Validating algorithms under CLIA and GLP requirements

Validating algorithms under CLIA and GLP requirements

Loading
Loading Social Plug-ins...
Language: English
Save to myLibrary Download PDF
Go to Page # Page of 52

Description: Validating algorithms under CLIA and GLP requirements, Flagship Bio sciences, 3 kinds of people, Vendor validation gap with CLIA. US Food and Drug Administration, Medical Device Manufacturers Regulations, Digital Pathology FDA 510k Clearances Immunohistochemistry, Hematology and Pathology Devices Advisory Committee, CAP Anatomic Pathology Checklist* that applies to Digital Pathology, ASCO/CAP Guidelines IHC ER/PgR Breast [2010], Recommendations for validating estrogen.

 
Author: Steve Potts (Fellow) | Visits: 3088 | Page Views: 3368
Domain:  Medicine Category: Biotech/Pharma Subcategory: FDA Practice and Patent Law 
Upload Date:
Short URL: http://www.wesrch.com/medical/pdfME1XXFXMHKMET
Loading
Loading...



px *        px *

* Default width and height in pixels. Change it to your required dimensions.

 
Contents:
Validating algorithms under CLIA and GLP requirements
27 July 2010

Steve Potts, Ph.D. MBA CEO Flagship Biosciences LLC

www.flagshipbio.com

Flagship Biosciences
A digital pathology and image analysis service provider
� Slide scanning & hosting � Advanced image analysis � Digital pathology conferencing

A combination of technology, talent, experience, expertise Fast and secure IT infrastructure GLP and CLIA-compliant histology and pathology
� EPL, Inc (GLP histology and pathology) � EPL Archives (GLP archives) � Vitro Molecular (CLIA and IHC development)

Biomarker development � EMT companion diagnostic � Histone profiling (PrognosDx)

www.flagshipbio.com

Disclosure

Flagship Biosciences has no financial connection (other than being a customer) to Aperio
� We use Aperio scanners for hardware, Aperio software for cell and area measurements and Visiopharm software for advanced image analysis and stereology. � Also developing our own in-house algorithms with the Aperio SDK and other technologies

Flagship is a pathologist-owned company seeking to meet practical needs and gaps we see in the industry NOW

www.flagshipbio.com

3 kinds of people in this audience

Clinical � "I know more than I want to know about CLIA, but what is GLP?" Preclinical (safety studies in pharma) "I know all I want to know about GLP, but what is CLIA?" Academic Research and Pharma Discovery Groups "I don't care about GLP or CLIA..."

www.flagshipbio.com

Regulatory groups involved
FDA

CDER Drugs

CDRH, Devices

CBER, Biologics

CVM, Veterinary
www.flagshipbio.com

CFSAN, Food

NCTR

Outline
Vendor validation gap with CLIA (clinical workflows) � Medical device manufacturer regulations for DP � Clinical lab regulations under CLIA Vendor validation gap with GLP (preclinical safety studies) � State of adoption of digital slides in preclinical � Peer review & image analysis � Validation required in GLP The path forward: � SOPs for in-house GLP image analysis � IA validation service for CLIA
www.flagshipbio.com

Vendor validation gap with CLIA (clinical)
Medical device manufacturers can only market specific uses of the technology Aperio example: � "The exact intended use and label indication for each clearance is available at http://www.aperio.com/other/regulatory.asp." � IHC � IHC � IHC � IHC � IHC HER2 Image Analysis HER2 Manual Read of Digital Slides ER/PR Image Analysis PR Breast Tissue Manual Read of Digital Slides HER2 Breast Tissue Tunable Image Analysis

www.flagshipbio.com

Vendor validation gap with CLIA (clinical)
Nuclear algorithm tuned for ER / PR What about when you are counting Ki67?
� Need to modify averaging radius for more blurring effect to compensate for pixelated Ki67 staining � Need to modify curvature threshold for less side-by-side cell segmentation

But no validation is supplied out of the box What about inflammatory cells, TUNEL, many other cell counting applications... ?

A cleared algorithm validated for PR (left) will need to be adjusted for the classic "speckling" and "spotty" staining commonly seen in Ki67 (right)

www.flagshipbio.com

Vendor validation gap with CLIA (clinical)
Membrane algorithm tuned for HER2 What about EGFR...or phosphomarkers...or your novel membrane targets?

