Targeted multigene deep sequencing of Bruton tyrosine kinase inhibitor–resistant chronic lymphocytic leukemia with disease progression and Richter transformation
Abstract
Background
In a proportion of patients with chronic lymphocytic leukemia (CLL), resistance to Bruton tyrosine kinase (BTK) inhibitors (BTKi) is attributed to acquired BTK/phospholipase C gamma 2 (PLCG2) mutations. However, knowledge regarding additional genetic lesions associated with BTK/PLCG2 mutations, and gene mutations in patients lacking BTK/PLCG2 mutations, is limited.
Methods
Using targeted deep sequencing, mutations in 29 genes associated with CLL and/or the BCR signaling pathway were assessed in patients with CLL who developed resistance to BTK inhibition with ibrutinib/acalabrutinib at a single institution.
Results
The study group included 29 patients with BTKi-resistant CLL, 23 patients with disease progression, and 6 patients with Richter transformation (RT). The median times to disease progression and RT were 33.3 months and 13.3 months, respectively. In 11 patients, sequencing was possible at both baseline (prior to treatment with BTKi) and at time of disease progression/RT. Of these patients, 4 demonstrated BTK mutations at the time of disease progression/RT; patients without BTK mutations frequently acquired mutations associated with disease progression/RT in TP53, SF3B1, and CARD11, whereas additional mutations were rare in patients with BTK-mutated CLL. Sequencing of all 29 patients at the time of disease progression/RT identified BTK mutations at a frequency of 66%, including a novel V537I mutation. Among patients with disease progression, BTK mutations were observed in 16 patients (70%). The median time to disease progression was shorter in patients without BTK mutations compared with those with BTK-mutated CLL. Among patients with RT, SF3B1 mutations were more frequent than BTK mutations (67% vs 50%). Following BTKi discontinuation, we sequential mutation analysis was performed in 2 patients with RT and 3 patients with disease progression in the setting of persistent disease. Both patients with RT demonstrated disappearance of BTK and expansion of TP53 mutations. All 3 patients with disease progression received venetoclax and demonstrated suppression of BTK mutations.
Conclusions
Longitudinal, targeted, multigene deep sequencing is informative for the clinical monitoring of mutational evolution in patients with CLL who are receiving BTKi.
Introduction
The neoplastic B cells of chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) are driven by activation of B-cell receptor (BCR) signaling.1, 2 Bruton tyrosine kinase (BTK) encodes for a kinase that is a key component of the BCR signaling pathway. Inhibitors of BTK (BTKi) such as ibrutinib and acalabrutinib covalently bind to BTK, leading to abrogation of BCR signaling, and patients treated with BTKi achieve durable remissions.3-7
Despite the success of BTKi therapy, a subset of patients treated with BTKi eventually develop drug resistance. These patients relapse either as progression of CLL or undergo histologic transformation to a higher grade lymphoma, also known as Richter transformation (RT).2, 8-15 Patients with CLL who are resistant to BTKi have a poor outcome.8, 16, 17 Multiple studies have shown that a substantial percentage of these patients acquire mutations that lead to reactivation of BCR signaling and resistance. These include mutations in BTK (C481) that disrupt irreversible covalent binding of BTKi to the enzyme and mutations in phospholipase C gamma 2 (PLCG2) downstream of BTK, leading to reactivation of BCR signaling.2, 9, 10, 14, 18 Recently, studies have confirmed a high frequency of mutations in these 2 genes in larger series,2 especially in patients who develop progressive CLL. The absence of BTK mutations prior to the initiation of therapy at a burden detectable using standard techniques, combined with the available evidence regarding its role in driving BTKi resistance, support that these mutations are predominant mechanisms underlying BTKi resistance. However, there remains a subset of patients who develops resistance without acquiring either of these mutations15, 19; in these patients, the molecular alterations driving BTKi resistance to our knowledge still are unclear. One study by Burger et al has demonstrated expansion of del(8p) clones in 3 of 5 patients with CLL without BTK/PLCG2 mutations variably associated with gene mutations in E1A binding protein P300 (EP300), MLL2 (KMT2D), and eukaryotic translation initiation factor 2A (EIF2A).15 Furthermore, the role of BTK/PLCG2 mutations in drug resistance when present at a low allelic burden compared with tumor, as can be observed in a large subset of patients, is not clear. These findings suggest that additional genetic or epigenetic lesions may contribute to disease resistance to BTK inhibitors. However, to the best of our knowledge, there currently are limited data regarding additional molecular and cytogenetic lesions associated with BTK/PLCG2 mutations, and genetic mechanisms of disease relapse in patients lacking these mutations.
Understanding the mechanisms of disease resistance is essential for designing and developing therapies for these patients. In addition to mutations in BTK/PLCG2, it is important to identify genetic alterations in additional CLL-associated genes. In the current study, using targeted deep sequencing of 29 genes associated with CLL, including targets of the B-cell receptor signaling pathway, we demonstrate the spectrum of gene mutations associated with CLL in 29 patients who experienced disease progression or RT while receiving BTKi therapy (ibrutinib or acalabrutinib) in a single institution. Among 11 patients who had available paired samples for testing at baseline (prior to BTKi therapy) and at the time of BTKi resistance, we identified mutations that were associated with disease recurrence in addition to BTK. At the same time, we have demonstrated the feasibility of multigene, targeted next-generation sequencing (NGS)–based assays for the assessment of gene mutations in patients receiving BTK inhibitors.
Materials and Methods
Patient Cohort
The study group included patients with CLL who either developed disease progression or underwent histologic transformation to higher grade lymphoma while receiving BTKi-based therapy (ibrutinib or acalabrutinib) and who had DNA available for sequencing analysis at the time of disease recurrence/progression or histologic transformation. All samples were derived from a single institution. The study was approved by the institutional review board. Informed consent was obtained in accord with the Declaration of Helsinki. We reviewed the bone marrow (BM) and peripheral blood (PB) morphologic findings and the results of flow cytometry immunophenotypic studies for all cases. Clinical data were collected from the medical records.
