Genomic Evolution of Ibrutinib Resistant Clones in Waldenström Macroglobulinemia

Summary

Ibrutinib is highly active in Waldenström macroglobulinemia (WM) patients, but disease progression can occur due to acquired mutations in BTK, the target of ibrutinib, or PLCG2, the protein downstream BTK. However, not all resistant patients harbour these alterations. We have performed a whole exome sequencing study to identify alternative molecular mechanisms that can drive ibrutinib resistance. Our findings include deletions on chromosomes 6q, including homozygous deletions, and 8p, which encompass key regulators of BTK, MYD88/NF-κB, and apoptotic signalling. Moreover, we have identified recurring mutations in ubiquitin ligases, innate immune signalling, and TLR/MYD88 pathway regulators in ibrutinib resistant WM patients.

Somatic mutations in MYD88 occur in 95–97% of patients with Waldenström Macroglobulinemia (WM) (Treon et al, 2012; Jiménez et al, 2013). Secondary events involving heterozygous loss of chromosome 6q21–25 (del6q) or C-terminal domain mutations in CXCR4 are common in WM, accounting for 40–50% and 30–40% of MYD88-mutated patients, respectively (Xu et al, 2016; Guerrera et al, 2018). MYD88 mutations are activating, and trigger constitutive NF-κB activation through BTK and IRAK1/IRAK4 (Yang et al, 2013). BTK inhibitors including ibrutinib, acalabrutinib and zanubrutinib are highly active in MYD88 mutated but not MYD88^wild-type WM patients (Argyropoulos & Palomba, 2018). CXCR4 mutations are associated with a lower rate of major responses and shorter progression free survival (Treon et al, 2018). Acquired mutations in BTK at the binding site of ibrutinib (BTK^Cys481), or its downstream mediator PLCG2, have been identified in half of WM patients progressing on ibrutinib (Xu et al, 2017) leading to enhanced growth and survival through the re-activation of MAPK3 and MAPK1 (ERK1/2) signalling (Chen et al, 2018). Re-activation of ERK1/2 triggers cytokine release that can confer ibrutinib resistance to neighbouring BTK^wild-type cells. Not all WM patients progressing on ibrutinib harbour these alterations. We therefore performed whole-exome sequencing (WES) to study the genomics evolution of ibrutinib resistant WM.

Five previously treated WM patients who progressed while on ibrutinib were included in this study. Clinical characteristics of this cohort are presented in Table I. Ibrutinib was administered at 420 mg/day until disease progression. Median time to progression was 23 months (range 9.1–37.1). Bone marrow (BM) samples were collected at baseline and at the time of clinical progression for three patients. For two patients, only samples at time of progression were available. CD19-selected WM cells were obtained from BM samples using magnetic-activated cell sorting (Miltenyi-Biotec, Auburn CA). CD19-depleted peripheral blood cells from all 5 patients were used as germline controls. The study was approved by the Dana-Farber/Harvard Cancer Centre Institutional Review Board, and all samples were collected following informed consent.

Table I.

Clinical and biological characteristics of WM patients progressing on ibrutinib.

	IBR1		IBR2		IBR3		IBR4		IBR5
characteristics	Baseline	Progression	Baseline	Progression	Baseline	Progression	Baseline	Progression	Baseline	Progression
Age (years)	77	-	61	-	66	-	78	-	73	-
Gender	Male	-	Male	-	Male	-	Male	-	Male	-
Serum IgM (mg/dl)	4130	NA	3630	NA	2490	NA	1300	NA	5790	3970
Serum IgM M-spike (g/dl)	2.21	1.12	2.09	NA	1.41	NA	0.98	0.81	3.62	NA
Hemoglobin (g/dl)	10.3	10.5	12.3	10.7	9.1	7.8	8.9	8.4	10.8	7.7
Platelets (x109/l)	230		83		265		133		126
Serum β2-microglobulin (mg/l)	4.5		2.5		5.6		14.2		3.9
Bone marrow involvement (%)	30	70	70	90	40		5	60	30	80
Prior therapies	Fludarabine, rituximab, CPR, bendamustine- R, bortezomib/dex, tositumomab	-	Rituximab, ofatumumab, C, pentostatin	-	Rituximab	-	Cladribine, C, rituximab, IFNalpha, bendamustine, bortezomib/dex	-	Rituximab, chlorambucil, bendamustine	-
Time on ibrutinib (months)	9.6	-	37.1	-	36.4	-	9.1	-	23.0	-
Best response to ibrutinib	PR	-	PR	-	VGPR	-	VGPR	-	PR	-
Other events supporting progression						Splenic enlargement New pleural effusion
MYD88 status (% alelle frequency)	L265P (36%)	(47%)	L265P (94%)	(100%)	S243N (85%)	(79%)	NA	L265P (29%)	NA	L265P (83%)
CXCR4 status (% alelle frequency)	Wild-type	-	S338X (62%)	(41%)	D185Y (7%)	Wild-type	NA	Wild-type	NA	S338fs (37%)

NA, not available; C, cyclophosphamide; P, prednisone; R, rituximab; dex, dexamethasone; IFN, interferon; PR, partial response; VGPR, very good partial response; fs, frameshift mutation present at this amino acid site.