P53 staining in advanced breast cancer

An algorithm validated and cleared for HER2 (left) will need to be modified for use in EGFR (right), and may not be appropriate

www.flagshipbio.com

US Food and Drug Administration

IMPORTANT DISTINCTION: Medical Device Manufacturers are regulated by FDA
Can only promote FDA cleared or approved products in the clinical market.

Clinical Laboratories are regulated by CLIA (not FDA)
Have a choice of using non-FDA cleared or approved test systems (see 493.1253) � a question of the laboratory's comfort level.

www.flagshipbio.com

Medical Device Manufacturers Regulations


US Food and Drug Administration (FDA)

www.flagshipbio.com

Digital Pathology FDA 510k Clearances Immunohistochemistry (IHC)
www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm ScanScope XT System (Aperio) 2009/08 2008/10 2008/08 2007/12 2007/10 PATHIAM (Bioimagene) 2009/02 2007/02 VIAS (Tripath) 2006/09 2006/04 2005/08 2005/05 K062428 K053520 K051282 K050012 Breast Breast Breast Breast P53 Ki-67 Her2/neu ER/PR Ventana Ventana Ventana Ventana Image Analysis - System Image Analysis - System Image Analysis - System Image Analysis - System K080910 K062756 Breast Breast Her2/neu Her2/neu Dako Dako Image Analysis - System Image Analysis - SW K080564 K080254 K073667 K071671 K071128 Breast Breast Breast Breast Breast Her2/neu PR ER/PR Her2/neu Her2/neu Dako Dako Dako Dako Dako Tunable Image Analysis - System Reading on Monitor - System Image Analysis - System Reading on Monitor - System Image Analysis - System

ARIOL (Applied Imaging) 2004/03 2004/01 K033200 K031715 Breast Breast ER/PR Her2/neu Dako Dako Image Analysis - System Image Analysis - System

ACIS (Clarient/Chroma Vision) 2004/02 2003/12 QCA (Cell Analysis) 2003/12 Date 510(k) Number K031363 Tissue Stain Breast ER Dako Reagent Image Analysis - SW Application K012138 K032113 Breast Breast ER/PR Dako Image Analysis - System Image Analysis - System Her2/neu Dako

www.flagshipbio.com

IHC Study Design Example Breast � HER2 � Dako
www.aperio.com/pathology-events/webinar_clinical.asp

I. Substantial Equivalence to Manual Microscopy

1) Manual Microscopy

MM 1, 2, 3

2) Reading Slides on a Computer Monitor

DR 1, 2, 3

3 Clinical Sites
- 80 Slides equal distribution - 80 Slides equal distribution - 80 Slides target population
www.flagshipbio.com

3) Image Analysis IA 1, 2, 3

Hematology and Pathology Devices Advisory Committee � 22&23 October 2009
www.fda.gov/AdvisoryCommittees/CommitteesMeetingMaterials/MedicalDevices/MedicalDevicesAdvisoryCommittee/HematologyandPathologyDevicesPanel/ucm146748.htm

2 Day Advisory Panel meeting on Digital Pathology Whole Slide Imaging (WSI) to gather expert information for how to ensure that the current standards of Safety and Effectiveness of routine surgical pathology will not be compromised if WSI is used instead of the conventional light microscope. Is digital WSI a class II device (req. 510k), or a class III device (req. PMA)? How should a manufacturer validate the performance of a digital WSI system? Presentation and discussions about:
- Principles of light microscopy (the reference method) - Principles of digital WSI - Possible criteria and studies for analyzing the accuracy and reproducibility of the diagnostic performance of pathologists - Objective and subjective aspects of reading slides

Status-Quo: FDA has not yet released any guidelines or provided a clear path on what manufacturers have to do to get FDA clearance/approval for digital pathology systems for primary diagnosis.
www.flagshipbio.com

Hematology and Pathology Devices Advisory Committee � 22&23 October 2009

STATEMENT by the College of American Pathologists
Potential Public Health Implications for not establishing proper validation.
Test Conditions: � Focus on use of the tool in the context of how a diagnosis is rendered in clinical practice. � All slides from each case using each modality (vs. "representative" slide) Sample Size: � Large sample size to detect small performance differences. Specimen Types: � Different specimen types require different capabilities and validation procedures. Diagnostic Spectrum: � Challenge set that reflects the spectrum of diagnostic labels and secondary measures. Evaluation Criteria: � Intra-pathologist variability, not accuracy of diagnosis.

