Download Using Informatics to Increase Efficiencies

Using Informatics to Increase Efficiencies:
Predictive Analytics in the Laboratory
Nephi Walton M.D. M.S.
Washington University – St Louis Children’s Hospital
Genetic Testing
Contrary to popular belief ….
Whole exome testing does not really cover the whole
exome
Whole genome testing does not really cover the
whole genome
Sequencing is not an exact science but is actually a
probability that a base has a certain value
Different genetic testing companies can give you
different results
And lastly….
Genetic Testing
Genetic mutations do not usually give you super powers
Genetic Testing
Allows for long term management, surveillance, and
prognosis
Informs reproductive decisions and familial risk
Gives families an answer
Ends the “diagnostic odyssey”
Gives us information to create treatments catered to
individuals – Personalized/Precision Medicine
Studying Genetic
Testing in Epilepsy
Our diagnostic yield for epilepsy was low
We were sending unnecessary tests
We were using the wrong panel sent to the wrong
lab
I came across a case that embodied the whole
problem I was trying to address…
A Case Study
8 Month old baby
Brought in for seizures and unresponsiveness.
The child was otherwise healthy with no dysmorphic
features.
EEG – Focal Abnormality
MRI – Normal
Discharged home on medications
A Case Study
Child admitted back to hospital for back to back
seizures that were not responsive to medication. The
neurology team ordered:
CMA
Sequencing for genes related to Dravet Syndrome
Metabolic workup
Cost ~ $8,000.00
Results
CMA - Variant found in normal parent.
Dravet sequencing – Negative.
Metabolic Workup – Normal.
A Case Study
Patient Returns
Genetics Consulted to explain the CMA
Small childhood epilepsy panel sent
Cost ~ $6000.00
Results – Epilepsy Panel
Two variants of unknown significance
Parent has one variant- likely benign.
Next Visit
Whole Exome Sequencing sent
Cost $9000.00
WES – Results - 2 years
later
Pathogenic Mutation in the GABRA1 Gene
GABRA1
GAMMA-AMINOBUTYRIC ACID RECEPTOR
Gamma-aminobutyric acid (GABA) is the major
inhibitory neurotransmitter in the human brain
where it acts at GABA-A receptors, which are ligandgated chloride channels.
First described in Epilepsy in 1996
Analysis
Multiple Large Epilepsy Panels would have found this
mutation up front at a cost of less than $6000.00
End Result
~$17,000.00 in unnecessary testing
Diagnosis delayed for 2 Years
… this in on only one patient
This is not a unique case.
What went wrong here
CMA – Standard of care everyone with epilepsy gets
a CMA ordered first
Single Gene Testing in Epilepsy
Epilepsy panels have significant variation in genes
covered.
And thus starts my
quest…
Off to ACMG to collect information on genetic
testing offered, methods, and costs.
Data Mining our own data to pull results of our own
testing.
Building a large phenotype genotype database to
better assess diagnostic yields based on phenotype.
Assessing the factors that are important for choosing
genetic testing laboratories and tests
Lessons learned from
ACMG
Pricing varies widely, is very subjective, and usually not
advertised
Testing methodologies and interpretation varied widely
Coverage of genes varied widely some covered only <50%
of gene
Sales people lie
If you want the best scientific information approach the
most unattractive person in the booth.
Step 2 -Digging into Data
3 Different EMR’s owned by two different institutions
No data standards for genetic testing
Description of genetic testing in the medical record varied
widely and was subject to intern knowledge:
From “SCN1A sequencing sent to Athena Laboratories” to
“genetic test” or “misc. test”
90% of descriptions of the same test were unique
In the interest of time I looked at only tests that did not
require manual review to attain an unbiased sample.
Single Gene Testing
Dravet’s Syndrome
Thought to be a well defined clinical syndrome
typically tested by sending SCN1A, or a small panel of
3 genes - SCNIA, SCN1B, and GABRG2.
Results
Total tests sent combined = 32
Cost of SCN1A > $2200.00
Cost of Small Panel 3 genes - $6130.00 (more than 500
gene panel)
Positive = 0, VUS = 4
Diagnostic Yield 0%
Wasted Money > $130,000.00
Results of Most Commonly
Ordered Epilepsy Panel
Early Infantile Epileptic Encephalopathy
Cost $4500.00
Total Tests = 23
Positive = 3
VUS = 4
Diagnostic Yield 13%
Results - WES
62 tests on patients with epilepsy
Diagnostic Tests: 22
VUS: 30
VUS Rate: 48%
Diagnostic Yield: 35%
Our diagnostic yield on exome results was near
reported ranges from 30 to 40%
Choosing Panels
Multiple companies with large epilepsy panels
Only 2 of them do sanger back-fill and del/dups
Largest panel 1000 genes
Fastest panel, 2 week turnaround for the most
common and actionable genes.
