Supplementary appendix

Transcription

1 Supplementary appendix This appendix formed part of the original submission and has been peer reviewed. We post it as supplied by the authors. Supplement to: Hoogman M, Bralten J, Hibar DP, et al. Subcortical brain volume differences in participants with attention deficit hyperactivity disorder in children and adults: a cross-sectional mega-analysis. Lancet Psychiatry 2017; published online Feb 15.

2 Supplementary Appendix Hoogman et al. Supplement to: Subcortical brain volume differences of participants with ADHD across the lifespan: an ENIGMA collaboration Content: smethods stable 1. Additional information on procedures and methods at the participating sites. stable 2. Clinical characteristics and IQ for cases and controls per cohort. stable 3. Results of the mega-analysis of subcortical left and right brain volumes. stable 4. Overview of average subcortical volumes and ICV per site, corrected for covariates. stable 5. Results of the mega-analysis with and without the addition of FreeSurfer versions as covariate stable 6. The contribution of handedness to the mega-analysis model stable 7. Meta-analysis: Effect sizes (Cohen s d) for the mean volume of each brain region, after controlling for age, sex, and intracranial volume stable 8. Linear effects of age and age*diagnosis stable 9. The results of the fractional polynomial analysis. stable 10. Results of the meta-analysis of the correlation between ADHD symptom scores and subcortical volumes and ICV that were significantly different in cases compared to controls. stable 11. Effect of comorbid disorders on subcortical volumes and ICV sfigure1. Boxplots of the bilateral subcortical volumes. sfigure2. Forest plots: results of the meta-analysis of case-control differences in subcortical brain volume. sfigure3. Number of subjects per age bin. sfigure4. Results of the fractional polynomial analysis for the subcortical brain volumes. sfigure5. Forest plot: result of the meta-analysis of the correlation between total number of ADHD symptoms and caudate volume. 1

3 smethods Explaining mega- and meta-analysis By mega-analysis we refer to the method of pooling all available individual data from all cohorts and analyzing the main effects in one single analysis. In the meta-analysis, we are first determining the main effects in each sample separately and then combine only those summary statistics in one meta-analysis. The advantage of the mega-analysis over the meta-analysis is the additive increase in degrees of freedom thereby increasing power, whereas the advantage of the meta-analysis is that site effects can be rigorously eliminated. Imaging quality control methods The following FreeSurfer labels were extracted: Left-Thalamus-Proper, Right-Thalamus-Proper, Left-Caudate, Right-Caudate, Left-Putamen, Right-Putamen, Left-Pallidum, Right-Pallidum, Left-Hippocampus, Right-Hippocampus, Left-Amygdala, Right-Amygdala, Left-Accumbens-area, Right-Accumbens-area. Next, by running a script called make_pngsfs.m segmentations were projected onto the t1 brain scan. These images were all inspected using SurfScan. Bad segmentations were excluded for further analysis. Statistical outliers were identified when a volume was smaller or larger than 1.5 times the interquartile range. Scripts and protocols are available on ( Statistical framework of the meta-analysis The R-package metaphor (version ) was used to perform an inverse variance-weighted, random-effects meta-analysis, in accordance with other ENIGMA Working Groups 2,3. A random-effects model was chosen, as we expected heterogeneity of effect sizes across samples. We fitted all random-effects models using the restricted maximum likelihood method (REML 4 ). The model includes diagnosis (case=1, control =0), age, sex and ICV. In the meta-analysis of ICV, ICV was omitted as covariate. To calculate Cohen s d effect size estimates, adjusted for age, sex, and ICV, we used the t-statistic from the Diagnosis predictor in the equation 5 for each sample separately. We also ran the model excluding the ICV term. To correct for testing 8 brain volumes, we used Bonferroni correction (significance threshold p=0.006). Extended modeling of age We calculated a moving average for the predicted volumes across age, incorporating sex, ICV, and site. The moving average was obtained by averaging brain volumes using a 5- year sliding window with 1-year increments. Secondly, we used fractional polynomials (Stata/SE 11.2, StataCorp LP, Texas, USA) to search for the optimal model of age for each volume using one- and two-term curvilinear models. The fractional polynomials were calculated separately for the participants with ADHD and healthy controls. These models all included age, sex, ICV, and site. The best-fitting model was chosen among the 8 possible one-term models and among 44 possible two-term models (for the following powers: -2, 1, 0.5, 0, 0.5, 1, 2, 3). Deviance for the linear model versus the one- and two-term models was calculated to identify the best model for each volume and participant group 6. Clinical measures and subcortical volume and ICV As ADHD symptom scores from different sites had been assessed using varying instruments and raters (see stable1 and 3), we performed a meta-analysis of the correlation between ADHD symptom scores and subcortical volumes and ICV, controlling for age, gender, and ICV (the latter only in the subcortical volume analyses), using Medcalc v12.5 (MedCalc Software, Ostend, Belgium). We restricted this random-effects meta-analysis to cases, because healthy controls had often been selected based on low ADHD symptom scores. To explore possible effects of the most frequent co-morbid disorders (see stable 1 & 3 for instruments used and prevalence in our cohorts) on the subcortical volumes and ICV, we first scored the psychiatric disorders as a positive lifetime history or negative lifetime history for all subjects based on the information we had from the clinical interviews. Subsequently, we added the following terms to the mega-analysis model: presence/absence of a psychiatric disorder, presence/absence of depression, presence/absence of anxiety disorder, presence/absence of substance use disorder. We did this in separate analyses because of collinearity 2

