Whole-genome risk prediction of common diseases in human preimplantation embryos.

Preimplantation genetic testing (PGT) of in-vitro-fertilized embryos has been proposed as a method to reduce transmission of common disease; however, more comprehensive embryo genetic assessment, combining the effects of common variants and rare variants, remains unavailable. Here, we used a combination of molecular and statistical techniques to reliably infer inherited genome sequence in 110 embryos and model susceptibility across 12 common conditions. We observed a genotype accuracy of 99.0-99.4% at sites relevant to polygenic risk scoring in cases from day-5 embryo biopsies and 97.2-99.1% in cases from day-3 embryo biopsies. Combining rare variants with polygenic risk score (PRS) magnifies predicted differences across sibling embryos. For example, in a couple with a pathogenic BRCA1 variant, we predicted a 15-fold difference in odds ratio (OR) across siblings when combining versus a 4.5-fold or 3-fold difference with BRCA1 or PRS alone. Our findings may inform the discussion of utility and implementation of genome-based PGT in clinical practice.

Validation of an Enhanced Version of a Single-Nucleotide Polymorphism-Based Noninvasive Prenatal Test for Detection of Fetal Aneuploidies.

OBJECTIVE: To validate an updated version (Version 2) of a single-nucleotide polymorphism (SNP)-based noninvasive prenatal test (NIPT) and to determine the likelihood of success when testing for fetal aneuploidies following a redraw. METHODS: Version 2 was analytically validated using 587 plasma samples with known genotype (184 trisomy 21, 37 trisomy 18, 15 trisomy 13, 9 monosomy X, 4 triploidy and 338 euploid). Sensitivity, specificity and no-call rate were calculated, and a fetal-fraction adjustment was applied to enable projection of these values in a commercial distribution. Likelihood of success of a second blood draw was computed based on fetal fraction and maternal weight from the first draw. RESULTS: Validation of this methodology yielded high sensitivities (≥99.4%) and specificities (100%) for all conditions tested with an observed no-call rate of 2.3%. The no-call threshold for sample calling was reduced to 2.8% fetal fraction. The redraw success rate was driven by higher initial fetal fractions and lower maternal weights, with the fetal fraction being the more significant variable. CONCLUSIONS: The enhanced version of this SNP-based NIPT method showed a reduced no-call rate and a reduced fetal-fraction threshold for sample calling in comparison to the earlier version, while maintaining high sensitivity and specificity.

Evidence of Selection against Complex Mitotic-Origin Aneuploidy during Preimplantation Development.

Whole-chromosome imbalances affect over half of early human embryos and are the leading cause of pregnancy loss. While these errors frequently arise in oocyte meiosis, many such whole-chromosome abnormalities affecting cleavage-stage embryos are the result of chromosome missegregation occurring during the initial mitotic cell divisions. The first wave of zygotic genome activation at the 4-8 cell stage results in the arrest of a large proportion of embryos, the vast majority of which contain whole-chromosome abnormalities. Thus, the full spectrum of meiotic and mitotic errors can only be detected by sampling after the initial cell divisions, but prior to this selective filter. Here, we apply 24-chromosome preimplantation genetic screening (PGS) to 28,052 single-cell day-3 blastomere biopsies and 18,387 multi-cell day-5 trophectoderm biopsies from 6,366 in vitro fertilization (IVF) cycles. We precisely characterize the rates and patterns of whole-chromosome abnormalities at each developmental stage and distinguish errors of meiotic and mitotic origin without embryo disaggregation, based on informative chromosomal signatures. We show that mitotic errors frequently involve multiple chromosome losses that are not biased toward maternal or paternal homologs. This outcome is characteristic of spindle abnormalities and chaotic cell division detected in previous studies. In contrast to meiotic errors, our data also show that mitotic errors are not significantly associated with maternal age. PGS patients referred due to previous IVF failure had elevated rates of mitotic error, while patients referred due to recurrent pregnancy loss had elevated rates of meiotic error, controlling for maternal age. These results support the conclusion that mitotic error is the predominant mechanism contributing to pregnancy losses occurring prior to blastocyst formation. This high-resolution view of the full spectrum of whole-chromosome abnormalities affecting early embryos provides insight into the cytogenetic mechanisms underlying their formation and the consequences for human fertility.

