Impact of ibrutinib dose adjustment on TTNT in first-line CLL/SLL: a real-world analysis using target trial emulation

Ibrutinib, a once-daily Bruton tyrosine kinase inhibitor (BTKi), is considered a standard-of-care first-line (1L) treatment for patients with chronic lymphocytic leukemia (CLL)/small lymphocytic lymphoma (SLL).¹ Across numerous phase 3 clinical trials in the 1L setting, ibrutinib is the only BTKi that has demonstrated improved overall survival (OS) relative to chemotherapy and/or chemoimmunotherapy treatments;^2-5 and results from a recent pooled analysis of patients with CLL/SLL from 3 phase 3 clinical studies (RESONATE-2, ECOG1912, and iLLUMINATE) indicated similar OS to an age-matched cohort from the general population.⁶ 

Ibrutinib has been approved in 1L CLL/SLL in the United States since 4 March 2016, and dosing flexibility (ie, adjustment to a daily dose of <420 mg/d) has since been introduced as part of the prescribing information.^7,8 Among other BTKis recommended for 1L therapy in patients with CLL/SLL, ibrutinib offers the most flexibility in terms of dosing.^7,8 Dosing flexibility with ibrutinib allows patients to adjust their daily dose to help prevent recurrence or worsening of adverse events while maintaining efficacy by allowing patients to stay on treatment and benefit from long-term treatment outcomes.^6,9,10 Dose adjustment (DA) strategies have also been highlighted as possible avenues to mitigate the risk of drug resistance, which can occur among BTKis.¹¹ Exploratory results from the RESONATE-2 trial, with up to 8 years of follow-up data in patients with CLL/SLL initiating single-agent 1L ibrutinib, indicated that progression-free survival (PFS) and OS were both similar among patients with and without DA.¹⁰ In a pooled analysis of 10 clinical trials, results showed that PFS was not affected by ibrutinib DA, and among patients with DA after a cardiac adverse event, recurrence of the event at the same or worse severity was less frequent vs those without DA.¹² 

In real-world studies, PFS often cannot be directly evaluated, but time to next treatment (TTNT; defined as the time from treatment initiation to the start of an alternative therapy for progressive CLL by the International Workshop on CLL) is a well-established^13-17 and clinically meaningful^18,19 proxy measure to evaluate progression using real-world data sources. In a recent real-world analysis of electronic medical records (EMRs) assessing outcomes in patients treated with 1L single-agent ibrutinib with or without DA, TTNT was found to be similar between the cohorts, and higher adherence was reported among patients with DA.²⁰ Finally, in a real-world analysis comparing patients with vs without a DA after an adverse event, patients with a DA were found to have longer TTNT than those without DA.²¹ 

Although real-world data on DA is starting to emerge, a formal comparison in an overall ibrutinib population (ie, regardless of adverse events), covering a diversified population, and addressing the methodological challenges associated with defining treatment strategies after the initiation of treatment is lacking. The current study aimed to fill this gap by comparing TTNT in patients with CLL/SLL initiating 1L single-agent ibrutinib at 420 mg/d and staying on this dose vs those having a DA within 3 to 12 months, using 2 distinct real-world databases covering a large and diversified population treated in community practices as well as teaching and nonteaching hospital systems.

To maximize generalizability and robustness of results, analyses were conducted in 2 databases: the Komodo Health payer-complete and Acentrus EMR databases.

The Komodo database (1 January 2015 to 30 April 2023) contains closed health insurance claims data from >140 million individuals in the US. The data set is derived from >150 private insurers and includes patients with commercial, individual, state exchange-purchased, Medicare Advantage, and Medicaid managed-care insurance coverage, and is geographically representative of the insured US population treated primarily in community practices. The data include comprehensive information on patient demographics, diagnoses, procedures, place of service, and medication use (eg, days of supply and quantity).

Acentrus EMR data from 1 January 2016 to 30 April 2022 were also used. Acentrus is a health system solution used by 128 000 prescribers, containing inpatient and outpatient data from 27 sites, including 10 National Cancer Institute–designated sites and 6 National Comprehensive Cancer Network members. It includes records of patients from 15 academic and 12 nonteaching hospital systems across 15 US states (ie, Arizona, California, Florida, Kansas, Massachusetts, Missouri, North Carolina, North Dakota, New York, Ohio, Tennessee, Texas, Virginia, Washington, and West Virginia). In addition to patients’ demographic and clinical characteristics (including diagnoses), the database provides information on medication orders, fills, and administrations (including dosing).

