Clinical and Biological Indicators of Duvelisib Efficacy in CLL from the Phase 3 DUO™ Study

Jennifer R. Brown, Ian W. Flinn, Matthew S. Davids, Peter Hillmen, Marco Montillo, Zsolt Nagy, Julio Delgado, Bryone Kuss, Constantine Tam, Ulrich Jaeger, Paolo Ghia, Stephan Stilgenbauer, Carol Moreno, NgocDiep Le, Stephanie Lustgarten, Kam Sprott, Jonathan Pachter, David T. Weaver, Florence Cymbalista

Research output: Contribution to journalMeeting Abstractpeer-review



In CLL, the timing and selection of therapy is largely informed by disease stage, prognostic molecular features, tumor load, and patient (pt) performance status. Duvelisib (DUV) is a novel, oral, dual PI3K-δ,γ inhibitor in clinical development for the treatment of CLL/SLL, FL, and other hematologic malignancies. Results from the Phase 3 DUO study (NCT02004522) for relapsed/refractory CLL/SLL showed that DUV monotherapy resulted in statistically significant improvement over ofatumumab (OFA) monotherapy for median progression-free survival (mPFS; 13.3 vs 9.9 months (mo); p<0.0001), overall response rate (ORR; 73.8% vs 45.3%; p<0.0001), and lymph node response rate (85.0% vs 15.7%; p<0.0001) per IRC assessments, with a median overall follow-up of 22.4 mo for both arms (Flinn et al., 2018). In order to identify pts from DUO who were more likely to respond to DUV, multiple markers, including tumor burden, cytogenetics, immune cell profiles, and serum factors were correlated with clinical responses in an exploratory analysis.


Of the 319 pts enrolled in the DUO study, 158 were treated with DUV and 155 with OFA. Tumor burden was assessed at baseline using target lymph node diameter and absolute lymphocyte count. Baseline blood samples from Cycle1 Day1 were analyzed for cytogenetic abnormalities (17p, 11q, 6q deletions; trisomy 12); immune cell counts by flow cytometry; and serum chemokine, cytokine, and serum factor levels. Univariate analysis of these markers was conducted for PFS and ORR. The statistically significant univariate markers were tested in a stepwise multivariate regression (MVR) model using binary high [H] or low [L] thresholds based on the median.


Tumor Burden: The mPFS (in mo) for DUV monotherapy did not appear to be impacted by baseline tumor burden: high (n=20), 16.6; medium (n=88), 12.7; and low (n=47), 15.1. Across these tumor burden levels, DUV maintained its PFS advantage over OFA.

Cytogenetics: DUV-treated pts with del(11q), a marker associated with poor CLL outcomes, had a longer mPFS (in mo) and ORR compared to pts without del(11q) (mPFS: 24.8, n=25 vs 12.7, n=81; HR 0.56, 95% CI [0.30,1.06]; ORR: 80.0% vs 69.1%, odds ratio, 0.56). In contrast, mPFS was shorter in OFA-treated del(11q) pts relative to those without this deletion (5.3, n=24 vs 11.3, n=71). Median PFS was significantly extended in the del(17p) subgroup for DUV vs OFA (16.6, n=27 vs 9.2, n=27; HR 0.42, [0.21,0.85]), and, to a lesser degree, in pts without del(17p) (13.1, n=83 vs 11.1, n=74; HR 0.55 [0.37,0.83]). For the 6 DUV pts with del(11q/17p), mPFS was 17.4 mo. A shorter mPFS was observed for DUV-treated pts with trisomy 12 compared with trisomy 12-negative pts (9.1, n=16 vs 16.5, n=73). For the 10 DUV pts with trisomy 12 as well as del(11q/17p), mPFS was more similar to the trisomy 12 group, at 9.1 mo.

Immune cell profiles and serum factors: Several immune cell profiles correlated with longer mPFS (in mo) for DUV, including Treg (16.4H vs 12.9L, HR 0.69 [0.43,1.12]) and monocytes (15.1L vs 12.7H, HR 0.82 [0.54,1.24]). Baseline chemokine and serum factor levels associated with longer mPFS with DUV included CCL3 (22.1L vs 12.2H, HR 0.60 [0.37,0.99]), IL2RA (16.3L vs 11.6H, HR 0.73 [0.47,1.14]), and TNFα (16.6L vs 12.2H, HR 0.78 [0.49,1.24]).

Multivariate regression model: The MVR model showed that the biomarker profile of trisomy 12-negative (HR -/+, 0.15), CCL22L (HR H/L, 2.23), TNFαL (HR H/L, 3.66), TregH (HR H/L, 0.31), and monocytesL (HR H/L, 2.45), was associated with longer PFS; TNFαL and TregH were the largest associations, with estimated 73% and 69% reductions in risk of progression or death, respectively. Similarly, stepwise models yielded TNFαL (odds ratio H/L, 0.37) and CXCL11H (odds ratio H/L, 2.99) as biomarkers for improvement in ORR. Based on these analyses, pts with TNFαL tend to have better efficacy profiles for PFS and/or ORR.

Conclusions: DUV monotherapy was highly efficacious in CLL/SLL pts with markers of poor response, including high tumor burden, del(17p), and del(11q) as shown by univariate analyses of tumor burden, cytogenetics, immune cell profiles, and serum factors. Multivariate analyses revealed a biomarker profile of CCL22L, TNFαL, TregH, monocytesL, and trisomy 12-negative that correlated with improved mPFS and/or ORR. Additional analyses are underway to characterize the predicted PFS for the biomarker profile.
Original languageEnglish
Pages (from-to)1856
Number of pages5
Issue numberS1
Publication statusPublished - 29 Nov 2018


  • Duvelisib Efficacy
  • PI3K-δ,γ inhibitor
  • hematologic malignancies


Dive into the research topics of 'Clinical and Biological Indicators of Duvelisib Efficacy in CLL from the Phase 3 DUO™ Study'. Together they form a unique fingerprint.

Cite this