Comparison of karyotype analysis and SNP-array in prenatal diagnosis of mosaicism aneuploidy

Ting Wang Guangdong Women and Children Hospital Huamei Huang Guangdong Women and Children Hospital Lihua Wu Guangdong Women and Children Hospital Yanlin Huang Guangdong Women and Children Hospital Xianzheng Li Guangdong Women and Children Hospital Li Guo (  guoli3861@163.com ) Guangdong Women and Children Hospital Hanbiao Chen Guangdong Women and Children Hospital Jian Lu Guangdong Women and Children Hospital Min Jian Guangdong Women and Children Hospital

Chorionic villus and amniotic uid were cultured in situ( ask) to obtain karyotypes, while cord blood was harvested in suspension for karyotype . The Gbanding karyotype analysis was performed according to standard laboratory protocol. No less than 50mg of villi tissue, 2mL cord blood and 15 to 20mL amniotic uid were aspirated into asepsis pipe and sent to the laboratory as soon as possible. All samples were divided into two parts for cytogenetic and molecular testing.Before cell culture, the maternal decidua and blood clots in villi were removed under stereomicroscope. Cord blood samples were directly inoculated into culture medium for cell culture,while amniotic uid was cultured with 1 ml suspension after centrifugation. All prenatal samples were cultured in 3 lines, of which 2 lines were used for report and 1 line was used as backup.The backup cells should be retained before the experimental results were obtained. CV and AF were harvested according to the standard operation procedure of in situ while karyotyping was performed with clonal analysis(one metaphase was analyzed for each independent colony, and at least 15 colonies were analyzed by two lines in total).The criterion for the prenatal diagnosis of chromosome mosaicism is to nd one or more cells with the same abnormality in at least two independent primary cultures and a two-stage approach is applied to distinguish pseudomosaicism from true mosaicism.When one or more cells or a clone are abnormal in the primary culture, additional cells from third line culture are studied [12,13].

SNP-array
The DNA from the uncultured samples was extracted by using QIAamp DNA Blood Mini Kit (QIAGEN,Germany) according to the manufacturer's recommended procedure.NANODROP 2000(Thermo ,USA) was employed to detect DNA concentration. CytoScan 750K chip (Affymetrix,USA) was used for SNP-array detection, including enzyme digestion, ligation, PCR ampli cation, puri cation, fragmentation, labeling, hybridization, washing, scanning. All operating procedures are strictly in accordance with the requirements of the manufacturer. The results were analyzed with ChAS3.1.0.15 (Affymetrix,USA) software.
Clinical signi cance was systematically assessed by comparing detected copy number variations (CNVs) with values in the literature and in the following public databases the International Standards for Cytogenomic Arrays(ISCA, https://www.iscaconsortium.org/), Online Mendelian Inheritance in Man( OMIM,http://www. omim.org),DECIPHER database (http://decipher. sanger.ac.uk/),database of genomic variants (DGV,http: //dgv.tcag.ca /dgv/app/home) and UCSC (http: //genome. ucsc.edu/). Generally,mosaicism as low as 10% to 20% had been proved to be undetectable by molecular plalform [14]. Mosaicism aneuploidy or CNVs are determined by in-house custom python algorithm from the Affymetrix software (SNP-array) according to the log test/reference ratio [15]. In case mosaicism was found in uncultured samples, the back-up cultured cells were processed to SNP-array analysis for con rmatory.

Follow-up
The details of the pregnancy outcome were collected from genetic consultant and primary referral doctors.Telephone follow-up was performed by senior nurses to obtain detailed prognostic informations including delivery, pre-and postnatal growth, major malformations, psychomotor development,and other diseases.

Identi cation of MCC
The use of short tandem repeat markers (STR) has been added to the genetic testing of chorionic villi to increase the detection of MCC. In case maternal blood was mixed in amniotic uid samples,SNP-array was performed with cultured cells to avoid MCC. Hemoglobin electrophoresis was used to identify the occurrence of such contamination in cord blood testing. MCC was not detected in all mosaicism cases involved in this study.

Statistical analysis
Kappa analysis was employed to measure the agreement of the mosaic positive rate between karyotype analysis and SNP-array.A kappa value of 0 means accidental agreement, while a value of 1 means complete agreement. Fisher's exact test was used to compare the counting data among the involved groups.Statistical signi cance was de ned as a p value < 0.05.Chi-square test was used for comparing the two groups, and Bonferroni test was used to correct the test level.All the analysis was performed with SPSS Statistics 25.0 software(IBM).