4 stable 1. Additional information on procedures and methods at the participating sites. Sample Reference Free- Surfer version Field strength of the MRI scanner Medication withheld during imaging Washout period medication before imaging Classification system for diagnosis Instrument for comorbidity assessment Instrument for symptom rating IQ instrument ADHD-WUE Conzelmann et al., Biol Psychiatry Tesla Partly hours to days DSM-IV SCID1 DSM-IV interview MWT-B ADHD-DUB1 McCarthy et al., JAMA Psych Tesla Yes 48 h DSM-IV SCID1 Conners Adult ADHD rating scale observer Verbal Comprehension, Perceptual Reasoning, Working Memory and Processing Speed subtests of WAIS-IV ADHD-DUB2 Frodl et al., 2010 Amico et al., Tesla No no washout DSM-IV SCID1 Conners Adult ADHD rating scale observer NA ADHD-Mattos Cocchi et al., J Neuroscience Tesla not medication on not applicable DSM-IV MINI K-SADS adapted for adults WASI ADHD200-KKI Tesla unknown unknown DSM-IV NA Conners Parent Rating Scale Revised Long version WISC-IV ADHD200-NYU Tesla yes 24h DSM-IV NA Conners Parent Rating Scale Revised Long version WASI ADHD200-Peking Tesla yes 48h DSM-IV NA ADHD rating scale WISCC-R ADHD200-OHSU Tesla yes 24-48h DSM-IV NA Conners rating scale 3rd edition Block Design, Vocabulary and Information subtests of WISC- IV 3

5 ADHD-UKA Vloet et al., 2010, Konrad et al., 2006, Herpertz 2008, Hubner et al., 2008, Krinzinger et al., Tesla Yes 48h ICD10 K-SADS and German K-Dips German Parental and Teacher Report on ADHD CPM (N = 30)/WASI (N = 49)/WISC-IV (N = 14) BergenadultADHD Dramsdahl et al., Front Psychiatry Tesla Partly 48h ICD-10 or DSM-IV NA NA WASI Bergen-SVG Tesla not on medication not applicable DSM-IV K-SADS-PL K-SADS PL WISC-IV 4

