Impaired nodal shrinkage and apoptosis define the independent adverse outcome of NOTCH1 mutated patients under ibrutinib therapy in chronic lymphocytic leukaemia

Abstract

The introduction of agents inhibiting the BCR-associated kinases such as ibrutinib has dramatically changed treatments algorithms of chronic lymphocytic leukaemia (CLL) as well as the role of different adverse prognosticators. We evaluated the efficacy of ibrutinib as single agent, in a real-life context, on 180 patients with CLL mostly pre-treated, recruited from three independent cohorts from Italy. Patients received 420 mg oral ibrutinib once daily until progression or occurrence of unacceptable side effects. Seventy-three patients discontinued ibrutinib for progression or for adverse events. NOTCH1 mutations (M) were correlated with a reduced redistribution lymphocytosis, calculated at 3 months on ibrutinib (p=0.022). Moreover, NOTCH1 mutated patients showed inferior nodal response at 6 months on ibrutinib compared to NOTCH1 wild type patients (p<0.0001). Significant shorter progression free survival (PFS) and overall survival (OS) were observed in NOTCH1 mutated patients (p=0.00002 and p=0.001). Interestingly, NOTCH1 M plus lower bax/bcl-2 ratio identified a CLL subset showing the worst PFS and OS (p=0.0002 and p=0.005). In multivariate analysis of PFS and OS, NOTCH1 M were confirmed an independent prognosticator (p=0.00006 and p=0.0039). In conclusion, NOTCH1 M are strongly associated with lower bax/bcl-2 ratio, consistent with a defective apoptosis, lower redistribution lymphocytosis and lower nodal shrinkage under ibrutinib treatment, this last responsible for partial responses, subsequent relapses, shorter PFS and OS. The therapeutic options for NOTCH1 mutated patients could be represented by either new small molecules combination approaches or from antibodies targeting NOTCH1.

Introduction

Chronic lymphocytic leukemia (CLL) is the most frequent adult leukemia in Western countries and it is characterized by an extremely heterogeneous clinical course. New molecular aberrations with negative prognostic value in CLL, such as NOTCH1, MYD88, TP53 and SF3B1 gene mutations, were identified in the last decade mainly thanks to the advent of next-generation sequencing (NGS).^1,2 In particular NOTCH1 mutations (M) are found in 10-14% of patients at diagnosis with frequency increasing with disease progression and during transformation to Richter syndrome.³ Furthermore NOTCH1 M are associated with the presence of trisomy 12 and with high CD49d expression which are negative prognostic factors in CLL. NOTCH1 M are also associated with an increased activation of the NF-kB pathway, promoting tumour cell proliferation and survival.⁴ NOTCH1 M were shown to affect the response to chemo-immunotherapy in CLL. In the CLL8 study Stilgenbauer et al.⁵ demonstrated that patients carrying NOTCH1 M did not benefit of the addition of rituximab to standard fludarabine and cyclophosphamide chemotherapy. Moreover, it has emerged that NOTCH1 M are associated with decreased duration of response in a large series of relapsed/refractory (R/R) patients treated with venetoclax.⁶

In a recent study, Tissino et al.⁷ have demonstrated that patients with CLL whose cells were characterized by high CD49d expression, underwent reduced lymphocytosis and inferior nodal response after treatment with ibrutinib. Several reports confirmed that in CLL the balance between the pro- and anti-apoptotic members of the BCL-2 family determines chemotherapy sensitivity and cell survival.^8,9 Noteworthy, we demonstrated that a low BAX/BCL-2 ratio had an additive negative prognostic impact in both TP53 M and NOTCH1 M patients with CLL treated with chemoimmunotherapy. ¹⁰ The recent introduction of novel B-cell receptor inhibitors such as ibrutinib and idelalisib and of novel potent oral BH3 peptidomimetics such as venetoclax in clinical practice, prompted us to evaluate the clinical impact of both NOTCH1 M and BAX/BCL-2 ratio in patients treated with targeted oral therapies and in particular in those treated with ibrutinib.

