Fit-For-Purpose PD-L1 Biomarker Testing For Patient Selection in Immuno-Oncology: Guidelines For Clinical Laboratories From the Canadian Association of Pathologists-Association Canadienne Des Pathologistes (CAP-ACP).

Since 2014, programmed cell death protein 1 (PD-1)/programmed cell death ligand 1 (PD-L1) checkpoint inhibitors have been approved by various regulatory agencies for the treatment of multiple cancers including melanoma, lung cancer, urothelial carcinoma, renal cell carcinoma, head and neck cancer, classical Hodgkin lymphoma, colorectal cancer, gastroesophageal cancer, hepatocellular cancer, and other solid tumors. Of these approved drug/disease combinations, a subset also has regulatory agency-approved, commercially available companion/complementary diagnostic assays that were clinically validated using data from their corresponding clinical trials. The objective of this document is to provide evidence-based guidance to assist clinical laboratories in establishing fit-for-purpose PD-L1 biomarker assays that can accurately identify patients with specific tumor types who may respond to specific approved immuno-oncology therapies targeting the PD-1/PD-L1 checkpoint. These recommendations are issued as 38 Guideline Statements that address (i) assay development for surgical pathology and cytopathology specimens, (ii) reporting elements, and (iii) quality assurance (including validation/verification, internal quality assurance, and external quality assurance). The intent of this work is to provide recommendations that are relevant to any tumor type, are universally applicable and can be implemented by any clinical immunohistochemistry laboratory performing predictive PD-L1 immunohistochemistry testing.

INTRODUCTION AND BACKGROUND

Programmed Cell Death Ligand 1 (PD-L1) Biology, Distribution in Normal Tissues, and Role in Immune Surveillance

The PD-L1 protein is encoded by the human CD274 gene, located on the short arm of chromosome 9. It was first identified in 1999 based on a homology search of the putative functionally related molecules B7-1/2.1 In 2000, programmed cell death protein 1 (PD-1) was identified as the receptor for the newly cloned PD-L1.2 In vivo loss of function studies performed on knockout mice showed that PD-L1 negatively regulates T-cell function.3 Early in situ protein expression by immunohistochemistry (IHC) on frozen tissues showed expression in macrophages.4 Subsequently, IHC on formalin-fixed paraffin-embedded tissues showed that while PD-L1 expression in benign tissues may be limited to hematolymphoid cells and placenta, it is expressed by a number of different epithelial, mesenchymal, neuroectodermal, hematopoietic, lymphoid, and germ cell tumors.5,6

The Evolution of PD-L1 as a Predictive Biomarker

PD-L1 has been evaluated as either a prognostic or a predictive biomarker in various malignant tumors.7–16 In contrast to previous immunotherapy checkpoint inhibitors, such as the anti-CTLA-4 drug ipilimumab,17 the use of the anti-PD-1 inhibitor pembrolizumab in advanced non–small cell lung cancer (NSCLC) became the first immunotherapy drug to require PD-L1 biomarker testing (using a clinically validated United States’ Food and Drug Administration (FDA)–approved IHC companion diagnostic assay) to determine patient eligibility based on results of the Keynote-001 trial, which demonstrated that the response to pembrolizumab is positively related to the level of PD-L1 expression.18 The Keynote 010 trial shifted the tumor proportion score (TPS) cutoff point for pembrolizumab as a second line treatment in NSCLC from 50% to 1%,19 whereas the Keynote 024 trial reintroduced the TPS≥50% in the first-line setting.20 Subsequent to NSCLC, other tumors that have PD-L1 companion diagnostic testing requirements include gastric/gastroesophageal junction adenocarcinoma (pembrolizumab), cervical carcinoma (pembrolizumab), urothelial carcinoma (pembrolizumab and atezolizumab), breast cancer (atezolizumab), and head and neck squamous cell carcinoma (pembrolizumab). Tumors that have complementary PD-L1 testing assays approved by the FDA include NSCLC (nivolumab, atezolizumab), head and neck squamous cell carcinoma (nivolumab), urothelial carcinoma (nivolumab, durvalumab, atezolizumab), melanoma (nivolumab). Real-world outcomes of patients with metastatic NSCLC treated with PD-1 inhibitors in the first year following United States regulatory approval indicate that PD-L1 expression is associated with better overall survival.21

The PD-L1 Companion/Complementary Diagnostic Testing Landscape

Currently, there are 5 PD-1/PD-L1 inhibitors that are approved by the FDA for the treatment of various cancers including melanoma, lung cancer, urothelial carcinoma, renal cell carcinoma, head and neck cancer, classical Hodgkin lymphoma, colorectal cancer, gastroesophageal cancer, hepatocellular cancer, and other solid tumors.22 Of these approved drug/disease combinations, a subset has FDA-approved, commercially available companion or complementary diagnostic assays that were clinically validated using data from their corresponding clinical trials or from related clinical studies. Although PD-L1 biomarker status to determine eligibility for pembrolizumab in NSCLC has significance for clinical laboratories as being the first companion diagnostic assay in immuno-oncology,18 the overall landscape suggests that PD-L1 testing will continue to expand.23–28 This likelihood is augmented by the possibility that (irrespective of the FDA) regulatory or funding agencies in countries other than the United States, may require demonstration of “PD-L1 positivity” by a properly validated IHC assay to approve or fund the use of anti-PD-L1/PD-1 therapies for specific patients as higher PD-L1 expression has been positively associated with outcome even in those drug-disease indications that currently do not require PD-L1 testing.29–34

