Risk of Neoplasia in Pediatric Patients Receiving Growth Hormone Therapy—A Report From the Pediatric Endocrine Society Drug and Therapeutics Committee

Abstract

Context:

GH and IGF-1 have been shown to affect tumor growth in vitro and in some animal models. This report summarizes the available evidence on whether GH therapy in childhood is associated with an increased risk of neoplasia during treatment or after treatment is completed.

Evidence Acquisition:

A PubMed search conducted through February 2014 retrieved original articles written in English addressing GH therapy and neoplasia risk. Subsequent searches were done to include additional relevant publications.

Evidence Synthesis:

In children without prior cancer or known risk factors for developing cancer, the clinical evidence does not affirm an association between GH therapy during childhood and neoplasia. GH therapy has not been reported to increase the risk for neoplasia in this population, although most of these data are derived from postmarketing surveillance studies lacking rigorous controls. In patients who are at higher risk for developing cancer, current evidence is insufficient to conclude whether or not GH further increases cancer risk. GH treatment of pediatric cancer survivors does not appear to increase the risk of recurrence but may increase their risk for subsequent primary neoplasms.

Conclusions:

In children without known risk factors for malignancy, GH therapy can be safely administered without concerns about an increased risk for neoplasia. GH use in children with medical diagnoses predisposing them to the development of malignancies should be critically analyzed on an individual basis, and if chosen, appropriate surveillance for malignancies should be undertaken. GH can be used to treat GH-deficient childhood cancer survivors who are in remission with the understanding that GH therapy may increase their risk for second neoplasms.

Pituitary-derived GH (pitGH) was initially used in humans in the 1950s, and recombinant human GH (rhGH) has been approved for clinical use in the United States since 1985 (1). As early as the 1970s, safety concerns have been raised about whether GH treatment increases the risk of cancer (2). Earlier case reports of patients developing leukemia while on GH (3) have been followed by epidemiological, in vitro, and animal model studies providing some biological plausibility for an association between GH, IGF-1, and cancer (4).

The purpose of this report is to provide a review of the clinical evidence analyzing the association between GH treatment during childhood and development of neoplasia during or after completion of GH therapy. Research in this area has focused on the development of malignancies during treatment; however, longer-term risks are discussed where evidence exists. Particular attention is paid to groups at higher risk for malignancies, such as those with prior cancers and those with conditions known to predispose to cancer.

Methods

A PubMed search was performed (up to February 2014) for original articles using search terms such as: “growth hormone (GH) treatment AND cancer,” “IGF-1 AND cancer,” “GH AND second malignancy,” and “pediatric cancer survivor AND GH.” Reports of the safety of GH treatment in patients with 19 conditions known to predispose to neoplasia took the following format: “name of condition AND GH.”¹ A secondary review of reference lists led to the identification of additional relevant articles. All articles identified were published in the English language. Key papers published in the older literature were included to ensure completeness and to provide historical perspective.

Molecular Signaling Within the GH/IGF-1 Axis and Cancer Risk

The GH receptor (GHR) is a single transmembrane class I cytokine receptor with no intrinsic tyrosine kinase activity (5, 6). Upon binding GH, the GHR activates downstream intermediates such as Janus kinase 2 and signal transduction and activator of transcription 5, ultimately leading to transcription of GH-dependent proteins such as IGF-1, acid labile subunit, and IGF binding protein-3 (IGFBP-3). IGF-1 and IGF-2 both activate the type 1 IGF receptor (IGF1R), a transmembrane heterotetramer with an intrinsic intracellular tyrosine kinase domain (7). IGF1R activation leads to phosphorylation cascades, particularly of MAPK and phosphatidylinositol 3′-kinase, that lead to cell proliferation, survival, and differentiation via effects on the nucleus and mitochondria (8).

Beyond this classical model of GH/IGF-1 axis function, additional levels of complexity enrich the system's ability to signal and coordinate with other regulators of cell fate. For example, the GHR can directly activate other intracellular signaling pathways, and GH can induce production of other growth factors (9, 10). The GH/IGF system interacts with other pathways, such as epidermal growth factor, to modulate each other's signaling (11). Furthermore, GH and IGF-1 effects can be induced by circulating hormone or by local autocrine/paracrine mechanisms (12).

