Development and Performance of a Comprehensive Targeted Sequencing Assay for Pan-Ethnic Screening of Carrier Status

The Journal of Molecular Diagnostics, Vol. 16, No. 3, May 2014

jmd.amjpathol.org

Development and Performance of a Comprehensive Targeted Sequencing Assay for Pan-Ethnic Screening of Carrier Status Alice K. Tanner,* C. Alexander Valencia,* Devin Rhodenizer,* Marina Espirages,* Cristina Da Silva,* Lisa Borsuk,y Sara Caldwell,y Edward Gregg,y Elizabeth Grimes,y Agnieszka M. Lichanska,y Leah Morris,y Anjan Purkayastha,y Brian Weslowski,y Clark Tibbetts,y Matthew C. Lorence,y and Madhuri Hegde* From the Emory Genetics Laboratory,* Department of Human Genetics, Emory University, Atlanta, Georgia; and TessArae LLC,y Potomac Falls, Virginia Accepted for publication December 11, 2013. Address correspondence to Madhuri Hegde, Ph.D., Emory University Department of Human Genetics, 615 Michael St, Ste 301, Atlanta, GA 30322. E-mail: mhegde@ emory.edu.

Identifying individuals as carriers of severe disease traits enables informed decision making about reproductive options. Although carrier screening has traditionally been based on ethnicity, the increasing ethnic admixture in the general population argues for the need for pan-ethnic carrier screening assays. Highly multiplexed mutation panels allow for rapid and efficient testing of hundreds of mutations concurrently. We report the development of the Pan-Ethnic Carrier Screening assay, a targeted sequencing assay for routine screening that simultaneously detects 461 common mutations in 91 different genes underlying severe, early-onset monogenic disorders. Mutation selection was aided by the use of an extensive mutation database from a clinical laboratory with expertise in newborn screening and lysosomal storage disease testing. The assay is based on the Affymetrix GeneChip microarray platform but generates genomic DNA sequence as the output. Analytical sensitivity and specificity, using genomic DNA from archived control cultures and from clinical specimens, was found to be >99% for all mutation types. This targeted sequencing assay has advantages over multiplex PCR and next-generation sequencing assays, including accuracy of mutation detection over a range of mutation types and ease of analysis and reporting of results. (J Mol Diagn 2014, 16: 350e360; http://dx.doi.org/ 10.1016/j.jmoldx.2013.12.003)

Carrier screening is genetic testing performed on asymptomatic adult individuals to determine their heterozygous status for mutations that can cause severe disease in their offspring. Identifying carrier individuals and couples provides them with a variety of reproductive options, including pre-implantation genetic diagnosis, use of donor gametes, prenatal testing, and adoption. Although carrier screening may be performed biochemically for some genetic disorders such as Tay-Sachs disease, for many disorders biochemical carrier testing is not available or carrier individuals may have inaccurate results, thus requiring molecular mutation analysis for accurate carrier detection. Carrier screening for a specific inherited disease can be performed when there is a known family history of the disease and causative mutations have been identified. In addition, carrier screening can be performed on individuals Copyright ª 2014 American Society for Investigative Pathology and the Association for Molecular Pathology. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jmoldx.2013.12.003

without a family history on the basis of their ethnicity and genetic diseases that are known to occur with higher frequency in those ethnicities. For example, the American Congress of Obstetricians and Gynecologists (ACOG) recommends that all individuals of Ashkenazi Jewish (AJ) descent who are pregnant or considering pregnancy be offered carrier screening for founder mutations for four diseases known to occur at higher frequency in that population [Tay-Sachs disease, Canavan disease, cystic fibrosis

Supported by Emory Genetics Laboratory operating funds. A.K.T. and C.A.V. contributed equally to this work. Disclosures: L.B., S.C., E.Gre., E.Gri., A.M.L., L.M., A.P., B.W., C.T., and M.C.L. are employed by TessArae, LLC., which owns the GeneCipher algorithms, the PECS array, the reference and mutant synthetic templates, the PECS multiplexed primer mixes, and the AlleleConfirm Kit used in this study.

Targeted Pan-Ethnic Carrier Screening (CF), and familial dysautonomia].1 The recommendations of the American College of Medical Genetics and Genomics (ACMG) for carrier screening in individuals of AJ descent include the four diseases recommended by ACOG and also include founder mutations for five additional diseases (Gaucher disease, Niemann-Pick disease type A, mucolipidosis IV, Fanconi anemia group C, and Bloom syndrome).2 Because other AJ founder mutations are identified in different diseases, these mutations and diseases are often added to AJ carrier screening panels offered by clinical testing laboratories. Other diseases for which carrier screening based on ethnicity is recommended include hemoglobinopathies and CF. Carrier screening for hemoglobinopathies is recommended for individuals of African, Southeast Asian, and Mediterranean descent.3 Although carrier screening for CF was initially recommended for the non-Hispanic white population and those of AJ descent, it is now recommended that it be offered to all individuals, given the increase in individuals of mixed ethnicity and the difficulties that may be present in determining a person’s ethnicity that is based on personal reporting.4 Given the changing ethnic admixture in the general population (the 2010 US Census indicated that more than half of the total US population growth between 2000 and 2010 was because of increases in the Hispanic population), the increasing use of donor gametes in assisted reproductive technology, and the decreasing cost of high throughput screening of many mutations simultaneously, it has been suggested that carrier screening be offered to the general population, regardless of ethnicity.5e7 Another argument for pan-ethnic carrier screening is that newborn screening (NBS) programs, which include screening for CF and hemoglobinopathies, are not ethnicity based. Reasons for this include that general population-based screening is more equitable in avoiding missed diagnoses of treatable diseases8 and the inaccuracy, incompleteness, and difficulty in obtaining ethnicity information,8e10 both of which also apply to carrier screening. In addition, as more mutations and diseases are added to current ethnicity-based carrier screening panels, carrier frequencies for some disorders may approach those in the general population, such as in AJ panels that include screening for spinal muscular atrophy and fragile X syndrome. The complexities of offering carrier screening for certain diseases only to specific ethnicities but offering carrier screening for other diseases to the general population could be alleviated by offering pan-ethnic carrier screening. The recent development of highly multiplexed carrier screening panels that are capable of concurrently reporting hundreds of causal mutations for many different monogenic diseases represents a far more informative and efficient approach than traditional sequential, single-gene testing. To this end, Emory Genetics Laboratory (EGL), in collaboration with TessArae, LLC, has developed a targeted sequencing assay for rapid routine screening that simultaneously detects 461 loci in 91 different disease-causing genes. This Pan-Ethnic Carrier Screen (PECS) assay focuses on disorders that are part of

The Journal of Molecular Diagnostics

-

jmd.amjpathol.org

recommended NBS panels in addition to several other metabolic and lysosomal storage diseases. The diseasecausing loci were chosen on the basis of prevalence in diverse ethnic populations and from EGL’s proprietary mutation database that was amassed from the global patient populations referred to EGL for clinical testing. The PECS assay excludes some disorders, included in other commercially available carrier screening panels, that target milder and/or later onset diseases or rare mutations that are restricted to small populations. The assay is performed on the Affymetrix GeneChip microarray platform, a proven and standard laboratory technology for gene expression and cytogenetic analysis,11 which uses a resequencing technology that generates gold standard genomic DNA sequence as the output,12 unlike other panels developed on multiplexed PCR or hybridization probe platforms. Results of an extensive validation study that found the analytical sensitivity and analytical specificity of the assay, as well as results from a set of patient specimens, are presented in this report.

