Diagnostic yields reported in the literature vary enormously — the same test applied to superficially similar patients can return yields from 10% to 72%. Before citing any yield number clinically, you must understand the six variables that drive this variance: **1. Cohort selection** is the biggest driver. Referral-centre or specialty-clinic cohorts are heavily enriched for complex, treatment-resistant, or multi-system disease — yielding 40–72% in some leukodystrophy series. Population-level or primary-care cohorts testing milder phenotypes yield 15–25% for the same test. Always ask: who was in the denominator? **2. Sequencing strategy — singleton vs. trio.** Sequencing a proband alone misses de novo variant detection opportunities and cannot phase compound heterozygous calls. Trio analysis (proband + both parents sequenced simultaneously) approximately doubles yield for de novo-enriched phenotypes: OR 2.04 (95% CI 1.62–2.56; Clark et al. 2018). For severe early-onset DEE, GDD, or complex MCA, trio sequencing is not optional. **3. Prior testing history.** A cohort enrolled because CMA was negative will show lower WES yield than first-tier WES — the CMA has already removed a fraction of diagnoses. Post-CMA-negative WES yield for ID is ~25–35%, while first-tier WES yield in the same phenotype is ~30–45%. **4. Gene and database maturation.** The number of validated gene-disease associations grows continuously. WES data from 2018 reanalyzed in 2024 yields new diagnoses in 10–25% of previously unsolved cases — some series report 25–56% incremental yield on reanalysis. A negative WES result is time-stamped, not permanent. **5. CMA platform.** SNP-based arrays detect regions of autozygosity (AOH) and uniparental disomy (UPD) — clinically important for imprinting disorders (Angelman, Prader-Willi) and recessive disease in non-consanguineous families. Oligo-only arrays miss this information. Platform matters when ordering. **6. Severity and syndromic burden.** Distinctive features, epilepsy comorbidity, multi-system involvement, and younger age at onset all independently predict higher yield in multivariate analyses. Mild isolated phenotypes (isolated ASD without ID, mild GDD) consistently show the lowest yields across all platforms.

**Chromosomal Microarray (CMA)** detects copy number variants (CNVs) ≥50–200 kb, aneuploidy, and — on SNP platforms — regions of autozygosity and UPD. It does not detect single-nucleotide variants, small indels, balanced rearrangements, or most repeat expansions. It is inexpensive, rapid, and has decades of clinical validation. Pooled yield across broad pediatric NDD cohorts: ~10% (95% CI 8–12%; Clark et al. 2018, n=20,068). **Whole Exome Sequencing (WES)** captures the ~2% of the genome encoding proteins, detecting SNVs and small indels in coding regions. It misses deep intronic variants, regulatory elements, most balanced structural variants, and repeat expansions. CNV detection from WES is possible but less sensitive than CMA. Pooled yield across pediatric NDD cohorts: ~36% (95% CI 33–40%; Clark et al. 2018; Pandey et al. 2025 meta-analysis of 24,631 probands). WES is now the standard first-tier test for undiagnosed NDD/ID at most academic centres. **Whole Genome Sequencing (WGS)** sequences the entire genome, additionally detecting: (a) balanced structural variants (inversions, translocations), (b) deep intronic and regulatory variants, (c) mitochondrial variants at improved depth, (d) short tandem repeat screening — modern WGS pipelines may screen for some STR disorders using tools like ExpansionHunter, though sensitivity is variable. WGS allows simultaneous SNV + CNV analysis on one platform. Pooled yield across NDD cohorts: ~41% (95% CI 34–48%; Clark et al. 2018). Notably, head-to-head meta-analysis shows WGS vs. WES yield is not significantly different overall (adjusted OR 1.13, p=0.50; Nurchis et al. 2023), suggesting the incremental gain is concentrated in specific subgroups: post-WES-negative patients (~10–20% incremental yield), leukodystrophies with suspected deep intronic pathology, and atypical presentations. **The repeat expansion caveat — critical for practice**: Standard WES does NOT detect trinucleotide or pentanucleotide repeat expansions. This includes: Friedreich ataxia (FXN GAA), most SCA types (CAG repeats), CANVAS (RFC1 AAGGG biallelic repeat), FXTAS/fragile X (FMR1 CGG premutation), myotonic dystrophy (DM1 CTG, DM2 CCTG), Huntington disease (HTT CAG), C9orf72 ALS/FTD, and FAME (SAMD12/MARCHF6 TTTCA). Modern WGS pipelines may screen for some short tandem repeat (STR) disorders using computational tools (e.g., ExpansionHunter, STRipy), but sensitivity is variable and not yet validated for all loci — dedicated repeat-primed PCR, Southern blot, or long-read sequencing (Oxford Nanopore or PacBio) remain the gold standard. Always assess whether the clinical phenotype suggests a repeat disorder before declaring testing negative as 'final.'

