Safety of Growth Hormone Replacement Therapy in Childhood-Onset Craniopharyngioma: A Systematic Review and Cohort Study

Introduction: Survival of childhood-onset craniopharyngioma (cCP) is excellent; however, many survivors suffer from hypothalamic-pituitary dysfunction. Growth hormone replacement therapy (GHRT) is of high importance for linear growth and metabolic outcome. Optimal timing for initiation of GHRT in cCP is on debate because of concerns regarding tumor progression or recurrence. Methods: A systematic review and cohort studys were performed for the effect and timing of GHRT on overall mortality, tumor progression/recurrence, and secondary tumors in cCP. Within the cohort, cCP receiving GHRT ≤1 year after diagnosis were compared to those receiving GHRT >1 year after diagnosis. Results: Evidence of 18 included studies, reporting on 6,603 cCP with GHRT, suggests that GHRT does not increase the risk for overall mortality, progression, or recurrent disease. One study evaluated timing of GHRT and progression/recurrence-free survival and found no increased risk with earlier initiation. One study reported a higher than expected prevalence of secondary intracranial tumors compared to a healthy population, possibly confounded by radiotherapy. In our cohort, 75 of 87 cCP (86.2%) received GHRT for median of 4.9 years [0.0–17.1]. No effect of timing of GHRT was found on mortality, progression/recurrence-free survival, or secondary tumors. Conclusion: Although the quality of the evidence is low, the available evidence suggests no effect of GHRT or its timing on mortality, tumor progression/recurrence, or secondary neoplasms in cCP. These results support early initiation of GHRT in cCP aiming to optimize linear growth and metabolic outcome. Prospective studies are needed to increase the level of evidence upon the optimal timing to start GHRT in cCP patients.


Introduction
Survival of childhood-onset craniopharyngioma (cCP) is excellent; however, quality of life is often deprived because of hypothalamic-pituitary damage [1].As severity of hypothalamic damage has been related to the extent of neurosurgery, it has been advocated to shift the neuro-surgical approach from gross-total resection (GTR) to limited resection (LR) [2,3].A drawback of LR and accepting tumor rest is, however, an increased risk for growth of tumor rest, necessitating second surgery and/or radiotherapy [4].
Of all pituitary deficiencies, growth hormone deficiency (GHD) is most common in cCP and is observed in 26-75% at diagnosis and in 70-92% after tumor treatment [4,5].GHD has a significant negative effect on longitudinal growth [6].In addition, GHD has negative effects on lipid and glucose homeostasis, lean and fat mass, as well as on bone mineral density and possibly even on cognitive outcome of the young child [7][8][9][10][11].To prevent such consequences of GHD, timely replacement of GHD is of great importance.
Craniopharyngioma cells have been shown to express high levels of insulin-like growth factor-1 and growth hormone (GH) receptors, and therefore it has been questioned whether GH replacement therapy (GHRT) influences cCP recurrence or regrowth [5].A recent consensus statement of the Growth Research Society concluded that: "although GH and insulin-like growth factor-1 have been shown in vitro to have a "permissive role" for carcinogenesis, there are no clinical studies that show that GHRT in patients surviving a brain tumor, after achieving complete remission, leads to increased recurrence rates" [12].Data on cancer survivors are limited, but more solid data were produced in low-grade tumors, such as in adult benign pituitary adenomas, in which no association was found between GHRT and tumor recurrence [13,14].With these results in mind, GHRT is considered safe in low-grade tumors, such as cCP.
The effect of timing of the start of GHRT is however still an issue of debate; the Lawson Wilkins Pediatric Endocrinology Society Drug and Therapeutics Committee recommends to start GHRT 1 year after completion of tumor therapy, but the Growth Hormone Research Society (GRS) recommends that this interval may be as short as 3 months in children with radiologically proven stable residue of craniopharyngioma who have significant growth failure and metabolic disturbance [15].Despite this recommendation, the optimal timing of the start of GHRT is often debated within the multidisciplinary teams, outweighing the potential benefits against the harms of GHRT.Especially for children with proven GHD and CP, to wait a year to commence with GHRT may significantly reduce height potential and have negative metabolic effects.To determine the optimal timing of the start of GHRT, there is a need for more studies evaluating the safety of GHRT and its timing.For this reason, we performed a cohort studys evaluating the effect of GHRT and its timing on mortality, progression/ recurrence of cCP, and/or secondary neoplasms.In parallel, we conducted a systematic review to provide a comprehensive overview of the current state of the art on the safety of GHRT and its timing.

