What a PD-L1 Score Means for Immunotherapy

PD-L1 is a protein that acts as a brake. When it sits on the surface of a tumor cell and meets PD-1 on an immune T cell, the T cell stands down. Some cancers exploit this deliberately, producing PD-L1 to switch off the immune response aimed at them.

Checkpoint inhibitor drugs block that handshake and release the brake. Pembrolizumab, nivolumab, atezolizumab and durvalumab are examples. Testing for PD-L1 tries to work out, in advance, whose cancer relies on this particular brake.

Your pathology report will use one of three systems. They are not interchangeable.

TPS, the tumor proportion score, counts only cancer cells: the percentage showing PD-L1 staining on their surface. It is used mainly in non-small cell lung cancer.

CPS, the combined positive score, counts PD-L1-positive cells of three types: tumor cells, lymphocytes and macrophages. That count is divided by the total number of tumor cells, then multiplied by 100. CPS is used in gastric, esophageal, cervical, head and neck and triple-negative breast cancers. Immune cells are counted on top of that fraction but not on the bottom. Because of that, the result can go above 100. When it does, it is capped at 100. A CPS is a number, not a percentage.

IC, the immune cell score, measures something different. It looks at the tumor area and asks what share of it is taken up by PD-L1-positive immune cells. It is used with atezolizumab, and the cutoff differs by cancer: IC of 5% or above in urothelial (bladder) cancer, IC of 1% or above in triple-negative breast cancer.

There is no single PD-L1 threshold. Each was established in the trial that led to each approval.

In non-small cell lung cancer, a TPS of 50% or above supports pembrolizumab used alone. The FDA later expanded that approval. Now a TPS of 1% or above also qualifies, using the 22C3 test. In triple-negative breast cancer, a CPS of 10 or above supports pembrolizumab plus chemotherapy. In head and neck cancer, a CPS of 1 or above supports pembrolizumab used on its own as first-line treatment; pembrolizumab combined with platinum and fluorouracil is approved there without any PD-L1 requirement. In gastric and esophageal cancers, the CPS threshold is 1, 5 or 10, depending on the regimen used. In cervical cancer, a CPS of 1 or above is used.

The antibody used matters too. Four clones dominate: 22C3, 28-8, SP263 and SP142. Each was developed alongside a particular drug. They do not give identical results on identical tissue. SP142 in particular tends to score lower than the others. Because of this, a score from one assay cannot be read across to a drug that was validated with a different one.

PD-L1 is one of the better biomarkers available. It is still far from decisive.

PD-L1 expression can vary within a single tumor. A biopsy taken from one edge may not represent the whole. Expression also varies between the primary tumor and its metastases. It can change over time, including after treatment. Because of this, a tissue sample from a diagnosis two years ago may no longer describe your cancer today. Biology is also only part of the story. The immune system's ability to respond depends on many things that PD-L1 staining does not capture.

The practical consequence follows from this. People with high scores sometimes do not respond. People with negative scores sometimes respond very well. If your score came back low, that is genuinely not the end of the conversation.

PD-L1 is usually interpreted alongside other results. One example is mismatch repair deficiency, also called high microsatellite instability, or dMMR/MSI-H. In colorectal and endometrial cancers, this predicts checkpoint inhibitor benefit more strongly than PD-L1 does. Tumor mutational burden is used in some settings too. In lung cancer, a driver mutation such as EGFR or ALK changes the plan. When one is found, targeted therapy is usually tried first, whatever the PD-L1 score says.

Many immunotherapy combinations are approved without any PD-L1 requirement at all. For those treatments, the score may shape expectations. It does not decide who qualifies.

A PD-L1 score is a statement about likelihood. It comes from a specific test, for a specific drug, in a specific cancer. It is worth knowing which assay produced your result. It is worth knowing which cutoff applies to your case. Those two details determine what the number actually means for your options.

• FDA — PD-L1 IHC 22C3 pharmDx (TPS and CPS scoring)
• FDA — Ventana PD-L1 SP142 Assay (IC scoring)
• Maule et al. 2022 (PMC) — PD-L1 antibody clone comparison
• FDA — Pembrolizumab for HER2-positive gastric/GEJ adenocarcinoma expressing PD-L1 (CPS ≥1)
• FDA — KEYTRUDA (pembrolizumab) prescribing information (CPS cutoffs by indication)
• NCI — Immune Checkpoint Inhibitors
• NCI Dictionary — PD-L1