Implementing PD-L1 Testing: Challenges For Clinical Laboratories

There are multiple different commercially available PD-L1 IHC companion/complementary diagnostic assays (PD-L1 IHC Kits). These PD-L1 assays were developed and clinically validated by clinical trials35–41 for different drug-disease combinations, using different PD-L1 primary antibody (Ab) clones on different IHC platforms with different IHC protocols and requiring different readout criteria for what is considered to be “positive” in different disease contexts. PD-L1 IHC Kits are reasonably considered reference standard assays for their respective drug-disease indications based on the Clinical and Laboratory Standards Institute (CLSI) guidelines for determining the diagnostic accuracy of qualitative assays. However, because the analytical sensitivity and specificity (including the acceptable threshold range relevant to low limit of detection) are not precisely defined for these assays, the question of whether or not the different assays can be used in place of one another (ie, are interchangeable) or whether laboratory-developed tests (LDTs) can be successfully validated, inevitably arises.

These questions represent significant challenges to clinical laboratories charged with the task of providing meaningful, reliable, and informative PD-L1 testing results in an environment where such results have an impact on eligibility for therapy and where oncologists and patients, rightfully, have high expectations of test accuracy. From the laboratory perspective, the challenges in establishing fit-for-purpose PD-L1 testing are multifaceted with biological, technical, interpretive, and regulatory factors coming together to form what amounts to an empirical gauntlet for laboratories. As such, there is an emerging need for guidance in this matter, a clear roadmap for the selection, validation/verification and reporting of PD-L1 IHC assays that are fit-for-purpose and evidence-based.

OBJECTIVE

The scope of these Canadian Association of Pathologists-Association Canadienne Des Pathologistes (CAP-ACP) recommendations is to provide guidance to assist clinical laboratories in the setup and implementation of fit-for-purpose PD-L1 biomarker assays that can accurately identify patients with specific tumor types who may respond to specific Health Canada-approved immuno-oncology therapies targeting the PD-1/PD-L1 checkpoint. These recommendations are issued as guideline statements (GSs) that address (i) assay selection and development for surgical pathology and cytopathology specimens, (ii) reporting elements, and (iii) quality assurance [including validation/verification, internal quality assurance, and external quality assurance (EQA)]. The intent of this work is to provide recommendations that are relevant to any tumor type, are universally applicable and can be implemented by any clinical IHC laboratory performing predictive PD-L1 IHC testing.

These recommendations do not apply to (and should not inform) drug selection by oncologists for any patient populations; rather, drug selection for specific disease indications is relevant to these recommendations only to the extent that the drug-disease combinations impact the selection and implementation of an appropriate fit-for-purpose PD-L1 testing strategy. Importantly, these recommendations are not designed to address specific IHC protocols or protocol components. Similarly, they do not address the issues of whether PD-L1 testing should be performed or the context(s) in which PD-L1 testing may be informative.

The GSs address 5 key questions:

Which PD-L1 assay(s) should be used to predict response to anti-PD-1/PD-L1 immunotherapies?

What is the quality of statistical methodologies employed to evaluate PD-L1 assay performance in interchangeability assessments?

Were specific diagnostic assays (IHC protocol conditions and specific readout) used and stated by clinical trials where a specific drug and a specific disease were evaluated?

How should the results of predictive PD-L1 assays be reported?

What measures/practices are necessary to ensure the quality of PD-L1 testing for patient selection in immunotherapy?

METHODOLOGY

The methodology is detailed in the Supplementary Files (Supplemental Digital Content 1, http://links.lww.com/AIMM/A241). Systematic and targeted review of published evidence, grading of evidence, development of recommendations, and grading of recommendations was done according to published guidelines with some modifications 42–53 in relation to “implementability” of the recommendations to daily practice in clinical IHC laboratories.54 A systematic review of published literature for key questions 1 and 2 resulted in the initiation of a meta-analysis of published test comparisons with emphasis on diagnostic accuracy being a principal criterion for test interchangeability; see Supplementary Materials—Key Questions (Supplemental Digital Content 1, http://links.lww.com/AIMM/A241).

HOW TO USE THIS DOCUMENT

The main outputs of this initiative are the 38 GSs, which are found in the section below and summarized in Table 1. Each GS is accompanied by an explanatory note where the authors felt it was necessary.

