The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 1, e409–e410 doi:10.1210/clinem/dgaa735 Letter to the Editor Response Letter to the Editor Response Response to Letter to the Editor: “A Genome- Wide Pharmacogenetic Study of Growth Hormone Responsiveness” Downloaded from https://academic.oup.com/jcem/article/106/1/e409/5974059 by guest on 29 April 2021 Joel N. Hirschhorn,1 Andrew Dauber,2 Laura Audi,3 Sailaja Vedantam,1 Michael B. Ranke,4 Alexander A.L. Jorge,5 Anders Lindberg,6 Cecilia Camacho-Hübner,7 and Michael P. Wajnrajch7 1 Division of Endocrinology, Boston Children’s Hospital, and Program in Medical and Population Genetics, Broad Institute, Harvard Medical School, Boston, Massachusetts 02115; 2Division of Endocrinology, Children’s National Hospital, Washington, DC 20010; 3Department of Pediatrics, Institut de Recerca (VHIR), Hospital Vall d’Hebron, Centre for Biomedical Research on Rare Diseases (CIBERER), Autonomous University, Barcelona, Spain; 4University Children´s Hospital, Paediatric Endocrinology, Tübingen, Germany; 5Unidade de Endocrinologia do Desenvolvimento (LIM42), Hospital das Clinicas da Faculdade de Medicina da Universidade de Sao Paulo, Sao Paulo, Brazil; 6Pfizer, Data Management, Sollentuna Sweden; and 77Pfizer Inc, Rare Disease, New York 10017 ORCiD numbers: 0000-0003-4890-0262 (A. Dauber); 0000-0002-8211-9237 (M. P. Wajnrajch). Received: 25 September 2020; Editorial Decision: 8 October 2020; First Published Online: 11 November 2020; Corrected and Typeset: 11 November 2020. Pierre Bougnères correctly points out the limitations of effect: A single polygene score of 697 height-associated our genome-wide association study (GWAS) of variants variants reflects the combined action of all 697 variants affecting growth hormone (GH) response, all of which (explaining ~16% of the variance in height [2]). we raised in our “Discussion,” including a relatively small In specific reference to the GHRd3 variant, although sample size for GWAS (1). Our main goal was to carry out testing this variant was not the main goal of our study, our re- the first genome-wide search powered to detect variants sults still provide additional information on the likely strength with moderate to large effects on GH response, and to test of association with GH responsiveness. The meta-analysis some specific hypotheses. Although our study size is small cited by Bougnères (3), which encompassed 1680 children compared with most GWAS of polygenic traits, it was ad- with a wide range of diagnoses and ancestries, reported a equately powered to detect larger effects that have been 0.075 smaller change in height SD score after 1 year in chil- seen in other pharmacogenetic settings. Furthermore, the dren who were homozygous for the more common (minor sample sizes needed for replication of individual variants allele frequency ~0.72) wild-type (WT) allele, and a decrease that have themselves been proposed to have large effects, of –0.159 in the subsample with GH deficiency. In our study, such as the GHRd3 variant, is much smaller than the the SD of the change in height SD score was 0.4, meaning study sizes needed for comprehensive GWAS discovery, that the effect sizes of the WT allele reported in the meta- for which effect sizes are more modest and appropriate analysis corresponds to a z score of –0.075/0.4 = –0.188 in all correction for multiple testing is required. We were also individuals and –0.159/0.4 = –0.40 in individuals with GH well powered to test the hypothesis that genetically de- deficiency. These effect sizes correspond to approximately termined height influences GH responsiveness, because 0.9% and approximately 4% of the variance in GH response we effectively examined a single hypothesis with large explained by this one variant. With these effect sizes, we had ISSN Print 0021-972X ISSN Online 1945-7197 Printed in USA © The Author(s) 2020. Published by Oxford University Press on behalf of the Endocrine Society. https://academic.oup.com/jcem   e409 This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. e410  The Journal of Clinical Endocrinology & Metabolism, 2021, Vol. 106, No. 1 62% and 92% power to achieve a P value of less than .05 Pharmaceuticals Group, 235 E 42nd St, MS 235-10-01, New York, in our sample of 614 children, of whom 276 had GH defi- NY 10017, USA. E-mail: michael.wajnrajch@pfizer.com. Disclosure Summary: J.N.H. received research support from ciency. In our primary analysis, European-ancestry–only ana- Pfizer Inc for this study and is on the scientific advisory board of lysis, and European-ancestry GH deficiency–only analysis, Camp4 Therapeutics; A.D. has received consulting fees from OPKO we actually observed slight trends in the opposite direction Biologics, Sandoz, Novo Nordisk, Ipsen, and