The GH/IGF axis has been implicated in cancer progression, including enhanced tumorigenesis, metastasis, resistance to chemotherapy and radiotherapy, and transformation (13). Underlying mechanisms include increased expression of ligands and receptors, loss of heterozygosity of the IGF-2 locus, and increased IGF1R gene copy number (12). Alterations in expression of the IGFBPs or in local production of IGFBP proteases also modify the local concentrations of bioavailable IGF. Additionally, the GH/IGF system interacts with oncogenes and tumor suppressors. For example, IGF-1 was required for transformation of mouse embryonic fibroblasts by Simian Virus 40 (SV40) large tumor antigen, a proto-oncogene derived from the polyomavirus SV40 (14). Conversely, some tumor suppressors, whose loss of function underlies certain tumor predisposition syndromes (15–29), function in part by inhibiting IGF action (Table 1). Because of the many roles the components of the GH/IGF axis may play in cancer progression, it has become the target of investigation for potential cancer therapy (30).

Table 1.

Interactions Between Tumor Suppressors and GH/IGF System Signaling

Tumor Suppressor Gene	OMIM No.	Cancer Predisposition/ Tumor Syndrome	Signaling Interaction With the GH/IGF-I Axis
WT1	607102	Denys Drash syndrome, Frasier syndrome	WT1 suppresses transcription of IGF1R (16)
TP53	191170	Li-Fraumeni syndrome	p53 suppresses transcription of IGF1R (17) and IGF2 (15)
			p53 induces transcription of IGFBP3 (23) and IGFBP2 (19)
PTEN	601728	PTEN hamartoma tumor syndrome	PTEN dephosphorylates (ie, inactivates) AKT, whose phosphorylation by phosphatidylinositol 3-kinase (PI3-K) construes one of the major IGF1R signaling pathways (22, 25)
VHL	608537	von Hippel Lindau disease	VHL leads to ubiquitination of HIF1α (hypoxia-inducible factor 1α) and decreased production of VEGF (vascular endothelial growth factor); IGF-1 induces VEGF, which can contribute to tumor growth by stimulating angiogenesis (27, 29)
			VHL also leads to dissociation of PKCδ (protein kinase Cδ) from IGF1R, inhibiting IGF1R signaling in renal cancer (20)
APC	611731	APC-associated polyposis conditions	APC is a negative regulator of the canonical Wnt signaling pathway by functioning in a multiprotein destruction complex that proteolyzes β-catenin, whereas IGF-1 stabilizes β-catenin protein and was shown to enrich human colon cancer stem cells in a β-catenin dependent manner (18, 21, 24, 26, 28)

Although there has been no molecular evidence to date that GH or IGF-1 cause the initial development of cancer, there is ample evidence that they can function as permissive factors making aberrant cells more aggressive (31). Yet, raising IGF-1 levels with rhGH treatment does not necessarily translate into increased cancer risk for three reasons: 1) the relative contribution of systemic vs local IGF-1 levels to carcinogenesis is unresolved; 2) GH also induces IGFBP-3, which may play a protective role through competitive inhibition of IGF/IGF1R signaling and through proapoptotic actions (32); and 3) the temporal relationship between IGF-1 elevation and vulnerability to increased cancer risk is unknown (10).

The Epidemiological Association Between High GH States and Development of Cancer

Epidemiological studies of adults without GH excess disorders have shown that persons with IGF-1 levels in the upper normal range have increased rates of certain neoplasms compared to those with IGF-1 levels in the lower normal range. Although some studies, including meta-analyses, have demonstrated an association between high IGF-1 states and the occurrence of prostate cancer (4, 33), colorectal polyps and cancer (34, 35), and premenopausal breast cancer (4, 36), other more recent studies have failed to find such associations, particularly for prostate cancer (37, 38). Furthermore, single nucleotide polymorphisms expected to result in increased IGF-1 concentrations were not associated with higher rates of colorectal cancer (39). The extent to which genetic polymorphisms in the GH/IGF-1 signaling pathway may modulate the GH/IGF-1 axis and interact with the environment to impact cancer risk has not been fully elucidated and remains an area of investigation (40–42).

