Methods for Reanalysis of Exome Sequencing Emerging

Clinical whole-exome sequencing (WES) offers diagnosis for approximately 30 percent of patients, but it has exponentially increased the number of variants identified—many of which are of unknown clinical significance. The continuous evolution of genomic knowledge necessitates reanalysis of exome sequencing results over time.

Despite recognition that reanalysis with the latest variant databases may enable increased diagnostic yield, reanalysis poses many practical challenges including workflow and time constraints, questions about the effective frequency of reanlaysis, and issues of patient follow-up. There are currently no standards for reanalysis. DTET examined reports from several leading institutions that presented their reanalysis findings at the American College of Medical Genetics and Genomics Annual Clinical Meeting (Charlotte, N.C.; April 10-14). These case reports show some emerging solutions for the challenge of reanalysis of exome sequences.

Automated Annotation for Reanalysis of WES Data
Automated bioinformatics pipeline may enable providers to efficiently and effectively characterize annotation changes, supplement existing genomic results with these changes, and filter the results for changes of interest, according to a presentation by Charu Kaiwar, M.D., from the Mayo Clinic.

Kaiwar reported on a process that enters all negative exome cases into a pipeline capable of providing periodic, time stamped differential outputs. The so-called Automated Reanalysis Pipeline performs systematic reanalysis of clinical exomes against continually updated knowledge sources. The annotation changes are traceable back to their original downloaded source. The Mayo researchers report the system is scalable and can be integrated into an existing bioinformatics network.

The final output of the pipeline contains variants flagged using user-defined criteria based on gene-variant-disease associations in public databases, population frequency, and predicted impact of variant on protein function.

The researchers reported on a pilot of 29 previously sequenced patients in the Mayo Clinic Diagnostic Odyssey cohort. The updated re-annotation was performed at an average of 12.5 months after initial evaluation. A total of 87 variants in ClinVar and 599 variants in the Online Mendelian Inheritance in Man (OMIM) filters were flagged for manual screening against phenotypes. Two variants—both single hits in autosomal recessive diseases—were relevant to patient phenotype. The researchers say further evaluation of the genes for copy number variants and other abnormalities are underway.

Using Electronic Health Records for Delivering Variant Updates
Changes in the MedSeq genome interpretation pipeline over the course of the study (July 2013 to February 2015) required reanalysis of the first 100 MedSeq genomes (50 healthy individuals and 50 with a cardiomyopathy diagnosis) using the updated pipeline (August 2015).

While the initial analysis yielded 1,201 variants, including 189 novel predicted null variants, only 230 unique variants (85 novel null and 59 previously reported) met criteria for return of results to subjects. Reanalysis enabled an additional 22 participants to receive updated results. These updates were delivered directly through the electronic health record through system delivered alerts and included reclassification of variants that were previously reported (n=10), newly identified variants as a result of reanalysis (n=9), and three cases of both.

"The findings in this study demonstrate the ability to comprehensively and efficiently interpret human genomes for both screening and diagnosis as well as implement a model for delivering regular variant updates through the electronic health record," concluded study presenter Kalotina Machini, Ph.D., from Brigham and Women's Hospital in Boston, Mass.

Reanalysis for Rare Disease Identification
The Undiagnosed Diseases Network (UDN) in a National Institutes of Health-funded research program in which patients with previously undiagnosed disease receive access to a cross-disciplinary network of research and clinical investigators.

Clinical evaluation and phenotyping, including genetic analysis with WES, is conducted by the HudsonAlpha Institute of Biotechnology (Huntsville, Alabama). Of the initial 270 cases assigned for genetic analysis, initial analysis included return of 144 clinical reports with primary findings for 127 variants connected to the patients' phenotypes.

Given additional discovery of gene-disease associations in the two years since the UDN began, HudsonAlpha began reanalysis of unsolved cases based on newly published findings both in literature and databases such as the Genome Aggregation Database and ClinVar. Additionally, new tools now enable scans of whole-genome sequencing data for structural variation and detection of repeat expansions.

Using novel secondary analysis methods, the institute reported identifying genomic variations, including mosaicism, repeat expansions, deletions, and translocations, likely to explain disease in an additional 17 patients. Reanalysis also yielded more than 50 candidate variants of interest for further investigation, that will possibly further advance the knowledge base for rare diseases, the authors say.

Reanalysis of Pediatric Patients
Laboratory-initiated reanalysis enables a time and cost effective, as well as a sensitive approach to increasing the diagnostic yield of WES, according to a presentation by Jill Murrell, Ph.D., from the Children's Hospital of Philadelphia (Pennsylvania).

Primary WES analyses was performed on 700 consecutive patients with clinical indications (neurodevelopmental disorders, multiple congenital anomalies, and immunodeficiencies) for WES testing. The overall diagnostic yield (associated with phenotype) was 30.1 percent.

The researchers developed an automated algorithm for reanalysis that utilized all of the genotype, phenotype, and test interpretation data generated during primary analysis and correlated patient data with terms generated from PubMed, OMIM, ClinVar, and Human Gene Mutation Databases. Identified variants were manually reevaluated.

Reanalysis was performed on 240 of the 443 nondiagnostic WES cases. Thirty-eight novel diagnoses were identified, resulting in a 15.8 percent increase in diagnostic yield. Novel disease genes, which had not been known to be associated with disease at the primary review were identified in 17 of the 38 cases. Reported candidate genes were reclassified as diagnostic in seven cases due to new evidence published in the literature. The researchers say that if reanalysis were performed monthly, approximately five variants for proband-only WES and 0.5 variants for trio WES would require evaluation.

Takeaway: Case reports presented by leading institutions highlight emerging solutions to address the challenge of reanalysis of exome sequences. While methodology varies by institution, all report an increase in diagnostic yield with reanalysis.