJ, Chang CH, Goldman M, et al: Quality of long-term survival following irradiation for intracranial tumors in children under the age of two. Int J Radiat Oncol Biol Phys 17:727-736, 1981 22. Koocher GP, O'Malley JE: The Damocles Syndrome: Psychosocial Consequences of Surviving Childhood Cancer. New York, NY, McGraw-Hill, 1981 23. Miller RW: Deaths from childhood leukemia and solid tumors among twins and other sibs in the United States, 1960-67. JNCI 46:203-209, 1971 24. Draper GJ, Heaf MM, Kinnear-Wilson LM: Occurrence of childhood cancers among sibs and estimation of familial risks. J Med Genet 14:81-90, 1977 25. Li FP, Winston KR, Gimbere K: Follow-up of children with brain tumors. Cancer 54:135-138, 1984
26. Tebbi CK, Bromberg C, Piedmonte M: Long-term vocational adjustment of cancer patients diagnosed during adolescence. Cancer 63:213-218, 1989 27. Allen LS, Hines M, Shryne JE, et al: Two sexually dimorphic cell groups in the human brain. J Neurosci 9:497-506, 1989 28. Adams RD, Hochberg F, Webster HD: Neoplastic disease of the brain, in Isselbacher KJ, Adams RD, Braunwald E, et al (eds): Harrison's Principles of Internal Medicine. New York, NY, McGraw-Hill, 1980, pp 19511960 29. Lannering B, Marky I, Lundberg A, et al: Long-term sequelae after pediatric brain tumors: Their effect on disability and quality of life. Med Pediatr Oncol 18:304-310, 1990 30. Peckham VC, Meadows AT, Bartel N, et al: Educational late effects in long-term survivors of childhood acute lymphocytic leukemia. Pediatrics 81:127-133, 1988 31. Kun LE, Mulhern RK, Crisco JJ: Quality of life in children treated for intellectual outcome in pediatric brain tumor patients. Neurosurgery 21:638-644, 1987 32. Dowell RE, Copeland DR: Cerebral pathology and neuropsychological effects: Differential effects of cranial radiation-as a function of age. Am J Pediatr Hematol Oncol 9:68-72, 1987 33. Choux M, Lena G, Hassoun J: Prognosis and longterm follow-up in patients with medulloblastoma. Clin Neurosurg 30:246-277, 1983 34. Monaco GP: Parent self-help groups for the families of children with cancer. CA 38:169-175, 1988
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
A
confidence interval [CI], 4.6 to 25.7), to have a health condition that affected their ability to work (OR, 5.9; CI, 3.7 to 9.4), to be unable to drive (OR, 28.8; CI, 6.9 to 119.9), or to describe their current health as poor (OR, 7.8; CI, 1.7 to 35.7). Unfavorable outcomes were more frequent in male survivors than in females, in those with supratentorial tumors compared with infratentorial ones, and in those who received radiation therapy. As clinicians consider improving therapies, they should anticipate late effects, such as those we observed, and attempt to target subgroups for interventions that may improve subsequent quality of life. J Clin Oncol 9:592-599, 1991. This is a US government work. There are no restrictions on its use.
FTER LEUKEMIA, brain and CNS tumors are the leading cause of cancer mortality in children younger than 15 years old.' Survival rates after diagnosis of pediatric CNS tumors have increased over the last 2 decades: 5-year relative survival in white patients younger than age 15 improved from 35% from 1960 to 1963 to 56%
increasing numbers of cancer survivors. As part of a multifaceted investigation of late effects of childhood and adolescent cancer, we studied what may be the largest cohort of long-term survivors of pediatric CNS tumors yet assembled to evaluate quality of life into adulthood.
from 1979 to 1984.2 But survival alone does not indicate the quality of the life that is lengthened by successful therapy. Although quality of life has a clear intuitive
MATERIALS AND METHODS
meaning, it is defined differently in many studies that attempt its assessment."