Additionally, epidemiological studies have suggested an association between acromegaly and risk for neoplasia, including colorectal polyps and cancer (43–45), central nervous system (CNS) tumors, and thyroid cancer (46–51). High IGF-1 levels have been proposed as a mechanism predisposing patients with acromegaly to the development of cancer; however, this has not been demonstrated conclusively (31, 52, 53), and the causative role of IGF-1 and GH in oncogenesis remains controversial (54). In fact, IGF-1 has failed to induce neoplasia in colonic epithelia from acromegalics in vitro (55). Alternate mechanisms, such as genomic instability and increased DNA oxidative damage, have been proposed as well (56). Furthermore, treatment of some patients with acromegaly with cranial irradiation potentially confounds any association with CNS neoplasms. Finally, there may be unidentified underlying risk factors, such as genetic polymorphisms, that could simultaneously predispose to acromegaly and other neoplasia (52).

Conversely, when 222 patients with severe congenital IGF-1 deficiency were compared to relatives without IGF-1 deficiency, none of the IGF-1-deficient subjects had a history of cancer, compared with 9–24% of their relatives (57). Similar results were reported in an Ecuadorian cohort with IGF-1 deficiency due to a GHR mutation (58). Although these reports support an association between low IGF-1 and decreased risk for neoplasia, the cohorts are too small to draw definitive conclusions.

Risk of De Novo Leukemia and Other Cancers in Patients Without Tumor Predisposition Syndromes, Prior Cancers, or Other Risk Factors for Malignancy

After the initial report of leukemia in a Japanese patient treated with GH (59), the risk of leukemia associated with GH treatment in children who lack known risk factors for the development of cancer has been of interest (60–67). Although the risk of nonhematogenous de novo cancer has not been studied as extensively as the risk of leukemia in pediatric patients on rhGH, the issue has been addressed by several GH postmarketing studies (67–70). Table 2 summarizes the published clinical data describing risk of neoplasia in children treated with GH. In most cases, these are postmarketing surveillance studies (64–67) with numerous subjects and many person-years of follow-up. However, there are limitations inherent in their design, including documentation of adverse events by voluntary, sporadic reports; lack of follow-up after completion of rhGH treatment; and lack of truly comparable control populations, necessitating the use of historical controls. In general, these studies have found no association between GH treatment in pediatric patients and the concurrent development of leukemia or other malignancies.

Table 2.

Studies Addressing GH Therapy and De Novo Cancer Risk in Patients Treated With GH During Childhood