WES was performed using the Haloplex Human exome capture kit (Agilent Technologies, Santa Clara, CA) and 150 bp paired-end sequencing. Data was analysed following the GATK Best Practice Guidelines (Broad Institute, Cambridge, MA). Reads were aligned to GRCh37/HG19 using BWA. Small variants were analysed using Strelka (Illumina, Inc., San Diego, CA) and MutSigCV was used to identify significantly mutated genes (Benjamini-Hochberg q-value≤0.1). Variants found in the Exome Aggregation Consortium (ExAC) database with allele frequency >0.01% and variants with scaled phred Combined Annotation Dependent Depletion (CADD) score <20 were removed from the analysis. Copy number alterations (CNA) were called using Control-FREEC (BoevaLab, Institut Curie, Paris). Kolmogorov-Smirnov and Wilcoxon rank-sum tests were applied to identify significant CNA (p-values<0.01) and all results were adjusted for multiple hypothesis testing using the Benjamini-Hochberg procedure. Complete lists of CNA and somatic small variants are available in Supplemental Tables I and II, respectively. Validation of CNA in chromosome 8p was performed in tumour and germline samples with the TaqMan RT-PCR assays for copy number: Hs00497516_cn (DOK2) and Hs00132843_cn (TNFRSF10B), and expression: Hs00929587_m1 (DOK2) and Hs00366278_m1 (TNFRSF10B) (Thermo Fisher Scientific, Waltham MA).

CNA analysis identified del6q in all 5 patients, including homozygous deletions in three of them (IBR2, IBR3 and IBR4) at relapse. In IBR2, a heterozygous del6q at baseline evolved to homozygous loss at the time of ibrutinib progression. IBR3 demonstrated a subclonal homozygous deletion in a third of the tumour population at baseline that increased at the time of disease progression (Fig 1A and Fig S1). While del6q is present in up to half of WM patients at diagnosis, it is usually heterozygous. The progressive expansion of tumour clones with homozygous del6q in patients developing ibrutinib resistance is of great interest since many important regulators of signalling impacted by ibrutinib reside within this domain. Among these, the inhibitor of BTK (IBTK); inhibitors of MYD88/NF-κB (TNFAIP3, HIVEP2, TRAF3IP2, IRAK1BP1) signalling; as well as important regulators of apoptosis (FOXO3, BCLAF1, PERP) (Guerrera et al, 2018). We also observed del8p in 4/5 (80%) patients at ibrutinib progression with the remaining patient (IBR1) having a microdeletion as well. Del8p was present at baseline in 2 patients (IBR1 and IBR2), and absent in the other (IBR3). Genes lost in del8p patients included BLK, DOK2, and TNFRSF10A/B. BLK plays an important role in B-cell proliferation and differentiation and is a target of ibrutinib (Berglöf et al, 2015). DOK2 is a negative regulator of TLR/MYD88 signalling, and loss of DOK2 triggers ERK1/2 signalling in LPS stimulated cells (Shinohara et al, 2005), similar to the observed increased in ERK1/2 signalling in BTK^Cys481 mutated WM and ABC DLBCL cells (Chen et al, 2018). In the patient with the del8p microdeletion (IBR1), TNFRSF10A/B but not DOK2 or BLK were impacted. While the size of the deletion was unchanged at progression, this patient did acquire a truncating mutation in DOK2. Del8p was also observed in 3/5 CLL patients who progressed on ibrutinib and underwent WES (Burger et al, 2016). Abrogation of apoptotic signalling related to loss of TRAIL receptor (i.e., TNFRSF10A and TNFRSF10B) was suspected to contribute to ibrutinib resistance. We therefore decided to validate our findings and analyse the expression of two of these genes (DOK2 and TNFRSFB) in patients progressing versus patients responding to ibrutinib therapy (VGPR). Although the high frequency of del8p was confirmed in 3/5 patients for DOK2 and 4/5 for TNFRSFB (Fig S2), no differences were observed regarding expression of these genes between patients progressing compared to patients with continued response (Fig S3). Additional studies into the functional significance of this event in a larger population are warranted. Larger studies may also clarify whether the presence of these alterations at baseline have an impact on the depth of response to ibrutinib, similarly to what was observed for MYD88 and CXCR4 mutations.

Fig 1. Copy number and genomic alterations identified by whole exome sequencing in Waldenström Macroglobulinemia patients progressing on ibrutinib.