Nevertheless, clini-
cal reports based on relatively small numbers of patients suggest that quality of life is impaired in survivors of pediatric and adolescent CNS tumors.6- 11 Few studies followed these patients into adulthood. Further description of the adult lives of
these long-term survivors may permit identification of subgroups for early intervention and ulti-
mately improve the quality of life for the ever-
From the Cancer ControlApplications, ClinicalEpidemiology and Biometry Branches, National Cancer Institute, National Institutesof Health, Bethesda, MD. SubmittedApril 4, 1990; accepted October 16, 1990. Presentedin part at the Meeting of the American Association for CancerResearch, May 1989. No reprintsavailable. This is a US government work. There are no restrictionson its use. 0732-183X/91/0904-0007$0.00/0
592
The National Cancer Institute collaborated with five cancer registries, and as part of a large historical cohort study,"•" we interviewed 342 adults (or their proxies) who had CNS tumors diagnosed before the age of 20 years between 1945 and 1974, survived at least 5 years, and reached 21 years of age by December 31, 1979. Eligible survivors were told that they were contacted because they had a serious childhood illness that required hospitalization. Up to two sibling controls were selected (N = 479) among those born before 1961 and at least 19 years old by December 31, 1979; preference was given to full siblings, same sex, and closest dates of birth. Medical records were abstracted for details of diagnosis, treatment, physical examination, medical history, and death certificate information. Subjects were interviewed in person when possible; the interview covered medical and reproductive experiences and also included questions on functioning. Proxy interviews were conducted for 81 (24%) survivors and 26 (5%) controls. The potential bias of proxy compared with direct interviews was addressed with separate analyses. Quality of life was defined in several different ways. Among the questions asked were "Have you ever been unable to take a job, had to change jobs, or had to stop working for at least 3 months because of any health condition (not including pregnancy)?"; "Could you tell me if any of the following conditions ever affected your ability to work: sight problems, hearing problems, or emotional problems?"; "How much help do you need to drive a car?";
Journalof Clinical Oncology, Vol 9, No 4 (April), 1991: pp 592-599
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
593
LONG-TERM SURVIVORS OF PEDIATRIC CNS TUMORS "Compared to other persons your age would you say that your health is excellent, good, fair, or poor?" Interview responses were used to determine the following: employment status (never employed, considered as unemployed), the need to change jobs or stop working because of a health condition (including specific questions about emotional, sight, and hearing problems), the ability to drive a car without assistance, income level in most recent job, selfdescriptions of health with positive or negative adjectives, marital status, highest level of school attended, and the need for proxy interview because of mental incompetence. Histologic confirmation of a second primary cancer diagnosed at least 5 years after the first was obtained from medical records. The risk of death was assessed with a hazards ratio (HR) generated from a proportional hazards linear regression model stratified by family (PROC PHGLM14; data did not violate the proportional hazards assumptions) and KaplanMeier survival estimates. The risk of a second tumor in survivors or a first tumor in controls was assessed similarly. Risk for other outcomes was measured with odds ratios (OR) generated from a conditional logistic regression model (PROC MCSTRAT)15 to compare survivors with their matched sibling controls. Sex differences were assessed by comparing male or female survivors with their matched controls, regardless of sex, and similar stratified analyses were performed to assess tumor site (supratentorial v infratentorial) and a history of radiation therapy compared with no radiotherapy.16 When matched analyses were impractical (because of small numbers) or inappropriate (age at diagnosis has no meaning in controls), odds ratios, X2 analyses, and t-tests were calculated."7 The accepted level of statistical significance was alpha less than .05 for two-tailed comparisons of means or 95% confidence intervals (CIs). RESULTS The average age at CNS tumor diagnosis was 11.3 years; sibling controls were an average 12.2 years old at the survivor's diagnosis. Survivors were 32.0 years old on average at last follow-up (interview or death) and controls were 33.2. Survivors were about two inches shorter (P < .01) and weighed 9 lb less (P < .01) than controls. Only 4% of subjects were nonwhite. The 342 tumors were astrocytomas (56%); other gliomas (8%); ependymoma (8%); medulloblastomas (7%); and oligodendrogliomas, meningiomas, hemangioblastomas, and others (21%). Tumor sites were supratentorial (18%); infratentorial (39%); cranial, not otherwise specified (36%); and extracranial (7%). Medical record abstracts did not permit reliable assessment of the degree of histologic malignancy, size of tumors, or the severity of symptoms at diagnosis. Surgery was all or part of their treatment for 94% of survivors; other treatments were radiother-