Caution: FDA approval for primary diagnosis for a specific tissue application may results in significant off-label use.
www.flagshipbio.com

Lab Regulations
� � �

US Food and Drug Administration (FDA) Clinical Laboratory Improvement Amendments (CLIA) '88 College of American Pathologists (CAP) American Society of Clinical Oncology (ASCO)/CAP New IHC HER2 and ER/PR Guidelines



www.flagshipbio.com

Clinical Laboratory Improvement Amendments (CLIA)` 88
All Clinical Labs need a CLIA license to operate and to be able to bill Medicare & Medicaid

Sections in the CLIA standard that apply to Digital Pathology
493.1105 Retention Requirements
(a)(6) Test reports (a)(7) Slide, block, and tissue retention

493.1251 Procedure Manual
(b)(5) Calibration and calibration verification (b)(6) Reportable range for test results for the test system (c) Manufacturer's test system instructions or operator manuals

493.1252 Test Systems, Equipment, Instruments, Reagents, Materials, and Supplies 493.1253 Establishment and Verification of Performance Specifications
(b)(1) Verification of performance specifications (b)(2) Establishment of performance specifications

493.1254 Maintenance and Function Checks 493.1255 Calibration and Calibration Verification Procedures 493.1256 Control Procedures 493.1291 Test Report

www.flagshipbio.com

College of American Pathologists (CAP)

Almost all Clinical Labs seek CAP accreditation, requires inspection based on check-list

CAP Anatomic Pathology Checklist* that applies to Digital Pathology
#1 Procedure Manual
ANP.07328

#2 Quality Management
ANP.10050 ANP.10200 & 10250

#3 Surgical Pathology Reports
ANP.12500

#4 Immunologic and Molecular Methods � Predictive Markers
ANP.22988 ASCO/CAP HER2 guidelines added (ANP.22989-999) ANP.22997 ANP.22999

#5 Instruments and Equipment ASCO/CAP Guidelines are becoming part of the Check-List

A committee has been formed that is looking into adding Digital Imaging and Image Analysis to the Check-List.
*09/27/2007 Edition

www.flagshipbio.com

ASCO/CAP Guidelines � IHC HER2 Breast [2007]*

Approximately 20% of current HER2 testing may be inaccurate HER2 on all invasive breast cancers No Gold Standard for HER2 I. Preanalytic (Tissue Preparation) II. Analytic (Staining)
95% concordance with another validated test (FISH, ISH, IHC) for pos. and neg. cases

III. Postanalytic (Interpretation)
Modified Scoring Scheme for 3+ with > 30% (vs. 10%) threshold for 3+ cells Equivocal reflex testing with another test (gene vs. protein)

* Guideline for HER2 Testing in Breast Cancer www.flagshipbio.com Wolff et al: Arch Pathol Lab Med � Vol 131, January 2007

ASCO/CAP Guidelines � IHC ER/PgR Breast [2010]*

Up to 20% of current IHC determinations of ER/PR testing worldwide may be inaccurate (false negative or false positive) ER/PR on all invasive breast cancers and breast cancer recurrences No Gold Standard for ER/PR (clinically validated = clinical benefit from endocrine therapy) "Image Analysis holds promise for improving inter- and intra-observer reproducibility, but controversy exists about how imaging should be implemented at this time."

* Breast Cancer Hormone Receptor Guideline, IHC www.flagshipbio.com Hammond et al: Arch Pathol Lab Med (2010)

ASCO/CAP Guidelines � IHC ER/PgR Breast [2010]

I. Preanalytic (Tissue Preparation)
- Tissue handling
Large, preferable multiple core biopsies of tumor preferred Time from Tissue Removal to Fixation < 1 hour Oriented, inked for surgical margin assessment, and sectioned at 5 mm intervals Small portion of tumor and fibrous normal tissue together (if possible) to ensure normal breast elements are available as an int. pos. control

- Type of fixative

10% NBF (neutral buffered formalin) Alternatively � validation against NBF fixation

- Duration of fix.