Materials and Methods Microarray Design The PECS assay is performed with an Affymetrix microarraybased targeted DNA sequencing platform. For each targeted mutation there are two detector tiles on the microarray. One detector tile is identical to the reference (wild-type; WT) allele sequence, and the other tile is identical to the mutant allele sequence. Each detector tile represents DNA sequences from the reference human genome sequence build 37 (hg19), and definitions of most of the targeted mutations are found in the Human Genome Mutation Database (BioBase). These detector tiles comprise overlapping sets of eight 25-base long oligonucleotide probes per base of sequence to be interrogated. The detector tiles enable reading of at least 24 bp, and, in a few cases, up to 101 bp of template DNA sequences in the immediate proximity of each targeted mutation locus. For a small number of loci representing large (>300 bp) deletions, the reference detector tile represents one of the two breakpoint sequences, whereas the mutant tile represents the junction sequence of the two breakpoints, and genomic sequences hybridizing to these tiles are amplified by two different primer pairs instead of the same primer pair as with the single nucleotide variations or small insertions and deletions.

Synthetic Template Design and Construction To verify the presence and to assess the performance of all of the reference and mutant detector tiles on the array, 24 different synthetic DNA sequences that contain concatenated sets of approximately 20 different reference allele sequences each were manufactured by OriGene Technologies, Inc (Rockville, MD) and cloned into plasmids (pWT). A corresponding set of 24 different synthetic DNA sequences that contain concatenated sets of the mutant allele sequences were

351

Tanner et al manufactured and cloned into plasmids (pMUT). Together these 48 plasmids represented the detector tile sequences for all of the 461 targeted mutation loci on the microarray.

Amplification Primer Design Candidate amplification primer pairs for targeted mutations were identified from the reference genome DNA sequence (GRCh37) with the use of the Oligo primer analysis software version 7.41 (Molecular Biology Insights, Inc, Cascade, CO). The initial primer selection round identified one or more primer pairs for each target mutation that met multiple default criteria as follows: i) amplicon length between 250 and 400 bp (unless sparse search results required longer); ii) predicted Tm > 50 C and Tm < 68 C with primer pair Tm difference <2 C; and iii) secondary structure having autologous DG > 2.5 kcal/mol and heterologous DG > 6.0 kcal/mol as inferred from nearest neighbor free energy change values, including mismatches (DG). Further screening of candidate primer pairs included target gene specificity for literature references to pseudogenes. Several hundreds of candidate primer pairs were then evaluated in singleplex amplifications from control human genomic DNA template (BioChain Institute Inc, Newark, CA), accepting those yielding predominately single amplicon products of predicted lengths. Amplification was performed in 384-well microtiter plates, using a temperature gradient thermal cycler, and quantification of amplicon yields was performed by capillary electrophoresis (CaliperLife Sciences). Satisfactory candidate primer pairs were next evaluated for amplification in 1 of the 32 multiplex categories according to observed optimal Tm and need for enhancer reagent for GC-rich primers. These multiplex amplifications were evaluated by hybridization and sequence analysis on the PECS microarray. Fine-tuning of the multiplex configuration involved limited relocations of individual primer pairs to different primer pair mixtures, or addition of particular primer pairs to multiple wells to increase amplicon yields. Final configuration primer sets were compared with a catalog of known genomic sequence polymorphisms (NCBI dbSNP build 137). In the few cases in which primer sequences overlapped common (>1% allele frequency) polymorphisms, the primers were modified as mixed nucleotides matching both alleles cited in dbSNP.

371 different primer pairs in 32 individual highly multiplexed reactions in a 384-well plate with the use of one of two different amplification master mixes (either with or without GC-Enhancer; Roche Applied Sciences, Indianapolis, IN). Each reaction consisted of 25 ng/mL of genomic DNA and between 1 and 18 different primer pairs, using 35 cycles of 95 C for 30 seconds, followed by a thermal gradient of 50 C to 65 C for 90 seconds, followed by 72 C for 150 seconds on a C1000 Touch thermal cycler (Bio-Rad, Hercules, CA). After amplification, the products from the 32 individual reactions were pooled and purified with a Wizard SV spin column (Promega, Madison, WI) according to the manufacturer’s instructions, and the concentration was determined from a 1:10 dilution of the products in nuclease-free water on a Nanodrop ND-1000 instrument. Purified products (20 ng) were fragmented and end-labeled with biotinylated ddUTP by using the Resequencing Assay Kit (Affymetrix, Santa Clara, CA) according to the recommended Affymetrix

Preparation of Genomic DNA, Amplification, and Microarray Hybridization Genomic DNA templates were purchased from the Coriell Institute for Medical Research (Camden, NJ) or extracted from patient samples by using the OrageneDx kit (Orasure Technologies, Bethlehem, PA) according to the manufacturers’ instructions. Extracted genomic DNA was diluted to a final concentration of 5 ng/mL immediately before use in the assay. Genomic DNA samples were assayed as outlined in Figure 1. Briefly, total genomic DNA was amplified by using

Figure 1 Overview of the PECS assay protocol. Genomic DNA samples are amplified in 32 highly multiplexed reactions, pooled, fragmented, labeled, and hybridized overnight to the PECS array. Arrays are washed, stained, and scanned the next morning, and the resulting .cel files are analyzed by the GeneCipher analysis algorithms. ON, overnight; PECS, Pan-Ethnic Carrier Screen; PDF, portable document format; QC, quality control.

352

jmd.amjpathol.org

-

The Journal of Molecular Diagnostics

Targeted Pan-Ethnic Carrier Screening protocol and hybridized to the PECS array (TessArae, LLC, Potomac Falls, VA). Microarray hybridization and processing were per the Affymetrix protocol.

