## Introduction

A growing body of evidence is providing support for the use of clinical genomic sequencing early in the diagnostic trajectory of patients with rare genetic diseases. Early genomic testing of children has high diagnostic and clinical utility1,2 and is cost-effective compared with the cost of the current standard diagnostic pathway.3,4,5 However, a number of uncertainties remain around the use of genomic sequencing in clinical care.

First, the longer-term impact of diagnostic genomic results on patient management, health-care service provision and family decision-making is not known. Rare diseases are commonly perceived either as “untreatable” or requiring expensive therapies that may impact on constrained health-care systems. Anecdotal evidence from families highlights the benefits of putting an end to the “diagnostic odyssey,” targeting treatment and enabling reproductive planning,6 but to date studies have predominantly looked at the immediate impact on care,1,2,4,7,8,9,10 and only one study has examined downstream patient care and cost consequences over a period of a year.11

Second, the impact of “uninformative” results on further investigation and health-care activity is poorly understood. An uninformative result could either lead to further investigation and health-care activity focused on achieving a diagnosis or, alternatively, diminished diagnostic efforts. Stored genomic data provide unique opportunities for reanalysis in the light of new gene discoveries, improved bioinformatics techniques, and changes in clinical circumstances. Guidelines propose that storage and reanalysis is good practice,12 and data from several studies indicate diagnostic effectiveness.13,14,15 However, laboratory practices vary from not offering reanalysis to automated reanalysis at no additional charge to reanalysis on clinician request at defined time intervals and at a cost.16 The cost-effectiveness and clinical impact of reanalysis compared with ongoing standard investigation (e.g., candidate gene testing) have not been determined, but are necessary to inform clinical and laboratory policies.

### Cost–utility analysis of the change in patient management and reproductive planning

Our analysis demonstrates that genomic testing followed by changes in management as a result of WES diagnosis resulted in a cost saving of AU$1577.88 (95% CI: –AU$205,449.67 to $19,780.25) per additional QALY gained over standard diagnostic care (Table 3). #### Cost–utility analysis account for changed management, cascade testing, and reproductive planning in patients and first-degree relatives Accounting for cascade testing, reproductive service utilization, and outcomes in first-degree relatives, in addition to the impact of altered management in the patients resulted in an incremental cost of AU$8118.70 (95% CI: AU$1961.84 to$38,943.90) per QALY gained (Table 3).

Two cost-effectiveness planes of genomic testing resulting in diagnosis are presented in Supplementary Fig. 2: one considering only the change in clinical management as a result of diagnosis and the other considering all health benefits of having a diagnosis including changes in management, cascade testing, and reproductive outcomes. When only the change in clinical management was considered, achieving a diagnosis as a result of genomic testing was dominant (less cost and an increase in QALY) in 48.5% of the simulations. After accounting for cascade testing and reproductive planning/outcomes in addition to the impact of altered management as a result of genomic testing, 97.8% of the simulation results fall into the top-right quadrant, which suggests that there is greater additional cost per QALY gained when these costs are considered.

## Discussion

Genetic conditions are the leading cause of death in infants, and children with genetic conditions are typically high users of tertiary pediatric care services. Genomic testing is a transformative technology that is expected to have a major effect on tertiary pediatric care. However, evidence is necessary to support the case to funders for its implementation, and for services to guide clinical and laboratory decision-making. Our study provides evidence of the broader long-term impacts of this technology on a range of clinical and health economic parameters. In addition, we provide evidence to support policy development and service provision in an area of highly variable clinical and laboratory practice: the reanalysis of stored genomic data.