TABLE 1

Guideline Statements

TERMINOLOGY

Accuracy

The closeness of a single result of a measurement and a true value.55,56

Biomarker

A physiological analyte that is objectively measured and evaluated as an indicator of normal biological and pathogenic processes or pharmacological responses to a specified therapeutic intervention.56

Candidate Assay

An assay under evaluation also referred to as an “index assay.”55,57

Characteristics of Validation

Validation of laboratory assays cannot be performed without defining validation characteristics. There are 4 essential characteristics of validation of laboratory assays including (i) sphere of validation, (ii) type of validation, (iii) scope of validation, and (iv) tier of validation. Tiers apply mostly to technical validation.58

Clinical Validation of Predictive Biomarker

The process of demonstrating how robustly and reliably the biomarker result predict the clinical outcome of interest. This is also termed as “qualification of predictive biomarker” (see below).59

Clinical Validity (Synonym—Clinical Utility)

The assay’s ability to add value to patient management decision-making compared with current practices.60

Companion Diagnostic IHC Assay For PD-L1

An IHC assay that provides information for the effective use of a corresponding anti-PD-1/PD-L1 therapeutic product, linked to the specific drug within its approved labeling.61

Comparator Assay

An assay that was designated as the true value also referred to as “reference standard” or “diagnostic accuracy criteria.”55,57

Complementary Diagnostic IHC Assay For PD-L1

An IHC assay that can aid in therapeutic decision-making for anti-PD-1/PD-L1 therapeutic products but is not required before prescribing a drug.62

Diagnostic Accuracy

The extent of agreement between a candidate assay (ie, index assay) and a comparator assay (ie, reference standard or other diagnostic accuracy criteria).55,57

Diagnostic Accuracy Criteria

The best currently available criteria for establishing the presence or absence of the condition, event, or characteristic.55

Diagnostic Biomarker

A biomarker that is used to identify disease.63

Diagnostic Sensitivity

The proportion of those with the target condition (as defined by a reference standard) who test positive with a candidate test.55,57,64

Diagnostic Specificity

The proportion of those without the target condition (as defined by a reference standard) who test negative with a candidate assay.55,57,64

Diagnostic Validation of a Biomarker

The process of demonstrating how robustly and reliably the assay results correlate with the diagnosis of interest.58

EQA Accuracy

For PD-L1, this is demonstrated by using 20 positive and 20 negative cases65 (based on a fit-for-purpose cutoff point) of tumors for which the PD-L1 biomarker will be used and the results compared with those obtained by another laboratory that is successfully using an already validated assay; positive cases are selected as such to span the entire reportable range of PD-L1 expression and positive cases usually originate from the institutional tissue archive. The aim of EQA accuracy assessment is to demonstrate that the LDT protocol performs as it should in a specific type of tumor (with specificity for predefined types of cells/tissues, appropriate subcellular signal localization, etc.).66 This is not to be confused with indirect clinical validation in which a larger number of randomly selected cases is used and evaluated against a reference standard IHC assay (see below).55,57

Indirect Clinical Validation of a Predictive Biomarker Assay

Validation of diagnostic accuracy of a biomarker assay against the results of a designated, previously clinically validated reference standard assay or “diagnostic accuracy criteria” (see above) or designated reference standard biomarker assay. The previously validated biomarker assay may or may not be regulatory agency-approved and it may or may not be employing the same methodology (eg, IHC, but also fluorescence in situ hybridization, etc.), but it must be already qualified/validated in a prospective clinical trial. Indirect clinical validation demonstrates that the assay in question (candidate assay) has identical or nearly identical diagnostic sensitivity and specificity as the reference standard assay or “diagnostic accuracy criteria” (see above) or designated reference standard assay (comparator assay), where the comparator assay has already established link(s) to clinical outcomes; it attempts to answer the question of whether (or to what degree) a candidate assay may identify the same patients (as being “positive” or “negative”) as the comparator assay.

“Interchangeable” PD-L1 IHC Predictive Biomarker Assay

This definition is adapted to be analogous to the use of the term “interchangeable” in drug development and approval. It refers to PD-L1 IHC assays that have demonstrated essentially identical performance in clinical trials for the same disease and the same drug. Demonstration of “correlation,” “similarity,” or “overall agreement” must not be interpreted as a demonstration of identical clinical outcomes, but rather technical performance. Indirect clinical validation is not sufficient for the clinical qualification of a biomarker assay and cannot be used as a basis for designating a biomarker assay as “interchangeable.”59 This applies to both, FDA-approved kits as well as LDTs.

LDT

An LDT, as it pertains to this document, is any IHC assay designed to detect and report the expression of PD-L1 protein in tumor cells and/or immune cells for predicting potential response to a regulatory agency–approved PD-L1/PD-1 checkpoint inhibitor with at least one of the following characteristics: (1) A testing laboratory develops and validates a PD-L1 IHC assay from first principles using separately purchased, commercially available components (aka. “de novo LDT”); (2) A testing laboratory adds/subtracts/modifies any manufacturer-specified preanalytical, analytical, or postanalytical component/aspect of a commercially available, regulatory agency–approved PD-L1 IHC assay/in vitro diagnostic device (aka. “kit-derived LDT”); (3) A testing laboratory performs PD-L1 testing using a commercially available, regulatory agency–approved PD-L1 IHC assay/in vitro diagnostic device in accordance with the manufacturer’s specifications but for a purpose other than that intended by the manufacturer (aka. “kit-derived LDT”). This definition of LDT is adapted from the Canadian Standards Association/Standards Council of Canada’s standard Z316.8-18: Requirements for the design, development, and validation of LDTs.67

PD-L1 IHC Assay

Any IHC assay where the purpose is to demonstrate expression of the PD-L1 protein.