apy (54%), radiation plus chemotherapy (2%), and chemotherapy only (1%). Of those who received radiotherapy, the mean total dose was 4,940 cGy. Use of radiotherapy was not related to the age at diagnosis, but was more likely in cases diagnosed in recent years (x 2 for trend; P = .01). Deaths and Second Cancers Death occurred before the interview in 47 survivors and seven controls. As adults, survivors were more likely to die than controls at any given age (HR, 14.0; CI, 4.9 to 39.6) (Fig 1). Among survivors, those who received radiation therapy were more likely to die than those who did not (HR, 5.0; CI, 2.3 to 10.7). After review of death certificates, only three deaths seemed unrelated to the history of cancer (one death due to a car accident with alcoholism listed as "other condition," one to being hit by a car as a pedestrian, and one to a stab wound not considered a suicide); the remaining 44 deaths were directly due to the primary tumor, its treatment, or a recurrence. Of the seven dead controls, one died from a brain tumor (Table 1 and text below); other causes of death in controls included accidental fall, skull fractures, pulmonary edema, and suicide (amobarbital poisoning). Second primary tumors were diagnosed more than 5 years after the first tumor diagnosis in five survivors, and four controls had cancer in adulthood (Table 1). As adults, survivors were no more likely to develop a second tumor than controls were to develop a first tumor at any given age (HR, 2.6; CI, 0.5 to 14.2). The boy with a hemangioblas-
e a
iE mi U.
cn
Age at Death (yr) Fig 1. Survival after CNS tumors (Kaplan-Meier curves). To be included in the study, subjects had to have survived at least 5 years and have reached age 21 years. Average follow-up time was 13 years after entry. (---) Sibling controls, (-) CNS tumor survivors (I, 95% confidence interval).
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
594
MOSTOW ET AL Table 1. Second Tumors in Survivors of Childhood and Adolescent CNS Tumors and Any Tumor in Their Sibling Controls Diagnosis First Tumor Diagnosis Type (site)
Year
Age (years)
Hemangioblastomat
1969
15
Female
Optic glioma
1953
3
Female Female
Astrocytoma (temporal lobe) Astrocytoma (spinal cord)
1954 1956
5 19
Female
Ependymoma
1956
6
Seminoma Renal cell carcinoma
1981 1976
35 21
Astrocytoma Hemangioblastoma
1976 1965
31 20
Sex Survivors Male
Sibling controls Male Female Male Male
Interval in Years*
Second Tumor
Renal cell carcinoma Pheochromocytoma Von Hippel-Lindau syndrome Neurofibrosarcoma (thigh) Neurofibromatosis Papillary thyroid carcinoma Breast (infiltrating ductal) Li-Fraumeni syndrome Oligodendroglioma
9 9 10 20 24 9
Survivor Oligodendroglioma Astrocytoma Von Hippel-Lindau syndrome Glioma Hemangioblastomat Von Hippel-Lindau syndrome
*Years from initial diagnosis to diagnosis of second cancer. tSame family as survivor noted above. Von Hippel-Lindau syndrome (see text).
toma as his first tumor had von Hippel-Lindau syndrome, which may include angiomas, cysts, pheochromocytomas, and tumors of the liver, pancreas, and kidneys (Table 1). The survivor with the optic glioma had neurofibromatosis type 1 with cafe au lait spots since birth and multiple neurofibrosarcomas on the left thigh as her second tumor. None of these five survivors received chemotherapy for their primary tumor. Only the survivor of the optic glioma received radiotherapy for her primary tumor, and she was the only one of the five who was dead at the time of interview. Among the four controls with confirmed tumors, the control with a hemangioblastoma was from the family with von Hippel-Lindau syndrome, which is transmitted with autosomal dominant inheritance. Interview Responses
Survivors differed from their siblings in many measures of adult functioning (Table 2). The greatest relative difference was in hearing problems that affected the survivors' ability to work or caused them to stop working. Employment difficulties were considerable; survivors were 11 times more likely never to have had a job and were at increased risk for a health condition that affected their ability to work or caused them to stop working. Although only 13% of the survivors could
not drive, inability to drive was almost 30 times
more likely for survivors than controls. Survivors rated their health as fair or poor more often than controls; they were also more likely never to have married. The mean highest grade of education Table 2. Quality-of-Life Measures in CNS Tumor Survivors Compared With Sibling Controls
Quality-of-Life Measure
Employment Never employed Health condition led to job change/stop working Conditions affecting work Emotional problems Sight problems Hearing problems Cannot drive a car Most recent income under $15,000 per year Health perception Excellent Good Fair Poor Never married
Crude Percentage of Survivors
Odds* Ratio
95% Confidence Interval*
15
10.8
4.6-25.7
37
5.9
12 24 8 13
2.4 7.7 45.2 28.8
76
3.3
2.3-4.9
38 42 15 5 42
0.6 1.0 3.8 7.8 3.6
0.4-0.8 0.7-35.7 1.9-7.5 1.7-35.7 2.5-5.3
3.7-9.4 1.4-4.0 4.2-14.0 6.1-333 6.9-119
*Derived from a stratified analysis that controlled for family; compares survivors with controls.