Fluorescence in Situ Hybridization
Interphase fluorescence in situ hybridization (FISH) analysis was performed on all cases using a US Food and Drug Administration–approved multicolor probe panel for the detection of 11q22.3 (ataxia telangiectasia mutated [ATM]), 13q14.3 (D13S319 locus located between RB1 and D13S25 loci), 13q34 (subtelomere LAMP1), 17p13.1 (TP53), and the centromeric region of chromosome 12 (D12Z3) (Abbott Molecular, Des Plaines, Illinois). At least 200 nuclei were analyzed for each probe and the percentage of abnormal nuclei divided by the total nuclei scored was calculated. The normal cutoff value for each probe set was established following standard laboratory procedures.20
Molecular Analysis
NGS-based mutation analysis
For sequencing, we used either PB or BM samples from patients with CLL who had at least 10% clonal CD19/CD5–positive cells as evaluated by flow cytometry. In a subset of patients with available samples from baseline (prior to receiving BTKi therapy) and at the time of disease progression or transformation to high-grade lymphoma (after BTKi therapy), we performed sequencing at both time points and others when available. None of the samples underwent B-cell enrichment or sorting.
Genomic DNA was extracted from the whole BM or PB mononuclear cell using an Autopure extractor (Qiagen, Valencia, California) and quantified using the Qubit DNA BR assay kit (Life Technologies, Carlsbad, California). Approximately 200 ng of genomic DNA was used for library preparation using the HaloPlex Target Enrichment System (Agilent Technologies, Santa Clara, California) using a custom-designed gene panel composed of the coding regions of 29 CLL-related genes (including 3530 amplicons that cover a total region of 82 kilobases) in a clinical molecular diagnostics laboratory. The panel was designed to detect the known clinically significant mutations in the targeted genes. Sequencing was performed on Miseq sequencer (Illumina, San Diego, California) for a total of 302 cycles of bidirectional sequencing. This high-throughput, targeted, high-coverage, NGS-based assay examined the following genes: ATM, BIRC3, BTK, CALR, CARD11, CD79A, CD79B, CHD2, CSMD3, CXCR4, DDX3X, EZH2, FAT1, FBXW7, KLHL6, LRP1B, MAPK1, MUC2, MYD88, NOTCH1, PLCG2, PLEKHG5, POT1, SF3B1, SPEN, TGM7, TP53, XPO1, and ZMYM3. The covered regions are shown in Supporting Table S1. A minimum of 250 read coverage was required for variant calling. Specifically, a median of 3000× coverage was obtained for BTK (exon 8 [codons 197-259] and exons 15-16 [codons 450-544]) and PLCG2 (exon 19 [codons 645-685]), with a predicted detection sensitivity of 1% for BTK and PLCG2. MiSeq Reporter software (Illumina) was used for variant calling against human genome build 19 (hg19) as the reference genome. The Integrative Genomics Viewer (IGV) was used for visualization of calls and read alignment. An in-house postvariant calling analysis tool, OncoSeek (version 1.6.1.400; OncoSeek Limited, Hong Kong), was used for annotation. The somatic nature of a mutation was determined by searching the Catalogue Of Somatic Mutations In Cancer (COSMIC) and other available databases and the literature. Single-nucleotide polymorphisms listed in the Short Genetic Variations database (dbSNP) 137/138 and Genome 10K genome projects were excluded.21, 22
Somatic Hypermutation
Total RNA from baseline samples was extracted from PB or BM specimens followed by reverse transcription. Sequence analysis of the immunoglobulin heavy-chain variable region (IGHV) was performed by Sanger sequencing as described elsewhere.23 Patient sequences were compared with reference sequences using the V-BASE 2 database. Deviation from the germline >2% was considered mutated, and a deviation of ≤ 2% was considered unmutated.
Results
Clinical Characteristics
In total, there were 29 patients with CLL who either underwent progression (23 patients) or transformation to high-grade lymphoma/RT (6 patients) while receiving treatment with a BTKi-based regimen and who had DNA available for targeted NGS-based sequencing at the time of disease progression or histologic transformation. A total of 27 patients were treated with ibrutinib and 2 patients received acalabrutinib. For comparison, we used a control group of 62 patients with CLL who did not develop progress while receiving therapy. The baseline clinicopathologic characteristics of the study and control groups including FISH (prior to the initiation of therapy with a BTKi) and IGHV mutation status are shown in Table 1. Patients who developed disease progression while receiving BTKi therapy had a significantly higher frequency of del(17p) (62% vs 29%; P = .008) compared with those who did not progress. Both patients with disease progression and non-progressors demonstrated a high frequency of germline IGHV mutations (78% vs 87%) without any significant differences. The variable region of the immunoglobulin heavy chain utilized in patients with disease progression in decreasing order of frequency included VH 1-69, 3-11, 3-30 and 3-48. Among patients with CLL who never developed disease progression while receiving BTKi therapy, the variable region used in decreasing order of frequency included VH 3-23, 3-33, and 3-30. The median duration of treatment until disease progression was 33.3 months (range, 13.6-70.3 months). The median duration of treatment with BTKi until histologic transformation was shorter at 13.3 months (range, 1.7-30.8 months). Over a median follow-up duration of 40.6 months, 14 of 35 patients died.
