Growth Hormone Post-Marketing Surveillance: Safety, Sales, and the Unfinished Task Ahead

Twenty-five years ago, the pediatric endocrinology community was stunned by news of Creutzfeldt-Jakob disease transmitted by treatment with pituitary-derived GH (pit-GH). With remarkable serendipity, recombinant DNA-derived human GH (rhGH) became available within months. In an environment fertile with concerns about unexpected adverse effects, postmarketing surveillance studies (PMSSs) were established, managed, and supported by manufacturers of rhGH, of which the Genentech National Cooperative Growth Study (NCGS) has been the largest and most comprehensive in scope. This action appeared prescient when, in the late 1980s, a possible link between GH therapy and new-onset leukemia was published (1). This association was later disproved (2, 3), but a number of possible rhGH-associated adverse effects were subsequently identified, systematically tracked, analyzed, and reported. In this issue of JCEM, Bell et al. (4) report more than 20 yr of NCGS safety data covering approximately 55,000 patients and nearly 200,000 patient-years of rhGH exposure. The gathering and analysis of this vast amount of drug experience information will likely not be accomplished again, making its value, in some respects, inestimable. It validates for the sponsoring company and participating physicians the rationale, commitment of resources, and hopes for the NCGS that marked its inception. The data confirm and extend prior NCGS reports (5, 6), summarized by these general clinical implications for rhGH therapy: 1) certain adverse effects associated with rapid growth (scoliosis progression, slipped epiphyses) and others of unknown mechanism (intracranial hypertension, pancreatitis) occur rarely and merit anticipatory guidance and close monitoring; 2) overt hyperglycemia is rare, but insulin sensitivity is reduced, and a possible increase in risk for type 2 diabetes mellitus (T2DM) may be obscured by its rising general pediatric population incidence; 3) in the rare occurrence when a child already at higher risk for sudden death dies during rhGH therapy, a possible contributing role of rhGH (e.g. reduced cortisol availability in hypopituitary children or airway compromise in children with Prader-Willi syndrome) cannot be excluded; and 4) rhGH does not increase risk for new malignancy in children without risk factors, but may slightly increase or hasten the onset of second malignancies in patients previously treated for cancer. Although the data on the whole are reassuring, caution is warranted when extrapolating these findings to future safety of rhGH. Like any drug, rhGH could cause adverse events that otherwise occur rarely in the reference population (e.g. malignancy) and/or increase the frequency of relatively common events (e.g. T2DM). In addition, adverse events can occur shortly after initiation of drug use, only with long-term use, or remotely after the drug has been discontinued (7). This full spectrum of potential rhGH adverse effects is not comprehensively elucidated by PMSSs due to: 1) inherent weaknesses in patient cohort surveillance dependent on physician reports; 2) changes in rhGH dosage and/or recipient characteristics that may alter risk for adverse effects; and 3) failure to capture adverse events that only become manifest after treatment. A separate but important issue unique to rhGH treatment is to define a “tolerable” level of risk for the newest and potentially largest rhGH-treated group in the future—essentially healthy, but short children. As noted by others (8), there are potential pitfalls in relying on data from PMSSs such as the NCGS. Patient enrollment is often incomplete; approximately 75% of eligible rhGH product-treated children are tracked by the NCGS (9). Drug exposure can be widely variable (some NCGS subjects received just a few injections of rhGH), complicating calculation of “patient-years” and potentially underestimating risk assessment. Variable compliance by physician “investigators” in completing and submitting forms can lead to underreporting of adverse events, and their description as related or unrelated to drug treatment is subjective. Finally, and perhaps most importantly, adverse events are only included in the NCGS risk analysis if they occur during rhGH treatment; detection of subsequent adverse effects depends on physician-initiated reports to monitoring agencies. Change in the treatment under investigation during the study period (e.g. higher rhGH dosage) complicates application of present findings to assessment of future risk. The allure of maximum height attainment and rhGH dose-responsiveness of this aim (10) have led to testing and acceptance of progressively higher rhGH doses. Most patient-years analyzed in the NCGS involve administration of rhGH doses used for hormone replacement for GH deficiency (GHD) (≤0.3 mg/kg · wk), whereas the expanding cohort of non-GHD recipients typically requires pharmacologically induced states of mild GH excess to overcome growth-restricting conditions: Turner syndrome (TS) (0.375 mg/kg · wk), intrauterine growth retardation (IUGR) (up to 0.48 mg/kg · wk), and idiopathic short stature (ISS) (0.375 mg/kg · wk). Use of even higher dosages has been reported in pubertal children with ISS (0.6 mg/kg · wk) (11). Because spontaneous GH excess is associated with known morbidities, it is plausible that high-dose rhGH treatment could increase remote metabolic or malignancy risks not detected in the NCGS. Change in recipient characteristics due to ethnic demographics and rising childhood obesity rates could lead to rhGH-induced diabetes risk that differs from NCGS estimates. Administration of rhGH to non-GHD prepubertal children is associated with increases in fasting insulin levels, mimicking (in ISS subjects) increased hepatic glucose production and fasting insulin levels during normal puberty (12). Insulin levels usually remain within the normal range and return toward baseline when rhGH is withdrawn, although decreased insulin sensitivity may persist in young women with TS (13). Even in patients at greater risk for disordered carbohydrate metabolism (e.g. TS, Prader-Willi syndrome, IUGR), rhGH-associated diabetes is rare. However, the Pharmacia KIGS PMSS reported a T2DM incidence of 34.4 cases per 100,000 yr of rhGH treatment, 6-fold higher than reported in children not treated with rhGH. T2DM did not resolve after rhGH therapy was