Clinical-Genomics HL7 SIG

Clinical-Genomics HL7 SIG 1

Clinical-Genomics HL7 SIGThe Tissue Typing Use Case

Amnon Shabo1, Shosh Israel2, Guy Karlebach1

1IBM Research Lab in Haifa, 2Hadassah University Hospital

Presented by Amnon Shabo

SHAMAN = Secured Health and Medical Access NetworkIMR = Integrated Medical Records Middleware

In collaboration with the Hadassah University Hospital in Jerusalem

Haifa Labs Integration of multiple sources of data; transformation to standards; full-text indexation

Watson/YorktownLabs

Processing of personal genomic and proteomic data


Types of Genomic Data• DNA Sequences

• Personal SNPs (Single Nucleotide Polymorphism)

• Programmatic / manual annotation (e.g., SNPs combination x could possibly lead to mutation y)

• Gene expression levels

• Proteomic (proteins translated w/SNPs)


The Case for Clinical-Genomics• Clinical-Genomics: the use of information obtained from DNA sequencing, patterns of gene expression & resulted proteins for healthcare purposes

• Personalized Medicine– Detect sensitivities/allergies beforehand– Drug Selection by clinicians

• Pharmacogenomics– Improve drug development based on clinical-genomics correlations

– Personal customization of drugs

• Preventive Care


Gene Expression in Cancer• Differences between normal tissue vs.

premalignant lesion vs. neoplastic tissue – markers of diagnostic value– targets for drug research– evolution of cancer

• Differences between responders vs. non-responders for a standard therapy

• Development of drug-resistance

• Correlation of gene expression patterns with presentation or evolution:– long vs. short survivors– metastatic vs. non-metastatic– clinical or pathological grades


Differential Display• Difference between banding patterns of cDNA from tumor tissue and normal tissue on polyacrylamide gel can point to a protein that could potentially be the target of a therapeutic antibody.

• DNA microarrays are also employed to examine the genetic expression of thousands of potential antigens and determine which are present in abnormal (tumor) tissue but not in normal tissue.


Using Databases• Vast databases of genetic information contribute to genomic research

• Search for potential antigens can be as easy as an online search

• HLA Database example: (part of the IMGT - international immunogentics project)

http://www.ebi.ac.uk/imgt/hla/


Clinical-Genomics InterrelationsBi-directional relationships:

• Genomics Clinical– Personal SNPs could be interpreted as mutations and thus indicate possible diseases/sensitivities

• Clinical Genomics– Patient & family history leads to genetic testing order

– Crosschecking of genomics results


SNPs Interpretation• SNPs as known mutations (might imply the develop. of diseases)

• Unknown SNPs: – in significant segments of the gene(possibly imply individual differences)

– in gene segments that translate to inactive parts of the proteins(thought to be insignificant)

• SNPs as normal polymorphisms


CG Uses: From Clinical to ForensicThese pictures describes paternity casework autoRADS - the left picture shows a case of paternity exclusion and the right one a case of paternity inclusion.

Taken from the site of Genelex, a company which offers, among other genomic services, paternity testing (see http://www.genelex.com/).


Variety of MethodsSTR (short tandem repeats )

STR’s are short sequences that are easy to detect and its specific pattern of repetitions could identify a gene without needing to

sequence the entire gene.


HL7 Specs for Clinical-Genomics• Create a DIM for Clinical-Genomics

• Derive R-MIMs and message types

• Clinical-Genomic Documents (CDA L3!)

• Review / Utilize the followingemerging bio-informatics standards– BSML (Bioinformatic Sequence Markup Language)

– MAGE-ML (Microarray and GeneExpression Markup Language)

Problem: These standards are not necessarily patient-based.


BSML: Sequencing Markup<Sequence id="_2" db-source="GMS" length="51" representation="raw" molecule="dna" topology="linear"

alignment-sequence="_"> <Feature-tables>

<Feature-table>- <Feature title="gms:sequence">

<Interval-loc startpos="1" endpos="51" /> </Feature> <Feature title="gms:new_fragment">

<Interval-loc startpos="1" endpos="51" /> </Feature> <Feature title="gms:annotation" value="possible somatic mutation cell line #4 end-

11thxml" /> <Feature title="/gms:new_fragment" /> <Feature title="/gms:sequence"/>

</Feature-table> </Feature-tables> <Seqdata>

AGGAATCAGAAAGGACACTCTGGACTTCAGCCAACAGGATACCTGAGCTGA</Seq-data>

</Sequence>


MAGE-ML: Gene Expression• Gene Description:

<reporter id="1051_g_at"> <rep_des V="Source: Human melanoma antigen

recognized by T-cells (MART-1) mRNA." /> </reporter>

• Gene Expression Levels:

<reporter id="32847_at" accession="U48959"><NormalizedIntensity value="0.235" /> <Control value="230.972" /> <Raw value="54.3" /> <T-testPValue value="no replicates" /> <PresentAbsentCall value="A" />

</reporter>


Analogy to Imaging IntegrationHL7DICOM relationship:

existing standardsIMAGINGDICOM

GENOMICSBSML;MAGE;I3C Efforts

Mass data

Summarized data

Pixels

Radiologist-Report

Sequences;Gene- Expression;ProteinsGenomicist-Report


Current Experimentations at IBM Research• A clinical point of view

– Bone-marrow transplantation center in Israel• Donor-recipient matching: tissue typing• Reporting to international BMT registry

• A research point of view– Research center in Canada

• Focusing on heart&lung diseases• Trying to find clinical-genomic interrelations

• Using clinical data from patient records compared with healthy people

• Using genomic data, mainly gene expression levels and proteins


Collaboration with Hadassah• Information exchange

– Report to international registries (IBMTR) • Standardization

– Transform to HL7-CDA documents (L.13)• Indexing

– Index all data including semi-structured data• Annotation

– Integrating the personal genomic data • Visualization

– Visualizing the integrated BMT documents

…agctgaa…SNPs


The BMT Procedure

Pre-BMT

BMT

Post-BMT

–Matching a donor or autologous transplant–Conditioning

•Irradiation•Chemotherapy•GVHD (Graft vs. Host Disease) Prophylaxis

–Substance donated•Bone-marrow•Peripheral blood stem cells•Cord blood stem cells•Donor lymphocytes

-Transplant

–Control of GVHD and other complications–Hematopoietic Reconstitution–Engraftment and Chimerism


New Trends in BMTMini-allografts (mini-transplantations)

• Immunosuppression instead of total conditioning (destroying the entire immune system)

• Infusing donor lymphocytes to attack tumors, cancerous cells, autoimmune artifacts and infectious pathogens

• Stopping the donor lymphocytes once they’re done with the patient disease source, so that they won’t attack the patient normal cells using ‘suicide genes’

• Striking a balance between to 2 immune systems


The HLA-Typing Use Case• HLA = Human Leucocytes Antigens; determine the personal fingerprint distinguishing between self and non-self

• HLA-Typing methods move from serology (antibodies) to molecular (DNA) and recently to DNA sequencing yielding higher levels of typing resolution

• Common Triggers: donor-recipient matching, familial relationships, disease association


Donor Matching• HLA (Human Leukocytes Antigens)

– HLA Typing– DNA typing

– About 6 important loci, each can have dozens of different antigens (alleles)

– Haplotype – common set of antigens

• Relatives versus unrelated donation• Donor banks• Search engines

• Lack of donors to minorities


HLA Alleles in the Family


Differences in Antigens

Class I:

Variables exons: 2,3,4

Allelic polymorphism is concentrated in the peptide (antigen) binding site:

Class II

Variables exons: 2


The HLA-Typing Triggers• Donor-Recipient Matching

– Bone-Marrow transplant• Full match (identical twin) • Avoid GVHD and Promote GVM • Precise and personal match rather than full match

– Organ transplant (cross-match: antibodies) • Living donor: also HLA typing before transplant

• Select the best treatment for the individual patient-donor matching

• HLA-typing is done for post-transplant Info. • Forensic Scenarios

– Paternity disputes – Crime suspects

(HLA is one component of known genetic markers)


Personal Rather than Full Match

Personal match could be beneficial to to new trends in BMT:

• HLA - A & B versus C:– When there is a match in HLA A & B:– Mismatch in HLA-C might promote GVL (Graft vs. Leukemia)

• Mini-transplants:– Avoid full-match (even when identical twin is available)


Data of Interest• Class I allele sequences (all cells):

– HLA-A– HLA-B– HLA-C

• Class II allele sequences (certain cells from the immune system):– HLA-DR (most important)– HLA-DQ (the contribution is not proven but can verify the DR match since there there is strong linkage)

– HLA-DP (usually is not being typed)

• might sequence only the polymorphic segments (e.g., exon 2 in class II and exon 2-4 in class I), each exon is about a 300 nucleotides, because SNPs in other segments are not important to the matching


New Naming Convention• Letter designates the membrane locus

• Full allele name: eight digits

– First 2 digits defining the allele family and where possible corresponding to the serological family

– Third and fourth digits describing coding variation

– Fifth and sixth digits describing synonymous variation

– Seventh and eighth digits describing variation in introns


Sequencing Data Example:Generic Meta Data:

– Local Names: DRB1*110101– IMGT/HLA No: HLA00756– Class: II– Assigned: 01-AUG-1989– Last Aligned: 17-OCT-2002– Component Entries: AF029281

AJ297587 – Cell SequenceDerived From: 34A2, FPAF

– Known Ethnic Origin of Cells: Caucasoid

– Length: 801 bps


Sequencing Data Example:

IMGT-HLA SEQUENCE DATABASE.htm

DRB1*110101

SNPs



IMGT-HLA SEQUENCE DATABASE.htm

SNP-Resulted Protein Sequence



IMGT-HLA SEQUENCE DATABASE2.htm

DRB1*110401

SNP



IMGT-HLA SEQUENCE DATABASE2.htm

SNP-Resulted Protein Sequence


Testing Kit Output Example

- Sample ID - Kit Name- Name - Kit Lot Number- Ethnic Group - Kit Expires- Donor/Patient - DNA Extraction- Purpose of Test - DNA Quality- Test Date - DNA Concentration- Test By - Review Date- Comments - Reviewed BySerology Results:HLA A: B: C: DR: DQ: Positive Lanes:Kit-specific

data


Tissue Typing Report- Recipient- Subject- Specific Alleles

- Record Number- Molecular Sample

- Date- Disease

- Patient Result

- Specific Alleles

- Possible combinations

- Siblings- Unrelated Donors


Search for Unrelated Donor• Banks of potential donors (volunteers)

• Each donor was tested only for HLA Class I

• When a patient needs a donor:– The transplant facility searches the donor banks to

find a donor (direct access to the donor banks databases)

– The search is based on Class I matching

– If appropriate donors are found – then the searching transplant facility initiates a request to the respective donor banks, asking for Class II typing

– Each approached donor bank is moving the request to the tissue typing lab where the DNA samples reside

– Class II matching results are returned to the searching facility and if the donor with the best match in both class I & II is approached


Donor Banks

Search for Unrelated DonorTransplant Center (TC) searches for

donorsDonor Banks

Patient Class I HLA

Class I Matching donors

Donor Bank

Request for HLA class II typing

TC chooses potential donors

Tissue Typing Lab Class II Typing

TC chooses

best donor

Class II Matching donors


Genomic Data in a Clinical Docs

• A DNA Testing Device – raw DNA sequences

• Reports from service units, e.g., tissue typing, should answer questions such as patient-donor matching, fatherhood, etc.

• Embedding annotated results received from a DNA lab in a CDA document

• Linking genomic annotations and clinical data (external links?)


Matching Option Notations• Different notations for coarse-grain results:

– possibilities from the A24 antigen family could be represented differently by different kits on the same patient DNA tested:• A*2402101-06/08-11N/13-15/17/18/20-23/25-36N• A*2402101-06/08-11N/13-15/17/18/20-23/25-31

– Pair combinations (inherited alleles):• DRB1*0402 AND DRB1*0408orDRB1*0404/44 AND DRB1*0414

Kit A:Exact combinationKit B:

two possiblecombinations or


Report Example – Unrelated DonorsThe Patient

Unrelated Donor 1

Unrelated Donor 2

Unrelated Donor 3


Class I vs. Class II Antigens• A 4-digit resolution level is common in class II antigens as they have been discovered more lately

• It’s desired that class I antigens will report in 4 –digits as well as they are more crucial to BMT success

• 4-digits reporting requires molecular and sequencing procedures

• 4-digits reporting still not common in class I


Clinical-Genomic Data in CDA?• What should go into a clinical document (extent of detail)?

• Programmatic and manual annotation at different levels?

• The users of such integrated documents: clinicians? genomicists? patients? Medico-ethical issues!

• HL7-Association semantics that represents the interrelations of clinical-genomics


First Attempts using CDA…• GMS

– Genetic Messaging System– From the computational biology center in IBM Watson– Example: integrating the genomic annotation and analysis of the

personal DNA sequences, into the clinical document (CDA format)

<levelone> <clinical_document_header>  </clinical_document_header> <body>   <gms:dna sequence="2" base="802" locus="1"> <gms:annotation>

possible somatic mutation cell line #4 end-11th

</gms:annotation> AGGAATCAGAAAGGACACTCTGGACTTCAGCCAACAGGATACCTGAGCTGA... <gms:automated_annotation> </body></levelone>

CDA L1


And the Work Just Begins…• Use Cases in Detail & Taxonomy

• High-Level CG Model and HL7-DIM

• Messages

• Documents

• Prototyping info. Exchange using specs