Data Analysis After microarray scanning, the resulting data (.cel) file was analyzed with the GeneCipher data analysis algorithms (TessArae, LLC). GeneCipher assesses the signal intensity from all probes comprising all of the detector tiles on the array and converts the signal intensity from the probes into homozygous nucleotide assignments (A, C, G, or T) or, when indicated by the data, into heterozygous assignments (R, Y, K, M, S, or W) at each position. If the aggregate probe data cannot support one of these calls, a less-stringent model reports lower quality (a, c, g, or t) base calls or uncalled bases (N). An assay-wide DNA sequence data quality metric, the average reference quality (ARQ) metric, evaluates the DNA sequences that are read and reported across the centers of all of the reference allele detector tiles as an assessment of the assay performance. For any genomic DNA template, the majority (>99%) of all targeted mutation loci are likely to be WT, reporting sequences that typically match or nearly match the corresponding reference allele detector tiles. The ARQ metric for each PECS assay is based on scoring of base calls from the ninth base to the 17th base reported from each reference allele detector tile, as þ1 for each called base (A, C, G, T) in the 9-bp scoring interval; no increment of score for ambiguous or low-quality salvaged base calls (R, Y, K, M, S, or W, and a, c, g, or t); and 1 penalty score for each uncalled base (N). Thus, the DNA sequence quality metric for individual loci (RQ) and the assay-wide ARQ metric are reported in the range of 9 to þ9. The highest level of DNA sequence quality for individual tiles is 8 to 9, and a minimum satisfactory score for an entire assay would be ARQ  8.5. If the ARQ score is satisfactory, GeneCipher evaluates the sequences read from each pair of reference and mutant allele detector tiles as a genotype signature. If a mutation is a single nucleotide substitution, then WT genomic templates read WT sequence from the reference allele tile. Sequences read from the mutant allele may also match the reference allele tile. In other cases, the mismatch of the mutant allele leads to significantly deteriorated sequences read from the mutant allele detector tile. A heterozygous (HET) template reads matching sequences from both detector tiles, except at the center nucleotide (base 13) mutation locus. This base is the ambiguous base assignment that represents the distinguishing bases of the two alleles (as R, Y, K, M, S, or W). A homozygote (HOM) template typically reads the reference and mutant allele detector tiles in a manner complementary to the WT rule above. If the target mutation is an insertion or deletion (indel), a WT template or HOM template reads the matching sequence from the reference or mutant allele detector tiles, respectively. A HET indel template reads the matching sequences from both reference and mutant allele detector tiles. When the template is reported HOM for certain

The Journal of Molecular Diagnostics

-

jmd.amjpathol.org

deletion loci, other targeted loci within the span of the deletion are not present in the template and are reported as not present. When the template is reported HET for certain deletion loci, other targeted loci within the span of the deletion appear to be HOM because there is only one allele present and it is the mutated allele. In such contingency cases, the HOM call is reported as HEMI. GeneCipher will report no allele call (NC) for a target locus in instances that lack satisfactory DNA sequence data quality to otherwise report WT, HET, HEMI, or HOM. The number of NCs per assay is correlated with lower assay ARQ metrics, and the number of NCs rises rapidly for assays that report ARQ < 8.5 (data not shown). GeneCipher further evaluates quantitative relationships among signals from a larger set of probes in the localized area of the first nucleotide that distinguishes the reference allele from the mutant allele. This statistical metric is highly informative for the WT, HET, or HOM status of any single nucleotide substitution, deletion, and/or insertion mutation type. Supplemental Figure S1 illustrates the typical relationship between the allele status and the statistical metric for the hemochromatosis gene c.187C>G (p.H63D) locus. The final determination of the mutation type is accomplished through serial analysis of the nucleotide assignments, genotype signature, and statistical metric, one target locus at a time, independent of reported results from all other targeted loci. GeneCipher then automatically generates a report that lists all of the detected mutant loci with the corresponding genotype.

Strategy for Detection of Large Indel Mutations Nine of the targeted mutations of the PECS assay panel represent large deletions of reference allele sequences, from 619 bp (hemoglobin, beta gene del619bp) to 232 kb (gap junction protein, beta 6, 30 kDA gene del232kb ex1-ex6). Two tiles represent the reference and mutant alleles of each of these large deletion mutations, as is the case for all targeted mutations of the PECS assay. For each large deletion mutation, a mutant allele detector tile spans the specified deletion junction as illustrated in Supplemental Figure S2. Amplification primers for those alleles are located near (200 bp) the deletion junction, in the upstream (50 or left) and downstream (30 or right) genome sequence. The reference allele of some of these large deletions could be amplified by the same primers as the mutant allele. Amplification of the reference allele in this manner, however, would be less efficient than amplification of the mutant allele, because the reference allele would yield a much larger amplicon that contained the otherwise deleted reference sequence between the WT deletion breakpoints. Instead, separate primer pairs were selected to amplify short sequences (<400 bp) spanning the left and the right WT deletion breakpoints. The PECS assay uses either the left or the right deletion breakpoint amplicon, depending on which (if either) consistently performs better in assays with genomic DNA templates, to detect the WT allele. Because two different amplicons (with independent yields in each assay) are used for the reference and mutant alleles of

353

Tanner et al Table 1 Monogenic Diseases and Associated Genes Targeted by the PECS Assay and the Number of Mutations Targeted in Each Targeted mutations

Category/disease or syndrome

Gene

Cystic fibrosis Lysosomal storage disorders Alpha-Mannosidosis Cystinosis Fabry disease Gaucher disease GM1 gangliosidosis/MPS type IVB Krabbe disease Metachromatic leukodystrophy Mucolipidosis type II/IIIA Mucolipidosis IV Niemann-Pick disease type A/B Niemann-Pick disease type C Pompe disease (GSD II) Sandhoff disease Tay-Sachs disease MPS syndromes MPS type I (Hurler/Scheie syndrome) MPS type II (Hunter syndrome) MPS type IIIA (Sanfilippo syndrome type A) MPS type IIIB (Sanfilippo syndrome type B) MPS type IVA (Morquio syndrome type A) MPS type VI (Maroteaux-Lamy syndrome) MPS type VII (Sly syndrome) Amino acid metabolism disorders ASL deficiency Citrullinemia type 1 Homocystinuria MSUD type 1A MSUD type 1B MSUD type 2 Phenylketonuria OTC deficiency Tyrosinemia type I Organic acid disorders Fumarase deficiency Glutaric acidemia I Isovaleric acidemia Ketothiolase deficiency Methylmalonic aciduria, MUT-related (MMA) Methylmalonic aciduria, MMAA-related (MMA) Methylmalonic aciduria, MMAB-related (MMA) Methylmalonic aciduria, clbC type (MMA) Fatty acid oxidation disorders Carnitine deficiency (carnitine uptake defect) CPT IA deficiency CPT II deficiency Long chain 3-hydroxyacyl-CoA dehydrogenase deficiency