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
595
LONG-TERM SURVIVORS OF PEDIATRIC CNS TUMORS completed was 12 for survivors and 13 for siblings
(P < .01).
Table 4. Quality-of-Life Measures in CNS Tumor Survivors Compared With Sibling Controls by Tumor Site Crude
Differences by Sex, Tumor Type, and Treatment Male survivors differed from females on several measures of adult functioning (Table 3). Males
were at much greater risk of unemployment, having a health condition that affected employability, having emotional problems, or perceiving their health as fair or poor rather than excellent or good. Females were more likely to be earning less than $15,000 at their most recent job. Hearing problems and difficulty driving a car seemed to occur with equal frequency in both sexes. For instance, 9% of the male survivors versus 7% of the female survivors said that hearing problems had led to job difficulties; the corresponding percentages for driving a car were 16% and 19%, respectively. Sex-specific rates could not be estimated for hearing problems because none of the 11 female survivors had a sibling with hearing problems; a similar situation existed for male Table 3. Quality-of-Life Measures in CNS Tumor Survivors Compared to Sibling Controls by Sex
Quality-of-Life Measure
Never employed Males Females Health condition led to job change/stop working Males Females Emotional problems Males Females Sight problems Males Females Most recent income < $15,000 Males Females Health perceived as fair or poor compared to excellent or good Males Females Never married Males Females
Crude Percentage of Survivors
Odds* Ratio
95% Confidence Interval*
14 17
31.6 6.7
4.2-237.7 2.5-18.0
42 30
11.6 3.2
5.3-25.7 1.8-6.0
13 11
3.2 1.7
1.5-6.8 0.8-3.7
28 18
19.4 4.2
6.0-63.3 2.0-8.7
65 90
2.1 9.3
1.3-3.3 4.0-21.9
21 17
15.5 2.8
3.6-66.8 1.3-6.1
56 24
6.4 1.5
3.8-10.9 0.8-2.7
*Derived from a stratified analysis that controlled for family; compares survivors with controls.
Quality-of-Life Measure
Never employed Supratentorial Infratentorial Health condition led to job change/stop working Supratentorial Infratentorial Emotional problems Supratentorial Infratentorial Sight problems Supratentorial Infratentorial Most recent income < $15,000 Suprotentorial Infratentorial Health perceived as fair or poor compared to excellent or good Supratentorial Infratentorial Never married Supratentorial Infratentorial
Percentage of Survivors
Odds* Ratio
95% Confidence Interval*
25 8
17.7 5.0
2.3-138 1.3-18.7
53 28
17.8 4.3
4.2-75.9 2.0-9.3
23 10
19.8 2.5
2.6-153 0.9-7.1
33 18
25.7 9.3
3.4-195 2.7-31.8
79 71
7.1 2.3
2.4-21.1 1.3-3.9
11 21
6.0 4.8
0.6-57.7 1.9-12.2
34 40
5.8 3.0
1.8-17.8 1.7-5.5
*Derived from a stratified analysis that controlled for family; compares survivors with controls.