| Clinicopathologic Parameter |
CLL Patients With Progression/RT N = 29 |
CLL Patients Who Did Not Develop Disease Progression/RT N = 62 |
|
|---|---|---|---|
| Age | Median age (range), y | 67.5 (39-84) | 67 (43-83) |
| Sex | Men | 19 (66%) | 46 (74%) |
| Women | 10 (34%) | 16 (26%) | |
| Rai stage | 0, I, or II | 13 (45%) | 30 (48%) |
| III or IV | 16 (55%) | 32 (52%) | |
| BTKi | Ibrutinib | 27 | 62 (100%) |
| Acalabrutinib | 2 | 0 | |
| Disease recurrence | CLL progression | 23 | NA |
| RT | 6 | NA | |
| FISH studies (prior to BTKi therapy) | Negative | 1 (4%) | 10 (16%) |
| ATM deletion | 9 (35%) | 24 (39%) | |
| Trisomy 12 | 4 (15%) | 10 (16%) | |
| Deletion D13S319 | 19 (73%) | 33 (53%) | |
| Deletion 17p | 16 (62%)a | 18 (29%)a | |
| IGHV mutation | Germline | 18/23 (78%) | 39/46 (85%) |
| Mutated | 5/23 (22%) | 7/46 (15%) | |
| Follow-up | No. died | 11 (38%) | 30 (48%) |
| Median follow-up duration from the onset of BTKi treatment (range), mo | 40.6 (2-99) | 27 (2-71) | |
| Median duration of BTKi treatment until disease progression or RT (range), mo | 25.4 (1.7-70.3) | NA | |
| Median duration of BTKi treatment until disease progression (range), mo | 33.3 (13.6-70.3) | NA | |
| Median duration of BTKi treatment until RT (range), mo | 13.3 (1.7-30.8) | NA | |
- Abbreviations: ATM, ataxia telangiectasia mutated; BTK, Bruton tyrosine kinase; BTKi, Bruton tyrosine kinase inhibitor; CLL, chronic lymphocytic leukemia; FISH, fluorescence in situ hybridization; IGHV, immunoglobulin heavy-chain variable region; NA, not applicable; RT, Richter transformation.
- a P < .001.
Molecular Analysis
Comparison of mutation changes at baseline and disease progression/RT in paired CLL samples
First, to identify disease progression or RT–associated gene mutations, we analyzed the mutation patterns in 11 patients on whom paired DNA samples were available (Fig. 1) (Table 2): 1 sample before the initiation of BTKi therapy and 1 sample at the time of disease progression (9 patients) or RT (2 patients). Of these 11 patients, 9 patients (82%) demonstrated gene mutation changes at the time of disease progression/RT, whereas 2 patients did not demonstrate any changes in gene mutations. All 9 patients gained mutations in ≥ 1 genes. Disease progression/RT–associated mutations included BTK (4 patients; 36%), TP53 (3 patients), SF3B1 (2 patients), POT1 (1 patient), CHD2 (1 patient), MUC2 (1 patient), NOTCH1 (1 patient), and CARD11 (1 patient). Recurrent alterations that were observed in ≥ 2 patients included mutations in BTK, TP53, and SF3B1. Three patients demonstrated loss of previously detected gene mutations at the time of disease progression/RT: BIRC3 (2 patients) and TP53 (1 patient).
| Patient Identifier/Age, Years/Sex | IGHV | VH Family | BTKi Duration, Months | Before BTKi | Disease Progression/RT Mutations (Allelic Burden) | Status, Time Since Resistance |
|---|---|---|---|---|---|---|
| 6/59/Man | G | 3-48 | 53.3 | TP53 p.C275F 0.02 | Alive, 28 mo | |
| TP53 p.C275Y 0.10 | TP53 p.C275Y 0.01 | |||||
| TP53 p.R175H 0.05 | ||||||
| TP53 p.Y234H 0.12 | ||||||
| TP53 p.R273C 0.12 | ||||||
| TP53 p.P278S 0.02 | ||||||
| TP53 p.H193D 0.28 | ||||||
| CHD2 p.S658fs 0.05 | ||||||
| BTK p.C481S 0.51 | ||||||
| BTK p.C481S 0.02 | ||||||
| 9/76/Woman | NA | NA | 22.4 | TP53 p.A86fs 0.82 | TP53 p.A86fs 0.21 | Died, 18 mo |
| TP53 p.R273P 0.02 | FBXW7 p.I257T 0.14 | |||||
| FBXW7 p.I257T 0.47 | FBXW7 p.K374N 0.17 | |||||
| FBXW7 p.K374N 0.46 | BTK p.V537I 0.02 | |||||
| 11/72/Woman | G | 2-5 | 39.2 | TP53 p.R110P 0.01 | TP53 p.R110P 0.05 | Alive, 6 mo |
| TP53 p.P177_C182del 0.12 | TP53 p.P177_C182del 0.20 | |||||
| DDX3X p.V321del 0.16 | DDX3X p.V321del 0.37 | |||||
| SF3B1 p.K700E 0.16 | SF3B1 p.K700E 0.35 | |||||
| BTK p.C481S 0.06 | ||||||
| BTK p.C481F 0.05 | ||||||
| 1/65/Woman | G | 4-39 | 40.4 | BIRC3 p.Q547fs 0.14 | TP53 p.F134L 0.46 | Alive, 8 mo |
| TP53 p.F134L 0.49 | POT1 p.G272C 0.46 | |||||
| POT1 p.G272C 0.45 | BTK p.C481Y 0.16 | |||||
| BTK p.C481F 0.07 | ||||||
| 7/58/Man | G | 3-48 | 27.5 | TP53 p.G245V 0.02 | TP53 p.G245V 0.02 | Died, 1 mo |
| TP53 p.Y220C 0.46 | TP53 p.Y220C 0.46 | |||||
| TP53 p.M237V 0.10 | TP53 p.M237V 0.08 | |||||
| TP53 p.G105R 0.03 | TP53 p.G105R 0.03 | |||||
| TP53 p.V157F 0.02 | ||||||
| BIRC3 p.Q547fs 0.04 | BIRC3 p.Q547fs 0.05 | |||||
| BIRC3 p.V568G 0.17 | BIRC3 p.V568G 0.17 | |||||
| 4/50/Woman | G | 4-31 | 14.7 | XPO1 p.E571K 0.19 | XPO1 p.E571K 0.05 | Alive, 32 mo |
| TP53 p.D281H 0.31 | ||||||
| NOTCH1 p.P2415del 0.29 | ||||||
| NOTCH1 p.P2514fs 0.19 | ||||||
| MUC2 p.P1523S 0.11 | ||||||
| TP53 p.C176S 0.02 | ||||||
| 5/68/Man | M | 3-30 | 13.6 | None | SF3B1 p.K700E 0.11 | Alive, 28 mo |
| POT1 p.G40E 0.07 | ||||||
| 8/57/Man | G | 3-23 | 53.9 | ATM p.P1526fs 0.32 | ATM p.P1526fs 0.93 | Alive, 26 mo |
| TP53 p.G226fs 0.05 | ||||||
| 2/68/Man | G | 1-69 | 30.7 | TP53 p.C277F 0.48 | TP53 p.C277F 0.62 | Alive, 1 mo |
| SPEN p.R1366Q 0.27 | SPEN p.R1366Q 0.36 | |||||
| FBXW7 p.R505C 33.30 | FBXW7 p.R505C 0.54 | |||||
| CARD11 p.N280_L283delinsKQEQ 0.02 | ||||||
| 3/60/Woman | G | 3-30 | 7.4 | SF3B1 p.N626Y 0.12 | SF3B1 p.N626Y 0.14 | Alive, 10 mo |
| SF3B1 p.K700E 0.35 | SF3B1 p.K700E 0.32 | |||||
| 10/80/Woman | G | 1-69 | 11.7 | TP53 p.P278T 0.44 | TP53 p.P278T 0.14 | Died, 2 mo |
| BIRC3 p.R434fs 0.02 | TP53 p.A74fs 0.28 | |||||
| BIRC3 p.F526fs 0.02 | SF3B1 p.K700E 0.01 |
- Abbreviations: G, germline, defined as ≤ 2% deviation from the germline immunoglobulin heavy-chain variable region sequence; IGHV, immunoglobulin heavy-chain variable region; M, mutated, defined as a >2% deviation from the germline IGHV sequence; RT, Richter transformation; VH, variable region of Ig heavy chain; NA, not available.