PD-L1 IHC Biomarker

Any PD-L1 IHC assay where both IHC protocol and IHC readout are fit-for-purpose based on demonstrated evidence.

Predictive Biomarker

A biomarker used to identify individuals who are more likely than similar individuals without the biomarker to experience a favorable or unfavorable effect from exposure to a medical product or an environmental agent.68

Prognostic Biomarker

A biomarker that provides information on the likely patient health outcome (eg, disease recurrence or progression) regardless of the treatment.68

Qualification

A conclusion, based on a formal regulatory process, that within the stated context of use, a medical product development tool can be relied upon to have a specific interpretation and application in medical product development and regulatory review63; when applied to a predictive biomarker, it refers to “clinical validation.”68–70

Regulatory Agency–Approved IHC Kit/Assay

An IHC kit/assay that was approved by a regulatory agency for a specific purpose.

Replacement Assay

See “Candidate assay.”

“Repurposed” PD-L1 IHC Predictive Biomarker Assay

Analogous to repurposed drugs, repurposed predictive biomarker assays those that have been originally developed and qualified in a clinical trial for 1 purpose (specific drug and disease) but were also qualified in a different clinical trial for a different purpose.71,72 Such a repurposed PD-L1 IHC biomarker assay is expected to have identical IHC protocol conditions, but may have different readout and different reporting requirements, for example, the PD-L1 IHC 22C3 pharmDx assay for NSCLC was repurposed as a biomarker assay for PD-L1 detection in gastric cancer for immunotherapy with pembrolizumab; a different readout was designated to this assay for this new purpose.73

Revalidation (technical/analytical)

In clinical IHC, technical/analytical revalidation is divided into primary, secondary, and tertiary revalidation depending on the trigger that initiated revalidation.58

Scope of Technical/Analytical Validation

In clinical IHC, the scope of technical/analytical validation is divided into initial validation and revalidation.58

Spheres of Validation

IHC validation has the following spheres: clinical, indirect clinical/diagnostic, technical/analytical.58

Technical/Analytical Validation of IHC Assay

Assessment of performance characteristics of an assay, including analytical sensitivity, analytical specificity, analytical reproducibility, EQA and/or non-IHC methodology accuracy, reportable range, extended analytical specificity, and preanalytical robustness.58,59,66,74

Tiers of Technical/Analytical Validation

Tier 1 (synonym: “verification”; including analytical sensitivity, analytical specificity, and reproducibility), tier 2 (including EQA or non-IHC methodology accuracy, reportable range, and extended analytical specificity), tier 3 (validation for preanalytical robustness including analytical sensitivity, analytical specificity for different clinically and institutionally applicable preanalytical conditions, reagents, and times).58 Verification is typically performed by using control materials such as immunohistochemistry critical assay performance controls (iCAPCs).64

Types of Validation

IHC validation type refers to the subject of validation. This includes validation of reagents (eg, buffer, primary Ab validation), lot-to-lot validation of primary Ab, validation of IHC protocol, validation of pathologist's readout, validation of instruments, etc.58

Validation

The process of assay/test validation establishes the clinical and analytical performance characteristics of an assay/test as well as the assay/test limitations. This involves confirmation, through the provision of objective evidence, that the requirements for a specific intended use or application have been fulfilled.58,75

Verification

Verification is performed to ensure that the laboratory can meet or exceed the Health Canada-approved, FDA-approved, or CE-marked performance characteristics established by the assay/test manufacturer; this involves confirmation, through the provision of objective evidence, that specified requirements have been fulfilled.75 Verification is also known as tier 1 technical/analytical validation (see above).58

RESULTS

A total of 38 GSs were generated along 3 broad streams: (1) Recommendations for assay and sample selection, (2) Recommendations for assay/test reporting, and (3) Recommendations for quality assurance.

Recommendations for Assay and Sample Selection

Assay Selection

GS01. Strong recommendation—Selection of a PD-L1 IHC assay as a predictive biomarker for anti-PD-1/PD-L1 treatment must be made in a fit-for-purpose manner in accordance with the 3D approach (Drug-Disease-Diagnostic assay).18,23,29–34,76–86

Explanatory note: In biomarker-driven companion/complementary diagnostic testing, “purpose” is tripartite and consists of (i) the Drug, (ii) the Disease, and (iii) the Diagnostic assay. What links the 3D’s of purpose together is the clinical trial where the efficacy of the Drug was established in participants with a specific Disease and where biomarker testing results with a specific Diagnostic assay on biospecimens from trial participants could successfully separate participants who showed clinical response to the Drug from participants who did not show clinical response to the Drug. A Diagnostic assay that has been validated with biospecimens from responders and nonresponders in a clinical trial designed to assess the efficacy of a drug in a specific disease, is considered to be “clinically validated” or “qualified.”

GS02. Strong recommendation—Health Canada-approved, FDA-approved, and CE-marked PD-L1 IHC Kits that are validated in clinical trials for specific purposes must be used for those specific purposes.76–78,86–90

Explanatory note: “PD-L1 IHC Kit” is a commercially manufactured and marketed Diagnostic assay that uses IHC to detect certain expression patterns of PD-L1 protein and that was clinically validated (see GS01) with biospecimens from responders and nonresponders of a clinical trial for a specific drug in a specific disease population.