Pfizer; A.A.L.J. re- (z scores of 0.024, 0.002, and 0.011, with P values of .66, ceived speaker fees from Sandoz and Novo Nordisk; M.B.R. has .97, and .89, respectively). None of the 95% CIs around these received speaker fees from Novo Nordisk, Merck Serono, Sandoz, z scores include the effect sizes reported in the meta-analysis, and Mediagnost. M.P.W. and C.C.H. are employees of Pfizer Inc; A.L. was an employee of Pfizer Health AB; and S.V. and L.A. have suggesting that the meta-analysis, at a minimum, has over- nothing to declare. estimated the effect of the GHRd3 variant. Of note, in Wassenaar et al (3), the combined meta-analysis P values for association with 2 related measures of GH re- References Downloaded from https://academic.oup.com/jcem/article/106/1/e409/5974059 by guest on 29 April 2021 sponsiveness ranged from .03 to .004, and the most signifi- 1. Dauber A, Meng Y, Audi L, et al. A genome-wide pharmacogen- cant of these was driven largely by the study that first reported etic study of growth hormone responsiveness. J Clin Endocrinol the association (4). We and others have previously noted that Metab. 2020;105(10):dgaa443. most candidate gene association studies that used a P value of 2. Wood AR, Esko T, Yang J, et al; Electronic Medical Records and less than .05 as a threshold for declaring significance turned Genomics (eMEMERGEGE) Consortium; MIGen Consortium; out to be false positives (5-8), which is why the thresholds for PAGEGE Consortium; LifeLines Cohort Study. Defining the role of common variation in the genomic and biological archi- declaring significance have since become much more strin- tecture of adult human height. Nat Genet. 2014;46(11): gent (genome-wide significance, typically P less than 5 × 10–8 1173-1186. [9]). We also note that if our study, with a total sample size of 3. Wassenaar MJ, Dekkers OM, Pereira AM, et al. Impact of the more than 600 individuals, were added to the meta-analysis in exon 3-deleted growth hormone (GH) receptor polymorphism Wassenaar et al, the combined P value would move even fur- on baseline height and the growth response to recombinant ther away from this threshold of significance. To borrow the human GH therapy in GH-deficient (GHD) and non-GHD chil- metaphor of David and Goliath from Bougnères, a few spe- dren with short stature: a systematic review and meta-analysis. J Clin Endocrinol Metab. 2009;94(10):3721-3730. cial candidate gene association studies hit their target, but most 4. Dos Santos C, Essioux L, Teinturier C, Tauber M, Goffin V, turned out to have gone wide of the mark. Although we cannot Bougnères P. A common polymorphism of the growth hormone definitively say that the association with GHRd3 and GH re- receptor is associated with increased responsiveness to growth sponsiveness is a false positive, our data strongly indicate that hormone. Nat Genet. 2004;36(7):720-724. the effect size is not as large as has been previously suggested. 5. Hirschhorn JN, Gajdos ZK. Genome-wide association studies: Finally, with respect to control of population substruc- results from the first few years and potential implications for ture, we analyzed European-ancestry individuals separately clinical medicine. Annu Rev Med. 2011;62:11-24. and included principal components of ancestry in our ana- 6. Hirschhorn JN, Lohmueller K, Byrne E, Hirschhorn K. A com- prehensive review of genetic association studies. Genet Med. lysis. The paper referenced by Bougnères (10) describes the 2002;4(2):45-61. effects of a subtle signal of population stratification on tests 7. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, of selection that integrate information across genome-wide Rothman N. Assessing the probability that a positive report is sets of summary association statistics for adult height; subtle false: an approach for molecular epidemiology studies. J Natl residual stratification is unlikely to have affected the repli- Cancer Inst. 2004;96(6):434-442. cation, or lack thereof, of individual variants in our study. 8. Chanock SJ, Manolio T, Boehnke M, et al; NCI-NHGRI Working Group on Replication in Association Studies. Replicating genotype-phenotype associations. Nature. Acknowledgments 2007;447(7145):655-660. Financial Support: This work was supported by Pfizer. 9. McCarthy MI, Abecasis GR, Cardon LR, et al. Genome-wide as- sociation studies for complex traits: consensus, uncertainty and challenges. Nat Rev Genet. 2008;9(5):356-369. Additional Information 10. Sohail M, Maier RM, Ganna A, et al. Polygenic adaptation Correspondence and Reprint Requests: Michael P. Wajnrajch, on height is overestimated due to uncorrected stratification in MD, MPA, Department of Pediatric Endocrinology, Pfizer Bio- genome-wide association studies. eLife. 2019;8:e39702.