Data in the Komodo and Acentrus databases are deidentified and comply with requirements for protection of patient privacy under the Health Insurance Portability and Accountability Act.

A retrospective cohort study design emulating a target trial was separately implemented in the Komodo and Acentrus databases. The index date was defined as the date of initiation of single-agent ibrutinib in 1L at a dose of 420 mg/d. In Komodo, the daily dose of ibrutinib was calculated based on days of supply and quantity information available in pharmacy claims; in Acentrus, the daily dose was calculated based on prescription information, dosing quantity, and dosing frequency information available for fills, orders, and administrations.

A washout period of ≥12 months without use of any antineoplastic agents before the index date was required to confirm ibrutinib use in 1L therapy (a definition commonly used in other real-world studies).^22-24 The 12-month period before the index date was also used to establish the baseline period for the evaluation of baseline demographic and clinical characteristics (Figure 1). After the index date, a window of 28 days was used to confirm that no other antineoplastic agents were used and that ibrutinib was not used as part of a combination therapy.

The following 2 treatment strategies were compared: (1) treatment for 3 months after the index date at 420 mg/d on single-agent ibrutinib followed by DA to a dose <420 mg/d occurring between 3 and 12 months after the index date, (2) sustained treatment for ≥12 months after the index date at 420 mg/d on single-agent ibrutinib. A minimum of 3 months with the standard approved ibrutinib dose of 420 mg/d before DA was required based on evidence from pharmacokinetic/pharmacodynamic modeling suggesting that starting at a higher dose intensity is critical to achieving optimal outcomes.²⁵ 

TTNT was evaluated over the period spanning from the index date until administrative censoring based on the earliest of the initiation of a next treatment, DA outside of 3 to 12 months, treatment discontinuation, participation in a clinical trial, end of continuous enrollment (only available in Komodo), death (only available in Acentrus), or end of data availability. Treatment discontinuation was defined as having a treatment gap of ≥365 days between the last day of ibrutinib supply and the date of the next ibrutinib fill or the end of follow-up.

Because TTNT was used as a proxy for disease progression, administrative censoring was also applied to both cohorts if 1 of the following events occurred: (1) within-class BTKi switch (ie, the next treatment was also a BTKi), with patients censored at the time of switch, and (2) anti-CD20 antibody or venetoclax add-on within 6 months after the index date, with patients censored at the time of add-on. Beyond the first 6 months after the index date, anti-CD20 or venetoclax add-on was considered as a next treatment. Patients with a within-class BTKi switch were censored because switching may have been due to tolerability rather than progression of disease.^26-29 Patients with an anti-CD20 or venetoclax add-on were censored within the first 6 months because this may have indicated late initiation of 1L combination therapy regimen rather than disease progression. The 6-month timeframe was selected based on recommendations from prescribing guidelines regarding the potential initiation of these additional agents (cycles 1-2 for anti-CD20, or cycles 3-4 for venetoclax, allowing until 6 months after the index date if there are delays in the initiation of the combination agent).^30-33 

Patients were required to meet the following inclusion criteria: ≥2 diagnoses for CLL/SLL ≥30 days apart, including 1 diagnosis before the index date; ≥1 order, fill, or administration for ibrutinib; ≥12 months of continuous enrollment (Komodo) or data availability (Acentrus) before the index date (to confirm use of ibrutinib in 1L); ≥28 days of data availability after the index date (to confirm single-agent treatment with ibrutinib); and age of ≥18 years as of the index date. Patients with other cancer diagnoses before the index date were excluded from analyses (see Figure 2 for details).

TTNT was defined as the time between the initiation of 1L ibrutinib and the initiation of the next treatment or the addition of another treatment to the regimen. The addition of another treatment had to be beyond the first 6 months for anti-CD20 and venetoclax, or beyond the first 28 days for other non-BTKi agents used to treat CLL/SLL.

