Exome Sequencing for Prenatal Detection of Genetic Abnormalities in Fetal Ultrasound Anomalies: An Economic Evaluation

Introduction In light of the prospective Prenatal Assessment of Genomes and Exomes (PAGE) study, this paper aimed to determine the additional costs of using exome sequencing (ES) alongside or in place of chromosomal microarray (CMA) in a fetus with an identified congenital anomaly. Methods A decision tree was populated using data from a prospective cohort of women undergoing invasive diagnostic testing. Four testing strategies were evaluated: CMA, ES, CMA followed by ES (“stepwise”); CMA and ES combined. Results When ES is priced at GBP 2,100 (EUR 2,407/USD 2,694), performing ES alone prenatally would cost a further GBP 31,410 (EUR 36,001/USD 40,289) per additional genetic diagnosis, whereas the stepwise would cost a further GBP 24,657 (EUR 28,261/USD 31,627) per additional genetic diagnosis. When ES is priced at GBP 966 (EUR 1,107/USD 1,239), performing ES alone prenatally would cost a further GBP 11,532 (EUR 13,217/USD 14,792) per additional genetic diagnosis, whereas the stepwise would cost a further additional GBP 11,639 (EUR 13,340/USD 14,929) per additional genetic diagnosis. The sub-group analysis suggests that performing stepwise on cases indicative of multiple anomalies at ultrasound scan (USS) compared to cases indicative of a single anomaly, is more cost-effective compared to using ES alone. Discussion/Conclusion Performing ES alongside CMA is more cost-effective than ES alone, which can potentially lead to improvements in pregnancy management. The direct effects of test results on pregnancy outcomes were not examined; therefore, further research is recommended to examine changes on the projected incremental cost-effectiveness ratios.

each hour of patient contact) and £137 per hour for a medical consultant.

Post-Partum Care
The cost of post-partum care varied depending on the unit the baby was admitted to and the amount of time spent.

Rates
Table S12-S13 presents the findings for A2. Briefly, 13.60% of all cases were identified to have an anomaly. CMA alone identified approximately 26% of all possible cases and ES alone identified approximately 82% of all possible cases. The stepwise and the combined strategies identified all possible cases.

Ratio (ICER)
The findings presented in Table S12-S13 show that compared to the cost of CMA, ES alone was found to be the least costly strategy.
Even so, the stepwise was able to identify all of the possible anomalies at an incremental cost of £266 (€305/$341), when compared to ES alone. The ICER derived for the stepwise is therefore lower than the ICER derived for ES alone. This implies that it is more costeffective to employ the stepwise, as opposed to ES alone, in order to identify an additional genetic diagnosis. In this case, the dominant strategy is the stepwise. This can be seen graphically in Figure S3.

Analysis
In the limited deterministic sensitivity analysis (DSA), five additional scenarios were investigated following the base case analysis in A2 (Table S12-S13). Each scenario tested a reduction in the cost of ES of up to 50% by 10% decrements. The pattern of dominance remained consistent with the base case analysis in all but one scenario. When the cost of ES was reduced by 50% £1050[€1,203/$1,347]), the pattern of dominance changes, such that ES was no longer a dominated strategy. This is clear, as the ICER for ES alone is lower than the ICER for the stepwise. This therefore implies that, based on the values of the ICER's, ES alone can be considered as a suitable strategy. However, if the WTP threshold exceeds £10581(€12127/$13,572), the stepwise will still remain the preferred strategy. This is because the stepwise yields more, in terms of effectiveness.