stopped. The possibility that rhGH treatment might accelerate appearance of T2DM in predisposed individuals (14) is particularly relevant to an increasingly obese and diabetes-prone treatment population. Lack of posttreatment follow-up weakens the ability of the NCGS to detect possible effects of rhGH on both T2DM and malignancy occurrence. Mitogenic and antiapoptotic actions of GH and IGF-I create a theoretical risk that rhGH treatment could enhance posttreatment tumor development. Concerns were heightened by the report of a nearly 3-fold increase in cancer mortality (based on only four cases) in 1848 U.K. pit-GH-treated patients (15). In contrast, a report of more than 6000 patients treated with pit-GH or rhGH between 1963 and 1996 found no increase in the rate of colorectaI cancer (16). This NCGS report confirms prior studies finding no increased “on-treatment” risk of any new leukemia (2) or extracranial, nonleukemia malignancies (17) in rhGH-treated patients without risk factors for cancer. However, it does not alleviate concern that rhGH treatment has a small, but significant enhancing effect on the occurrence of second tumors in children previously treated for malignancy (18). Moreover, because studies of non-GH-treated populations suggest that high-normal levels of free IGF-I may increase rates of breast and prostate cancers, and IGF-I levels in and above the high normal range are seen in some rhGH-treated children, a potential relationship among GH, IGF-I, and neoplasia requires continued vigilance (19). The NCGS originated to monitor rhGH safety, and the report by Bell et al. (4) convincingly affirms its success in accomplishing this task. It also shows, even with the caveats raised, that rhGH treatment has, thus far, a remarkable record of safety. Along the way, however, an additional unintended outcome has emerged from the availability of the NCGS database and its use by physicians and industry—facilitation of rhGH use beyond labeled indications. The history of rhGH is a paradigm of “expansive biotechnology” wherein a biomedical technology, originally designed for treatment of disease, expands, with the encouragement of well-intended physicians and support of industry, into treatment of conditions for which the conceptual boundary between disease and variation, and therefore between treatment and enhancement, is blurred (20). Physicians who, on their own initiative, use Genentech product rhGH for “off-label” causes of inadequate growth appropriately enter these patients into the NCGS database. These single patient and small cohort experiences (usually noncontrolled and observational) are available for query, compilation, analysis, and publication by physician-investigators. This database access has led, according to a PubMed search for “NCGS,” to 34 articles about positive effects of rhGH treatment (compared with 13 about safety), often in non-Food and Drug Administration (FDA)-approved (at the time) conditions: e.g. chondrodysplasia (1996), Noonan syndrome (1996), IUGR (1996), myelomeningocele (1996), glucocorticoid dependency (1998; first author was this reviewer), ISS treated for attention deficit disorder (1998), cystic fibrosis (1997), Robinow syndrome (1999), Aarskog syndrome (1999), and ISS pubertal dosing (2009). The likely result of these favorable rhGH treatment experiences, presented and published by “thought leaders,” is reinforcement of off-label prescribing by other pediatric endocrinologists—perhaps explaining why a 2005 NCGS paper included 8018 children with ISS who began rhGH treatment before the FDA approval for ISS treatment occurred (21). The resulting changes in the composition of the ever expanding rhGH treatment cohort require ongoing adjustment and analysis of condition-specific risk-benefit relationships. Thus, whereas the NCGS report by Bell et al. (4) provides reassurance, results must be viewed in the context of increased rhGH dosage and range of recipients, spurred by industry-physician collaboration, raising new safety questions. Safety is a relative concept that is affected by illness severity and availability of alternative effective treatments (including no treatment). For a child with growth failure and GHD after recovery from central nervous system malignancy, a doubling of the risk for a second malignancy from approximately three to approximately six per 1000 patient-years (18) may be an acceptable trade-off for growth restoration. For a child with ISS, on the other hand, even the smallest risk for a long-term adverse effect may not be outweighed by an unpredictable and poorly defined benefit for an otherwise healthy child. To confirm the negligible risk considered tolerable in this situation will require many years of posttreatment follow-up. This task will not be accomplished by the NCGS, which, as recently announced by Genentech, will be closed, due to the judgment that additional data collection and analysis will not provide new insights to answer safety and efficacy questions. Soon after prior constraints of pit-GH therapy were removed by the abundance of rhGH, the notion was put forth that, for height promotion, the degree of disability and responsiveness to rhGH treatment (rather than the underlying etiology of short stature) should be the keys to determining entitlement (22). Although subsequent FDA approval of rhGH treatment for several non-GHD conditions validated that position, new questions have emerged about the value of rhGH-induced height promotion in improving health and well-being and its justification when compared with costs and (known and unknown) risks. Given the modest effects of rhGH on improving adult height in most patients without GHD and still unproven beneficial effects on quality of life, the question posed then—will rhGH treatment for short stature be a panacea or Pandora’s box?—has been partially answered. To show now that rhGH therapy does not open a door to future unforeseen adverse outcomes, the global pediatric endocrine community must pick up where the NCGS and other rhGH PMSSs leave off, by conducting collaborative follow-up of children long after their treatment with rhGH is completed. Disclosure Summary: The author’s institution has received grant support from Genentech, Inc. for participation in the NCGS. GH deficiency idiopathic short stature intrauterine growth retardation pituitary-derived GH postmarketing surveillance study recombinant human GH type 2 diabetes mellitus Turner syndrome