Noninvasive single‐cell‐based prenatal genetic testing: A proof of concept clinical study

To clinically assess a cell‐based noninvasive prenatal genetic test using sequence‐based copy number analysis of single trophoblasts from maternal blood.


| INTRODUCTION
Currently, the most widely used laboratory procedures for prenatal genetic testing are two diagnostic tests, amniocentesis and chorionic villus sampling (CVS), and a single screening test, cell-free noninvasive prenatal testing (cfNIPT).Tests performed on CVS and amniocentesis specimens are considered diagnostic but are invasive and carry a small risk of pregnancy loss. 1 In contrast, cfNIPT is a risk-free screening test, but its ability to detect chromosomal deletions and duplications is inferior to what is achievable with CVS and amniocentesis.The cfNIPT also produces varying frequencies of false positive results, depending on the nature of the abnormality detected and the genomic region involved.The availability of a prenatal genetic test that is risk-free and provides high resolution detection of pathogenic deletions and duplications would be a superior option for patients and clinicians alike.
The potential to use circulating fetal cells in maternal blood to derive noninvasive prenatal genetic results has been hypothesized since at least 1969, 2 but various technological challenges, primarily the extreme rarity of fetal cells in the maternal circulation, have hindered the introduction of such a test into the clinic.Although there is evidence that fetal nucleated red blood cells (fnRBC), fetal lymphocytes, and fetal hematopoietic stem progenitor cells (HSPC) can be recovered from maternal blood, none of these have achieved full clinical utility.Kølvraa and colleagues presented evidence in 2014 that extravillous trophoblasts might be a suitable target for cellbased noninvasive prenatal testing. 3,4Using CD105 and CD141 as surface markers and cytokeratin as a cytoplasmic marker for these trophoblasts, they detected fetal cells in 91% of 78 samples, which was confirmed by XY fluorescence in situ hybridization (FISH).
Another key development came in 2011 when it was demonstrated that DNA from single human tumor cells could be used for nextgeneration sequencing (NGS) that would be clinically impactful. 5rthermore, it was shown in 2012 that clinically significant deletions and duplications could be detected in single human lymphoblasts at a resolution of 1 Mb using array comparative genomic hybridization (CGH). 6In 2016, two collaborative publications reported further progress in using isolated circulating trophoblasts to detect fetal aneuploidy and deletions. 7,8This proof of feasibility for prenatal diagnosis was further confirmed using NanoVelcro microchips. 9Results from genomic copy number analysis of trophoblasts from maternal blood compared favorably with those from CVS or amniocentesis. 10Most of these reports focused on detection of aneuploidies, deletions, and duplications, but the ability to detect monogenic disorders has also been reported. 11,12Most recently, two groups have reported continuing success in performing cell-based NIPT. 13,14[17] We report a series of clinical studies designed to demonstrate the feasibility of a clinical high-resolution noninvasive single-cell-based prenatal test using whole genome NGS for the detection of aneuploidy and deletion/duplication.

| Study design
Three studies (Table 1) were carried out for the evaluation of the cell-based noninvasive Luna Prenatal Test (hereafter referred to as the Luna test).These included a no-follow-up (NFU) series obtained from routine pregnancies, a follow-up (FU) series scheduled to have CVS or amniocentesis, and a spike-in series of cells from seven lymphoblast or fibroblast cell lines with known aneuploidy or deletion/duplication (del/dup).The NFU series was designed to assess the success rate for recovering trophoblast cells and for achieving NGS analyses that could be evaluated for either aneuploidy only, or for aneuploidy plus del/dup at a resolution of 1.5 Mb for deletions and 2 Mb for duplications.Blood was collected between Jul. 2021 and Jan. -305 sample for shipping.Each sample was processed with the methods that were later used for the launch of the Luna clinical test.About a third of the way through the study, an initiative to collect a second blood sample was implemented for samples where 0 or 1 cells were recovered in the first sample.An overview of sample characteristics in the NFU series is provided in Table 2.
The FU series was designed to determine the performance of the Luna test compared with CVS or amniocentesis results in the same cases.A total of 243 blood samples (7 unusable) were received from three recruitment sites: Columbia University Irving Medical Center (Dr.Ronald Wapner), Icahn School of Medicine at Mount Sinai (Dr. Joanne Stone), and Yale School of Medicine (Dr.Audrey Merriam).
Samples were collected between Aug. 2021 and Sep.2022, and an overview of sample characteristics is shown in Table 3.Details of cells recovered are provided in Appendix 1.Any woman-planning to undergo CVS or amniocentesis at these sites was eligible for participation, and the indications for performing testing were heterogeneous.Many had increased risk including cytogenetic risk, monogenic risk, advanced maternal age, or abnormal ultrasound.In all cases, blood was drawn prior to the invasive procedure.The results from CVS or amniocentesis served as accepted standard of care data for comparison with the Luna test results.There was no attempt for rapid results, and no Luna results were returned to patients.This comparison would allow us to make a preliminary estimate of the accuracy, specificity, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) of the Luna test relative to CVS/ amniocentesis (CVS/amnio).
For the spike in samples, tissue culture cells were labeled with tracking dye and added to control whole blood.These samples were then processed through the steps of the protocol, including fixation, lysis, permeabilization, and staining, and then picked individually based on staining with tracker dye without an enrichment step.
Protocols for amplification and DNA sequencing were identical to those for samples from pregnant women.

| Protocol
Four 10 mL tubes of maternal blood (Cell-Free DNA BCT®, Streck) were carefully collected to maximize free flow with immediate inversion to avoid any clotting; blood was shipped overnight at ambient temperature on the day it was collected to Luna Genetics in Houston, TX.Blood was processed with a mild fixation of 2% paraformaldehyde for 10 min followed by a 5-fold-volume increase resulting in a 0.08% Triton X-100 concentration for 4 min as a lysis procedure to eliminate red blood cells (RBC). 7,8Cells were permeabilized with a final concentration of 1X BD Perm/Wash Buffer 554723 (BD Biosciences) and stained with two antibodies against T A B L E 1 Three groups of samples for testing.

Sample number Characteristics
No  S1), and presumptive fetal trophoblasts were manually picked into microfuge tubes using a micromanipulator within the CytePicker.Occasionally, cells were picked as trophoblast doublets, on the assumption that they were previously adjacent cells in the placenta.Individual cells or doublets were subjected to whole genome amplification using the PicoPLEX® Single Cell WGA Kit (Takara), and a DNA library suitable for NGS was prepared using the Nextera XT DNA Library Preparation Kit (Illumina).DNA was sequenced (single-end, read length of 100 bp) to obtain 1-5 million reads on an Illumina MiniSeq or NextSeq 2000, and the read data was analyzed using NxClinical (NxC) software (BioDiscovery).Up to five cells were analyzed for each case of singleton fetus (additional cells were analyzed in cases of twin pregnancy or suspected mosaicism, when possible).All single-cell copy number profiles were plotted and analyzed against both male and female reference data sets, with the male reference used in all figures.Reference set controls were prepared from a pool of 10-20 male or female trophoblasts from lowrisk pregnancies where no copy number abnormalities were evident.All cells were examined using both log ratio and copy number settings in NxC, with visual analysis performed on one chromosome at a time, noting any major deflections from normal copy number.For deletion/duplication analysis, the highest quality cell was selected, and each call made by the software was manually examined for assignment as a real copy number variant (CNV) versus technical artifact.Subsequently, each additional cell was reviewed in the same manner, checking for concordance against the highest quality cell and excluding calls that were deemed technical artifact.
Preliminary categorical assignment of each CNV call was made by a genomics specialist (see 2.4 Scoring Quality of NGS data).Copy number gains or losses meeting quality criteria were flagged for final review and interpretation by an ABMGG-certified lab director.

| Fetal cell genotyping
Maternal genomic DNA was extracted concurrently from the blood sample and analyzed using a custom Illumina Global Screening Array v2 SNP array at a CLIA-accredited reference lab.The full SNP data for the mother was compared with limited SNP data for each cell from single-cell NGS reads.Each cell was documented by genotyping to be of fetal and not maternal origin based on hundreds to thousands of SNP alleles present in the cell but not in the mother, whereas maternal cells had negligible such alleles.Egg donation and surrogate maternal variables were considered.