Systematic Review
The Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) checklist was used to structure the systematic review [16].A literature search was conducted using two electronic databases (MEDLINE and Embase) (from January 1, 1980 up to December 31, 2021).We developed a search strategy in collaboration with a trained clinical librarian using Medical Subject Headings/ Emtree terms and text words related to GHRT and cCP.Details of the search strategy are shown in online supplementary table 1 (for all online suppl.material, see https://doi.org/10.1159/000531226).The reference lists from relevant reviews and included studies were screened for additional studies that were not identified by our initial search strategy.We used the following predefined selection criteria: (1) study population of at least 20 patients diagnosed with cCP (at age <18 years), (2) patients who received GHRT for GHD, (3) at least one outcome (mortality, tumor progression/tumor recurrence, or secondary neoplasm), as defined by the authors, on safety of GH was reported, and (4) published in the English language.Only studies with original data were considered (e.g., RCTs or cohort studies).When multiple articles with (partially) overlapping study populations were identified, the article with the most recent publication date or with the longest follow-up time was included.When the amount of overlap was unclear, we included both studies reporting the possibility of overlap.The following data were extracted: study population, comparison group, follow-up duration, type of cCP treatment, information on GHRT, and relevant outcomes (mortality, tumor progression/tumor recurrence, or secondary neoplasms).All effect measures reported of the outcome were included (e.g., number of events, prevalence, risk ratio, progression-free survival [PFS]).Risk of bias assessment was based on previously described checklists according to evidence-based medicine criteria (online suppl.Table 2) [17,18].Two reviewers (J.v.S. and E.K.) evaluated studies independently using Rayann (i.e., study selection (both title/abstract and full text phase), data extraction, and risk of bias assessment), and disagreements were resolved by consensus.was conducted.With the use of PALGA (Pathologisch Anatomisch Landelijk Geautomatiseerd Archief), the Dutch cohort of children (≤18 years) diagnosed with cCP during the period 2004-2021 was retrieved [20].All patients known at the Princess Máxima Center for pediatric oncology, who had been diagnosed with a cCP from 2004 to 2021 (either proven histologically or clinically very likely based on imaging results) and had a minimal follow-up of 1 year were included.Informed consent was obtained from the legal guardian and/or patient (if age 12 and up) for the use of data.

Data Collection
Data on sex at birth, age, tumor and treatment-related variables, and endocrine outcomes were collected.Data on GHRT were collected, including type of GHRT, time between diagnosis and start of GHRT, and duration of GHRT.For timing of GHRT, GHRT was divided into two subgroups: GHRT started within 1 year after cCP diagnosis and GHRT started later than 1 year after cCP diagnosis.

Treatment
Type of resection was based on the report of the neurosurgeon in combination with the report of the radiologist, or, if not available, based on the documentation of the treating oncologist.The upfront surgical procedure (within the first 3 months after cCP diagnosis) was classified into cyst drainage/fenestration (eCD/ F), LR, near total resection (NTR), and GTR.An endoscopic cyst drainage or fenestration without further resection of tumor tissue reported by the neurosurgeon and/or radiologist, was scored as eCD/F.LR was defined as less than 95% of the tumor resected or a partial resection (with obvious residual tumor mass present) reported by the neurosurgeon and/or radiologist.NTR was defined as more than 95% of the tumor resected, but not all tumor tissue removed.If exact percentages were not reported, NTR was based on neurosurgical and/or radiologist reporting: NTR or reports stating that almost all tumor was removed and little, minimal, or minor tumor tissue had been left (residual small enhancement lesion but no apparent tumor mass).GTR was defined as having removed all tumor tissue, stated both by the neurosurgeon and radiologist.If more than one surgical procedure was performed within the first 3 months, the most invasive type of resection was classified.

