Roche uPath RUO User manual

Brand: Roche

Model: uPath RUO

Type: User manual

uPath PD-L1 (SP263) Image Analysis

for Non-Small Cell Lung Cancer

Algorithm Guide (RUO)

Table of Contents

Introduction 1

Algorithm Summary and Explanation 2

Intended Use 3

Intended Use of product 3

Purpose of Algorithm Guide 3

Test Principles 4

Limitations 5

Data Security 6

Workﬂow for using Image Analysis Algorithm 7

Pathologist Workﬂow 8

PD-L1 Image Analysis 12

Staining Characteristics 15

PD-L1 Evaluation 15

Troubleshooting 26

Introduction

Note: uPath PD-L1 (SP263) image analysis for NSCLC is an

adjunctive computer-assisted methodology to aid in the

acquisition and measurement of images from microscope glass

slides of tissue specimens IHC-stained for the presence of PD-L1

protein. To assure the validity of image analysis scores, it is the

responsibility of the pathologist to verify agreement by employing

appropriate controls as speciﬁed in the instructions for the

VENTANA PD-L1 (SP263) Assay.

The Roche uPath RUO enterprise software (uPath RUO

enterprise software) with the uPath PD-L1 (SP263) image

analysis for Non-Small Cell Lung Cancer ((uPath PD-L1 (SP263)

image analysis for NSCLC)) is a software system designed

to assist in the consistent quantitative assessment of protein

expression in immunohistochemically (IHC) stained histologic

sections from formalin-ﬁxed, parafﬁn-embedded (FFPE) normal

and neoplastic tissues. The Roche uPath RUO enterprise

software can be used for review of digitized images of histologic

sections with image analysis algorithms, or without image

analysis algorithms (Digital Read applications). Digital Read

applications present images on the computer screen in the same

manner as one would see with a manual microscope, inclusive

of the pathologist’s ability to select any areas of interest and the

option of various magniﬁcation levels.

For the image analysis applications, the pathologist may use

the system software to select and outline one or several regions

of interest (ROIs), and each ROI may be viewed at various

magniﬁcations and then analyzed by the software. A count of the

total number of target cells and the number interpreted by the

algorithm as positive and negative is generated. The uPath PD-

L1 (SP263) image analysis for NSCLC reports the PD-L1 Tumor

Cell (TC) score (0-100%) for a particular ROI, or an aggregate

score for all selected ROIs for that slide. The pathologist can

accept the score provided by the algorithm, or may override the

score with a different score. The system requires competent

human intervention at all steps in the analysis process, and the

software makes no independent interpretations of the data. The

Roche uPath RUO enterprise software consists of image analysis

algorithms and software with a user interface.

Roche uPath RUO enterprise software is an end-to-end digital

pathology software solution that allows pathology laboratories

to acquire, manage, view, analyze, share, and report on digital

images of pathology specimens. Using the uPath RUO enterprise

software, the pathologist can view digital images at various

magniﬁcations (as previously stated), add annotations, make

tissue section measurements, perform image analysis, and

generate reports.

uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide 1

2 uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide

Algorithm Summary and Explanation

The Roche uPath RUO enterprise software with the uPath

PD-L1 (SP263) image analysis for NSCLC is a software system

designed to assist the pathologist in the consistent quantitative

assessment of protein expression in IHC stained histologic

sections from FFPE normal and neoplastic tissues. The Roche

uPath enterprise software can be used for review of digitized

images of histologic sections with image analysis algorithms, or

without image analysis algorithms (Digital Read applications).

Digital Read applications present images on the computer

screen in the same manner as one would see with a manual

microscope, inclusive of the pathologist’s ability to select any

areas of interest and the option of various magniﬁcation levels.

For the image analysis applications, the pathologist may use

the system software to select and outline one or several regions

of interest (ROIs), and each ROIs may be viewed at various

magniﬁcations and then analyzed by the software. A count of

the total number of target cells and the number interpreted by

the algorithm as positive and negative is generated. The uPath

PD-L1 (SP263) image analysis for NSCLC also reports the PD-L1

TC score (0-100%) for a particular ROI, or an aggregate score

for all selected ROIs for that slide. The pathologist can accept

the score provided by the algorithm, or may override the score

with a different score. The system requires competent human

intervention at all steps in the analysis process, and the software

makes no independent interpretations of the data. The Roche

uPath RUO enterprise software consists of image analysis

algorithms and software with a user interface.