survivors and the ability to drive a car. Male survivors were more likely never to have married than female survivors (P < .001). Both tumor site and treatment were associated with altered quality-of-life measures (Tables 4 and 5). Survivors of supratentorial tumors and those who received radiation therapy were generally at greater risk for adverse outcomes than survivors of infratentorial tumors and those who received no radiation when each was compared with controls. Although sight problems were most likely among supratentorial tumor survivors compared with controls, the greatest relative difference between supratentorial and infratentorial tumor survivors was for emotional problems. Survivors who were treated with radiotherapy seemed to have poorer life experiences in most of the measures we looked at (Table 5), unemployment was nearly four times as frequent, and ratings of poor health status were very much more frequent in these survivors compared with those who did not receive radiotherapy. Hearing problems occurred in 12% of survi-
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
596
MOSTOW ET AL
Table 5. Quality-of-Life Measures in CNS Tumor Survivors Compared with Sibling Controls by Treatment
Quality-of-Life Measure
Never employed Radiotherapy No radiotherapy Health condition led to job change/stop working Radiotherapy No radiotherapy Emotional problems Radiotherapy No radiotherapy Sight problems Radiotherapy No radiotherapy Most recent income < $15,000 Radiotherapy No radiotherapy Health perceived as fair or poor compared to excellent or good Radiotherapy No radiotherapy Never married Radiotherapy No radiotherapy
Crude Percentage of Survivors
Odds* Ratio
95% Confidence Interval*
21 8
16.0 4.5
4.8-52.8 1.2-17.2
47 23
6.8 4.7
3.7-12.5 2.0-10.9
16 8
3.0 1.6
1.5-5.9 0.7-3.8
30 15
9.3 5.1
4.4-19.8 1.9-13.9
83 69
5.6 2.1
3.0-10.3 1.3-3.6
30 10
13.1 1.0
4.6-37.2 0.3-2.8
47 35
4.9 2.5
2.8-8.5 1.4-4.3
*Derived from a stratified analysis controlled for family; compares survivors with controls.
vors treated with radiotherapy compared with 4% of those who were not, and inability to drive was increased twofold (24% v 12%, respectively). The mean highest grade of education was grade 11 for those survivors who received radiation and 13 for those who did not (P < .01). No histologic diagnosis was consistently associated with increased risk for any of the outcomes when survivors were compared with controls in matched analyses, although matched analyses stratified by histology were less meaningful because most survivors had astrocytomas. Among survivors, however, medulloblastoma survivors were most likely to have had every one of the unfavorable outcomes listed in Table 4. Survivors diagnosed at younger ages were at increased risk for having never been employed, having a lower final level of education achieved, or never marrying (each tested with XZ for trend; P < .01). Age at diagnosis had no bearing on other measures.
Mental Incompetence Mental incompetence was given as the reason for a proxy interview by 28 survivors and two controls. Considering only subjects who were alive at the time of interview, survivors were more likely to be mentally incompetent than matched sibling controls (OR, 16.2; CI, 3.8 to 69.8). Compared with survivors with normal mental functioning, mentally incompetent survivors were younger at diagnosis (mean ages, 8.6 and 11.1 years, respectively; P = .02), more likely to have supratentorial tumors (OR, 8.7; CI, 3.0 to 24.7), and more likely to have been treated with radiotherapy (OR, 5.3; CI, 2.1 to 13.2). Male and female survivors were equally likely to be mentally incompetent (OR, 1.8; CI, 0.8 to 4.1). DISCUSSION Survival for 5 years after diagnosis was a requirement for inclusion in the study. Thus, this cohort was not representative of all incident pediatric and adolescent CNS tumors. Our 342 survivors comprised about 2% of all incident CNS tumors in the United States diagnosed between 1945 and 1975 and at ages younger than 20 years, assuming 1,100 new cases annually and 45% with 5-year survival."8 Astrocytoma was the leading cell type in this cohort of survivors, as it is in incidence series. 19 Survivors diagnosed more recently may already be benefiting from improved imaging techniques that detect tumors earlier, as well as surgical procedures and therapeutic regimens designed to reduce late morbidity. This study may serve as a historical comparison when patients diagnosed today become long-term survivors. In our study, matched analyses tended to be less likely to result in statistically significant risk estimates than unmatched analyses, but matching survivors to their specific siblings was consistent with the original study design. To the extent that any adult's quality of life could have been influenced by a variety of factors in childhood and adolescence (eg, neighborhood, schools attended, and parental influences-their occupations, marital status, education level, and attitudes toward their children), sibling matching assured that these factors were similar in survivors and controls. When proxy interviews were excluded from analyses, risk estimates for adverse outcomes were consistently lower, though conclusions of statistical significance were never changed. In other