The aims of this study were: i) to analyse the correlations between NOTCH1 M and other biological parameters including CD49d expression and the BAX/BCL-2 ratio; ii) to address the impact of NOTCH1 M both on redistribution lymphocytosis and on nodal responses after treatment with ibrutinib; iii) to evaluate the impact of NOTCH1 M and BAX/BCL-2 ratio on the overall response rate (ORR) to ibrutinib, progression free survival (PFS) and overall survival (OS); iiii) to assess whether NOTCH1 M could be considered an independent prognostic factor.

Methods

Study design and patients

In this study we retrospectively analysed 180 patients with CLL exposed to treatment with ibrutinib. Patients were recruited from three independent cohorts from Italy (Rome Tor Vergata University, Rome Cattolica Sacro Cuore University and Catania Ferrarotto Hospital), between 2014 and 2019. Informed consent was obtained in accordance with the Declaration of Helsinki. The study was performed under the Institutional Review Board of the Centro di Riferimento Oncologico (IRCSS) of Aviano (approval numbers: IRB-05-2010 and IRB-05-2015).

Patients were 122 males and 58 females with a median age of 69 years (range, 36-90). According the modified Rai staging system, ¹¹ 134 patients had an intermediate risk and 46 a high risk stage. All these parameters were considered at the time ibrutinib was initiated.

All patients received 420 mg oral ibrutinib (Imbruvica; Janssen, Beerse, Belgium) once daily until progression or occurrence of unacceptable side effects. Median number of previous chemotherapy regimens were two (range, 0-4). Patients receiving first-line ibrutinib were 26 (14.4%), of whom 24 of 26 (92%) were TP53 mutated. Median follow-up was 25 months (range, 10-61). Seventy-three patients (40.6%) discontinued ibrutinib either for progression (n=42) or for adverse events (n=31) (Table 1): 32 patients were subsequently treated with venetoclax (11 for toxicity [grade 3 or 4 World Health Organization] and 21 for progression of disease), five patients were treated with idelalisib, and the remaining 36 patients received other lines of chemotherapy (n= 12) or no therapy (n=24). The clinical characteristics of patients are reported in Table 1. The clinical assessment of patients with CLL to establish diagnosis and response to therapy were based both on the International Workshop on Chronic Lymphocytic Leukemia (iwCLL) criteria.¹² The clinical impact of NOTCH1 M and BAX/BCL-2 ratio on ibrutinib treatment was evaluated by measuring the kinetics of absolute lymphocyte counts (ALC), the reduction of lymphadenopathy, and the clinical outcome, as defined by ORR, PFS and OS.

Chronic lymphocytic leukemia characterization

Flow cytometry was employed for immunophenotypical CLL characterization and was performed with FACSCalibur or FACSCanto I flow cytometer. BCL-2 and BAX oncoproteins were analysed by flow cytometry in samples taken before starting ibrutinib. BAX/BCL-2 ratio was calculated dividing mean fluorescence intensity (MFI) of BAX by MFI of BCL-2 on CLL cells, as previously described.¹⁰ The threshold of positivity was set at ≥1.5. Immunoglobulin heavy-chain variable region gene (IGHV) mutational status was performed by NGS, as previously described.^13,14 TP53 exons 2 to 11 mutational status and NOTCH1 exon 34 and 3’ untranslated (UTR) region mutational status were analysed by NGS, as previously described.^4,7 CLL samples were considered mutated for NOTCH1 i.e., NOTCH1 M, if exceeding a variant allele frequency (VAF) of 1%.^15,16

Table 1.Patient characteristics (n = 180)

Redistribution lymphocytosis and nodal response

The redistribution lymphocytosis was calculated as percent variation of ALC over the baseline values. Nodal response was calculated as percent reduction in sum of the product of diameter (SPD) values on the major lymph node regions over the baseline measurement, as reported previously.¹⁷

Additional details on the employed procedures and methods are reported in the Online Supplementary Materials and Methods.

