The Application Value of Chromosome Microarray Analysis in Prenatal Diagnosis of Clinically Relevant Copy Number Variations in Fetuses

Abstract

Objective

Exploring the clinical application value of chromosome microarray analysis (CMA) technology in detecting clinically relevant copy number variations (CNVs) in fetuses and evaluating its efficacy in prenatal diagnosis.

Methods

A retrospective analysis was conducted on 2860 pregnant women undergoing prenatal diagnosis between January 2019 and August 2023. All cases received concurrent G-banded karyotyping and CMA. Detection rates were compared and stratified by clinical indications (eg, structural anomalies, increased nuchal translucency NT). CNVs were classified per ACMG guidelines. Parental origin verification and neonatal follow-up (6 months to 2 years) were performed to correlate CNVs with postnatal phenotypes.

Results

CMA yielded a significantly higher overall detection rate than karyotyping (14.41% vs 10.10%, P<0.05), identifying 238 pathogenic (8.32%), 62 likely pathogenic (2.17%), and 112 VUS (3.92%) cases. The highest CMA yield occurred in fetuses with structural anomalies (26.85%), particularly multi-system defects (25.16% P/LP yield). Pathogenic yields also rose progressively with NT thickness. Of the 300 P/LP CNVs, 54.00% were microdeletions and 61.33% were <5 Mb, with DiGeorge syndrome being most prevalent (5.67%). De novo P/LP CNVs showed a significantly higher rate of abnormal postnatal phenotypes (40.00%) compared to inherited cases (10.53%, P<0.05). For VUS, inherited cases had a significantly higher postnatal normality rate (93.75%) than de novo variants (66.67%, P<0.05).

Conclusion

Based on large sample data, this study confirms that CMA technology can not only improve the overall detection rate of fetal chromosomal abnormalities, but also effectively identify submicroscopic CNVs within the blind spots of traditional karyotype analysis. It serves as a high-value first-line tool for fetuses with multisystem anomalies or increased NT. Parental origin verification is critical for refining VUS interpretation and guiding pregnancy management.

Introduction

Birth defects represent a critical global public health challenge, profoundly affecting both population quality and pediatric health outcomes.¹ The primary objective of prenatal diagnosis is the timely identification and assessment of fetal developmental abnormalities to provide a robust scientific foundation for clinical intervention. While routine prenatal ultrasound is effective in detecting structural anomalies, its diagnostic utility remains limited regarding underlying genetic etiologies.² Chromosomal aneuploidy is a prevalent form of genetic abnormality for which traditional G-banded karyotyping has long served as the “gold standard.” However, karyotyping is constrained by inherently low resolution (typically 5–10 Mb), laborious laboratory procedures, and prolonged turnaround times, rendering it incapable of detecting submicroscopic chromosomal structural variations.³

With the continuous evolution of prenatal diagnostic technologies, chromosomal microarray analysis (CMA) has emerged as a high-resolution, genome-wide screening tool that offers significant advantages in identifying submicroscopic aberrations.⁴ The sub-microscopic structural CNVs refer to copy number variations that cannot be detected by traditional G-banding karyotype analysis (with a resolution of 5–10 Mb) and need to be identified through molecular techniques such as CMA, typically with segment sizes <5–10 Mb.⁵ The filtering threshold commonly used in conventional whole-genome CMA testing is typically set at ≥50 kb. This setting effectively identifies clinically relevant microdeletions and microduplications, with some variations involving the exon regions of genes.⁶ In contrast, CNVs in non-exon regions have relatively lower pathogenic risks if they do not affect regulatory elements. Research has shown that intragenic CNVs account for a significant proportion of pathogenic CNVs. Some research reports indicated that this proportion can reach around 10%. Approximately one quarter of these involve a single exon, highlighting the clinical value of conducting high-resolution analysis of sub-microscopic structural CNVs.⁷ Compared with traditional karyotyping analysis, CMA technology can detect CNVs of smaller fragments involving microdeletions and microduplications in exon regions that cannot be identified by traditional karyotyping analysis, providing a new direction for exploring the genetic causes of fetal structural abnormalities.^8,9 Therefore, the identification of submicroscopic structural CNVs (particularly those involving exon regions that are clinically relevant) is of crucial importance for precise genetic diagnosis. CMA technology has certain clinical value in detecting submicroscopic CNVs. Based on different chip types, CMA can detect genomic imbalanced small fragment duplications or deletions as low as 1 kb at the highest resolution, with some variations involving exon regions.¹⁰ This study focused on clinically relevant CNVs that could be clearly interpreted under the conventional clinical filtering threshold (≥ 50 kb). Empirical evidence suggests that in cases where fetal ultrasound is abnormal, but the karyotype remains unremarkable, CMA provides an incremental diagnostic yield of approximately 4–6%; this efficacy is particularly pronounced in fetuses presenting with multisystem structural anomalies.¹¹ However, most existing studies have focused on the detection efficacy of CMA for whole-genomic CNVs, while there are still limited studies specifically addressing the clinical value of its detection of sub-microscopic structural CNVs (particularly clinically relevant variants involving exon regions), especially regarding the distribution characteristics and clinical interpretation strategies in fetuses with different prenatal diagnostic indications. Clarifying the incremental value of CMA in detecting submicroscopic structural CNVs is of crucial importance for optimizing prenatal genetic counseling.