Tumor Progression/Recurrence
Tumor progression was defined as growth of residual tumor volume (cystic and/or solid) after any cCP treatment (excluding GTR) on magnetic resonance imaging scans as reported by a neuroradiologist for which an intervention was needed.Recurrence of tumor was defined as newly identified cCP after a GTR on imaging scans, as reported by a neuroradiologist for which an intervention was needed.

Statistical Analysis
Data are presented as mean ± SD or median [range] for continuous data, depending on the distribution.Data are presented as percentages for categorical variables.Between-group differences were evaluated by Student's t test for continuous data with a normal distribution, Mann-Whitney U test for continuous data with a skewed distribution, and by χ 2 test or Fisher's exact test for categorical data.To assess violation of normality distribution, QQ plot of the residuals and the Shapiro-Wilk's test were employed.Between group differences were evaluated by oneway analysis of variance for continuous data with a normal distribution, Kruskal-Wallis test for continuous data with a skewed distribution (skew variables were not further transformed), and by χ 2 test or Fisher's exact test for categorical data.To study the effect of possible risk factors on the outcome, univariate and multivariable Cox regression proportional hazard models were estimated.Independent variables to be included in the multivariable model were selected by estimating the univariate model and by considering the clinical relevance of each variable.The Cox regression model with initiation of GHRT as a time-dependent covariate was used to generate estimates of relative risk for tumor progression/recurrence in children with cCP.To estimate time to tumor progression/recurrence, Kaplan-Meier was used.Landmark analysis was used to study the impact of starting GHRT at different times after diagnosis on PFS.A p value of <0.05 was considered statistically significant.Analyses were performed by using SPSS version 27.0.

Study Characteristics
All included studies were retrospective (n = 4) or prospective (n = 14) cohort studies.Four studies contained a control group with cCP patients not receiving GHRT.In total 6,603 cCP patients with GHRT were included (the number in each study varied between 22 and 1,038) and 108 cCP patients without GHRT.If reported, mean and median follow-up time from cCP diagnosis ranged between 3.2 and 15.9 years.Treatment consisted of surgery only or radiotherapy only or a combination of both, in the nine studies reporting on treatment modalities.Mean or median time between cCP diagnosis and start of GHRT ranged between 0.4 and 3.0 years.Due to missing data and clinical heterogeneity, pooling of data was not possible, and therefore, we provide descriptive results.For detailed information see Tables 1-3.

Mortality
Mortality of cCP patients was evaluated in total in six studies, including 2,438 cCP with GHRT [21,22,32,[36][37][38].Prevalence of mortality of cCP was reported in Safety of GHRT in Childhood-Onset Craniopharyngioma five of the six studies and ranged between 0.2% and 6.4% in patients with cCP and GHRT.One study included cCP patients with and without GH, however, no report on the significance level of the difference between both groups was given.Two studies reported a significantly increased mortality risk in the patients with cCP and GHRT (both univariate analyses), compared to healthy controls.Reported mortality causes were death due to the craniopharyngioma itself, adrenal crisis, trauma, and infected ventricular-peritoneal shunt.Detailed information is reported in Table 1.No study reported on mortality in relation to the timing of GHRT.

Tumor Progression and Tumor Recurrence
Sixteen studies, including 6,405 cCP patients with GHRT, reported on tumor progression and/or recurrence in relation to GHRT [21-28, 30-35, 38].In none of the studies the definition for the outcome "tumor progression" or "tumor recurrence" was provided.Four studies included a control group consisting of cCP patients without GHRT.Tumor progression/recurrence rate ranged between 4.2% and 50.0% in cCP patients with GHRT, compared to 39.4-57.1% in cCP without GHRT (reported in two studies).Two of the four studies includ-ing a control group reported the significance level of the difference between the cCP with and without GHRT and both reported no significant difference on tumor progression/recurrence rate (one study performed univariate analysis, one study performed a multivariate analysis, adjusted for degree of resection, radiotherapy, gender, age at diagnosis).One study reported on timing of GHRT and progression/recurrence-free survival and found no statistically significant difference between starting GHRT 1 year after cCP diagnosis versus 1-3 years, 3-5 years, and >5 years [26].Detailed information is reported in Table 2.