Results

NOTCH1 mutations and BAX/BCL-2: correlations with other biological parameters

Sixty-five patients were NOTCH1 M (65 of 180, 36.11%), with VAF levels >1 (Online Supplementary Table 2S). With regard to the distribution of VAF levels, 21 patients had VAF between 1% and 10%, seven patients between 10.5% and 20% and 37 patients above 20%. Fifty-six NOTCH1 M cases bore a single mutation, eight cases two mutations and one case three mutations. NOTCH1 M cases were classified as follows: 45 delCT, six frameshift other than delCT (FS), 8 3’-UTR and six considering both missense (one) and nonsense (five) mutations (Online Supplementary Table S2). Seventy-four patients showed a BAX/BCL-2 ratio lower than 1.5 (74 of 113, 65.5%). NOTCH1 M were significantly associated with SPD ratio<1.5: in fact, 34 of 38 NOTCH1 M patients showed BAX/BCL-2 ratio less than 1.5 (P=0.0001). Moreover, NOTCH1 M were strongly correlated with CD49d expression: 51 patients were both NOTCH1 M and CD49d ≥30% (P=0.0001). Furthermore, a significant correlation was found between a lower BAX/BCL-2 ratio and CD38>30% (41 of 54; P=0.030) as well as between CD38>30% and NOTCH1 M (27 of 38 patients; P=0.0004) (Table 2; Online Supplementary Table S3).

Trisomy 12 was confirmed to be strongly correlated with NOTCH1 M (18 of 23; P=0.0002). There was only a trend towards significant association between NOTCH1 M and IGHV UM status (48 of 62; P=0.08). On the other hand, IGHV UM status was correlated with lower BAX/BCL-2 ratio (58 of 81; P=0.030). TP53 M and/or del17p were found in 66 of 178 patients (37.1%). Noteworthy, 23 of 178 patients (13%) were simultaneously NOTCH1 and TP53 mutated. The distribution of clinical and biological prognostic factors according to NOTCH1 M is shown in Table 2. The distribution of prognostic factors according to the BAX/BCL-2 ratio and CD38 was obtained in 113 patients from Rome and shown in Table 2 and the Online Supplementary Table S3.

Relevance of NOTCH1 mutations as biological prognostic parameter

The mean peripheral lymphocyte percentage change from baseline, calculated at 3 months on ibrutinib, was lower in NOTCH1 M patients than in NOTCH1 wild-type (WT) patients (14% vs. 54%; P=0.022, Mann-Whitney test), thus confirming a reduced redistribution lymphocytosis (Figure 1A).

Moreover, the mean percent SPD change, calculated at 6 months on ibrutinib, was lower in NOTCH1 M patients than in NOTCH1 WT patients (53% vs. 80%; P<0.0001, Mann- Whitney test), confirming a significant poor nodal response (Figure 1B). Moreover, we compared NOTCH1 M plus lower BAX/BCL-2 ratio versus NOTCH1 M plus higher BAX/BCL-2 ratio with respect to redistribution lymphocytosis and lymph node shrinkage. No significant differences were found between these two subsets (Online Supplementary Figure S3).

Figure 1.Box plots by NOTCH1 wild-type and mutated groups showed significant lower redistribution lymphocytosis after 3 months on ibrutinib treatment in NOTCH1 M patients (A) and equally lower sum of the product of diameter (SPD) values after 6 months in NOTCH1 M patients (B). pts: points; WT: wild-type; M: mutated; B LYMPHS: B lymphocytes.

Table 2.Distribution of prognostic factors in chronic lymphocytic leukemia according to NOTCH1 mutations

NOTCH1 mutations, BAX/BCL-2 ratio and their impact on clinical outcome

According to clinical endpoints, ORR was 91% [complete response (CR): 18%, partial response (PR): 28%, PR with lymphocytosis (PR-L): 45%] (Table 1). The estimated 2-year and 4-year OS were 84% and 71%, respectively (Table 1; Online Supplementary Figure S4). Noteworthy, OS was longer in patients previously treated with one line of chemoimmunotherapy before ibrutinib (P=0.02, Online Supplementary Figure S5). PR and PR-L were significantly correlated with NOTCH1 M (30 of 65 and 22 of 65, respectively; P=0.00001, Online Supplementary Table S4). Of note, PR, PRL and chemoresistance were also associated with lower BAX7BCL-2 ratio (23 of 29, 33 of 52 and nine of nine, respectively; P=0.002, Online Supplementary Table S5). Interestingly, discontinuation due to disease progression was more frequent in NOTCH1 M patients than in NOTCH1 WT patients (P=0.034, Online Supplementary Table S4). Significant shorter PFS and OS were observed in NOTCH1 M patients (34% vs. 76% and 56% vs. 83% at 3 years, respectively; P=0.00002 and P=0.001; Figure 2A and B). There were no significant differences among VAF range 1-10%, 10.5-20% and above 20% with respect to PFS and OS, as shown in the Online Supplementary Figures S6 and S7. Moreover, we restricted the analysis of NOTCH1 to the relapse setting only (154 of 180 patients) obtaining similar significant results regarding PFS and OS (Online Supplementary Figures S8 and S9).