Roche uPath RUO enterprise software is an end-to-end digital

pathology software solution that allows pathology laboratories

to acquire, manage, view, analyze, share, and report on digital

images of pathology specimens. Using the uPath RUO enterprise

software, the pathologist can view digital images at various

magniﬁcations (as previously stated), add annotations, make

tissue section measurements, perform image analysis, and

generate reports.

Intended Use of product

The Roche Digital Pathology Solution provides image acquisition

and image analysis applications. The following automated image

analysis application is included with the Roche Digital Pathology

Solution:

The uPath PD-L1 (SP263) image analysis, Non-Small Cell

Lung Cancer (NSCLC) is an adjunctive computer-assisted

methodology that supports the detection and measurement of

PD-L1 protein in formalin-ﬁxed, parafﬁn-embedded normal and

neoplastic tissue.

For Research Use Only. Not for use in diagnostic procedures.

Purpose of Algorithm Guide

This guide is intended to:

• Background information on the intended use of the algorithm,

limitations, and how the algorithm works

• Present Performance Characteristics.

• Show Step-wise directions for running uPath PD-L1 (SP263)

image analysis for NSCLC.

• Provide pathologists with a tool to facilitate the use of the uPath

PD-L1 (SP263) image analysis for NSCLC on FFPE NSCLC

sections stained with the VENTANA PD-L1 (SP263) Assay.

• Provide photographic images that illustrate how uPath PD-L1

(SP263) image analysis for NSCLC should be used.

• Provide example images of challenging cases to provide

guidance in how to use the algorithm in their evaluation.

Intended Use

uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide 3

The Roche uPath RUO enterprise software with uPath PD-L1

(SP263) image analysis for NSCLC employs image analysis

techniques to obtain PD-L1 TC positive scores.

The PD-L1 image analysis algorithms use pre-deﬁned

parameters to score PD-L1 images. The steps involved in image

analysis are:

• Identiﬁcation of white space and automatic exclusion from

analysis.

• Detection of cell locations in the entire image.

• Classiﬁcation of cells as tumor cells or other cell types.

• Identifying stained versus unstained tumor cells.

• Computing the stained TC percentage score which is the

percentage of tumor cells that are stained.

How PD-L1 image analysis algorithm identiﬁes tumor cells and

how the score is calculated:

• The algorithm identiﬁes tumor cells using color, intensity, size

and morphological features.

• Identiﬁed tumor cells are classiﬁed as stained using a pre-set

threshold that is in line with the assay method sheet.

• For the calculation of percentage of stained tumor cells, the

number of stained tumor cells is divided by the total number of

tumor cells.

PD-L1 image analysis algorithms’ white space determination:

• The algorithms will automatically exclude white space.

Artifacts such as dirt, scratches or ink may not be automatically

excluded.

• The white space areas determined by the algorithm may be

reviewed by the user using a false color overlay; see “PD-L1

Image Analysis: False Color Overlay” on page 13.

PD-L1 image analysis algorithms score:

• The algorithms will report out a score to one digit in the tenths

place, e.g. “4.8%”.

• It is not recommended to round up or down the score. For

example, “4.8%” should not be rounded up to 5%.

Test Principles

4 uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide

The algorithms are designed to work for VENTANA PD-L1

(SP263) Assay. The test results are only as good as the quality

and accuracy of the IHC slide that is imaged, and the subsequent

image that is analyzed.

The pathologist must validate the VENTANA PD-L1 (SP263)

Assay staining run by manual microscopic examination of the

PD-L1 control slides to verify that the expected results have been

obtained before images from patient slides are acquired by the

uPath RUO enterprise software.

The pathologist must follow the manufacturer’s

recommendations for the VENTANA PD-L1 (SP263) Assay

including using all the positive and negative quality control

materials for each staining run.

If the control slides are not acceptable with manual

microscopic examination, the patient tissues need to be re-

stained with acceptable results. (See the VENTANA PD-L1

(SP263) Assay method sheet for details about quality control

recommendations.)

The pathologist must follow the recommendations for VENTANA

PD-L1 (SP263) Assay interpretation as recommended in the

VENTANA PD-L1 (SP263) Assay method sheet (available at www.

ventana.com).

If the images captured have different staining (nuclear,

cytoplasm, etc.), incorrect results will be generated.

uPath PD-L1 (SP263) image analysis for NSCLC will reject nuclei

that are elongated regardless of the overall shape of the cell.

For this reason, tumors containing large numbers of cells with

elongated nuclei may need to be evaluated manually.

Although it is required for macrophages to be excluded from

the region of analysis, it is not always possible to capture all

macrophages. Thus, the algorithm result is inﬂuenced by

remaining macrophages in the region of analysis. This is critical

when a patient score is close to the cutoff (either 1% or 50%).