Secondary Neoplasms
Two studies were found to report on secondary neoplasms after GHRT in cCP, including 1,738 cCP with GHRT [21,38].In the study of Bell et al. [16], 3 out of 994 cCP patients with GH (0.3%) developed a secondary neoplasm.No control group was included.In the study of Yuen et al. [38], who reported on a standardized incidence ratio (univariate) comparing cCP with GHRT to a healthy population, 2 of 744 cCP (0.27%) developed a new intracranial tumor and 4 of 744 cCP (0.54%) a new extracranial tumor.When compared to the general population, the risk     for secondary intracranial tumors was increased with a standardized incidence ratio of 11.2 (1.26-40.50)(p = 0.035).In addition, they performed an univariate analysis for risk of secondary intracranial tumors, combining cCP and adult CP with GHRT, and found a trend toward an increased risk of radiotherapy (HR 3.47 [0.92-13.10],p = 0.067).For extracranial tumors no significant difference was identified.Detailed information is reported in Table 3.No study reported on secondary neoplasms in relation to timing of GHRT.

Risk of Bias Assessment according to Evidence-Based
Medicine Criteria [14,15] Of the studies reporting on overall mortality, in one study there was a high risk of selection bias, because patient inclusion from the original cohort of eligible cCP was <75% (17%).In the other 5/6 studies, the risk of selection bias was unclear (83%).The risk of attrition bias (≥75% of the outcome was assessed in the study cohort) and detection bias (outcome assessors blinded for important determinants) was low in all studies reporting on mortality.Confounding bias was applicable in two studies, in both there was a high risk for confounding bias because no prognostic factors (e.g., diversity in treatment) were taken into account.In the studies reporting on tumor progression/ recurrence, the risk of selection bias was unclear in 14/16 studies (88%), high in one study (6%), and low in another (6%).The risk of attrition bias was low in 15/16 studies (94%) and unclear in the other one (6%).Presence of detection bias was unclear in all studies.In all studies reporting on tumor progression/recurrence for which confounding bias was applicable, there was a high risk of confounding.In the studies reporting on the risk to develop a secondary neoplasm, selection bias and detection bias was unclear in both studies, while the risk of attrition bias was low.The risk of confounding, applicable to one study, was high.Detailed information on the reasoning for all risk of bias judgments is included in Tables 1-3.

Patient Characteristics
In total, 166 children (aged <18 years) had been diagnosed with histology proven cCP in the Netherlands in the period 2004-2021, and were potentially eligible for inclusion in the study [26].Of these, 105/166 (63%) were known in the Princess Máxima Center and could be asked informed consent for use of the data (examined and confirmed eligible).In total, 87/105 patients (83%) approved consent for use of data for this study and could thus be

Safety of GHRT in Childhood-Onset Craniopharyngioma
Table 3.
Studies reporting on the risk of secondary neoplasm in cCP with GHRT included.Of all 87, data were available for analysis.Reasons for not obtaining consent were: refusal to participate (n = 10, 10%) and not able to retrieve for obtaining consent for use of data/moved abroad (n = 7, 7%).
cCP Treatment Of 87 cCP patients, mean age at diagnosis was 7.39 years ±3.67 and mean age at follow-up was 14.19 years ±4.98.Of all, 26 (29.9%) had been treated with eCD/F, 24 patients (27.6%) with LR, 14 (16.1%) with NTR, and 22 (25.3%) with GTR.One patient was given a wait and see policy.Seven patients (8.0%) received local RT immediately after their first surgery.Another 33 patients (37.9%) received RT at time of progression/ recurrence.Of the 65 patients with limited surgery (wait and see, eCD/F, LR, or NTR), 7 (10.8%)patients received RT directly following first surgery, and 31 (47.7%) for tumor progression/recurrence at follow-up with a mean delay of 2.90 years ±2.71 after cCP diagnosis.Twenty-seven cCP (41.5%) had not received any RT at last moment of follow-up, after a mean follow-up time of 4.71 years ±3.39.Of the 22 patients who were given GTR, 2 patients (9.0%) received RT at follow-up, with a mean delay of 2.03 years ±0.75.Interferon alpha for cystic cCP was given in 11.5% as initial therapy, and in 6.9% during follow-up at time of cyst growth.None of the patients died or developed a secondary tumor during follow-up.