First Author, Year (Ref.)	GH Source	Study Design	Data Source	Control Subjects	Case Subjects or Cohort Description	Outcome Variables	Results and Conclusions
Watanabe, 1989 (60)	pitGH +/− rhGH	Case control study	GH-treated patients in Japanese clinics	10 matched controls of GH users	5 cases of leukemia among GH users between 1985 and 1987	Risk factors for leukemia in GH users	Duration and total dose of GH did not show a dose-dependent relationship with leukemia occurrence.
Stahnke, 1992 (61)	pitGH +/− rhGH	Case series	Compilation of cases of leukemia in GH users observed worldwide	General population in respective countries	Multinational cohort of 31 patients with leukemia during or after completion of GH therapy; 15 had other risk factors for leukemia	Risk for leukemia in GH users; incidence rate was compared to expected rates based on general incidence of leukemia	GH treatment is most probably not inducing leukemia.
Fradkin, 1993 (62)	pitGH +/− rhGH	Cohort study	Patient interviews, self-reports, and medical records	SEER (general population in United States)	US-based cohort of 6284 patients (82% had no organic cause for GHD) 59 376 patient-years	Risk for leukemia and lymphoma; incidence rates were compared to expected rates based on SEER data	GH does not increase risk for leukemia/lymphoma in patients with idiopathic GH deficiency.
Nishi, 1999 (63)	pitGH +/− rhGH	Cohort study	Japanese GH treatment study committee	General population in Japan	Japanese cohort of 32 000 patients (no. of patients without neoplasia risk factors was not reported); 9 of 15 patients developing leukemia had no risk factors; of these, 5 developed leukemia 0.9–9.8 y after stopping GH	Risk for leukemia during GH therapy and for patient-years at risk including the years after cessation of GH therapy	In patients without risk factors, SIR for leukemia during GH therapy was 0.75–1.00 (95% CI, 0.2–2.57) and during entire follow-up was 0.97–1.29 (95% CI, 0.44–2.45). GH does not increase risk for leukemia in patients without risk factors.
Swerdlow, 2002 (68)	pitGH	Cohort study	Data obtained from medical/research records and national registries	British national cancer incidence and mortality rates in general population specific by age, sex, and calendar year	UK-based cohort of 1849 patients (53% had idiopathic GHD; 496 patients had high-risk factors). Data from 1959–1985. Analysis completed for the entire cohort and for those without high-risk factors	Cancer incidence (SIR) and cancer mortality (SMR)	Excluding the 496 high-risk patients, SIR for colorectal cancer was 11.1 (95% CI, 1.3–39.9) based on 2 patients with colon cancer. Other cancer incidence rates were not increased. When all cancer types were combined, SIR and SMR were not significant at 1.4 (95% CI, 0.5–2.8) and 2.3 (95% CI, 0.8–5.0). Two cases are too few to allow firm conclusions. These effects cannot be generalized to rhGH recipients.
Tuffli, 1995 (70)	rhGH	Postmarketing surveillance study	NCGS postmarketing surveillance study	SEER (general population in United States)	US and Canadian cohort of 12 209 patients; 51 000 patient-years; 10 new cases of nonleukemic extracranial neoplasms reported	Incidence of nonleukemic extracranial neoplasms	GH does not increase risk for nonleukemic extracranial neoplasms.
Allen, 1997 (64)	rhGH	Postmarketing surveillance study	NCGS postmarketing surveillance study	SEER (general population in United States)	US and Canadian cohort of 21 705 patients with no risk factors; 67 773 patient-years of GH therapy, 119 846 patient-years of risk	SIR for leukemia during GH therapy and for patient-years at risk, including the years after cessation of GH therapy	SIR for neoplasm during patient-years of GH therapy was 0.94 (95% CI, 0.11–3.4). SIR for neoplasm during patient-years at risk was 0.88 (95% CI, 0.18–2.57). GH does not increase risk for leukemia in patients without risk factors.
Bell, 2010 (67)	rhGH	Postmarketing surveillance study	NCGS postmarketing surveillance study; data from 1985–2006	Age-matched general population	US and Canadian cohort of 54 996 patients (42.5% idiopathic GHD); 178 464 patient-years of GH therapy among those without risk factors; 15 new cases of leukemia during GH therapy, of which 3 patients had no risk factors	Risk for leukemia during GH therapy in patients without risk factors	SIR for de novo leukemia in patients without risk factors was 0.54 (95% CI, 0.11–1.58). SIR for de novo malignancies in these patients was 1.12 (95% CI, 0.75–1.61). No significant increase in risk for leukemia or de novo malignancies among GH users without additional risk factors during GH therapy.
Popovic, 2010 (66)	rhGH	Postmarketing surveillance study	KIMS postmarketing surveillance study in adults	325 patients with idiopathic GHD (34% adult-onset GHD) but without malignancy in KIMS database	Multinational cohort of 100 patients with de novo malignancy (96% with adult-onset GHD)	Association between IGF-1, IGFBP-2, IGFBP-3, and de novo cancer risk in adults receiving GH therapy	IGF-1 levels within normal reference range during GH therapy were not associated with de novo malignancy; high IGFBP-2 and IGFBP-3 levels may be associated with increased cancer risk.
Wilton, 2010 (69)	rhGH	Postmarketing surveillance study	KIGS postmarketing surveillance study	Cancer incidence among general population in respective countries	Multinational cohort of 58 603 patients with no risk factors (54% idiopathic GHD); 197 173 patient-years of follow-up	SIR for any de novo cancer during GH therapy	SIR for new malignant neoplasms was 1.26 (95% CI, 0.86–1.78). GH does not increase risk for de novo cancer in patients without risk factors.

Abbreviations: SEER, Surveillance, Epidemiology and End Results; NCGS, National Cooperative Growth Study; KIGS, Pfizer International Growth Database; KIMS, Pfizer International Metabolic Database. Patient-years on GH therapy indicates the duration patients were observed between the start and the end of GH therapy. Patient-years at risk indicates the duration patients were observed between the start of GH therapy and the completion of follow-up (includes years of follow-up off GH therapy).