6 DAT-London Paloyelis et al., JAACAP Tesla Yes 48h DSM-IV NA NA Vocabulary, Similarities, Picture Completion and Block Design subtests of WISC/WAIS IMpACT-NL Hoogman et al., AMJP Tesla Yes 24h DSM-IV SCID1&2 DSM-IV interview Vocabulary and block design subtests of WAIS MGH Seidman et al., Biol. Psychiatry Tesla Yes 24h DSM-IV SCID1 DSM-IV interview Vocabulary and block design subtests of WAIS NICHE de Zeeuw et al, PloS One Tesla partly 0-24h DSM-IV DISC-IV NA Vocabulary and block design subtests of WISC- III NYU ADHD UAB-ADHD ZI-CAPS ADHD-Rubia NeuroImage- ADAM Tesla Yes 24h DSM-IV SCID NA Tesla Yes 48h DSM-IV NA NA ODD and CD with Tesla Yes 48h DSM-IV structured clinical interview Lim et al, Psychological Tesla Yes 48h DSM-IV Co-morbid disorders Medicine 2015 were exclusion criteria von Rhein et al, ECAP 2014 Kiddie SADS SDQ for Hyperactive impulsive symptoms and Conners Parent Rating scale revised for Inattentive symptoms Tesla Yes 48h DSM-IV K-SADS-PL Algorithm Von Rhein, see reference WASI WISC Subscales HAWIK-IV WASI Vocabulary and block design subtest of WAIS/WISC of NeuroImage-NIJM von Rhein et al, ECAP Tesla Yes 48h DSM-IV K-SADS-PL Algorithm Von Rhein, see reference Vocabulary and block design subtest of WAIS/(WISC 5

7 NIH MTA Shaw et al, Biological psychiatry Tamm et al, Drug and Alc. Dep Tesla Yes 36h DSM-IV DICA NA Subtests of WISC Tesla Yes 24h DSM-IV NA NA WISC-III full version (N = 87)/subtests of WISC-III (N = 42) All adult ADHD cases were (retrospectively) diagnosed with ADHD in childhood. SCID: Structured Clinical Interview for DSM disorders, MINI: M.I.N.I. International Neuropsychiatric Interview, K-SADS: Kiddie Schedule for Affective Disorders and Schizophrenia; K-DIPS: Kinder Diagnostische Interview bei psychischen Störungen, DISC-IV: Diagnostic Interview Schedule for Children, K-SADS-PL; Kiddie Schedule for Affective Disorders and Schizophrenia, Present and Lifetime; DICA: Diagnostic Interview for Children and Adolescents. SDQ: Strenghts and Difficulties questionnaire. MWT-B: Mehrfachwahl-Wortschatz-Intelligenz-Test, WAIS-IV: Wechsler Adult Intelligence Scale Fourth Edition, WASI: Weschler Abbreviated Scale of Intelligence, WISC-IV: Wechsler Intelligence Scale for Children Fourth Edition, WISCC-R: Wechsler Intelligence Scale for Chinese Children-Revised, CPM: Colored Progressive Matrices, WAIS-III: Wechsler Adult Intelligence Scale Thirth Edition, WISC-III: Wechsler Intelligence Scale for Children Thirth Edition, HAWIK-IV: Hamburg-Weschsler-Intelligentztest fuer Kinder-IV 6