Based on this, this study aimed to explore the clinical application value of CMA technology in detecting fetal submicroscopic structural CNVs, analyze the detection efficacy of CMA under different prenatal diagnostic indications, evaluate the complementary role of CMA technology and traditional karyotype analysis in prenatal diagnosis, and thereby provide a basis for optimizing prenatal genetic diagnosis strategies and guiding clinical precise genetic counseling.

Materials and Methods

Study Population

A total of 2860 pregnant women who underwent prenatal diagnosis at our institution between January 2019 and August 2023 were retrospectively enrolled. The participant inclusion flow chart is presented in Figure 1.

Figure 1.

Flowchart of the Screening Process for Pregnant Women.

Inclusion criteria were: (1) singleton pregnancy; (2) gestational age of 16–24 weeks; (3) indications for invasive prenatal diagnosis (amniocentesis, cordocentesis, or chorionic villus sampling), including ultrasound-detected structural anomalies, nuchal translucency (NT) thickness ≥ 2.5 mm, high-risk non-invasive prenatal testing (NIPT) results, high-risk maternal serum screening, advanced maternal age (≥ 35 years at delivery), adverse obstetric history, parental chromosomal abnormalities or maternal anxiety requesting the procedure; and (4) concurrent completion of both G-banded karyotyping and CMA.

Exclusion criteria were: (1) multiple gestations; (2) failure to complete both testing modalities; (3) incomplete clinical data, lack of key demographic information, or unclear prenatal indications; (4) loss to follow-up or refusal to participate in follow-up studies; and (5) inadequate sample quality failing to meet testing requirements (eg, DNA degradation or insufficient concentration) The reasons for selecting patients for invasive prenatal diagnosis in this study were as follows: Firstly, the research objective was to evaluate the clinical application value of CMA in prenatal diagnosis. The invasive procedures (amniocentesis, umbilical cord blood sampling, chorionic villi puncture) are necessary for obtaining fetal samples for CMA testing, so patients who have undergone invasive diagnoses must be included. Secondly, invasive prenatal diagnosis is applicable to pregnant women with clear indications for prenatal diagnosis (such as ultrasound structural abnormalities, NT thickening, high-risk NIPT results, advanced maternal age, etc). The indication distribution of the patients included in this study conformed to clinical norms and was representative. Thirdly, this study required both traditional karyotyping analysis and CMA testing, and only invasive procedures could obtain sufficient quantity and quality of fetal DNA samples to meet the requirements of both testing methods. The CNVs detected in this study were defined as genomic fragments with a size of ≥ 50 kb and a continuous probe count of ≥ 25, covering some clinically relevant variations involving exon regions, but not specifically referring to exon-level resolution.

Research Methods

Study Design

In this retrospective study, all participants received standardized genetic counseling from certified counselors prior to invasive procedures. Counseling addressed the indications for testing, the principles and limitations of karyotyping versus CMA, potential clinical implications of the findings, procedure-related risks, and potential management pathways. After signing informed consent, all participants underwent both traditional karyotyping and CMA testing.

The study analysis included: (1) comparison of detection rates and characterization of discordant results between the two methods; (2) analysis of CMA positivity rates across different indications; (3) subgroup analysis of structural anomalies (single-system vs multi-system) and specific organ systems; (4) correlation analysis between NT thickness (subdivided into 2.5–3.0mm, 3.0–3.4mm, ≥3.5mm. Among them, 2.5–3.4mm was considered the critical thickening range, while ≥ 3.5mm indicated a significant thickening.) and CMA abnormality rates; (5) classification of copy number variations (CNVs) by pathogenicity, type, size, and microdeletion/duplication syndromes; (6) parental origin verification for variants of uncertain significance (VUS) and pathogenic/likely pathogenic (P/LP) CNVs to analyze the association between parental origin and postnatal phenotype; and (7) long-term follow-up (6 months to 2 years post-birth) to evaluate the correlation between pregnancy outcomes, CNVs, and neonatal phenotypes.

This study combined different prenatal diagnostic indicators (such as ultrasound structural abnormalities, NT thickening, high-risk NIPT, high-risk serological screening, advanced maternal age, adverse pregnancy history, maternal anxiety, etc) into the same analysis cohort. The reason for this were as follows: firstly, the above indicators are common indications for invasive prenatal diagnosis in clinical practice, and combined analysis can reflect the overall detection efficiency of CMA in the real world; Secondly, one of the main objectives of this study was to evaluate the differences in the detection rate of CMA in different subgroups of indications. Therefore, it was necessary to compare and analyze each subgroup within the same cohort, rather than focusing solely on a single indication; Thirdly, combined analysis could provide a larger sample size, improve statistical efficiency, and provide a foundation for subsequent subgroup analysis.