- At the time of disease progression/RT, mutations shown in bold with underline (in red) indicate “new onset” whereas mutations with underline (in blue) represent those that disappeared.
We compared the pattern of concurrent gene mutation changes between patients who acquired BTK mutations and those who did not. Four of 9 patients acquired mutations in BTK at the time of disease progression/RT. Among patients who acquired BTK mutations, a new mutation (in the CHD2 gene) was detected in only 1 patient (25%); 2 patients lost previously detected mutations (in TP53 and BIRC3) whereas no other changes were noted in the remaining patient. Seven patients did not acquire BTK mutations at the time of disease progression/RT. Of these, 5 patients (71%) gained additional mutation(s) in various genes that included TP53, SF3B1, POT1, MUC2, and CARD11 compared with their pre-BTKi mutation profile; one patient, who concurrently acquired mutations in TP53 (additional) and SF3B1, demonstrated loss of a previously detected BIRC3 mutation. No changes in the pre-BTKi and post-disease progression/RT mutation profile were noted in patients 7 and 3. Comparison of FISH studies demonstrated new trisomy 12 and ATM loss during transformation in only 1 patient (patient 10). Details regarding paired mutation analysis in 2 patients with RT are described further in the subsequent section.
Comparing the mutation profile between pre-BTKi and post–disease progression/showed that the most dynamic changes were noted in the TP53 gene: patients 4 and 8 gained new mutations at the time of disease progression and patient 10 also developed an additional TP53 mutation during histologic transformation along with an SF3B1 mutation, whereas patient 9 demonstrated disappearance of 1 of the 2 TP53 mutations detected previously. Patient 6 had multiple TP53 subclonal mutations before the initiation of therapy with BTKi. At the time of disease progression, the patient lost 3 previously detected TP53 mutations while acquiring 3 new mutations; gain of an additional mutation was noted when sequencing was performed 3 months after disease progression. Loss of BIRC3 mutations was noted in 2 patients at the time of disease progression/RT.
None of the 11 patients demonstrated mutations in BTK or PLCG2 at baseline prior to the initiation of BTKi therapy. In addition, baseline samples from 6 additional patients with CLL (3 of whom developed disease progression and 3 of whom developed RT; data not shown) were available, none of which demonstrated BTK or PLCG2 mutations prior to the initiation of BTKi therapy.
Mutational Landscape in Patients With CLL at the Time of Disease Progression or RT
Collectively, 29 patients had DNA available for mutation analysis at the time of disease progression (23 patients) or RT (6 patients) (Fig. 2) (see Supporting Table S2). All patients had at least 1 gene mutation at the time of disease progression or RT. The highest frequency of mutations was observed in BTK (noted in 19 of 29 patients [65.6%] at the time of disease progression). The frequencies of BTK mutations in patients who developed disease progression and underwent RT were 69.6% (16 of 23 patients) and 50% (3 of 6 patients), respectively. Both patients receiving acalabrutinib developed BTK mutations at the time of disease progression and RT, respectively. All but 1 patient had BTK mutations localized to the hotspot codon C481; the distribution of amino acid substitutions showed C481S (22 patients), C481F (7 patients), C481Y (4 patients), and C481R (2 patients). In one patient who developed disease progression while receiving ibrutinib, a BTK mutation was noted in codon 537 (V537I). Ten of 19 patients with BTK mutations had >1 mutation with different amino acid changes. This included a single patient with 5 different mutations at a low allelic burden (<5%). The median allelic frequency for BTK mutations was 11% (range, 1%-91%) when only the BTK mutation at the highest allelic burden per patient was considered. CARD11 mutations were observed in 2 patients. None of the patients in the study cohort demonstrated PLCG2 mutations.
A total of 23 patients underwent disease progression. Recurrent gene mutations at the time of disease progression in decreasing order of frequency included BTK (16 patients), TP53 (15 patients), SF3B1 (5 patients), NOTCH1 (4 patients), SPEN (3 patients), CARD11 (2 patients), FBXW7 (2 patients), ATM (2 patients), and POT1 (2 patients). Among patients with CLL who developed disease progression, the median time to CLL progression in patients with BTK mutations was 39.8 months (range, 14.2-70.3 months) whereas the median time to disease progression in patients who did not have BTK mutations was shorter at 22.6 months (range, 13.6-53.9 months) (P = .07).
