Growth Hormone Replacement Therapy in Children With Medulloblastoma: Use and Effect on Tumor Control By Roger J. Packer, James M. Boyett, Anna J. Janss, Theodora Stavrou, Larry Kun, Jeffrey Wisoff, Carolyn Russo, Russell Geyer, Peter Phillips, Mark Kieran, Mark Greenberg, Stewart Goldman, Douglas Hyder, Richard Heideman, Dana Jones-Wallace, Gilbert P. August, Sharon H. Smith, and Thomas Moshang Purpose: Progress has been made in the treatment of medulloblastoma, the most common childhood malignant brain tumor: However, many long-term survivors will have posttherapy growth hormone insufficiency with resultant linear growth retardation. Growth hormone replacement therapy (GHRT) may significantly improve growth, but there is often reluctance to initiate GHRT because of concerns of an increased likelihood of tumor relapse. Patients and Methods: This study retrospectively reviewed the use of GHRT for survivors of medulloblastoma in 11 neuro-oncology centers in North America who received initial treatment for disease between 1980 and 1993 to determine its impact on disease control. A Landmark analysis was used to evaluate the relative risk of relapse in surviving patients. Results: Five hundred forty-five consecutive patients less than 15 years of age at diagnosis were identified. Six-year progression-free survival (mean ⴞ SD) was 40% ⴞ 5% in children less than 3 years of age at

diagnosis compared with 59% ⴞ 3% for older patients. Older patients with total or near-total resections (P ⴝ .003) and localized disease at diagnosis (P < .0001) had the highest likelihood of survival. One hundred seventy patients (33% ⴞ 3% of the cohort) received GHRT. GHRT use varied widely among institutions, ranging from 5% to 73%. GHRT was begun a mean of 3.9 years after diagnosis, later in children younger than 3 years at diagnosis (5.4 years). By Landmark analyses, for those surviving 2, 3, and 5 years after diagnosis, there was no evidence that GHRT increased the rate of disease relapse. Conclusion: This large retrospective review demonstrates that GHRT is underutilized in survivors of medulloblastoma and is used relatively late in the course of the illness. GHRT is not associated with an increased likelihood of disease relapse. J Clin Oncol 19:480-487. © 2001 by American Society of Clinical Oncology.

EDULLOBLASTOMA IS the most common malignant brain tumor of childhood and the one for which the most progress has been made over the past three decades.1 The tumor arises at all ages and peaks in incidence between the ages of 3 and 7. The majority of tumors

will require treatment with craniospinal radiation therapy supplemented with local boost radiation therapy to the primary tumor site.1 Many children will also receive adjuvant chemotherapy. Treatment is now successful in between 60% and 80% of children, however, long term survivors often have significant sequelae.1,2 One of the most common sequelae is growth hormone insufficiency with resultant retarded linear growth.3 After conventional doses of craniospinal radiation therapy (36 Gy), supplemented with local boost radiotherapy to a dose of 55 Gy, nearly all survivors diagnosed before puberty will suffer some degree of growth insufficiency.3-5 For these children, growth hormone treatment is often indicated. However, primarily because of concerns that growth hormone may act as a mitogen and stimulate tumor growth, there is often reluctance on the part of the physician and/or parents to begin growth hormone replacement thereapy (GHRT). The period of time between completion of treatment and recommendations to initiate GHRT also significantly varies. Some neuro-oncology centers use GHRT as early as 1 year after initial treatment, whereas other institutions arbitrarily do not start GHRT for at least 2 to 3 years after the completion of all treatment. In an attempt to better elucidate the use of GHRT in children surviving medulloblastoma, to determine the pat-