| Scoring Quality of NGS data
Every cell subjected to NGS analysis was given one of three scores for deletion/duplication (del/dup) and/or aneuploidy as below by a genomics specialist: 1. Scorable for aneuploidy and 1.

| RESULTS
Blood was obtained after overnight shipment.Among the 243 pregnancies scheduled for CVS or amniocentesis (FU series), Luna results were normal for 160 singleton cases.Additionally, among 158 low-risk patients not undergoing invasive testing (NFU series), normal results were obtained in 134 cases.Finally, single cells from seven chromosomally abnormal cell lines were processed through the same workflow to generate spike-in data.Figure 1A shows four normal trophoblast cells from one singleton pregnancy and Figure 1B shows single abnormal trophoblast cells, each from a different singleton pregnancy, demonstrating the robust detection of the three most common forms of autosomal trisomy (T13, T18, and T21).Two biological processes, apoptosis and S phase of the cell cycle, are found to interfere with cell analysis at a variable frequency (5%-10% for each) in circulating trophoblasts.Figure S2A demonstrates a case with two normal cells and one apoptotic cell.Figure S2B shows cultured normal lymphoblasts (GM12878) in the G1,S, and G2/M phases of the cell cycle analyzed with FACS to document DAPI intensity.Two cells from each phase were analyzed by NGS. Figure S2C demonstrates two cells from a case with trisomy 16; the trisomy is detected in both cells, although one of the cells is in the S phase.

| Analysis of 158 no-follow-up (NFU) samples from 35 states
For the NFU series, there were 158 first draw samples and 26 redraw samples from 158 pregnancies.The overall information for NGS scorable cells is shown in Figure 2 indicating the number and quality of trophoblasts for each case.Focusing on the combined data from the first draw and redraws, one or more scorable cells were obtained for 91% of the samples so that a report with results could be issued Examples of normal and abnormal cells analyzed with the single cell prenatal genetic test.Panel (A) includes four cells from a singleton pregnancy with a normal male fetus compared with a male control using a log ratio plot.A genomic copy number plot for all chromosomes is shown from chromosome 1 on the left to X and Y on the right using the NxClinical software.No gains or losses of copy number above the reporting cutoff are observed in any cells.T01 to T06 identify specific trophoblasts from this case.Panel (B) shows plots from three different cases for the three most common forms of trisomy as indicated.All are plotted against a male control; the T13 and T21 fetuses are male and the T18 female.Each plot shows a clear gain in copy number for the relevant chromosome.Aneuploidy abnormalities in the 158 NFU series included two with T21, three with 45,X (one mosaic), and two samples with mosaic trisomy 20 (Table 5).Subchromosomal findings (Table 5) included one each with deletion of 3p26.3-p26.1,deletion of 15q11.2-q13.1 (Prader-Willi/Angelman region), the duplication of 16p13.11,and duplication of 18p11.32.Three of the subchromosomal events were interpreted as pathogenic or likely pathogenic, and one (dup18) was considered likely benign.Because of the concordance of multiple cells within each abnormal sample (Figures 3 and S3), we believe that these findings are real and not artifacts, but they could be subject to mosaicism and not present in CVS/amnio had they been performed.
Follow-up contact was not available under the IRB provisions, and thus results could not be tested further.9][20] The 1.9 Mb duplication of 16p13.112][23] The 6.3 Mb deletion of 15q11.2-q13was interpreted as pathogenic and consistent with a pathogenic class 1 deletion (BP1-BP3) causing either Prader-Willi or Angelman syndrome, 24,25 which could not be distinguished according to the parent of origin based on the information available.

| Concordance between cells in a case
Of the 158 singleton NFU pregnancy cases, 118 cases (75%) had two or more reportable cells.This count excludes three mosaic cases (two T20 and one 45, X), which by definition have cells of differing results.
All cells (total 404 cells) within each of those 118 cases showed concordant results for aneuploidy.There were 74 cases with two or more cells of high quality (score 1).All 210 of these score 1 cells agreed as to the presence or absence of any pathogenic changes. -30 When scoring for clinically significant copy number abnormalities, the concordance between cells within the same sample was 100% (excluding the three samples with placental mosaicism).

