Background - Although there has been a revolution in the treatment of chronic lymphocytic leukemia (CLL), the challenge remains to identify the right drugs for the right patients. It is widely accepted that CIT, including the 'gold standard' fludarabine, cyclophosphamide and rituximab (FCR), is contraindicated for patients with TP53 disruption and, more recently, unmutated IGHV genes. Also patients with short, dysfunctional telomeres were shown to have inferior outcomes when treated with FCR-based regimens. To date, a role for CD49d in this setting has not been established.

Aims - Here we evaluated the ability of telomere lenght (TL) and CD49d to cooperate with IGHV gene status to predict progression-free survival (PFS) in patients treated with FCR-based regimens in the frontline setting in three UK trials, ARCTIC, ADMIRE and CLL4.

Methods - The study included a discovery cohort of 245 CLL treated with FCR/FCR-like regimens according to the two UK trials ARCTIC and ADMIRE. As there was no significant difference in PFS between the three arms of the study (P = 0.97), analysis was performed on the combined cohort. The median follow-up was 77.5 months with 157 progressions and 76 deaths. Twenty-nine patients were TP53 deleted and/or mutated, with shorter PFS compared to cases without TP53 disruption (final cohort, 216 TP53 wild-type CLL). The validation cohort was composed of 119 CLL samples derived from patients randomised to receive fludarabine, cyclophosphamide (FC) from the UK CLL4 trial. The median follow-up was 67.2 months with 99 progressions and 77 deaths. Fifteen CLL were TP53 mutated/deleted, with shorter PFS compared to cases without TP53 disruption (final cohort, 104 TP53 wild-type CLL). TL was measured using the high-throughput STELA assay and patients were bifurcated into two groups with either short telomeres inside the fusogenic range (TL-IFR) or long telomeres outside the fusogenic range (TL-OFR). CD49d was measured by flow cytometry and dichotomized as CD49dpos and CD49dneg based on the established 30% cut-off. For IGHV gene status, the 2% cutoff was used to split patients in mutated (IGHV-M) and ummutated (IGHV-UM).

Results - In the 216 CLL with wild-type TP53 status from the ARCTIC/ADMIRE trials, CD49d expression was a predictor of PFS (P=0.02; HR=1.46 [1.03-2.06]). In keeping with previous reports, patients with IGHV-UM genes (P<.0001; HR=2.53 [1.79-3.58]) or TL-IFR (P=0.0002; HR=1.97 [1.30-2.98]) showed also significantly shorter PFS. This data was used as a starting point for a risk-stratification algorithm (Figure 1A). IGHV-UM cases could not be further dissected by TL or CD49d expression (P=0.76 log-rank test), their 8-year PFS being just 19.0%; (HR=5.58 [3.70-8.42]). In contrast, the IGHV-M group could be stratified by TL and CD49d expression (P<0.001 log-rank test). In particular, 13/84 (15.5%) of IGHV-M CLL with TL-IFR showed a median PFS of 3.0 years with a 8-year PFS of 15.4% (HR=6.45 [1.84-22.58]), similar to IGHV-UM cases (P = 0.19). Patients with IGHV-M genes and TL-OFR could be further stratified by CD49d expression into categories with different 8-year PFS: 43.1% in IGHV-M/TL-OFR/CD49dpos cases (HR=2.52 [1.08-5.89]), and 75.5% in the IGHV-M/TL-OFR/CD49dneg reference group (Table 1). Figure 1B shows the overlaid Kaplan-Meier curves for this hierarchical stratification. The proposed algorithm was then evaluated in the UK CLL4 trial. In keeping with the known inferiority of FC when compared with FCR, the 8-year PFS was just 5.2% in the IGHV-UM subset, irrespective of TL or CD49d expression. In concordance with the discovery cohort, the IGHV-M subset with TL-IFR had a similar median PFS to IGHV-UM cases (P=0.84). Furthermore, CD49d expression was able to stratify cases with TL-OFR; in particular, 18/104 (17.3%) were IGHV-M/TL-OFR/CD49dneg CLL which had a 8-year PFS of 77.8% (summary in Table 1 and Figure 1C).

Conclusion - Our analysis shows that only IGHV-M/TL-OFR/ CD49dneg patients may benefit from CIT; this group represented just 56/321 cases (17.4%) of the combined cohort, suggesting that approximately 82% of patients should be considered for alternative therapies. Incorporation of this algorithm into clinical trial design and real-world practice would enable rational, risk-adapted, clinical management with the aim of treating all CLL patients with the optimal therapeutic regimen in the frontline setting.

Disclosures

Norris:TeloNostiX Ltd: Current equity holder in private company, Patents & Royalties. Hillmen:Gilead: Other: Financial or material support, Research Funding; Alexion: Consultancy, Research Funding, Speakers Bureau; Apellis: Consultancy, Research Funding, Speakers Bureau; AstraZeneca: Consultancy, Speakers Bureau; Acerta: Other: Financial or material support; Roche: Consultancy, Other: Financial or material support, Research Funding, Speakers Bureau; AbbVie: Consultancy, Other: Financial or material support, Research Funding, Speakers Bureau; Pharmacyclics: Other: Financial or material support, Research Funding; Janssen: Consultancy, Other: Financial or material support, Research Funding, Speakers Bureau. Rawstron:BD Biosciences (Intrasure): Patents & Royalties. Baird:TeloNostiX Ltd: Current equity holder in private company, Patents & Royalties. Fegan:TeloNostiX Ltd: Current equity holder in private company, Patents & Royalties. Pepper:TeloNostiX Ltd: Current equity holder in private company, Patents & Royalties.

Author notes

*

Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract