Meta-analysis of clinical trial safety data in a drug development program: answers to frequently asked questions.

BACKGROUND: Meta-analyses of clinical trial safety data have risen in importance beyond regulatory submissions. During drug development, sponsors need to recognize safety signals early and adjust the development program accordingly, so as to facilitate the assessment of causality. Once a product is marketed, sponsors add postapproval clinical trial data to the body of information to help understand existing safety concerns or those that arise from other postapproval data sources, such as spontaneous reports. PURPOSE: This article focuses on common questions encountered when designing and performing a meta-analysis of clinical trial safety data. Although far from an exhaustive set of questions, they touch on some basic and often misunderstood features of conducting such meta-analyses. METHODS: The authors reviewed the current literature and used their combined experience with regulatory and other uses of meta-analysis to answer common questions that arise when performing meta-analyses of safety data. RESULTS: We addressed the following topics: choice of studies to pool, effects of the method of ascertainment, use of patient-level data compared to trial-level data, the need (or not) for multiplicity adjustments, heterogeneity of effects and sources of it, and choice of fixed effects versus random effects. LIMITATIONS: The list of topics is not exhaustive and the opinions offered represent only our perspective; we recognize that there may be other valid perspectives. CONCLUSIONS: Meta-analysis can be a valuable tool for evaluating safety questions, but a number of methodological choices need to be made in designing and conducting any meta-analysis. This article provides advice on some of the more commonly encountered choices.

Comparison of Bayesian and frequentist meta-analytical approaches for analyzing time to event data.

Using meta-analysis in health care research is a common practice. Here we are interested in methods used for analysis of time-to-event data. Particularly, we are interested in their performance when there is a low event rate. We consider three methods based on the Cox proportional hazards model, including a Bayesian approach. A formal comparison of the methods is conducted using a simulation study. In our simulation we model two treatments and consider several scenarios.

Planning and core analyses for periodic aggregate safety data reviews.

BACKGROUND: In 2009, the Safety Planning, Evaluation and Reporting Team gave detailed recommendations for a well-planned and systematic approach for safety data collection and analysis. Important aspects of this approach included regular reviews of aggregate data by a multidisciplinary team focusing on safety. PURPOSE: This article provides information to facilitate the planning and implementation of aggregate data reviews. METHODS: Our recommendations are based on experience of the authors and review of relevant literature. RESULTS: We present information regarding the planning of aggregate data reviews as well as examples of data displays that are useful for many different compounds. A subset of these data displays could form a set of 'core' analyses to be generated for aggregate data reviews.

Comparison of several imputation methods for missing baseline data in propensity scores analysis of binary outcome.

We performed a simulation study comparing the statistical properties of the estimated log odds ratio from propensity scores analyses of a binary response variable, in which missing baseline data had been imputed using a simple imputation scheme (Treatment Mean Imputation), compared with three ways of performing multiple imputation (MI) and with a Complete Case analysis. MI that included treatment (treated/untreated) and outcome (for our analyses, outcome was adverse event [yes/no]) in the imputer's model had the best statistical properties of the imputation schemes we studied. MI is feasible to use in situations where one has just a few outcomes to analyze. We also found that Treatment Mean Imputation performed quite well and is a reasonable alternative to MI in situations where it is not feasible to use MI. Treatment Mean Imputation performed better than MI methods that did not include both the treatment and outcome in the imputer's model.

Recommendations for safety planning, data collection, evaluation and reporting during drug, biologic and vaccine development: a report of the safety planning, evaluation, and reporting team.