In addition, performance of the Roche uPath RUO enterprise

software for uPath PD-L1 (SP263) image analysis for NSCLC with

the following types of lung cancers has not been evaluated: small

cell carcinoma, fetal adenocarcinoma, enteric adenocarcinoma,

squamous cell carcinoma in situ (SCIS), combined small cell

carcinoma, large cell neuroendocrine carcinoma, typical

carcinoid tumor, atypical carcinoid tumor, pleomorphic

carcinoma, spindle cell carcinoma, giant cell carcinoma,

carcinosarcoma, pulmonary blastoma, lymphoepithelioma-

like carcinoma, NUT carcinoma, mucoepidermoid carcinoma,

adenoid cystic carcinoma, and epithelial-myoepithelial

carcinoma.

This device has not been tested, or its safety and effectiveness

validated, when used with a personal computer (PC) from home.

According to the 1988 Clinical Laboratory Improvement

Amendments (CLIA ‘88), each laboratory that introduces

an FDA cleared system must demonstrate that it can obtain

performance speciﬁcations comparable to those established by

the manufacturer.

Limitations

uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide 5

Data loss or unavailability of the instrument due to malicious

software or unauthorized instrument access.

Malicious software or unauthorized instrument access can result

in data loss or instrument unavailability.

To avoid infection by malicious software or the unauthorized

access and misuse of the instrument, the following

recommendations are essential:

• Do not install or run any other software on the instrument.

• Make sure that other computers and services on the network

are properly secured and protected against malicious software

and unauthorized access. For example, the laboratory

information system (LIS), archiving share, backup share, or

service.

• Customers are responsible for the security of their local area

network, especially in protecting it against malicious software

and attacks. This protection might include measures, such as

a ﬁrewall, to separate the device from uncontrolled networks.

This protection might also include measures that ensure that the

connected network is free of malicious code.

• Restrict physical access to the instrument and all attached IT

infrastructure (computers, cables, network equipment, and so

on).

• Make sure that instrument backup and archive ﬁles are

protected from any unauthorized access and disaster. This list

includes the remote storage location, disaster discovery sites,

and the secure transfer of backup ﬁles.

• If possible, use a ﬁrewall to restrict network trafﬁc.

USB ﬂash drives can be used for several kinds of backups and

restores. Wrong handling of a USB ﬂash drive may result in data

loss or malfunction of the instrument.

• Use only USB ﬂash drives that are tested and installed by your

local Roche service representative.

• At any one time only one USB device can be in use. Before

inserting a USB ﬂash drive, check that no other USB device is

inserted.

• Before removing a USB ﬂash drive, choose the Eject button in

Windows.

• To prevent a virus from infecting the software, use the USB ﬂash

drive exclusively on the instrument. Do not store other data on

this USB ﬂash drive.

Data Security

6 uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide

Workﬂow for using Image Analysis Algorithm

Materials needed

• Roche uPath RUO enterprise software

• uPath PD-L1 (SP263) image analysis for NSCLC

• NSCLC slides stained with the VENTANA PD-L1 (SP263) Assay (using OptiView Detection Kit) on the BenchMark ULTRA instrument

• VENTANA DP 200 slide scanner (slide scanner)

Workﬂow

1. VENTANA PD-L1 (SP263) Assay is performed using a BenchMark ULTRA instrument on a glass slide of a patient’s NSCLC tissue.

2. Image acquisition (whole slide scanning) is performed using the slide scanner at 20x magniﬁcation at one z-plane.

3. Once digital images are acquired, these images are transferred/uploaded from the computer associated with the VENTANA DP 200

slide scanner to the image management system (IMS) on a centralized server. The centralized server may be a physical server or a

Virtual Machine. This server also houses the uPath RUO enterprise software.

4. Following transfer to the server, a case will be created in the software. Case creation can occur automatically using identiﬁcation

information (i.e., tissue type and primary antibody) contained in the barcode label of the glass slide or entered manually into the

software (refer to Roche uPath RUO enterprise software User Guide (PN 1018184EN)).

5. If uPath PD-L1 (SP263) image analysis for NSCLC is installed and a 20x image is accessioned with the appropriate stain and tissue

type, the software then automatically triggers Whole Slide Analysis (WSA).

6. WSA automatically analyzes the entire scanned image.

7. Once WSA is completed with the computation on the entire slide, the pathologist is notiﬁed within the software that “analysis is

complete.” The pathologist is able to select speciﬁc ROIs to score. A pathologist may select a single ROI of any size or multiple ROIs. If

multiple ROIs are selected, a composite score is provided as well as individual scores for each ROI.