Duration of Tissue Fixation > 6 hours and < 72hours Adequate volume of fixative (opt. 10x specimen)

www.flagshipbio.com

ASCO/CAP Guidelines � IHC ER/PgR Breast [2010]

II. Analytic (Staining)
- Antibodies
***Note:
Enforcing FDA clearances and approvals

Clinically-Validated Antibodies with published reports*
ER: 1D5, 6F11, SP1, 1D5+ER.2.123 PR: 1294, 1A6, 312 ***No research use only, investigational use only, nor laboratory-developed antibodies. 90% concordance ER/PR pos. / 95% concordance ER/PR neg.

*Note:
FDA clearance not sufficient e.g. PR 363

Alternatively � show concordance with a clinically validated assay

- Control samples

Pos. and Neg. Batch Control incl. intermediate reactivity Monitor assay performance **Note:
Score Percentage and Intensity ** Use of Image Analysis suggested.

On-slide external controls and internal normal epithelial elements Validate proper assay performance
www.flagshipbio.com

ASCO/CAP Guidelines � IHC ER/PgR Breast [2010]

III. Postanalytic (Interpretation)
- Interpretation
**Note:
Case for digitization of slides

Storage of slides for > 6 weeks before analysis is not recommended** Tumor containing areas of entire Slide to be reviewed Scores: Percentage (%)
(Estimation or Quantification*)

*Note:

Interpretation:

Use of Image Intensity (strong, medium, weak) Analysis encouraged! - Low percentage Optional Score - Multiple pathologists (e.g. Allred, H-Score, Quick Score) Called a "desirable method".

Pos 1% Neg < 1% Uninterpretable

Confirmatory Testing when staining does not fit clinical profile

www.flagshipbio.com

ASCO/CAP Guidelines � IHC ER/PgR Breast [2010]

III. Postanalytic (Interpretation)
- Reporting
Accessioning
Patient identification Physician identification Date of procedure Clinical indication for biopsy Specimen site and type Collection time Time sample placed in fixture Type of fixture Fixation duration

IHC Report
Patient identification* Physician identification* Date of service* Specimen site and type* Specimen identification (case and block nb) Fixative Cold ischemia time (time: removal - fixation) Duration of fixation Staining method - Primary antibody and vendor - Assay details and other reagents/vendors - References supporting validation of assay Status of FDA approval Controls (high, low-level, neg.,int. elements or normal breast tissue) Adequacy of sample for evaluation Results* - Percentage and Intensity - Interpretation - Pos., Neg., uninterpretable - Internal and external controls (pos. neg. not present) - Standard assay conditions met/not met - Optional score - Comment:
Explain reason for uninterpretable result Unusual conditions Provide correlation with histologic type of tumor

*minimum, others need to be available in lab

www.flagshipbio.com

ASCO/CAP Guidelines � IHC ER/PgR Breast [2010]

III. Postanalytic (Interpretation)
- Internal quality control and validation
Comprehensive quality control program for all aspects of the total test
- Validation of test method before reporting patient results (Re-Validation of test whenever there is a significant change to the test system) - Standard Operating Procedures with appr. elements and sign-off - Qualifications, responsibilities, and training of personnel - Proper labeling of samples and reagents - Equipment calibration, maintenance, QC, and remedial actions - Internal QA plan for entire testing process and evidence that it is followed - Quality of tests for interpretation - Ongoing competency assessment of technologists and pathologists - Report adequacy and quality - Recordkeeping for entire test process and record retention - Accurate, timely submission of results - Periodic Trend Analysis (based on patient population) - Regular, Ongoing assay reassessments at least semiannually

Mandatory Proficiency Testing (>90%) twice per year CAP Lab Accreditation � inspection every other year, self every year CAP Certification program for Pathologists
www.flagshipbio.com

ASCO/CAP Guidelines � IHC ER/PgR Breast [2010]