BACKGROUND: The Safety Planning, Evaluation and Reporting Team (SPERT) was formed in 2006 by the Pharmaceutical Research and Manufacturers of America. PURPOSE: SPERT's goal was to propose a pharmaceutical industry standard for safety planning, data collection, evaluation, and reporting, beginning with planning first-in-human studies and continuing through the planning of the post-product-approval period. METHODS: SPERT's recommendations are based on our review of relevant literature and on consensus reached in our discussions. RESULTS: An important recommendation is that sponsors create a Program Safety Analysis Plan early in development. We also give recommendations for the planning of repeated, cumulative meta-analyses of the safety data obtained from the studies conducted within the development program. These include clear definitions of adverse events of special interest and standardization of many aspects of data collection and study design. We describe a 3-tier system for signal detection and analysis of adverse events and highlight proposals for reducing "false positive" safety findings. We recommend that sponsors review the aggregated safety data on a regular and ongoing basis throughout the development program, rather than waiting until the time of submission. LIMITATIONS: We recognize that there may be other valid approaches. CONCLUSIONS: The proactive approach we advocate has the potential to benefit patients and health care providers by providing more comprehensive safety information at the time of new product marketing and beyond.

Genotypes and phenotypes in children with short stature: clinical indicators of SHOX haploinsufficiency.

BACKGROUND: Short stature affects approximately 2% of children, representing one of the more frequent disorders for which clinical attention is sought during childhood. Despite assumed genetic heterogeneity, mutations or deletions of the short stature homeobox-containing gene (SHOX) are found quite frequently in subjects with short stature. Haploinsufficiency of the SHOX gene causes short stature with highly variable clinical severity, ranging from isolated short stature without dysmorphic features to Léri-Weill syndrome, and with no functional copy of the SHOX gene, Langer syndrome. METHODS: To characterise the clinical and molecular spectrum of SHOX deficiency in childhood we assessed the association between genotype and phenotype in a large cohort of children of short stature from 14 countries. RESULTS: Screening of 1608 unrelated individuals with sporadic or familial short stature revealed SHOX mutations or deletions in 68 individuals (4.2%): complete deletions in 48 (70.6%), partial deletions in 4 (5.9%) and point mutations in 16 individuals (23.5%). Although mean height standard deviation score (SDS) was not different between participants of short stature with or without identified SHOX gene defects (-2.6 vs -2.6), detailed examination revealed that certain bone deformities and dysmorphic signs, such as short forearm and lower leg, cubitus valgus, Madelung deformity, high-arched palate and muscular hypertrophy, differed markedly between participants with or without SHOX gene defects (p<0.001). Phenotypic data were also compared for 33 children with Turner syndrome in whom haploinsufficiency of SHOX is thought to be responsible for the height deficit. CONCLUSION: A phenotype scoring system was developed that could assist in identifying the most appropriate subjects for SHOX testing. This study offers a detailed genotype-phenotype analysis in a large cohort of children of short stature, and provides quantitative clinical guidelines for testing of the SHOX gene.

Growth hormone is effective in treatment of short stature associated with short stature homeobox-containing gene deficiency: Two-year results of a randomized, controlled, multicenter trial.

BACKGROUND: The short stature homeobox-containing gene, SHOX, located on the distal ends of the X and Y chromosomes, encodes a homeodomain transcription factor responsible for a significant proportion of long-bone growth. Patients with mutations or deletions of SHOX, including those with Turner syndrome (TS) who are haplo-insufficient for SHOX, have variable degrees of growth impairment, with or without a spectrum of skeletal anomalies consistent with dyschondrosteosis. OBJECTIVE: Our objective was to determine the efficacy of GH in treating short stature associated with short stature homeobox-containing gene deficiency (SHOX-D). DESIGN AND METHODS: Fifty-two prepubertal subjects (24 male, 28 female; age, 3.0-12.3 yr) with a molecularly proven SHOX gene defect and height below the third percentile for age and gender (or height below the 10th percentile and height velocity below the 25th percentile) were randomized to either a GH-treatment group (n = 27) or an untreated control group (n = 25) for 2 yr. To compare the GH treatment effect between subjects with SHOX-D and those with TS, a third study group, 26 patients with TS aged 4.5-11.8 yr, also received GH. Between-group comparisons of first-year and second-year height velocity, height sd score, and height gain (cm) were performed using analysis of covariance accounting for diagnosis, sex, and baseline age. RESULTS: The GH-treated SHOX-D group had a significantly greater first-year height velocity than the untreated control group (mean +/- se, 8.7 +/- 0.3 vs. 5.2 +/- 0.2 cm/yr; P < 0.001) and similar first-year height velocity to GH-treated subjects with TS (8.9 +/- 0.4 cm/yr; P = 0.592). GH-treated subjects also had significantly greater second-year height velocity (7.3 +/- 0.2 vs. 5.4 +/- 0.2 cm/yr; P < 0.001), second-year height sd score (-2.1 +/- 0.2 vs.-3.0 +/- 0.2; P < 0.001) and second-year height gain (16.4 +/- 0.4 vs. 10.5 +/- 0.4 cm; P < 0.001) than untreated subjects. CONCLUSIONS: This large-scale, randomized, multicenter clinical trial in subjects with SHOX-D demonstrates marked, highly significant, GH-stimulated increases in height velocity and height SDS during the 2-yr study period. The efficacy of GH treatment in subjects with SHOX-D was equivalent to that seen in subjects with TS. We conclude that GH is effective in improving the linear growth of patients with various forms of SHOX-D.