| Analysis of 243 follow-up (FU) samples with CVS and amniocentesis comparison
The results for the 243 cases in the FU series are summarized in  which would not have detected a deletion of this size.No scorable trophoblasts were recovered in 39 of 236 usable samples.
There were 10 cases with differences between the Luna test and CVS/amnio data (Table 7), all of which were compatible with placental mosaicism.We interpret all results from all procedures to be real and accurate with findings compatible with placental mosaicism.Placental mosaicism occurred in 7 of 236 Luna cases (3.0%) and in 3 of 188 (1.6%) CVS cases (total 4.6%).In one noteworthy case, the Luna test detected T13 in four out of 4 trophoblast cells, but a karyotype on cultured cells from amniocentesis was normal (Table 7).This case documents why the Luna test is not diagnostic for aneuploidy (see Discussion).Interestingly, no case showed placental mosaicism in both the Luna test and CVS, suggesting that neither the current Luna test nor CVS alone detect all placental mosaicism for aneuploidy.
There were six cases (Table 8) with a Luna test result but no CVS/ amnio result.These included three trisomy cases (one T13, one T21, and one T22) in which there was a fetal demise before a scheduled CVS/amnio.There were two Luna cases (one T18 and one normal Luna result but abnormal ultrasound heart defect) where pregnancies were terminated based on existing pathological information without CVS/ amnio.There was one normal Luna case with a failed CVS.
Details for 15 twin cases are provided in Table S1.Although more extensive genotyping could clarify monozygous from dizygous twins and distinguish the same sex dizygous twins, this study was not set up to make these distinctions.Based on CVS/amnio data, there were six cases of opposite sex twins, and the Luna test found cells of both sexes in four cases but identified only male cells in one case and only female cells in another case.For cases of the same sex twins (four female and two male pairs), the Luna test was not set up to distinguish the two twins, and it is likely often the case that cells were recovered from only one twin.One of these cases is the twins with Williams syndrome deletion detailed above.In one case of male/male twins, the Luna test could not distinguish one twin from the other, but CVS/amnio showed mosaic T2 in one twin but no T2 in the other twin and no T2 in the Luna cells.There were no unexpected sex differences, such as male cells identified by Luna in confirmed female/ female twin pairs or vice versa.

| Conclusions for CVS/amnio comparison
The Luna lab was blinded to all results of CVS/amnio until it submitted reports to a joint database.In general, the agreement between the Luna test and CVS/amnio for singleton pregnancies was 100%, with the caveat of 10 cases involving placental mosaicism that led to differing results.In addition, there were two cases reported as normal by Luna, which were reported as having a balanced translocation by CVS/amnio, and these were counted as agreement.There was one case where Luna detected a 17p12 deletion, but only karyotype was performed on the CVS, and this was not counted as a disagreement.There were 39 cases in which no cells were recovered (Table 6) using the Luna test, but data was available for CVS/amnio.
Of the 39, seven did not meet sample acceptance criteria and were rejected, 22 failed to recover cells, and in 10 cases cells were recovered and subjected to NGS, but all cells were unscorable.For the 22 patients with failure to recover cells, we reviewed whether these were associated with any pathological state.There were 12 normal male and 8 normal female results, two male T21, one mosaic 45, X, and one unbalanced translocation.The numbers were too small to be statistically significant, but there was no strong evidence for concern.Of the 243 samples listed in Table 6, seven samples were rejected and 32 recovered no scorable cells.For the 204 cases where Luna results were informative, 123 (60%) had at least one cell scorable for del/dup, whereas 81 (40%) could be scored for aneuploidy only and not for deletion/duplication.This was noticeably different from the 15.8% for the NFU series.This difference may have been related in part to the slightly higher percentage of samples beyond 14 weeks of gestation and the percentage of samples with lower blood volumes.

