Long-term outcomes with continuous venetoclax for relapsed chronic lymphocytic leukemia

Key Points

• •Approximately one-third of patients with R/R CLL treated with continuous venetoclax experience a “long-term” response of >5 years.
• •Long-term responders remain at risk of relapse, infections, hematological toxicity, and emergence of BCL2 mutations.

Visual Abstract

Abstract

Venetoclax was first approved for continuous use in relapsed/refractory (R/R) chronic lymphocytic leukemia (CLL), but the efficacy and consequences of very long-term BCL2 inhibition are unknown. We describe the frequency, characteristics, and outcomes of patients with R/R CLL treated with >5 years of continuous venetoclax. Long-term responders (>5 years of continuous therapy without progressive disease [PD]) were identified from a cohort of 86 patients with R/R CLL treated with continuous venetoclax with or without rituximab. Landmark analyses at 2 and 5 years assessed association between undetectable measurable residual disease (uMRD; 10⁻⁴ peripheral blood flow cytometry) and progression-free survival (PFS). Next-generation sequencing was performed at PD for BCL2 and TP53 mutations. Twenty-nine patients (33%) were long-term responders. Compared to those with PD within 5 years, they were more likely to have mutated immunoglobulin heavy chain variable region (44% vs 14%; P = .029), noncomplex karyotype (89 vs 50%; P = .043), and uMRD (79 vs 30%; P < .001). At a median follow-up of 11.1 years, 76% had discontinued venetoclax, mostly due to PD. In a landmark analysis of patients continuing venetoclax beyond 2 years, 5-year PFS was 87% for those with uMRD vs 45% without (hazard ratio, 1.30; 95% confidence interval, 1.12-1.52; P = .001). BCL2 mutations were detected in 4 of 8 patients at PD. Grade ≥3 toxicities, including infection (51%), neutropenia (20%), and thrombocytopenia (10%), occurred at consistent frequency throughout treatment. Long-term responses to continuous venetoclax occur in approximately one-third of patients with R/R CLL. Sustained uMRD confers the most favorable outcomes, although all patients show a continuous risk of relapse, infection, and cytopenias.

Introduction

The treatment of chronic lymphocytic leukemia (CLL) has been revolutionized by the development of venetoclax, an orally bioavailable BCL2 inhibitor, and Bruton tyrosine kinase inhibitors (BTKis).¹ When venetoclax is used continuously for relapsed/refractory (R/R) CLL, monotherapy yields an overall response rate (ORR) of 73.5% to 80%, complete response rate (CRR) of 15% to 20%, and median progression-free survival (mPFS) of 25 to 28 months, including in patients whose CLL harbors del17p and/or TP53 mutation.2, 3, 4 Depth of response is augmented by addition of rituximab (ORR, 86%; CRR, 51%; mPFS, 5.6 years).⁵ Although initially studied and approved as continuous therapy, venetoclax is now mostly delivered for a fixed duration in combination with obinutuzumab or BTKi6, 7, 8, 9, 10 for up-front disease and with rituximab for relapsed disease.¹¹ Several important observations have emerged from the use of fixed-duration venetoclax-rituximab in R/R CLL. Firstly, a negative prognosis is imparted by immunoglobulin heavy chain variable region (IGHV)-unmutated status (IGHV-U; 5-year PFS, 28.7% vs 78.7%), TP53 aberrancy, complex karyotype (CK; 5-year PFS, 17.7% vs 48.5%), and bulky lymphadenopathy (>5 cm).11, 12, 13 Secondly, achieving undetectable minimal residual disease (uMRD) below a threshold of 10⁻⁴ in the peripheral blood (PB) at the end of treatment (EOT) predicts a superior outcome (5-year PFS from EOT, 40.7 vs 12.5%).¹³ Thirdly, toxicities from venetoclax-rituximab are predictable and mainly limited to cytopenias and infections. Finally, BCL2 resistance mutations (eg, G101V) are uncommon at the time of progressive disease (PD) after fixed-duration therapy, and most patients respond to either venetoclax retreatment or subsequent BTKi.¹⁴ In the frontline treatment of CLL, venetoclax and obinutuzumab achieves a CRR of 85% to 87%, EOT PB uMRD of 76% to 86.5%, and an mPFS ranging from 76 months in older patients to 70% at 5 years in younger patients.^10,15, 16, 17 As in relapsed disease, the duration of response is influenced by attainment of UMRD, IGHV status, and TP53 aberrancy.^10,15, 16, 17

In contrast to fixed-duration therapy, less is known about the predictive power of uMRD, late toxicity, and patterns of disease progression when venetoclax is used continuously, particularly outside of the well-described del(17p) cohort.¹⁸ Therefore, we studied a cohort of patients with R/R CLL treated with continuous venetoclax on early-phase clinical trials for >5 years. Establishing the characteristics, outcomes, and toxicity of this group is important for several reasons. Firstly, there is an evolving role for venetoclax retreatment and MRD-guided treatment duration that may result in longer net exposure to venetoclax in the future.19, 20, 21 Secondly, the long-term toxicity profile is relevant for other malignancies in which continuous venetoclax has an established (eg, acute myeloid leukemia) or evolving (eg, mantle cell lymphoma) role.^22,23 Finally, there may be subpopulations of patients with CLL who benefit from continuous venetoclax over fixed-duration therapy.

Methods

Patients

All patients with R/R CLL treated with continuous venetoclax with or without rituximab in 1 of 4 early-phase clinical trials (M12-175, NCT01328626; M13-982, NCT01889186; M13-365, NCT01682616; M15-889, NCT 02980731) from December 2011 onward at 2 institutions (Peter MacCallum Cancer Centre and The Royal Melbourne Hospital) were evaluated. Eligibility criteria and details of these 4 trials have previously been published.^2,3,24,25 Briefly, all 4 trials enrolled patients with R/R CLL [M13-982 included only patients with disease harboring del(17p)]. Venetoclax was given as monotherapy in 3 trials and in combination with rituximab (6-9 monthly doses) in M13-365 trial. Doses were between 200 and 600 mg in phase 1 studies (M12-175 and M13-365) and 400 mg in other studies. Cessation vs continuation of venetoclax at the time of PD, timing of next therapy, and choice of next therapy were at investigator and patient discretion.

All patients provided written informed consent and study protocols were approved by local institutional review boards and conducted in accordance with the Declaration of Helsinki.

Baseline data

Data were collected up until January 2026 through trial assessments and medical records (after trial completion). “Long-term responders” were defined by continuous venetoclax for >5 years without PD. Patients discontinuing drug for reasons other than PD or death (eg, intolerance and patient preference) were excluded from comparative analysis.

Baseline clinicopathologic characteristics were collected including age, prior lines of therapy (PLOT), fludarabine refractoriness (defined by failure to respond or progression within 6 months of completion), bulky adenopathy (>5 cm), CK (≥3 chromosomal aberrations on conventional stimulated karyotyping), highly CK (≥5 chromosomal aberrations on conventional stimulated karyotyping), TP53 aberrancy [TP53 mutation and/or del(17p)], and IGHV status. TP53 mutational status was assessed using targeted next-generation sequencing (NGS) of all coding exons of TP53 (NM_000546.5), with variant curation based on the Catalogue of Somatic Mutations in Cancer (https://cancer.sanger.ac.uk/cosmic). Fluorescence in situ hybridization was performed for del(17p) according to standard practice. Response and PD were assessed according to the 2018 International Workshop on Chronic Lymphocytic Leukemia (iwCLL) criteria.²⁶ PB MRD was assessed using multiparameter flow cytometry by European Research Initiative in CLL methodology. uMRD was defined as <1 leukemic cell in 10⁴ nucleated cells, with an assay sensitivity of at least 10⁻⁴. PB MRD and bone marrow (BM) assessments were mandated during the initial portion of each individual clinical trial. Beyond this time point, the frequency of PB uMRD and BM assessments was at the clinician’s discretion.

Long-term responders

For long-term responders (>5 years), additional efficacy data were collected, including time to detectable MRD (MRD⁺), PFS, overall survival (OS), timing and reason for venetoclax cessation, time to next treatment (TTNT), and response to next treatment. PFS was measured from the first dose of venetoclax to PD or death. OS was measured from the first dose of venetoclax to death. TTNT was measured from the first dose of venetoclax to the initiation of subsequent therapy. Formal response assessment with computed tomography scans were conducted as per the original trial protocols.

Toxicity data were collected and graded according to the National Cancer Institute Common Terminology for Adverse Events, version 4.0. Grade ≥3 infection, second malignancies, and hypogammaglobulinemia (serum immunoglobulin G [IgG] <7 g/L) were assessed from the time of venetoclax commencement to cessation or last follow-up. Grade ≥3 cytopenias were assessed from 5 years after therapy to cessation or last follow-up.

Testing for BCL2 and TP53 mutations was performed by NGS for long-term responders with iWCLL PD while on continuous venetoclax. The estimated limit of detection across the entire BCL2 coding region was 0.5% variant allele frequency (VAF).

Statistics

Characteristics of long-term responders were compared to those with PD within <5 years using the Fisher exact test (categorical variables) and the 2-sided t test (continuous variables). PFS and OS were analyzed using the Kaplan-Meier methods, with estimation of corresponding 95% confidence intervals (CI). Landmark analysis was undertaken at 2 and 5 years from venetoclax initiation to assess the impact of uMRD on PFS. Patients were included if they were progression free, remained on venetoclax, and had available MRD data within 6 months of the landmark. Associations between PFS and uMRD were assessed using Cox proportional hazards model and restricted mean survival time (RMST). All statistics were performed using Stata for windows (StataCorp 2024, College Station, TX).

Results

Characteristics of long-term responders

Eighty-six patients with R/R CLL were treated with continuous venetoclax across 4 clinical trials (M12-175, n = 37; M13-982, n = 11; M13-365, n = 16; M15-889, n = 22): 81% (70) as monotherapy and 19% (16) in combination with rituximab (Figure 1). The median age at venetoclax initiation was 67 years (range, 45-87), and the median PLOT was 3 (range, 1-8). The cohort was enriched for high-risk features: 55% with TP53 aberrant, 75% (37/49 assessed) with IGHV-U, and 39% (23/59 assessed) with CK. All patients were BTKi naïve.

Figure 1.

Patient disposition.^auMRD defined by <1 CLL cells per in 10⁴ total nucleated cells by flow cytometry in PB. FU, follow-up; Ven, venetoclax.

Among these patients, 29 patients (33%) continued venetoclax for >5 years and are henceforth referred to as “long-term responders.” Four patients ceased venetoclax before 5 years for reasons other than PD and were excluded from comparative analysis (allogeneic stem cell transplant, cardiomyopathy, prostate cancer, and patient preference in uMRD CR, n = 1 each). Long-term responders had higher incidence of IHCV region-mutated (IGHV-M) CLL (44% vs 14%; P = .029), attainment of CR (52% vs 26%; P = .012), and PB uMRD (76% vs 30%; P = .043) than patients with PD before 5 years. CK was less common among long-term responders (16 vs 50%; P = .043). Importantly, neither TP53 wild-type status nor achievement of CR was a prerequisite for long-term response, with 41% of long-term responders having TP53 aberrant disease and only 52% achieving CR as best response (Table 1). Of long-term responses, 6 of 29 (20%) occurred in patients who never achieved uMRD. These 6 patients were all assessed for uMRD, albeit at differing frequencies, and showed a mixture of IGHV status (3/6 IGHV-U, 2/6-IGHV-M, and 1/6 IGHV unknown) with otherwise favorable genomics (0/6 CK and 1/6 TP53 aberrant).

Table 1.

Characteristics of long-term responders and comparator cohort

	Venetoclax >5 y (n = 29)	Venetoclax <5 y (n = 53)	n	P value
Pretreatment clinical variables
Age, median, y	67.7	67.3	82	.93
PLOT, median	3	3	82	.67
Fludarabine refractory, n (%)	9 (31)	23 (43)	82	.35
Bulk >5 cm, n (%)	10 (34)	25 (47)	82	.35
IGHV-M, n (%)	7 (44)	4 (14)	45	.03
TP53 aberrant, n (%)	12 (41)	29 (54)	82	.36
CK, ≥3 abnormalities, n (%)	3 (16)	18 (50)	55	.02
HCK, ≥5 abnormalities, n (%)	2 (10)	8 (22)	55	.47
Treatment variable, with rituximab, n (%)	4 (14)	12 (22)	82	.40
Response variable, n (%)
CR	15 (52)	14 (26)	82	.03
uMRD achieved (PB; MRD 10−4)	23 (79)	16 (30)	82	.00

Continuous variables (age and PLOT) were assessed with 2-sided t test, and categorical variables were accessed with Fisher exact test. Bold indicates statsitically significant findings.

HCK, highly CK.

Outcomes of long-term responders

Among long-term responders, the median survivor follow-up was 11.1 years (range, 5.5-12.4), which is an additional 6.1 years beyond the 5-year landmark. At last follow-up, 22 of 29 (76%) long-term responders had ceased venetoclax (Figure 1). The median total duration of venetoclax treatment was 7.9 years (range, 5.1-12.3). Venetoclax was mostly ceased for PD (12/22), with other causes including development of therapy-related myeloid neoplasm (t-MN) (4/22), death without PD (2/22; 1 infection and 1 unknown cause), and increasing frailty (2/22; aged 93 and 73 years). Drug cessation due to toxicity occurred in 2 patients: one with recurrent bronchitis and the other with progressive cytopenias without BM evaluation. Seven patients remained on venetoclax at last follow-up (median treatment duration, 10.9 years; [range, 7.8-12.4]; Figure 1). Of 11 deaths (38% of the long-term responders), only 2 were attributable to PD, and both were from Richter transformation (RT; Figure 2). Other causes of death included t-MN (n = 4), infection (n = 1), and unrelated medical conditions (n = 4; dementia, heart failure, stroke, and unknown cause).

Figure 2.

Outcomes of long-term responders. Swimmer plot for patients continuing Ven beyond 5 years. Ven, venetoclax.

From the 5-year time point, the mPFS was an additional 3 years (95% CI, 2.3-3.8), and the median OS was not reached. The 2-year PFS and OS after the 5-year time point were 74% (95% CI, 56-87) and 85% (95% CI, 66-94), respectively (Figure 3). There was a continual pattern of relapse, with no plateau observed (Figure 3A).

Figure 3.

Kaplan-Meier analysis of PFS and OS. (A) PFS and (B) OS for 29 patients who remained on venetoclax for >5 years. The mPFS was 2.98 years (95% CI, 2.33-3.80). The median OS was not reached.

Progressions on long-term venetoclax beyond 5 years of therapy

Twelve instances of PD occurred beyond 5 years: histologically RT in 2 (17%), CLL in 9 (75%), and “accelerated” CLL in 1 (8%). RT was rapidly fatal in both instances (Figure 2). Of 9 patients with PD after prior uMRD attainment, all had MRD relapse preceding iwCLL-defined PD, with a median latency time of 3.6 years (range, 2.1-6.1). Five of 9 underwent treatment aiming to restore uMRD (addition of rituximab alone, n = 3; addition of rituximab plus venetoclax dose increase, n = 2), with variable responses (restoration of uMRD, n = 2 [duration, 18 and 34 months]; reduction of MRD but >10⁻⁴, n = 2; no response, n = 1), as previously published.²⁷

Twelve patients received a subsequent line of therapy, with a median TTNT after PD of 5.47 months (range, 0.3-23.4). The next treatment was acalabrutinib in 6 patients (CLL; 1 CR, 3 partial response, and 2 stable disease [SD]) and zanubrutinib in 2 (CLL; 1 partial response and 1 SD). One patient each received ibrutinib (RT; PD), R-CHOP (rituximab, cyclophosphamide, doxorubicin, Oncovin [vincristine], and prednisone; RT; PD), chimeric antigen receptor T cells (CLL; PD), and sonrotoclax (CLL; response unknown; Figure 4A). In 11 response-evaluable patients, the ORR was 45%. Among the 8 cases of progressive CLL treated with BTKis, 5 responded (ORR, 63%), and 3 had SD. At a median follow-up of 23.3 months from commencing the next therapy, mPFS2 and OS were not reached. At 2 years post commencing next therapy, PFS was 65% (95% CI, 24-88) and OS was 82% (95% CI, 45-95; Figure 4B). Of the 3 progression events occurring on postvenetoclax therapy, 2 were in patients with RT at the time of original PD, and the third patient had RT at second PD (Figure 4B).

Figure 4.

Outcomes of patients with disease progression after long-term Ven. (A) Swimmer plot for patients after PD (n = 12). (B) Kaplan-Meier OS analysis for patients with disease progression on Ven (n = 12). CAR-T, chimeric antigen receptor T‑cell therapy; Ven, venetoclax.

Molecular characteristics of progressors

At PD, NGS was performed in 8 of 12 patients. Acquired BCL2 resistance mutations were detected in 4 patients (50%), with multiple distinct mutations in 2 patients and 7 unique mutations in aggregate. The VAF ranged from <1% to 30%. The most common mutations were Arg107_Arg110 (n = 2) and Val156Asp (n = 2), along with other mutations including Asp103Tyr, Asp103Glu, and Ala113Gly (n = 1 each; supplemental Table 1). Three of 8 patients had acquired TP53 mutations (VAF, 2%, 19%, and not reported) at the time of PD, which were not detected at baseline, including 1 patient with concurrent BCL2 mutation. There were 2 patients with BAX mutation detected at the time of PD, presumably in the myeloid compartment.

Long-term uMRD responders

Of 7 patients with ongoing response, only 2 had enduring PB uMRD at the last follow-up (Figure 1), one of whom had confirmed uMRD in BM. Both patients had favorable genomics, with IGHV-M in 2 of 2 tested, TP53 wild type in 2 of 2, and CK in none. Neither of these 2 patients experienced grade ≥3 cytopenias or infections. Two patients died of t-MN while in ongoing uMRD CR.

Prognostic value of uMRD at 2-year and 5-year time point with continuous venetoclax

Of 86 patients commencing continuous venetoclax, 54 remained progression free at the 2-year landmark, and 43 had MRD data meeting the criteria for inclusion in landmark analysis (23 uMRD and 20 MRD⁺). The 5-year PFS (from the time of venetoclax commencement) was 87.0% (95% CI, 64.8-95.6) in those with uMRD and 45.1% (95% CI, 22.1-65.7) in those without (hazard ratio [HR], 0.77; 95% CI, 0.37-1.61; P = .50; Figure 5A). Because the Kaplan-Meier curves cross, RMST was used to compare the impact of MRD status at 2 years on PFS. Over the additional 3 years of follow-up, RMST was significantly longer in the uMRD group (HR, 1.30; 95% CI, 1.12-1.52; P = .001).

Figure 5.

PFS based on MRD status at 2 years and 5 years. (A) Landmark analysis of iwCLL progression stratified by MRD status at 2 years (± 6 months) for patients who remained on venetoclax and are progression free (n = 43). Patients who had disease progression or came off venetoclax before 2 years were excluded (n = 46). (B) Landmark analysis of iwCLL progression stratified by MRD status at 5 years for patients who remained on venetoclax and were progression free (n = 29).

At the 5-year landmark, only 9 of 29 patients maintained uMRD. MRD⁺ status at the 5-year landmark was significantly associated with shorter PFS (HR, 0.229; CI, 0.074-0.7012; P = .0045; Figure 5B).

Toxicity in long-term responders

Infectious adverse events

Of 29 patients, 27 had data regarding infections available. Fourteen patients (52%) experienced a total of 26 grade ≥3 infections while receiving venetoclax (1 event in 6 patients and ≥2 events in 14 patients). Most events were grade 3, with 4 grade 4 infections (Escherichia coli bloodstream infection, disseminated adenovirus, Pneumocystis pneumonia, and cholangitis) and 1 grade 5 infection (pneumonia). Nine patients (31%) experienced grade ≥3 infection after 5 years. This group had significantly higher rate of grade ≥3 neutropenia (66.7% vs 20%; P = .032) than those who did not experience grade ≥3 infection after 5 years but otherwise had similar age, PLOT, fludarabine exposure, and incidence of hypogammaglobulinemia (supplemental Table 3). The frequency of grade ≥3 infection in the first 5 years of venetoclax treatment (0.8 infections per 100 patient months) was comparable to the rate observed beyond 5 years (1.2 infections per 100 patient months). There was no plateau in time to first grade ≥3 infection (Figure 6).

Figure 6.

Cumulative incidence of first grade ≥3 infection. Reverse Kaplan-Meier curve showing cumulative incidence of grade ≥3 infection.

Table 2 shows distribution of 26 infections by site, time, and risk factor. Pulmonary infections predominated, particularly after year 5. Neutropenia (neutrophil count of <1 × 10⁹/L) was present in 5 of 19 patients (26%), lymphopenia (lymphocyte count of <1 × 10⁹/L) in 8 of 19 patients (42%), and hypogammaglobulinemia (IgG level of <7 g/L) in 12 of 15 patients (12/15) with contemporaneous data available. Five patients (18.5%) experienced infections classified as opportunistic infections (OI), with 1 case each of Pneumocystis pneumonia at 6.5 years, fungal pneumonia (species unknown) at 5.6 years, disseminated zoster at 2.8 years, ocular toxoplasmosis at 2 months, and disseminated adenovirus at 2.5 months (supplemental Table 2).

Table 2.

Timing, site, type, and risk factors for grade 3 or higher infections in long-term responders

	Total	Year 1-5	Year ≥6
Grade ≥3 infections	26	15	11
Site
Pulmonary	5	1	4
Febrile neutropenia (culture negative)	4	3	1
Gastrointestinal	4	4	0
Urinary tract	4	2	2
Skin/soft tissue	3	1	2
Bloodstream	3	2	1
Ear, nose, throat	2	1	1
Ocular	1	1	0
Opportunistic
Yes	5	3	2
No	21	12	9
Risk factor at time
ANC <1 × 109/L (n = 19)	5	3	2
ALC <1 × 109/L (n = 19)	8	4	4
Hypogammaglobulinemia (n = 15)	12	3	9

A total of 27 patients were included. Two patients did not provide consent for long-term toxicity follow-up.

ALC, absolute lymphocyte count; ANC, absolute neutrophil count.

Noninfectious adverse effects

Hypogammaglobulinemia (IgG level of <7 g/L) at any time point during venetoclax treatment occurred in 19 of 29 patients (66%; baseline rate of hypogammaglobulinemia at venetoclax commencement unknown), with 10 of 29 (35%) requiring immunoglobulin replacement. Neutropenia after 5-year time point occurred in 69% of patients (grade 3, 20%; grade 4, 14%) and thrombocytopenia in 69% (grade 3, 10%; grade 4, 7%). Six of 10 patients (60%) with grade ≥3 neutropenia after 5 years also suffered from grade ≥3 infection, and 4 of 10 (40%) went on to develop t-MN.

There were 16 second malignancies across 14 patients; 6 nonmelanoma skin cancers (NMSC), 4 t-MN, 2 prostate cancers, and 1 each of parotid cancer, lung cancer, melanoma, and ocular surface squamous neoplasia. The frequency of NMSC was 20% (6/29), and that of non-NMSC malignancy was 35% (10/29). All 4 patients with t-MN had at least 3 PLOT, were fludarabine exposed, and died. TP53 aberrancy was evident in 3 t-MN, but BAX mutations were not detected.

Discussion

We have demonstrated that long-term responses >5 years are achieved in approximately one-third of patients with R/R CLL treated with continuous venetoclax. Our cohort has been followed for a median of 11.1 years from treatment commencement and exposed to a median of 7.9 years of venetoclax, making this, to our knowledge, the longest treatment duration and follow-up of a continuously treated venetoclax cohort in the published literature.^5,18

Unsurprisingly, long-term responders had significantly higher rates of favorable biological features such as IGHV-M, non-CK, and best response of CR than those progressing within 5 years. These prognostic factors are well established in both fixed-duration and continuous therapies.^4,12,15,28 Less expected, however, is the observation that a subset of patients who do not attain uMRD can nevertheless experience very long-term remissions.

With extended follow-up, there is no apparent plateau in PFS, and even patients with the deepest responses to venetoclax appear to eventually relapse. This is corroborated by long-term follow-up of patients with disease harboring del(17p) treated with continuous venetoclax.¹⁸ The rate of RT in “long-term responders” (2/12 [17%] PD events) is comparable with that reported in other heavily chemotherapy pretreated populations²⁹ and lower than what is reported in patients with disease progressing early on continuous venetoclax (17/25 [68%]).³⁰ This observation aligns with what was seen in the 6-year follow-up of M13-982, in which 25% of progression events were RT, but 83% of these occurred in the first 2 years.¹⁸ Five of 8 patients with non-RT PD in our study responded to subsequent therapy with BTKis, and the remaining 3 had SD. Overall, this provides reassurance that long-term venetoclax use does not select for a highly aggressive or treatment-resistant clone at relapse.

BCL2 resistance mutations occurred in 4 of 8 patients (50%) tested at the time of PD. With the caveat of small numbers, the frequency of BCl2 mutation appears similar to other cohorts in which progression events occurred substantially earlier.^5,31,32 Notably, the G101V mutation (the most common BCL2 mutation) was not detected in our cohort, and Arg107_Arg110 duplication and Val156Asp occurred twice. Further work is required to determine whether different BCL2 mutations predominate depending on the time of progression.

The results of the 5-year landmark analysis showed a statistically significant association between uMRD and PFS (HR, 0.229; CI, 0.074-0.712; P = .004). Although intuitive, to our knowledge, this is a novel finding that has not been previously demonstrated in studies with shorter follow-up. Whether such patients would fare equally well with treatment cessation after a period of sustained uMRD remains unknown.

Importantly, our cohort also provides the longest toxicity data for continuous venetoclax. Previously, Davids et al have published up to 56 months of follow-up with a median 16 months of duration of venetoclax, reporting grade 3 infections in 22% of patients.³³ Here, with much longer follow-up, grade 3 infections occurred in approximately half of patients. Importantly, first episodes of grade ≥3 infection (Figure 5) and OI emerged even after 5 years of venetoclax therapy (Table 2). A similar phenomenon has been seen with continuous ibrutinib in R/R CLL, in which cumulative incidence of OI was 4.7% at 5 years.³⁴ Infectious toxicity is likely related to the cumulative effects of prior treatment (eg, fludarabine), CLL itself (eg, T-cell dysfunction), and the effects of venetoclax on nonmalignant lymphocytes.^35,36 Given that venetoclax may also have some restorative effects on the dysfunctional immune landscape in CLL,³⁷ detailed comparison of the incidence of infection in patients treated with fixed-duration vs continuous therapy would be of value.

Second primary malignancies were relatively frequent in long-term responders, with 20% of patients developing NMSC and 35% developing another malignancy. When accounting for different follow-up periods, this appears similar to what has been reported with continuous ibrutinib.³⁶ t-MN occurred in 14% of “long-term responders,” which is comparable to incidence reported in fludarabine-treated patients with CLL in the prevenetoclax era.³⁸ All 4 cases had prior fludarabine exposure and were fatal. Importantly, BAX mutations were not reported in patients with t-MN, supporting the notion that BAX-mutated clonal hematopoiesis is a distinct phenomenon from chemotherapy-induced t-MN.³⁹

The major strength of our study is the maturity of follow-up. However, there are several limitations. Firstly, genomic data are incomplete, with several patients lacking baseline IGHV status and 4 of 12 patients with PD lacking BCL2 sequencing. Secondly, the nonuniform nature of MRD testing necessitated the exclusion of several patients (10/53) from the 2-year landmark analysis and may have affected reported incidence of uMRD. Thirdly, incidence of infection and cytopenias is subject to the inherent biases of retrospectively collected data. Finally, many of our observations regarding long-term venetoclax efficacy and toxicity are in the context of heavily chemotherapy pretreated patients and may be less relevant for contemporaneous CLL cohorts.

In conclusion, long-term responses >5 years occur in approximately one-third of patients with R/R CLL treated with continuous venetoclax. Despite favorable genomics and high rates of uMRD, long-term responders demonstrate a continual pattern of relapse with high rates of BCL2 mutations and persistent risk of infections/cytopenias. Whether a subgroup of patients with R/R CLL benefit from continuous venetoclax over fixed-duration therapy remains unclear. This question is now being examined in frontline CLL, with ongoing phase 3 trials randomizing patients to MRD-guided or fixed duration venetoclax combinations.⁴⁰

Conflict-of-interest disclosure: S.M. reports honoraria from Janssen and Kite/Gilead. T.E.L. reports honoraria from AbbVie. M.A.A. reports honoraria from Takeda, CSL Behring, Sobi, Sano, Kite/Gilead, AbbVie, BeOne, Janssen, AstraZeneca, Novartis, and Roche. A.W. reports research funding from AbbVie; advisory board fees from AbbVie, AstraZeneca, and Roche; and honorara from AbbVie, AstraZeneca, Roche, BeiGene, Janssen, and Novartis. P.B. reports consulting or advisory fees or honoraria from Adaptive Biotechnologies, Amgen, AstraZeneca, BeOne, Johnson & Johnson, and Servier. P.A.T. reports honoraria from AbbVie, AstraZeneca, BeiGene, Genmab, Janssen, Lily, Merck, and Roche; speaker fees from AbbVie, AstraZeneca, BeiGene, Jannsen, Merck, and Roche; and travel support from Merck. M.A.A., T.E.L., and A.W.R. are employees of the Walter and Eliza Hall Institute of Medical research, which received milestone and royalty payments in relation to venetoclax and has received financial benefits related to these payments. A.W.R. and J.F.S. are inventors on a patent related to venetoclax. A.W.R. has received research funding from Johnson & Johnson. The remaining authors declare no competing financial interests.