Additionally, there are insufficient data in children to support the practice of titrating rhGH doses to normalize IGF-1 levels as a means to modify the risk of malignancy. However, in patients who are treated with rhGH for GH deficiency, it may be reasonable to attempt to maintain age- and Tanner stage-appropriate levels of IGF-1 in children undergoing treatment with rhGH; the potential risks conferred by titrating to higher levels are not well-defined. This practice has also been proposed for the management of idiopathic short stature and small for gestational age patients (71–73), although evidence that this practice improves safety outcomes is not available.

Taking a longer view, one considers whether pediatric GH treatment affects cancer risk later in adulthood. Among 1352 pediatric patients treated with pitGH in the United Kingdom compared with the general population, there was no increase in the standardized incidence rate (SIR) for cancers at all sites, but the SIR for colorectal cancer was significantly increased. It is notable that this observation was based upon the occurrence of two cases of colorectal cancer, one of which occurred in an individual with possible familial adenomatous polyposis (68).

In another study, when adults with hypopituitarism were followed prospectively during GH treatment for a mean duration of 59.9 months, their risk of developing a neoplasm was similar to that of the general population; however, the GH-treated cohort was small (n = 289), and the duration of follow-up was relatively brief (74). Likewise, cancer incidence was not significantly elevated in GH-treated adult hypopituitary subjects (n = 6840; mean follow-up, 3.7 y) within the multinational Hypopituitary Control and Complications Study (HypoCCS), with the exception of increased incidence in the subgroup of those with childhood-onset hypopituitarism treated in the United States (SIR, 2.74; 95% confidence interval [CI], 1.18–5.41) (75). HypoCCS included subjects with hypopituitarism due to congenital causes and neoplastic causes. Some of the observed cases contributing to this elevated SIR were subsequent neoplasms. When all HypoCCS subjects treated for childhood-onset GH deficiency (COGHD; n = 1204, including 389 diagnosed with idiopathic COGHD) were analyzed during a mean follow-up of 3.7 years, there was no increased risk of cancer incidence; the overall cancer SIR for COGHD was 0.27 (95% CI, 0.01–1.50), and there were no new cancer diagnoses among the idiopathic COGHD subjects (76).

A recent systematic review of literature through September 2013 raised concern about a statistically significant increase in overall cancer incidence (SIR, 2.74; 95% CI, 1.18–5.41) and second neoplasms (relative risk [RR], 1.99; 95% CI, 1.28–3.08); however, the authors cautioned that several confounders and biases may affect this analysis (77).

Thus, there is no strong, conclusive association between pediatric GH treatment and the development of cancer in GH-deficient adults; however, further research is needed.

Cancer-Related Mortality in Patients Without Tumor Predisposition Syndromes, Prior Cancers, or Other Risk Factors for Malignancy Treated With rhGH

Another clinically relevant question is whether rhGH treatment results in higher rates of cancer-related mortality. A UK-based study comparing cancer risk and mortality in 1849 patients who received pitGH to the general population found an increased standardized mortality ratio (SMR) due to colorectal cancer and Hodgkin disease (68). The number of events in the cohort overall was small, and the influence of the duration of GH treatment could not be assessed because, although treatment duration with pitGH was documented, similar data for rhGH treatment were not available. More recently, reports from the Safety and Appropriateness of Growth Hormone Treatments in Europe (SAGhE) study showed conflicting results from two different cohorts within this group. In the French cohort (78), among 6465 subjects with a history of GH treatment in childhood, but who did not have additional risk factors for malignancy (included diagnoses were idiopathic short stature, idiopathic GH deficiency, small for gestational age, or neurosecretory dysfunction), there was no increase in overall cancer-related mortality but a significant increase in bone tumor-related mortality (SMR, 5.0; 95% CI, 1.01–14.63). The bone tumor deaths in the French study occurred in three subjects, and analysis of a similar cohort from three other European countries (2543 subjects) demonstrated no increase in overall cancer-related mortality or bone tumor-related mortality (79).

GH Treatment and Cancer Risk in Patients With Conditions Known to Predispose to Tumor Development

There are a number of monogenic tumor syndromes that predispose an affected individual to malignancies (80–83). In some cases, the syndrome itself, such as neurofibromatosis type 1 (NF1), is associated with an increased prevalence of short stature and GH deficiency (84–89). In most tumor-predisposition syndromes, however, GH secretion abnormalities are not inherent to the primary disease, limiting the available data on rhGH treatment in such patients.