Definition of VALIDATION OF A TEST
... "There is no universally acceptable procedure for validating tests. The process for validating tests must take into account the purpose for which a test is intended to be used, claims made about the test, and the risks that may prevent the test from serving its intended purpose or meeting performance claims. Even FDA-approved and FDA-cleared tests require limited revalidation in clinical laboratories (a process often referred to as verification) to establish that local implementation of the test can reproduce a manufacturer's validated claims. Tests that use reagents or equipment that have not been validated (such as RUOs or IUOs) typically pose increased risks that require more extensive validation, as do tests used in more loosely controlled settings. The determination of whether a test has been adequately validated requires professional judgment."

www.flagshipbio.com

Recommendations for validating estrogen and progesterone receptor immunohistochemistry assays.*

Referenced by ASCO/CAP ER/PgR Guideline

Initial Test Validation
GOAL: Reasonable, but not absolute, assurance that a test is performing as intended. METHOD: Test Results to be compared with a standard � ideally clinical outcome, but not practical. Compare to: - Another laboratory that validated against clinical outcome - Another laboratory conforming with ASCO/CAP guidelines and having performed proper validation. - Another laboratory with an alternative, clinically validated method. - Tissue challenges used in a Proficiency-Testing Program - Validation tissue provided by an organization such as CAP or the NIST If the laboratory wants to use special techniques for scoring, such as image analysis, they must use them in the validation study.
* Validating ER and PgR Immunohistochemistry Assays Fritzgibbons et al: Arch Pathol Lab Med (in press) www.flagshipbio.com

Recommendations for validating estrogen and progesterone receptor immunohistochemistry assays.

How many tests will a lab have to do as part of the initial validation? "Five won't cut it"

FDA cleared or approved Test
40 Specimens
20 Negative (< 1%) 20 Positive ( 1%) incl. 5 Weak Positive (1-10%)
Not more than 10 specimens per run Runs on multiple days with multiple personnel

Lab developed or modified Test
80 Specimens
40 Negative (< 1%) 40 Positive ( 1%) incl. 10 Weak Positive (1-10%)
Not more than 20 specimens per run Runs on multiple days with multiple personnel

Acceptance Criteria* (with 1% cut-off) 90% Positive Percent Agreement 95% Negative Percent Agreement

Acceptance Criteria* (with 1% cut-off) 90% Positive Percent Agreement 95% Negative Percent Agreement

or use verification procedure from package insert

www.flagshipbio.com

Recommendations for validating estrogen and progesterone receptor immunohistochemistry assays.

Ongoing Assay Assessment
Monitor pos. and neg. ER rates (Trend Analysis) - compare to expectations - 2x / year Monitor concord. between ER/PR results and gene expression 95% (if performed) External Proficiency Testing Program 90% Monitor ER/PR results by pathologist � acceptable variations est. by Lab Dir. � 2x / year

Pathologist Skill Validation
(not participated in the initial test validation) 40 Specimens
20 Negative (< 1%) 20 Positive ( 1%) some Weak Positive (1-10%) Acceptance Criteria (with 1% cut-off) 2 incorrect assessments
www.flagshipbio.com

Outline
Vendor validation gap with CLIA (clinical workflows) � Medical device manufacturer regulations for DP � Clinical lab regulations under CLIA Vendor validation gap with GLP (preclinical safety studies) � State of adoption of digital slides in preclinical � Peer review & image analysis � Validation required in GLP The path forward: � SOPs for in-house GLP image analysis � IA validation service for CLIA
www.flagshipbio.com

Adoption in preclinical toxicology

ALL GLP SLIDES SCANNED

Automated flagging of compound induced abnormalities Automated detection of histology slide irregularities (e.g. rips, folds, bubbles, etc)

Scope of Usage

SLIDES REQUIRING ALGORITHMS SCANNED ABNORMAL SLIDES SCANNED

Remote quantitative IHC and morphology Quantitative morphology Quantitative IHC Peer Reviews Weekly digital slide conferences between pharma and CROs Searchable preclinical digital slide databases (abnormals only)

Telepathology (remotely controlled scopes) Digital photographs

Ranking and organizing of abnormals across studies electronically Standardization of pathologist scoring across sites and studies