Price is important but not as important as accuracy
and customer service
CMA Data
2900 CMA’s from past 5 years
Phenotypes of all patients were standardized based
on indications on the CMA paperwork
Manual chart review performed on patients with
epilepsy and cardiac problems to further refine
phenotypes
Tool built that allows retrieval of diagnostic yield
based on phenotype
Results - CMA
406 patients with epilepsy
Diagnostic yield with developmental delay – 27%
Diagnostic yield with congenital anomalies – 41%
Diagnostic yield with both of the above – 64%
102 with isolated epilepsy
Diagnostic yield of CMA - 0%
5 VUS – also found in phenotypically normal parent
Why is CMA indicated in
Isolated Epilepsy?
In published studies 2- 3% of patients with isolated
epilepsy were found to have deletions primarily in
15q11.2 and 16p13.11.
Several patients in our population had similar
findings but all had developmental delay as well.
Carolien G. F., et al. “Recurrent microdeletions at 15q11.2 and 16p13.11 predispose to idiopathic generalized epilepsies”, Brain Jan 2010, 133 (1) 23-32
Mefford, Heather C., et al. "Genome-wide copy number variation in epilepsy: novel susceptibility loci in idiopathic generalized and focal epilepsies."
PLoS Genet 6.5 (2010)
Large NGS Panels vs Smaller Panels with
Sanger Backfill and Deletion/Duplication
Advantages of Large NGS Panels
Cover more genes
Disadvantages of Large NGS Panels
More Variants of Unknown Significance
Incomplete Coverage of Genes
Miss 10% of Positive Results Without Del/Dup Studies*
WES vs Large Panels
Advantages of Whole Exome Sequencing
Complete Coverage – Higher Diagnostic Yield
Parental Testing Included
Disadvantages of Whole Exome Sequencing
Higher Rate of VUS
More Complicated Counseling Process
Expensive
Slow Turn Around Time
Reasons for Current Lab Selection
The majority of tests and laboratories used in the
testing of epilepsy were used for historical reasons.
There was little if any comparison of laboratories
performed and very few people were aware of the
costs of sending these tests.
Lack of time to research tests and risk of using a
laboratory they did not have experience with were
common reasons cited for not considering other
options.
Reasons for Current Lab Selection
Neurologists and geneticists tended to use different
laboratories based on what they had used in the past.
There were significant differences in testing behaviors
between the two groups and some differences even
among those in different specialties within neurology.
There seemed to be a perception that our overall
diagnostic yield was higher than it actually was.
Physicians tended to be more open to changing practice
after seeing results of the testing being sent.
Problems Identified with Genetic
Testing Laboratories
Poor reporting
Questionable methodology
Poor customer service – unable to reach a genetic counselor
Poor customer service – unable to reach a human
Aggressive pursuit of patients by billing department
Aggressive sales people
Exceeding stated turn around times
Cost and difficulty in testing parent for VUS
High prices
Factors for Selecting Laboratories
Reporting Quality
Reliability of Lab
Reputation of Lab
Turn Around Time
Customer Service
Access to Genetic Counselors
Offers Prior Authorization Service
Free Parental Testing for Variants of Unknown Significance
Access to Data for Research
Factors for Selecting Panels
Number of Genes Covered
Sanger Back Fill for Regions of Poor Coverage
Sanger confirmation of findings
Deletion Duplication Performed or Offered
Fast Turn Around Time
Price
Discussion
Based on our findings we decided patients with isolated epilepsy that have
had time to show normal development should not have CMA as the first
line of testing.
In patients with epilepsy accompanied with developmental delay or any
congenital anomalies CMA is indicated and was shown to be very high yield.
As a next step we felt starting with a large gene panel that included
parental testing to better assess variants of unknown significance provided
the largest coverage up front at a lower cost than WES without requiring
the extensive counseling and long turn around time that goes along with
WES.
We felt that reflexing to deletion duplication studies in the most common
genes in epilepsy was important as deletions and duplications compromise
about 10% of positive findings in these genes.
Discussion
If the aforementioned studies are negative we then
reflex to exome for the price difference between the
original NGS panel and the exome.
Working with the laboratory to reflex to WES for the
price difference between the tests made the decision
of not going straight to exome easier and eliminates
the costs associated with sending two expensive
tests. This makes sense as most NGS tests are ran on
an exome backbone and the additional work is based
on interpretation of the remaining genes.
Epilepsy Algorithm
Changing Practice
Forced change not accepted well
Physicians open to change when presented with data
Physicians should be part of the decision making process
Physicians as a general rule want to provide the best care
for the patients
Physicians tend to worry less about cost unless it
translates to cost for the patient’
Physician are often not aware of costs
Tool Built
Allows for analysis of results of test based on patient
phenotype
Allows for calculation of a probability of a positive
result given a specific phenotype
Limited internal use has saved money
Challenges
Messy non-standard data about tests
Non-standard phenotype information
Pricing information not available in EHR hidden from
physician
Genetic Testing is very dynamic field and something
has probably changed since I have started this
presentation
Future Implications