| Calculation of sensitivity and specificity for luna test versus CVS/amnio
For the 243 samples received (236 usable), 160 agreed for normal, 14 agreed for trisomy, and 1 agreed for twins with Williams syndrome.Although there were no disagreements for the twin cases, the same sex twins were not distinguished from each other by the Luna test.Thus, only twins where both sexes were identified in the Luna test and in the follow-up test were counted for these sensitivity and specificity estimates.If cases interpreted as placental mosaicism were excluded (Figure 5A), the preliminary estimates of accuracy, sensitivity, specificity, PPV, and NPV were all 100%, but the total number of cases was limited, and the confidence intervals were significantly wide.If one counts mosaic cases as false positives or  false negatives, the preliminary estimates for accuracy, sensitivity, specificity, PPV, and NPV are all less than 100%, as shown in Figure 5B.One Luna result had all four cells with T13 and a normal amnio result and would be a false positive by any measure.The consequences of the T13 in the placenta may have implications for the course and outcome of the pregnancy.All other Luna positives that were interpreted as placental mosaicism involved 45,X or rare autosomal trisomy (RAT), where the possibility of mosaicism is high and would likely be followed up by CVS/amnio.For the three cases where mosaicism was absent on the Luna test but present on direct CVS, these emphasize that the fewer the cells recovered by the Luna test, the higher the probability of missing placental mosaicism.Of course, the Luna test does not detect type 2 mesenchymal mosaicism.