Table 3.Multivariate Cox regression analysis

Additive prognostic properties of NOTCH1 mutations and BAX/BCL-2 ratio

In order to obtain a better refinement in the prognostic assessment of PFS and OS, we combined the values of the BAX/BCL-2 ratio with those of NOTCH1. Within the subset of 113 patients from Rome, shorter PFS and OS were detected both in patients with NOTCH1 M (46% vs. 83% and 68% vs. 86% at 3 years, respectively; P=0.0019 and P=0.031, Online Supplementary Figure S10A and B) and with lower BAX/BCL-2 ratio (60% vs. 97% and 72% vs. 94% at 3 years, respectively; P=0.019 and P=0.013, Figure 3A and B). Therefore, higher or lower BAX/BCL-2 ratio combined with NOTCH1 WT or NOTCH1 M identified two subsets of patients, the former with the best prognosis and the latter with the worst prognosis with respect to both PFS (97% vs. 42%; P=0.0002, Figure 4A) and OS (94% vs. 63%;P=0.005, Figure 4B), confirming the true additive prognostic properties of these two prognosticators.

Multivariate analysis

The clinical impact of NOTCH1 as independent prognosticator was checked by multivariate Cox proportional hazards analysis applied to models including two other prognosticators proven to be significant in univariate analysis (Table 2). With respect to PFS, NOTCH1 M (P=0.0002) were confirmed as an adverse independent prognostic factor (P=0.00006) together with >2 previous lines of therapy (P=0.004) and TP53 M (P=0.028) (Table 3). Similarly, in a multivariate analysis of OS, NOTCH1 M retained an independent prognostic value (P=0.0039) together with >2 previous lines of therapy (p=0.015) and TP53 M (p=0.047) (Table 3). NOTCH1 M and >2 previous lines of therapy were confirmed as independent prognosticators for PFS (P=0.035 and P=0.015, respectively) also in a model that included the BAX7BCL-2 ratio, available in a smaller subset of cases (n=113, Online Supplementary Table S6). Conversely, in the same subset of patients, no factor emerged as independent prognosticator for OS (Online Supplementary Table S6).

Discussion

In the present study we evaluated the efficacy of ibrutinib treatment in the high-risk NOTCH1 M CLL group and correlated NOTCH1 M to BAX/BCL-2 ratio, a value reflecting the susceptibility of cells to apoptosis. Efficacy of ibrutinib remained high at 4-year follow-up in almost all pre-treated patients with CLL, with 71% of patients alive and progression free, similarly to other studies.¹⁷ Moreover, ibrutinib was more effective in patients previously treated with only one line therapy, compared to patients previously treated with >2 lines of therapy (Online Supplementary Figure S5). On the other hand, the clinical outcome was similar for patients receiving first-line ibrutinib and patients with one previous therapy, probably due to the high incidence of TP53 mutated patients (24 of 26) in first-line setting (Online Supplementary Figure S5).

Figure 2.Progression-free survival and overall survival curves based on NOTCH1. Kaplan-Meier plot comparing progression-free survival (PFS) (A) and overall survival (OS) (B) based on NOTCH1. NOTCH1 mutated (NOTCH1 M) patients experienced both a shorter PFS and OS.

In CLL, the frequency of NOTCH1 M cases between 6-12% if evaluated at presentation, increases to about 15-20% in the context of fludarabine refractory patients.^18,19 The higher frequency of NOTCH1 M characterizing our cohort of patients (36%) could be attributed both to the previous lines of chemotherapy and to the very low cut-off (>1%) chosen for NOTCH1 M. The adverse clinical outcome of patients with NOTCH1 M CLL was confirmed in univariate analysis in several independent cohorts of patients treated with chemo-immunotherapy.^20-24 Since clonal CLL cells accumulate because of prolonged survival due to impairment of apoptosis, the analysis of the BAX/BCL-2 ratio could be a valid tool to provide information on the chemo-sensitivity of CLL cells.^9,10

We addressed the clinical impact of both NOTCH1 M, evaluated by NGS, and BAX/BCL-2 ratio, determined by flow cytometry, in patients with CLL homogeneously treated with ibrutinib, mainly in a (R/R) setting. Determination of both parameters was done prior to starting ibrutinib therapy.