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
597
LONG-TERM SURVIVORS OF PEDIATRIC CNS TUMORS
words, survivors interviewed by proxy were more likely to have adverse outcomes. We included proxy interviews to reflect the experience of all pediatric CNS tumor survivors and to minimize the selection bias from analyzing only individuals doing well enough to be interviewed. To the degree that siblings may be affected by the cancer diagnosis in a family member, they may not be the best control group for psychosocial outcomes. However, population-based data are not available that could be matched to our group of cancer survivors. Comparing survivors with their sibling(s) may be less than ideal, but this comparison is often unavoidable among parents, teachers, and close community members. Thus, sibling controls may be more useful for family counseling and similar interventions. Since OR express relative odds, a very low risk in controls can be just as important as high risk in survivors. For example, only two of the 453 sibling controls were unable to drive. Thus, the OR was high even though the majority of survivors were able to drive. We have presented OR that express risks compared with matched sibling controls because these are the comparisons that will be made by parents, survivors, siblings, and teachers. The survival rates beyond age 20 years (Fig 1) suggest that, as adults, long-term survivors of pediatric and adolescent CNS tumors have a sustained excess risk of death, compared with their siblings. Other authors who have followed CNS survivors for long periods 20' 21 noted that death rates seemed to diminish over time. Although we lack details on severity of the disease at diagnosis, the reported causes of death suggest continued complications owing to the original tumor, its treatment, or both. Again, the increased risk of death was associated with radiation therapy. The uncertain future after a tumor diagnosis-termed the Damocles syndrome-is a burden that may be carried forever and includes concerns about recurrence of disease and/or second primary tumors.22 Our survivors showed no excess of adultonset tumors, and two of the five observed were explained by mendelian disorders complicated by multiple neoplasms, namely the von HippelLandau syndrome and Recklinghausen neurofibromatosis. Familial clustering was also represented by the family with Li-Fraumeni syndrome (Table 1). The apparent excess of cancer among controls in our study, four cases in 479 siblings, is explained
in part by familial clustering, which may account for some of the excess cancer risk seen in siblings of children with CNS tumors.23'24 Our findings on societal functioning show that when these survivors of pediatric and adolescent CNS tumors were compared with their siblings, the survivors were more likely to have a number of adverse outcomes as adults. Still, 85% of the survivors had some employment, 83% could drive a car, and 80% described their health as excellent or good (Table 2). Stable employment is one measure of security and social independence. Although jobs differ in physical and intellectual demands, survivors were at significantly greater risk for being permanently unemployed, consistent with prior reports. 25 A recent analysis of 40 adolescent non-CNS cancer survivors showed unemployment to be almost twice as likely in survivors than controls; furthermore, 7% reported discrimination in hiring based on their cancer history.26 Occupational therapists, parents, and others with the opportunity to help develop vocational skills might devote extra attention to recovering pediatric and adolescent CNS tumor patients to attenuate long-term disabilities. Male survivors had consistently greater risk for adverse outcomes than female survivors who were functioning almost as well as their siblings. It is unlikely that males had more severe disease and more aggressive treatment than females because the distribution of major