Three Generation Scenario with Likely Genetic Predisposion

Scenario:Family History with BCRA1 risk prediction

- Three generation scenario with likely genetic predisposion:

For extra credit, make use of a family from theCanonical Pedigree Project. The HAPI server has been pre-populated with the Green and Brown families.

Simplified Family Tree for the Greens and the Browns
Index / Name / Gender / DOB / Died / Notes
1 / Peter Green / male / 1936-10-28
2 / Bertha Green / female / 1937-10-28
3 / Eric Brown / male / 1935-10-29
4 / Mary Brown / female / 1936-10-28
5 / Harry Green / male / 1960-10-22
6 / Cindy Green / female / 1962-10-22 / 2005-10-22 / Dx Breast Cancer, age 36
7 / Paula Brown / female / 1960-10-22 / 2005-10-22 / Dx Breast Cancer, age 37
8 / Amy Brown / female / 1958-10-23 / 2006-10-23 / Dx Breast Cancer, age 42
9 / Frank Brown / male / 1956-10-23 / 2001-10-23 / Dx Prostate Cancer, age 39
10 / Jane Green / female / 1985-10-16 /

Dx Breast Cancer, age 30.
Likely BRCA1 variation that predisposes carriers to breast, ovarian, and prostate cancer.
Mutation risk, according to BRCAPRO, is 35%.

- Minimal History:

o Genetic family members diagnosed with a condition (e.g. breast cancer) at certain age (e.g. age 25)

o Outcome (e.g. in remission for 23 years with no sign of relapse)

- Risk assessment:

o Blood draw (e.g. for genetic profiling)

o Assessment (e.g. level of risk)

- Action plan:

o Patient education (e.g. breast self assessment ...)

o Referral for genetic counseling

(

Simplified family tree for Greens and Browns
Index / Name / Gender / DOB / Deceased Date / fatherID / motherID / Relationship with Proband / Notes
1 / Peter Green / male / 10/28/36 / PGRFTH
2 / Bertha Green / female / 10/28/37 / PGRFTH
3 / Eric Brown / male / 10/29/35 / MGRFTH
4 / Mary Brown / female / 10/28/36 / MGRMTH
5 / Harry Green / male / 10/22/60 / 1 / 2 / FTH
6 / Cindy Green / female / 10/22/62 / 10/22/05 / 3 / 4 / MTH / Dx Breast Cancer, age 36
7 / Paula Brown / female / 10/22/60 / 10/22/05 / 3 / 4 / MAUNT / Dx Breast Cancer, age 37
8 / Amy Brown / female / 10/23/58 / 10/23/06 / 3 / 4 / MAUNT / Dx Breast Cancer, age 42
9 / Frank Brown / male / 10/23/56 / 10/23/01 / 3 / 4 / MUNCLE / Dx Prostate Cancer, age 39
10 / Jane Green / Female / 10/16/85 / 5 / 6 / SELF / • Dx Breast Cancer, age 30.
• Likely BRCA1 variation that predisposes carriers to breast, ovarian, and prostate cancer.
• Mutation risk, according to BRCAPRO, is 35%.
Column / FHIR mapping
Index / FamilyMemberHistory.id.value
Name / FamilyMemberHistory.name.value
Gender / FamilyMemberHistory.gender.value
DOB / FamilyMemberHistory.ageString
Deceased Date / Patient.deceasedBoolean=true;
fatherID / FamilyMemberHistory.extension.(url=identifier).valueCodableConcept.coding.code.value
motherID / FamilyMemberHistory.extension.(url=identifier).valueCodableConcept.coding.code.value
relationship / FamilyMemberHistory.relationship.coding.code.value
notes / multiple

FamilyMemberHistoryxmlns="
<id value="1" />
<extension url="
<extension url="type">
<valueString value="father"/>
</extension>
<extension url="identifier">
<valueCodeableConcept
<coding>
<system value="Id"/>
<code value="5"/>
<display value="Father of the patient"/>
</coding>
</valueCodeableConcept
</extension>
</extension>
<extension url="
<extension url="type">
<valueString value="mother"/>
</extension>
<extension url="identifier">
<valueCodeableConcept
<coding>
<system value="Id"/>
<code value="6"/>
<display value="Mother of the patient"/>
</coding>
</valueCodeableConcept
</extension>
</extension>
<patient>
<reference value="10"/>
<display value="Jane Green"/>
</patient>
<status value="completed"/>
<name value="Jane Green"/>
<relationship>
<coding>
<system value=" />
<code value="SELF" />
<display value="SELF" />
</coding>
</relationship>
<gender value="female" />
<ageString value="30" />
<deceasedBoolean value="false" />
</FamilyMemberHistory

The proposed model system architecture will integrate patient data available from multiple sources, including; Family History, predictive risk and available genetic test data. Family history data will be generatedwith HL7 FHIR standards to simulate pedigree informationthatcapable of being transmitted to the predictive models to calculate potential onset of breast cancer and the probability of BRCA1/BRCA2 mutations. The goal is to have the data integrated so as predictive risk may be determined, the analyticsare informative to decide if clinical genetic samples should be run and specialty advice sought.