| Clinical and analytical performance with spike-in cells
Given that it is not feasible to obtain samples from women carrying fetuses with a wide range of deletions and duplications, we used human cultured cells (lymphoblast or fibroblast obtained from biobanks at the Coriell Institute for Medical Research) with known aneuploidy or deletion or duplication as shown in Table 9.Out of a total of 148 scorable cells, known abnormalities were detected and called by the NxC software in every cell except one unscorable cell.
An example of the data for a 1.5-Mb deletion cell line is shown in Figure S4).Thus, the preliminary analytic sensitivity was near 100% with the expected finding being reported in every cell.This included inter-prep and intra-prep replicates.Additionally, reproducibility and repeatability studies were performed using the same sample set to demonstrate the precision of results across multiple replicates of the assay (inter-run precision).Three replicates of a single sample on a single run were performed to demonstrate the precision of test results under the same operating conditions (intra-run precision).The observed results from each of the runs performed were 100% concordant.In addition, we performed a Limit of Detection Analysis by spiking in Coriell cell lines harboring characterized copy number abnormalities, which confirmed that our clinical Luna test can detect copy number losses down to a deletion of 1.5 Mb or a duplication of 2.0 Mb (Table 9 and Figure S4).with increasing maternal weight, 26 while maternal weight appears to be minimal or absent as a modifying factor for single cell testing 27 ; this is consistent with our experience (data not shown).For cfNIPT, maternal autoimmune disease and therapy with low molecular weight heparin decrease fetal fraction, 26 and we speculate that these variables would not affect single-cell testing.
The two primary weaknesses of the Luna test are (1) failure to recover sufficient numbers of trophoblasts in some cases, and (2) the fact that many cells were scorable for aneuploidy only and not for del/dup.The failure to recover cells of sufficient quality can be addressed in part through a repeat blood draw at 14-15 days after the initial phlebotomy.In 26 cases with 0-1 cells recovered, a redraw was processed and resulted in additional scorable cells in 77% of cases.This will not solve all cases and further improvements are needed.There is no significant evidence that failure to recover cells is an indication of any abnormality with the pregnancy, although data are limited, and this may occur more frequently in normal pregnancies, as the number fetal cells are reported to be increased with some aneuploid fetuses. 28Even when an acceptable number of scorable cells is recovered, they may be few in number such as three or less, which may cause failure to detect aneuploidy mosaicism, especially for RAT, 45,X, and T13.The failure of the Luna test to detect rare autosomal trisomy was seen in three cases in Table 7, where only one cell was recovered with the Luna test.At present, a relatively large blood volume (40 mL) is being collected, but this remains a superior sample collection method when compared to the risks of CVS or amniocentesis.The inability to detect the smallest deletions and duplications such as single exon events remains weakness relative to CVS and amniocentesis, but this could be addressed by deeper sequencing of single cells.The current cost of the test limits its uptake in many healthcare settings.
From the data in the NFU series in Figure 2 and based on other publications, 7,13,14 we suggest that the optimal time to perform the Luna test is 9-14 weeks of gestation, though it is also feasible to perform the test as early as 8 weeks or as late as 22 weeks or later.If cells are obtained, the Luna test is reliable, but there is a slightly increased risk of failure to obtain scorable cells outside the 9-14week range.
The FU series demonstrates complete agreement when the Luna test and CVS or amniocentesis are performed on the same singleton pregnancy, apart from 10 cases of placental mosaicism (Table 6 and Figure 5).We accept that most experts would classify disagreements between the Luna test and amnio/CVS as false positives or false negatives for the Luna test.We prefer to think of these as true results for the placenta and true results for the fetus, with differences requiring combined interpretation.Many possible improvements can be envisioned for cell-based noninvasive prenatal testing.The potential to improve the recovery of trophoblasts and/or recovery of other fetal cell types from a given volume of blood is suggested by credible reports of 3-6 fetal cells per mL of maternal blood, 32 compared to the recovery of 0.1-0.2scorable cells per mL of maternal blood reported here and compared to 0.18/cells per mL reported by the Menarini lab 13 and 0.42 cells per mL (not corrected for scorability) reported by the ARCEDI lab. 8We observed a 9% failure rate of the NFU series and 14% in the FU series, while Menarini 13 reports a failure rate of 27%-34%, and ARCEDI 11,14  embryos generated in vitro. 34Based on this study, the occurrence of type 1 or 3 mosaicism in the placenta is not always detected by either the Luna test or CVS and is likely slightly higher than is often reported 35 ; here we observed 10 cases out of 188 potentially informative cases (5.3%).
There is some debate over whether it is desirable or undesirable to detect placental mosaicism for rare autosomal trisomy when present.Some laboratory study only cultured cells on CVS knowing that type 1 mosaicism may be missed.The data reported here suggests that the frequency of detection of RAT is higher than what is seen for type 1 on CVS.In such cases, we propose that the correct diagnosis (truth) is placental mosaicism, and all the observed data are true and accurate.More recent studies argue that detecting placental mosaicism for RAT is in the best interest for optimal management of a pregnancy. 36,37ll-based NIPT is a promising approach to genetic prenatal diagnosis, although the current inability to recover 5-10 cells in >95% of cases limit its widespread adoption.If sufficient cells could be recovered, it is feasible to perform high-coverage whole genome sequencing (WGS) on trophoblasts 38 raising the possibility of noninvasive trio WGS in the first trimester.
2022 from healthy pregnant volunteers who were distributed across 35 states in the USA.Participants were recruited through social media and virtually consented by a genetic counselor employed by Luna.A phlebotomist visited the home, drew blood, and delivered the BELLAIR ET AL.
were often cells in the S phase of the cell cycle where numerous small genomic segments not yet replicated cannot be distinguished from small deletions.Cells scorable for aneuploidy and 1.5 Mb del/ 2.0 Mb dup resolution had very high-quality NGS data and very few putative gains or losses called by the NxC software.The final interpretation for the case was based on the cumulative findings in all the scorable cells.Each case was signed out as normal (no pathogenic copy number abnormalities detected) or as having a specific copy number abnormality.The numbers of cells with a score of 1 (scorable for aneuploidy and del/dup at 1.5/2.0Mb)or a score of 2 (aneuploidy only resolution) is specified in the figures and tables below.For ethics approval, the samples from the Follow-Up (FU) series were collected under a protocol approved by the Columbia University Medical Center IRB protocol AAAT6189 and WCG ID 20193442.The No Follow-up (NFU) series was approved by the WCG IRB ID 20216940.

F I G U R E 2
Cell analysis for 158 cases from across the US with no follow-up CVS or amniocentesis.Samples were collected between Jul. 2021 and Jan. 2022.Each column is a case.Cells are designated by vertical bars as follows: Cells from the first blood draw are in green and cells from the redraw are in brown.Cells scorable for aneuploidy and del/dup are in dark colors and cells scorable for aneuploidy only are in light colors.Cases with abnormalities are indicated in black font.[Colour figure can be viewed at wileyonlinelibrary.com]T A B L E 4 Reportable cells from 158 cases in the no-follow-up (NFU) series.