8 stable 2. Clinical characteristics and IQ for cases and controls per cohort Cohort ADHD symptom score in cases only* N Lifetime co-morbid disorder present* N Lifetime co-morbid depression present* N Lifetime co-morbid anxiety* N Lifetime co-morbid SUD* IQ* ADHD-WUE cases ADHD-WUE controls ADHD-DUB1 cases ADHD-DUB1 controls ADHD-DUB2 cases ADHD-DUB2 controls ADHD-Mattos cases ADHD-Mattos controls ADHD200-KKI cases ADHD200-KKI controls ADHD200-NYU cases ADHD200-NYU controls ADHD200-Peking cases ADHD200-Peking controls ADHD200-OHSU cases ADHD200-OHSU controls ADHD-UKA cases Mean (SD; n) Yes/No/Unknown Yes/No/Unknown Yes/No/Unknown Yes/No/Unknown Mean (SD; n) 13 1 (2 6;37) 41/19/2 30/30/2 14/45/3 7/52/3 117 (16;56) - 3/53/0 3/53/0 0/56/0 0/56/0 114 (14;50) (22 9;34) 17/19/0 0/32/4 0/32/4 15/19/2 113 (12;36) - 8/31/0 0/39/0 1/38/0 9/29/1 103 (13;33) 30 5 (9 1;20) 7/13/0 7/13/0 1/19/0 0/20/0 NA - 0/0/0 0/0/0 0/0/0 0/0/0 NA 12 4 (2 4;17) 8/9/0 3/14/0 7/10/0 3/14/0 114 (8;17) - 0/0/0 0/0/0 0/0/0 0/0/0 NA (19 6;22) 0/0/25 0/0/25 0/0/25 0/0/ (10;69) - 0/0/69 0/0/69 0/0/69 0/0/ (15;25) (17 4;148) 0/0/151 0/0/151 0/0/151 0/0/ (14;102) - 0/0/109 0/0/109 0/0/109 0/0/ (14;146) 50 5 (8 6;95) 0/0/102 0/0/102 0/0/102 0/0/ (13;142) - 0/0/143 0/0/143 0/0/143 0/0/ (13;102) (18 6;36) 0/0/42 0/0/42 0/0/42 0/0/ (13;67) - 0/0/67 0/0/67 0/0/67 0/0/ (14;42) 13 8 (1 8;102) 34/67/1 0/101/1 6/95/1 0/101/1 110 (13;79) 7

9 ADHD-UKA controls Bergen-adultADHD cases Bergen-adultADHD controls Bergen-SVG cases Bergen-SVG controls DAT-London cases DAT-London controls IMpACT-NL cases IMpACT-NL controls MGH-ADHD cases MGH-ADHD controls NICHE cases NICHE controls NYU ADHD cases NYU controls UAB-ADHD cases UAB-ADHD controls ZI-CAPS cases ZI-CAPS controls ADHD-Rubia cases - 0/0/79 0/0/79 0/0/79 0/0/ (12;85) NA 0/0/38 0/0/38 0/0/38 0/0/ (9;37) - 0/0/43 0/0/43 0/0/43 0/0/ (14;29) 9 6 (3 6;25) 0/0/25 0/0/25 0/0/25 0/0/25 96 (9;29) - 0/0/29 0/0/29 0/0/29 0/0/29 97 (12;24) NA 0/0/27 0/0/27 0/0/27 0/0/ (11;29) - 0/0/29 0/0/29 0/0/29 0/0/ (11;27) 12 9 (3 3;104) 79/46/0 57/68/0 25/100/0 22/103/0 110 (15;120) - 23/97/0 12/108/0 6/114/0 6/114/0 107 (15;125) 9 4 (4 2;79) 61/18/0 46/33/0 17/62/0 38/41/0 113 (13;69) - 29/40/0 15/54/0 2/67/0 21/48/0 115 (13;78) NA 34/44/0 0/78/0 0/78/0 0/78/0 106 (13;80) - 0/80/0 0/80/0 0/80/0 0/80/0 102 (15;78) NA 20/20/0 14/26/0 5/35/0 9/31/0 112 (11;39) - 7/33/0 3/37/0 2/38/0 3/37/0 111 (12;39) NA 0/0/103 0/0/103 0/0/103 0/0/103 NA - 0/0/95 0/0/95 0/0/95 0/0/ (16;13) 8 8 (4 6;22) 0/0/22 0/0/22 0/0/22 0/0/ (19;13) - 0/0/13 0/0/13 0/0/13 0/0/ (16;22) 88 8 (11 4;36) 0/44/0 0/44/0 0/44/0 0/44/0 110 (11;33) 8