CFTR

75

MAN2B1 CTNS GLA GBA GLB1 GALC ARSA GNPTAB MCOLN1 SMPD1 NPC1 GAA HEXB HEXA

3 10 7 11 3 5 11 1 2 7 4 3 1 9

IDUA IDS SGSH

4 3 7

NAGLU

7

GALNS ARSB GUSB

3 3 4

ASL ASS1 CBS BCKDHA BCKDHB DBT PAH OTC FAH

4 4 5 1 3 1 18 2 4

FH GCDH IVD ACAT1 MUT

1 2 1 3 6

MMAA

1

MMAB

2

MMACHC

1

SLC22A5

3

CPT1A CPT2 HADHA

2 19 1

Table 1

(continued ) Gene

Medium chain acyl-CoA dehydrogenase deficiency Short chain acyl-CoA dehydrogenase deficiency Very long chain acyl-CoA dehydrogenase deficiency Glycogen storage diseases GSD type Ia GSD type Ib GSD type III Carbohydrate metabolism disorders Galactosemia Hereditary fructose intolerance Congenital disorders of glycosylation Congenital disorder of glycosylation type Ia Congenital disorder of glycosylation type Ib Hearing loss GJB2-related nonsyndromic hearing loss GJB6-related nonsyndromic hearing loss Pendred syndrome Usher syndrome type IF Usher syndrome type IIIA Muscle disorders Inclusion body myopathy 2 McArdle disease (GSD V) Muscle-eye-brain disease Nemaline myopathy Blood disorders Beta thalassemia/sickle cell disease Factor V Leiden thrombophilia Factor XI deficiency Fanconi anemia type C Glucose-6-phosphate dehydrogenase deficiency Other ABCC8-related hyperinsulinism Alpha-1 antitrypsin deficiency Ataxia-telangiectasia Ataxia with vitamin E deficiency Autoimmune polyendocrine syndrome type I Autosomal recessive polycystic kidney disease Biotinidase deficiency Bloom syndrome Canavan disease Cartilage-hair hypoplasia Familial dysautonomia Familial mediterranean fever Herlitz junctional epidermolysis bullosa, LAMC2-related Herlitz junctional epidermolysis bullosa, LAMB3-related

ACADM

8

ACADS

1

ACADVL

1

G6PC SLC37A4 AGL

10 2 8

(table continues)

354

Targeted mutations

Category/disease or syndrome

GALT ALDOB

9 5

PMM2

4

MPI

1

GJB2 GJB6 SLC26A4 PCDH15 USH3A

6 1 4 1 3

GNE PYGM POMGNT1 NEB

1 5 1 1

HBB F5 F11 FANCC G6PD

27 1 4 4 3

ABCC8 SERPINA1 ATM TTPA AIRE

3 2 1 1 1

PKHD1

5

BTD BLM ASPA RMRP IKBKAP MEFV LAMC2

7 2 4 2 2 12 1

LAMB3

4

(table continues)

jmd.amjpathol.org

-

The Journal of Molecular Diagnostics

Targeted Pan-Ethnic Carrier Screening

Results Table 1

(continued ) Targeted mutations

Category/disease or syndrome

Gene

HFE-associated hereditary hemochromatosis Joubert syndrome 2 MSUD type 3, aka DLD deficiency Nijmegen breakage syndrome Rhizomelic chondrodysplasia punctata type 1 Sjogren-Larsson syndrome Smith-Lemli-Opitz syndrome Sulfate transporter-related osteochondrodysplasia, including achondrogenesis type 1B, diastrophic dysplasia, and recessive multiple epiphyseal dysplasia Wilson disease Zellweger syndrome (infantile refsum disease) PECS assay totals, genes

HFE

2

TMEM216 DLD NBN PEX7

1 2 1 4

ALDH3A2 DHCR7 SLC26A2

1 13 4

ATP7B PEX1 91

6 2 461

CPT, carnitine palmitoyltransferase; GSD, glycogen storage disease; MPS, mucopolysaccharidosis; MSUD, maple syrup urine disease.

the large deletions, the GeneCipher algorithm gives greater weight in detecting and reporting mutant alleles to the sequence signatures read from the detector tiles as the genotype signature than to the statistical metric. In general, a WT template will provide a matching sequence to the reference tile (left or right deletion breakpoint) and a poor sequence match to the left or right half of the mutant allele deletion junction tile. Similarly, a HOM template will provide a matching sequence to the mutant allele deletion junction tile while reporting a poor sequence on the left or right half of the reference allele deletion breakpoint tile. A HET template typically reports a matching or nearly matching sequence to both the reference and the mutant allele detector tiles.

Orthologous Confirmation of Detected Alleles All mutant loci detected by PECS assays were subjected to Sanger dideoxy sequencing with the use of the Big Dye Terminator version 3.1 Cycle Sequencing kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions for one-half reaction on an ABI PRISM 3500 Genetic Analyzer using POP-7 Polymer with a 50-cm capillary. To date, >90% of the mutant loci detected have been subjected to confirmation by Sanger dideoxy sequencing (and the remaining loci are in progress). In many cases, sequence confirmation of mutant allele calls led to sequence confirmation of incidental WT calls because some of the loci interrogated by the assay are clustered in the same amplicons. The primers developed for orthologous confirmation of detected loci are available for purchase in 96-well plates as the AlelleConfirm Kit and enable amplification and sequencing of each PECS locus individually.

The Journal of Molecular Diagnostics

-

jmd.amjpathol.org

Targeted Disease Genes and Mutations The PECS assay enabled simultaneous screening of genomic DNA templates for 461 mutations in 91 monogenic inherited disease genes. These represented 13 categories of monogenic disorders for which there were common or ethnic-specific mutations, including metabolic disorders, lysosomal storage disorders, CF, mucopolysaccharidosis syndromes, blood disorders, glycogen storage diseases, congenital disorders of glycosylation, and AJ disorders. The monogenic disorders and associated genes, as well as the number of mutations detected in each gene, are listed in Table 1. Detailed descriptions of the targeted mutations are provided in Supplemental Table S1.

Verification of the Reference and Mutant Allele Detector Tile Sequences To verify the specified detector tile sequences used on the array, PECS assays were performed with synthetic DNA sequences that represented the reference (pWT) and mutant (pMUT) allele sequences of all 461 loci on the array. Assays performed with pWT and pMUT templates reported sequences that matched the reference or mutant allele detector tiles, respectively, verifying that the specified detector tile sequences were present on the array and that the sequence information derived from each detector tile when assayed with the corresponding synthetic template was correct. Results from this large series of control assays are summarized for reference allele tile verifications in Supplemental Table S2 and for mutant allele tile verifications in Supplemental Table S3.