Conventional Chromosomal Karyotyping

Standard G-banding was performed following cell culture, harvesting, hypotonic treatment, fixation, and slide preparation. Analysis was conducted using an Olympus BX53 microscope, with at least 20 metaphase cells analyzed per sample at a resolution of 320–400 bands. Karyotypes were reported according to the International System for Human Cytogenetic Nomenclature (ISCN 2020).

CMA Technical Workflow

Genome-wide CNV detection was performed as follows:

1. Sample Collection and DNA Extraction: Samples included amniotic fluid (n = 2587), umbilical cord blood (n = 221), and chorionic villi (n = 52). Genomic DNA (gDNA) was extracted using the QIAamp DNA Mini Kit (Qiagen, Germany) according to the manufacturer’s protocol.

2. Quality Control: DNA purity was verified using a NanoDrop 2000 (A260/A280 ratio: 1.8–2.0). Concentration was quantified using a Qubit 4.0 Fluorometer and the dsDNA HS Assay Kit (Thermo Fisher Scientific, USA). Requirements for hybridization were a concentration ≥ 50 ng/μL and total mass ≥ 500 ng.

3. Microarray Platform: The Affymetrix CytoScan™ HD platform (Thermo Fisher Scientific, USA) was utilized. The workflow-digestion, ligation, PCR amplification, purification, fragmentation, labeling, hybridization, washing, staining, and scanning (GeneChip^® Scanner 3000 7G)-followed the manufacturer’s standard instructions.

4. Data Analysis: Data were processed using Chromosome Analysis Suite (ChAS) v4.2. CNV filters were set at ≥ 50 kb in size and ≥ 25 contiguous probes. Clinical annotation utilized DECIPHER, DGV, OMIM, ClinGen, and ClinVar databases. Classification included: Pathogenicity (5 classes per ACMG), Type (microdeletion, microduplication, or combined), Size (<5Mb, 5–10Mb, >10Mb), and Syndrome classification.

The “sub-microscopic structural CNVs” analyzed in this study were defined as copy number variations that were ≥ 50 kb in size, ≥ 25 consecutive probes, confirmed to cover one or more gene exon regions after database annotation, and might be associated with known genetic diseases or fetal phenotypes. This definition was set for clinical interpretation purposes and did not refer to the technical exon-level resolution detection. Although the highest resolution of the Affymetrix CytoScan™ HD chip can reach 1 kb, but in this study, a filtering threshold of ≥ 50 kb was used to balance the diagnostic rate of clinically reportable variations and the risk of false positives. The focus was on identifying sub-microscopic CNVs that cannot be detected by traditional karyotype analysis, including microdeletions/microduplications in some exon regions.

Pathogenicity Classification

Following ACMG guidelines,¹² CNVs were categorized into five classes: (1) Pathogenic (P): containing known pathogenic genes or regions highly correlated with fetal phenotype; (2) Likely Pathogenic (LP): containing potentially pathogenic genes/regions with slightly weaker evidence than P; (3) Variant of Uncertain Significance (VUS): genetic changes where the association with the disease/abnormality cannot currently be determined; (4) Likely Benign; and (5) Benign. This study focused primarily on P, LP, and VUS categories.

Parental Origin Verification and Follow-up

For detected VUS and P/LP cases, parental peripheral blood was analyzed via CMA to determine if the variation was inherited (paternal/maternal) or de novo. In conjunction with parental phenotypes, the clinical significance of VUS was re-evaluated, and the distribution of parental origins was recorded.

Follow-up Study

Follow-up was conducted until 6 months to 2 years post-birth. Neonatal assessment included physical examination (to screen for congenital malformations, dysmorphic features, and growth parameters), targeted ultrasound (heart, brain, abdomen, etc), and standardized developmental scales (the Gesell Developmental Schedule). Statistical indicators included: (1) continued pregnancy vs termination rates; (2) postnatal normal phenotype rates for VUS based on parental origin; (3) types and incidence of abnormal phenotypes in P/LP cases; and (4) phenotypic differences between de novo and inherited pathogenic variations.

Statistical Analysis

Data were analyzed using SPSS 25.0. Continuous data were first tested for normality; normally distributed data were expressed as mean ±standard deviation and compared using t-tests. Non-normally distributed data were expressed as median (interquartile range) and compared using the Mann–Whitney U-test. Categorical variables were expressed as percentages (%) and analyzed using the χ² test or Fisher’s exact test. A p-value < 0.05 was considered statistically significant.

Result

Baseline Characteristics

This retrospective study included 2860 singleton pregnancies undergoing invasive prenatal diagnosis. Maternal characteristics are summarized in Table 1. The mean maternal age was 32.4±4.8 years, with 966 cases (33.78%) classified as advanced maternal age (≥35 years at the expected date of confinement). Sampling was primarily performed between 19 and 21 gestational weeks (1287 cases, 45.00%). The leading indication for prenatal diagnosis was ultrasound-detected fetal structural anomalies (648 cases, 22.66%), followed by advanced maternal age (621 cases, 21.71%) and high-risk serum screening results (523 cases, 18.29%). The vast majority of samples (2587 cases, 90.45%) were obtained via amniocentesis.