Whole genome prediction for preimplantation genetic diagnosis.

BACKGROUND: Preimplantation genetic diagnosis (PGD) enables profiling of embryos for genetic disorders prior to implantation. The majority of PGD testing is restricted in the scope of variants assayed or by the availability of extended family members. While recent advances in single cell sequencing show promise, they remain limited by bias in DNA amplification and the rapid turnaround time (<36 h) required for fresh embryo transfer. Here, we describe and validate a method for inferring the inherited whole genome sequence of an embryo for preimplantation genetic diagnosis (PGD). METHODS: We combine haplotype-resolved, parental genome sequencing with rapid embryo genotyping to predict the whole genome sequence of a day-5 human embryo in a couple at risk of transmitting alpha-thalassemia. RESULTS: Inheritance was predicted at approximately 3 million paternally and/or maternally heterozygous sites with greater than 99% accuracy. Furthermore, we successfully phase and predict the transmission of an HBA1/HBA2 deletion from each parent. CONCLUSIONS: Our results suggest that preimplantation whole genome prediction may facilitate the comprehensive diagnosis of diseases with a known genetic basis in embryos.

Common variants spanning PLK4 are associated with mitotic-origin aneuploidy in human embryos.

Aneuploidy, the inheritance of an atypical chromosome complement, is common in early human development and is the primary cause of pregnancy loss. By screening day-3 embryos during in vitro fertilization cycles, we identified an association between aneuploidy of putative mitotic origin and linked genetic variants on chromosome 4 of maternal genomes. This associated region contains a candidate gene, Polo-like kinase 4 (PLK4), that plays a well-characterized role in centriole duplication and has the ability to alter mitotic fidelity upon minor dysregulation. Mothers with the high-risk genotypes contributed fewer embryos for testing at day 5, suggesting that their embryos are less likely to survive to blastocyst formation. The associated region coincides with a signature of a selective sweep in ancient humans, suggesting that the causal variant was either the target of selection or hitchhiked to substantial frequency.

Expanding the scope of noninvasive prenatal testing: detection of fetal microdeletion syndromes.

OBJECTIVE: The purpose of this study was to estimate the performance of a single-nucleotide polymorphism (SNP)-based noninvasive prenatal test for 5 microdeletion syndromes. STUDY DESIGN: Four hundred sixty-nine samples (358 plasma samples from pregnant women, 111 artificial plasma mixtures) were amplified with the use of a massively multiplexed polymerase chain reaction, sequenced, and analyzed with the use of the Next-generation Aneuploidy Test Using SNPs algorithm for the presence or absence of deletions of 22q11.2, 1p36, distal 5p, and the Prader-Willi/Angelman region. RESULTS: Detection rates were 97.8% for a 22q11.2 deletion (45/46) and 100% for Prader-Willi (15/15), Angelman (21/21), 1p36 deletion (1/1), and cri-du-chat syndromes (24/24). False-positive rates were 0.76% for 22q11.2 deletion syndrome (3/397) and 0.24% for cri-du-chat syndrome (1/419). No false positives occurred for Prader-Willi (0/428), Angelman (0/442), or 1p36 deletion syndromes (0/422). CONCLUSION: SNP-based noninvasive prenatal microdeletion screening is highly accurate. Because clinically relevant microdeletions and duplications occur in >1% of pregnancies, regardless of maternal age, noninvasive screening for the general pregnant population should be considered.

Clinical experience and follow-up with large scale single-nucleotide polymorphism-based noninvasive prenatal aneuploidy testing.