histologic tumor types and treatments was similar in each sex. Psychosocial interactions between parents and affected children may have long-term effects; if female survivors were given more support, attention, or different treatment, this might have reduced their risk for adverse outcomes. Alternatively, sexspecific differences in brain composition and structure 27 could explain the female's superior ability to recover from CNS tumors. Improved attention to psychosocial aspects of care could reduce the risks for adverse outcomes in both sexes, but male survivors seemed to be in greater need of assistance. Supratentorial tumors are more likely to result in higher function deficits (eg, intellect, memory, emotion) than infratentorial tumors (mostly cerebellar), which tend to affect motor coordination.28 More severe disability has been noted for other survivors of pediatric supratentorial compared with infratentorial brain tumors. 29 However, infra-
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
598
MOSTOW ET AL
tentorial tumors can also affect higher cortical function since any intracranial space-occupying lesion can affect the entire brain. This may partially explain the fact that infratentorial survivors were also at significant risk for most of the same adverse outcomes as supratentorial survivors. Still, the greater impact of supratentorial tumors in this study might be because of more severe long-term effects on intellectual and emotional development. Pediatric brain tumor survivors who had motor impairment as their primary long-term deficit generally fared better than those with intellectual and psychologic-emotional sequelae.2 9 We should not attribute reduced quality of life simply to radiation therapy; radiotherapy in these CNS tumor survivors might simply indicate increased severity or reduced operability of the initial tumor. Although it is difficult to separate the effect of the tumor from the treatment, other studies have documented significant intellectual and social deficits in survivors who had received
opment. Physical disfigurement or handicap may also diminish a survivor's desire and ability to interact socially. Although our measures were crude surrogates for the highly subjective assessment of any individual's quality of life, our results suggest that there is considerable room to improve the quality of the future lives of current patients. The ideal goal of designing improvements in therapy should be a life similar to that expected for siblings and peers. Survivors and their families may take solace in knowing that the majority of individuals in our cohort were functioning well as adults. Parent self-help groups are an important part of families' social support systems,34 and they may use the quality-of-life measures presented here as the basis for discussions with their children and their health-care providers. Most importantly, identification of differences may permit interventions (eg, physical and occupational therapy and special education) to minimize possible late effects.
radiotherapy especially early in life.30 3- 3 However,
one recent study suggests that several factors other than radiotherapy contribute to diminished neurodevelopmental function."0 The effect of age at diagnosis may reflect the sensitivity of the developing brain to radiation. It could also represent impaired development of social skills at crucial points of psychosocial devel-
ACKNOWLEDGMENT The authors are grateful to Mary McAdams, IMS, for technical assistance, and acknowledge the assistance of the patients and staff of the following registries: the California State Department of Health Services, Yale University, Connecticut Tumor Registry, University of Kansas, University of Iowa, and The University of Texas MD Anderson Cancer Center.
REFERENCES 1. Young JL, Ries LG, Silverberg E, et al: Cancer incidence, survival, and mortality for children younger than age 15 years. Cancer 58:598-602, 1986 2. US Department of Health and Human Services, Public Health Service, National Institutes of Health, National Cancer Institute: 1987 Annual Cancer Statistics Review. NIH Publication No. 88-2789 3. Mor V, Guadagnoli E: Quality of life measurement: A psychometric tower of Babel. J Clin Epidemiol 41:10551058, 1988 4. Calman KC: Definitions and dimensions of quality of life, in Aronson NK and Beckman J (eds): Quality of Life of Cancer Patients. New York, NY, Raven, 1987, pp 1-9 5. Donovan K, Sanson-Fisher RW, Redman S: Measuring quality of life in cancer patients. J Clin Oncol 7:959-968, 1989 6. Bamford FN, Morris Jones P, Pearson D, et al: Residual disabilities in children treated for intracranial space-occupying lesions. Cancer 37:1149-1151, 1976 7. Bordeaux JD, Dowell RE, Copeland DR, et al: A prospective study of neuropsychological sequelae in children with brain tumors. J Child Neurol 3:63-68, 1988 8. Ellenberg L, McComb JG, Siegel SE, et al: Factors
