# PD-L1 immunohistochemistry comparability study in real-life clinical samples: results of Blueprint phase 2 project

Source: https://onco.cc/key-papers/paper-blueprint-phase-2-pd-l1-assays-jto-2018/  
OnCo record `paper-blueprint-phase-2-pd-l1-assays-jto-2018` (Key paper). Data CC BY-NC 4.0, attribute "Data from OnCo (onco.cc)"; commercial use needs a licence.

## TL;DR

Five different stains for the same protein were run on 81 real lung cancer samples and read by pathologists from thirteen countries. Three of the five agree closely on tumour cells, one stains less and one stains more, and none of them can be read reliably on immune cells.

## Summary

Eighty-one lung cancer specimens of various histological and sample types were stained with all five trial-validated PD-L1 assays (22C3, 28-8, SP142, SP263 and 73-10) and evaluated by an international panel of pathologists, with parallel assessment of digital images and of cytology cell blocks. The 22C3, 28-8 and SP263 assays gave highly comparable staining of tumour cells; SP142 was less sensitive and 73-10 more sensitive. Glass slide and digital scoring were highly concordant, with correlation above 0.96. Reliability between pathologists was very strong for tumour-cell scoring, with overall intraclass correlation 0.86 to 0.93, poor for immune-cell scoring at 0.18 to 0.19, and good for scoring on cytological cell block material at 0.78 to 0.85.

## Fields

- Kind: Key paper
- Last checked: 2026-09-25
- Journal: Journal of Thoracic Oncology
- Year: 2018
- DOI: 10.1016/j.jtho.2018.05.013
- Authors: Tsao MS, Kerr KM, Kockx M, et al.
- Findings: 22C3, 28-8 and SP263 are interchangeable for tumour-cell scoring; SP142 is less sensitive and 73-10 more sensitive.; Agreement between pathologists on tumour cells is excellent (0.86 to 0.93) and on immune cells is poor (0.18 to 0.19).; Cytology cell blocks are an acceptable substrate (0.78 to 0.85).; Digital images can be scored in place of glass slides.
- What it means: It is the reason a laboratory can run one stain and report against several drug labels, and the reason any decision resting on an immune-cell score rests on the least reproducible number in lung pathology.
- Caveats: Eighty-one specimens, chosen to span sample types rather than to represent a consecutive series.; Comparability is analytical: it shows the stains agree, not that the thresholds predict benefit equally.; The SP142 immune-cell score used for atezolizumab was not made reliable by this work.

## Sources

- Tsao et al., J Thorac Oncol 2018: Blueprint phase 2, five PD-L1 assays on 81 real-world lung cancer specimens: https://doi.org/10.1016/j.jtho.2018.05.013
- PubMed: https://pubmed.ncbi.nlm.nih.gov/29800747/

## Connected records

- biomarkers: [PD-L1 IC score (immune-cell score, SP142)](https://onco.cc/biomarkers/pd-l1-ic-score/), [PD-L1 TC score (tumour-cell score, SP263 and 28-8)](https://onco.cc/biomarkers/pd-l1-tc-score/), [PD-L1 TPS (tumour proportion score)](https://onco.cc/biomarkers/pd-l1-tps/)
- cancers: [Non-small-cell lung cancer](https://onco.cc/cancers/nsclc/)
- technologies: [Histopathology & immunohistochemistry](https://onco.cc/technologies/histopathology-ihc/), [Immune checkpoint inhibitors](https://onco.cc/technologies/checkpoint-inhibitor/)
- targets: [PD-L1](https://onco.cc/targets/pdl1/)
- pathways: [PD-1 / PD-L1 immune checkpoint & T-cell activation](https://onco.cc/pathways/pd1-checkpoint/)
- terms: [Biopsy](https://onco.cc/terms/biopsy/), [Immunohistochemistry (IHC)](https://onco.cc/terms/ihc/)
- journals: [Journal of Thoracic Oncology](https://onco.cc/journals/journal-of-thoracic-oncology/)

---
JSON: https://onco.cc/api/v1/entities/paper-blueprint-phase-2-pd-l1-assays-jto-2018.json