Tumor Progression/Recurrence
Of the 87 patients, 53 patients (60.9%) had one or more tumor progression or recurrent events after a mean period of 1.09 years ±0.92 (after diagnosis) that resulted in additional treatment.Of these, 48 (55.2%) cCP patients had tumor progression and 5 (5.7%) tumor recurrence.In total, there were 100 events of progression/recurrences requiring treatment in these 53 patients.In 56.0% surgery was the treatment of first choice, in 25.0% radiotherapy, in 8.0% surgery plus RT, in 10.0% interferon alpha, and in one event tocilizumab.Of the 40 patients treated with radiotherapy (seven immediately after initial treatment, 33 at follow-up), 11 patients (27.5%) developed a progression/recurrence event after RT requiring treatment.

GHRT and cCP Progression/Recurrence
In total 78/87 (89.7%) cCP were diagnosed with GHD, of whom 75 (86.2%)received GHRT at any point during follow-up.In 1 patient using GHRT, the starting date of GHRT was unknown and was therefore excluded for further analysis.One patient had already started GHRT before cCP diagnosis.At cCP diagnosis, GHRT was paused for 5 months and restarted after no signs of tumor residue.
Using landmark analysis with the landmark point set at 6 months, 1 year (Fig. 2b), or 1.5 years, progression/ recurrence-free survival did not differ between cCP who started with GHRT before the landmark versus Safety of GHRT in Childhood-Onset Craniopharyngioma cCP who started with GHRT after the landmark ([n = 70] p = 0.894, [n = 55] p = 0.073, and [n = 49] p = 0.244, respectively).Multivariate Cox regression (with landmark set at 1 year from cCP diagnosis), including starting time of GHRT and limited surgery (eCD/F, LR, or NTR), showed a trend toward increased risk of limited surgery for tumor progression (HR 2.81 95% CI 0.95-8.27,p = 0.061).
In the multivariate Cox regression, using timing of GHRT as a time-dependent covariate, extent of surgery at diagnosis (eCD/F, LR, or NTR) was a significant risk factor for developing progression/recurrence (HR 8.09 CI 95% 3.09-21.21).Radiotherapy at moment of diagnosis was a protective factor for tumor progression/ recurrence (HR 0.27 95% CI 0.08-0.89).No increased risk for developing progression/recurrence was found for GHRT, timing of GHRT, or GHRT duration (Table 5).

Other Endocrine Deficiencies
Of the 87 patients, 87.4% had been diagnosed with thyroid-stimulating hormone deficiency, 75.9% with adrenocorticotropic hormone deficiency, 49.4% with gonadotrophin deficiency, 73.6% with diabetes insipidus, and 8.0% with central precocious puberty.All patients had received adequate substitution therapy with levo-thyroxine, hydrocortisone, sex steroids, or desmopressin when indicated.Of the 7 patients (8.0%) diagnosed with central precocious puberty, all had received GnRH analogues.

Discussion
Delay in treatment of GHD in childhood may severely affect final height, especially when this occurs during pubertal development [39].In addition, GHD may have detrimental effect on metabolic health, which is of special importance in cCP with hypothalamic damage [40].For fear of tumor progression or recurrence, treatment with GHRT, and its timing, may be a subject of discussion between the oncologist and the parents [41].The results of this systematic review of current literature combined with the results of our own cohort studys, suggest that GHRT in children treated for cCP does not increase the risk for overall mortality, progression or recurrent disease, independent of timing of GHRT.
In one study, a statistically significant increased risk for a secondary intracranial tumor compared to age-and sexspecific general population rates was found for cCP with GHRT, with a trend toward radiotherapy as confounding risk factor [38].Other large cohort studies have shown that GHRT in cancer survivors does not increase the risk for secondary tumors or recurrence, but that the risk for secondary tumors is mainly related to the given radiotherapy [42,43].Taking these reassuring data into account, it may even be advocated, from a physiological point of view, that GHRT should be given to all children with GHD from time of diagnosis to optimize metabolic outcome, irrespective of underlying disease.Future research, including prospective designs, should address the possibility of more early initiation of GHRT.
The data provided in this manuscript will help to adequately balance the benefits and risks for starting GHRT in children with cCP.Benefits of GHRT in children with GHD are improvement of final height, bone strength, and body composition including BMI [22,[44][45][46][47][48][49].In cCP patients, quality of life is mainly disturbed due to hypothalamic-pituitary dysfunction, resulting in hypothalamic obesity [1].Improving body composition and BMI in these children is thus of special importance and early initiation of GHRT may positively influence this outcome.Data supporting the effects of GHRT on body composition in children with hypothalamic obesity is, however, inconsistent, but it must be considered that hypothalamic obesity is a highly complex disease with many factors contributing to its etiology and solely evaluating the effect of GHRT is, therefore, very challenging [47,50,51].
Because of the advocated shifting in treatment regimen from GTR to limited surgery, the number of cCP patients with residual disease is expected to increase.The argument favoring LR is the increased risk for hypothalamic dysfunction associated with complete resection [52].Hypothalamic damage is the most important adverse outcome impacting QoL and LR has shown to be preventive for hypothalamic damage to some degree [53,54].
Of our cohort, 48/89 (53.9%) had tumor progression and 5/89 (5.6%) had tumor recurrence requiring treatment.Compared to literature, our progression/ recurrence rate seems to be higher, with overall tumor progression/recurrence rates reported from 4.2% to 50.0% [21, 23-28, 30, 31, 33-35, 38].This difference may be explained by the fact that, in our cohort, only 25.3% of the patients received GTR.Not surprisingly, in our multivariate analysis, limited surgery was indeed a risk factor for tumor progression/recurrence.
The optimal timing of radiotherapy in cCP also remains a matter of debate [55].To wait with radiotherapy until disease progression can be beneficial and even crucial in young children who are still in the essential process of brain development.In this regard, the possible beneficial effects of radiotherapy (tumor control) must be balanced against its possible adverse effects, such as additional hypothalamic-pituitary damage, cognitive decline, vascular injuries, and secondary neoplasms [56,57].With this respect, an important observation in our cohort was the fact that of the 65 patients who did not receive GTR, 27 (41.5%)had not received additional radiotherapy at last moment of follow-up.Of these, 13/27 (48.1%) did not develop any tumor progression event (mean follow-up time 4.24 years ±4.22), of which 8/13 (61.5%) had received GHRT.The other 14/27 (51.9%) cCP developed tumor progression, of which 9/14 (64.3%) had received GHRT.On the other hand, 26.3% of patients treated with LR and additional radiation (n = 38), still developed a recurrent disease, which has also been reported by others [29,58,59].Possibly, the margins of the radiation field may have contributed to this rate and this should be evaluated in future research.