The NGS approach used for NOTCH1 M analysis allowed detection of allele frequency as low as 1%, highlighting the presence of subclonal mutations in 32% of total NOTCH1 M cases.^1,16,26 Of note, subclonal NOTCH1 M (i.e., VAF<10%) had similar prognostic impact as clonal mutations (Online Supplementary Figures S6 and S7); consistently a receiver operating characteristic curve analysis confirmed the use of 1% as optimal cut-off (Online Supplementary Figures S2). In this context, detection of NOTCH1 M by NGS could be viewed as a useful tool for clinical follow-up of patients as well as for minimal residual disease studies, although the latter use remains speculative at the moment. From a biological point of view, we found a significant relationship between NOTCH1 and some other prognosticators. In particular, a significant correlation between NOTCH1 M and higher CD49d or CD38 expressions was observed, as well a trend towards an association between NOTCH1 M and IGHV UM status, in keeping with previous observations by us and others.^27,21,4,28 Further, co-occurrence of NOTCH1 M and TP53 M characterized 13% of our patients (23 of 178), a rather high percentage if compared to previous reports where concomitant NOTCH1 M and TP53 M, preferentially affecting the same leukemic cells,²⁹ accounted for 1.2-2.6% of CLL patients.^20,23 This may be due to the high number of pre-treated patients and to the low cut-off chosen by us for NOTCH1 M detection. We confirmed that NOTCH1 M were strongly correlated with trisomy 12, in line with previous reports describing a high NOTCH1 M rate in CLL cases with isolated trisomy 12 and a lower frequency in cases characterized by additional chromosomal abnormalities.^30-32 In particular, a mutation frequency of 41.9% was reported in aggressive trisomy 12 cases, suggesting a pivotal role of NOTCH1 activation in this group.³³ Moreover, we observed here a lower BAX/BCL-2 ratio in NOTCH1 M patients, in keeping with our previous studies showing NOTCH1-dependent activation of the NF-kB pathway that may result in the upregulation of target genes, including BCL-2.⁴

Figure 3.Progression-free survival and overall survival curves based on BAX/BCL-2 ratio within the subset of 113 patients from Rome. Kaplan-Meier plot comparing progression-free survival (PFS) (A) and overall survival (OS) (B) based on the BAX/BCL-2 ratio. Patients with a BAX/BCL-2 ratio <1.5 experienced both shorter PFS and OS.

The strong correlation between lower BAX/BCL-2 ratio and NOTCH1 M suggests that the poor prognosis of NOTCH1 M patients may be related to the lack of apoptosis, although these observations need further confirmation.

The variability in the degree and kinetics of ibrutinibinduced recirculation lymphocytosis has been highlighted by several studies,^34,35 and was also confirmed in the present study. Here we show that at 3 months on ibrutinib, the typical ibrutinib-induced peak of lymphocytosis is observed in NOTCH1 WT patients, but not in NOTCH1 M cases. Moreover, even though the analysis of nodal response confirmed an overall significant reduction in organomegaly and lymph node size in most cases at 6 months on ibrutinib, NOTCH1 M cases experienced a significant lower nodal response compared to NOTCH1 WT cases. These results may be explained by the strong correlation between CD49d overexpression and NOTCH1 M (51 of 65 cases), in line with the reported involvement of the NOTCH1 pathway in the regulation of CD49d expression.⁴ Consistently, CD49d associates with nodal presentation and subsequent development of lymphadenopathy in patients with CLL.³⁶ Moreover, CD49d expression identifies cases with reduced lymphocytosis and inferior nodal response upon ibrutinib treatment, suggesting the retention of CD49d-expressing cells in tissue sites via activated VLA-4 .⁷

Figure 4.Progression-free survival and overall survival curves in relation to combined BAX/BCL-2 ratio and NOTCH1. Progression-free survival (PFS) and overall survival (OS) were shorter within the NOTCH1 mutated (NOTCH1 M) plus BAX/BCL-2 <1.5 subgroup (A-B), showing additive prognostic properties.

Consistently with the high frequency of pre-treated patients in our cohort (154 of 180), the OS values at 2 and 4 years (84% and 71% respectively), were similar to those reported for the phase III RESONATE study in patients with previously treated CLL/SLL.³⁷

We have recently reported that NOTCH1 M identify a subgroup of patients with CLL with worse prognosis in the setting of a rituximab-based induction and consolidation treatment.³⁸ Here, we described a negative prognostic impact of NOTCH1 M also in the ibrutinib setting. Our findings differ from those resulting from the extended follow up from the RESONATE study of relapsed/refractory CLL, where the presence of NOTCH1 M did not negatively affect the efficacy of ibrutinib on disease progression outcomes. ³⁷ This difference can be explained by the very low cut-off (>1%) chosen for NOTCH1 M in our study, although for the validation of these findings additional independent cohorts are needed.

The here reported capacity of BAX/BCL-2 index to identify patients with a different response to ibrutinib could be of interest in the light of the treatments protocols associating B-cell receptor inhibitors and BH3 mimetics such as venetoclax.³⁹

Moreover, the additive negative prognostic value of NOTCH1 M and low BAX/BCL-2 ratio described by us, further support the rationale to improve the efficacy of ibrutinib by using the BCL-2 inhibitor venetoclax in patients with NOTCH1 mutated CLL.¹⁰ Interestingly, an additive prognostic impact of the combination of BAX/BCL-2 and NOTCH1 M in the setting of chemo-immunotherapy was also reported by us.¹⁰

Several independent cohorts of patients confirmed the adverse clinical outcome of NOTCH1 M with CLL in univariate analysis,^20-23,40 although conflicting results are reported about its independent prognostic effect. In particular, NOTCH1 M did not retain independent significance as a predictor of time-to-first treatment in one of the largest series of patients with CLL,⁴¹ while in another study it emerged as an independent predictor of shorter survival, along with TP53 abnormalities.⁴² Here NOTCH1 M were confirmed to be an independent prognostic factor together with previous lines of therapy and TP53 both with respect to PFS and OS. The apparent higher prognostic impact of NOTCH1 M compared to TP53 mutation, as emerged in our multivariable analysis, may be explained by the greater number of TP53 mutated cases treated first line with ibrutinib, hence with a better prognosis than NOTCH1 M cases that were more frequently treated with ibrutinib in second or further lines of therapy.

The current use of B-cell receptor and BCL-2 inhibitors led to high-rate improvement of outcome in CLL. However, several issues remain, resulting in resistance/progression thus limiting the eradication of the tumour. The growing evidence for a critical role of the NOTCH1 pathway in CLL makes this cancer gene a target to design tailored treatments for this peculiar subset through specific NOTCH1- targeted therapies. In this context, γ-secretase inhibitors are the most extensively explored anti-NOTCH1 molecules and their combination with fludarabine demonstrated antitumour effects in primary CLL with NOTCH1 M.⁴³ Noteworthy, a humanized antibody targeting NOTCH1 (clinicaltrials gov. Identifier: OMP-52M51) entered phase I trial in relapsed/refractory lymphoid malignancies.⁴⁴ However, to date, the future treatment of CLL with NOTCH1 M relies on the association of small molecule inhibitors targeting both the BCR pathway and the antiapoptotic BCL-2 protein.

Footnotes

• Received February 29, 2020
• Accepted July 20, 2020

Correspondence

GIOVANNI DEL POETA g.delpoeta@tin.it

Disclosures

No conflicts of interest to disclose.

Contributions

GDP and VG designed the study, interpreted data, performed statistical analysis, wrote the manuscript and gave final approval of the manuscript; AB and AZ contributed to study design and data interpretation and to write the manuscript; LL, AC and MIDP contributed to interpret the data and to write the manuscript; AZ, FMR and GDP obtained flow cytometric data; FMR performed FISH cytogenetic analysis; VG, FP and RB investigated IGHV, NOTCH1 and TP53 mutations; FB, SA, GG, AV contributed to study design and data interpretation; II, MP, PdF, MC recruited the patients and collected clinical data.