The following is a summary of “Identification of a predictive phosphoproteomic signature of response to atezolizumab and bevacizumab (AB) in patients with advanced hepatocellular carcinoma (aHCC),” published in the 2024 ASCO Annual Meeting under issue of Oncology by Sarker et al.
The number of treatments for advanced hepatocellular carcinoma (aHCC) has grown recently. The atezolizumab and bevacizumab (AB) combination became standard in 2020, but most patients see progression within a year, and 25% don’t respond. There is also a lack of predictive biomarkers for treatment selection.
Researchers conducted a retrospective study using phosphoproteomics on formalin-fixed and paraffin-embedded (FFPE) resected and Tru-Cut liver biopsies from patients with aHCC treated with AB to build a model predicting treatment response.
They extracted 10×10 µm sections of proteins from FFPE biopsies of 30 patients (etiology: 16 viral, 14 non-viral). After reversing crosslinks and tryptic digestion, mass spectrometry cleaned, enriched, and quantified peptides. Patients were divided into good responders (GRG, n=20, DoR >7.5 months) and poor responders (PRG, n=10, DoR <7.5 months), then used these features to train a random forest prediction model.
The results showed that 40 phosphopeptides were selected to build the AB response model, including previously described phosphorylation sites, pGStA1-3S202 and pHSPB1S9. The model accurately predicted outcomes for all good (n=20) and 7 poor responders, showing 100% sensitivity, 87% precision, and 70% specificity. It effectively stratified patients with log-rank (P<0.001, HR<0.1) with similar performances across viral ( DoR 17.1 vs. 0 months) and non-viral etiology (DoR 14.4 vs. 3.1 months). Notably, kinase-substrate enrichment analysis revealed significant (P<0.01) differences, including increased RAF-MEK-ERK pathway activity in some poor responders. The increased activity suggested that patients in PRG may have shown sensitivity to drugs targeting RAF kinases, such as sorafenib.
Investigators concluded that they’ve developed a preliminary model predicting AB response in aHCC using routine FFPE biopsies. After validating it in more patient groups, this model could fill a gap for clinical response biomarkers in aHCC.