Assay Performance DNA Sequence Data Quality and Reporting Detected Alleles To evaluate the performance of the PECS assay, 110 genomic DNA templates acquired from Coriell and 46 genomic DNA Table 2 PECS Assay Results from Carrier Screening of 110 Positive Control Archival DNA Templates (Coriell) and 46 Clinical Laboratory DNA Templates (EGL) PECS assay calls at 464 loci

Coriell archive templates

EGL clinical templates

HET HEMI HOM NC (Uncalled allele) Not present WT Total PECS Calls

272 4 36 297 5 50,096 50,710*

118 2 5 104 1 20,976 21,206y

*110 templates  461 loci per assay Z 50,710 test reports. y 46 templates  461 loci per assay Z 21,206 test reports. EGL, Emory Genetics Laboratory; HEMI, targeted loci within the span of the deletion appear to be HOM because there is only one allele present and it is the mutated allele; HET, heterozygous; HOM, homozygous; NC, not called PECS, Pan-Ethnic Carrier Screen; WT, wild-type.

355

Tanner et al Table 3 Confirmation Testing of PECS Assay Calls by Conventional Sanger Sequencing of the Same Templates PECS assay reports Source of genomic DNA

Total for confirmation, n

Sequencing confirmations HET, n

Coriell archive templates (n Z 110) HET 261 253 HEMI 4 0 HOM 35 1 WT 2804 0 EGL clinical templates (n Z 46) HET 99 96 HEMI 2 0 HOM 5 1 WT 336 0

HEMI, n

HOM, n

WT, n

0 4 0 0

0 0 34 0

8 0 0 2804

0 2 0 0

0 0 4 0

3 0 0 336

EGL, Emory Genetics Laboratory; HEMI, targeted loci within the span of the deletion appear to be HOM because there is only one allele present and it is the mutated allele; HET, heterozygous; HOM, homozygous; NC, not called; PECS, Pan-Ethnic Carrier Screen; WT, wild-type.

templates from clinical specimens at EGL were tested. Table 2 summarizes all 71,916 calls from these 156 different templates at 461 target mutation loci per template. The average ARQ metric for all 156 templates was ARQav Z 8.69  0.18, and the total number of detected mutant alleles, either HET, HEMI, or HOM, was 440 (average of 2.8 per assay). Assays of these 156 genomic DNA templates reported a total of 465 NCs (average of 3.0 per template assay). Supplemental Table S4 contains the nucleotide sequence reported from both the reference and mutant allele tiles for all 461 target mutations in each of the 156 templates, including the ARQ score, genotype signature, and statistical metric value. We note that the frequency of HOM calls is significantly greater among the Coriell Archive templates (36/110; 33%) than among the EGL clinical templates (5/46; 11%). The probability that the frequencies of HOM calls of each cohort are drawn from the same distribution of frequencies is P < 0.003 (t-test). This outcome is consistent with the expectation that the Coriell templates are derived from both carriers and affected individuals, whereas the EGL templates are derived from unaffected carriers. In contrast, the frequency of HET calls (unaffected carrier status) for the Coriell Archive templates (271/110 Z 2.5 HET per template) is not significantly different from the frequency of HET calls among the EGL clinical templates (119/46 Z 2.6 HET per template). The probability that the frequencies of HET calls of each cohort are drawn from the same distribution of frequencies is P < 0.75 (t-test), consistent with a null hypothesis assumption of no significant difference.

The same sequencing ladders used for confirmation of assayreported mutant alleles also enabled confirmation of WT calls at other mutation-proximal loci. Together with additional confirmations of other loci reporting WT, a total of 3546 Sanger sequencing confirmation tests of PECS assay WT and mutant allele calls are summarized in Table 3. With respect to confirmatory Sanger sequencing, the PECS reports of mutant allele detection (HET/HOM) or absence (WT) included no false negative results and 11 false positive results. This corresponds to 99.69% concordance of results from the two orthologous sets of sequencing-based analyses. The PECS assay and GeneCipher data analysis algorithm have been designed to minimize the likelihood of any false negative report, which is a critical aspect of a clinical laboratory assay’s performance. In turn, this leads to a very low rate of false positive reports, and the effectiveness of this approach is reflected in the data shown in Table 3. The analytical sensitivity and analytical specificity of the PECS assay is evaluated for the detection and reporting of mutant alleles in genomic DNA templates (carrier state), compared with reference standard results of Sanger sequencing of the same templates. Table 4 presents traditional 2  2 square analysis tables for presence (HET, HEMI, or HOM) or absence (WT) of mutant alleles as reported by PECS assays and as reported by the Sanger sequencing analysis of the same templates. The calculated analytical sensitivity approaches 100% (95% CI, 98.8%e100.0%), whereas the analytical specificity of the PECS assay is 99.7% (95% CI, 99.4%e99.8%) for the combined results from 110 archival DNA templates and 46 clinical genomic DNA templates. A similar assessment of PECS assay performance was made for the differentiation of carrier genotypes HET (and HEMI) or HOM in all 156 templates that reported mutant alleles, as shown in Table 5. The data presented here are Table 4 Detection and Reporting of Mutant Alleles by PECS Assay Compared with Benchmark Analysis of the Same Templates by Sanger DNA Sequencing Mutant Mutant allele allele Total, present, n absent, n n

Templates Coriell archive templates (n Z 110) PECS assay positive PECS assay negative Total EGL clinical templates (n Z 46) PECS assay positive PECS assay negative Total

292 0 292

8 2804 2812

300 2804 3104

103 0 103

3 336 339

106 336 442

Analytical Sensitivity and Analytical Specificity of PECS Assay Of the total 437 PECS-reported mutant allele calls tallied as HET, HEMI, or HOM in Table 2, 406 (93%) have been subjected to orthologous confirmation by Sanger sequencing.

Combining results from the 110 Coriell archive templates and the 46 EGL clinical templates leads to analytical sensitivity that approaches 100.0% (95% CI, 98.8%e100.0%) and the analytical specificity is 99.7% (95% CI, <99.4%e99.8%). The analytical sensitivity was 100.0% for both the Coriell archive templates and the EGL clinical templates, and the analytical specificity was 99.7% and 99.1%, respectively. EGL, Emory Genetics Laboratory; PECS, Pan-Ethnic Carrier Screen.