Results
Overall data A owchart of mosaic results after conventional karyotype and SNP-array was shown in Fig. 2.In all the involved samples ,14,805 cases(959 from CV,9034 from AF and 4812 from CB) were both performed with SNP-array and karyotype analysis,of which 169 mosaic cases(19 CV,104 AF and 46 CB) were identi ed by SNP-array or karyotype analysis. Mosaicism was found in both karyotype analysis and SNP-array in 99(0.67%,99/14,805) cases(10 CV,59 AF and 30 CB) of prenatal samples. In the remaining 70 cases of mosaicism, the results of karyotype analysis and SNP-array were discrepant with ten cases (1.04%,10/959),forty-ve cases (0.5%,45/9034) and fteen cases (0.31%,15/4812) dectected from CV,AF and CB respectively.The discrepant cases were divided into three groups:(I) Mosaic karyotype with normal SNP-array;(II) Mosaic SNP-array with normal karyotype;(III) Mosaic karyotype with non-mosaic abnormal SNP-array. Mosaicism dectected in both SNP-array and karyotype Ninety-nine prenatal samples of mosaicism were detected in all cell populations by both SNP-array(uncultured cell) and karyotype analysis. The mosaic levles ranged from 31-89% and the mean was 63%.The majority of the mosaic samples were sex chromosome mosaicism accounting for 60.6%(60/99).Of the remaining 39 cases, common aneuploidy(trisomy (t) 21, t18, and t13) accounted for 13.1%(13/99), other whole chromosome aneuploidy for 12.1% (12/99) and partial chromosome aberrations for 14.1% (14/99).Of these, 66 pregnancies were terminated, 22 pregnant women gave birth to normal infants, and 11 were lost to follow-up as shown in Fig. 3 .
In group I (Table 1), fty-six cases of prenatal samples(7 from CV,35 from AF and 14 from CB) yield a normal SNP-array results in uncultured cells but mosaic karyotype in long-term culture. The mosaic levels ranged from 2-50% and the mean was 17%.All the back-up cells from long-term culture were performed with SNP-array for comparison and mosaicism were also detected in ve cases.Of the cases in group I,60.7%(34/56) samples involved sex chromosome mosaicism in karyotype analysis while SNP-array presented normal results.Of the remaining 22 cases, common aneuploidy accounted for 5.4%(3/56), other whole chromosome aneuploidy for 16.1%(9/56) and partial chromosome aberrations for 17.9% (10/56).The pregnancy outcomes of group I were 15 TOP,34 live born and 7 not available. 47,XX,+16(case 4) karyotype was respectively found in only one isolated culture, but it did not meet the reporting criteria. A total of 100 cultured cord blood lymphocytes were counted and one supernumerary marker chromosome(SMC) was found in case 5. However, it was not reported because only one SMC was recognized in a culture. Follow-up uorescence in situ hybridization(FISH) con rmed that the SMC was isochromosome 12p.All the back-up culture of these ve cases were performed with SNP-array for com rmation and no mosaicism was found. b:A total of 100 cultured cord blood lymphocytes were counted and one supernumerary marker chromosome(SMC) was found.
However, it was not reported because only one SMC was recognized in a culture. Follow-up FISH con rmed that the SMC was isochromosome 12p.
Mosaic karyotype with non-mosaic abnormal SNP-array In group III (Table 3),nine cases yielded mosaic karyotype from long-term culture but SNP-array showed non-mosaic and abnormal results.Of these,six cases(66.7%,6/9) were sex chromosome mosaicism in karyotype analysis but whole sex chromosome aneuploidy in SNP-array.The SNP-array result of case 3 demonstrated a double micro duplication in chromosome 4 and 16 while the mosaic SMC were found in karyotype analysis.Case 8 and 9 also showed the micro deletion and duplication in chromosome 17 and 3 respectively.

Data analysis
The data analysis was shown in Table 4 and Table 5.Comparing the agreement between karyotype analysis and SNP-array, the kappa value is 0.004 (P < 0.001), which means "slight agreement" [16]. The mosaic positive rate of karyotype analysis only was 1.11% (164/ 14,805), which was signi cantly higher than that of SNP-array (0.7%, 104/14,805), and the difference was statistically signi cant (P < 0.001). The mosaic positive rate of combination with SNP-array and karyotype was 1.14% (169/ 14,805), which was higher than that of SNP-array(0.7%).