Six patients underwent RT/diffuse large B-cell lymphoma (DLBCL), all involving the BM. In 2 patients, DLBCL also was diagnosed concurrently in a supraclavicular lymph node. At the time of RT arising within the setting of BTKi therapy, the highest frequency of mutations was noted in SF3B1 (4 patients) followed by TP53 (3 patients), BTK (3 patients), ATM (2 patients), and BIRC3 (2 patients). Three of 6 patients had BTK mutations (median variant allele frequency [VAF], 2.3% [range, 1.2%-7.2%]). In all 3 patients, BTK mutations were present at a low allelic burden compared with the tumor burden. We also quantified the large cell component by histologic evaluation. The large cell component in patient 30 represented approximately 5% to 10% of the overall tumor, confirmed by the presence of MYC rearrangement in approximately 5% of nuclei; a BTK mutation was noted at a mutant allelic frequency of 3%. It was not possible to determine whether a BTK mutation was present within the large cells. In patient 34, a BTK mutation was noted at an allelic burden of 1.2%. In patient 24, who was treated with acalabrutinib, the large cell component represented approximately 10% by histologic evaluation. At the time of RT, 2 different BTK mutations (C481S and C481F) were noted at 7.2% and 3.4%, respectively. In 2 patients who experienced RT (patients 10 and 3), paired pre-BTKi and posttransformation samples were available. In patient 10, SF3B1 and an additional TP53 mutation were acquired at the time of RT compared with the sample obtained before treatment with ibrutinib, whereas a BIRC3 mutation was lost. In patient 3, there was no change noted in the mutation profile. Representative images of RT are shown in Figure 3, and specific clinicopathologic details are provided in Table 3.
| Identifier/Age, Years/Sex | IGHV | Duration of BTKi, Months | BM Diagnosis | FCM Findings | Cell-of-Origin Classification for DLBCL | BCL2 | Ki-67 | MYC | Karyotype | FISH | NGS Results | Status, Time Since RT |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 30/64/Man | NA | 24.4 | CLL and focal DLBCL (MYC rearrangement, 5%) | CD5+/CD23+/kappa | ABC (CD10-/BCL6+/MUM1+) | ND | 70% | 5%-10% | 48,XY,+X,add(1)(p13),t(4;9)(q21;p22),der(12)t(1;12)(p22;q22),+mar[cp14]/46,XY[6] | D13S319 single (9%); D13S319 both (71%) | BTK p.C481S (<5%); ATM p.L2445R; ATM splice mutation; SF3B1 p.K700E | Alive, 17 mo |
| CLL | CD5+/CD23+/kappa | ND | <5% | ND | 46,XY[20] | D13S319 single (44%); D13S319 both (12%) | ND | |||||
| 34/69/Man | Mutated (2.7%, VH 2-5) | 30.8 | DLBCL | CD5+/CD23+/kappa | ABC (CD10-/BCL6-/MUM1-) | ND | ND | Approximately 20% | 45,X,-X,ins(5;?)(q22;?),del(8)(p21),+12,del(14)(q24),-17,der(19)t(17;19)(q21;p13.3)[15]/46,XX[5] | Trisomy 12 (41%); TP53 (39%) | BTK p.C481S (1.2%); TP53 p.R248G; TP53 p.R248Q; TP53 p.P278S; MUC2 p.P1510S; ZMYM3 p.G49fs; SPEN p.V1211fs; SPEN p.A2804fs | Alive, 12 mo |
| CLL | CD5+/CD23+/kappa | ND | <5% | ND | ND | ND | ND | |||||
| 24/66/Woman | Germline (VH 1-69) | 25.4 | DLBCL | CD5+/CD23+/kappa | ABC (CD10-/BCL6+/MUM1+) | ND | 60% | ND | 86~88,XXX,-X,del(3)(p13p25)x2,add(4)(q35)x2,-5,-6,del(6)(q13q23)x2,-8,-8,-8,add(9)(p22)x2,-10,del(11)(q22q23)x2,-12,-13,del(13)(q12q22)x2,-14,-15,-16,add(16)(q24)x2,-17,add(17)(p11.2)x2,-19,add(19)(p13.3)x2,-20,-21,+5~11mar[cp9]/46,XX[11] | Tetraploid cell population (86%) with deletions of 4 copies of D13S319 loci, 2 copies of ATM and TP53 | ATM p.R2032S (<5%); BTK p.C481S; BTK p.C481F; TP53 p.Y220C (<5%); BIRC3 p.R432fs (<5%) | Alive, 21 mo |
| CLL | CD5+/CD23+/kappa | ND | <5% | ND | 46,XX[20] | D13S319 single and both; ATM | ND | |||||
| 3/60/Woman | Germline (VH 3-30) | 7.4 | DLBCL | CD5+/CD23+/lambda | GCB (CD10-/BCL6+/MUM1-) | Positive | 40% | ND | 45,XX,der(13;17)(q10;q10),add(21)(p12)[6]/45,idem,i(15)(q10)[7]44~45,XX,der(13;17)(q10;q10),add(15)(p13)[cp2]/46,XX[7] | TP53 (7%) | SF3B1 p.K700E; SF3B1 p.N626Y | Alive, 17 mo |
| CLL | NA | ND | <5% | ND | 45,XX,der(13;17)(q10;q10),add(21)(p12)[8]/46,idem,+15[2]/46,XX[10] | D13S319 single (22.5%); TP53 (76.5%) | SF3B1 p.K700E; SF3B1 p.N626Y | |||||
| 10/80/Woman | Germline (VH 1-69) | 11.7 | DLBCL | CD5dim/CD23dim/lambda | ABC (CD10-/BCL6-/MUM1-) | ND | ND | Approximately 20% | 82~84,XXXX,add(1)(p21)x2,-2,del(4)(q21),del(4)(q25),-5,-6,-6,-8,-9, add(9)(p24),add(9)(q34),-10,-12,-13,-16,-17,-18,+3mar[cp2]/46,XX[18] | ATM (2 extra copies, 7.5%); chromosome 12 (2-3 extra copies, 10.5%); D13S319 and LAMP (extra copies) | TP53 p.P278T; TP53 p.A74fs; SF3B1 p.K700E | Died, 13 mo |
| CLL | CD5dim/CD23dim/lambda | ND | <5% | ND | 46,XX,del(3)(p13)[3]/46,XX[17] | D13S319 (84.5%) | TP53 p.P278T; BIRC3 p.R434fs; BIRC3 p.F526fs | |||||
| 35/69/Man | Mutated (5%, VH 1-69) | 30.8 | DLBCL | CD5+/CD23+/kappa | ABC (CD10-/BCL6+/MUM1+) | ND | 90% | ND | ND | BIRC3 p.C588R; SF3B1 p.G740E; NOTCH1 p.S2467*; NOTCH1 p.P2514fs | Alive, 0.4 mo | |
| CLL | CD5+/CD23+/kappa | ND | <5% | ND | ND | ATM (60%); D13S319 single (91%) | ND |
- Abbreviations: +, positive; -, negative; BM, bone marrow; BTKi, Bruton tyrosine kinase inhibitor; CLL, chronic lymphocytic leukemia; DLBCL, diffuse large B-cell lymphoma; FCM, flow cytometry; FISH, fluorescence in situ hybridization; IGHV, immunoglobulin heavy-chain variable region; LAMP, lysosomal associated membrane protein; NA, not available; ND, not done; NGS, next-generation sequencing; RT, Richter transformation; VH, variable region of Ig heavy chain.
