PD-L1, a transmembrane protein, binds to PD-1 to transmit inhibitory signals, helping tumor cells evade T-cell attack. Blocking the PD-1/PD-L1 pathway is effective in cancer treatment. PD-L1 expression levels predict immunotherapy efficacy. Guidelines recommend using PD-L1 as a biomarker for immunotherapy, with immunohistochemistry as the primary evaluation method.
Currently, several detection reagents produced by Denmark’s DAKO, the United States’ Ventana, Xiamen Aid, and Suzhou MyGene have been approved in China for companion diagnostics or supplementary diagnostics of immune checkpoint inhibitors such as nivolumab. These products, compared to general immunohistochemistry detection reagents, have three prominent characteristics: rapid changes in intended use, complex scoring rules, and positive judgment values derived from clinical trials of drugs.
NMPA released an article in the “review forum” on its website. Below is the summary:
Product Technical Requirements
Product technical requirements should mainly include the following performance indicators: appearance, conformity, intra-batch repeatability, inter-batch repeatability, and stability. Considering the differences in PD-L1 expression in different tumor tissues, the “conformity” and “repeatability” indicators should clearly specify the use of applicable tumor tissue types containing different expression levels for detection.
Analytical Performance Studies
(1) Immunoreactivity
Normal tissues: Specificity evaluation should be conducted on 30 types of normal human tissues, with no fewer than 3 cases of each tissue type, and the staining location and characteristics should be described.
Abnormal tissues: Specificity evaluation should be conducted on related benign and malignant lesions. Relevant literature that adopts the declared product as test material can be submitted to verify the immunoreactivity of antibody reagents or detection kits with abnormal tissues. Abnormal tissues include adrenal cortex carcinoma, astrocytoma, basal cell carcinoma, squamous cell carcinoma, large cell carcinoma, thymoma, papillary adenocarcinoma, small cell undifferentiated carcinoma, squamous cell carcinoma, endometrial clear cell carcinoma, transitional cell carcinoma, lymphoma, malignant ependymoma, and stromal sarcoma.
(2) Specificity
- Cross-reactivity studies
Cross-reactivity studies with PD-1 and PD-L2 should be conducted, utilizing immunohistochemistry methods to detect cells/tissues expressing different proteins and verifying antibody specificity. Other reasonable methods can also be used for cross-reactivity studies.
- Specific recognition studies
Immunoblotting or enzyme-linked immunosorbent assays should be used to verify the specific detection performance of PD-L1. Specificity can be verified by adding PD-L1 antigen for detection or testing PD-L1 gene knockout tumor cells.
(3) Precision
Considering the significant differences in scoring rules for different PD-L1 detection reagents and the apparent differences in the interpretation of PD-L1 expression in clinical settings between labs and personnel, precision evaluation of such reagents should be more comprehensive, considering various influencing factors in the study design. Precision studies should be conducted using a randomized blind method with samples being expected samples with removed identification information (unstained samples), including negative, positive, and a certain number (recommended not less than 20% of the total sample size) of samples near the positive judgment value. Study data should include staining intensity, scoring results, and staining location, with reasonable statistical methods (e.g., Wilson Score method) used to calculate negative percent agreement (NPA), positive percent agreement (PPA), overall agreement (OPA), and bilateral 95% confidence intervals, and statistics on consistency rates for different expression levels. Appropriate acceptance criteria should be established, with the lower limit of the 95% confidence interval for consistency rates not less than 85%.
- Internal reproducibility
(1) Within-run precision: Conduct repeatability studies on continuous sections of expected tissue type samples.
(2) Between-run precision: Verify the main variables that may affect detection precision, including batches (not less than 3 batches), applicable models, operating methods (manual/automatic), inter-day and inter-reader variability. A reasonable clearing period should be set between two readings by the same reader.
- External reproducibility
Studies should be conducted in at least 3 external testing laboratories, including evaluations between non-consecutive days, different readers, and different external laboratories.
(4) Sensitivity
Study the detection ability for expected tissue type samples of different stages, primary and metastatic tumors. Study samples should include different expression levels (0 to 100%) and staining intensity, and calculate the positive rate of all samples and the proportion of positive samples with different staining intensities.
(5) Participation in domestic and international quality control activities
Enterprises are encouraged to actively participate in quality control activities of domestic (National Pathology Quality Control Evaluation Center) and international quality control organizations (NordiQC, UK NEQUAS, CAP, etc.) and submit quality control result reports.
Positive Value Study
The positive judgment value of such products is determined in the clinical trials of the drug, derived from the efficacy data of the drug’s clinical trials. Typical positive judgment value studies include establishing and validating two parts, with samples used in the two parts being independent of each other.
Samples used to establish the positive judgment value should meet the design of drug clinical trials and consider factors such as different geographical regions, ethnicities (races), genders, ages, and different pathological stages. Objective response rate (ORR)-based analysis methods, such as receiver operating characteristic curve (ROC) analysis, can be used to select and determine a reasonable positive judgment value.
Applicants can also refer to previous experiences or similar marketed products and use ROC analysis methods to propose the positive judgment value of the declared product, recruit subjects directly into major efficacy clinical trials, and validate the reference interval.