Curves
The PSA was undertaken to obtain the differences in the costs and effectiveness between each strategy in order to produce an incremental cost-effectiveness plane. Figure S5 shows the mean incremental costs and incremental effectiveness between CMA alone and ES alone for A2. The graph shows a large amount of parameter uncertainty, as the mean incremental costs and incremental effectiveness falls in the north east and the north-west quadrant of the plane. This indicates that ES is certain to be more costly than CMA, but there is a small probability, consistent with the data available, that ES is also less effective than CMA. Figure S6 shows the mean incremental costs and incremental effectiveness between CMA alone and the stepwise. The graph shows some parameter uncertainty, although all points fall within the north east region of the plane. This implies that in all cases the stepwise will identify more abnormalities than CMA alone, but at an additional cost. Figure S7 shows the mean incremental costs and incremental effectiveness between ES alone and the stepwise. There is a large amount of parameter uncertainty despite all points falling within the north east region of the plane. This implies that in all cases the stepwise will identify more abnormalities than ES alone, but at an additional cost.

Effectiveness
Acceptability Figure S8 presents the CEAC for A2, when the cost of ES is (€2,407/$2,694). At a WTP of £20,000 (€22,923/$25,654) the probability that the stepwise is cost-effective is approximately 36% and the probability that CMA alone is cost-effective is 56%. At a WTP of

Curve (CEAC)
£30,000 (€34,385/$38,481) the probability that the stepwise cost-effective is 90% and the probability that CMA alone is cost-effective is 6%. Figure S9 presents the CEAC for A2, when the cost of ES is £1,050 (€ 1,203/$1,347). At a WTP of £20,000 (€22,923/$25,654) the probability that the stepwise is cost-effective is approximately 79% and the probability that CMA alone is cost-effective is less than 1%.
At a WTP of £30,000 (€34,385/$38,481) the probability that the stepwise cost-effective is 94% and the probability that CMA alone is cost-effective is 0%.   Table S15.

Rates
The detection rates varied between strategies and subgroups (see Table S14). CMA detected between 2-10% of all phenotypes identified using USS. ES alone detected between 3-15% of all phenotypes identified using USS. The stepwise and the combined detected between 3-19% of all phenotypes identified using USS. It should be noted that all test strategies had a detection rate of approximately 3% for the brain subgroup. This is because both CMA and ES identified the same positive cases within the sample.

Ratio (ICER)
Table S14 presents the ICER's for each subgroup. The ICER for ES alone and the combined approach exceeded the stepwise for all subgroups but the Brain. This means that compared to CMA, the stepwise is the dominant strategy. The results suggest that if the WTP is at least £36,900 (€42,290/$47,330), the stepwise will be the preferred option for all subgroups.
The findings associated to the Brain subgroup show the ICER's to be undefined. This is because there was no additional effectiveness per strategy, only additional costs. The findings therefore suggest that CMA will remain as the preferred strategy should the USS be indicative of an anomaly associated with the Brain, as CMA absolutely dominates all other strategies. Even so, the sample size for this subgroup was only 26. This might have been too small to show a true depiction of the cost-effectiveness of detecting an anomaly associated with the Brain.
The ICER's of each subgroup differed, meaning the necessary WTP for a strategy to be cost-effective varied. This was due to the different incremental costs and incremental effectiveness for each subgroup. It might therefore be more cost-effective to undertake the stepwise on selected subgroups, where the incremental costs are lower and the effectiveness is significantly greater. For example, cases with multiple anomalies detected by USS will likely be preferred over cases with a single anomaly detected, as the incremental effectiveness is greater and the incremental costs are lower, which in turn, produces a lower ICER.
Furthermore, if the WTP threshold is £30,000 (€34,385/$38,481), based on the ICERS alone, employing the stepwise would not be recommended if a single anomaly, abdominal, gastro, or brain related anomaly is identified by USS. This is because the ICER's for these sub-groups exceed the maximum threshold.