**NICU / Neonatal Encephalopathy**: For neonates with unexplained encephalopathy, seizures, or hypotonia without clear perinatal cause, CMA yields 8–15% (aneuploidies, large CNVs); rapid WES (rWES) 20–35%, rising to ~50% in highly selected acute cohorts; rapid WGS (rWGS) 35–50%, with some studies ranging 20–72% depending on cohort selection. In a 2023 JAMA RCT (Maron et al.), rWGS and rWES showed comparable yields. Approximately 18% of NICU admissions are estimated to carry a Mendelian disease. Critically, a molecular diagnosis changed management in 38–50% of diagnosed cases — the highest clinical utility figures in all of paediatric genetics. **Developmental and Epileptic Encephalopathy (DEE — broadly)**: This heterogeneous group (EIMFS, Dravet, LGS, EIDEE) has CMA yield 8–15%, WES 24–40%, WGS 35–50%. Within DEE, the most specific syndromes have the highest yields: EIMFS and Dravet ~78%, early infantile DEE ~43%, IESS 17–26% (see below). In a 2025 monocentric cohort of 1,000 paediatric epilepsy patients (Lemoinne et al.), overall NGS yield was 37%. **Infantile Epileptic Spasms Syndrome (IESS / West syndrome)**: The most precisely studied epilepsy subgroup. Meta-analysis of 13 studies (n=629 for CMA, n=799 for WES): CMA yield 14% (95% CI 11–16%), WES yield 26% (95% CI 21–31%). WGS yield comparable to WES at 19–25%. TSC inclusion/exclusion significantly affects these numbers. Crucially, a genetic diagnosis enables precision therapy decisions in up to 61.6% of genetically explained IESS cases. **New-Onset Focal Epilepsy (non-DEE, normal development)**: Lower yield reflects a more benign, heterogeneous phenotype. CMA 5–8%, WES 10–15%, WGS 15–20%. Gene panel is a reasonable first step; escalate to WES/WGS if panel negative or phenotype evolves. **New-Onset Generalized Epilepsy (non-DEE)**: CMA 5–10%, WES 12–18%, WGS 18–25%. Point mutations in SCN1A, GABRA1, GABRG2, and related genes dominate — CMA rarely diagnostic here; panel or WES is preferable. **Refractory / Drug-Resistant Epilepsy (any onset, no clear structural cause)**: Drug resistance is an independent predictor of higher yield in multivariate analyses. CMA 8–12%, WES 24–40%, WGS 35–48%. Across all epilepsy phenotypes, the Sheidley et al. 2022 meta-analysis (154 studies, n=39,094) reports: CMA 9%, WES 24%, WGS 48%.