10 ADHD-Rubia controls NeuroImage-ADAM cases NeuroImage-ADAM controls NeuroImage-NIJM cases NeuroImage-NIJM controls NIH cases NIH controls MTA cases - 0/33/0 0/33/0 0/33/0 0/33/0 92 (12;41) 13 4 (3 0;97) 31/64/2 0/0/97 0/0/97 0/0/ (13;85) - 14/71/0 0/0/85 0/0/85 0/0/85 95 (14;97) 13 2 (2 8;139) 62/73/4 1/0/138 0/0/139 0/0/ (14;39) - 5/34/0 0/0/39 0/0/39 0/0/39 98 (15;139) NA 0/0/251 0/0/251 0/0/251 0/0/ (11;235) - 0/0/251 0/0/251 0/0/251 0/0/ (15;233) NA 0/0/88 0/0/88 0/0/88 0/0/ (22;40) MTA controls *see for instruments used stable 1, NA is Not Available - 0/0/41 0/0/41 0/0/41 0/0/ (15;87) 9

11 stable 3. Results of the mega-analysis of subcortical left and right brain volumes. Side p-value for Diagnosis Cohen s d Accumbens Left 2 92x Right Amygdala Left 4 13x Right 6 35x Caudate Left Right Hippocampus Left Right Pallidum Left Right Putamen Left Right Thalamus # Left Right # thalamus volume was not available from the NIH sample. 10

12 stable 4. Overview of average subcortical volumes and ICV per site, corrected for covariates. Accumbens Amygdala Caudate Hippocampus Pallidum Putamen Thalamus ICV Site Dx Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD ADHD-WUE ADHD-DUB1 ADHD-DUB2 ADHD-Mattos ADHD200-KKI ADHD200-NYU ADHD200-Peking ADHD200-OHSU Cases 462,3 74,4 1447,0 131,7 3520,6 335,4 4287,9 300,9 1414,7 143,3 4794,2 470,0 8012,7 540, ,8 Controls 448,9 74,4 1479,4 131,7 3527,2 335,4 4239,4 301,0 1391,9 143,4 4857,3 470,1 8088,7 540, ,2 Cases 712,9 90,5 1804,0 135,7 3998,4 371,7 4724,1 351,5 1805,7 191,7 6289,8 555,5 8207,1 523, ,9 Controls 717,5 90,5 1844,2 135,6 4135,7 371,6 4732,5 351,5 1802,9 191,7 6318,4 555,3 8370,2 522, ,1 Cases 524,1 57,9 1566,8 82,5 3840,4 432,2 4474,2 296,8 1580,7 153,0 5413,9 550,4 8626,2 694, ,3 Controls NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA Cases 630,8 62,2 1745,9 108,5 4453,0 472,4 4509,8 340,9 2084,3 229,1 5996,5 484,4 8120,5 446, ,9 Controls NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA Cases 688,1 80,1 1544,3 139,9 4139,4 359,3 3879,0 288,5 2037,5 189,1 6602,3 477,9 7463,0 450, ,3 Controls 719,9 79,6 1589,4 138,3 4250,1 357,3 3850,4 287,3 2082,5 188,4 6757,4 476,2 7546,2 448, ,4 Cases 721,6 95,8 1472,5 165,1 4067,9 424,1 3734,1 336,8 1894,2 183,7 6561,9 576,5 7162,0 470, ,7 Controls 749,0 96,8 1447,9 166,8 4028,1 429,5 3829,6 340,5 1919,3 185,7 6639,6 582,4 7208,1 475, ,2 Cases 692,0 94,0 1667,7 170,6 4066,5 399,9 4118,1 288,1 1829,0 192,2 6519,6 603,4 7630,3 493, ,2 Controls 687,0 93,0 1663,4 168,6 4033,3 395,9 4141,5 285,3 1821,8 190,2 6571,3 597,5 7643,2 488, ,8 Cases 703,1 90,1 1551,2 163,6 4033,7 443,7 4223,3 354,6 1926,2 162,2 6167,4 537,1 7662,7 501, ,1 Controls 720,6 89,4 1641,0 162,1 4249,3 439,4 4230,2 351,3 1919,4 160,3 6323,2 531,9 7598,0 497, ,7 ADHD-UKA Cases 646,7 81,6 1631,4 172,1 3999,7 435,6 4422,2 310,9 1791,5 176,5 5958,9 546,5 8247,0 497, ,0 11