Primary Screening Decision Logic and Data flow for Risk

Family History has been long discussed and resources for pedigree have evolved, here we request insertion of information that will be required to id a patient, and link specific pedigree relationships so as risk may be determined. Two element of information are sought from the clinicians A) applicable patient coding so as to link a patient with other persons in the pedigree and B) discussion as to answers and wording to assessment questions so as they may be standardized. these elements will allow a standardized metadata file to be transmitted for analytic programs for query and reinsertion to PHR.

Patient data:

Patient.ID
Patient.name
Patient.gender
Patient.birthDate
Patient.deceasedBoolean
FamilyMemberHistory.extension.(url=type).valueString
FamilyMemberHistory.extension.(url=identifier).valueCodableConcept.coding.code.value
FamilyMemberHistory.extension.(url=type).valueString
FamilyMemberHistory.extension.(url=identifier).valueCodableConcept.coding.code.value

Father data:

FamilyMemberHistory.id.value
patient.extension..extension(url=type).valueString
patient.extension.extension(url=identifier).valueCodableConcept.coding.code.value
FamilyMemberHistory.extension.(url=type).valueString
FamilyMemberHistory.extension.(url=identifier).valueCodableConcept.coding.code.value
FamilyMemberHistory.patient.reference.value
FamilyMemberHistory.patient.display.value
FamilyMemberHistory.name.value
FamilyMemberHistory.gender.value
FamilyMemberHistory.ageString
FamilyMemberHistory.status.value
FamilyMemberHistory.relationship.coding.code.value

Mother data

Example condition entry format:

FamilyMemberHistory.condition.coding.code.code.value
FamilyMemberHistory.condition.coding.code.display.value
FamilyMemberHistory.condition.onsetString.value
FamilyMemberHistory.codition.note.text

Screeningtools employ a simple graphical user interface to collect a detailed three generation pedigree and determine the number of blood relatives who have or have had breast and/or ovarian cancer. Genetic risk algorithms require information about parents, siblings, children, grandparents, aunts, uncles, nieces, nephews, and grandchildren. Information from the patient should reflect data about herself and family members in both maternal and paternal lines:

Vital status (living or dead);

Age (if still living);

Whether the individual has or has had breast and/or ovarian cancer;

Whether individuals with breast cancer had one or both breasts involved;

The age at diagnosis of each individual with breast and/or ovarian cancer;

Any history of chemoprevention or risk reduction surgery;

Family history (eg. Ashkenazi Jewish descent)

In addition, the screening tools usuallyinclude the Gail Model algorithm, which combines the patient’s family history with information from the patient’s own medical history to predict a woman’s risk of developing breast cancer in the next five and ten year time periods (Costantino, et al., 1999; NCI, 2011). The number of women with increased risk of breast cancer by the Gail Model calculation is larger than the number at risk for the hereditary syndromes. It s also pertinent to identify this group because treatment with chemoprevention using selective estrogen receptor modulators, such as Tamoxifen and Raloxifene, can reduce the risk of breast cancer by nearly 50% in these women (Smith & Good, 2003; Vogel, et al., 2006). The screening tool will therefore also ask patients to input:

their age at menarche,

age at first live birth,

race,

number of breast biopsies (if any) and

number of biopsies showing atypical hyperplasia (if any)

A genetic risk of 10% or greater and/or Gail Model risk of 1.7% or greater will be scored as ―positive, since these threshold values indicate the need to refer the patient to a specialist for further evaluation. In those cases, the system will generate a graphical risk report in an easily understandable format for the clinician to use in subsequent discussions with the patient.