Curve (CEAC)
A PSA was undertaken to examine the level of uncertainty surrounding the parameters used for each phenotype subgroup. The PSA enabled the model to derive a CEAC for each group, which states the probability each strategy will be cost-effective. Figure S10 presents the CEAC for the multiple anomaly subgroup. The figure shows that at a WTP of £20,000 (€22,923/$25,654), the probability that CMA is cost-effective is 29% and the probability that the stepwise is cost-effective is 56%. At a WTP of £30,000 (€34,385/$38,481), Figure S10 illustrates that the probability that CMA is cost-effective is 7%, the probability that ES alone is costeffective is 10%, and the probability that stepwise is cost-effective is 83%. Figure S11 presents the CEAC for the multiple anomaly subgroup. The figure shows that at a WTP of £20,000 (€22,923/$25,654), the probability that CMA is cost-effective is 95%, whereas the probability that stepwise is cost-effective is 4%. At a WTP of £30,000 (€34,385/$38,481), the figure indicates that the probability that CMA is cost-effective is 64%, whereas the probability that stepwise is cost-effective is 35%. If the WTP is at least £35,000 (€40,115/$44,894), the probability that stepwise is cost-effective exceeds the probability that CMA is cost-effective. The findings suggest that the stepwise is likely to be more cost-effective when the USS is indicative of multiple anomalies compared to the single anomaly subgroup, as the probability is greater amongst the group. Nonetheless, the sample size of the multiple anomaly subgroup was much lower compared to the single anomaly subgroup. Therefore, the test results of the samples may have subsequently influenced the PSA findings.

Phenotype: Skeletal/Limb/Spinal
Figure S12 presents the CEAC for the subgroup associated with skeletal, limb and spinal anomalies. At a WTP of £20,000 (€22,923/$25,654) the probability that the stepwise is cost-effective is approximately 15% and the probability that CMA alone is costeffective is 74%. At a WTP of £30,000 (€34,385/$38,481) the probability that the stepwise cost-effective is 32% and the probability that CMA alone is cost-effective is 56%. If the WTP is greater than £40,000 (€45,846 / $51,308), the probability that the stepwise is costeffective will exceed the probability that CMA alone is cost-effective.

Phenotype: Cardiac
Figure S13 presents the CEAC for the subgroup associated with cardiac anomalies. At a WTP of £20,000 (€22,923/$25,654) the probability that the stepwise and CMA alone are cost-effective is approximately 45%. At a WTP of £30,000 (€34,385/$38,481) the probability that the stepwise cost-effective is 76% and the probability that CMA alone is cost-effective is 17%. As the WTP increases, the probability that the stepwise is cost-effective becomes closer to 100%.

Phenotype: Abdominal/Gastro
Figure S14 presents the CEAC for the subgroup associated with abdominal and gastro anomalies. At a WTP of £20,000 (€22,923/$25,654) the probability that the stepwise is cost-effective is approximately 20% and the probability that CMA alone is cost-effective is 80%. At a WTP of £30,000 (€34,385/$38,481) the probability that the stepwise cost-effective is 32% and the probability that CMA alone is costeffective is 67%. If the WTP is greater than £50,000 (€57,308/ $64,135), the probability that the stepwise is cost-effective will exceed the probability that CMA alone is cost-effective.

Phenotype: Nuchal Translucency
Figure S15 presents the CEAC for the subgroup associated with nuchal translucency anomalies. At a WTP of £20,000 (€22,923/$25,654) the probability that the stepwise is cost-effective is 20% and the probability that CMA alone is cost-effective is 61%. At a WTP of £30,000 (€34,385/$38,481) the probability that the stepwise cost-effective is 40% and the probability that CMA alone is cost-effective is 35%.

Phenotype: Brain
Figure S16 presents the CEAC for the subgroup associated with brain anomalies. At a WTP of £20,000 (€22,923/$25,654) and £30,000 (€34,385/$38,481), the probability that the stepwise is cost-effective is 0% and the probability that CMA alone is cost-effective is 100%.
There is no threshold whereby the probability of cost-effectiveness changes. This is because the only difference between each strategy was the cost.

Figure S1. Model Decision Tree
The decision tree was built for the purpose of the study.  Stepwise CMA and ES