Staining

• Tissue preparation and staining should follow the recommendations provided in the VENTANA PD-L1 (SP263) Assay method sheet.

• The uPath PD-L1 (SP263) for NSCLC requires use of the VENTANA PD-L1 (SP263) Assay, and any additional material or supplies listed

in the VENTANA PD-L1 (SP263) Assay method sheet, to stain tissues prior to analysis. The VENTANA PD-L1 (SP263) Assay detects

PD-L1 protein in FFPE NSCLC tissue stained with OptiView DAB IHC Detection Kit on a BenchMark ULTRA instrument.

Image Capture

The VENTANA DP 200 slide scanner is required for scanning of the slides. Images are required to be scanned at 20x magniﬁcation. It is

recommended that the tissue be free of folds and ink. If large sections of the image are out of focus, it is recommended that the slides

be rescanned. For further information on scanning, please refer to the VENTANA DP 200 slide scanner User Guide (PN 1016906EN).

General Navigation: uPath RUO enterprise software

Please refer to Roche uPath RUO enterprise software User Guide for more information regarding the uPath RUO enterprise software

interface. uPath RUO enterprise software is meant to be customizable to individual and site needs including but not limited to: report

conﬁguration, user interface. Within this algorithm guide we will focus on the tools necessary for using the uPath PD-L1 (SP263)

image analysis for NSCLC only.

uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide 7

Pathologist Workﬂow

Opening a Case

Images of NSCLC stained with VENTANA PD-L1 (SP263) Assay can be accessed by double-clicking on a case (Figure 1);

Or by selecting a case and pressing the viewer tab within uPath RUO enterprise software (Figure 2).

A screen with all images associated with a case will appear (Figure 3).

Figure 1

Figure 3

Figure 2

8 uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide

Once a glass slide stained with the VENTANA PD-L1 (SP263) Assay is scanned on a VENTANA DP 200 slide scanner at 20x, the

image is imported into the software and associated with a case. The algorithm will automatically begin performing the WSA. Times to

complete the WSA precomputing step depend on server speciﬁcations, image sizes and number of images in the queue. “waiting to

start auto analysis” speciﬁes the images are in queue and yet to be analyzed and “analyzing” is used when WSA is being performed

(Figures 4 & 5).

Once the image is completely analyzed via WSA in the software, “analysis successful” will be displayed underneath the slide image

within the viewer in the uPath RUO enterprise software (Figure 6). Images cannot be scored prior to successful WSA completion.

Figure 6

Figure 5Figure 4

uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide 9

Drawing Whole Tumor ROI(s): Selecting Tumor Area

Use the Freehand tool button within the ROI dropdown menu to select the tumor area(s) on the IHC slide image (Figure 7) to be

analyzed. Figure 8 illustrates an image that has a single ROI drawn. Additional ROIs can be drawn. Each area selected will result in a

ROI to appear in the SLIDE PANEL (Figure 9).

When drawing the ROIs, the exclusion of necrotic tissue, ink, stained tissue folds and other artifacts is recommended. Speciﬁc areas to

be avoided or omitted as well as examples will be addressed later in this algorithm guide.

Figure 8

Figure 9

Figure 7

10 uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide

Drawing Whole Tumor ROI(s): Exclusion Area

Use the Freehand exclusion tool within the Exclusion dropdown menu (Figure 10) to exclude areas of necrotic tissue, ink, stained

tissue folds and other artifacts within the previously selected whole tumor ROI. Speciﬁc areas to be avoided or omitted as well as

examples of these areas will be addressed later in the algorithm guide (Figure 11). If large areas of the image are blurry or out of focus,

rescan the slide.

Excluded areas will not be analyzed by the algorithm, and the stained and unstained tumor cells within this area will be excluded

from the total analysis area. If the ROI has already been analyzed, and an exclusion is used, the ROI will need to be reanalyzed and the

overlay and score will be updated appropriately.

Drawing Whole Tumor ROI(s): Deletion

If a selected whole tumor ROI is not optimal, it can be deleted. Select the whole tumor ROI by clicking on the center of the ROI on the

image and then clicking the Delete button within the SLIDE PANEL or within the slide image near the ROI. A conﬁrmation window will

appear. Select Conﬁrm to delete the selected ROI. Select Cancel to retain the ROI.

Figure 11

Figure 10

uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide 11

Once all whole tumor ROI(s) and/or exclusion areas have been drawn, the image is ready to be analyzed. Select the whole tumor ROI

by clicking on the center of the ROI to be analyzed or clicking on the ROI in the SLIDE PANEL. For each ROI, select the Image Analysis

button either within the SLIDE PANEL (Figure 12) or next to the ROI.