M

From the Departments of Neurology and Pediatrics, Children’s National Medical Center, The George Washington University, Washington, DC; Departments of Biosatistics and Neuro-Oncology, St Jude Children’s Research Hospital, Memphis, TN; Departments of Oncology, Neurology, and Endocrinology, Children’s Hospital of Philadelphia, Philadelphia, PA; Department of Neurosurgery, New York University, New York; Children’s Hospital of Buffalo and Roswell Park Cancer Institute, Buffalo, NY; Department of Neurosurgery, University of California at San Francisco, San Francisco; Children’s Hospital of Los Angeles, Los Angeles, CA; Children’s Memorial Hospital, Chicago, IL; Department of Pediatrics, Children’s Hospital and Regional Medical Center, Seattle, WA; Dana-Farber Cancer Institute, Boston, MA; Hospital for Sick Children, Toronto, Canada. Submitted May 15, 2000; accepted September 13, 2000. Partially supported by grant no. CA21765 and by the American Lebanese Syrian Associated Charities. Address reprint requests to Roger J. Packer, MD, Department of Neurology, Children’s National Medical Center, 111 Michigan Avenue, NW, Washington, DC 20010; email [email protected]. © 2001 by American Society of Clinical Oncology. 0732-183X/01/1902-480

480

Journal of Clinical Oncology, Vol 19, No 2 (January 15), 2001: pp 480-487

Downloaded from ascopubs.org by 37.44.207.45 on January 20, 2017 from 037.044.207.045 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

481

GROWTH HORMONE FOR MEDULLOBLASTOMA

tern of growth hormone use and to assess its impact, if any, on disease control in children with medulloblastoma, this study retrospectively reviewed the experience at 11 major neuro-oncology programs in North America. PATIENTS AND METHODS Eleven institutions in the United States and Canada with wellestablished pediatric brain tumor programs retrospectively reviewed all medulloblastoma patients diagnosed and treated at their institutions between 1980 and 1993. In an attempt to include only prepubertal children, only children 15 years of age or younger were eligible for review. To be eligible for study, all patients must have received their primary treatment at the institution and must have had a histologically confirmed medulloblastoma. Primitive neuroectodermal tumors that arose outside the posterior fossa (such as pineoblastoma and supratentorial primitive neuroectodermal tumors) were excluded from study (see Appendix). Information obtained on retrospective chart review included date of birth; date of diagnosis; extent of tumor at diagnosis, designated as a metastatic (M) stage (M0 ⫽ no dissemination, M1 ⫽ positive cytology for tumor in the CSF only, and M2-3 ⫽ evidence of metastatic disease on myelography or magnetic resonance imaging of the spine); extent of tumor resection (total or near total resection, subtotal resection, or biopsy); use of radiation, including dose to the craniospinal axis and local tumor site; use of chemotherapy; date GHRT was started; date of tumor progression; and date of last contact for patients who had not progressed.

Statistical Considerations Distributions of progression-free survival (PFS) were estimated by the Kaplan-Meier method. PFS was measured from date of diagnosis (and specific landmarks, as explained below) until failure (death or disease progression) or date of last contact. The prognostic implications of age at diagnosis, extent of tumor resection, and M stage were evaluated by the Mantel-Haenszel statistic.6 Because of known important treatment differences, all analyses of the prognostic implication of GHRT were conducted separately for the two groups of patients defined by whether or not patients were diagnosed before their third birthday (infants). For analysis of use of GHRT, patients were also separated into those between 3 and 12 years of age and those older than 12 years because some patients older than 12 might have already begun puberty and may have been less likely to receive GHRT. In addition, because children with medulloblastoma are at risk for premature onset of puberty, an analysis was also performed evaluating only those children 10 years of age or younger at initiation of treatment. Analyses of outcome for all older children were stratified by M stage and extent of resection. The Cox life-table regression model with initiation of GHRT as a time-dependent covariate was used to generate estimates of relative risk between patients who received growth hormone and those who did not.7 The Landmark method8 was used to graphically display the impact of GHRT on PFS as a function of initiation of GHRT. All children at least 3 years of age and surviving failure-free at the landmarks of 2 years, 3 years, and 5 years after diagnosis were analyzed for subsequent PFS according to whether or not GHRT had been initiated before the landmark. Patients who began GHRT after the landmark were censored at that point for the subsequent estimate of PFS. The Landmark analyses was stratified by M stage and extent of tumor resection. Similarly, infants less than 3 years of age at the time of diagnosis were analyzed separately at the landmark of 5 years. The

cumulative incidences of use of GHRT were estimated considering death and progression as competing risks.9

RESULTS

The 11 participating institutions provided data on 575 patients. Of these, 30 were ineligible (two were diagnosed before 1980; one was older than 15 years at diagnosis; and 27 were diagnosed after 1993). Thus, this study identified 545 consecutive patients diagnosed between 1980 and 1993 who were 15 years of age or younger at the time of their diagnosis of medulloblastoma. Four hundred forty-five children were 10 years of age or younger at time of diagnosis. Data was relatively complete, but there was some missing information for the 545 eligible patients, including date of progression,6 extent of surgical resection,5 M stage,10 and date of start of GHRT.2 Representativeness of Cohort Because this is a retrospective multi-institutional study, several parameters demonstrated to be prognostic factors in other studies were investigated to evaluate the representativeness of this cohort. Four hundred sixteen children were at least 3 years of age at the time of diagnosis, and 129 infants (23.7%) were younger. Patients were a mean age of 6.2 years at the time of diagnosis. Table 1 lists the distribution of extent of resection and presenting M stage by the two groups defined by age at diagnosis. Fifty-nine percent of infants had a total or near total resection of their primary tumor compared with 67% of older children. Sixty-three percent of infants presented with no metastatic disease (M0) compared with 73% of older children. Effects of Known Prognostic Factors on PFS Figure 1 shows Kaplan-Meier estimates of PFS for the prognostic factors investigated. Patients less than 3 years of age at the time of diagnosis had a 6-year PFS (mean ⫾ SD) of 40% ⫾ 5% compared with 59% ⫾ 3% for older patients. Extent of resection was a significant predictor of PFS in older children (P ⫽ .0003) but not in infants P ⫽ .56). In children at least 3 years of age at diagnosis, 65% ⫾ 3% were alive and free of progressive disease 6 years after diagnosis after a total, or near total, resection compared with 45% ⫾ 5% of patients who had a subtotal resection or biopsy. M stage was an important predictor of PFS in both older children (P ⬍ .0001) and infants (P ⫽ .020). In children greater than 3 years of age, those with M0 disease had a 6-year, PFS rate of 66% ⫾ 3% compared with 39% ⫾ 5% for those patients with M1⫹ disease. GHRT The use of GHRT before progression in medulloblastoma patients varied widely among institutions for both infants

Downloaded from ascopubs.org by 37.44.207.45 on January 20, 2017 from 037.044.207.045 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

482

PACKER ET AL Table 1.

Staging Information Patients

Age at Diagnosis ⬍3 years Presenting Features

Extent of resection* Total/near total Subtotal Biopsy M stage† M0 M1 M2-M4

ⱖ3 years

Totals

No.

%

No

%

No.

%

75 46 6

59.1 36.2 4.7

278 128 7

67.3 31.0 1.7

353 174 13

65.4 32.2 2.4

79 21 26

62.7 16.7 20.6

297 29 83

72.6 7.1 20.3

376 50 109

70.3 9.4 20.4

*Missing extent of resection for five patients. †Missing M stage for 10 patients.

and older children. Patients began GHRT at a mean of 10.1 years of age and were treated with GHRT for a mean of 49 months (range, 1 to 125 months; SD of 27.8 months). Final cumulative incidences of the use of GHRT before progression for infants and older (3 to 12 years of age) children are shown in panel B of Fig 2 for each of the 11 institutions. For older children, the cumulative incidences varied from 5% to 73%; the cumulative incidence of GHRT was 33.3%. There were 34 children older than 12 at diagnosis, and five of these received GHRT. In addition, eight older children received GHRT after progression (Table 2). If only children 10 years of age or younger were analyzed, the cumulative indices varied from 6% to 77%; the cumulative incidence of GHRT was 36.4%. In infants, the use of GHRT varied from 0% to 33%. Thirty-five infants received growth hormone before progression, and all had been treated with craniospinal radiation therapy at some point. Six of these infants received GHRT after an initial progression (Table 2). As shown in panel C of Fig 2 the times of initiation of GHRT in older children (at least 3 years of age) varied widely among institutions. The median times ranged from 1.72 to 5.7 years after diagnosis. In those children 10 years of age or younger, the median times of initiation ranged from 1.6 to 6.2 years after diagnosis. The 75th percentiles varied from 2.63 to 7.9 years after diagnosis. Furthermore, the differences between the 25th and 75th percentiles illustrate considerable intra-institutional variability for a number of institutions. The median age of older children at the time of initiation of GHRT was 10.9 years. For infants (data not shown) inter-institutional variability in the initiation of GHRT also exists. The median times of initiation of GHRT varied from 3.72 to 9.04 years after diagnosis. The median age of infants at the time of initiation of GHRT was 7.1 years.