GS03. Strong recommendation—Health Canada-approved, FDA-approved, and CE-marked PD-L1 IHC Kits that are validated in clinical trials for specific purposes, when not used for these specific purposes, but used for a different purpose (ie, in accordance with the 3D approach for a different disease and/or different drug), must be considered as LDTs and are not to be considered as Health Canada-approved, FDA-approved, or CE-marked for this new purpose.86–88,90–93

Explanatory note: If Diagnostic assay #1, which was clinically validated with results from its clinical trial (ie, clinical trial #1) that assessed the efficacy of Drug #1 in Disease #1, is used to predict potential response to Drug #2 in Disease #2, then Diagnostic assay #1 is considered a LDT in the context of Drug #2 in Disease #2. In the context of Drug #1 in Disease #1 though, Diagnostic assay #1 is a fit-for-purpose companion/complementary diagnostic assay (eg, PD-L1 IHC Kit #1). Such “kit-derived LDTs” that were also indirectly clinically validated for some purposes (that were not included in the kit label) in published literature, may be acceptable for those limited additional applications. However, just like any other LDT, “kit-derived LDTs” need to be technically validated by the laboratory that will be performing the PD-L1 testing before being put in clinical use.

GS04. Strong recommendation—PD-L1 IHC LDTs that are properly validated for a specific purpose according to Section 3.1 (Recommendations for Quality Assurance - Validation) may be used for that purpose, even if not Health Canada-approved, FDA-approved, or CE-marked for this new purpose. This applies, whether they are developed with the same or a different primary Ab (clone or polyclonal Ab) compared with IVDs marketed for the same purpose.58,59,65,94

Explanatory note: If a biomarker test is required or desired to predict for potential response to a specific approved Drug in a specific Disease AND the laboratory does not wish to or is not able to use the specific PD-L1 IHC Kit that was clinically validated by the associated clinical trial, then the laboratory may use an LDT so long as the LDT is validated in a fit-for-purpose manner in accordance with Section 3.1 (Recommendations for Quality Assurance - Validation). An LDT may be developed using any anti-PD-L1 primary Ab clone as long as the LDT is properly validated. Well-designed, fit-for-purpose (properly validated) de novo LDTs are favored over kit-derived LDTs (see GS03). See GS26 for the definition of a “properly validated LDT.”

GS05. Recommendation—Selection of a “replacement assay” for any given purpose (as defined by the 3D approach) should be based on demonstrated diagnostic test accuracy (sensitivity and specificity) against a designated reference standard (or other diagnostic accuracy criteria) for that specific purpose; neither clone nor readout need to be identical, but high sensitivity and specificity (see Section 3.1; Recommendations for Quality Assurance - Validation) should be achieved for direct or indirect clinical validation by the selected replacement (candidate) assay.55

Explanatory note: This statement reiterates that a properly validated PD-L1 IHC LDT is one that shows evidence of success for both technical validation and clinical validation (direct or indirect). It further elaborates that the evidence generated by indirect clinical validation is proof of diagnostic accuracy, which consists of 2 elements: diagnostic sensitivity and diagnostic specificity. Diagnostic sensitivity and specificity should both be ≥90%. For indirect clinical validation, this means that at least 90% of cases that were read as being positive based on results generated by the reference standard PD-L1 IHC Kit must also be read as being positive based on results generated by the corresponding PD-L1 IHC LDT. Similarly, at least 90% of cases that were read as being negative based on results generated by the reference standard PD-L1 IHC Kit must also be read as being negative based on results generated by the corresponding PD-L1 IHC LDT. The tissues for indirect clinical validation may originate from the institution that is performing the indirect clinical validation for a newly developed LDT, but it also may originate from a proficiency testing (PT) program’s reference laboratory as well as from multiinstitutional sources.

GS06. Strong recommendation—Published results of properly conducted studies of direct or indirect clinical validation and/or meta-analyses of such studies, that address diagnostic accuracy of candidate assays, must guide selection of potential “replacement assays” for any specific purpose (eg, both Dako PD-L1 IHC 28-8 pharmDx and VENTANA PD-L1 (SP263) showed acceptable diagnostic accuracy for the 50% TPS cut-off for PD-L1 IHC 22C3, but not for the 1% cut-off).90

Explanatory note: A systematic review and meta-analysis for PD-L1 replacement assays indicated that the evidence does not support interchangeability of the assays when based on diagnostic accuracy.90

GS07. Recommendation—For Health Canada-approved, FDA-approved, and CE-marked PD-L1 IHC Kits, the clinical laboratory should validate the assay as an LDT when any pre-analytical parameters fall outside of what is recommended in the Kit’s instructions for use.58,59,87,88,92,94,95

Explanatory note: Validation refers to both clinical validation (direct or indirect) and technical validation. As explained above, any parameter outside of the declared label of a Kit renders the assay to an LDT and therefore, at a minimum it requires indirect clinical validation and technical validation. See also GS30.