F I G U R E 3
Detection of a Prader-Willi/Angelman deletion in all four available cells.The fetus is female, and data are plotted against a male control.The deleted region is in the red box on the genomic plots.In the insets, the relevant region on chromosome 15 is zoomed in for each cell.[Colour figure can be viewed at wileyonlinelibrary.com]

F I G U R E 4
Detection of a Williams syndrome deletion (7q11.2) in two cells from a monochorionic twin pregnancy.Panel (A) shows a view of chromosomes 6 to 13 from genomic plots, panel (B) chromosomes 6 to 8, and panel (C) a subregion of chromosome 7. [Colour figure can be viewed at wileyonlinelibrary.com]T A B L E 7 Cases with differences compatible with placental mosaicism or rare autosomal trisomy/RAT (10 cases).

F I G U R E 5
Statistical analysis of comparison of Luna test to CVS and amniocentesis.(A) Excluding mosaicism.(B) Including mosaicism.T A B L E 9 Cell line IDs and associated abnormalities.
Subject and sample information associated with the no follow up (NFU) study cases.

Table 6 .
Information on the type of invasive procedure, gestation, and Chromosomal abnormalities a in 158 pregnancies in the no follow-up (NFU) series.
tests performed is provided in Appendix 1.The CVS/amnio samples were analyzed for karyotype/aneuploidy FISH only in 61% and by array þ/− karyotype in 39%.For 160 singleton pregnancies (83 female and 77 male), Luna results were normal, and sex agreed with CVS/amnio results.For 14 aneuploidy cases, nine T21, three T18, one T13, and one XXX, Luna test results agreed with those from CVS/ amnio.For one case, the chromosome analysis of CVS tissue showed T21 in all 20 cells analyzed, with four cells also showing T8, while T8 was not detected in any cells analyzed by the Luna test.This case was group showed a deletion of 1.5 Mb at 17p12 consistent with a diagnosis of hereditary neuropathy with liability to pressure palsies (OMIM 162500) in five cells on the Luna test.Two cells were scored as aneuploidy only due to quality thresholds, but the deletion was still visually evident.Unfortunately, this finding could not be confirmed because only karyotype analysis was performed on the cultured CVS, T A B L E 5 a Dup 18p11.32 nonpathogenic not included.
Summary of 243 cases attempting to compare the Luna test with CVS or amniocentesis.a T A B L E 6 a See the text of Results for further explanation.b Some cases are listed twice indicated as counted above.BELLAIR ET AL.
9,33AIR ET AL. normal male pregnancies.9,33Recovery of other cell types such as fnRBC, fetal lymphocytes, or fetal hematopoietic stem progenitor cells (HSPC) is theoretically possible and desirable, but the authors are not aware of reliable recovery of such cells in clinically usable numbers in the first trimester.Non-trophoblast cells would be especially helpful in addressing placental mosaicism.Processes for downstream analyses of recovered cells could also be improved.Particularly, cells in S phase are currently not scorable for small deletions.We have observed that deeper sequencing of cells in the S phase can provide improved del/dup detection (data not shown).If the focus is on detecting inherited or de novo monogenic mutations, S phase cells are not at a disadvantage.The use of cell-based noninvasive prenatal testing is impacted by placental aneuploidy mosaicism, but only a small fraction of pregnancies are involved (5.3% in this study).Placental mosaicism is classified as type 1 if confined to trophoblasts, type 2 if confined to the chorionic villus stroma, and type 3 if both trophoblast cells and the villus stroma are involved.Since the Luna test utilizes trophoblasts for analysis, it will permit detection of types 1 and 3 mosaicism, but not type 2. CVS can detect all types of placental mosaicism with evidence that direct cell analysis queries primarily cytotrophoblasts, but long-term culture queries primarily mesenchymal cells.We observed multiple examples of both Luna test-negative and CVS/ amnio-positive and Luna test-positive and CVS/amnio-negative results.This is not surprising as placental mosaicism represents a complex sampling problem, and few cells are available in the Luna test.It is known that aneuploidy mosaicism is quite common in day 5 a failure rate of 7.6%-10%.Hou et al., in a correction to a publication, reported 1.1-3.1 trophoblasts per mL of blood in six 314 -