affecting intellectual outcome in pediatric brain tumor patients. Neurosurg 21:638-644, 1987 9. Lebaron S, Zeltzer PM, Zeltzer LK, et al: Assessment of quality of survival in children with medulloblastoma and cerebellar astrocytoma. Cancer 62:1215-1222, 1988 10. Mulhern RK, Horowitz ME, Kovnar EH, et al: Neurodevelopmental status of infants and young children treated for brain tumors with preirradiation chemotherapy. J Clin Oncol 7:1660-1666, 1989 11. Packer RJ, Meadows AT, Rorke LB, et al: Long-term sequelae of cancer treatment on the central nervous system in childhood. Med Pediatr Oncol 15:241-253, 1987 12. Mulvihill JJ, Myers MH, Connelly RR, et al: Cancer in offspring of long-term survivors of childhood and adolescent cancer. Lancet 2:813-817, 1987 13. Byrne J, Mulvihill JJ, Myers MH, et al: Effects of treatment on fertility in long-term survivors of childhood or adolescent cancer. New Engl J Med 317:1315-1321, 1987 14. Harrell FE: The PHGLM procedure, in Hastings RP (ed): SUGI Supplemental Library User's Guide, Version 5 Edition. Cary, NC, SAS Institute, Inc, 1986, pp 437-466 15. Naessens JM, Offord KP, Scott WF, et al: The MCSTRAT procedure, in Hastings RP (ed): SUGI Supple-
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
599
LONG-TERM SURVIVORS OF PEDIATRIC CNS TUMORS mental Library User's Guide, Version 5 Edition. Cary, NC, SAS Institute, Inc, 1986, pp 307-328 16. Kleinbaum DG, Kupper LL, Morgensten H: Epidemiologic Research. Principles and Quantitative Methods. New York, NY, Van Nostrand Reinhold, 1982 17. SAS Institute Inc: SAS User's Guide: Statistics, Version 5 Edition. Cary, NC, SAS Institute, Inc, 1985, pp 403-432 18. Li FP: Cancers in children, in Schottenfeld D, Fraumeni JF Jr (eds): Cancer Epidemiology and Prevention. Philadelphia, PA, Saunders, 1982, pp 1012-1024 19. Duffner PK, Cohen ME, Freeman AI: Pediatric brain tumors: An overview. CA 35:287-301, 1985 20. Hawkins MM: Long-term survival and cure after childhood cancer. Arch Dis Child 64:789-807, 1989 21. Spunberg JJ, Chang CH, Goldman M, et al: Quality of long-term survival following irradiation for intracranial tumors in children under the age of two. Int J Radiat Oncol Biol Phys 17:727-736, 1981 22. Koocher GP, O'Malley JE: The Damocles Syndrome: Psychosocial Consequences of Surviving Childhood Cancer. New York, NY, McGraw-Hill, 1981 23. Miller RW: Deaths from childhood leukemia and solid tumors among twins and other sibs in the United States, 1960-67. JNCI 46:203-209, 1971 24. Draper GJ, Heaf MM, Kinnear-Wilson LM: Occurrence of childhood cancers among sibs and estimation of familial risks. J Med Genet 14:81-90, 1977 25. Li FP, Winston KR, Gimbere K: Follow-up of children with brain tumors. Cancer 54:135-138, 1984
26. Tebbi CK, Bromberg C, Piedmonte M: Long-term vocational adjustment of cancer patients diagnosed during adolescence. Cancer 63:213-218, 1989 27. Allen LS, Hines M, Shryne JE, et al: Two sexually dimorphic cell groups in the human brain. J Neurosci 9:497-506, 1989 28. Adams RD, Hochberg F, Webster HD: Neoplastic disease of the brain, in Isselbacher KJ, Adams RD, Braunwald E, et al (eds): Harrison's Principles of Internal Medicine. New York, NY, McGraw-Hill, 1980, pp 19511960 29. Lannering B, Marky I, Lundberg A, et al: Long-term sequelae after pediatric brain tumors: Their effect on disability and quality of life. Med Pediatr Oncol 18:304-310, 1990 30. Peckham VC, Meadows AT, Bartel N, et al: Educational late effects in long-term survivors of childhood acute lymphocytic leukemia. Pediatrics 81:127-133, 1988 31. Kun LE, Mulhern RK, Crisco JJ: Quality of life in children treated for intellectual outcome in pediatric brain tumor patients. Neurosurgery 21:638-644, 1987 32. Dowell RE, Copeland DR: Cerebral pathology and neuropsychological effects: Differential effects of cranial radiation-as a function of age. Am J Pediatr Hematol Oncol 9:68-72, 1987 33. Choux M, Lena G, Hassoun J: Prognosis and longterm follow-up in patients with medulloblastoma. Clin Neurosurg 30:246-277, 1983 34. Monaco GP: Parent self-help groups for the families of children with cancer. CA 38:169-175, 1988
Downloaded from ascopubs.org by UNIVERSITY LIVERPOOL on April 20, 2019 from 154.059.124.102 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
![[CNS tumors in childhood].](/assets/img/hold.png)
