A PET tracer that sees PD-L1, where FDG sees nothing useful

Design

prospective paired-tracer imaging study with immunohistochemistry reference, DeLong test and multivariable logistic regression

Population

34 patients with histologically confirmed head and neck squamous cell carcinoma (median age 59, 26 men); 37 lesions; 17 followed through chemo-immunotherapy

Primary outcome

discrimination of PD-L1-positive lesions and prediction of chemo-immunotherapy response

Effect

SUVmax 2.7 vs 1.7 in PD-L1-positive vs negative lesions (P < 0.001); AUC 0.92 vs 0.69 for FDG (P = 0.003); response prediction odds ratio 12.42 (95% CI 2.08–74.15)

FDG shows metabolic activity, which is why it says nothing about whether a tumour expresses programmed death-ligand 1 (PD-L1). A prospective study imaged 34 patients with histologically confirmed head and neck squamous cell carcinoma with both 68Ga-DOTA-WL12, a PD-L1-targeted tracer, and FDG, with immunohistochemistry as the reference.

Across 37 lesions, DOTA-WL12 maximum standardised uptake value was higher in PD-L1-positive than negative lesions (2.7 vs 1.7, P < 0.001), while FDG uptake did not differ at all. For identifying PD-L1 positivity the area under the curve was 0.92 for DOTA-WL12 against 0.69 for FDG (P = 0.003). In the 17 patients who went on to chemo-immunotherapy with complete follow-up, DOTA-WL12 uptake was higher in responders than non-responders (2.7 vs 1.3, P = 0.002) and was the only independent predictor of response (odds ratio 12.42, 95% confidence interval 2.08 to 74.15).

This is an early prospective study and the response analysis rests on 17 patients, which is why the odds ratio has a confidence interval spanning a factor of thirty-five. What it establishes is that the tracer binds what it claims to bind in patients, and that the signal is not something FDG was ever going to provide. Whether it should change who receives immunotherapy is a question for a much larger study — and, in India, the tracer is a research agent rather than something a department can order.

• Do not read FDG uptake as any indication of PD-L1 status; it did not discriminate at all here.
• Continue to base immunotherapy decisions on tissue immunohistochemistry.
• Note the value of a whole-body tracer where lesions differ: PD-L1 expression is heterogeneous between sites.
• Treat the response prediction as hypothesis-generating at 17 patients.
• Availability is the limiting factor — this is a research tracer, not a clinical service, in most of the world.

Why it matters

It shows a PET signal that carries biological information FDG structurally cannot give, in a cancer where immunotherapy selection is currently biopsy-bound.

Don't overread it

Seventeen patients in the response analysis — this identifies a candidate imaging biomarker, it does not qualify one.

The statistics, in plain English

An area under the curve of 0.92 against 0.69, tested formally with DeLong, is a real difference in discrimination and not a marginal one. The response result is far weaker: an odds ratio of 12.42 with a confidence interval from 2.08 to 74.15 comes from 17 patients, and an interval that wide means the direction is probably right and the magnitude is unknown. With 37 lesions from 34 patients, lesions within the same patient are not independent, which the lesion-level analysis does not fully account for.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.