From a clinical perspective, despite NF1 being the best-studied tumor-predisposition syndrome with respect to rhGH safety, and whereas GH treatment does not appear to worsen the clinical manifestations of NF1 (90), few studies have reported on cancer risk. GH receptors have been identified in plexiform neurofibromas of NF1 (91). However, plexiform neurofibroma tumor burden does not increase with advancing puberty (92), a time of physiologically increased GH secretion.

Noonan syndrome is an approved indication for rhGH treatment in the United States; however, Noonan syndrome, specifically mutations in PTPN11, has been associated with increased risk of de novo malignancy compared to the general population, regardless of the use of GH therapy (93). This association is plausible because the causative mutations in Noonan syndrome disrupt the RAS-MAPK pathway, which is important for cellular differentiation and proliferation. Whether rhGH treatment further augments this risk is not known.

Short stature and GH deficiency (GHD) are also prevalent in Fanconi anemia (85, 89), a DNA-breakage syndrome associated with an increased risk of malignancies (94). Treatment of such children with rhGH after successful hematopoietic cell transplantation has been reported to be effective in a small series (95). Although there is currently insufficient evidence to conclude whether GH treatment increases cancer risk in patients with Fanconi anemia, recent guidelines have discussed the use of rhGH treatment in patients who are GH deficient, making note that, even after transplant, patients retain a higher risk of malignancy relative to the general population (96).

Although not a monogenic hereditary disorder, inflammatory bowel disease (IBD) is a chronic illness associated with higher risk for malignancy, particularly colorectal cancer (97, 98). Additionally, some treatments for IBD, such as thiopurines, predispose IBD patients to leukemia and myelodysplastic syndrome (99). Although not approved in the United States for this indication, rhGH treatment has been proposed by some as a means to improve lean body mass and linear growth in pediatric patients with IBD (100–102). It is not known whether rhGH treatment in children with IBD affects their risk of developing cancer.

Unfortunately, there is insufficient data available for clinicians to make evidence-based decisions regarding the safety of rhGH therapy in children at risk for malignancies because of an underlying genetic disease or a predisposing medical condition. In the absence of such evidence, clinicians should consider whether raising IGF-1 levels might constitute an additional risk for neoplasia in patients predisposed to tumor development (Table 1). In those patients in whom IGF-1 could play a role in tumor growth based on their underlying condition, if treatment with rhGH is initiated, maintaining IGF-1 levels in an age-appropriate range and undertaking routine tumor screening according to guidelines available for a given tumor-predisposition condition is prudent despite a lack of evidence to support or refute this practice.

GH Treatment and the Risk of Recurrence in Pediatric Cancer Survivors

Whether or not treatment with GH alters the risk of recurrence of the primary cancer in survivors of pediatric cancer has long been a concern of patients, caregivers, and providers. The Childhood Cancer Survivor Study (CCSS) reported on the risk of disease recurrence with GH treatment across many pediatric cancer diagnoses in a large, multi-institutional North American cohort that includes more than 14 000 5-year survivors of pediatric cancer (103). In the CCSS, the use of GH treatment was validated (n = 361), and the RR of recurrence in those treated with GH replacement therapy was 0.83 (95% CI, 0.37–1.86), compared to survivors not treated with GH (104). Furthermore, the RR of recurrence was not increased for any specific cancer diagnosis.

Survivors of medulloblastoma, most of whom are treated with craniospinal irradiation, are the most common group of CNS tumor survivors to receive treatment with GH. In a study of 545 medulloblastoma survivors, there was no worsening of event-free survival and no increased recurrence risk among the 170 survivors treated with GH (105). Additionally, several other studies have reported no increased risk of recurrence with GH treatment in survivors of malignant brain tumors (68, 106–108).

Several reports have also shown no increased risk of recurrence of leukemia with GH treatment. In a review of 910 patients treated for acute lymphoblastic leukemia (ALL) at St. Jude Children's Hospital from 1978–1989, 47 patients received rhGH therapy for GH deficiency. When compared to those patients who did not receive GH therapy, GH therapy was not associated with an increased risk for leukemia relapse (109). The median time after remission for GH therapy initiation was 7 years (range, 4.3–11.4 y), and the median duration of GH therapy was 4.5 years (range, 1–8 y).