Growth hormone treatment of early growth failure in toddlers with Turner syndrome: a randomized, controlled, multicenter trial.

CONTEXT: Typically, growth failure in Turner syndrome (TS) begins prenatally, and height sd score (SDS) declines progressively from birth. OBJECTIVE: This study aimed to determine whether GH treatment initiated before 4 yr of age in girls with TS could prevent subsequent growth failure. Secondary objectives were to identify factors associated with treatment response, to determine whether outcome could be predicted by a regression model using these factors, and to assess the safety of GH treatment in this young cohort. DESIGN: This study was a prospective, randomized, controlled, open-label, multicenter clinical trial (Toddler Turner Study, August 1999 to August 2003). SETTING: The study was conducted at 11 U.S. pediatric endocrine centers. SUBJECTS: Eighty-eight girls with TS, aged 9 months to 4 yr, were enrolled. INTERVENTIONS: Interventions comprised recombinant GH (50 mug/kg.d; n = 45) or no treatment (n = 43) for 2 yr. MAIN OUTCOME MEASURE: The main outcome measure was baseline-to-2-yr change in height SDS. RESULTS: Short stature was evident at baseline (mean length/height SDS = -1.6 +/- 1.0 at mean age 24.0 +/- 12.1 months). Mean height SDS increased in the GH group from -1.4 +/- 1.0 to -0.3 +/- 1.1 (1.1 SDS gain), whereas it decreased in the control group from -1.8 +/- 1.1 to -2.2 +/- 1.2 (0.5 SDS decline), resulting in a 2-yr between-group difference of 1.6 +/- 0.6 SDS (P < 0.0001). The baseline variable that correlated most strongly with 2-yr height gain was the difference between mid-parental height SDS and subjects' height SDS (r = 0.32; P = 0.04). Although attained height SDS at 2 yr could be predicted with good accuracy using baseline variables alone (R(2) = 0.81; P < 0.0001), prediction of 2-yr change in height SDS required inclusion of initial treatment response data (4-month or 1-yr height velocity) in the model (R(2) = 0.54; P < 0.0001). No new or unexpected safety signals associated with GH treatment were detected. CONCLUSION: Early GH treatment can correct growth failure and normalize height in infants and toddlers with TS.

Effect of growth hormone dose on bone maturation and puberty in children with idiopathic short stature.