**Global Developmental Delay (GDD — infant/toddler, unexplained)**: CMA 8–12%. WES first-tier 30–40%, rising to 50–61% in trio + CNV-seq (Zhang multicenter 2024, n=434). WGS 35–45%. Trio approach nearly doubles yield vs. singleton (OR ~2.04). Moderate-to-severe impairment, age 12–24 months, and craniofacial abnormalities are the strongest predictors of high yield. **Intellectual Disability (ID — child/adolescent, unexplained)**: CMA 8–12%. WES first-tier 30–45%; post-CMA-negative WES 25–35%. WGS first-tier 35–50%. The ACMG 2021 guideline supports WES/WGS as first- or second-tier for ID (Manickam et al.). The UK/Ireland DDD study (2023, Wright et al.) demonstrated 41% WGS trio yield in a large ID cohort. The French DEFIDIAG 2025 RCT of WGS-trio first-line for ID showed significantly higher yield than standard testing strategy in both previously-tested and never-tested groups. **Isolated ASD (without ID, epilepsy, or distinctive features)**: Lowest yield in neurodevelopment. CMA 5–8%, WES 10–15%, WGS 12–18%. This reflects the polygenic and multifactorial architecture of isolated ASD. However, with comorbid ID (~20–30% yield) or comorbid epilepsy, ASD approaches GDD/ID yields. CMA retains value even without diagnosis for recurrence counseling. **Hypotonia ± Motor Delay (infant/toddler)**: CMA 10–15%, WES 30–45%, WGS 35–50%. Heterogeneous — NMJ, congenital myopathy, and congenital neuropathy variants may need RNA-seq as an adjunct (splicing variants missed by WES). A 2022 Neurology consensus review on neonatal hypotonia supports early WES/WGS. **Multiple Congenital Anomalies (MCA ± DD ± hypotonia)**: CMA is a productive first step at 15–25%. WES 35–50%, WGS 40–55%. Combined WGS yield in MCA+ID can reach 50–62% in monocentric cohorts (2024). Consanguinity raises the yield of recessive diagnoses substantially. **Cerebral Palsy (unexplained; no clear perinatal/structural risk factor)**: CMA ~8–12% (JAMA Neurol 2022 meta-analysis). WES yield: overall 31.1% (95% CI 24.2–38.6%; JAMA Pediatr 2023 meta-analysis, Gonzalez-Mantilla et al.); pediatric-specific 34.8%; with strict exclusion criteria (no perinatal risk) 42.1%. WGS comparable. CP without clear risk factors has highest yield; CP + ID yields 37.8% vs. 17.6% without ID. **Macrocephaly (isolated or with other features)**: CMA 10–20% (higher with MCA). WES 20–40% (isolated macrocephaly lower; syndromic/NDD ~35–45%). WGS 25–45%. CMA detects PTEN-region duplications, NF1 region, and 16p11.2 dup. PTEN, PIK3CA (somatic; may need dedicated somatic panel), NF1, and RAS-MAPK pathway genes require sequencing. Somatic mosaic overgrowth syndromes (PIK3CA, AKT3) may need deep-coverage sequencing or tissue biopsy.

**Episodic Ataxia (child/adolescent)**: CMA rarely diagnostic (5–8%) — episodic ataxias are almost exclusively caused by point mutations in ion channel or pump genes. WES yield 20–35% (varies by clinical pre-selection). WGS 20–40%, slightly higher for structural variant detection. Key genes: KCNA1 (EA1), CACNA1A (EA2), ATP1A3 (AHC/RDP spectrum), SCN8A. Important caveat: CACNA1A repeat expansions causing EA2-like phenotype and FHM subtypes are not reliably detected by standard WES/WGS. A targeted gene panel is competitive with WES in a well-defined episodic ataxia phenotype. **Progressive (Hereditary) Ataxia (child/adolescent)**: CMA 5–10%. WES 21–50% (wide range depending on cohort selection and prior testing). In specialty hereditary ataxia cohorts, WES yields 40–50%. Fogel et al. 2020 (UCLA cohort) defined a diagnostic ceiling of ~50% for WES in properly selected ataxia cohorts. WGS 40–55%, with the additional benefit of detecting regulatory variants and deep intronic changes in genes such as POLG and AFG3L2. **Critical repeat-expansion caveat for ataxia**: The most common hereditary ataxias worldwide are NOT diagnosed by WES or standard WGS: - **Friedreich ataxia (FA)**: GAA trinucleotide repeat expansion in FXN intron 1 — requires repeat-primed PCR - **Spinocerebellar ataxias (SCA1/2/3/6/7/10/17/36)**: CAG or other repeat expansions — require repeat analysis - **CANVAS**: Biallelic AAGGG repeat expansion in RFC1 — requires dedicated Southern blot or long-read sequencing - **FXTAS**: FMR1 CGG premutation — requires FMR1 PCR/Southern blot - **FAME**: TTTCA repeat expansions in SAMD12, MARCHF6 — long-read sequencing required In a patient with progressive ataxia and negative WES, always ask: has dedicated repeat expansion testing been sent? **Leukodystrophy / Leukoencephalopathy (MRI-selected, genetic suspected)**: CMA has low standalone yield for most leukodystrophies (10–15%) but is useful as an adjunct to detect PLP1 duplication (Pelizaeus-Merzbacher), ALD deletion (ABCD1 in X-linked ALD), and MLC-associated deletions. WES in MRI-selected cohorts: **50–72%** — among the highest yields in clinical neurogenetics. The GWMD cohort (Neurology 2022, n=126) reported 72% overall; 77% for onset <3 years; 85% in the hypomyelination subgroup. WGS in dedicated leukodystrophy programmes: **72–90%+** (Zerem et al. 2023: 89.6% with all modalities combined including RNA-seq). WGS advantages are critical here: deep intronic variants in EIF2B genes (VWM) and POLR3A/B, regulatory region variants, and pseudogene-masking resolution. **The leukodystrophy key principle**: MRI pattern recognition dramatically narrows the differential and should guide testing. After correct MRI categorisation (hypomyelination vs. demyelination vs. cystic vs. vacuolating), directing WES/WGS to the appropriate gene-disease context yields 50–89% — making leukodystrophy one of the highest-yield scenarios in all of neurology.

**Summary Pooled Yields Across Pediatric Neurogenetics / NDD Cohorts**: | Test | Pooled Yield | Source | |---|---|---| | CMA | ~10% (95% CI 8–12%) | Clark et al. 2018, n=20,068 | | WES | ~36% (95% CI 33–40%) | Clark et al. 2018; Pandey et al. 2025, n=24,631 | | WGS | ~41% (95% CI 34–48%); NSD vs. WES | Clark et al. 2018; Nurchis et al. 2023 (OR 1.13, p=0.50) | | rWGS (NICU) | 35–50%; comparable to rWES in RCT | Maron et al. JAMA 2023; NICUSeq 2021 | | All epilepsy | CMA 9%, WES 24%, WGS 48% | Sheidley et al. Epilepsia 2022, n=39,094 | **Six Critical Pedagogical Points**: **1. CMA is not obsolete.** It uniquely detects aneuploidy, large CNVs, UPD (SNP arrays), and regions challenging for WES (segmental duplications). In MCA, IESS, and ID, CMA yield rivals or complements WES at lower cost. Many labs now perform simultaneous CNV-seq with WES. **2. Trio sequencing is not optional for severe early-onset phenotypes.** De novo variants explain the majority of severe DEE, early-onset ID, and complex MCA. Singleton WES misses phasing, parent-of-origin information, and has roughly half the de novo detection rate of trio analysis. **3. WGS > WES for specific scenarios**: post-negative-WES cohorts (~10–20% incremental yield); leukodystrophies with suspected deep intronic/regulatory pathology; atypical CP; combined SNV+CNV analysis on one platform. **4. Repeat expansion disorders are a separate testing universe.** Friedreich ataxia, most SCA types, CANVAS, FXTAS, DM1/DM2, HD, C9orf72, FAME — none reliably detected by standard short-read WES or WGS. Always assess whether the phenotype suggests a repeat disorder and order dedicated testing. **5. Yield is dynamic.** Reanalysis of existing WES/WGS at 12–24 months yields new diagnoses in ~10–25% of previously unsolved cases (some series: 25–56% incremental on reanalysis of older WES). A negative WES is time-stamped, not permanent. **6. Leukodystrophy is the special case.** After MRI pattern recognition, WES/WGS in the correct phenotypic context yields diagnoses in 50–89% — among the highest yields in clinical neurogenetics. MRI pattern should guide, not replace, genetic testing. **Clinical utility ≠ diagnostic yield**: Reaching a diagnosis changes management in 2–50% of diagnosed cases depending on phenotype. High-impact examples include: SCN8A → quinidine; KCNQ2 → retigabine/carbamazepine; pyruvate dehydrogenase deficiency → ketogenic diet; SLC6A1 → valproate avoidance; GLUT1 → ketogenic diet; NICU diagnosis → withdrawal of life support or initiation of targeted therapy.