Effects of Growth Hormone Use on PFS There was no statistical evidence of association between the use of GHRT and PFS in either infants (P ⫽ .71) or older children (P ⫽ .138). There was also no statistical evidence of association between the use of GHRT and PFS if analysis was restricted to those between 3 and 10 years of age (P ⫽ .084). The estimates and 95% confidence intervals (CI) for the relative risk of GHRT usage were 0.710 (95% CI, 0.648 to 4.267) for infants and 0.648 (95% CI, 0.365 to 1.150) for older children. The CI for the older children even suggests an advantage in PFS for those older children who received GHRT. However, there was an inherent bias in this analysis because children had to survive for greater than 2 years of age to be a candidate for receiving growth hormone. This possible association was further investigated with landmark analyses, which are displayed in Fig 3. Panel A of Fig 3 shows that of the older children who were free of progression 2 years after diagnosis, 24 had begun GHRT by that landmark. There is no statistical evidence of a difference in subsequent PFS (P ⫽ .55). Panel B shows similar results of the landmark of 3 years after diagnosis. On the other hand, panel C shows that of the older children who were progression-free 5 years after diagnosis, 85 had begun GHRT before that point and there was statistical evidence of a difference in subsequent PFS (P ⫽ .019). The children who received GHRT had a better outcome. Findings were similar, independent if children between 3 and 15 years of age at diagnosis, 3 and 12 years of age at diagnosis, or 3 and 10 years of age at diagnosis were analyzed. Panel D shows that for infants who were free of disease at 5 years after diagnosis, 13 had begun GHRT by the landmark. Because of limited statistical power, one cannot conclude with confidence that the two groups have the same outcome; however, the estimates do suggest that the infants who

Downloaded from ascopubs.org by 37.44.207.45 on January 20, 2017 from 037.044.207.045 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

483

GROWTH HORMONE FOR MEDULLOBLASTOMA

DISCUSSION

Fig 1. Panel A: Kaplan-Meier estimates of the distributions of PFS for infants less than 3 years of age and older at diagnosis. Panels B and D: Estimates of PFS according to extent of tumor resection for older children and infants, respectively. Panels C and E: Estimates of PFS according to M stage for older children and infants, respectively.

received GHRT did not have an inferior outcome (P ⫽ .46). There is no statistical evidence of a difference in subsequent PFS (P ⫽ .46).