CONTEXT: GH at 0.22 mg/kg.wk has been shown to have no effect on pubertal onset or pace, whereas GH at 0.5 mg/kg x wk has been shown to advance pubertal onset and bone maturation. OBJECTIVES: Our objectives were to determine whether 0.37 mg/kg x wk GH advanced pubertal onset, pace, or bone maturation relative to 0.24 mg/kg x wk GH; whether 0.37 mg/kg x wk GH led to pubertal onset at an inappropriately early age; and whether age at start of GH therapy influenced pubertal onset. DESIGN: We conducted a randomized, open-label study to final height. PATIENTS: We studied children with idiopathic short stature. INTERVENTION: Patients were treated with 0.24 mg/kg x wk, 0.24 increasing to 0.37 mg/kg x wk, or 0.37 mg/kg x wk. MAIN OUTCOME MEASURES: We assessed age at pubertal onset and rates of bone maturation, Tanner stage development, and increase in testicular volume (boys only). RESULTS: For the primary comparison between the 0.24 and 0.37 mg/kg x wk dose groups, median ages of pubertal onset (in years) were similar (13.7 vs. 13.5 for boys and 11.7 vs. 11.4 for girls) and were greater than those for the general population for each sex. Age at start of GH therapy did not appear to influence pubertal onset for either sex. Rates of pubertal pace and bone maturation were not significantly different between the 0.24 and 0.37 mg/kg x wk dose groups for either sex. CONCLUSION: GH at 0.37 mg/kg x wk does not appear to accelerate pubertal onset, pace, or bone maturation compared with GH at 0.24 mg/kg x wk in patients with idiopathic short stature. From a clinical standpoint, our results suggest that the approved dose range of up to 0.37 mg/kg x wk GH does not lead to pubertal onset at an inappropriately early age.

Safety of growth hormone treatment in pediatric patients with idiopathic short stature.

CONTEXT: Recombinant human GH was approved by the United States Food and Drug Administration in 2003 for the treatment of idiopathic short stature (ISS). However, to date, the safety of GH in this patient population has not been rigorously studied. OBJECTIVE: The objective of this study was to address the safety of GH treatment in children with ISS compared with GH safety in patient populations for which GH has been approved previously: Turner syndrome (TS) and GH deficiency (GHD). DESIGN/SETTING: The rates of serious adverse events (SAEs) and adverse events (AEs) of particular relevance to GH-treated populations were compared across the three patient populations among five multicenter GH registration studies. PATIENTS: Children with ISS, TS, or GHD were studied. INTERVENTION: Treatment consisted of GH doses ranging from 0.18-0.37 mg/kg.wk. MAIN OUTCOME MEASURES: The main outcome measures were rates of SAEs and AEs of special relevance to patients receiving GH. Laboratory measures of carbohydrate metabolism were used as outcome measures for the ISS studies. RESULTS: Within the ISS studies, comprising one double-blind, placebo-controlled study and one open-label, dose-response study, SAEs (mainly hospitalizations for accidental injury or acute illness unrelated to GH exposure) were reported for 13-14% of GH-treated patients. Overall AE rates (serious and nonserious) as well as rates of potentially GH-related AEs were similar in the GHD, TS, and ISS studies (for ISS studies combined: otitis media, 8%; scoliosis, 3%; hypothyroidism, 0.7%; changes in carbohydrate metabolism, 0.7%; hypertension, 0.4%). Measures of carbohydrate metabolism were not affected by GH treatment in patients with ISS. There was no significant GH effect on fasting blood glucose in either study (GH dose range, 0.22-0.37 mg/kg.wk) or on insulin sensitivity (placebo-controlled study only). CONCLUSION: GH appears safe in ISS; however, the studies were not powered to assess the frequency of rare GH-related events, and longer-term follow-up studies of GH-treated patients with ISS are warranted.

Growth hormone and low dose estrogen in Turner syndrome: results of a United States multi-center trial to near-final height.

A cardinal clinical feature of Turner syndrome (TS) is linear growth failure resulting in extreme short stature: the median adult height of untreated women with TS is 143 cm, 20 cm (8 in.) below that of the general female population. In the largest multicenter, randomized, long-term, dose-response study conducted in the United States, 232 subjects with TS received either 0.27 or 0.36 mg/kg.wk of recombinant human GH with either low dose ethinyl E2 or oral placebo. The study was placebo-controlled for both GH and estrogen for the first 18 months and remained placebo-controlled for estrogen for its duration. The near-final height of the 99 subjects whose bone age was at least 14 yr was 148.7 +/- 6.1 cm after 5.5 +/- 1.8 yr of GH started at a mean age of 10.9 +/- 2.3 yr; this represents an average increase of 1.3 +/- 0.6 SD scores from baseline (TS standard). Height was greater than 152.4 cm (60 in.) in 29% of subjects compared with the expected 5% of untreated patients. Mean near-final heights of subjects who received the lower GH dose, with or without estrogen, were 145.1 +/- 5.4 and 149.9 +/- 6.0 cm, respectively; those who received the higher GH dose with or without estrogen achieved mean near-final heights of 149.1 +/- 6.0 and 150.4 +/- 6.0 cm, respectively. Factors that most impacted outcome were younger age, lower bone age/chronological age ratio, lower body weight, and greater height SD score at study entry. This study demonstrates significant GH-induced improvement in height SD score, with correction of height to within the normal channels for a significant number of patients, and provides evidence of a GH dose-response effect. These data also indicate that early administration of estrogen, even at relatively low doses, does not improve gain in near-final height in patients with TS.