Before digital slides

After digital slides
Potts Genetic Engineering News 2008: 28(15)

www.flagshipbio.com

Information on whole slide imaging in GLP
Resources for GLP
� Premarket specification to industry of what was required from digital pathology vendors in December 2007 (PDF) � Overview of GLP and IQ/OQ/PQ with digital pathology (PDF) � Invited presentation at PhRMA (PDF) (Amgen, GSK, Aperio) � Invited presentation at New York Academy of Sciences (PDF) (Biogen-Idec, Aperio) � ACVP 2008 Annual Meeting (PDF) � Poster presentation at STP (PDF) � Poster presentation at ESTP (PDF) � For an overview of the 21 CFR Part 11 and equivalent geographic regulations, please see the following: US 21 CFR Part 11 (PDF) European Union Annex 11 (English) (PDF) European Union Annex 11 (German) (PDF) Japanese ERES guidelines (Japanese) (PDF) Available at http://www.aperio.com/applications/pharma-biotech-glp.asp

www.flagshipbio.com

PhRMA presentation 2008

www.flagshipbio.com

Emerging need for peer review validation

ALL GLP SLIDES SCANNED

Automated flagging of compound induced abnormalities Automated detection of histology slide irregularities (e.g. rips, folds, bubbles, etc)

Scope of Usage

SLIDES REQUIRING ALGORITHMS SCANNED ABNORMAL SLIDES SCANNED

Remote quantitative IHC and morphology Quantitative morphology Quantitative IHC Peer Reviews Weekly digital slide conferences between pharma and CROs Searchable preclinical digital slide databases (abnormals only)

Telepathology (remotely controlled scopes) Digital photographs

Ranking and organizing of abnormals across studies electronically Standardization of pathologist scoring across sites and studies

Before digital slides

After digital slides
Potts Genetic Engineering News 2008: 28(15)

www.flagshipbio.com

VIPER: Validation of peer reviews
Flagship Biosciences is leading a consortium of 6 pharmas and one medical device pathologist participants to look at workflow and validation of whole slide images in peer reviews � Workflow � Pathologist time versus glass slides � 20x versus 40x? � Randomization and blinding strategies � Validation approaches � International bandwidth issues Kickoff meeting ACVP Dec 2010 Next meeting STP June Donated primate study with full dual peer review
www.flagshipbio.com

Collaboration with EPL in peer reviews

www.flagshipbio.com

Emerging use of image analysis in GLP
ALL GLP SLIDES SCANNED Automated flagging of compound induced abnormalities Automated detection of histology slide irregularities (e.g. rips, folds, bubbles, etc)

Scope of Usage

SLIDES REQUIRING ALGORITHMS SCANNED ABNORMAL SLIDES SCANNED

Remote quantitative IHC and morphology Quantitative morphology Quantitative IHC Peer reviews Weekly digital slide conferences between pharma and CROs Searchable preclinical digital slide databases (abnormals only)

Telepathology (remotely controlled scopes) Digital photographs

Ranking and organizing of abnormals across studies electronically Standardization of pathologist scoring across sites and studies

Before digital slides

After digital slides
Potts Genetic Engineering News 2008: 28(15)

www.flagshipbio.com

Image analysis quantitation in preclinical

Periarteriolar lymphoid tissue (green) and red pulp (red) in a mouse spleen.

Thymus cortex to medulla ratio analysis

Courtesy Frank Voelker and Rob Diller, www.flagshipbiosciences.com www.flagshipbio.com

Vendor validation gap with GLP
For GLP environment, some vendors supply IQ/OQ/PQ validation process � This does not cover image analysis Emerging area is use of IA on toxicology studies, to supplement historic minimal/mild/moderate/marked qualitative scoring with computer assisted scoring � Both area and cell based, highly diverse solutions See Flagship's image analysis blog at www.flagshipbio.com No out-of-the-box algorithms from any vendor, the problem is too diverse

www.flagshipbio.com

21 CFR 11 Functionality
Category Details

System logins and passwords

Timeouts, password rules enforcement, administrator privileges, etc. Configurable roles (pathologist can sign off, technician can scan and QA slides, etc.) Images and bar codes cannot be tampered with, 1D & 2D barcoding. IMAGE SIGNATURE Application and file level logging

Access control and user privileges

Auto detection of data tampering

Comprehensive audit log

Electronic signature records

Complete sign-off workflow

www.flagshipbio.com

IQ/OQ/PQ validation
IQ: Have all system components been installed correctly?