GS08. Expert opinion—For LDTs that employ the same primary antibody clone as a Health Canada-approved, FDA-approved, and CE-marked PD-L1 IHC Kit, the same pre-analytical caveats found in the corresponding Health Canada-approved, FDA-approved, and CE-marked PD-L1 IHC Kit apply.59,96

Explanatory note: Every monoclonal primary Ab is designed for a unique epitope that has its own biochemical characteristics including preanalytical robustness. When preanalytical robustness of 1 primary Ab (eg, 22C3 clone) is tested, these results do not necessarily apply to other clones developed for different epitopes of the same molecule (eg, 28-8, SP142, and SP263, all of which are developed to detect the PD-L1 molecule but bind to different epitopes).

Sample Selection

GS09. Recommendation—If more than one tissue block is available for a given tumor, the most representative sample should be tested.97–122 More than one block may be tested when the reporting pathologist determines that such additional testing is necessary to establish the PD-L1 status of the tumor. If additional blocks from the same sample are tested, the results from all tested blocks should be combined as if it were present in a single paraffin block.

Explanatory note: Depending on the pathologist’s assessment, testing of additional paraffin blocks may be required to establish the PD-L1 status of the tumor.

GS10. Expert opinion—For samples where the initial level contains less than the minimum required number of tumor (or other target) cells as defined by the interpretive manual for a specific assay and purpose, testing of additional levels may be performed with extra tissue containing additional tumor cells to reach the required threshold. If additional levels are tested, all tissue from all levels may be combined as if it were present in a single level.

No Explanatory note.

GS11. Recommendation—If more than one sample is available at the same time (eg, biopsies from two different sites, or biopsy and cytology samples, or primary and corresponding metastatic tumors), the most representative sample should be tested. More than one sample may be tested when the reporting pathologist determines that such additional testing is necessary to establish the PD-L1 status of the tumor.97–122

No Explanatory note.

GS12. Recommendation—It is acceptable to test a tumor sample at the time of consideration for anti-PD-1/PD-L1 immunotherapy even if an older tumor sample was previously tested.122

No Explanatory note.

GS13. Recommendation—The use of Health Canada-approved, FDA-approved, or CE-marked PD-L1 IHC Kits that were validated for formalin-fixed paraffin-embedded biopsy samples, and not specifically validated for cytology samples, may be used for cytology samples if (i) they were processed according to the same pre-analytical conditions as required by the PD-L1 IHC Kits, and (ii) the readout is compatible with the type of cytology samples considering the lack of tissue architecture.108,123–130

Explanatory note: PD-L1 IHC Kits are typically validated for formalin-fixed paraffin-embedded surgical pathology samples and not for “cytology” samples. This would exclude smears since regardless of fixative, smears are typically not paraffin-embedded. However, cytology specimens that are immediately fixed in 10% neutral buffered formalin and then spun down into a cell pellet that is subsequently processed/paraffin-embedded (similar to typical surgical pathology specimens) and where tumor cells can be clearly distinguished from inflammatory cells and other cells (eg, mesothelial cells), are essentially small biopsies and may be tested. However, if the laboratory intends to perform PD-L1 testing on i) smears (regardless of fixative) or ii) cytologic specimens not fixed in 10% neutral buffered formalin, then the PD-L1 IHC Kit becomes an LDT in the context of the nonqualifying “cytology” specimen-type and indirect clinical validation is required.

GS14. Expert opinion—For the use of PD-L1 LDTs on cytology samples,58,65,76,90 GS30 applies.

Explanatory note: PD-L1 IHC LDTs are considered properly validated for cytology samples when evidence of clinical validation (direct or indirect) and technical validation is provided and is relevant to the cytology samples on which the laboratory plans to perform testing. Also, see GS26.

GS15. Expert opinion—For samples where identification of appropriate cell types is problematic (eg, tumor cell vs. macrophage in NSCLC), multiplexing (eg, double IHC) may be used, not instead of, but in addition to the usual PD-L1 assay that is performed by the laboratory, so long as the multiplexed assay has undergone appropriate technical validation.58,65

No Explanatory note.

Recommendations for Assay/Test Reporting

Reporting of Predictive PD-L1 Assays

GS16. Expert opinion—Pathologists issuing reports for PD-L1 testing should successfully undergo training that is (i) endorsed by the manufacturer of an approved PD-L1 IHC Kit, or (ii) endorsed by a professional organization (eg, Canadian Association of Pathologists, International Quality Network for Pathology, etc.), or (iii) provided by a pathologist who has undergone training as indicated in (i) or (ii) and in a train-the-trainer format. The optimal goal for such training should be to achieve 90% sensitivity and 90% specificity of the previously validated readout.131–136

Explanatory note: Readout accuracy (sensitivity and specificity) needs to be demonstrated for relevant cutoff(s), rather than for overall agreement with a designated reference standard or concordance. Documentation and retention of readout validation evidence apply as per relevant documentation and retention of evidence of any IHC assay validation.

GS17. Expert opinion—Reporting of computer-assisted PD-L1 readout (image analysis) for the purpose of patient selection for immunotherapy is not recommended at this time.137–142

Explanatory note: Some PD-L1 readout methods may be more amenable to image analysis than others; therefore, any recommendations at this time may not be universally applicable for all different types of readouts. At this time, there is insufficient evidence to recommend the use of image analysis for the readout of PD-L1 assays with confidence. Pathologist-assisted image analysis may prove to be a valuable tool as it further develops.