Which patients do not require a GH stimulation test for the diagnosis of adult GH deficiency?

Adult GH deficiency (GHD) is currently diagnosed in patients with either a history of childhood-onset GHD or acquired hypothalamic-pituitary disease by GH stimulation testing. However, GH stimulation tests are invasive, time consuming, and associated with side effects. Based on preliminary analyses of patients enrolled in the U.S. Hypopituitary Control and Complications Study (HypoCCS), we proposed the presence of adult GHD could be predicted with 95% accuracy by the presence of three or more pituitary hormone deficiencies (PHDs) or a serum IGF-I concentration less than 84 microg/liter (11 nmol/liter). To validate the diagnostic utility of these criteria, we studied results obtained in 817 adult patients (mean [SD] age: 46.4 [15.7] yr, body mass index: 30.1 [7.2] kg/m(2)) enrolled in HypoCCS who had serum GH concentrations from stimulation tests (11 different tests used, excluding clonidine) and serum IGF-I (competitive binding RIA) measured at the central laboratory (Esoterix Endocrinology, Calabasas Hills, CA). When patients were stratified into subgroups on the basis of the presence of zero, one, two, three, and four additional PHDs, median (25th, 75th percentile) peak GH levels (micrograms per liter) were 3.5 (0.85, 7.1), 0.73 (0.18, 4.2), 0.29 (0.05, 1.4), 0.06 (0.025, 0.295), and 0.025 (0.025, 0.07), respectively. The mean log (peak GH) concentration was significantly different among the subgroups (P < 0.05). The proportion of patients in each group with severe GHD diagnosed by stimulation testing (peak GH < 2.5 microg/liter) was 41%, 67%, 83%, 96%, and 99% for patients with zero, one, two, three, and four PHDs, respectively. The positive predictive values (PPVs) for GHD of three PHDs, four PHDs, and serum IGF-I less than 84 microg/liter were 96%, 99%, and 96%, respectively. The PPV of these three diagnostic criteria was also 95% or more after excluding the data originally used to identify these potential predictors. Taken together, the presence of either three or four additional PHDs or IGF-I less than 84 microg/liter (55% of the patients met at least one of these criteria) reliably predicted GHD with a high PPV (95%), high specificity (89%), and moderate sensitivity (69%). We concluded that patients with an appropriate clinical history and either the presence of three or four additional PHDs or serum IGF-I less than 84 microg/liter (measured in the Esoterix assay) do not require GH stimulation testing for the diagnosis of adult GHD. In clinical practice, we suggest that other causes of low serum IGF-I should be excluded before applying these diagnostic criteria.

Psychological adaptation in children with idiopathic short stature treated with growth hormone or placebo.