Although no studies specifically address how long one should wait between completion of cancer therapy and initiation of GH treatment, clinical guidelines from the Pediatric Endocrine Society (110) and expert opinion (111), including those of the authors, suggest waiting 1 year after cancer treatment is complete to ensure that there is not an early cancer recurrence. A retrospective review of medulloblastoma survivors showed no difference in the risk for recurrence among those treated with GH 1, 2, or 3 years after completion of cancer therapy (105). Reported patterns of clinical practice have been similar (Ref. 112; and personal communication from B. S. Miller). In addition, it should be acknowledged that tumor recurrence beyond 1 year remains possible with or without GH treatment. Other factors to consider when deciding the timing of GH therapy include the chronological and skeletal ages, pubertal status, current height, primary tumor type, overall oncological prognosis, risk of relapse, and the goals of the patient and caregivers.

GH Treatment and the Risk of Subsequent Neoplasms in Pediatric Cancer Survivors

Survivors of pediatric cancer are at risk for GH deficiency primarily due to tumor location or due to cranial, craniospinal, or whole-body irradiation (113). Such patients may also be at increased risk for the development of subsequent primary neoplasms due to prior chemotherapy and/or radiotherapy. Multiple studies have addressed whether or not GH treatment augments this risk.

Within the CCSS, irrespective of GH treatment, pediatric cancer survivors have been reported to be at increased risk to develop a new neoplasm compared with the general population not treated for childhood cancer (SIR, 6.0; 95% CI, 5.5–6.4) (114). The most common types of subsequent neoplasms were nonmelanoma skin cancer and breast cancer. Three reports from the CCSS have addressed the association of GH treatment and the development of subsequent primary tumors (104, 115, 116). When pediatric survivors within the CCSS were evaluated, the risk of any subsequent neoplasm associated with exposure to GH treatment was initially reported to be elevated with RR = 3.21 (95% CI, 1.88–5.46) after a median follow-up of 6.2 years (104). Second neoplasm types included meningioma, osteogenic sarcoma, glial tumors, and others. After a 32-month extension of follow-up within the same cohort, this was revised to a RR of 2.15 (95% CI, 1.3–3.5) (116). Likewise, in reports from GH registries, 5-year cumulative incidence of subsequent neoplasms in patients treated with GH was 6.2% in pediatric participants in GeNeSIS (Genetics and Neuroendocrinology of Short Stature International Study) and 4.8% in adult participants in HypoCCS. These findings were thought to be concordant with the prior CCSS findings, but no control group was available (117).

Irradiation is a well-established risk factor for subsequent intracranial neoplasms and GHD. Hence, several studies have reported on the risk of subsequent primary neoplasms in the CNS, specifically meningioma and glioma (118). Secondary gliomas occur much earlier and may present while patients are receiving GH for pediatric indications. On the other hand, meningioma more commonly occurs long after cancer treatment, and cases of secondary meningiomas are still accumulating in this cohort after more than 30 years of follow-up. Nevertheless, in an updated analysis from the CCSS focused exclusively on the risk of secondary CNS tumors, there was no increase in the rates of meningioma, glioma, or any CNS subsequent neoplasm associated with GH treatment after adjusting for the CNS radiation dose and the duration of follow-up after radiotherapy (115). This analysis differed from prior CCSS reports in that the duration of follow-up was longer, specific types of subsequent neoplasm outcomes were analyzed separately (eg, meningioma), and the results of the multivariate model were controlled for cranial radiation dose, as opposed to a dichotomized radiation exposure variable.

Furthermore, in a cohort of pediatric (n = 41) and adult (n = 69) patients treated with CNS irradiation, after a median follow-up of 14.5 years, there was no increase in the occurrence of subsequent primary tumors in those treated with GH (median duration, 8 y) compared to matched controls (119). Likewise, in a single institution cohort of nonpituitary brain tumor patients treated with adult GH replacement (n = 60; 75% diagnosed at <20 y of age, 23 started GH therapy during childhood), there was no association between GH treatment and the development of subsequent neoplasms after a median follow-up time of 17.4 years (120).