Table 1.

Baseline Characteristics of the Study Population (n = 2860)

Characteristic	Category	Number of Cases	Percentage (%)
Maternal Age (years)	<35	1894	66.22
	≥35	966	33.78
Gestational Age (weeks)	16–18	1025	35.84
	19–21	1287	45.00
	22–24	548	19.16
Indications for Prenatal Diagnosis	Ultrasound structural abnormalities	648	22.66
	Increased NT (≥ 3.5mm)	392	13.71
	High Risk NIPT result	435	15.21
	High-risk serum screening	523	18.29
	Advanced maternal age	621	21.71
	Adverse obstetric history	164	5.73
	Other*	77	2.69
Sampling Method	Amniotic fluid	2587	90.45
	Cord Blood	221	7.73
	Chorionic Vili	52	1.82

Note: *Others include parental chromosomal abnormalities and maternal anxiety.

Abbreviations: NT, nuchal translucency; NIPT, non-invasive prenatal testing.

Comparison of CMA and G-Banded Karyotyping

CMA identified 412 chromosomal abnormalities across the 2860 samples, yielding an overall detection rate of 14.41%. This included 238 (8.32%) pathogenic CNVs, 62 (2.17%) likely pathogenic CNVs, and 112 (3.92%) variants of uncertain clinical significance (VUS). In contrast, conventional karyotyping detected 289 abnormalities (10.10%), comprising 92 cases of aneuploidy (3.22%) and 197 structural abnormalities, such as balanced translocations and inversions (6.89%).

The diagnostic yield of CMA was significantly higher than that of karyotyping (P < 0.05). Specifically, among the 1796 samples with normal karyotypes, CMA identified an additional 207 abnormalities (11.53%), including 47 (2.62%) P-CNVs, 31 (1.73%) LP-CNVs, 112 (6.24%) VUS, and 17 (0.95%) regions of loss of heterozygosity (LOH). Conversely, karyotyping detected 39 abnormalities in CMA-normal samples (n=2448); all of these were either balanced structural rearrangements (n=28) or low-level mosaicism (n=11), and no pathogenic CNVs were identified in these cases. Overall, CMA provided an incremental yield of 123 (4.30%) clinically significant abnormalities (P/LP CNVs) over conventional karyotyping (Table 2).

Table 2.

Comparison of Diagnostic Yield Between CMA and Conventional Karyotyping

Detection Method	Abnormal Detection Rate, n (%)	Clinically Significant CNVs (P/LP) n (%)	VUS Detection Rate, n (%)
Conventional Karyotype Analysis	289 (10.10)	289 (10.10)	–
CMA	412 (14.41)	300 (10.49)	112 (3.92)
χ2Value	24.596	–	–
P-value	<0.001	<0.001	–

Abbreviations: CMA, chromosomal microarray analysis; CNVs, copy number variations; VUS, variants of uncertain clinical significance.

Diagnostic Yield by Prenatal Indication

Distribution of Indications

CMA positivity rates varied significantly across indications (Table 3). The highest yield was observed in fetuses with structural anomalies (26.85%), where P/LP CNVs accounted for 18.52% of cases. Other indications included increased NT (16.33%), high-risk NIPT (12.41%), and other indications (high-risk serum screening, advanced maternal age, and adverse obstetric history and other, 8.66%).

Table 3.

Comparison of CMA Results Across Different Prenatal Diagnostic Indications

Prenatal Diagnostic Indication	Total (n)	P/LP CNVs	VUS	Overall Abnormality Detection
Ultrasound structural abnormalities	648	120 (18.52%)	54 (8.33%)	174 (26.85%)
Increased NT (≥2.5mm)	392	42 (10.71%)	22 (5.61%)	64 (16.33%)
High-risk NIPT result	435	38 (8.74%)	16 (3.68%)	54 (12.41%)
Other indications (High-risk serum screening, Advanced maternal age, Adverse obstetric history, others)	1385	100 (7.22%)	20 (1.44%)	120 (8.66%)
Total	2860	300 (10.49%)	112 (3.92%)	412 (14.41%)

Abbreviations: P/LP, pathogenic/likely pathogenic; CNVs, copy number variations; VUS, variants of uncertain significance; NT, nuchal translucency; NIPT, non-invasive prenatal testing.

Ultrasound Abnormality Subgroups

The 648 fetuses with structural anomalies were categorized into single-system (n=493) and multi-system (n=155) groups. The P/LP CNV detection rate was significantly higher in the multi-system group than in the single-system group (25.16% vs 15.21%; χ²=12.68, P<0.001). In terms of overall abnormal detection rates, multiple-system structural abnormalities accounted for 36.77%, while single-system structural abnormalities accounted for 22.51% (Table 4).