All statistical analyses were conducted separately for patients in the Komodo and Acentrus cohorts. To account for immortal time bias in analyses (ie, patients with DA are required to survive on 1L therapy until DA) as well as overlap between follow-up time and definition of treatment strategies, a target trial emulation approach was used (as summarized in supplemental Table 1) in which all patients were cloned at the index date and contributed follow-up to both treatment strategy arms.^34-36 Each clone was assigned to 1 treatment strategy, such that all patients contributed 1 clone to each of the 2 treatment strategies of interest: (1) DA, and (2) no DA. Follow-up time for a given clone ended when the patient’s observed treatment path deviated from the assigned treatment strategy (ie, artificial censoring; see supplemental Figure 1 for additional details on patient cloning).

To account for the artificial censoring introduced by patient cloning, inverse probability of censoring weights (IPCW) was used, in which uncensored patients were upweighted to represent censored patients with similar characteristics (thus mitigating potential informative censoring bias).³⁷ Weights were calculated as the inverse probability of remaining uncensored for each 30-day interval from the index date until the end of the follow-up period, given baseline characteristics and time-varying confounders, using a pooled logistic regression model.^38,39 

TTNT between DA and no DA treatment strategies was compared using IPCW-weighted pooled logistic regression models, in which the odds ratios were used to approximate the hazard ratios (HRs).⁴⁰ Robust variance estimation was used to calculate 95% confidence intervals (CIs) for HRs (see supplemental Table 2 for the list of covariates used in the models).

Predicted IPCW-weighted Kaplan-Meier survival curves were generated to describe TTNT for each treatment strategy by calculating the predicted probabilities of survival from the IPCW-weighted pooled logistic regression models (see supplemental Table 2 for more details on the calculation of predicted survival probabilities).

Sensitivity analyses using different time periods to capture DA (3-6 and 3-9 months) were also conducted to evaluate the robustness of results. Results were also replicated in the subgroup of patients classified as being at high risk for a cardiovascular (CV) event, based on the presence of preexisting CV conditions or a high CV risk score (see supplemental Table 2 for the list of conditions and CV risk scores considered).

In Komodo, 6464 patients met all inclusion criteria related to being an adult with CLL/SLL initiating ibrutinib. Of these patients, 3731 (57.7%) remained in the study after applying the exclusion criteria, of which 3342 (89.6%) initiated 1L single-agent ibrutinib with a starting dose of 420 mg/d (Figure 2). In Acentrus, 2276 patients met all inclusion criteria related to being an adult with CLL/SLL initiating ibrutinib. Of these patients, 2255 (99.1%) remained in the study after applying the exclusion criteria, of which 1171 (51.9%) initiated 1L single-agent ibrutinib with a starting dose of 420 mg/d (Figure 2).

In Komodo, the mean age was 67.5 years, with 37.6% being female. The mean baseline Quan-Charlson Comorbidity Index (Quan-CCI) score was 3.0, 10.5% had baseline valvular disease, 6.5% had baseline renal impairment/dialysis, 7.6% had baseline atrial fibrillation, and 79.0% of patients were classified as being at high risk for a CV event. Baseline characteristics in the high CV risk subgroup were comparable with those of the overall cohort; for instance, the mean age was 69.4 years, 37.6% were female, and the mean Quan-CCI was 3.1 (Table 1). In Acentrus, the mean age was 70.4 years, with 34.6% being female. The mean Quan-CCI score was 3.0, 5.0% had baseline valvular disease, 4.0% had baseline renal impairment/dialysis, 7.3% had baseline atrial fibrillation, and 61.8% of patients were classified as being at high risk for a CV event. In the high CV risk subgroup, characteristics were also generally similar (eg, 33.0% female), except mean age (75.0 years) and mean Quan-CCI (3.5), which were higher than in the overall cohort (Table 1).

In Komodo, over a median follow-up of 19.6 months, 608 (18.2%) patients had a DA after the index date (519 [85.4%] adjusted their dose to 280 mg/d), with a mean (median) time to DA of 9.9 (10.7) months. Of these patients, 281 (46.2%) had a DA between 3 and 12 months after the index date (3-6 months: 141 [23.2%] patients; 3-9 months: 218 [35.9%] patients; Table 2). In Acentrus, over a median follow-up of 30.9 months, 229 (19.6%) patients had a DA after the index date (178 [77.7%] adjusted their dose to 280 mg/d), with a mean (median) time to DA of 9.0 (5.5) months. Of these patients, 111 (48.5%) had a DA between 3 and 12 months after the index date (3-6 months: 66 [28.8%] patients; 3-9 months: 87 [38.0%] patients; Table 2).