GS18. Recommendation—A structured (synoptic) format should be used for PD-L1 reporting. Searchable formats are recommended to facilitate audit of institutional or pathologist-specific prevalence.131,143–147

No Explanatory note.

GS19. Recommendation—In the comments section of the report, the following information should be included: (i) the drug and disease for which the assay was run, and (ii) the assay that was performed, and on what platform, (iii) the clinically relevant cutoff point for the 3D combination, (iv) the readout parameter being reported (eg, Tumour Proportion Score, Combined Positive Score, Immune Cells), and (v) the clinically relevant cutoff category within which the result falls.148–157

No Explanatory note.

GS20. Expert opinion—If the specimen was inadequate for testing, this may be stated in the report along with the reason why the specimen was inadequate.

No Explanatory note.

GS21. Expert opinion—Results with cytology samples may be reported in the same manner recommended for solid tissue samples (see GS19).

No Explanatory note.

GS22. Expert opinion—Results with fresh-cut sections from archived paraffin blocks or previously-cut and stored unstained sections that are older than the PD-L1 IHC Kit manufacturer’s recommendation may be reported only if positive and it is the pathologist’s assessment that, based on presented evidence, the result is not false-positive (eg, pattern of staining being membranous, appropriate internal and external controls are present and show expected range of positive and negative result). “Positive result” is here defined as a positive result for a specific purpose and a specific patient context (eg, the staining of inflammatory cells has no relevance for some PD-L1 assays while for others, they are the principal cell type assessed to determine whether the sample is to be designated as being a “positive” result). Therefore, some positivity identified in a lesion does not qualify the results as reportable. Negative results may be reported as “no result” and potential for false-negative results with older samples may be included in the comment section.158,159

No Explanatory note.

Recommendations for Quality Assurance

Validation

GS23. Strong recommendation—Validation of PD-L1 IHC predictive assays must be fit-for-purpose in accordance with the 3D (Drug-Disease-Diagnostic assay) approach.58,65,76,77,94

Explanatory note: This statement mirrors GS01 but from the perspective of the laboratory. The Drug-Disease combination informs the laboratory of the PD-L1 IHC Kit that can either (i) be verified and run as the companion/complementary Diagnostic assay for the chosen Drug-Disease combination, or (ii) be used as the reference standard assay if the laboratory decides to develop, validate, run, and maintain an LDT for the chosen Drug-Disease combination. The Disease component of the 3D approach also informs the laboratory of the type of validation cases necessary to collect in order to create the tissue tools that will allow for ongoing monitoring of analytical sensitivity and specificity on a daily basis and of diagnostic sensitivity and specificity on a periodic basis. Also, see Explanatory note GS01.

GS24. Strong recommendation—When a laboratory decides that an LDT will be used instead of a regulatory agency–approved IHC kit for the same purpose, the IHC laboratory must provide evidence of successful methodology transfer. This must include (i) technical validation and (ii) indirect clinical validation by using the relevant reference standard where not already published or established (ie, the regulatory agency–approved kit for the same purpose).58,59,65,77–79,86,91

No Explanatory note.

GS25. Recommendation—Indirect clinical validation can be performed when there is a recognized reference standard or gold standard that has been validated in a published clinical trial for that specific purpose. Samples for indirect clinical validation should include at least 50 randomly selected, clinically relevant positive cases, and at least 50 negative cases identified in the screening process.55

Explanatory note: The tissues for indirect clinical validation may originate from the institution that is performing the indirect clinical validation for a newly developed LDT, but it also may originate from a PT program’s reference laboratory as well as from multiinstitutional sources. See Terminology section for more information about indirect clinical validation and explanatory notes for GS05 and GS26.

GS26. Strong recommendation—PD-L1 IHC LDTs that have been clinically validated or indirectly clinically validated in published literature must still be technically validated by the clinical IHC laboratory before being put in use as a PD-L1 predictive biomarker assay.58,59,65,86

Explanatory note: A properly validated LDT in the context of PD-L1 being a patient selection biomarker in immuno-oncology is one that has successfully undergone both clinical validation (direct or indirect) and technical validation. Clinical validation may be direct (where the reference standard result is based on the clinical responses of clinical trial participants) or indirect (where the reference standard result is based on the result generated by the companion diagnostic assay that was developed from clinical responses of clinical trial participants). Evidence for clinical validation (direct or indirect) of an LDT may be generated by the laboratory or alternatively, if available, may be derived from the literature. Clinical validation (direct or indirect) evaluates the total test. Evidence for technical validation must be generated by the laboratory—such evidence cannot be derived from literature. Technical validation is performed on the analytical phase of the total test, namely the protocol and the readout. Therefore, direct clinical validation is based on clinical outcomes (the study of patients), indirect clinical validation on the reference standard assay results (the study of cases), and technical validation on the protocol and/or readout results with validation samples (ie, QA/QC tissue tools).

