An Age-Based Framework for Evaluating Genome-Scale Sequencing Results in Newborn Screening

Abstract

Objective: To assess the performance of a standardized, age-based metric for scoring clinical actionability to evaluate conditions for inclusion in newborn screening and compare it with the results from other contemporary methods.

Study design: The North Carolina Newborn Exome Sequencing for Universal Screening study developed an age-based, semiquantitative metric to assess the clinical actionability of gene-disease pairs and classify them with respect to age of onset or timing of interventions. This categorization was compared with the gold standard Recommended Uniform Screening Panel and other methods to evaluate gene-disease pairs for newborn genomic sequencing.

Results: We assessed 822 gene-disease pairs, enriched for pediatric onset of disease and suspected actionability. Of these, 466 were classified as having childhood onset and high actionability, analogous to conditions selected for the Recommended Uniform Screening Panel core panel. Another 245 were classified as having childhood onset and low to no actionability, 25 were classified as having adult onset and high actionability, 19 were classified as having adult onset and low to no actionability, and 67 were excluded due to controversial evidence and/or prenatal onset.

Conclusions: This study describes a novel method to facilitate decisions about the potential use of genomic sequencing for newborn screening. These categories may assist parents and physicians in making informed decisions about the disclosure of results from voluntary genomic sequencing in children.

Keywords: ASQM; NC NEXUS; NGS-NBS; NSIGHT; actionability; decision-making; disclosure; genetics.

Figure 1.. Implementation of the Age-based Semi-Quantitative Metric (ASQM)

A.) The four NC NEXUS categories, separated on dimensions of actionability and age of onset/intervention. ASQM scores were assigned by a multi-disciplinary review committee. The gray box represents the gene-disease pairs with scores of 9–11, for which the ASQM score alone was not sufficient to make a classification of Category 1 or Category 2. Separation of categories 1 and 2 from categories 3 and 4 was made at age of onset/intervention of 18 because of the ethical principle of preserving future autonomy for children undergoing genetic testing,(32) but we also collected more granular data regarding ages of onset and intervention in anticipation that these might be useful for defining age-targeted screening panels (Table 2). B.) Examples of scores for conditions in Category 1 and Category 2. Scores for each criterion and the underlying rationale are provided. Left, PKU is a classic example of a high-scoring condition that is found on the RUSP and was placed in Category 1, pediatric conditions with high actionability. Right, Tay-Sachs is a severe childhood condition with no efficacious intervention and therefore a low-scoring condition placed in Category 2, pediatric conditions with low or no actionability.

Figure 2.. Nuances of categorical designation and scoring rubric

A.) Consensus categorization of gene-disease pairs scoring 9–11. After scoring conditions, the review committee identified that there was no single threshold that was adequate to delineate actionability among those gene-conditions having scores of 9, 10, or 11. For this reason, each gene-condition with childhood age of onset/intervention and score in this range was further discussed and a consensus decision among the committee was made regarding placement in category 1 or 2. As would be predicted, the majority of those with a score of 11 were assigned to Category 1 and the majority of those with a score of 9 were assigned to Category 2, illustrating that scores in this range reflect a grey zone that may not capture the entirety of the decision-making process. B.) Comparison of ASQM scores based on condition-specific rubrics. Each horizontal bar represents the total score for a gene-disease pair broken down into its scores for each ASQM criterion. This process can also be illustrated by the example of Primary Ciliary Dyskinesia (PCD) for which 35 genes have been published as causative. Specialty-area experts for clinical and molecular knowledge of PCD were consulted to develop our approach for scoring and categorization, and determined that presymptomatic intervention would be somewhat beneficial despite the paucity of evidence in the scientific literature. Although there is substantial locus heterogeneity for PCD, we considered the condition as a whole to have generally equivalent efficacy and acceptability of medical intervention regardless of the genetic cause, based on the highly similar clinical presentations. Therefore, scores varied most significantly in terms of the knowledge base establishing a particular gene as causative (Table 2). Genes associated with PCD for which the knowledge score was at least 1 received a total ASQM score ≥9. The decision was made for all of these gene-disease pairs to be placed in Category 1, acknowledging the potential for benefit through presymptomatic intervention.

Figure 3.. Comparison of ASQM scores across different lists of conditions

Box and whisker plots summarizing ASQM scores for A.) genes curated into NC NEXUS Categories 1–4, B.) 34 genes related to 25 primary RUSP conditions (32 were placed in Category 1 and 2 were assigned to Category 2) and 35 genes related to 22 secondary RUSP conditions (18 pairs were assigned to Category 1, 15 pairs were assigned to Category 2, and 2 pairs (ACAD8: isobutyryl-CoA dehydrogenase and AUH: 3-methylglutaconic aciduria, type I) were not assigned to any category and are only on the diagnostic list, and C.) gene-disease associations curated by the BabySeq Project into Categories A, B, and C. D.) Comparing gene-disease pairs categorized by both NC NEXUS and BabySeq, 244 gene-disease pairs were placed in both NC NEXUS category 1 and BabySeq category A, suggesting that this group may consist of conditions with definitive or strong clinical validity, high penetrance, and high actionability. Similarly, 48 gene-disease pairs were placed in NC NEXUS category 1 and BabySeq category B, indicating that they may have slightly lower knowledge base and/or reduced penetrance, but have strong consensus to be actionable in childhood. Together, these 292 conditions (indicated by dark shading) might represent a consensus list of gene-disease pairs for inclusion in NGS-NBS. In contrast, 73 gene-disease pairs were placed in NC NEXUS category 2 and BabySeq category A, suggesting that these disorders may be strongly associated with highly penetrant childhood diseases without substantial actionability. Similarly, 52 gene-disease pairs were placed in NC NEXUS category 2 and BabySeq category C, indicating that they are likely associated with lower strength of evidence, lower penetrance, and/or lower overall actionability. Together, these 125 conditions (indicated by medium shading) might comprise a group of gene-disease pairs that would not generally meet consensus criteria for newborn screening. The 16 gene-disease pairs that were placed in BabySeq category C and either NC NEXUS category 3 or category 4 (indicated by light shading) are consensus adult-onset conditions, with or without actionability. In contrast, 45 gene-disease pairs were categorized in NC NEXUS category 1 for return to all participants, but placed in BabySeq category C and not returned to any participants; 13 gene-disease pairs were placed in BabySeq category B (potentially actionable), but in NC NEXUS category 2 (low or no actionability); and 12 gene-disease pairs were placed in BabySeq categories A and B, but were assigned to NC NEXUS categories 3 and 4 due to adult onset, with or without actionability. Additionally, 11 gene-disease pairs were evaluated but were not assigned a NC NEXUS category whereas 3 of these gene-disease pairs were placed in BabySeq Category A and 8 were placed in Category C (Table 2).