Of note, we sequenced 10 patients who did not develop BTKi resistance while receiving BTKi therapy using the same NGS panel. The following mutations were noted: TP53 (5 patients; 50%); SF3B1 (3 patients; 30%); NOTCH1 (2 patients; 20%); BIRC3 (2 patients; 20%); and a mutation each in XPO1, FBXW7, CARD11, SPEN, POT1, and MUC2. None showed mutations in BTK or PLCG2. Further, TP53 sequencing data (from other NGS panels that included the entire coding region of TP53) were available for 27 patients. TP53 mutations were noted in 14 of these 27 patients (52%), which was not significantly different from the study group.
Clonal Evolution After Stopping BTKi Therapy
After discontinuation of BTKi, we performed sequencing at additional time points in 2 patients with RT and 3 patients with disease progression. The clonal evolution patterns in the patients with RT are illustrated in Figures 4A and 4B and elaborated below. Patient 24 had been treated multiple times before initiating treatment with acalabrutinib. The patient achieved an excellent response for 1 year followed by resistance. The BM samples demonstrated an increased number of large cells consistent with RT (approximately 10% large cells). At the time of RT, the tumor demonstrated mutations in TP53 (VAF, 73%), ATM (VAF, 7%), and BIRC3 (VAF, 6%), as well as 2 different BTK mutations (C481S [VAF, 7%] and C481F [VAF, 4%]) noted at 7.2% and 3.4%, respectively. FISH studies demonstrated deletions of ATM, D13S319, and TP53 genes. After the diagnosis of RT, treatment with acalabrutinib was stopped. Over the next several months, the patient showed persistent RT with large cells eventually replacing the entire BM. Sequencing was performed at multiple time points; the patient received treatment with idelalisib, obinutuzumab, and venetoclax followed by nivolumab over this time period. Gene mutations in ATM, BIRC3, TP53, and BTK persisted for 4 months after treatment when acalabrutinib was stopped (a BTK mutation was detectable at a 0.9% allelic burden at 4 months only by manual review of Integrated Genomic Viewer). At 16 months post RT, mutations in BTK, ATM, and BIRC3 were not detectable despite the persistence of RT whereas a TP53 mutation expanded. Repeat sequencing at 19 months post RT confirmed this finding (Fig. 4A). Similarly, patient 34, who also was treated multiple times prior to receiving ibrutinib, developed RT involving the BM; supraclavicular lymph nodes demonstrated CLL without transformation. Four months post RT and after stopping ibrutinib, the BTK mutation disappeared whereas selected TP53 mutant clones expanded. The patient had persistent RT that eventually responded to rituximab, etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin (R-EPOCH); nivolumab; and venetoclax followed by a matched unrelated donor stem cell transplantation (Fig. 4B).
We performed serial sequencing in 3 patients with CLL who developed disease progression (Figs. 4C-4E). All 3 patients received venetoclax immediately after discontinuation of BTKi therapy. In all 3 patients, BTK mutations decreased in VAF, indicating that venetoclax is capable of suppressing BTK mutant clones.
Discussion
Understanding the diverse mechanisms of resistance to BTK inhibitor therapy, in addition to BTK and PLCG2 mutations already reported in the literature, is essential for the rational design of therapies to treat these patients. However, to the best of our knowledge, there is limited knowledge regarding additional genetic mutations associated with BTK/PLCG2 mutations, and gene mutations in patients lacking BTK/PLCG2 mutations. In this single-institution study of 29 CLL patients with BTKi resistance, using targeted deep sequencing, we demonstrated a spectrum of gene mutations in 29 CLL-related and BCR signaling pathway genes. In 11 patients for whom baseline samples were available prior to BTKi treatment, we identified progression/RT–associated mutations in genes other than BTK, which were far more frequent in patients with CLL who did not develop BTK mutations. We sequenced 6 patients who developed RT and found that SF3B1 mutations were more frequent than BTK mutations, unlike in patients with progression. To our knowledge, the current study is the first to report BTK mutations in patients with CLL who were treated with acalabrutinib (a second-generation selective BTK inhibitor that is more potent than ibrutinib). Last, the results of the current study independently confirm that BTK mutations contribute to ibrutinib resistance in at least a subset of patients, consistent with previously reported findings.8, 9, 11, 14, 15, 24
The overall frequency of BTK mutations in the current study was 65.6%, which is consistent with the frequencies of 53.8% to 80.4% reported by 2 other large series.2, 14 Our clinically validated NGS assay is designed to simultaneously detect somatic mutations in 29 CLL-related genes that include mutational hotspots of BTK exons 8 (codons 197-259) and 15 to 16 (codons 450-544). Mutations outside of the kinase domain, such as T316, which is known to confer resistance, are rare.12 We found a novel V537I mutation that to our knowledge has not been reported previously. This mutation is located within the kinase domain of the BTK gene, close to C481, and potentially can affect the binding of the drug. Functional studies are needed to confirm the role of this mutation in BTKi resistance. The median coverage of 3000× for BTK allowed for the detection of mutations at an allelic burden of 1% to 5%. Consistent with other studies, multiple different BTK mutations were noted in the majority of patients, suggesting that the origin of these subclones is likely independent and induced by the effect of BTKi.14, 25 In one patient, we observed 5 different BTK mutations. One patient who underwent mutational analysis demonstrated the same BTK mutation 6 months prior to demonstrating clinical signs of disease progression. Another novel observation was the detection of BTK mutations in 2 patients with CLL who were treated with acalabrutinib and developed BTKi resistance (one with progressive disease and other with RT; details shown in Fig. 4A). Acalabrutinib is a highly selective BTKi, and has minimal off-target activity.