(A) Depicts copy number alterations in WM patients progressing on ibrutinib. Each square represents a chromosome. Losses are coloured in blue, gains in red, and green corresponds to normal copy number. Deletion of chromosome 6q was present in 100% patients at baseline and progression, including homozygous loss in 33% cases at baseline and in 60% at the time of ibrutinib progression. In IBR3 the homozygous deletion was inferred to be subclonal, affecting to a third of the tumour population at baseline and increasing at progression. Del8p was observed in 2/3 cases (66%) at baseline and in 4/5 cases (80%) at progression, and includes BLK, and DOK2. The patient without del8p (IBR1) harboured a truncating mutation in DOK2. Finally, 3q amplification was absent at baseline, but present in 40% patients at the time of progression. (B) Depicts genomic alterations in WM patients progressing on ibrutinib. Patients (represented with lines) are displayed by time to progression (TTP). For IBR1, IBR2 and IBR3, the proportion of persistent mutations (horizontal bars) is shown with respect to the mutations present at only one of the time points (vertical bars). On the right, the most remarkable alterations are shown for each patient: first, recurrent copy number variations and then, somatic mutations. Patients with mutated BTK^Cys481 also harboured other mutations which affect the B-cell receptor pathway. Other recurrently mutated genes were also identified (i.e. ITCH, RNF19B and TOLLIP), as well as copy number alterations, such as 3q amplification with loss of heterozygosity, with an embedded uniparental disomy that included TBL1XR1 in the two patients lacking BTK mutations.

In two patients (IBR1 and IBR3), we also found chromosome 3q amplifications that encompassed a copy number neutral loss of heterozygosity region of approximately 7 million base pairs that included TBL1XR1, a transcriptional regulator of NF-κB and WNT signalling (Jung et al, 2017). Mutations of TBL1XR1 are the most frequent variants found in MYD88^wild-type WM patients (Hunter et al, 2018), who show poor responses to ibrutinib monotherapy (Treon et al, 2018), though no somatic mutations in TBL1XR1 were identified in this study. No other recurrent CNA were detected, though two patients (IBR4 and IBR5) displayed multiple deletions or gains involving large chromosomal regions (Fig 1A and Fig S1).

Samples taken at ibrutinib progression showed a high proportion (median 87%; range 81%–89%) of acquired versus persistent mutations (median 13%, range 11%–19%). Most persistent mutations (60%) maintained the same allele frequency from baseline and did not enrich with the acquired mutations, suggesting that a high level of clonal heterogeneity persisted from baseline despite long exposure to ibrutinib therapy. Acquired mutations were more subclonal than persistent alterations (median allele frequency of 8.4% versus 13.9%, p=0.08). Among variants acquired at progression, BTK mutations were identified in 3 of 5 patients (IBR2, IBR4, and IBR5). Two of them (IBR2, IBR5) had BTK^Cys481Ser mutations at the ibrutinib binding domain, while the other patient (IBR4) had a BTK^Thr62Ala mutation. Both BTK^Cys481 mutated patients also harboured alterations in genes related to B-cell receptor signalling, such as PLCG2^Tyr495His, LYN^Ala2Stop and LYN^Ala139Thr, and CD79B^Asp33Tyr, and were CXCR4-mutated.

In patients without BTK mutations, we searched for other alterations that could contribute to ibrutinib resistance. For this, we focused on variants likely to be deleterious based on the CADD score, present only at progression, and affecting genes expressed in WM. We identified several candidates including ITCH^Ala646Ser (n=2), an ubiquitin ligase whose substrates are CXCR4, LYN or SYK; RNF19B^Arg30Gly (n=2), another ubiquitin ligase involved in STAT1-mediated transcriptional activity; FCRL3^Glu694Gln (n=1), a protein that modulates innate immune signalling in B-cells; BIRC2^Ala506Glu (n=1), a regulator of alternative NF-κB, and MAPK signalling; and negative regulators of TLR signalling including TOLLIP, seen in two patients as TOLLIP^Met242Arg and TOLLIP^Arg228His, and DOK2^Tyr345Stop in one patient (IBR1) (Fig 1B). In another patient (IBR3), a truncating SYK^Tyr526Stop at baseline was not detectable at time of progression on ibrutinib. Since SYK^Tyr526 is a critical phosphorylation site required for BCR activation and substrate recruitment, this clonal evolution could signify enrichment of clones capable of functional BCR/SYK signalling (Bohnenberger et al, 2011).

These studies provide novel insights into clonal evolution of ibrutinib resistant WM. Ibrutinib resistant WM is still an uncommon event. This limited the number eligible study patients and precluded definitive analysis. While resistance in some samples can be explained by acquired alterations in BTK and PLCG2 (Xu et al, 2017), identification of alternative mechanisms of resistance for the rest of the population is crucial as the number of WM patients on ibrutinib for multiple years continues to grow. Key findings include the recognition that del6q and del8p may accompany ibrutinib resistance, a notable finding since these regions encompass many key regulators of BTK, MYD88/NF-κB, and apoptotic signalling. Moreover, these studies also identified recurring mutations in ubiquitin ligases, innate immune signalling, and TLR/MYD88 pathway regulators in ibrutinib resistant WM patients.