In Komodo, a total of 463 (13.9%) patients initiated a next treatment, of whom 198 (42.8%) initiated a venetoclax-based regimen as their next treatment. Other common second-line regimens included anti-CD20 monotherapy (n = 94 [20.3%]), chemoimmunotherapy (n = 76 [16.4%]), and BTKi + anti-CD20 combination therapies (n = 35 [7.6%]). In Acentrus, a total of 90 (7.7%) patients initiated a next treatment, of whom 59 (65.6%) initiated a venetoclax-based regimen as their next treatment. Other common second-line regimens included anti-CD20 monotherapy (n = 14 [15.6%]), chemoimmunotherapy (n = 8 [8.9%]), and BTKi + anti-CD20 combination therapies (n = 2 [2.2%]).

In both databases, following adjustment in IPCW-weighted pooled regression models, DA was not associated with an increased risk of having a next treatment (adjusted HR in Komodo: 0.95 [95% CI, 0.80-1.14]; P = .61; adjusted HR in Acentrus: 1.14 [95% CI, 0.80-1.62]; P = .47), and predicted survival curves were similar between DA and no DA treatment strategies (Figure 3).

In sensitivity analyses using different time periods of 3 to 6 and 3 to 9 months to evaluate DA, results remained nonsignificant. When considering a DA period of 3 to 6 months, the HR was 0.95 (95% CI, 0.76-1.18; P = .64) in Komodo, and 1.30 (95% CI, 0.86-1.97; P = .21) in Acentrus. When considering a DA period of 3 to 9 months, the HR was 0.99 (95% CI, 0.82-1.21; P = .96) in Komodo and 1.14 (95% CI, 0.78-1.67; P = .51) in Acentrus (Figure 4). Results were also consistent and nonsignificant in the subgroup of patients at high risk for a CV event. In this subgroup, when considering a DA period of 3 to 12 months, the HR was 0.93 (95% CI, 0.76-1.13; P = .47) in Komodo, and 1.12 (95% CI, 0.72-1.73; P = .63) in Acentrus. When considering a DA period of 3 to 6 months, the HR was 0.94 (95% CI, 0.74-1.19, P = .59) in Komodo, and 1.26 (95% CI, 0.76-2.11; P = .37) in Acentrus. When considering a DA period of 3 to 9 months, the HR was 0.99 (95% CI, 0.76-1.22; P = .89) in Komodo and 1.01 (95% CI, 0.63-1.61; P = .96) in Acentrus (Figure 4).

This was a large real-world analysis of a diversified population of patients with CLL/SLL in 2 separate databases covering patients with various types of insurance and treated in various settings such as community practices, as well as academic and nonteaching hospital systems. In this analysis, a small proportion of patients required DA (18.2% in Komodo, and 19.6% in Acentrus), indicating that ibrutinib is well tolerated in the real world. This is consistent with the proportion of patients with DA reported in a post hoc analysis of the RESONATE-2 clinical trial (∼20%)¹⁰ and several real-world studies in 1L and relapsed/refractory settings.^9,21,41,42 

Furthermore, this study showed that undergoing DA within 3 to 12 months of initiating 1L ibrutinib yielded similar clinical outcomes as having no DA, that is, there were no significant differences in TTNT between DA and no DA strategies in both the Komodo and Acentrus databases. These findings were also consistent for the different time periods used to capture DA (ie, 3-6 and 3-9 months) and in the high CV risk subgroup. This study also meets an important need for a formal real-world comparison of outcomes among patients with DA vs patients without DA using statistical adjustment that addresses potential biases associated with immortal time bias and overlap between the identification of patient cohorts and follow-up. Furthermore, although differences between the 2 databases can be explained as a function of the diverse patient population as well as data availability (eg, longer TTNT in Acentrus vs Komodo may be explained by the academic vs community setting, earlier years of data availability in Komodo, and the lower proportion of patients with high CV risk in Acentrus), the consistency of findings across 2 distinct databases, including in sensitivity and subgroup analyses, is a strength of the study and speaks to the external validity of the results given the representation from both community and academic/nonteaching hospital systems.