In the current study, none of the patients had PLCG2 mutations. Reported frequencies of PLCG2 mutations in patients ranged between 19.6%2 and 53.8%,14 with the variability related to the differences in the coverage of the mutational hotspot and the level of the sensitivity of the assays. A high percentage of PLCG2 mutations was observed in patients with BTK mutations.2, 14, 24 The reasons for the discrepancy between the findings of the current study and others could be multifactorial. In contrast to most reported studies,2, 8, 9, 12, 24 we did not perform B-cell sorting and tumor enrichment prior to mutation analysis. Furthermore, our assay covered only exon 19 (codons 645-685) of PLCG2 (median coverage: ×1800), and can miss less frequent mutations such as P664, R665, S707, L845F, D993, D1140, and M1141 and deletions of S707 and A708.2, 8, 9, 11, 14, 15, 18, 24 In addition to features inherent to the assay design, the current study cohort included a mix of patients who developed disease progression as well as those whose disease transformed while they were receiving treatment with a BTKi. Studies have shown that acquired BTK/PLCG2 resistance mutations are more frequent in patients with CLL who are undergoing disease progression compared with patients who experience RT.8, 14, 15, 25
None of the patients in the current study cohort demonstrated mutations in BTK or PLCG2 at baseline prior to the initiation of BTKi therapy. In fact, from a total database of 652 patients (including approximately 400 treatment naive patients) who underwent mutational profiling using this panel, mutations in BTK or PLCG2 were never observed.
The BTK C481 mutation directly disrupts the binding site of BTKi, leading to activation of BCR signaling and cellular proliferation. A high association between BTK mutations and CLL progression suggests a significant role in disease progression. However, a subset of patients who develop disease progression have low-level BTK mutant clones. In such cases, it is possible that BTK mutations may cooperate with additional genetic events that drive disease progression. Furthermore, there is a subset of patients with CLL who progress without acquiring either BTK or PLCG2 mutations. These findings suggest an important role for other genomic alterations or that BTK mutations cooperate with other genetic events.25 To date, most of the reported studies in the literature have assessed for mutations in BTK and PLCG2 mutations alone using deep sequencing. Few studies have assessed concurrent mutations in other genes over the course of BTKi therapy by either targeted NGS or whole-exome analysis; < 25 patients have been reported in total.9, 12, 15, 19, 25 After reviewing the reported findings in the literature, in composite, mutations in BTK,9, 12, 15, 25 PLCG2, MLL2/KMT2D,2, 15 EP300,15 RAS,2, 15, 25 PLCO,25 TP53, and del(8p) alteration have been reported in at least 2 patients (recurrent disease); nonrecurrent but progression–associated mutations also have been reported in EIF2A, ribosomal protein S15 (RPS15), XPO1, FBXW7, NPM1, MUC2, and IRF4 genes.
In the current study, to identify mutations associated with disease progression/RT, we evaluated the genomic mutation profiles of paired tumor samples in 11 patients before BTKi and at the time of disease progression. At disease progression, 4 of 11 patients developed BTK mutations. Patients who acquired BTK mutations rarely demonstrated additional gene mutations; only 1 of 4 patients (25%) demonstrated a gain of a CHD2 mutation. This suggests that the BTK mutation by itself, at least when present at a significant tumor burden, can cause disease resistance and relapse and does not need additional cooperating drivers, which is consistent with reported findings from cell lines.9, 11 In contrast, 5 of 7 patients (71%) who did not acquire BTK mutations demonstrated new mutations in other genes at the time of disease progression/RT (TP53, SF3B1, NOTCH1, POT1, and CARD11). More important, the median time to disease progression in patients who acquired BTK mutations tended to be longer compared with BTKi-resistant patients who did not acquire BTK mutations (P = .07). This suggests that BTK mutations develop as a result of long-term exposure to the drug. However, in the interim, some patients acquire additional gene mutations (characteristically associated with adverse prognosis) that abolish response to BTKi, akin to RT, and demonstrate an even shorter time to transformation. These data further support alternate mechanisms of drug resistance other than BTK mutations in a subset of patients.26 However, the number of patients in the current study is small to draw definitive conclusions.
It is interesting to note that patients with CLL who gained BTK mutations also demonstrated loss of preexisting mutations, including loss of a TP53 mutation in 1 patient and notably the loss of BIRC3 mutations in 2 patients. Another study also demonstrated loss of pretreatment del(17q)/TP53–mutant clones in 2 patients and loss of trisomy 12 in 1 patient, all of whom acquired BTK mutations.25 This finding indicates that BTKi therapy is capable of eliminating these clones. In 2 patients (patients 7 and 3), we did not identify any changes in the gene mutation or copy number profiles by FISH between samples obtained before BTKi therapy and after disease progression/RT. Patient 3 also had concurrent RT in a lymph node. Hence, it would be of interest to sequence the lymph node to identify any potential effect of the microenvironment.
Clonal mutation evolution, especially in genes that confer aggressive outcomes such as SF3B1, NOTCH1, and TP53, occurring while patients are undergoing therapy with BTKi and its relevance in the pathobiology of disease progression remains to be explored. Sequential sampling to determine the pattern of mutation evolution from a larger cohort of patients receiving BTKi (alone or in combination) over a longer follow-up will help us to understand the role of these mutations as well as provide a basis for the addition of novel agents in combination with BTKi therapy to prevent the development of disease resistance. For example, the results of a recent study demonstrated that NOTCH1 mutations in combination with a low BAX/BCL-2 ratio conferred poor overall survival and progression-free survival in patients with CLL who are treated with ibrutinib and support the use of the BCL-2 inhibitor venetoclax with ibrutinib for a better outcome.27 In the current study, in a subset of patients who acquired BTK mutations and drug resistance, we demonstrated that treatment with venetoclax led to the suppression of BTK clones. This finding indicates that venetoclax and other novel agents will form a platform for future clinical trials.