13 Controls 700,4 82,3 1742,5 173,5 4097,0 438,5 4629,2 313,1 1722,6 178,3 6212,4 551,2 8115,7 502, ,1 Bergen-adultADHD Bergen-SVG DAT-London IMpACT-NL MGH NICHE NYU ADHD UAB ADHD ZI-CAPS Cases 689,7 78,0 1674,7 155,3 3903,4 436,9 4273,5 383,4 1457,4 152,4 5848,4 476,1 7359,9 506, ,8 Controls 715,1 77,8 1728,2 154,8 4066,0 435,7 4341,1 382,9 1522,0 152,1 5976,7 476,1 7492,5 505, ,5 Cases 757,8 87,7 1719,3 145,1 4101,0 420,9 4444,3 337,6 1730,0 146,2 6592,3 551,4 7517,2 452, ,8 Controls 805,1 87,4 1720,4 145,1 4201,0 420,1 4261,5 337,3 1683,3 146,1 6258,7 550,9 7371,1 452, ,1 Cases 486,4 56,8 1292,4 110,4 3038,1 348,5 3558,7 256,1 1310,0 112,7 4509,1 355,8 6464,5 471, ,6 Controls 468,2 56,7 1275,9 110,1 3126,4 348,1 3689,4 254,5 1285,3 112,6 4571,5 354,6 6720,9 469, ,3 Cases 585,4 68,4 1398,1 127,8 3846,2 339,2 3836,2 285,3 1878,0 163,8 5532,5 502,1 7838,1 577, ,4 Controls 584,0 68,4 1402,1 127,8 3870,2 339,1 3856,6 285,3 1883,2 163,8 5596,5 502,1 7855,3 577, ,4 Cases 809,5 114,0 1504,1 180,6 3489,8 357,3 3891,2 288,5 1706,5 153,6 6131,3 506,6 6720,7 543, ,6 Controls 820,1 114,2 1535,0 180,9 3485,1 357,7 3895,5 289,2 1692,1 153,8 6224,8 507,2 6608,7 544, ,0 Cases 600,9 67,1 1582,1 127,3 4087,7 503,5 4365,1 298,5 1898,6 159,5 5956,5 467,8 8117,7 522, ,4 Controls 608,9 67,1 1584,4 127,2 4107,2 503,4 4339,0 298,4 1909,8 159,4 5921,5 467,6 8193,0 522, ,1 Cases 704,8 80,9 1614,2 192,3 3880,5 436,6 4102,7 340,3 1676,7 189,1 6207,1 535,9 7238,1 605, ,6 Controls 718,1 80,9 1570,8 192,3 3835,6 436,6 3977,9 340,3 1685,8 189,1 6137,1 535,8 7279,8 605, ,1 Cases 652,9 112,9 1604,6 159,1 3673,6 423,1 4345,1 338,3 1685,8 212,0 5795,4 595,3 7326,9 585, ,3 Controls 692,9 113,0 1657,4 159,3 3786,3 423,4 4368,6 338,5 1684,4 212,2 5978,1 595,8 7437,2 585, ,4 Cases 721,0 110,3 1724,9 138,5 3989,5 414,3 4495,5 303,9 1614,0 152,3 6170,7 682,2 8076,0 478, ,1 Controls 653,5 115,0 1718,5 144,1 3883,0 431,8 4563,4 314,8 1612,5 157,3 6171,7 711,1 7990,7 491, ,7 ADHD-Rubia Cases 755,4 88,7 1768,9 172,9 4005,9 426,9 4521,7 340,4 1652,7 142,6 6441,9 600,9 7429,1 463, ,5 12