Once the PD-L1 Analysis is complete, the result will appear in the SLIDE PANEL in two locations: under Slide Score and next to the

ROIs (Figure 13). The slide score is based on a summation of the tumor cell positivity status across all selected ROI(s), which is the

score that will appear in the report.

You can also see more detailed information in the SLIDE SCORE ﬂipout and the ROI DETAILS Flip-out by clicking the pencil or word

bubble icon. The information displayed here can be conﬁgured in the administrator settings (Figure 14). Please see Roche uPath RUO

enterprise software User Guide for how to conﬁgure the information within the SLIDE SCORE Flip Out.

Figure 12

Figure 14

PD-L1 Image Analysis

Figure 13

12 uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide

PD-L1 Image Analysis: False Color Overlay

Once all ROI(s) Analysis button has been pressed and analyzed, a color overlay will be displayed on the ROI (Figure 15).

In the images below (Figure 16), red dot represent tumor cells the algorithms predicts to be positively stained and blue represents the

negative tumor cells identiﬁed by the image analysis algorithm. The yellow represents areas with no tissue.

When grabbing the image (pressing the left mouse button and moving the image) the overlay will disappear. When the mouse button

is no longer pressed, the overlay reappears.

Figure 15

Figure 16: VENTANA DP 200 slide scanner images of VENTANA PD-L1 (SP263) Assay-stained NSCLC tissue in uPath RUO enterprise

software after analysis. Red overlay represents cells determined to be positively stained tumor cells; blue overlay represents cells

determined to be negatively stained tumor cells; yellow overlay represents large areas without tissue.

uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide 13

Manual Override Slide Score Entry

Scores can be manually overridden by clicking on the pencil icon within the SLIDE PANEL next to the Slide score.

The SLIDE SCORE ﬂip-out will appear (Figure 17). On the SLIDE SCORE ﬂip-out, select the EDIT button. The SLIDE SCORE ﬂip-out

now allows the user to type in a manual score (Figure 18). The Comments ﬁeld allows notes regarding the case and/or decision to

override the automated score. For PD-L1, scores of 0-100% may be manually entered. Upon entry of a Manual Override score, click on

the CONFIRM notiﬁcation. A conﬁrmation message will appear, press “Yes.”

The score within the SLIDE PANEL will reﬂect the manual override score. The user will have the option to re-analyze the image by

pressing the bar graph button. You will also notice that the image analysis provided score next to the ROI(s) will not be present (Figure

19).

Figure 18Figure 17

Figure 19

14 uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide

Please See VENTANA PD-L1 (SP263) Assay method sheet (PN 1014258EN) and Interpretation Guide (PN 1015317EN).

PD-L1 Evaluation

NSCLC neoplastic cells labeled with the VENTANA PD-L1 (SP263) Assay are evaluated for percent positivity of the tumor cells

with membrane staining at any intensity of the diaminobenzidine (DAB) signal. The immunohistochemical staining in NSCLC is

membranous and/or cytoplasmic, and may be expressed homogeneously or heterogeneously throughout the neoplasm. Membrane

staining can have a discontinuous, circumferential or basolateral pattern.

Staining Characteristics

Figure 20: Scanned images of IHC stained NSCLC tissue in uPath RUO enterprise software; before (above) analysis and after (below)

analysis. Red overlay represents cells determined to be positively stained tumor cells and blue overlay represents cells determined to

be negatively stained tumor cells.

uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide 15

Cytoplasmic staining is generally diffuse with some cases displaying a ﬁnely granular quality. Rare cases have shown a peri-nuclear

dot-like body staining with variable intensity. The total percentage of tumor membrane signal intensities is visually estimated and used

to generate the PD-L1 expression level. Tumor cell cytoplasmic staining is disregarded for determining PD-L1 expression. An isotype-

matched negative control antibody is used to evaluate the presence of background in test samples and establish a staining intensity

baseline.

Figure 21: Scanned images of IHC stained NSCLC tissue in uPath RUO enterprise software before (above) analysis and after (below)

analysis. Red overlay represents cells determined to be positively stained tumor cells and blue overlay represents cells determined to

be negatively stained tumor cells.

16 uPath PD-L1 (SP263) image analysis for Non-Small Cell Lung Cancer Algorithm Guide

Page is loading ...

Roche uPath RUO User manual

Roche uPath RUO User manual

Roche uPath RUO User manual

Related papers

Other documents