For our diagnosed patients, the promise of personalized care as a result of genomic sequencing was realized for a large proportion of the cohort. Importantly, the treatments and surveillance initiated were generally of a modest cost and did not result in increased hospital service use. Overall, the changes in clinical management resulting from genomic sequencing across the entire cohort of 80 were found to result in a cost saving of $1578 per QALY gained at one-year follow-up. Unexpectedly, we found that uninformative WES results had value in supporting the diagnosis of non-Mendelian conditions in 7 patients (8.75% of the cohort). The uninformative genomic sequencing result, combined with the clinical course, enabled an alternate, non-genetic diagnosis to be made, releasing the child from further investigations and unnecessary treatments. Early genomic sequencing offers the greatest opportunities to influence outcomes and reduce the costs of other investigations by making timely diagnoses.1,2,8,10 One of the challenges of initiating genomic testing early in the diagnostic trajectory is that clinical presentations are often poorly differentiated and incomplete, making it difficult to distinguish genetic from non-genetic etiologies. Performing genomic testing in a small number of patients whose subsequent course suggests a non-monogenic etiology may be an inherent risk of early test initiation, but the uninformative results can still be beneficial in contributing to diagnostic certainty, albeit for non-monogenic conditions. We found that reanalysis of existing WES data was a highly effective diagnostic strategy in patients with uninformative results but ongoing suspicion of a monogenic condition. The diagnostic yield we obtained (14%) is consistent with 3 recent studies totaling over 1000 reanalyzed cases, reporting additional diagnostic yields ranging between 10 and 21% 20 to 36 months after initial analysis.13,14,15 The majority of reanalysis diagnoses have been due to the identification of variants in genes associated with disease following the initial WES analysis. The value of genomic data reanalysis is increasingly acknowledged, but a major challenge for implementation is the development of sustainable models for the laboratory and clinical delivery of this labor-intensive service. The coupling of reanalysis with follow-up clinical genetics appointments at intervals of one to two years provides the opportunity to re-examine the genomic data not only in the context of evolving genetics knowledge, but also in the context of evolving patient phenotype information and family priorities. This model was more cost-effective than undertaking reanalysis every 6 months, with a cost saving of AU$1058 per additional diagnosis compared with usual diagnostic care.

For the infants’ families, the major impact of early genomic diagnosis observed in this cohort is the restoration of parental reproductive confidence. As the study participants underwent WES in the first two years of life, all parents were of reproductive age. A molecular diagnosis in the child was strongly associated with increased use of reproductive genetic services and with subsequent pregnancy, with eight more ongoing pregnancies in the families where the child had a diagnosis compared with those without. All pregnant couples at increased risk of recurrence utilized either PGD or PND, with one couple continuing an affected pregnancy because treatment for the condition was available. The only termination of pregnancy occurred in the “undiagnosed” group, based on uncertainty regarding the recurrence risk. This pattern reflects previous observations from the 1980s of couples postponing and avoiding having more children when PND for a genetic condition was not available.23 Including the cost of cascade testing and reproductive services use resulted in an additional cost of AU\$8118 per QALY gained in this cohort over the follow-up period. However, the lifetime value of the additional eight children to families, communities, and Australian society is expected to far exceed the cost of providing reproductive genetic services, and indeed the cost of providing genomic sequencing in the cohort.

This study is limited by its small size, and bootstrapping was conducted to provide an estimate of certainty. The results are not generalizable to patients undergoing genomic testing for different indications or using different service delivery models, as diagnostic yields and the clinical impact of results are likely to be different.24 Collection of clinical utility and cost-effectiveness data in larger clinically ascertained cohorts followed up over extended periods of time will be needed to clarify the lifetime impacts of genomic testing. As clinically ascertained cohorts tend to be small, we encourage standardized data collection around clinical utility parameters and health-care costs across diverse health-care settings to facilitate data analysis and comparison.

Our work provides evidence of the longer-term outcomes of genomic sequencing for patients, their families, and health-care costs. This follow-up study demonstrates that this transformative technology results in improved patient management without a major increase in health-care costs, and in restoration of reproductive confidence in families. The results further strengthen the case for the early use of genomic testing in the diagnostic trajectory. In addition, we show that reanalysis at 18 months—around the time of clinical review—is a cost-effective model for the storage and re-examination of genomic data in clinical service delivery. As health systems worldwide consider the implementation of genomic sequencing in routine clinical practice, these results provide valuable evidence to guide funders and services in the cost-effective use of this technology.