Distinct from brain tumor survivors, in 910 survivors of childhood ALL in a single institution study with 47 survivors treated with GH, there was no association between GH treatment and the development of second malignancies after 11 years of follow-up (109).

Thus, although some studies have found no association between GH treatment and the occurrence of a second primary neoplasm in pediatric cancer survivors, others have reported an increased risk. The effect of GH on this risk appears to decline with longer duration of follow-up, and the relationship between pediatric GH treatment and the development of subsequent neoplasms many years after completion of GH treatment remains inconclusive. Likewise, the role of adult GH replacement on the development of subsequent neoplasms in pediatric cancer survivors warrants additional investigation.

GH Treatment and the Risk of Recurrence in Craniopharyngioma Patients

Craniopharyngioma, a sellar and/or suprasellar CNS tumor that is histologically nonmalignant, has a high rate of recurrence after complete resection and a high rate of progression after incomplete resection (121). Craniopharyngioma is frequently associated with GH deficiency. Postmarketing surveillance studies have reported on the rates of craniopharyngioma recurrence (67, 108). The National Cooperative Growth Study (NCGS) reported that, among craniopharyngioma patients treated with GH, the recurrence rate was 8.7% (67). Recurrence rates in survivors of intracranial neoplasms were higher in the Pfizer International Growth Database (KIGS) database, likely due to longer duration of follow-up, but they were not thought to differ from historical controls (108). Both of these studies are limited by the lack of appropriate control populations.

Several short- and long-term cohort studies in adult and pediatric-aged craniopharyngioma patients have failed to demonstrate an adverse effect of GH therapy on recurrence rates, disease progression, or event-free survival (121–125). Some of these studies did not include untreated controls, and in others subjects were not randomized to GH or no GH treatment. Any bias introduced by nonrandom selection of subjects for treatment could not be accounted for in the analysis (121).

It is unclear from current research how long GH-deficient patients with craniopharyngioma should be observed for stability after initial treatment of the tumor before initiating GH treatment. Finally, some patients with craniopharyngioma are treated with irradiation, increasing their risk to develop subsequent primary neoplasms; the impact of GH on this risk in craniopharyngioma patients has not been reported.

Summary of Findings

Although GH and IGF-1 signaling pathways are implicated mechanistically in tumor growth and in the pathogenesis of some tumor predisposition syndromes, the available clinical and epidemiological data do not conclusively establish a causal role for GH treatment in the development of new malignancies. Based on the available clinical evidence, we summarize the following:

• 1. In general, data do not support an association between GH treatment and risk of de novo development of a cancer in an otherwise healthy child with no prior history of cancer and no known predisposition to cancer.
• 2. In patients with known tumor predisposition due to genetic or other medical conditions, patients and caregivers should be informed that the risks of developing cancer associated with GH treatment have not been adequately studied in these populations. Consideration of mechanistic plausibility for a possible effect and routine surveillance for malignancies based on condition-specific guidelines are appropriate.
• 3. GH-deficient pediatric cancer survivors who have completed cancer therapy and are without signs of active neoplastic disease may be candidates for rhGH treatment without concern that the GH treatment raises their intrinsic risk of recurrence. Patients and caregivers should be informed that there may be an increase in the risk overall for subsequent primary neoplasms.
• 4. The association between rhGH treatment in adulthood and development of malignancies may differ from that in patients treated with rhGH for pediatric indications. This issue was not addressed in this manuscript.

Conclusions

Despite plausible molecular mechanisms and epidemiological evidence associating increased IGF-1 levels with the development of cancer, GH treatment during childhood and adolescence has not been associated with the de novo development of cancer. There may be a small increased risk for subsequent neoplasms overall in pediatric cancer survivors; this risk decreases with time. However, there is no specific increase in the rates of subsequent CNS neoplasms. As the proposed indications for GH treatment expand, further research should investigate whether or not other underlying medical conditions modify the risk of GH treatment.

Because most of the cancers associated with acromegaly and/or high GH/IGF-1 levels are cancers seen predominantly in adults, long-term studies are needed to determine whether treatment with rhGH in childhood is associated with an increased risk for the common adult cancers. Additionally, further investigation is needed to determine whether adult GH replacement or the duration of treatment is associated with the development of malignancies, particularly for those tumor types previously reported to be associated with high levels of GH and/or IGF-1.