A data-driven evaluation of the size and content of expanded carrier screening panels

Abstract

Purpose: The American College of Obstetricians and Gynecologists (ACOG) proposed seven criteria for expanded carrier screening (ECS) panel design. To ensure that screening for a condition is sufficiently sensitive to identify carriers and reduce residual risk of noncarriers, one criterion requires a per-condition carrier rate greater than 1 in 100. However, it is unestablished whether this threshold corresponds with a loss in clinical detection. The impact of the proposed panel design criteria on at-risk couple detection warrants data-driven evaluation.

Methods: Carrier rates and at-risk couple rates were calculated in 56,281 patients who underwent a 176-condition ECS and were evaluated for panels satisfying various criteria. Condition-specific clinical detection rates were estimated via simulation.

Results: Different interpretations of the 1-in-100 criterion have variable impact: a compliant panel would include between 3 and 38 conditions, identify 11-81% fewer at-risk couples, and detect 36-79% fewer carriers than a 176-condition panel. If the carrier rate threshold must be exceeded in all ethnicities, ECS panels would lack prevalent conditions like cystic fibrosis. Simulations suggest that the clinical detection rate remains >84% for conditions with carrier rates as low as 1 in 1000.

Conclusion: The 1-in-100 criterion limits at-risk couple detection and should be reconsidered.

Keywords: clinical detection rate; clinical guidelines; clinical utility; expanded carrier screening; genetic testing.

Fig. 1

Impact of American College of Obstetricians and Gynecologists (ACOG) guidelines on panel size, carrier rates, and at-risk couple rates. We consider four panels: the full 176-condition panel, the subset of conditions that meet the first six ACOG criteria (excluding the 1-in-100 criteria), the subset of conditions that meet the ethnicity-specific requirements in ACOG guidelines 691 (ref. ), and a panel that includes only cystic fibrosis (CF) and spinal muscular atrophy (SMA). (a) The number of diseases that meet criteria. (b) U.S.-weighted panel carrier rates. (c) US-weighted at-risk couple rates. (d) The number of observed at-risk couples that would be identified (green) or omitted (red) by the indicated panel. Specific conditions of observed at-risk couples are shown in Figure S1. Horizontal lines show respective numbers for the 176-condition panel.

Fig. 2

Carrier rates vary widely by ethnicity. (a) Self-reported ethnicity-specific carrier rates for the five conditions with the highest US-weighted carrier rates in the 176-condition panel for which an ethnicity-specific carrier rate is less than 1 in 100 in at least one ethnicity. Ethnicities with carrier rates below a 1-in-100 carrier rate are denoted with Xs; ethnicities with carrier rates above the 1-in-100 carrier rate are denoted with circles. (b) The relative difference in carrier rates between the highest and lowest ethnicity-specific carrier rates for each condition in (a). CFTR cystic fibrosis, GJB2 GJB2-related DFNB1 nonsyndromic hearing loss and deafness, HBB Hb β chain-related hemoglobinopathy, PAH phenylalanine hydroxylase deficiency, DHCR7 Smith–Lemli–Opitz syndrome. AF African, AJ Ashkenazi Jewish, CO Caucasian/other, EA East Asian, HIS Hispanic, ME Middle Eastern, NE Northern European, SA South Asian, SE Southern European, SEA Southeast Asian.

Fig. 3

Impact of different 1-in-100 carrier rate threshold definitions on carrier rates and at-risk couple rates. Included conditions met the first six American College of Obstetricians and Gynecologists (ACOG) criteria (Fig. 1) and different definitions of the 1-in-100 carrier rate ACOG criteria (x-axis). Shading of green bars denotes different carrier rate thresholds for X-linked conditions: a carrier rate threshold of 1 in 100 (light green) and a carrier rate threshold of 1 in 10,000 (dark green). (a) The number of diseases that meet criteria. (b) US-weighted panel carrier rates. (c) US-weighted at-risk couple rates. (d) The number of observed at-risk couples that would be identified by the panel subset. Specific conditions of observed at-risk couples are shown in Figure S1. Horizontal lines show respective numbers for the 176-condition panel.

Fig. 4

The relationship between panel size, panel carrier rate, and at-risk couple rate. (a) The panel carrier rate (green) and panel at-risk couple rate (purple) are plotted as a function of the carrier rate threshold and panel size (x-axis). Conditions are ordered from most to least prevalent based on US-weighted carrier rate. Because X-linked and autosomal recessive (AR) conditions are inherited differently, carrier rates for X-linked conditions were transformed to corresponding AR carrier rates based on their at-risk couple rates. For example, for an X-linked condition with a carrier rate and at-risk couple rate of 1 in 10,000, the carrier rate would be transformed to 1 in 100. (b) Ratio of marginal panel carrier rate to marginal at-risk couple rate as a function of the panel size determined by carrier rate threshold (x-axis). The panel carrier rate was calculated from condition-specific US-weighted carrier rates. At-risk couple rates were calculated as the square of the US-weighted carrier rate.

Fig. 5

Estimation of clinical detection rate. (a, b) A model schematic showing how clinical detection rate is estimated for a hypothetical autosomal recessive (AR) condition. We assume a condition has six pathogenic variants and a carrier rate of 1 in 10,000, resulting in a prevalence of 1 in 400,000,000. (a) Assumed number of pathogenic variants, including both observed variants (purple, variants denoted with O) and a minority of unobserved variants (green, variants denoted with U). (b) Simulations of the expected number of reported cases (assuming all cases will be reported). The estimated clinical detection rate is defined as the sum of the variant frequencies for variants that can be classified as pathogenic, determined by three or more estimated case reports (shown in blue). Variants whose pathogenicity cannot be determined are shown in red or have no reported cases. (c, d) Estimated clinical detection rates for (c) AR conditions and (d) X-linked conditions on the 176-condition panel. US-weighted carrier rates and estimated clinical detection rate for each condition are shown when three reported cases are needed to determine pathogenicity for each variant. Dots show median estimated clinical detection rate from 10,000 iterations per condition, and lines show corresponding 95% confidence intervals. We excluded X-linked severe combined immunodeficiency and X-linked ornithine transcarbamylase deficiency from this analysis because we did not observe any carriers of these X-linked conditions during the study period (see Supplementary Text S3). Conditions and corresponding clinical detection rate estimates are provided in Table S4.