OBJECTIVE: We sought to report on laboratory and clinical experience following 6 months of clinical implementation of a single-nucleotide polymorphism-based noninvasive prenatal aneuploidy test in high- and low-risk women. STUDY DESIGN: All samples received from March through September 2013 and drawn ≥9 weeks' gestation were included. Samples that passed quality control were analyzed for trisomy 21, trisomy 18, trisomy 13, and monosomy X. Results were reported as high or low risk for fetal aneuploidy for each interrogated chromosome. Relationships between fetal fraction and gestational age and maternal weight were analyzed. Follow-up on outcome was sought for a subset of high-risk cases. False-negative results were reported voluntarily by providers. Positive predictive value (PPV) was calculated from cases with an available prenatal or postnatal karyotype or clinical evaluation at birth. RESULTS: Samples were received from 31,030 patients, 30,705 met study criteria, and 28,739 passed quality-control metrics and received a report detailing aneuploidy risk. Fetal fraction correlated positively with gestational age, and negatively with maternal weight. In all, 507 patients received a high-risk result for any of the 4 tested conditions (324 trisomy 21, 82 trisomy 18, 41 trisomy 13, 61 monosomy X; including 1 double aneuploidy case). Within the 17,885 cases included in follow-up analysis, 356 were high risk, and outcome information revealed 184 (51.7%) true positives, 38 (10.7%) false positives, 19 (5.3%) with ultrasound findings suggestive of aneuploidy, 36 (10.1%) spontaneous abortions without karyotype confirmation, 22 (6.2%) terminations without karyotype confirmation, and 57 (16.0%) lost to follow-up. This yielded an 82.9% PPV for all aneuploidies, and a 90.9% PPV for trisomy 21. The overall PPV for women aged ≥35 years was similar to the PPV for women aged <35 years. Two patients were reported as false negatives. CONCLUSION: The data from this large-scale report on clinical application of a commercially available noninvasive prenatal test suggest that the clinical performance of this single-nucleotide polymorphism-based noninvasive prenatal test in a mixed high- and low-risk population is consistent with performance in validation studies.

Single-nucleotide polymorphism-based noninvasive prenatal screening in a high-risk and low-risk cohort.

OBJECTIVE: To estimate performance of a single-nucleotide polymorphism-based noninvasive prenatal screen for fetal aneuploidy in high-risk and low-risk populations on single venopuncture. METHODS: One thousand sixty-four maternal blood samples from 7 weeks of gestation and beyond were included; 1,051 were within specifications and 518 (49.3%) were low risk. Cell-free DNA was amplified, sequenced, and analyzed using the Next-generation Aneuploidy Test Using SNPs algorithm. Samples were called as trisomies 21, 18, 13, or monosomy X, or euploid, and male or female. RESULTS: Nine hundred sixty-six samples (91.9%) successfully generated a cell-free DNA result. Among these, sensitivity was 100% for trisomy 21 (58/58, confidence interval [CI] 93.8-100%), trisomy 13 (12/12, CI 73.5-100%), and fetal sex (358/358 female, CI 99.0-100%; 418/418 male, CI 99.1-100%), 96.0% for trisomy 18 (24/25, CI 79.7-99.9%), and 90% for monosomy X (9/10, CI 55.5-99.8%). Specificity for trisomies 21 and 13 was 100% (905/905, CI 99.6-100%; and 953/953, CI 99.6-100%, respectively) and for trisomy 18 and monosomy X was 99.9% (938/939, CI 99.4-100%; and 953/954, CI 99.4-100%, respectively). However, 16% (20/125) of aneuploid samples did not return a result; 50% (10/20) had a fetal fraction below the 1.5th percentile of euploid pregnancies. Aneuploidy rate was significantly higher in these samples (P<.001, odds ratio 9.2, CI 4.4-19.0). Sensitivity and specificity did not differ in low-risk and high-risk populations. CONCLUSIONS: This noninvasive prenatal screen performed with high sensitivity and specificity in high-risk and low-risk cohorts. Aneuploid samples were significantly more likely to not return a result; the number of aneuploidy samples was especially increased among samples with low fetal fraction. This underscores the importance of redraws or, in rare cases, invasive procedures based on low fetal fraction. LEVEL OF EVIDENCE: II.