Table 4.

Analysis of CMA Results by Ultrasound Abnormality Subgroup

Ultrasound Abnormality Subtype	Total (n)	P/LP CNVs	VUS	Overall Abnormality Detection
Multisystem anomalies	155	39 (25.16%)	18 (11.61%)	57 (36.77%)
Isolated (Single system) anomalies	493	75 (15.21%)	36 (7.30%)	111 (22.51%)
χ2-value		12.68
P-value		<0.001

Abbreviations: P/LP, pathogenic/likely pathogenic; CNVs, copy number variations; VUS, variants of uncertain significance.

NT Thickening Subgroups

Among 392 cases of increased NT, fetuses were divided into three subgroups: 2.5–3.0 mm (n = 127), 3.0–3.4 mm (n = 143), and ≥ 3.5 mm (n = 122). Detection rates rose progressively with increasing NT thickness, with rates of 11.02% for 2.5–3.0 mm, 15.38% for 3.0–3.4 mm, and 24.59% for ≥ 3.5 mm. The ≥3.5 mm subgroup showed a significantly higher pathogenic yield (18.03%) compared to the other subgroups (P<0.05; Table 5).

Table 5.

Analysis of CMA Results Across Different NT Thickness Subgroups

NT Thickness (mm)	Total	P/LP CNVs		VUS		Overall Detection
		(n)	Percentage (%)	(n)	Percentage (%)	(n)	Percentage (%)
2.5–3.0	127	8	6.30	4	3.15	2	1.57
3.0–3.4	143	15	10.49	6	4.19	11	7.69
≥3.5	122	22	18.03	9	7.38	8	6.56
χ2-value
P-value

Note: The overall detection rate and P/LP yield show a significant upward trend as NT thickness increases.

Abbreviations: NT, nuchal translucency; P/LP, pathogenic/likely pathogenic; CNVs, copy number variations; VUS, variants of uncertain significance.

CNV Characteristics and Syndrome Distribution

Distribution of CNV Types

Among the 300 P/LP CNVs, 162 (54.00%) were microdeletions, 123 (41.00%) were microduplications, and 15 (5.00%) involved concurrent microdeletions and duplications. The positive predictive values (PPVs) for these three types were 56.32%, 50.41%, and 80.00%, respectively; however, these differences were not statistically significant (χ²= 3.21, P > 0.05).

CNV Size Distribution

Regarding fragment size, 184 cases (61.33%) were < 5 Mb, 79 (26.33%) were 5–10 Mb, and 37 (12.34%) were > 10 Mb. The proportion of P/LP findings within these size categories was 53.26%, 60.76%, and 56.76%, respectively. No significant correlation was found between CNV size and pathogenicity (χ²= 0.58, P > 0.05), although the pathogenic rate for the 5–10 Mb fragment size was slightly higher than that of other sizes (Table 6).

Table 6.

Distribution Characteristics of CNV Types and Fragment Sizes (n=300)

Classification	Subgroup	(n)	Frequency (%)	Pathogenicity Rate (%)	PPV (%)
CNVs Type	Microdeletion	162	54.00	55.56	56.32
	Microduplication	123	41.00	48.78	50.41
	Microdeletion + Microduplication	15	5.00	80.00	80.00
	χ2-value			3.21	3.41
	P-value			0.201	0.181
Fragment Size	<5Mb	184	61.33	53.26	53.54
	5–10Mb	79	26.33	60.76	60.00
	>10Mb	37	12.34	56.76	56.76
	χ2-value			0.58	0.465
	P-value			0.748	0.792

Abbreviations: CNVs, copy number variations; PPV, positive predictive value; Mb, megabase.

Distribution of Common Pathogenic Syndromes

We identified 23 distinct microdeletion/duplication syndromes among the 300 P/LP cases. The most frequent were DiGeorge syndrome (22q11.21 deletion, n=17; 5.67%), Wolf-Hirschhorn syndrome (4p16.3 deletion, n=12; 4.00%), 15q11q13 duplication syndrome (n=9; 3.00%), and Potocki-Lupski syndrome (17p11.2 duplication, n=8; 2.67%, Table 7).

Table 7.

Distribution of Common Pathogenic CNV Syndromes (n=300)

Syndrome	Chromosomal Locus	(n)	Frequency (%)
DiGeorge syndrome	22q11.21(del)	17	5.67
Wolf-Hirschhorn syndrome	4p16.3(del)	12	4.00
15q11q13 duplication syndrome	15q11.2-q13.1(dup)	9	3.00
Potocki-Lupski duplication syndrome	17p11.2(dup)	8	2.67
Other syndromes	–	254	84.66
Total		300	100

Abbreviations: del, deletion; dup, duplication; RCAD, Renal Cysts and Diabetes syndrome.