At the completion of the risk calculation, a notification should be sent to the primary provider and theriskpercentagesarerecordedinthepatient’selectronic record. Ideally, Positiveresultswillalso trigger generation of an alert to the physician (or radiologist) to notify him/her of the positive result. The alert will include links to concise summaries of the information needed for discussions with the patient. The summary will include:

Previously generated reports (if any);
Information about the biological basis and medical significance of the finding;
Information to put the screening result in a larger perspective (i.e. the majority of screen positive patients actually do not have a BRCA1 or BRCA2 mutation since the screening tool favors sensitivity over specificity);
Information about the reasons for referral to a genetic counselor due to increased genetic risk or a medical oncologist due to increased Gail Model risk. The provider will discuss options for procedures that provide further risk evaluation and possible outcomes and opportunities for intervention depending on the subsequent results.

Eventually, if the patient elects to use the screening instrument at home via a web portal, similar information will be made available as a simple, concise electronic summary. This summary will also include online links to several evidence-based resources. These online sources provide detailed information on the topics of breast and ovarian cancer. Alternatively, The screening instrument will also offer the patient a chance to utilize the HHS My Family Health Portrait application. This can be used to create a more comprehensive pedigree with information that can be later used to calculate genetic risk for multiple diseases, both malignant and nonmalignant. This tool allows storage of the genetic risk data in the patient’s personal health record, such as Microsoft HealthVault. In addition, it has the functionality to report the data in the HL7 Version 3 Standard Pedigree format (2012). It also creates a printed version for discussion with the patient’s healthcare provider(s). The screening instrument’s expert content and technical requirements will lead to a widely accepted, robust, evidence-based tool that enables successful screening to identify women with a high risk for breast and ovarian cancer.

additional credit: BRCA Lab integration

Integrating genetic tests into CDS speeds delivery of services. When genetic test results are available, a treatment recommendation application can assist in alerting as to alternative treatments for the patient. For example, A patient with a positive breast cancer biopsy and BRCA gene mutation is also at risk for ovarian cancer. Knowing whether she carries the BRCA gene mutation may change her ovarian-cancer screening schedule. As such, genetic testing can greatly shorten the diagnostic process and help provide timely treatment.

BRCA Challenge (GA4GH) is an international effort to advance understanding of breast, ovarian, and other cancers by pooling genomic and clinical data on the genes BRCA1 and BRCA2. The BRCA Exchange, the first product of the BRCA Challenge, is a publicly accessible web portal that provides a simple interface for patients, clinicians, and researchers to access curated expert interpretations and some supporting evidence for genetic variants identified in BRCA1 and BRCA2.

The new release will include an additional tier that allows web users, after agreeing to a disclaimer and data use policy, to interact with data on genetic variants that have not yet been expert classified. This represents the first time that all publicly available BRCA1 and BRCA2 variant data contained in databases such as ClinVar, LOVD, ExAC, and the 1000 Genomes Project, have been pooled in a single, federated resource. Such federation allows researchers, clinicians, and others to simultaneously search across the world’s collective knowledge on the genes.

Variants can be found here:

Please see V3/FHIR mappings on next page when inputting data into ClinFHIR:

Courtesy of Kevin Power:

HL7 Version 3 Implementation Guide: Clinical Genomics; Family History/Pedigree Interoperability, Release 1 - US Realm / HL7 FHIR (FHIR DSTU2 (v1.0.0-6951))
FH Record / FH Record
id / Unique ID for FH instance / List.identifier / List.id is a technical id and would change for every server where the pedigree was stored. List.identifier is a business identifier and would remain consistent.
code / List.code / This would say something along the lines of "Family History", though it could be narrower (e.g. Genetic family history, Breast Cancer risk family history, etc.)
text / List.text / The narrative for the resource
statusCode / Completed, still active (RIM Act State Machine) / List.status? / It should match. There are a bunch of v3 statuses, but most don't make any sense for a pedigree. current=completed; retired=superseded; entered-in-error=nullified. Could be we'll want to add a "draft" too which would correspond to "new"
effectiveTime / List.date / This is when the list was prepared. There's also FamilyMemberHistory.date which captures the same information on a per-relative basis
confidentialityCode / List.meta.security, FamilyMemberHistory.meta.security / In FHIR, confidentiality is handled via tags. Note that all records would need their own tags. (And it's possible that some family members could be tagged differently than others.)
uncertaintyCode / Missing / No mapping - not clear what it means for uncertainty to be captured at the level of the entire history. Could add an extension if we really needed one.
languageCode / List.language, FamilyMemberHistory.language
methodCode / Program creating the family history data / Missing / Could capture this w/ Provenance or an extension
informat / Source of information from which FH was collected / This would also come from Provenance, though we could try to make the case for inclusion as part of the 80%
risk / Not sure what this is
component / Problem known by not attributed to a specific family member / Observation / The List can include a mixture of FamilyMemberHistory entries as well as Observation entries (the latter can say things like "no family history of breast cancer" or "paternal history of heart disease", etc.
patientPerson (proband) / Patient (proband)
id / Patient.identifier / Seems this must all be captured on the probands record? That is probably OK?
name / Patient.name
telecom / Patient.telecom
administrativeGenderCode / Patient.gender
birthTime / Patient.birthDate
deceasedInd / Patient.deceasedBoolean
deceasedTime / Patient.deceasedDate
raceCode / Observation.value or extension on Patient / Observations are probably better as the demographic racecode is fully self-asserted and may have little relationship to ancestry.
ethinicGroupCode / Observation.value or extension on Patient / Observations are probably better as the demographic racecode is fully self-asserted and may have little relationship to ancestry.
relative / 0..N(relative) / List<FamilyMemberHistory
clinicalObservation / 0..N(relative) / FamilyMemberHistory.condition
relative / Relative
code / HL7 Role Code (FamilyMember - Relationship to the patientPerson) / FamilyMemberHistory.relationship / HL7 v3 Value Setup FamilyMember)
id / FamilyMemberHistory.identifier / 0..*
name / FamilyMemberHistory.name / 0..* (Why 0..*?)
telecom / Missing
administrativeGenderCode / FamilyMemberHistory.gender / Mappings to V3
birthTime / FamilyMemberHistory.born
FamilyMemberHistory.age / Accounts for approximate age
deceasedInd / FamilyMemberHistory.deceased
deceasedTime / FamilyMemberHistory.deceased
raceCode / Extension pointing to Observation (with value) or extension on Patient / Observations are probably better as the demographic racecode is fully self-asserted and may have little relationship to ancestry.
ethinicGroupCode / Extension pointing to Observation (with value) or extension on Patient / As Observations or as extensions. (Observations are probably better as the demographic racecode is fully self-asserted and may have little relationship to ancestry.)
deceasedEstimatedAge / Estimated age of death / FamilyMemberHistory.deceasedQuantity
livingEstimatedAge / Estimated age of living relative whose birth date is unknown / FamilyMemberHistory.born[x]
FamilyMemberHistory.age[x] / Accounts for approximate age
relative / id of mother / FamilyMemberHistory.extension.geneticParents / Details
relative / id of father / FamilyMemberHistory.extension.geneticParents / Details
clinicalObservation / 0..N / FamilyMemberHistory.condition / 0..*
FamilyMemberHistory.note
FamilyMemberHistory.status
No children by choice or reason unknown could be a note, an extension or an observation
Infertility would be a condition
clinicalObservation / Conditions / Observations
code / SNOMED/ICD / FamilyMemberHistory.condition.code
dataEstimatedAge / Age of onset / FamilyMemberHistory.condition.onset / Accounts for approximate age
FamilyMemberHistory.condition.outcome
FamilyMemberHistory.condition.note
genomics / Variant details / FamilyMemberHistory.extenstion.geneticsObservations / Details
PedigreeAnalysisResults / RiskAssessment
ID / RiskAssessment.identifier
code / disease or variation for which probabilities/risks are calculated / RiskAssessment.condition / Does this have to be on the probands record of conditions? No. It's a risk for a condition, not necessarily a condition they have.
effectiveTime / RiskAssessment.date
negationInd / can indicate no risk / missing? / No risk would be reflected by risk probabilities of 0
methodCode / algorithm / RiskAssessment.method
RiskAssessment.performer / Performer could be a device (system) that performed the analysis
inputParameters / inputs to the algorithm / missing? / RiskAssessment.basis lets you point to the family history information, observations, questionnaires or other source information, but doesn't let you specifically flag any particular input as being a specific parameter name. If you wanted to do that, you'd need extensions.
components / 1..N / RiskAssessment.prediction
AnalysisResult / catcher for any that don't fit other classes / outcome
probability[].probability(Decimal/Range/CodeableConcept)
relativeRisk
when[].when(Period/Range) / While structured differently, it feels like we can express the same classes of risks with the prediction?
Note - RiskAssessment also allows for probabilityCodeableConcept, which V3 does not
Probability / probability of having
PercentageRisk / percentage of risk
RelativeRisk / relative risk
AgeAndProbability / links Age to Probability
RiskAssessment.basis / Information used in assessment
RiskAssessment.prediction.outcome / Possible outcome for the subject
RiskAssessment.predicition.rationale / Explanation of prediction
RiskAssessment.mitigation / How to reduce risk