GS27. Strong recommendation—Health Canada-approved, FDA-approved, and CE-marked PD-L1 IHC Kits are already directly or indirectly clinically validated in clinical trials; however, their analytical performance must be verified according to manufacturer’s instructions before use. If no specific instructions are provided by the manufacturer, general recommendations for Tier 1 technical validation apply (analytical sensitivity, analytical specificity, and reproducibility) by using iCAPCs or iCAPCs-like tissues/calibrators with at least 3 successful, independent verification runs.58,160

No Explanatory note.

GS28. Strong recommendation—As per recommendations above, if multiplex assays (eg, double IHC staining for PD-L1 and TTF-1) are to be used, they must have undergone technical validation. The possible impact of multiplexing on diagnostic accuracy of each biomarker must be assessed against single IHC staining using a Health Canada-approved, FDA-approved, or CE-marked IHC kit or other fit-for-purpose designated reference standard.58,65,94

No Explanatory note.

GS29. Strong recommendation—The PD-L1 IHC readout must be validated for diagnostic accuracy (readout accuracy) by each pathologist reporting PD-L1 IHC results as a categorical variable; in addition, validation of readout precision must also be conducted when applicable (eg, if reporting exact % positive cells as a continuous variable). This applies for both regulatory agency–approved IHC kits and LDTs.58,66,76,160

Explanatory note: There are online and in-person resources available for readout training. In addition, pathologists may avail themselves of EQA PT tools developed specifically for pathologist readout proficiency where and when available.

GS30. Strong recommendation—Laboratories using regulatory agency–approved PD-L1 IHC Kits in a manner that deviates from the manufacturer's instructions must validate the deviation.58,65,86,161

Explanatory note: Deviations may occur in the preanalytical phase, the analytical phase, or the postanalytical phase of the total test. See also GS07.

Quality Control

GS31. Strong recommendation—Laboratories must run external on-slide controls for every slide.58,64,66,162,163

No Explanatory note.

GS32. Strong recommendation—The minimum composition of external controls for FDA (or other regulatory body)-approved PD-L1 IHC Kits must be in compliance with the Kit manufacturer's instructions.148,155,156,161,164

No Explanatory note.

GS33. Expert opinion—The minimum composition of external on-slide controls for LDTs ideally will mirror the requirements of the corresponding fit-for-purpose FDA (or other regulatory body)-approved kit.

No Explanatory note.

GS34. Expert opinion—Where not already included as part of the required minimum composition, external on-slide controls for PD-L1 testing may include the following elements: (i) lesional tissue showing positivity for PD-L1 in a specified cell population, (ii) lesional tissue showing negativity for PD-L1 in a specified cell population, (iii) low expressor tissue in accordance with iCAPCs-like principles (eg, macrophages in germinal centres of tonsil).58,64,163

No Explanatory note.

GS35. Recommendation—The laboratory should assess the external on-slide control by light microscopy prior to a slide being released from the laboratory. When external on-slide controls demonstrate either unacceptable background or lack of signal in any positive control tissue, such IHC slides should be failed by the laboratory.64,162

No Explanatory note.

EQA—PT

GS36. Recommendation—All laboratories performing PD-L1 testing for patient selection should participate in fit-for-purpose PT at least every six months where available and required by relevant regulatory bodies.165–169

Explanatory note: A fit-for-purpose PT challenge follows the 3D approach. The Drug-Disease combination determines the Diagnostic assay that will be used to generate the reference standard results against which the results of the participating laboratories are measured. The Disease determines the selection of the tumor type and the readout that will be used in the PT run to calculate diagnostic accuracy.

GS37. Recommendation—Laboratories should select a PT program that ideally informs participating laboratories on the accuracy of the PD-L1 IHC protocol and the pathologists’ readout as determined by central expert assessment.165,170,171 If no such program is available/accessible, inter-laboratory exchange of samples and information with a provincial or national reference laboratory should be performed.172,173

No Explanatory note.

GS38. Strong recommendation—Participation in PT, even when PT is fit-for-purpose, is not acceptable as a substitute for clinical and technical validation.58,59,65,86

Explanatory note: When fit-for-purpose and when designed appropriately, PT may be a substitute for indirect clinical validation (but not for clinical validation or technical validation).

CONCLUSIONS

In situ biomarker testing is not defined solely by the detection of biological gene products in human tissue sections; rather, such testing must always be accompanied by the appropriate medical context to be meaningful for patient care. For diagnostic biomarker testing, the medical context may be the impact of tissue specificity (eg, tumor type) on the meaningfulness of test results, whereas, for predictive biomarker testing, the medical context typically also includes a specific therapeutic agent in addition to tumor type or tissue specificity. Therefore, the current reality of PD-L1 testing in immuno-oncology is such that detection of PD-L1 protein expression is only meaningful in the context of the tumor(s)/tissue type(s) being tested, for potential response to a particular therapeutic agent based on data generated by clinical trials. Given the evolving landscape for PD-L1 testing, the intention of the CAP-ACP National Standards Committee for High Complexity Testing is to periodically update these GSs as long as PD-L1 remains a relevant biomarker for patient selection in immuno-oncology.

Supplemental Digital Content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website, www.appliedimmunohist.com.