Parental Verification and Clinical Outcomes for VUS

Parental origin was verified as 86 of the 112 VUS cases (76.79%). Of these, 68 (79.07%) were inherited from phenotypically normal parents, and 18 (20.93%) were de novo. The rate of normal postnatal phenotypes (defined as live-born infants without clinical manifestations related to CNVs) was significantly higher in the inherited group (93.75%, calculated as [29+31] / [30+34]) than in the de novo group (66.67%, 8/12, P < 0.05; Table 8).

Table 8.

Relationship Between Parental Origin and Clinical Outcomes in VUS Cases

Parental Origin	(n)	Frequency (%)	Continued Pregnancy, n (%)	Termination of Pregnancy, n (%)	Normal Postnatal Phenotype, n (%)	Abnormal Postnatal Phenotype, n (%)
Paternal	32	37.21	30 (93.75)	2 (6.25)	29 (96.67)	1 (3.33)
Maternal	36	41.86	34 (94.44)	2 (5.56)	31 (91.18)	3 (8.82)
De novo	18	20.93	12 (66.67)	6 (33.33)	8 (66.67)	4 (33.33)
Total	86	100.00	76 (88.37)	10 (11.63)	68 (89.47)	8 (10.53)
Inherited (Total)	68	79.07	64 (94.12)	4 (5.88)	60 (93.75)	4 (6.25)

Notes: Parental origin verification was successfully performed in 86 of 112 cases (76.79%); the remaining 26 cases were excluded due to parental refusal of blood sampling or loss to follow-up. 64 cases inherited from the parental group continued pregnancy until delivery, of which 60 cases had normal phenotypes, with a normal phenotype rate of 93.75%. Normal postnatal phenotype is defined as the absence of clinical manifestations related to the detected VUS during the follow-up period (6 months to 2 years post-birth), as determined by pediatric physical examination, ultrasound, and developmental assessment.

Abbreviations: VUS, variant of uncertain clinical significance; n, number of cases.

Discrepancies Between CMA and Karyotyping

This study identified 15 cases with inconsistent results between CMA and conventional karyotyping. These included 9 cases where the karyotype was normal, but CMA detected pathogenic CNVs; 4 cases where karyotyping identified balanced translocations while CMA was normal; and 2 cases with other forms of inconsistency (Table 9).

Table 9.

Analysis of Cases with Discordant CMA and Conventional Karyotyping Results

Case No.	Indications	Karyotyping Results	CMA Results (Size)	Parental Origin	Pregnancy Outcome
1	Ultrasound: Cardiac anomaly	46, XN	16p13.11 dup (1.52Mb)	De novo	TOP
2	Ultrasound: Renal anomaly	46, XN	17q12 del (1.42Mb)	De novo	TOP
3	High-risk NIPT	46, XN	22q11.21 del (2.13Mb)	De novo	TOP
4	Advanced maternal age	46, XN	15q11.2 del (0.75Mb)	Maternal	CP, Normal phenotype
5	Ultrasound: CNS anomaly	46, XN	1q21.1 dup (1.12Mb)	De novo	TOP
6	Adverse obstetric history	46, X, t(X;18)(q22;q21)	Normal	Paternal	CP, Normal phenotype
7	High-risk serum screening	46, X, t(7;15)(q32;q26)	Normal	De novo	CP, Normal phenotype
8	Ultrasound: Multisystem anomaly	46, XN	8p23.1 del (3.76Mb)	De novo	TOP
9	Increased NT	46, XN	10q22.3q23.2 dup (5.32Mb)	De novo	TOP
10	Ultrasound: Facial anomaly	46, XN	4p16.3 del (2.85Mb)	De novo	TOP
11	Advanced maternal age	46, X, t(2;8)(p12;q24)	Normal	Paternal	CP, Normal phenotype
12	High-risk NIPT	46, XN	11q23.2q23.3 del (1.87Mb)	De novo	TOP
13	Ultrasound: Skeletal anomaly	47, XN, +mar	Normal	–	CP, Normal phenotype
14	High-risk serum screening	46, XN	7q11.23 dup (1.42Mb)	Maternal	CP, Normal phenotype
15	Adverse obstetric history	46, X, inv(9) (p12q13)	Normal	Paternal	CP, Normal phenotype

Follow-up and Association Between Phenotypes and CNVs

Among the 300 fetuses with P/LP CNVs, 227 (75.67%) underwent termination of pregnancy, while 73 (24.33%) continued to term. Among the 73 live births, 18 (24.66%) exhibited abnormal phenotypes associated with the CNVs. These included intellectual disability (n=6, associated with 15q11q13 duplication, 2q13 microduplication, etc), congenital heart disease (n=5, associated with 22q11.21 deletion, 7q11.23 duplication, etc), developmental language delay (n=4, associated with 16p11.2 and 15q13.3 deletions), and dysmorphic facial features (n=3, associated with 4p16.3 and 8p23.1 deletions).

The abnormal phenotype rate was 11.11% for cases inherited from the father and 10.00% for those from the mother. In contrast, the rate in the de novo group was 40.00%, representing a statistically significant difference compared to the inherited groups (χ²= 8.523, P < 0.05; Table 10 and Figure 2).