Discussion
Chromosomal mosaicism referred to the existence of two or more cell lines with different karyotypes in cultures.The impact of mosaicism on fetuses depended on the distribution of abnormal cell lines in fetal and embryonic external tissue, which was determined by the mutation occurred time during the development of the embryos [17].It was still a challenge to diagnosis a mosaic fetus prenatally due to the limitation of the technologies available,the tissues sampled and the timing of invasive prenatal diagnosis [18].SNP-array has the advantages of high throughput analysis, automation, rapid reporting cycle and avoidance of culturing fetal cells in most cases. Compared with array comparative genomic hybridization (aCGH), SNP-array technique can detect not only tiny chromosomal abnormalities that can not be recognized by karyotype analysis, but also uniparental disomy and triploid [19].To our knowledge,this is the rst study which compared the positive rate of prenatal mosaicism by SNP-array and karyotype analysis based on large data.
In the current study,we identi ed 99 cases of mosaicism which were recognized by both SNP-array and karyotype analysis.AF cells are closer to fetal cell lines than CV cells due to the origin of multiple embryonic layers, therefore the mosaicism detected in AF samples is more likely to represent true fetal mosaicism(TFM) than that detected in CV samples. Of these, we discovered that the mosaicism ratio of karyotype analysis were discordant with that of SNParray in several cases.For example,4 cases of Pallister-Killian Syndrome(PKS) were observed with extremely low levels of additional isochromosome 12p in karyotype analysis of CB while the mosaic ratios in SNP-array were 49%, 56%, 74%, 85%, respectively.This phenomenon is caused by the tissue-limited mosaic characteristic of PKS and the loss of isochromosome 12p after CB culture [20].Different mosaic ratio between cultured(karyotype analysis) and uncultured(SNP-array) prenatal samples may be due to the selective growth in the culture process. Cell culture can lead to the selection of euploidy over aneuploidy cells in vivo, which increased with the duration of culture [9]. A bias of the selection to a particular cell type was presented during the culture whereas SNP-array performed with direct samples from different cell lineages do not have such potential bias [21].In view of this, the results of SNP-array may be more reliable if the mosaic ratio detected by the two technologies are discordant.
Majority(40/70,57.1%) of the mosaic cases with discordant results(group I,II and III) involved sex chromosome mosaicism.The incidence of sex chromosome aneuploidy mosaicism in blood is higher than that of autosomes [22].It had been reported that low levels of sex chromosome mosaicism were found in 34% of infertile men who accepted assessment and those with low levels mosaicism were signi cantly older [23].The discrepancies in the cases of group I may be due to the limitations of SNP-array which could not detect low levels of mosaicism.In group II, we identi ed two cases with single clonal abnormality in karyotype analysis after long-term culture, but with high rate of mosaicism in SNP-array.According to the previous standard chromosome karyotype analysis [24], if only a single abnormal clone is found in the single line culture, the result should be reported as no abnormal.However, we highly recommended that the abnormal karyotype should be mentioned in the advisory opinion of the prenatal genetic counseling report due to the hints of molecular technology.In group III, three cases with CNVs that could not be detected by karyotype analysis were identi ed by SNP-array which had the higher resolution than conventional cytogenetics. CNVs found accidentally in SNP-array should be explained in accordance with the guidelines [25]. In prenatal samples, mosaicism detected in uncultured materials is likely to represent the presence of normal and abnormal cell lines in the fetus.Those cases(group II) with apparent mosaic SNP-array results but karyotype of long-term culture failed to detect the abnormal cell line may be due to the discrepancies in cell populations after culture [26].The mosaic rate of karyotype analysis in group I ranged from 2 to 50% with an average of 17% while SNP-array showed no abnormaliltis. Theoretically,SNP-array can detect mosaic rate as low as 5%, but in fact, affected by the experimental process and sample quality, it can only detect mosaicism with a level of more than 20% in the application of prenatal diagnosis [27].That may be the reason why the SNP-array in our study could not detect mosaicism in group I.
The current study demonstrated that the mosaic positive rate of karyotype analysis was higher than that of SNP-array (1.11% versus 0.7%), and the difference was statistically signi cant (P < 0.001). Karyotype analysis can detect mosaicism at a level as low as 2%(case 44 and 51 in group I) and this can be used as a supplement to SNP-array's limitation to detect low-level mosaicism.The advantages of SNP-array was high resolutions and rapid direct detection which could be used as a complement to karyotype analysis. Therefore,the joint detection may be the best method to detect mosaicism based on the comparisons in Table 5.
Genetic counseling for prenatal mosaicism was still a challenge due to the complex mechanisms and unpredictable outcomes.The live born rates of the 99 prenatal samples with both mosaic SNP-array and karyotype in this study were 25% (22/88,11 cases lost to follow-up),which were signi cant lower than that of discordant cases(65%,41/63,7 cases lost to follow-up).This illustrated that the double mosaic cases seem like to have poor prognosis.However, even with the most precise prenatal testing,TFM does not necessarily mean that the fetus will have any consequences in phenotype, since the origin or proportion of cells carrying abnormalities cannot be predicted [19].These limitations should be clearly conveyed to those couples involved by the genetic counselors. For the cases with very low level mosaicism,we recommend a comprehensive assessment by combining the results of cytomolecular genetic examination and ultrasound examination.Unless a severe malformations was detected, termination of pregnancy was not recommended.
In conclusion,the current study demonstrated that the combination of SNP-array and karyotype analysis may be the best strategy for the prenatal diagnosis of mosaicism aneuploidy.These two techniques have their own advantages and disadvantages, which can complement each other in clinical application.