Actionability of commercial laboratory sequencing panels for newborn screening and the importance of transparency for parental decision-making

Abstract

Background: Newborn screening aims to identify individual patients who could benefit from early management, treatment, and/or surveillance practices. As sequencing technologies have progressed and we move into the era of precision medicine, genomic sequencing has been introduced to this area with the hopes of detecting variants related to a vastly expanded number of conditions. Though implementation of genomic sequencing for newborn screening in public health and clinical settings is limited, commercial laboratories have begun to offer genomic screening panels for neonates.

Methods: We examined genes listed on four commercial laboratory genomic screening panels for neonates and assessed their clinical actionability using an established age-based semi-quantitative metric to categorize them. We identified genes that were included on multiple panels or distinct between panels.

Results: Three hundred and nine genes appeared on one or more commercial panels: 74 (23.9%) genes were included in all four commercial panels, 45 (14.6%) were on only three panels, 76 (24.6%) were on only two panels, and 114 (36.9%) genes were listed on only one of the four panels. Eighty-two genes (26.5%) listed on one or more panels were assessed by our method to be inappropriate for newborn screening and to require additional parental decision-making. Conversely, 249 genes that we previously identified as being highly actionable were not listed on any of the four commercial laboratory genomic screening panels.

Conclusions: Commercial neonatal genomic screening panels have heterogeneous content and may contain some conditions with lower actionability than would be expected for public health newborn screening; conversely, some conditions with higher actionability may be omitted from these panels. The lack of transparency about how conditions are selected suggests a need for greater detail about panel content in order for parents to make informed decisions. The nuanced activity of gene list selection for genomic screening should be iteratively refined with evidence-based approaches to provide maximal benefit and minimal harm to newborns.

Fig. 1

Comparison of genes on commercial laboratory NBS panels. Venn diagram summarizing all of the overlapping or distinct genes from BabyGenes Supplemental Newborn Screening, Sema4 Natalis, Fulgent Newborn Genetic Analysis NGS Panel, and PerkinElmer Expanded Newborn Screening and Gene Sequencing Panel using a free Ghent University Bioinformatics and Evolutionary Genomics software tool [23]. A single gene entry for HBA1/HBA2 on Sema4 Natalis was split for comparison, and the total number of genes on the panel changed from 166 to 167

Fig. 2

Comparison of ASQM total scores for genes on commercial laboratory NBS panels. Box and whisker plot of ASQM score distributions for 215 genes on PerkinElmer’s panel, 224 scores on Fulgent’s panel, 167 genes on Sema4’s panel, and 106 genes on BabyGenes’ panel. All genes that were scored using the ASQM were used for comparison, regardless of category. The gray box represents the area where genes-disease pairs scoring 9, 10, or 11 could not be automatically categorized and necessitated further discussion by the scoring committee prior to final categorization. Distributions varied significantly across all panels by Kruskal–Wallis test (p < 0.0001), and paired Mann–Whitney U tests identified significant differences between PerkinElmer and Fulgent panels' ASQM gene-disease pair scores (***p = 0.0001) and between PerkinElmer and Sema4 panels' ASQM gene-disease pair scores (****p < 0.0001)

Fig. 3

Comparison of ASQM scores of overlapping or distinct genes on commercial panels. Box and whisker plot of ASQM scores of genes that appeared on all four commercial laboratory NBS panels, only three commercial panels, only two commercial panels, and only one commercial panel. The gray box represents the area where genes-disease pairs scoring 9, 10, or 11 could not be automatically categorized and necessitated further discussion by the scoring committee prior to final categorization. Distributions varied significantly across all panels by Kruskal–Wallis test (p < 0.0001), and paired Mann–Whitney U tests identified significant differences in three out of four comparisons between scores of genes on all four panels versus only two panels (*p = 0.429), scores of genes on all four panels versus only one panel (****p < 0.0001), scores of genes on three panels versus only one panel (****p < 0.0001), and scores of genes on two panels versus only one panel (***p = 0.0001)

Fig. 4

Examples of different actionability scores based on defined phenotype. Bar graph of ASQM scores for eight genes with conditions that were separately reviewed by the scoring committee depicting individual criterion scores for the severity of disease, likelihood of presentation or penetrance, efficacy and acceptability of intervention, and knowledge base. The area between the vertical dashed lines represents gene-disease pairs scoring 9, 10, and 11 that underwent an additional round of discussion and review prior to categorization. Genes associated with early-onset conditions that scored 8 or below were automatically placed in ASQM Category 2 and pairs scoring 12 or greater were placed in ASQM Category 1. Assigned ASQM categories are labeled at the end of each bar for every gene-disease pair: NBS Cat 1 for Category 1 with early-onset and high actionability, Cat 2 for Category 2 with early-onset but lower or no actionability, and Cat 4 for Category 4 with adult-onset and lower or no actionability