Greater than one-third of patients with CLL that is resistant to BTKi therapy develop RT, most often DLBCL, and these patients have a dismal outcome.8, 13, 16 Overall, there is a shorter interval to RT in patients who are treated with BTKi therapy, suggesting a different genetic evolution pathway compared with disease progression,2, 8, 14 as was noted herein. In the current study, SF3B1 gene mutations were noted at a higher frequency in patients with RT (66% vs 22%; P = .06) compared with patients with CLL progression. In 1 of 2 patients, an SF3B1 mutation was acquired at the time of RT. BTK mutations were less frequent, noted in 3 of 6 patients with RT, all at a low allelic burden (VAF, < 10%). In 2 patients, sequencing at a subsequent time point after stopping BTKi therapy demonstrated the disappearance of BTK mutations, despite persistent RT (Figs. 4A and 4B). This suggests that BTK mutations may not play a significant role in the development of RT, which needs other genomic or epigenetic alterations compared with patients with CLL with progressive disease.8, 12, 25, 28 To the best of our knowledge, understanding of the genomic characterization of RT arising within the setting of BTKi therapy is limited, unlike RT in other settings, which has been relatively well studied.29-31 Approximately 80% of the large B-cell lymphomas arising from patients with CLL are clonally related.32 Our RT series lacks the V-D-J rearrangement data to determine the clonal relationship with underlying CLL. To overcome this, we used immunoglobulin light chain expression and FISH abnormalities in the paired samples to suggest clonal relationship. All 5 patients with available flow cytometry results demonstrated the same clonal light chain in paired CLL and RT samples. In all 4 patients with available paired FISH data, abnormalities noted during the CLL phase also were found to be present at the time of RT with or without additional aberrations (Table 3).
The mutation profile in patients with CLL is dynamic and complex and therefore it is of immense interest to understand the status of BTK mutations and mutational evolution patterns after discontinuation of BTKi therapy. Patients with disease progression often are treated with the Bcl-2 inhibitor venetoclax.33 In the current study, we examined the mutation evolution in 3 patients with CLL with disease progression after discontinuation of BTKi therapy and treatment with venetoclax. Although limited by the small number, the current study data demonstrate that venetoclax is able to suppress BTK mutant clones in patients who experience disease progression. Further studies are needed to support this observation and currently are underway.
A limitation of the current study is the lack of comprehensive copy number assessment. Using standard FISH studies, we noted that patients with CLL who developed disease recurrence had a significantly higher frequency of del(17q) compared with those patients who did not develop disease recurrence while receiving BTKi therapy over the same time period, confirming the observations of another study.34 Using standard targeted FISH studies, the majority of patients in the current study did not demonstrate changes prior to the initiation of BTKi therapy and at the time of disease progression. However, Burger et al reported recurrent del(8p) alterations leading to haploinsufficiency of TNF-related apoptosis-inducing ligand receptor (TRAIL-R) in patients with CLL with progressive disease who did not acquire BTK mutations.15 Hence, a comprehensive whole-genome copy number assessment using single-nucleotide polymorphism/array-comparative genomic hybridisation (array-CGH) will be helpful. Furthermore, we have not analyzed other genes such as EIF2A, RPS15, EP300, and MLL2/KMT2D that have been suggested in some of the patients in the current study.
In summary, the results of the current single-institution study add to the evidence that BTK mutations are associated with resistance to BTKi therapy (ibrutinib and acalabrutinib) in a significant subset of patients. Using targeted deep sequencing with a multigene panel in 11 patients with available baseline samples prior to BTKi therapy, we identified additional progression-specific gene mutations that are more frequent in patients with CLL who do not develop BTK mutations. Unlike patients with CLL with disease progression, SF3B1 mutations were found to be more frequent than BTK mutations in patients with CLL who developed RT. Clinical grade targeted deep sequencing assays are useful for monitoring alterations in multiple genes and clonal evolution in patients receiving BTKi therapy.
Funding Support
Supported by startup funds awarded to Rashmi Kanagal-Shamanna by the institution.
Conflict of Interest Disclosures
Nitin Jain has received research funding from Pharmacyclics, AbbVie, Genentech, Bristol-Myers Squibb, Pfizer, ADC Therapeutics, Incyte, Celgene, AstraZeneca, Servier, Verastem Inc, Cellectis, and Adaptive Biotechnologies and has acted as a member of the advisory boards of and received honoraria from Pharmacyclics, AbbVie, Pfizer, ADC Therapeutics, AstraZeneca, Servier, Verastem Inc, Novartis, Novimmune, Janssen, and Adaptive Biotechnologies for work performed outside of the current study. Hagop M. Kantarjian has received grants from AbbVie, Agios, Amgen, Ariad, Astex, Bristol-Myers Squibb, Cyclacel, Immunogen, Jazz Pharmaceuticals, and Pfizer; has received honoraria from AbbVie, Agios, Amgen, Immunogen, Orsinex, Pfizer, and Takeda; and has acted as a member of the advisory board for Actinium for work performed outside of the current study.
Author Contributions
Rashmi Kanagal-Shamanna, Preetesh Jain, and L. Jeffrey Medeiros designed the study. Rashmi Kanagal-Shamanna and L. Jeffrey Medeiros collected the results, analyzed the data and wrote the article. Rashmi Kanagal-Shamanna analyzed the molecular data. Rashmi Kanagal-Shamanna and Preetesh Jain collected clinical information. Keyur P. Patel, Mark Routbort, Carlos Bueso-Ramos, Tahani Alhalouli, Joseph D. Khoury, and Rajyalakshmi Luthra reviewed the molecular data. Alessandra Ferrajoli, Michael Keating, Nitin Jain, Jan Burger, Zeev Estrov, and William Wierda treated the patients and contributed patient samples. All authors reviewed the article and provided the final approval.
