Clinical methodology

This is a prospective multicenter observational study of the application of a laboratory developed test. Women at 31 clinical sites judged by their attending physicians to be at high risk for fetal aneuploidy, who met the inclusion criteria of the study protocol and who had made the decision to pursue CVS or AC were offered enrollment into the study. Blood samples (20-30 mL drawn into 2 or 3 ethylenediaminetetraacetic acid purple/lavender–top tubes) were drawn before the procedure in all instances.

The study participant was considered enrolled and assigned a bar-coded participant identification number after the informed consent was signed but before the sample was collected. The samples were immediately placed on wet ice and transferred to a standard refrigerator at the collection site where they were stored at a temperature between 0°C and 8°C. Within 6 hours all samples were transferred in a chill pack (at approximately 4°C) to a locally contracted laboratory for further processing to plasma. Samples were maintained at a central storage facility at –70°C or colder until transferred to the study sponsor for testing by MPS.

The results of the samples analyzed by the study sponsor were stored separately from the original karyotype results and both secured in password-protected databases. Karyotypes from the AC or CVS specimens were completed at independent commercial laboratories and the results reported to the patients and their providers.

Inclusion criteria were singleton pregnancy in a patient 18 years of age or older who had provided written informed consent and who had made the decision to pursue invasive prenatal diagnosis by CVS or AC. The participant was further judged to be at increased risk for fetal aneuploidy for 1 or more of the following conditions: maternal age 35 years or older at the estimated date of delivery, screen positive on first- or second-trimester serum biochemical screening tests, the presence of a fetal abnormality on ultrasound, or a personal or family history of a chromosomal abnormality.

Exclusion criteria were the inability to give written informed consent, multiple gestation, or fetal demise of an additional embryo during the current pregnancy at 8 weeks or farther in gestation. Gestational age was determined by a reliable menstrual history unless it differed from composite ultrasound measurement dating by more than 7 days, in which case the latter would be used to assign a gestational age.

Maternal demographic data and karyotype results were entered into password-protected databases and maintained by the Obstetrix Medical Group. All databases including study sponsor MPS test results, maternal demographic data, and karyotype results were stored separately until test performance was assessed by an independent biostatistician.

An oversight committee composed of 3 knowledgeable individuals uninvolved in any aspect of the study reviewed study subject enrollment and the prevalence rate of trisomy 21 at predetermined intervals to recommend the final sample size with the goal of identifying sufficient numbers of trisomy 21 fetuses (approximately 80-160); otherwise, the oversight committee was blinded to all study participant test results. A case prevalence rate of up to 4% was anticipated at the outset based on historical data at some of the participating sites. It was estimated that this would translate to a sample size of approximately 4000 patients.

This study was approved by the Western Institutional Review Board (WIRB protocol 20090261) or local institutional review boards. It was deemed a minimal-risk protocol because simple venipuncture is considered to be noninvasive, and the volume of blood drawn was 30 mL or less (CFR Part 812.3[k]). This study was registered with clinicaltrials.gov (identifier NCT00847990 ).

Laboratory methodology

All samples were collected and processed under the same protocol. A 10 mL aliquot of maternal whole blood was drawn into an ethylenediaminetetraacetic acid-K2 spray-dried Vacutainer (Becton Dickinson, Franklin Lakes, NJ), stored, and transported to the processing laboratory on wet ice. Within 6 hours of the blood draw, the maternal whole blood was centrifuged (Eppendorf 5810R plus swing-out rotor; Eppendorf AG, Hamburg, Germany) chilled (4°C) at 2500 × g for 10 minutes and the plasma was collected. The plasma was centrifuged a second time (Eppendorf 5810R plus fixed-angle rotor) at 4°C at 15,000 × g for 10 minutes. After the second spin, the plasma was removed from the pellet and distributed into 4 mL plasma bar-coded aliquots and immediately stored frozen at –70°C until the DNA extraction.

The MPS for aneuploidy detection was completed following a previously published method. The fetal fraction of the cfDNA was determined using a method relying on differentially methylated markers. DNA extraction and library preparation also followed a previously published method, and clustering and sequencing were performed using the HiSeq 2000 sequencers (Illumina Inc, San Diego, CA). Thirty-six cycles of single-read multiplexed sequencing (libraries pertaining to 12 samples including controls in each lane of a flow cell) were performed and image analysis and base calling were performed with the manufacturer-provided software (Illumina Inc). Sequences were aligned to the UCSC hg19 human reference genome using Bowtie version 2.

Data analysis was completed using sequence reads unique to the chromosomes of interest (21, 18, 13, X, Y) and then standardizing the fractional representation of each chromosome by comparison with a known euploid control group (z-scores). The standardization of chromosome representation was carried out by using the median chromosome representation as computed over the samples from a given flow cell and using a previously established estimate of variability of this representation.

This calculation does not rely on a group of control euploid samples; this is carried out for each flow cell and accounts, implicitly, for flow cell to flow cell variability (this is the same approach taken in clinical practice). Z-scores at or above 3 were considered indicative of trisomy 21 (z-scores of at or above 3.95 were considered indicative of trisomies 18 and 13). The normalization procedure for the sex chromosome aneuploidies (SCA) as well as the classification algorithm was completed in accordance with a previously published report.

All samples were required to meet quality control criteria. These included a minimum fetal fraction of 4.0% and a maximum fetal fraction of 50%; minimum fetal DNA per sample of 26 copies; minimum library concentration of 7.5 nmol; minimum number of autosomal-aligned reads of 9 million. Sequencing results that exhibited strong guanine-cytosine (GC) bias (as estimated from the shape of the counts per 50 kb bin vs GC content of each 50 kb bin) were rejected and the affected samples were reprocessed.

Discordant results resolution

A discordant result resolution plan was developed before the initiation of the sample testing by MPS. This process included investigation for transcription/clerical errors in the case report forms. Twenty-three database entry errors were identified and corrected by the principal investigator. These included 17 errors in fetal sex, 3 errors in complex karyotypes, and 3 errors in aneuploidy from terminations of pregnancy. All had formal karyotype reports available to assure correct entries. Also included was a review of sample collection procedures, an analysis of secondary aliquots, maternal DNA assessment (in buffy coat), and clinical patient follow-up after the birth.

A set of 52 single-nucleotide polymorphisms was used to identify potential chain of custody events outside the collection sites or the analytical laboratory, which would be considered part of the standard workflow. Affected samples outside this standard workflow were removed from the analysis. There were 3 affected samples whose analyses demonstrated that sample identification was incorrect and they qualified for removal from the analysis. Two samples had a genetic mismatch between the buffy coat and the sequencing library; the third sample had a sequencing library matching a library from a different patient. Karyotypes and sequencing results were concordant once the correct patient samples were identified, but because the chain of custody was compromised, they were excluded from further analysis.

Statistical analysis

The statistics were calculated treating the invasive procedure results as the gold standard. The exact 95% Clopper-Pearson confidence limits are shown alongside the binomial outcomes, which include sensitivity and specificity as well as positive and negative predictive values. Asymptotic 95% confidence limits, computed under a log transform, were also calculated. All statistics were generated using the frequency procedure, SAS/STAT software, version 9.3 of the SAS system (SAS Institute, Cary, NC). All statistical analyses were done by an independent statistician.