Short stature is a common sequelae of successful treatment of childhood brain tumors.3-5 The reasons for short stature in children with brain tumors is multifactorial and includes poor nutrition, abnormal vertebral growth in children receiving irradiation to the spinal column, the use of chemotherapy, and direct damage to the hypothalamicpituitary axis secondary to the tumor or surgery. However, the most common reason for short stature in children with brain tumors is the effects of cranial irradiation on hypothalamic/pituitary function. After cranial irradiation, short stature is most commonly secondary to growth hormone deficiency. The detrimental effects of growth hormone insufficiency are exacerbated by the increased incidence of precocious puberty in children who have received cranial irradiation. Although growth hormone insufficiency has been reported in a variety of different tumor types, most studies have dealt with the effects of cranial irradiation on growth hormone secretion in children surviving leukemias and medulloblastomas.3 The dose and dose fraction of cranial irradiation used effects the likelihood of the development of growth hormone insufficiency. Larger dose fractions are more likely to cause growth hormone insufficiency, and the greater the overall dose of radiation, the higher the incidence of growth hormone deficiency and the shorter the time interval between treatment and biochemical evidence of growth hormone deficiency.10-12 Hypothalamic/ pituitary doses of greater than 24 Gy, especially at doses greater than 30 Gy, are likely to result in abnormal growth hormone release to stimuli including arginine and insulin, at times with resultant reduction in growth velocity and overall short stature.10-12 Children with medulloblastomas are at high risk for this sequelae because patients are usually prepubertal at diagnosis and conventionally receive craniospinal radiation doses of 36 Gy. With improving survival rates for children with medulloblastomas, long-term sequelae takes on increasing importance. Despite the lack of any clear evidence that medulloblastomas contain growth hormone receptors and that tumor control is detrimentally effected by growth hormone replacement, there has been reluctance to use GHRT for children who are long-term survivors of medulloblastoma.13 This reluctance seems to be a result of a variety of different reasons. There has been the concern that growth hormone will induce new malignancies or activate the growth of a pre-existing tumor.13,14 A report from Japan suggest an increased incidence of leukemia in patients treated with growth hormone.14-16 Data from several national registries have not confirmed this relationship.15,16 In one study of 6,284 American recipients of growth hormone, six cases of

Downloaded from ascopubs.org by 37.44.207.45 on January 20, 2017 from 037.044.207.045 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

484

PACKER ET AL

Fig 2. Panel A shows the cumulative incidence functions for the initiation of GHRT for infants less than 3 years of age at diagnosis and for older children at least 3 years of age at diagnosis. Panel B shows final cumulative incidences for infants and older children treated at each of the 11 institutions. Panel C shows the 25th percentiles, median, and 75th percentiles of the distributions of initiation of GHRT in terms of years after diagnosis for children at least 3 years of age at diagnosis who were treated at each of the 11 institutions. The horizontal bars are at the medians; the lower and upper points of the vertical lines are the 25th and 75th percentiles, respectively.

Downloaded from ascopubs.org by 37.44.207.45 on January 20, 2017 from 037.044.207.045 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

485

GROWTH HORMONE FOR MEDULLOBLASTOMA Table 2.

Growth Hormone Use Patients

Age at Diagnosis ⬍3 years

ⱖ3 years

Totals

GHRT

No.

%

No.

%

No.

%

Never Before Progression After Progression

98 25 6

76.0 19.4 4.7

280 128 8

67.3 30.8 1.9

378 153 14

69.4 28.1 2.6

leukemia were found compared with an expected 2.26 cases. The increased risk was confined to subjects with a previous cranial tumor, including four with craniopharyngiomas and one with an astrocytoma. This information was confounded because four of the five patients had received cranial irradiation. Other long-term surveys of patients treated with growth hormone have not found an increased incidence of leukemia.17 An international database of 228 children treated with growth hormone for hypopituitarism, secondary to craniopharyngiomas, found 17 recurrences, including four children who had tumor recurrence before starting growth hormone relapse.18 A retrospective analysis of the National Cooperative Growth Hormone Study evaluated 1,262 children with brain tumors who were treated with growth hormone.13 Brain tumors recurred in 6.6% of these children, including 18% of children with low-grade gliomas, 7.2% in children with medulloblastomas, and 6.4% of those with craniopharyngiomas. The authors concluded that this recurrence rate suggested that growth hormone use was not associated with an increased incidence of tumor recurrence in children with brain tumors. Another reason for the reluctance of some to use GHRT for children with medulloblastoma has been studies that have demonstrated growth hormone receptors or other type of growth hormone-related receptors on brain tumors.19-21 Although insulin-like growth factor binding proteins have been demonstrated in a variety of different CNS tumors, this has primarily been documented in malignant gliomas and meningiomas.19-21 Despite the highly questionable rationale for avoiding GHRT in children with medulloblastoma, there has been continued reluctance to use such treatment by some centers, primarily because of the lack of any prospective, or for that matter retrospective, large studies demonstrating the safety of such treatment and the effects of such treatment on long-term tumor control. The results of this study confirm the variability and, in some cases, reluctance of centers to use growth hormone. The institutions chosen to participate in this study all had well-established pediatric brain tumor programs and were participating in innovative studies for children with brain tumors. Their use of GHRT varied