Safety of GHRT in Childhood-Onset Craniopharyngioma
The quality of the evidence we found by our systematic review, to answer the question on optimal timing of GHRT, is low.The one study we found did not report an increase in progression/recurrence-free survival with different starting times of GHRT after cCP diagnosis in a univariate analysis, but multivariate analysis was lacking [26,29].In our cohort, we found no difference in PFS between cCP patients who were given GHRT <1 year when compared to after 1 year.It was noticed, however, that the children who started GHRT >1 year after diagnosis, more often had developed tumor progression/recurrence already before starting GHRT (group B).These progression/ recurrence events before starting GHRT may have been the reason not to start GHRT and may thus be considered as an inherent bias.Therefore, we performed a landmark analysis excluding patients with progression/recurrence in the first year after diagnosis.Using this analysis, we did not identify any differences between the groups.Unfortunately, we were not able to perform an extensive cox regression with multiple covariates, as the number of events were limited.Future research should focus on prospective designs, starting GHRT irrespective of CP rest/regrowth or recurrence at different time intervals.
The results of the literature search and our cohort studys have inherent limitations.It must be acknowledged that the quality of the studies that were included in the systematic review was limited; most of the studies did not include an adequate control group (i.e., cCP patients with GHD but without GHRT), or adequate risk analysis (i.e., taking into account all relevant prognostic factors).In addition, there was risk of bias and the power of the study was often limited, a b Fig. 2. a Kaplan-Meier curve of progression/ recurrence-free survival in cCP.b Kaplan-Meier curve of progression/recurrence-free survival in cCP who started with GHRT within 1 year versus after 1 year of diagnosis (with landmark applied at 1 year after cCP diagnosis).cCP, childhood-onset craniopharyngioma; GHRT, growth hormone replacement therapy.
due to small sample size.Underreporting of complications cannot be precluded, as in most registries, data are based on information provided by the attending physician.Moreover, initiation of GHRT in cCP is dependent on the judgement of the physician, as individuals in cohorts are not randomized.Patients with prior recognized risk factors, such as former progression/recurrence, could have been excluded from receiving GHRT.Therefore, incidence of adverse events may have been underestimated.must be acknowledged, that there is a lack of long-term data on safety of GHRT in the literature search, as well as in our cohort studys.Future studies should therefore aim on a long-term registry design or perform data linkage studies; however, it may take many years for such studies to find moderate effects.Next, in our review, data of nine studies were based on five different multinational observational studies, thus possible overlap in patients must be taken into account when interpreting these results.Although the results of our cohort analysis are in line with the existing literature, it must be noted that the distribution between patients with GHRT versus no GHRT was very skewed, conclusions should, therefore, be drawn with caution.
The strengths of our review and cohort studys include the elaborate literature search, the dual assessment of obtained articles, and the high-quality assessment of timing of GHRT.The data provided will contribute to optimally balance the benefits against the potential risks for starting GHRT in cCP.Future studies are needed with preferably, large prospective randomized controlled trials, also focusing on the newest recommendation from the Growth Hormone Research Society stating that GH may be started as early as 3 months after cCP diagnosis.
In conclusion, the available evidence suggests that there is no increased risk of overall mortality, tumor progression/recurrence, or secondary neoplasms in cCP patients after starting GHRT and no effect of timing of GHRT.Although these results are reassuring, it must be kept in mind that the quality of the evidence is mostly low.Careful long-term surveillance after starting GHRT in cCP patients is crucial and there is need for prospective studies.The benefits and possible harms of commencing GHRT and its timing should be a shared decision of the health care professional together with the patient and the parents based on individual arguments, such as age, longitudinal growth, pubertal stage, presence and worries of tumor residue, and metabolic state.

Table 1 .
Studies reporting on overall mortality in cCP with GHRT

Table 2 .
Studies reporting on tumor progression or recurrence in cCP with GHRT DOI: 10.1159/000531226 van Schaik et al.

Table 2 (
National Cooperative Growth Study; GeNeSIS, Genetics and Neuroendocrinology of Short Stature International Study; PedGH, growth hormone treatment started at childhood; ContGH, continued growth hormone therapy in pediatric and adult life; AdultGH, growth hormone therapy started in adulthood; AE, adverse event.Numbers presented in mean ± SDS or median [range].*statistically significant.^Possibleoverlap of patients, data were all collected from the prospective cohort KIGS: Pfizer International Growth Database.^Possibleoverlap of patients, data were all collected from the prospective cohort NCGS: National Cooperative Growth Study.^Possibleoverlap of patients, data were all collected from the prospective cohort GeNeSIS: Genetics and Neuroendocrinology of Short Stature International Study.

Table 4 .
Patient characteristics at diagnosis and follow-up