356

jmd.amjpathol.org

-

The Journal of Molecular Diagnostics

Targeted Pan-Ethnic Carrier Screening Table 5 PECS Assay Differentiation of HET (and HEMI) from HOM Genotypes in 156 Templates Reporting Detection of Mutant Alleles

PECS assay HOM PECS assay HET Total

HOM by Sanger sequence

HET by Sanger sequence

Total

36 0 36

2 349 351

38 349 387

The analytical sensitivity approaches 100.0% (95% CI, 88.0%e100.0%), and the analytical specificity is 99.4% (95% CI, <97.7%e99.9%). HEMI, targeted loci within the span of the deletion appear to be HOM because there is only one allele present and it is the mutated allele; HET, heterozygous; HOM, homozygous; PECS, Pan-Ethnic Carrier Screen.

concordant, with the analytical sensitivity of correct differentiation of mutant alleles approaching 100% (95% CI, 88.0%e100.0%), whereas the analytical specificity is 99.4% (95% CI, 97.7%e99.9%). Although this evaluation was not a blinded, prospective clinical validation of assay performance, the number of unique samples, the diversity of their respective carrier states, and the >99% concordance of assay-reported mutant alleles (HET, HEMI, and HOM) with Sanger DNA sequence analysis of the same templates and loci suggest that a similarly high level of performance will be obtained in routine clinical implementation of the PECS assay.

Detection Rates of Mutant Alleles Reported by PECS Assay A total of 437 instances of 145 different mutant alleles were detected and reported in PECS assays of the 156 templates, an average of 2.8 mutant allele detections per assay. Supplemental Table S5 presents the number of instances that each of these 145 different mutant alleles were reported by the 156 PECS assays as HET, HEMI, or HOM.

Confirmation of Anecdotal Mutations in Archival Positive Control Templates Positive control templates from Coriell were originally selected for this study on the basis of catalog annotations of the 461 loci that are targeted by the PECS assay. PECS assay results were >98% concordant with annotated positive control mutations among the 110 archival DNA templates from Coriell. Concordant PECS assay results included the following: i) 73 Coriell HET annotations of templates that were concordant with 75 PECS assay HET calls; ii) 23 Coriell HOM annotations of templates that were concordant with 23 PECS assay HOM calls; iii) 2 Coriell annotations of HET deletion mutations that led to 2 contextually correct PECS assay HEMI calls; and iv) 1 Coriell template (NA12785) annotated HET ASPA p.Y231Y, a common silent polymorphism and not the assay-targeted ASPA p.Y231X mutation, that was correctly identified by PECS assay and reported as WT.

The Journal of Molecular Diagnostics

-

jmd.amjpathol.org

Seven discrepancies were identified between Coriell annotations of positive control templates and PECS assay results. Two of the seven Coriell HET annotations were called HOM by PECS assay. Subsequent Sanger sequencing of these two templates indicated one was HET but one was HOM. Five of the seven Coriell HET annotations were called WT by PECS assay. Subsequent Sanger sequencing of these five templates indicated that all were WT. It is possible that these discrepancies may be a result of secondary mutations at PECS assay or Sanger sequencing primer binding sites, leading to drop out of one of the two alleles or allele drop out in the Coriell samples because of repeat passages. However, all primer locations were carefully screened for reported polymorphisms, and primer sequences were modified to match both alleles (as described in Materials and Methods), making the likelihood low.

SRY Gene Sex-Specific Marker Sequences The PECS assay reports sex associated with each template by using sequences from two detector tiles that represent the Y chromosome-specific sex-determining region Y (SRY) gene. Samples from male donors typically report perfect or nearly perfect matches to the two detector tiles (49 of 50 called bases). Lacking the SRY gene, female templates typically report no recognizable sequence from these sex-specific detector tiles. Supplemental Table S6 shows representative results indicating sex of templates from 30 male and 30 female donors. Sex identification is an important context for specimen quality assurance and for the reporting of genotypes associated with X chromosome-linked genes. The PECS panel includes the following four X-linked genes: glucose-6phosphate dehydrogenase (G6PD); galactosidase, alpha; iduronate 2-sulfatase; and ornithine carbarmoyltransferase. Females with two X chromosomes may be either heterozygous or homozygous for one or more mutations in these genes, but males will either be WT or HEMI (one mutant allele copy and no reference allele copy). The PECS assay results for a male with mutation(s) in X chromosome-linked genes appear like HOM but are reported as HEMI. As an example, two of the 30 females identified in Supplemental Table S3 were identified as HET, in one case compound HET, for different targeted mutations of the G6PD gene: NA20270 G6PD HET c.202G>A p.G68R (Female), NA20270 G6PD HET c.376A>G p.N126D (Female), and NA04510 G6PD HET c.563C>T p.S188F (Female). Two males among the 30 identified in Supplemental Table S3 were identified as carriers of a G6PD mutation: CCR-NA16265 G6PD HOM->HEMI c.376A>G p.N126D (Male) and CCR-NA00325 G6PD HET c.563C>T p.S188F (Male). The first of these two males has the expected contextual override of default HOM call to the HEMI call for G6PD p.N126D. The second individual is unequivocally male by the reported SRY sex marker sequences, but his PECS assay reports HET (not HOM) for G6PD p.S188F. This unexpected

357

Tanner et al result is reconciled by reference to the archival annotation for the template as having Klinefelter syndrome, a karyotype with one Y chromosome (hence male) and two X chromosomes. In this situation the individual may be both male and HET for a mutation in an X chromosome-linked disease gene.

Discussion Design of PECS Panel: Selection of Targeted Disorders and Mutations Selection of the specific disorders to be tested is a critical factor in the development of a pan-ethnic carrier screening assay. The panel includes the AJ founder mutations recommended for screening by ACOG and ACMG, in addition to the 23 mutations in the CF transmembrane conductance regulator (ATP-binding cassette subfamily C, member 7) gene recommended for screening.1,2,4 Expanding beyond the ACMG and ACOG recommendations for carrier screening, the panel has significantly increased focus on other severe, early-onset conditions. Because a critical objective of NBS is early detection of infants at risk for life-threatening conditions, disorders included in the ACMG-recommended panel for NBS13 were considered for inclusion on the PECS assay. Although NBS focuses on disorders that are medically actionable, related disorders for which no treatment is currently available were also included for consideration. In general, less severe and/or adult-onset diseases were not included. After selection of disorders to be included, the second consideration for inclusion in the PECS panel is selection of the specific mutations to be targeted. An effective molecular carrier screen must target those mutations that are commonly encountered in affected individuals. Although some diseases, such as CF and sickle cell disease, have wellknown common disease-causing mutations, other diseases are primarily caused by private mutations in which each affected family carries its own unique mutation. Inclusion of private mutations was not considered, because they would not be expected to increase the clinical sensitivity of a carrier screen designed for the general population. For diseases in which common mutations found in specific ethnic groups or in multiple unrelated individuals are not sufficiently documented in the literature, EmBase,14 EGL’s mutation database of clinical samples from both the United States and abroad, was a valuable resource for selection of mutations. This resource was especially useful for mutations associated with metabolic and lysosomal storage diseases, given EGL’s pioneering experience to provide molecular analysis for these genes. Mutations that EGL has identified in multiple unrelated individuals were considered likely to be representative of mutations that may be common in the general population or within certain ethnicities. However, mutations known to be common within specific ethnicities were not included if the ethnicity in question did not constitute a significant portion of the