widely; in one institution only 6% of eligible patients who were progression-free after apparent successful treatment for medulloblastoma received growth hormone compared with more than 50% of children in another institution. In addition, the timing of growth hormone use in this group of children also significantly differed. Growth hormone was begun at a median of approximately 4 years after diagnosis, but children received growth hormone as early as 1 year after completion of treatment, whereas others did not receive growth hormone until 7 years later after diagnosis and treatment. To avoid an age bias in analysis of data, patients diagnosed between birth and 15 years of age were included in this study. The onset of puberty is difficult to predict in this population because the presence of the tumor in an adolescent may delay the onset of puberty, whereas the use of craniospinal radiation may result in an earlier onset of puberty. For those reasons, the use of GHRT and its effects on tumor control were analyzed for children diagnosed at 15 years of age or less at diagnosis, 12 years of age or less at diagnosis, and 10 years of age or less at diagnosis. For the group as a whole, even when stratified for the other risk factors, children who received growth hormone had a better overall 5-year PFS rate. However, there was obviously a bias in this analysis because children who had progressed early in the course of illness most likely would have succumbed to their disease and would not have been candidates to receive growth hormone. For this reason, the Cox model with a time-dependent covariate to graphically describe the findings was used to analyze the data. Landmark method of analysis was used, starting the reanalysis at specific time points after diagnosis, so that patients all had to survive and have no evidence of disease progression for a specific period of time before comparative analysis was performed. At the 2-, 3-, 4-, and 5-year points, there was no evidence that the use of GHRT made disease progression more likely. In fact, there was a trend for better 5-year PFS for those patients who received growth hormone compared with those who did not. This did not reach statistical significance in all analyses but raises the possibility of some selection bias and that healthier patients may have gone on to receive growth hormone therapy.

Downloaded from ascopubs.org by 37.44.207.45 on January 20, 2017 from 037.044.207.045 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

486

PACKER ET AL

Fig 3. Landmark analyses of PFS and GHRT for children at least 3 years of age at diagnosis are shown for the landmarks of 2, 3, and 5 years after diagnosis in panels A, B, and C, respectively. The landmark analysis for infants less than 3 years of age is shown in panel D for the landmark of 5 years after diagnosis.

Downloaded from ascopubs.org by 37.44.207.45 on January 20, 2017 from 037.044.207.045 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.

487

GROWTH HORMONE FOR MEDULLOBLASTOMA

The results of this study suggest that GHRT for prepubertal patients with growth hormone insufficiency is not associated with an increased likelihood of disease relapse, independent of age at diagnosis, and for those patients who could benefit from GHRT, such treatment should probably be instituted earlier and more frequently. More recent treatment trials are attempting to use a lower dose of craniospinal irradiation. If such trials prove to be successful, the overall frequency of growth hormone insufficiency, and thus the need for GHRT, will probably decrease. However, given the other potential beneficial effects of GHRT other than increases in linear growth, such as improved muscle strength, a lower risk of cardiovascular disease, and better overall quality-of-life, GHRT may take on increasing

importance in older, postpubertal patient populations.22 This study did not address long-term growth hormone use, except for prepubertal patients who were growth hormone deficient and had impaired longitudinal growth. The vast majority of children in this series were treated with 36 Gy of cranial irradiation. Prospective studies are needed to evaluate the benefits and safety of GHRT in children surviving medulloblastoma, especially in postpubertal patients, and our data would strongly suggest that such studies can be safely undertaken. ACKNOWLEDGMENT We thank Yingfu Li, MS, who assisted with the statistical analysis and provided technical support.

APPENDIX Institutions cooperating in this study included: Children’s National Medical Center, Washington, DC, University of Buffalo, Roswell Park Cancer Center, Buffalo; New York University, New York, NY; Children’s Memorial Hospital, Chicago, IL; St Jude Children’s Research Hospital, Memphis, TN; Children’s Hospital of Los Angeles, Los Angeles; University of California at San Francisco, San Francisco, CA; The Hospital for Sick Children, Toronto, Canada; Children’s Hospital and Dana-Farber Cancer Institute, Boston, MA; Children’s Hospital of Philadelphia, Philadelphia, PA; Seattle Children’s Hospital, Seattle, WA.