The findings of this study are consistent with the literature. The use of DA in continuous ibrutinib treatment among patients with CLL/SLL is based on the finding that >95% occupancy of the BTK target occurs at doses lower than the original recommended dose of 420 mg/d, allowing maintenance of treatment with continued clinical benefit and fewer off-target effects.^25,43 The long-term benefits of DA to maintain ibrutinib treatment and manage adverse events were demonstrated in a follow-up study of patients in the phase 3 RESONATE-2 study.¹⁰ In that study, 79 patients were treated with ibrutinib for ≥5 years; 16 patients (20%) had a DA because of adverse events and neither median PFS nor OS were significantly different between patients with or without DA. The study findings were also consistent with results from other large real-world studies that have compared clinical outcomes in ibrutinib-treated patients with CLL/SLL, both in terms of the proportion of patients with DA and the outcomes observed (ie, no difference between DA and no DA).^9,41,42 In 1 study of 197 ibrutinib-treated US patients with CLL (81% with relapsed-refractory disease; 19% in 1L), 19% had a DA, and overall response rates, OS, and PFS were not significantly different between DA and no DA.⁴¹ A study of 315 patients in the United Kingdom with relapsed-refractory CLL treated with ibrutinib also reported that there were no detrimental effects on OS in the 26% of patients with DA.⁴² In a more recent study of 70 US patients with CLL, 31.3% of patients had a DA, with no significant differences in overall response rates, median PFS, and OS between DA and no DA treatment strategies.⁹ This study also found that of patients initiating a second-line regimen, venetoclax-based regimens were the most common. This is in line with trial evidence supporting its use in the relapsed/refractory setting^44,45 as well as a previous real-world study, which reported that among patients treated with venetoclax in second or later lines, 68.8% of patients were treated with ibrutinib in the prior line.⁴⁶ Overall, this study adds to the literature by addressing study design and methodological considerations such as immortal time bias and time-varying confounders, which, to our knowledge, have not been previously implemented in other real-world studies, in addition to covering a large diversified population.

In addition to having no impact on TTNT, ibrutinib DA can be associated with other benefits. First, DA can help patients remain on treatment, thus improving treatment adherence. For instance, a previous study has shown that patients with DA had higher adherence than patients with no DA.²⁰ Adherence can be critical in achieving optimal outcomes, as has been shown in previous studies.^47-49 In addition to the once-daily frequency of administration of ibrutinib, guidelines for DA are more comprehensive than for other BTKis and there is flexibility in the dosage form (ie, pills or capsules), all of which may have helped improve adherence for ibrutinib. These elements may have also contributed to the higher adherence observed for ibrutinib compared with twice-daily acalabrutinib, as reported in a previous study.⁵⁰ Ibrutinib DA may also play a role in mitigating the risk of treatment resistance,¹¹ a topic increasingly discussed, particularly for some BTKis,⁵¹ as well as reducing the risk of adverse events.¹² In turn, a reduction in the risk of adverse events has been shown to lead to lower inpatient admissions and costs in patients treated for CLL/SLL,²⁴ which may help improve quality of life.⁵² This has been confirmed in other studies as well, in which, in addition to lower drug costs, ibrutinib DA has been shown to reduce the number of inpatient admissions and other services received, resulting in lower health care costs.^21,53 

This study is subject to some limitations. Komodo claims and Acentrus EMR data may contain omissions and inaccuracies; however, this is expected to apply to all patients regardless of the treatment strategy, and thus would have minimal impact on overarching conclusions. Acentrus is a provider-based data source; therefore, records are only available to the extent that visits are part of the network of academic and nonteaching hospital systems included in the data. In addition, disease and prognostic characteristics were not available in both databases.

This study used a washout period of 12 months to identify the use of ibrutinib in 1L therapy. Although this may have resulted in capturing patients in longer remission who had previously received a line of therapy more than 12 months before the index date, this washout period has been used in other real-world studies and would make it very unlikely for patients to have received prior ibrutinib or other BTKi treatments.^22-24 Prescriptions for ibrutinib were assumed to indicate their use; however, patients may not always adhere to the regimen as prescribed. In addition, reasons for DA or next treatment were not available in both databases, but not having this information does not alter the results of this study and instead demonstrates the comprehensive evaluation of DA regardless of the reason (eg, effectiveness, safety, or physician/patient request), with consistent findings across 2 distinct databases, showing that outcomes are comparable regardless of DA. Despite the use of cloning and IPCW, residual confounding due to changes in unobserved confounders over time may remain. Finally, although this study supports the use of ibrutinib DA in the treatment of CLL/SLL, it is not clear whether the results can be extended to other BTKi therapies available for CLL/SLL because prescribing guidelines for other BTKis are different.^54,55 

