Supplementary Material for: The effect of resolution level and targeted design in the diagnostic performance of prenatal chromosomal microarray analysis

Introduction This study was performed to assess the optimal resolution for prenatal testing by array comparative genomic hybridization (aCGH), aiming to balance between maximum diagnostic yield and minimal detection of variants of unknown significance (VOUS). Methods This was a prospective study using data of 2336 fetuses that underwent invasive prenatal diagnosis and the samples were analyzed by aCGH. In total, six different aCGH platforms were studied; four different resolutions (0.18 Mb, 0.5Mb, 1Mb, 2Mb) and two platform designs (whole genome [WG] and targeted). The results of these designs were compared based on their diagnostic yield and VOUS rate. The performance of the different designs was further analyzed according to indication for invasive testing. Results The diagnostic yield of CNVs increased with increasing level of analysis. The detection rates of clinically significant chromosomal abnormalities were almost the same across our targeted array designs; 7.2% with 0.18 Mb backbone/ 0.05 Mb, vs. 7.1% with 0.5 Mb backbone/ 0.05 Mb (p>0.05). However, a significant difference in the rate of VOUS was observed; 9.4% with 0.18 Mb backbone/ 0.05 Mb vs. 6% with 0.5 Mb backbone/ 0.05 Mb (p<0.001). After analyzing the results across different indications for testing, we found that the application of non-targeted platform designs and lower levels of resolution analysis (such as 1Mb WG, or 0.5MbL/1MbG WG) would offer similar diagnostic yield in most cases with major congenital anomalies, with lower VOUS rates. However, the sample size for many indication groups was too small to extract robust associations. Conclusion It appears that the targeted array platform with 0.5Mb backbone resolution and 0.05Mb on targeted gene-rich regions is optimal for routine CMA use in prenatal diagnosis. It may be beneficial to individualize the minimum resolution in specific referral indications, as the indications for invasive prenatal testing may be quite heterogeneous.

引言 本研究旨在评估用于产前检测的阵列比较基因组杂交（array comparative genomic hybridization, aCGH）的最优分辨率，以期在最大化诊断检出率与最小化意义未明变异（variants of unknown significance, VOUS）检出率之间达成平衡。
方法 本研究为前瞻性研究，纳入2336例接受侵入性产前诊断的胎儿样本数据，所有样本均通过aCGH进行分析。本研究共评估6种不同的aCGH平台：4种分辨率梯度（0.18 Mb、0.5 Mb、1 Mb、2 Mb）以及2种平台设计类型（全基因组[WG]与靶向型）。基于诊断检出率与VOUS检出率对不同平台设计的结果进行对比，并进一步根据侵入性产前检测的临床指征分层分析不同设计的性能表现。
结果 拷贝数变异（copy number variations, CNVs）的诊断检出率随分析分辨率提升而升高。临床意义明确的染色体异常检出率在不同靶向型阵列平台设计中基本一致：0.18 Mb骨架/0.05 Mb分辨率组为7.2%，0.5 Mb骨架/0.05 Mb分辨率组为7.1%（p>0.05）。但VOUS检出率存在显著统计学差异：0.18 Mb骨架/0.05 Mb分辨率组为9.4%，0.5 Mb骨架/0.05 Mb分辨率组为6%（p<0.001）。分层分析不同检测指征的结果后发现，对于多数严重先天性畸形病例，采用非靶向型平台设计与较低分辨率分析（如1 Mb全基因组平台，或0.5 Mb连锁/1 Mb基因座全基因组平台）可获得相近的诊断检出率，且VOUS检出率更低。不过多数检测指征组的样本量较小，难以得出稳健的统计学关联结论。
结论 兼具0.5 Mb骨架分辨率与靶向基因富集区域0.05 Mb分辨率的靶向型阵列平台，是产前诊断中常规使用染色体微阵列分析（chromosomal microarray analysis, CMA）的最优选择。由于侵入性产前检测的指征具有高度异质性，针对特定转诊指征个体化设置最低分辨率或可带来额外获益。