SNP-based non-invasive prenatal testing detects sex chromosome aneuploidies with high accuracy.

OBJECTIVE: This study aimed to develop a single-nucleotide polymorphism-based and informatics-based non-invasive prenatal test that detects sex chromosome aneuploidies early in pregnancy. METHODS: Sixteen aneuploid samples, including thirteen 45,X, two 47,XXY, and one 47,XYY, along with 185 euploid controls, were analyzed. Cell-free DNA was isolated from maternal plasma, amplified in a single multiplex polymerase chain reaction assay that targeted 19,488 polymorphic loci covering chromosomes 13, 18, 21, X, and Y, and sequenced. Sequencing results were analyzed using a Bayesian-based maximum likelihood statistical method to determine copy number of interrogated chromosomes, calculating sample-specific accuracies. RESULTS: Of the samples that passed a stringent quality control metric (93%), the algorithm correctly identified copy number at all five chromosomes in all but one of the 187 samples, for 934/935 correct calls as early as 9.4 weeks of gestation. We detected 45,X with 91.7% sensitivity (CI: 61.5-99.8%) and 100% specificity (CI: 97.9-100%), and 47,XXY and 47,XYY. The average calculated accuracy was 99.78%. CONCLUSION: This method non-invasively detected 45,X, 47,XXY, and 47,XYY fetuses from cell-free DNA isolated from maternal plasma with high calculated accuracies and thus offers a non-invasive method with the potential to function as a routine screen allowing for early prenatal detection of rarely diagnosed yet commonly occurring sex aneuploidies.

Noninvasive prenatal aneuploidy testing of chromosomes 13, 18, 21, X, and Y, using targeted sequencing of polymorphic loci.

OBJECTIVE: This study aims to develop a noninvasive prenatal test on the basis of the analysis of cell-free DNA in maternal blood to detect fetal aneuploidy at chromosomes 13, 18, 21, X, and Y. METHODS: A total of 166 samples from pregnant women, including 11 trisomy 21, three trisomy 18, two trisomy 13, two 45,X, and two 47,XXY samples, were analyzed using an informatics-based method. Cell-free DNA from maternal blood was isolated, amplified using a multiplex polymerase chain reaction (PCR) assay targeting 11,000 single nucleotide polymorphisms on chromosomes 13, 18, 21, X, and Y in a single reaction, and sequenced. A Bayesian-based maximum likelihood statistical method was applied to determine the chromosomal count of the five chromosomes interrogated in each sample, along with a sample-specific calculated accuracy for each test result. RESULTS: The algorithm correctly reported the chromosome copy number at all five chromosomes in 145 samples that passed a DNA quality test, for a total of 725/725 correct calls. The average calculated accuracy for these samples was 99.92%. Twenty-one samples did not pass the DNA quality test. CONCLUSIONS: This informatics-based method noninvasively detected fetuses with trisomy 13, 18, and 21, 45,X, and 47,XXY with high sample-specific calculated accuracies for each individual chromosome and across all five chromosomes.

Origins and rates of aneuploidy in human blastomeres.