Table 10.

Comparison of Follow-up Outcomes for P/LP CNVs Based on Parental Origin [n (%)]

Parental Origin	(n)	Termination of Pregnancy	Continued Pregnancy to Term	Abnormal Postnatal Phenotype
Inherited (Total)	136	98 (72.06)	38 (27.94)	4 (10.53)
Paternal	65	47 (72.31)	18 (27.69)	2 (11.11)
Maternal	71	51 (71.83)	20 (28.17)	2 (10.00)
De novo	164	129 (78.66)	35 (21.34)	14 (40.00)
Total	300	227 (75.67)	73 (24.33)	18 (24.66)

Note: The abnormal postnatal phenotype rate was calculated using the number of “Continued Pregnancy to Term” cases as the denominator.

Abbreviations: P/LP, pathogenic/likely pathogenic; CNVs, copy number variations; n, number of cases.

Figure 2.

Comparison of the rates of abnormal phenotypes in live-born fetuses associated with pathogenic or likely pathogenic copy number variations (CNVs) derived from different parental origins.

Discussion

Chromosomal abnormalities are a primary cause of birth defects and spontaneous abortions. While traditional G-banded karyotyping remains a “gold standard” for prenatal diagnosis, its limited resolution (typically 5–10 Mb) makes it insufficient for detecting submicroscopic copy number variations (CNVs).^13,14 In this study of 2860 pregnancies, chromosomal microarray analysis (CMA) achieved an overall abnormality detection rate of 14.41%, significantly higher than the 10.10% yield of karyotyping. This confirms that CMA effectively identifies microscopic aberrations that escape conventional detection.

As a high-resolution, genome-wide screening tool, CMA identifies microdeletion and microduplication syndromes that karyotyping cannot resolve.¹⁵ We identified an additional 123 cases (4.30%) of clinically significant abnormalities via CMA, an incremental yield consistent with existing literature.^16,17 The original contribution of this study lies in: based on a large-scale Chinese prenatal diagnosis cohort, it systematically quantified the detection rate of exon-related clinical CNVs (reaching 18.52% in fetuses with ultrasound structural abnormalities), and clarified the independent impact of parental origin on fetal prognosis. The abnormal phenotype rate of de novo P/LP CNVs (40.00%) was four times higher than that of the genetic group. These findings may have the following implications for prenatal diagnostic criteria: It is recommended to use CMA as the first-line diagnostic tool for fetal cases with ultrasound multisystem structural abnormalities and significant NT thickening. Incorporating parental source verification into the routine management process for P/LP CNVs and VUS can provide a basis for risk stratification in genetic counseling and pregnancy decision-making. Furthermore, these findings suggest that combining CMA with karyotyping offers irreplaceable value in bridging the technical gaps of traditional methods. Furthermore, CMA does not require cell culture, significantly shortening the testing cycle and providing a wider window for clinical decision-making in advanced-gestation pregnancies. It also demands lower sample quality, demonstrating higher success rates in miscarriage tissues unsuitable for culture.¹⁸ Research indicates that CMA success rates in such tissues are 30–40% higher than karyotyping,¹⁹ which is vital for analyzing pregnancy loss and guiding subsequent pregnancies. A previous study of 391 fetuses with structural anomalies²⁰ found a CMA diagnostic rate of 8.3%, with cardiovascular (12%) and multisystem (9.5%) anomalies showing the highest yields.

In our stratified analysis, the CMA detection rate for fetuses with structural anomalies was 26.85% (18.52% for P/LP CNVs). This is significantly higher than the 5.2% incremental yield reported in the systematic review by Hillman et al²¹ and the 9.1% P/LP CNV probability reported by de Wit et al.²² This discrepancy likely stems from a higher proportion of complex multisystem anomalies referred to our center and suggests that genetic etiology may vary across populations, necessitating evaluation based on local characteristics. Our subgroup analysis showed that multisystem and urogenital anomalies carried the highest genetic risk, reinforcing CMA as a first-line diagnostic tool for these cases. Additionally, in the adverse obstetric history cohort, the P/LP CNV yield was 6.71%, consistent with reports ranging from 5.1% to 17.43%.²³ For fetuses with increased nuchal translucency (NT 3.5 mm) and normal karyotypes, our CMA detection rate was 16.33%, aligning with studies suggesting an incremental yield of 9.1% in such cases.²⁴ Previous study has also confirmed that NT thickness within the critical thickening range of 2.5–3.4 mm is associated with an increased risk of fetal chromosomal abnormalities.²⁴ Our data provides direct evidence that even with a normal karyotype, CMA should be recommended for significantly thickened NT to exclude microdeletion/duplication syndromes. Finally, the finding that over 60% of P/LP CNVs were Mb and that DiGeorge syndrome was the most prevalent confirms CMA’s core advantage in identifying submicroscopic variations.