358

general US population, because inclusion of these mutations would not be relevant for most individuals tested. In addition, mutations for which no causal link with disease has been established or which represent known sequence polymorphisms or variants of unknown significance (VOUSs) were generally excluded from the PECS assay. Specific exceptions were made to include the arylsulfatase A gene pseudodeficiency alleles c.1049A>G (p.N350S) and c.*96A>G, and the hexosaminidase A pseudodeficiency alleles c.739C>T (p.R247W) and c.745C>T (p.R249W). Although these nucleotide changes do not cause disease in vivo, they produce a carrier result in biochemical assays and may be useful for interpretation of biochemical data in carrier individuals or their offspring. A final critical factor is the ability to add new disorders and mutations to increase the comprehensiveness and utility of the assay. Detector tile pairs for mutations in two genes, SMN1 (spinal muscular atrophy) and DHDDS (autosomalrecessive retinitis pigmentosa), are already present on the PECS array, and development of the amplification primers to include these new mutations in the PECS assay is in progress.

Analytical Validation of PECS Assay Performance We report two approaches to analytical validation of PECS assay performance. First, synthetic templates that contain all of the loci in the assay were used to verify the detection and reporting of each locus as a WT, HET, or HOM allele. Second, a number of genomic DNA samples that contain known mutations, when available, were used to confirm the results of the synthetic templates. In the validation study using characterized genomic templates from culture collections (110 samples) and previously characterized patient samples (46 samples), the analytical sensitivity and specificity were found to exceed 99% for all mutation types, including single nucleotide substitutions, small indels, as well as large deletions (>300 bp). There were 71,072 WT calls reported from the 156 PECS assays of genomic DNA templates, and of these, a total of 3140 WT calls were subjected to confirmation by Sanger sequencing with zero discrepancies. Combining the outcome of zero discrepancies with a simple square root estimate of the sampling error suggests that the PECS assay false negative rate is between 0.0% and 1.8%. The assessment of 100% analytical sensitivity (as shown in Table 4) from the entire data aggregate (156 templates) has a comparably wide 95% CI (98.8%e100%). The parameters embedded in the GeneCipher algorithms were intentionally set to call every locus either as mutant (HET) or uncalled (NC) unless the data clearly support a WT call. This minimizes the risk of a false negative result, leading to a slight increase in the risk of a false positive result or NC at any given locus, and the results of this study found that the false positive rate for both sample types is <1%. The analytical sensitivity and specificity of the PECS assay for distinguishing HET versus HOM loci also exceed 99%, with the false negative rate approaching zero. The

jmd.amjpathol.org

-

The Journal of Molecular Diagnostics

Targeted Pan-Ethnic Carrier Screening number of HOM calls in the aggregate data (as shown in Table 2) are not indicative of a high false positive rate, but merely show a threefold higher frequency of HOM genotypes for particular mutations among the archived positive control templates from Coriell, compared with the results from clinically unaffected individuals in the EGL cohort. This is to be expected because the templates represented in the Coriell archive include genomic samples from affected patients and their respective family members.

Consideration of Highly Multiplexed PCR Assays and NGS as Alternative or Complementary Platforms The PECS assay is similar to established high multiplicity PCR assays because each of these methodologies relies on locus-specific amplification of targets from genomic DNA templates. A critical advantage of PECS is that detection and identification of both WT and mutant alleles is based on assay-reported DNA sequences from the reference and mutant allele detector tiles, not simply a PCR-based detection or failure to detect amplification. If a multiplexed PCR assay fails to amplify a targeted allele, the failure to detect the allele does not necessarily indicate absence of that allele in the assay template. In addition, the presence of an unknown polymorphism near the site of any targeted mutation, which may lead to a negative or weakly positive result in a PCR-based assay, will not reduce the ability of the PECS assay to generate high-quality sequence data for accurate detection and identification of the allele(s). As a targeted mutation assay, the PECS assay has several advantages over next-generation sequencing (NGS)-based whole exome sequencing (WES) or whole genome sequencing (WGS). Indels and pseudogenes are not accurately sequenced, and the false positive variant rate in these regions is high. Further, every NGS-based assay has some amount of exon dropout that requires a Sanger-based fill-in strategy to complete the assay, which adds to the cost. In addition, WES/WGS routinely returns multiple VOUSs, that is, nucleotide changes for which the clinical presentation, if any, is uncertain. A VOUS result is not clinically beneficial for patients or providers when making reproductive decisions and may, in fact, lead to increased anxiety or concern. WES/ WGS data will also include information on those mutations that cause milder, later-onset diseases or that are susceptibility alleles rather than disease-causing mutations. Laboratories will then need to decide how to report these changes, and physicians will need to provide pretest and posttest counseling for VOUS results and susceptibility alleles, potentially creating further confusion and uncertainty for patients. If laboratories are following the American College of Genetics and Genomics recommendations on incidental findings and conducting WES/WGS to perform carrier screening, they will then need to analyze and report information on the loci in the recommended panel, even if those loci are not part of the carrier screening panel. In contrast, the PECS assay only includes known mutations that generally

The Journal of Molecular Diagnostics

-

jmd.amjpathol.org

cause severe and/or early-onset diseases and does not report nontargeted loci, eliminating the anxiety caused by VOUS and susceptibility alleles. In addition, data storage and data processing for acquisition and analysis of results from WES/WGS assay, or any NGS assay, require substantial information technology and bioinformatics support, in terms of hardware, software, and personnel. In contrast, GeneCipher analysis of PECS assay data automatically returns a clinical laboratory-friendly report that lists all of the detected mutations along with confidence metrics. GeneCipher/PECS provides a turnaround time that is typically <24 hours, from sample access to results. Although raw data acquisition from particular NGS platforms may only require several hours, the front-end sample handling and library productions and back-end bioinformatics reductions of raw data to genotypes typically require several days. Other advantages of the PECS assay over WES/WGS are that the protocol is highly automated with a throughput of up to 240 samples per week per technician and the cost per sample is the same whether performing the assay on a single patient sample or on dozens of samples, alleviating the need to wait for samples to accumulate to manage assay costs.