OBJECTIVE: To characterize chromosomal error types and parental origin of aneuploidy in cleavage-stage embryos using an informatics-based technique that enables the elucidation of aneuploidy-causing mechanisms. DESIGN: Analysis of blastomeres biopsied from cleavage-stage embryos for preimplantation genetic screening during IVF. SETTING: Laboratory. PATIENT(S): Couples undergoing IVF treatment. INTERVENTION(S): Two hundred seventy-four blastomeres were subjected to array-based genotyping and informatics-based techniques to characterize chromosomal error types and parental origin of aneuploidy across all 24 chromosomes. MAIN OUTCOME MEASURE(S): Chromosomal error types (monosomy vs. trisomy; mitotic vs. meiotic) and parental origin (maternal vs. paternal). RESULT(S): The rate of maternal meiotic trisomy rose significantly with age, whereas other types of trisomy showed no correlation with age. Trisomies were mostly maternal in origin, whereas paternal and maternal monosomies were roughly equal in frequency. No examples of paternal meiotic trisomy were observed. Segmental error rates were found to be independent of maternal age. CONCLUSION(S): All types of aneuploidy that rose with increasing maternal age can be attributed to disjunction errors during meiosis of the oocyte. Chromosome gains were predominantly maternal in origin and occurred during meiosis, whereas chromosome losses were not biased in terms of parental origin of the chromosome. The ability to determine the parental origin for each chromosome, as well as being able to detect whether multiple homologs from a single parent were present, allowed greater insights into the origin of aneuploidy.

Accurate prediction of HIV-1 drug response from the reverse transcriptase and protease amino acid sequences using sparse models created by convex optimization.

MOTIVATION: Genotype-phenotype modeling problems are often overcomplete, or ill-posed, since the number of potential predictors-genes, proteins, mutations and their interactions-is large relative to the number of measured outcomes. Such datasets can still be used to train sparse parameter models that generalize accurately, by exerting a principle similar to Occam's Razor: When many possible theories can explain the observations, the most simple is most likely to be correct. We apply this philosophy to modeling the drug response of Type-1 Human Immunodeficiency Virus (HIV-1). Owing to the decreasing expense of genetic sequencing relative to in vitro phenotype testing, a statistical model that reliably predicts viral drug response from genetic data is an important tool in the selection of antiretroviral therapy (ART). The optimization techniques described will have application to many genotype-phenotype modeling problems for the purpose of enhancing clinical decisions. RESULTS: We describe two regression techniques for predicting viral phenotype in response to ART from genetic sequence data. Both techniques employ convex optimization for the continuous subset selection of a sparse set of model parameters. The first technique, the least absolute shrinkage and selection operator, uses the l(1) norm loss function to create a sparse linear model; the second, the support vector machine with radial basis kernel functions, uses the epsilon-insensitive loss function to create a sparse non-linear model. The techniques are applied to predict the response of the HIV-1 virus to 10 reverse transcriptase inhibitor and 7 protease inhibitor drugs. The genetic data are derived from the HIV coding sequences for the reverse transcriptase and protease enzymes. When tested by cross-validation with actual laboratory measurements, these models predict drug response phenotype more accurately than models previously discussed in the literature, and other canonical techniques described here. Key features of the methods that enable this performance are the tendency to generate simple models where many of the parameters are zero, and the convexity of the cost function, which assures that we can find model parameters to globally minimize the cost function for a particular training dataset. AVAILABILITY: Results, tables and figures are available at ftp://ftp.genesecurity.net. SUPPLEMENTARY INFORMATION: An Appendix to accompany this article is available at Bioinformatics online.

Use of the l1 norm for selection of sparse parameter sets that accurately predict drug response phenotype from viral genetic sequences.

We describe the use of the l1 norm for selection of a sparse set of model parameters that are used in the prediction of viral drug response, based on genetic sequence data of the Human Immunodeficiency Virus (HIV) reverse-transcriptase enzyme. We discuss the use of the l1 norm in the Least Absolute Selection and Shrinkage Operator (LASSO) regression model and the Support Vector Machine model. When tested by cross-validation with laboratory measurements, these models predict viral phenotype, or resistance, in response to Reverse-Transcriptase Inhibitors (RTIs) more accurately than other known models. The l1 norm is the most selective convex function, which sets a large proportion of the parameters to zero and also assures that a single optimal solution will be found, given a particular model formulation and training data set. A statistical model that reliably predicts viral drug response is an important tool in the selection of Anti-Retroviral Therapy. These techniques have general application to modeling phenotype from complex genetic data.