The clinical management of variants of uncertain significance (VUS) remains a challenge, with our 3.92% detection rate aligning with international reports of 3–5%.^25,26 The novelty of this study lies in providing quantitative, large-scale evidence for parental origin verification within a Chinese population. Our follow-up showed that VUS inherited from phenotypically normal parents resulted in a 93.75% rate of normal postnatal outcomes, significantly higher than the 66.67% in de novo cases. The case of 16p11.2 duplication in our cohort vividly illustrates how parental verification can pivot management from potential termination to a continued pregnancy with a favorable outcome. While previous studies^27–29 confirmed the value of parental testing, our data translates this into a practical clinical pathway. Therefore, we support parental origin verification as a routine practice to transform abstract VUS into concrete risk assessments. The results of this study can be understood in the broader context of the evolution of prenatal genetic diagnosis techniques. From traditional karyotype analysis to CMA, and then to whole exome sequencing and whole genome sequencing, prenatal diagnosis is undergoing a profound transformation from cytogenetics to genomics. A systematic review indicates that genome-wide CNV analysis has become a key component of modern prenatal diagnosis, with an incremental diagnostic rate of approximately 3–5% for CMA in fetal structural abnormalities.³⁰ Another study has shown that in fetuses with ultrasound structural abnormalities, CMA as a first-line detection tool can significantly improve the detection rate of genetic abnormalities.³¹ On this basis, the emerging prenatal exome sequencing further increases the diagnostic rate to 29.45% and can identify CNV/SNV composite heterozygous variations that cannot be detected by CMA.³² Therefore, the results of this study are consistent with the trend of prenatal diagnostic technology evolution mentioned above, suggesting that the precise identification of sub-microscopic structural CNVs is an important step in the current transition from CMA to higher-resolution genomic methods.

This study has the following limitations. Firstly, this study was a single-center retrospective study. The sample only included patients who underwent invasive puncture due to meeting the indications for prenatal diagnosis. Pregnant women who did not undergo puncture or only received non-invasive screening were not included. Moreover, as this center is a regional prenatal diagnosis center, the proportion of complex structural abnormalities in the referred cases may be higher. Therefore, the results may have selection bias, and the estimated abnormal detection rate of CMA may be higher than that of the general pregnant population. The conclusion should be treated with caution when extrapolated. Secondly, different prenatal diagnostic indications (such as ultrasound structural abnormalities, NT thickening, high-risk NIPT results, advanced maternal age, etc) were combined into the same analysis cohort, resulting in significant differences in case composition among subgroups. Although this study controlled for some confounding factors through subgroup analysis, it could not completely rule out mutual influences among the various indications. Thirdly, the follow-up period (6 months to 2 years) may be insufficient for assessing neurodevelopmental outcomes. Some neurodevelopmental disorders (such as mild intellectual disability, learning difficulties, or behavioral abnormalities) might only become apparent during school age or even later. Therefore, the conclusions of this study are mainly applicable to abnormal phenotypes that can be identified within 6 months to 2 years, and there are limitations in evaluating long-term neurodevelopmental outcomes. Fourthly, in this study, 26 cases of VUS (23.21%) were unable to complete parental source verification due to one or both parents refusing to draw blood or being lost to follow-up, which may affect the completeness of VUS related conclusions. Future multicenter, large-scale prospective studies are needed to further validate the indications and strategies for the application of CMA in prenatal diagnosis, and to extend the follow-up period to evaluate long-term neurodevelopmental outcomes.

Conclusion

In conclusion, this study confirmed that the CMA technology can effectively identify sub-microscopic structural CNVs within the blind areas of traditional karyotype analysis based on a large sample of 2860 cases. The overall abnormal detection rate (14.41%) was significantly higher than that of karyotype analysis (10.10%), and 4.30% of clinically significant CNVs were additionally detected. The diagnostic efficacy of CMA was closely related to the indications for prenatal diagnosis. It had diagnostic value in fetuses with ultrasound structural abnormalities (especially multiple system structural abnormalities) and thickened NT. It may serve as the first-line genetic diagnostic tool for such fetuses. The parental origin of CNVs was closely related to the risk of fetal abnormal phenotypes: the abnormal phenotype rate of de novo P/LP CNVs (40.00%) was significantly higher than that of inherited CNVs from the parental group (10.53%). The normal phenotype rate of VUS fetuses inherited from parents with normal phenotypes reached 93.75% after birth. Therefore, the validation of parental origin may provide crucial decision-making basis for the clinical interpretation of CNVs and pregnancy management.

Funding Statement

There is no funding to report.

Data Sharing Statement

All data generated or analysed during this study are included in this published article.

Ethics Approval

This study had been approved by the Ethics Committee of Henan Provincial People’s Hospital (Approval number: 2023-135, Approval date: October 13, 2023). All procedures performed in studies involving human participants were in accordance with the standards upheld with those of the 1964 Helsinki Declaration and its later amendments for ethical research involving human subjects.

Consent to Participate

Written Informed consent was obtained from every human participant in the study and the patients participating in the study all agree to publish the research results.

Disclosure

The authors declare that they have no competing interests in this work.