REFERENCES 1. Packer RJ, Cogen P, Vezina G, et al: Medulloblastoma: Clinical and biologic aspects. Neurooncol 1:232-250, 1999 2. Packer RJ, Goldwein J, Nicholson HS, et al: Treatment of children with medulloblastoma with reduced-dose craniospinal radiation therapy and adjuvant chemotherapy: A Children’s Cancer Group study. J Clin Oncol 17:2127-2136, 1999 3. Packer RJ, Meadows AT, Rorke LB, et al: Long-term sequelae of cancer treatment on the central nervous system in childhood. Med Ped Oncol 15:241-253, 1987 4. Oberfield SE, Allen JC, Pollack J, et al: Long-term endocrine sequelae after treatment of medulloblastoma: Prospective study of growth and thyroid function. J Pediatr 108:219-223, 1986 5. Shalet SM, Gibson B, Swindell R, et al: Effect of spinal irradiation on growth. Arch Dis Child 62:461-464, 1987 6. Mantel N: Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 50:163170, 1966 7. Cox DR: Regression models and life tables (with discussion). J R Stat Soc B34:187-220, 1972 8. Anderson JR, Cain KC, Gelber RD: Analysis of survival by tumor response. J Clin Oncol 1:710-719, 1983 9. Kalbfleisch JD, Prentice RL: Statistical Analysis of Failure Time Data. New York, NY, Wiley, 1980, pp 163-178 10. Shalet SM, Beardwell CO, Morris Jones PH, et al: Growth hormone deficiency after treatment of acute leukemia in children. Arch Dis Child 51:489-494, 1976 11. Shalet SM, Beardwell CG, Pearson D, et al: The effect of varying doses of cerebral irradiation on growth hormone production in childhood. Clin Endocrinol 5:287-293, 1976

12. Dickinson WP, Berry DH, Dickinson L, et al: Differential effects of cranial radiation on growth hormone response to arginine and insulin infusion. J Pediatr 92:754-757, 1978 13. Moshang T Jr, Rundle AC, Graves DA, et al: Brain tumor recurrence in children treated with growth hormone: The National Cooperative Growth Study experience. J Pediatr 128:S4-S7, 1996 14. Ritzen EM: Does growth hormone increase the risk of malignancies? Horm Res 39:99-101, 1993 15. Watanabe S, Tsuneumatsu Y, Koniyama A, et al: Leukemia occurrence among growth hormone users. Lancet i:1159, 1988 (letter) 16. Fisher DA, Job J-C, Preece M, et al: Leukemia in patients treated with growth hormone. Lancet i:1159-1160, 1988 (letter) 17. Fradkin JE, Mills JL, Schonberger LB, et al: Risk of leukemia after treatment with pituitary growth hormone. JAMA 270:2829-3282, 1993 18. Wilton P, Price DA: Rate of relapse of craniopharyngioma in children treated with growth hormone. KIGS Biannu Rep No 8:1:4854, 1992 19. Unterman TG, Glick RP, Waites GT, et al: Production of insulin-like growth factor-binding proteins by human central nervous system tumors. Cancer Res 51:3030-3036, 1991 20. Pinski J, Schally AV, Halmos G, et al: Somatostatin analogues and bombesin/gastrin-releasing peptide antagonist RC-3095 inhibit the growth of human glioblastomas in vitro and in vivo. Cancer Res 54:5895-5901, 1994 21. Friend KE, Radinsky R, McCutcheon IE: Growth hormone receptor expression and function in meningiomas: Effect of a specific receptor antagonist. J Neurosurg 91:93-9, 1999 22. Vance ML, Mauras N: Growth hormone therapy in adults and children. N Engl J Med 341:206-216, 1999

Downloaded from ascopubs.org by 37.44.207.45 on January 20, 2017 from 037.044.207.045 Copyright © 2017 American Society of Clinical Oncology. All rights reserved.