Page 111 - 2020_11-Haematologica-web

Prognostic impact of CLL stereotyped subsets
hypermutation (SHM) status, genomic aberrations, and gene mutations such as SF3B1 and NOTCH1 detected by next-generation sequencing (NGS).1-4 Based on IGHV gene mutational status, CLL can be divided into mutated (M- CLL) and unmutated (U-CLL) with indolent and aggres- sive disease courses, respectively.5,6
The combinatorial and junctional diversity of the IGHV- IGHD-IGHJ recombination along with the SHM mecha- nism can lead to the potential synthesis of almost 1012 dif- ferent IG. At odds with this diversity, the B-cell receptor (BcR) IG of about one-third of CLL display highly homol- ogous variable heavy complementarity-determining region 3 (VH CDR3), which led to their being grouped into different subsets carrying (almost) identical alias stereotyped BcR IG.7-12 Subset #1 represents approximate- ly 5% of U-CLL and is characterized by the combination of a heavy chain IGHV1-5-7/IGHD6-19/IGHJ4 gene rearrangement with a light chain IGKV1-39/IGKJ1-2 gene rearrangement.10 The IGHV gene bears little or no SHM and the VH CDR3 length is 13 amino acids (aa) long. Subset #1 is associated with a poor outcome in terms of patient survival and short time-to-first-treatment (TTFT) in comparison to U-CLL using the same IGHV genes.13-15 Subset #2 represents 3% of all CLL and is defined by the IGHV3-21/IGLV3-21 combination with a short VH CDR3 of 9 aa.10 Of note, subset #2 comprises both U-CLL and M- CLL cases and has been associated with an aggressive clin- ical course, irrespective of SHM status.16-18 Subset #4 is the largest subset of M-CLL, carrying a BcR IG that consists of a heavy chain IGHV4-34/IGHD5-18/IGHJ6 gene rearrangement (20 aa long VH CDR3) and a light chain IGKV2-30/IGKJ1-2 rearrangement.10 In previous studies, subset #4 was associated with indolent disease, enriched in young patients, and a long TTFT.11,15,19 Finally, subset #8 is composed of cases with unmutated IGHV4-39/IGHD6- 13/IGHJ5 gene rearrangements. Furthermore, in a case control study, a strong association with Richter syndrome and poor outcome was reported.8,20
Using the data of four prospective multicenter clinical trials of the German CLL study group (GCLLSG), the pri- mary objective of the current study was to search for asso- ciations between the most common and best character- ized CLL stereotyped subsets (#1, 2, 4 and 8) and disease characteristics as well as outcome, and also compare these findings against non-subset U-CLL and M-CLL. The sec- ondary objective was to study the prognostic value of sub- set #2 irrespective of its IGHV mutational status.
Methods
Study population
To assess the prognostic value of the most prevalent CLL sub- sets, we evaluated patients recruited in four randomized phase III trials conducted by the GCLLSG. None of the patients had received any prior treatment and all had a diagnosis of CLL according to the International Workshop on Chronic Lymphocytic Leukemia criteria.1 All trials were approved by the leading ethics committee. Written informed consent was obtained from all patients according to the Declaration of Helsinki. To provide data about the clinical course of newly diagnosed patients not requiring treatment, we evaluated an “early-stage CLL cohort” of 710 asymptomatic Binet stage A patients from the CLL1-trial (clinical trials.gov identifier: NCT00262782). The study compared a ‘watch and wait’ strategy to upfront fludarabine (F) monotherapy.21 In
total, 639 (90.0%) patients were followed with the ‘watch and wait’ approach. Patients treated with F (71, 10.0%) were excluded from our analyses.
To evaluate the clinical course of patients needing treatment, we evaluated an “advanced-stage cohort” from three phase III trials enrolling patients with CLL requiring front-line treatment. The CLL8-trial (clinical trials.gov identifier: NCT00281918) included 817 fit patients and compared fludarabine and cyclophosphamide (FC) to FC plus rituximab (FCR).22,23 The CLL10-study (clinical trials.gov identifiers: NCT00262782 and NCT2000769522) included 561 fit patients and compared FCR to bendamustine and rituximab.24 Finally, the CLL11-study (clinical trials.gov identifier: NCT01010061) enrolled 781 unfit patients and compared chlorambucil with or without rituximab or obinutuzumab.25
To evaluate the first objective, we classified patients into six groups: subsets #1, #2, #4, #8, non-subset M-CLL, and U-CLL. To evaluate the second objective, we classified patients into three cat- egories: (i) subset #2; (ii) IGHV3-21 rearrangement not meeting the subset #2 criteria (“IGHV3-21”); and (iii) all other cases (“IGHV”). Each group was sub-divided into mutated (“m”) and unmutated (“u”) cases, resulting in six subgroups.
Biological markers and clinical characteristics
Baseline clinical and laboratory characteristics evaluated for associations and potential prognostic relevance are listed in Tables 1 and 2. Detailed descriptions of the diagnostic methods have been published previously.3,8,26-30
Statistical analysis
Time-to-first-treatment was defined as the time between diag- nosis and date of first treatment. For the advanced stage CLL cohort, time-to-next-treatment (TTNT), progression-free survival (PFS) and overall survival (OS) were calculated from randomiza- tion to the initiation of subsequent treatment, disease progression or death, and death, respectively. For the early stage cohort, PFS and OS were calculated from diagnosis. Subjects without an event were censored at the time of last assessment. Survival rates were estimated using the Kaplan-Meier method and compared using non-stratified log-rank tests. Hazard ratios and 95% confidence intervals were calculated using Cox proportional hazards regres- sion model.31 All variables that showed significant association with TTFT, TTNT, PFS or OS in univariate Cox regression analy- ses were included in multivariable analyses applying forward and backward stepwise selection procedures. Adjustments for multi- ple testing were not applied and all reported P-values have an exploratory character. We performed two analyses in both cohorts. In the first one, we compared subsets #1, 2, 4 and 8 indi- vidually to non-subset U-CLL and M-CLL, and in the second one, we evaluated the prognostic value of subset #2 in terms of muta- tional status and IGHV3-21 usage.
Results
Subset assignment and distribution overview
Of 639 patients from the CLL1 'watch and wait' cohort, immunogenetic data were available for 592 (92.6%); 402 (67.9%) patients carried mutated IGHV genes and 190 (32.1%) patients expressed unmutated IGHV genes. Eight cases (1.35%) were assigned to subset #1, 16 cases (2.7%) to subset #2, 8 cases (1.35%) to subset #4, and 2 cases (0.34%) to subset #8. With regard to the rest of the cases, 174 (29.4%) belonged to non-subset U-CLL and 384 (64.9%) to M-CLL. For the second analysis differentiating the impact of subset #2, we found 29 cases (4.9%)
haematologica | 2020; 105(11)
2599