14 Controls 746,0 89,1 1735,7 173,5 4062,2 428,7 4354,9 342,2 1672,8 143,3 6569,6 603,6 7426,6 465, ,4 NeuroImage-ADAM NeuroImage-NIJM NIH MTA Cases 686,1 78,8 1572,5 132,9 4347,6 460,0 4099,6 283,1 1888,9 179,5 6429,5 588,8 7961,7 508, ,2 Controls 683,0 78,9 1586,0 133,1 4277,6 460,7 4094,8 283,6 1886,3 179,9 6354,9 589,7 7820,9 509, ,4 Cases 640,6 73,4 1462,5 130,9 4076,3 342,3 3954,0 244,0 1811,0 168,9 6090,9 422,9 7736,1 498, ,9 Controls 640,3 73,8 1517,9 131,8 4125,2 344,9 4013,2 245,4 1792,4 170,1 6016,2 425,1 7778,4 503, ,6 Cases 647,6 81,0 1517,2 136,2 3803,4 396,7 4238,3 316,0 1651,3 145,7 5896,9 461,0 NA NA ,9 Controls 643,6 81,0 1544,5 136,1 3875,5 396,7 4293,2 315,8 1666,1 145,7 5954,5 461,0 NA NA ,0 Cases 612,8 98,9 1572,5 164,5 4027,4 432,2 4270,4 366,8 1861,3 231,4 5930,4 496,3 8100,8 794, ,5 Controls 629,3 100,5 1588,9 167,5 4120,9 438,9 4212,6 373,3 1806,8 234,9 6017,3 503,8 8101,4 806, ,9 Displayed are the average subcortical volumes and ICV. The subcortical volumes are adjusted for Age, Sex and ICV. ICV is adjusted for Age and Sex only. NA= not available. 13

15 stable 5. Results of the mega-analysis with and without the addition of FreeSurfer versions as covariate Model without FreeSurfer version Model with FreeSurfer version Cohen s d (90% CI) p-value Cohen s d (90% CI) p-value Accumbens ( ) 4 98x ( ) 5 34x10-5 Amygdala ( ) 3 69x ( ) 3 45x10-7 Caudate ( ) ( ) Hippocampus ( ) ( ) Pallidum ( ) ( ) 0 93 Putamen ( ) 6 36x ( ) 7 32x10-5 Thalamus ( ) ( )

16 stable5. The contribution of handedness to the mega-analysis model. Brain volume p-value handedness in the model Accumbens 0.81 Amygdala 0.46 Caudate 0.54 Hippocampus 0.36 Pallidum 0.99 Putamen 0.45 Thalamus

17 stable 6. Meta-analysis: Effect sizes (Cohen s d) for the mean volume of each brain region, after controlling for age, sex, and intracranial volume. Brain volume N Cases/Controls Cohen s d ± SE p-value 95% CI I 2 Accumbens 1650/ ± ( ) 46% Amygdala 1598/ ± ** ( ) 45% Caudate 1660/ ± ** ( ) 10% Hippocampus 1592/ ± ( ) 52% Pallidum 1651/ ± ( ) 8% Putamen 1661/ ± ** ( ) 4% Thalamus # 1406/ ± ( ) 9% ICV 1694/ ± ** ( ) 14% **significant at Bonferroni-corrected threshold (<0 006), # thalamus volume was not available from the NIH sample. I 2 represents the percentage of the observed variation due to differences between samples. Data from 21 cohorts were included, two were excluded because of their patient-only design. 16

18 stable7. Linear effects of age and age*diagnosis p-value for the term age in the model p-value for the term age*diagnosis in the model Accumbens 4 70x Amygdala Caudate 2 95x Hippocampus Pallidum 2 62x Putamen 1 30x Thalamus 1 03x ICV

19 stable 8. The results of the fractional polynomial analysis. Participants with ADHD Healthy controls Subcortical volume Optimal model of age controlling for sex, ICV, sample R 2 Optimal model of age controlling for sex, ICV, sample R 2 Accumbens Linear model of age 0 43 Linear model of age 0 48 Amygdala 2 term model of age (0,3) term model of age (-2) 0 46 Caudate Linear model of age 0 43 Linear model of age 0 32 Hippocampus 1 term model of age (-2) term model of age (0,0) 0 48 Pallidum Linear model of age 0 50 Linear model of age 0 52 Putamen Linear model of age term model of age (-1,-0) 0 50 Thalamus 2 term-model of age (-2,3) term-model of age (-1,3) 0 49 ICV 2 term-model of age (0 5, 0 5) term-model of age (0 5,1)

20 stable 9. Results of the meta-analysis of the correlation between ADHD symptom scores and subcortical volumes and ICV that were significantly different in cases compared to controls. Analysis in all cases Analysis in all childhood cases N Correlation coefficient 95%CI p-value random N Correlation 95%CI p-value random effects analysis coefficient effects analysis Accumbens Amygdala Caudate * Hippocampus Putamen ICV *Also see sfigure5 for the Forest plot. 19

21 stable 10. Effect of comorbid disorders on subcortical volumes and ICV; displayed are p-values. Any disorder Depression/dysthymia Anxiety SUD Accumbens Amygdala Caudate Hippocampus Putamen ICV Displayed are the p-values for the variables presence/absence of any psychiatric disorder, presence/absence of depression/dysthymia, presence/absence of anxiety disorder, and presence/absence of substance use disorder in the mega-analysis. 20

22 sfigure1. Boxplots of the bilateral subcortical volumes. 21

23 22

24 23

25 24

26 25

27 26

28 27

29 28

30 29

31 30

32 31

33 32

34 33

35 34

36 35

37 sfigure2. Forest plots: results of the meta-analysis of case-control differences in subcortical brain volume. 36

38 37

39 38

40 39

41 40

42 41

43 42

44 43

45 sfigure3. Number of subjects per age bin. 44

46 sfigure4. Results of the fractional polynomial analysis for the subcortical brain volumes. 45

47 46

48 47

49 48

50 49

51 50

52 51

53 52

54 53

55 54

56 55

57 56

58 57

59 58

60 59

61 60

62 sfigure5. Forest plot: result of the meta-analysis of the correlation between total number of ADHD symptoms and caudate volume. 61

63 References 1. Viechtbauer W. Conducting meta-analysis in R with the metafor package. Journal of Statistical Software 2010; 36(3): van Erp TG, Hibar DP, Rasmussen JM, et al. Subcortical brain volume abnormalities in 2028 individuals with schizophrenia and 2540 healthy controls via the ENIGMA consortium. Mol Psychiatry Schmaal L, Veltman DJ, van Erp TG, et al. Subcortical brain alterations in major depressive disorder: findings from the ENIGMA Major Depressive Disorder working group. Mol Psychiatry Harville DA. Maximum likelihood approaches to variance component estimation and to related problems. Journal of the American Statistical Association 1977; 72(358): Nakagawa S, Cuthill IC. Effect size, confidence interval and statistical significance: a practical guide for biologists. Biol Rev Camb Philos Soc 2007; 82(4): Sauerbrei W, Meier-Hirmer C, Brenner A, Royston P. Multivariable regression model building by using fractional polynomials: Description of SAS, STATA and R programs. Computational Statistics & Data Analysis 2006; 50(12):