Supplementary Appendix
Study Design and Patients
Approximately 4 months after the last patient underwent randomization, a protocol amendment allowed patients in the ofatumumab group who had PD, as confirmed by IRC, to receive ibrutinib. An independent data monitoring committee reviewed data from the protocol-specified interim analysis (January 2014), and recommended conclusion of the study and that all patients be permitted to crossover to ibrutinib.
The intent-to-treat (ITT) population was the primary population for analysis of efficacy endpoints and baseline characteristics. The primary endpoint was IRC-assessed PFS per IWCLL criteria with the 2012 clarification for treatment-related lymphocytosis (Hallek, Blood 2012). Key secondary endpoints included OS and ORR.
Briefly, 391 patients were randomized 1:1 to oral ibrutinib 420 mg once daily until disease progression (PD) or unacceptable toxicity, or intravenous ofatumumab for up to 24 weeks on a schedule of 300 mg followed by 2000 mg × 11 doses.
Procedures
Toxicity was graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) Version 4.0 and IWCLL criteria were used for hematologic toxicity. Post-baseline lymphocytosis was defined as an increase of at least 50% compared to baseline with an absolute count of ≥5 × 109/L. Immunoglobulin heavy chain variable region gene mutational analysis (IGHV) and CD38 expression were assessed by a central laboratory.
Mutation profiling results were reported by the Foundation One Heme Panel on evaluable baseline CD19+ enriched peripheral blood mononuclear cell (PBMC) samples. Technical aspects of the Foundation One Heme Panel are described previously (Frampton, Nature Biotechnology. 2013). Germline comparison was not performed; however, the panel has been developed to provide interpretation without a matched normal by removing germline variants from the 1000 Genomes Project (dbSNP135) (Frampton, Nature Biotechnology 2013). The panel reports known somatic and likely somatic variants of base substitutions, copy number changes (e.g. insertions, deletions), and/or rearrangements based on known driver alterations (COSMIC v62) highlighted as biologically significant.It also has been designed to evaluate the panel cancer-related genes, including TP53 and ATM, to a high uniform depth (targeting >500× coverage by non-PCR duplicate read pairs, with >99% of exons at coverage >100×) (Frampton, Nature Biotechnology 2013). During the validation, the panel was able to detect 99% of substitutions at MAF < 10% therefore supporting its capabilities to detect subclonal mutations.
Complex karyotype (CK) when performed was determined by local laboratories and did not routinely include stimulation. CK was defined as the presence of ≥3 cytogenetic abnormalities based on local karyotyping and reported by the investigator per local site assessment. Sites had the option to provide data from historical samples drawn within 90 days prior to randomization. Cytogenetic data were entered into the database by the site through electronic data capture. Stimulated karyotype analysis was not required.
Standard CLL fluorescence in-situ hybridization (FISH) probes were used by local laboratories to detect abnormalities in chromosomes 11q and 17p. Within 90 days prior to randomization, a peripheral blood sample or bone marrow sample (aspirate or biopsy) was tested for FISH analysis for stratification purpose. For subjects without lymphocytosis, it was requested that FISH be assessed from a bone marrow sample. A previous FISH and/or cytogenetic result demonstrating the presence of del17p per the assay specification was considered adequate for stratification provided appropriate documentation was available. Analysis of IGHV was performed by a central laboratory.
Statistical Analysis
PFS rate by subgroup at selected landmark point(s) was compared both between arms and within an arm based on the Z test. Overall PFS by subgroup was compared within an arm based on unstratified log-rank test. Overall response rate (ORR) by subgroup was compared both between arms and within an arm using Fisher’s exact test.
For multivariate Cox proportional hazards regression, the following parameters were tested: age (<65 vs. ≥65 y), Rai stage (0-II vs. III-IV), ECOG (≥1 vs 0), number of prior lines of therapy (1 vs >1), del11q (yes vs no), β2-microglobulin (yes vs no), disease refractory to purine analogues (yes vs no), del17p (yes vs no). Gene mutations (e.g., TP53 alone) and CK were not included in the multivariate analysis (MVA) given missing data in up to 24% of patients in the ibrutinib and ofatumumab arms. The threshold for the permittance of missing data in MVA is 10%.
For construction of observed and crossover-adjusted survival curves, a rank preserving structural failure time model (RPSFTM) was applied to estimate the counterfactual survival times for the ofatumumab patients. The model is fit under the assumption that patients who start on ibrutinib, stay on ibrutinib. Switching to subsequent therapy or discontinuing treatment is ignored. ‘Recensoring’ is applied to avoid informative censoring bias. This is done by recensoring the counterfactual survival times for the ofatumumab patients at their earliest possible censoring time (per guidelines from NICE Decision Support Unit ( accessed April 13, 2016).
Supplemental Figure Legends
Supplemental Figure 1. RESONATE™ PFS by subgroup or gene mutation
Supplemental Figure 2. PFS with ibrutinib by number of prior lines of therapy
Supplemental Figure 3. RESONATE™ observed and crossover-adjusted overall survival curves
Supplemental Figure 4. Overall survival with ibrutinib in patients by del17p/del11q subgroups
Supplemental Figure 5.Overall survival by number of prior therapies
Supplemental Figure 6. Lymphocytosis in ibrutinib-treated patients with mutated and unmutated IGHV
1