In conclusion, this large real-world comparative analysis of patients with CLL/SLL from community practices as well as academic and nonteaching hospital systems showed that a small proportion of patients had DA of 1L ibrutinib, and that undergoing DA yielded similar clinical outcomes compared with having no DA, with similar findings for patients at high risk for a CV event. This study findings, replicated across 2 separate large real-world databases, along with those from clinical trials and other real-world community oncology practices, suggest that a flexible dosing approach with ibrutinib is effective in allowing patients to achieve optimal outcomes while remaining on long-term continuous 1L treatment.

The authors thank Priyanka Gogna, an employee of Analysis Group, Inc, a consulting company that has received research funding from Janssen Scientific Affairs, LLC for the conduct of this study, for her contributions to the design and conduct of the study. Medical writing assistance was provided by Christopher Crotty, a former employee of Analysis Group, Inc.

This study was funded by Janssen Scientific Affairs, LLC, a Johnson & Johnson company.

The study sponsor was involved in several aspects of the research, including the study design, interpretation of data, writing of the manuscript, and decision to submit the manuscript for publication.

Contribution: All authors have provided final approval of this version to be published and agree to be accountable for all aspects of the work; N.G. contributed to interpretation of results and writing (review and editing) of the manuscript; R.W. contributed to conceptualization, methodology, formal analysis, investigation, visualization, interpretation of results, and writing (review and editing) of the manuscript; Z.P.Q. contributed to conceptualization, methodology, interpretation of results, project administration, and writing (review and editing) of the manuscript; Z.D. contributed to conceptualization, methodology, interpretation of results, project administration, writing (review and editing) of the manuscript; M.-H.L. contributed to conceptualization, methodology, formal analysis, visualization, interpretation of results, and writing (review and editing) of the manuscript; B.E. contributed to conceptualization, methodology, formal analysis, visualization, writing (review and editing) of the manuscript; B.M. contributed to conceptualization, methodology, formal analysis, visualization, and writing (review and editing) of the manuscript; J.H. contributed to conceptualization, methodology, writing (review and editing) of the manuscript, and project administration; A.B. contributed to interpretation of results and writing (review and editing) of the manuscript; H.M. contributed to interpretation of results and writing (review and editing) of the manuscript; and K.A.R. contributed to interpretation of results and writing (review and editing) of the manuscript.

Conflict-of-interest disclosure: M.-H.L., B.E., and B.M. are employees of Analysis Group, Inc, which has received research funding from Janssen Scientific Affairs, LLC. R.W., Z.P.Q., Z.D., J.H., A.B., and H.M. are employees of Janssen Scientific Affairs, LLC, and stockholders of Johnson & Johnson. N.G. reports consultancy for Seagen, TG Therapeutics, AstraZeneca, Pharmacyclics, Janssen, Bristol Myers Squibb, Gilead Sciences, Kite Pharma, BeiGene, Incyte, Lava Therapeutics, Incyte, Roche/Genentech, Novartis, Loxo Oncology, AbbVie, Genmab, Adaptive Biotech, and ADC Therapeutics; research funding from TG Therapeutics, Roche/Genentech, Bristol Myers Squibb, Gilead, MorphoSys, AbbVie, and Pharmacyclics; serving as a member of the speakers bureau of AstraZeneca, Janssen, Pharmacyclics, Kite Pharma, Bristol Myers Squibb, and Epizyme; and serving as an advisory committee member for the Roche NHL Solutions Panel. K.A.R. reports consultancy for AbbVie, Genentech, BeiGene, AstraZeneca, Janssen, Loxo@Lily, and Pharmacyclics; and research funding from AbbVie, Genentech, and Novartis.

Correspondence: Bruno Emond, Analysis Group, Inc, 1190 Avenue des Canadiens-de-Montréal, Tour Deloitte, Suite 1500, Montréal, QC H3B 0G7, Canada; email: bruno.emond@analysisgroup.com.