OQ: Do system components operate correctly?

PQ: Does the entire system perform as expected?

www.flagshipbio.com

SOP for pattern recognition
Write a workflow for your image analysis process Then re-write as an SOP Validation steps should follow the SOP steps

Workflow courtesy Dave Young, Flagship Biosciences www.flagshipbio.com

Tissue variability and acceptable error

Tissue variability & acceptable error

Find out what your manual accepted error rate is before digital slides and IA Estimate your error rates on image analysis projects, record them, make it part of the SOP

Is it better to perfectly analyze small "representative" areas or accept a level of error and run the entire tissue section? � Measure intratissue variability rates

www.flagshipbio.com

Measuring intratissue variability rates

Tissue variability & acceptable error

Courtesy Frank Voelker, Quantitation in Toxicologic Pathology, STP meeting 2008 www.flagshipbio.com

Example intratissue variability rates

Tissue variability & acceptable error

www.flagshipbio.com

Emerging need for stereology
Random 2D sampling is not enough Europe is leading in this area Stereology will be required in GLP settings and is already moving to the clinic in Europe

See Bob Dunstan's webinar (Biogen-Idec) on use of stereology (June 9,2010) � www.visiopharm.com

www.flagshipbio.com

Slide Storage
If you use image analysis, you will need to treat the slide as raw data.

Dunstan, Potts, NYAS Preclinical GLP 2007 www.flagshipbio.com Pyrrah, Blackmer, Potts, PhRMA Preclinical Session 2007

Outline
Vendor validation gap with CLIA (clinical workflows) � Medical device manufacturer regulations for DP � Clinical lab regulations under CLIA Vendor validation gap with GLP (preclinical safety studies) � State of adoption of digital slides in preclinical � Peer review & image analysis � Validation required in GLP The path forward: � Preclinical: SOPs for in-house GLP image analysis � Clinical and clinical trials: IA validation service for CLIA
www.flagshipbio.com

Flagship training and IA validation services
Requires a close dialog between image analysis experts and pathologists willing to embrace image analysis

Completed a virtual training trial run: 6 week session (two hours each Thursday) on image analysis training (led by pathologists and IA experts)
� http://www.flagshipbio.com/products-page/virtual-instructor-led-training/

www.flagshipbio.com

Flagship training and IA validation services
CLIA image analysis validation service 1. Multi-week session (two hours each Thursday on regulated image analysis concepts) 2. 3rd party vendor independent validation of image analysis in a regulated setting 3. Virtual exchange of digital slides and image analysis results between CLIA labs Guidance and Advice: Dave Eberhard, MD (UNC / LabCorp) Steve Schmechel, MD (Univ of Minnesota) Hadi Yaziji, MD (Vitro Molecular)
Yaziji, H. et al. Appl Immunohistochem Mol Morphol. 2008 Dec;16(6):513-20. Consensus recommendations on estrogen receptor testing in breast cancer by immunohistochemistry.

Please contact pathservices@flagshipbio.com for further information

www.flagshipbio.com

Conclusions
We've reviewed the vendor validation gap with CLIA (clinical workflows) � Medical device manufacturer regulations for DP � Clinical lab regulations under CLIA � NEXT STEPS: IA validation service We've reviewed the vendor validation gap with GLP (preclinical safety studies) � State of adoption of digital slides in preclinical � Peer review & image analysis � NEXT STEPS: In-house SOPs for GLP IA, Stereology

www.flagshipbio.com

Acknowledgements
Dr. Holger Lange Ole Eichhorn and clearances team at Aperio Dr. David Young, Frank Voelker, Trevor Johnson, Flagship Dr. Robert Dunstan, Biogen-Idec Dave Eberhard, MD (UNC / LabCorp) Steve Schmechel, MD (Univ of Minnesota) Hadi Yaziji, MD (Vitro Molecular) Neil Gray, AstraZeneca

www.flagshipbio.com

Subscribe
x