Functional Prognostic Model Predicts CLL PFS With Ibrutinib/Venetoclax

A functional prognostic model built from 6 laboratory features of leukemic cells and their microenvironment outperformed existing clinical scores in predicting progression-free survival (PFS) among patients with relapsed/refractory (R/R) chronic lymphocytic leukemia (CLL) treated with ibrutinib (Imbruvica; Pharmacyclics/Janssen) plus venetoclax (Venclexta; AbbVie/Genentech), according to a study published in HemaSphere.¹

The model combined ex vivo sensitivity to 3 drugs, 2 ratios of protein expression or phosphorylation, and the proportion of CD8-positive T cells. A higher proportion of CD8-positive T cells was associated with an increased risk of progression. Greater ex vivo drug sensitivity was associated with more favorable PFS. None of the established genetic markers, including immunoglobulin heavy chain variable (IGHV) mutational status and TP53 aberrations, were selected into the final model.

“Our findings demonstrate that patterns of ex vivo drug sensitivity profiles can predict response to a given but different treatment, as previously reported also for other hematologic malignancies,” the authors wrote. “The drugs that we identified as predictive share a common mechanism wherein they inhibit signals from the tumor microenvironment that the CLL cells depend on for survival.”

Historic Prognostic Markers Could Fall Short in Relapsed/Refractory CLL

Prognostic markers in CLL, such as unmutated IGHV and TP53 aberrations, were defined in the era of chemoimmunotherapy, and the authors noted that fixed-duration targeted combinations have produced deep responses even in patients with these markers. A reassessment of the International Prognostic Index for CLL (CLL-IPI) found that targeted therapies outperformed chemoimmunotherapy across all risk groups.² In the phase 2 VISION/HO141 trial (NCT03226301) of minimal residual disease-guided ibrutinib plus venetoclax, estimated 4-year PFS was 81% overall at a median follow-up of 51.7 months.³

“However, since response to treatment remains heterogeneous also in the era of targeted therapies, novel prognostic markers are needed to optimally tailor treatments,” the authors of the new study wrote.¹

How the Functional Prognostic Model for CLL Was Built

Researchers analyzed peripheral blood mononuclear cells collected before treatment from 177 patients with R/R CLL enrolled in VISION/HO141. The cohort was high risk, with 17% having del(17p), 25% having mutated TP53, and 63% having unmutated IGHV. Cells underwent ex vivo testing against 95 single agents and 87 drug combinations, flow cytometry profiling of 31 intracellular proteins and phosphoproteins, and immunophenotyping.

The first batch of 157 patients was split into training (n = 80) and test (n = 77) sets stratified by IGHV and TP53 status, and a second batch of 20 patients served as a blinded validation set. The final 6 features were:

• Sensitivity to VEGFR-2/EGFR inhibition
• Sensitivity to JAK1/2 plus Bcl-2 inhibition
• Sensitivity to Bruton tyrosine kinase (BTK) plus PI3K inhibition
• BCL-2 to BTK (pY223) ratio
• p90RSK (pS380) to PLCγ2 (pY759) ratio
• CD8-positive T-cell proportion

Functional Prognostic Model vs CLL-IPI for Predicting Progression-Free Survival

In the test set, the functional prognostic model achieved a Harrell C-index of 0.766 and a time-averaged area under the receiver operating characteristic curve (ROC-AUC) of 0.812. In the blinded validation set, the C-index was 0.815 and the time-averaged ROC-AUC was 0.805. In a combined test and validation subset of 66 patients with complete data for established scores, the functional prognostic model separated high- and low-risk groups (adjusted P = .011), whereas the CLL-IPI did not (adjusted P = .679).

Ex vivo sensitivity to the trial drugs alone did not predict outcome, with a hazard ratio of 0.993 for venetoclax (P = .852), 0.830 for ibrutinib (P = .224), and 0.968 for the combination (P = .391). Removing any 1 of the 3 data types made the PFS difference between risk groups lose significance, and removing drug sensitivity data reduced performance the most. In an external cohort of 26 patients with R/R CLL treated with umbralisib, the model predicted treatment response with an ROC-AUC of 0.706 (P = .043).

The authors acknowledged that all models showed some overfitting given the relatively small sample size and that the model was developed from predefined functional profiles, so other unknown features could improve predictions. The umbralisib cohort lacked PFS data, and the original risk threshold could not be tested there. The authors also noted that the assays use flow cytometry and a luminescent cell viability test and called the approach relatively low cost.

“Since our study was limited to retrospective analyses of R/R CLL, the model needs to be validated in independent patient cohorts, including targeted combination regimens for frontline CLL, and in a prospective clinical trial, before it may be implemented in clinical practice,” the authors concluded.

References

1. Hermansen JU, Kong W, Brodersen AM, et al. A functional prognostic model predicts progression-free survival in patients with relapsed chronic lymphocytic leukemia treated with ibrutinib + venetoclax. HemaSphere. 2026;10(9):e70467. doi:10.1002/hem3.70467

2. Langerbeins P, Giza A, Robrecht S, et al. Reassessing the chronic lymphocytic leukemia International Prognostic Index in the era of targeted therapies. Blood. 2024;143(25):2588-2598. doi:10.1182/blood.2023022564

3. Niemann CU, Dubois J, Nasserinejad K, et al. Long-term follow-up of MRD-guided ibrutinib plus venetoclax in relapsed CLL: phase 2 VISION/HO141 trial. Blood Adv. 2025;9(15):3665-3675. doi:10.1182/bloodadvances.2024015180