The influence of short stature on psychological adaptation in childhood and adolescence is controversial. GH is currently used to treat children with idiopathic short stature (ISS, also known as non-GH-deficient short stature). This study represents the first double-blind, placebo-controlled trial of the effects of GH on the psychological adaptation of children and adolescents with ISS, treated with GH until adult height was attained.Sixty-eight children (53 males, 15 females), 9-16 yr old, with marked ISS (measured height or predicted adult height -2.5 sd or less) received either GH 0.074 mg/kg or placebo sc three times per week until height velocity decreased to less than 1.5 cm/yr. Parents completed the Child Behavior Checklist (CBCL) and children the Self-Perception Profile (SPP) and Silhouette Apperception Technique at baseline and annually thereafter. Baseline behavioral/emotional adjustment (CBCL) and self-concept (SPP) scores for children with ISS were within the normative range. The two study groups exhibited similar behavioral and self-concept profiles (CBCL) during the first 2 yr of the study. However, CBCL behavior problems (internalizing, externalizing, and total problems) appeared to decline, in yr 3 and 4, in the GH-treated group relative to the placebo-treated group. Group differences in CBCL competency domains and the SPP were not observed at any point during the study. Short stature among children with ISS enrolled in this long-term, placebo-controlled study was not associated with problems in psychological adaptation or self-concept with the psychological instruments employed. GH treatment was associated with a trend toward improvement in problem behaviors, as measured by questionnaires (CBCL) completed by study participants' parents. It remains to be determined whether GH treatment significantly impacts adaptation, psychosocial function, or quality of life in children with ISS.

Effect of growth hormone treatment on adult height in peripubertal children with idiopathic short stature: a randomized, double-blind, placebo-controlled trial.

GH is often used to treat children with idiopathic short stature despite the lack of definitive, long-term studies of efficacy. We performed a randomized, double-blind, placebo-controlled trial to determine the effect of GH on adult height in peripubertal children. Subjects (n = 68; 53 males and 15 females), 9-16 yr old, with marked, idiopathic short stature [height or predicted height < or = -2.5 sd score (SDS)] received either GH (0.074 mg/kg) or placebo sc three times per week until they were near adult height. At study termination, adult height measurements were available for 33 patients after mean treatment duration of 4.4 yr. Adult height was greater in the GH-treated group (-1.81 +/- 0.11 SDS, least squares mean +/- sem) than in the placebo-treated group (-2.32 +/- 0.17 SDS) by 0.51 SDS (3.7 cm; P < 0.02; 95% confidence interval, 0.10-0.92 SDS). A similar GH effect was demonstrated in terms of adult height SDS minus baseline height SDS and adult height SDS minus baseline predicted height SDS. Modified intent-to-treat analysis in 62 patients treated for at least 6 months indicated a similar GH effect on last observed height SDS (0.52 SDS; 3.8 cm; P < 0.001; 95% confidence interval, 0.22-0.82 SDS) and no important dropout bias. In conclusion, GH treatment increases adult height in peripubertal children with marked idiopathic short stature.

Prospective safety surveillance of GH-deficient adults: comparison of GH-treated vs untreated patients.

CONTEXT: In clinical practice, the safety profile of GH replacement therapy for GH-deficient adults compared with no replacement therapy is unknown. OBJECTIVE: The objective of this study was to compare adverse events (AEs) in GH-deficient adults who were GH-treated with those in GH-deficient adults who did not receive GH replacement. DESIGN AND SETTING: This was a prospective observational study in the setting of US clinical practices. PATIENTS AND OUTCOME MEASURES: AEs were compared between GH-treated (n = 1988) and untreated (n = 442) GH-deficient adults after adjusting for baseline group differences and controlling the false discovery rate. The standardized mortality ratio was calculated using US mortality rates. RESULTS: After a mean follow-up of 2.3 years, there was no significant difference in rates of death, cancer, intracranial tumor growth or recurrence, diabetes, or cardiovascular events in GH-treated compared with untreated patients. The standardized mortality ratio was not increased in either group. Unexpected AEs (GH-treated vs untreated, P ≤ .05) included insomnia (6.4% vs 2.7%), dyspnea (4.2% vs 2.0%), anxiety (3.4% vs 0.9%), sleep apnea (3.3% vs 0.9%), and decreased libido (2.1% vs 0.2%). Some of these AEs were related to baseline risk factors (including obesity and cardiopulmonary disease), higher GH dose, or concomitant GH side effects. CONCLUSIONS: In GH-deficient adults, there was no evidence for a GH treatment effect on death, cancer, intracranial tumor recurrence, diabetes, or cardiovascular events, although the follow-up period was of insufficient duration to be conclusive for these long-term events. The identification of unexpected GH-related AEs reinforces the fact that patient selection and GH dose titration are important to ensure safety of adult GH replacement.