Education and Counseling Genetic counseling is an important component of carrier screening. Like any common mutation panel, PECS assay results can greatly reduce an individual’s chance of being a carrier for any of the included disorders, but it will not reduce the risk to zero. An individual who tests negative across targeted mutations of the PECS assay may still be a carrier of a mutation that is not included on the panel. It is imperative that laboratories and providers understand this limitation and counsel their clients accordingly. In some instances, biochemical carrier testing may be used in combination with the PECS assay to further determine a patient’s carrier risk. For example, individuals of AJ descent may also choose to have Tay-Sachs disease enzyme carrier testing, and individuals of African, Southeast Asian, or Mediterranean descent should be offered biochemical hemoglobinopathy carrier testing. The PECS assay complements but does not replace a family history. If an individual has a personal or family history of any of the disorders present on a carrier screening panel, their carrier risk because of that history is much greater than the general population risk. Other carrier testing may be appropriate, such as full gene sequence analysis, on the basis of the disorder and mutations present in the family. Because of these issues, carrier screening should take place in the context of genetic counseling and informed consent.

Large-Scale Impact Potential of PECS Assay and Limitations The PECS assay was primarily developed to facilitate rapid routine patient screening of a large number of disease-

359

Tanner et al causing mutations in a single low-cost assay. The assay has public health as well as medical (genetic) impact potential. Aggregated PECS assay data can support reliable estimates of carrier frequencies for many rare disorders through costeffective, large-scale analysis of patients’ genomic DNA across an extensive list of recessive disorders. Because the PECS assay is based on targeted genotyping, only those carriers of selected disease-causing mutations, not all carriers of any possible mutation associated with each disease, will be reported. PECS-derived disease frequencies are therefore expected to be the lower limits because other disease-causing alleles will not be included. However, by measuring the carrier frequency directly, the PECS assay will be a significant improvement over carrier rate estimates that have been based on incidence rates and challenged by the failure to recognize mild phenotypes or to include child mortality.15 When used in a clinical laboratory, and because not all loci were able to be validated on patient genomic DNA, confirmation of variant calls should be performed before reporting results. This confirmation can be conducted by Sanger sequence analysis or allele-specific PCR reactions for large deletions. TessArae has developed a set of primers that are based on those used for orthologous confirmation of detected alleles (as described in Materials and Methods). In conclusion, the need for pan-ethnic screening has increased because of more individuals reporting mixed racial ancestry, unclear ethnic classifications in medicine, and unknown ancestry through adoption or the use of donor gametes. Preliminary data on carrier screening in the general population indicate that conditions other than those recommended for carrier screening by professional societies are just as prevalent in the general population and that many diseases previously thought to be ethnicity specific are in fact found in individuals outside of those ethnicities.16 To address this need, we developed the PECS assay, which focuses on common mutations in severe, early-onset monogenic disorders. Validation studies found the analytical sensitivity and specificity to be >99% for all mutation types. Pan-ethnic carrier screening for a large number of genetic disorders is technically feasible with the PECS assay at a population level, whereas large-scale implementation of the PECS assay can define mutation frequencies, which will facilitate accurate pretest and posttest counseling in the settings of carrier screening and prenatal diagnosis.

References

Supplemental material for this article can be found at http://dx.doi.org/10.1016/j.jmoldx.2013.12.003.

1. ACOG Committee on Genetics: ACOG Committee Opinion No. 442: preconception and prenatal carrier screening for genetic diseases in individuals of Eastern European Jewish descent. Obstet Gynecol 2009, 114:950e953 2. Gross SJ, Pletcher BA, Monaghan KG; Professional Practice and Guidelines Committee: Carrier screening in individuals of Ashkenazi Jewish descent. Genet Med 2008, 10:54e56 3. ACOG Committee on Genetics: ACOG Practice Bulletin No. 78: hemoglobinopathies in pregnancy. Obstet Gynecol 2007, 109: 229e237 4. American College of Obstetricians and Gynecologists Committee on Genetics: ACOG Committee Opinion No. 486: update on carrier screening for cystic fibrosis. Obstet Gynecol 2011, 117:1028e1031 5. Ross LF: A re-examination of the use of ethnicity in prenatal carrier testing. Am J Med Genet A 2012, 158A:19e23 6. Callum P, Iger J, Ray M, Simes C, Falk RE: Outcome and experience of implementing spinal muscular atrophy carrier screening on sperm donors. Fertil Steril 2010, 94:1912e1914 7. Pletcher BA, Gross SJ, Monaghan KG, Driscoll DA, Watson MS: The future is now: carrier screening for all populations. Genet Med 2008, 10:33e36 8. Hinton CF, Grant AM, Grosse SD: Ethical implications and practical considerations of ethnically targeted screening for genetic disorders: the case of hemoglobinopathy screening. Ethn Health 2011, 16: 377e388 9. Dyson SM, Culley L, Gill C, Hubbard S, Kennefick A, Morris P, Rees D, Sutton F, Squire P: Ethnicity questions and antenatal screening for sickle cell/thalassemia [EQUANS] in England: a randomized controlled trial of two questionnaires. Ethn Health 2006, 11:169e189 10. Aspinall PJ, Dyson SM, Anionwu EN: The feasibility of using ethnicity as a primary tool for antenatal selective screening for sickle cell disorders: pointers from the research evidence. Soc Sci Med 2003, 56:285e297 11. Bissaillon CL, Tyropolis AM, Lebel KA, Moore F, Pfleuger SM, Lefferts JA, Tafe LJ, Steinmetz HB, Tsongalis GJ: Concurrent validation of chromosome microarray testing in separate molecular pathology laboratories (Abstract TT73). J Mol Diagn 2012, 14:746 12. Metzgar D, Myers CA, Russell KL, Faix D, Blair PJ, Brown J, Vo S, Swayne DE, Thomas C, Stenger DA, Lin B, Malanoski AP, Wang Z, Blaney KM, Long NC, Schnur JM, Saad MD, Borsuk LA, Lichanska AM, Lorence MC, Weslowski B, Schafer KO, Tibbetts C: Single assay for simultaneous detection and differential identification of human and avian influenza virus types, subtypes, and emergent variants. PLoS One 2010, 5:e8995 13. Newborn screening: toward a uniform screening panel and system. Genet Med 2006, 8(suppl 1):1Se252S 14. Bean LJH, Tinker SW, da Silva C, Hegde M: Free the data: one laboratory’s approach to knowledge-based genomic variant classification and preparation for EMR integration of genomic data. Hum Mutat 2013, 34:1183e1188 15. Lacassie Y, Storment JM, Lazarin GA: Serendipitous diagnosis of mild recessive multiple epiphyseal dysplasia through parental-targeted screening test. Am J Med Genet 2011, 155A:3136e3138 16. Lazarin GA, Haque IS, Nazareth S, Iori K, Patterson AS, Jacobson JL, Marshall JR, Seltzer WK, Patrizio P, Evans EA, Srinivasan BS: An empirical estimate of carrier frequencies for 400þ causal Mendelian variants: results from an ethnically diverse clinical sample of 23,453 individuals. Genet Med 2013, 15:178e186

360

jmd.amjpathol.org

Supplemental Data

-

The Journal of Molecular Diagnostics