Ibrutinib increases the risk of hypertension and atrial fibrillation: Systematic review and meta-analysis

1Ibrutinib is the first-in-class oral covalent inhibitor of Bruton's tyrosine kinase that has been approved for the treatment of patients with chronic lymphocytic leukemia (CLL), mantle cell lymphoma (MCL) and Waldenstrӧm’s macroglobulinemia (WM) due to its efficacy. Despite the significant favourable impact in the hematologic conditions, Ibrutinib increases significantly the risk of atrial fibrillation (AF). The mechanisms leading to AF are not well established, but it is known that arterial hypertension is associated with increased risk of this dysrhythmia. Yet, the safety data from ibrutinib ‘s trials regarding reported adverse events of hypertension is heterogenous.

2Therefore, we performed a systematic review of randomized controlled trials to evaluate the impact of ibrutinib in the incidence of reported hypertension and atrial fibrillation on randomized controlled trials, irrespective of the comparators (active or placebo) or population.

3In this study, we performed systematic review with meta-analysis which is a well-known method to evaluate specific safety aspects of drugs using the cumulative evidence from clinical trials.

4We made an electronic database search through MEDLINE, EMBASE, Central Register of Controlled Trials (CENTRAL) and ClinicalTrials.gov in August 2018 and updated in December 2018 using standardized methods We also performed extensive hand searching by screening references of included studies and review articles for additional citations.

5We considered eligible all randomized controlled trials (RCTs) comparing ibrutinib with any control group (placebo, no-treatment or standard care, non-pharmacological interventions or any active drug). All RCTs were considered for inclusion irrespective of patients’ baseline conditions, background therapy, ibrutinib dose, study follow-up or language of publication.

6The primary outcomes were the incidence of hypertension and atrial fibrillation. For both outcomes we used a broad and lenient definition of the conditions. Hypertension was defined as blood pressure increase reported by investigators as an adverse event (or serious adverse event). AF was defined as a rhythm disorder characterized by the presence of irregular RR intervals and no discernible, distinct P waves during at least 30 seconds by convention, or AF reported by investigators as an adverse event. Whenever possible the adverse events were reported according to the Common Terminology Criteria for Adverse Events (CTCAE).

7We considered reasonable our lenient inclusion criteria as ibrutinib was only studied in a few hematologic conditions (CLL, MCL, WM) and it is not expected that the risk for hypertension or atrial fibrillation is different amongst diseases. Furthermore, our inclusive criteria increase the power of findings.

8The records retrieved through electronic database search were screened independently by 2 authors. Suitable studies were evaluated for the inclusion in the review through full-text assessment. Study selection and data extraction were performed independently. If different data were available for the same trial, we considered the most recent report or the updated data from ClinicalTrials.gov.

9Study characteristics and results were extracted independently into a standardized form. Whenever possible modified intention-to-treat (patients randomized and treated at least once with the allocated drug) data was extracted for analysis.

10Risk of bias was evaluated through the Cochrane Risk of Bias Tool. Disagreements throughout this process were resolved by consensus.

11The incidence of hypertension or atrial fibrillation was treated as a dichotomous data and Risk Ratio (RR) was estimated. The 95% confidence interval (95%CI) was used to estimate the precision of pooled results from studies.

12The data analysis was performed through the Hartung-Knapp-Sidik-Jonkman (HKSJ) method in using R 3.5.2, OpenMetaAnalyst and OpenMEE. A secondary analysis was performed using the Mantel-Haenszel method and random effects models through RevMan version 5.3 (The Nordic Cochrane Centre, Copenhagen; The Cochrane Collaboration, 2014). A fixed 0.5 correction was added when one arm presented zero-events to avoid computational problems. Heterogeneity was assessed using the Chi square test and I2 statistic. The I2 statistics measures the percentage of total variation between studies attributed to interstudy heterogeneity rather than random, and we used the Sidik-Jonkman estimator to derive tau2 and subsequently I2. Statistical heterogeneity was considered substantial if I2> 50%.

13The 95% prediction intervals were estimated to assess dispersion of the effect size in different setting, deriving whether true effects are to be expected for 95% of similar newly conducted study. The 95% prediction intervals were put into perspective with the results of pooled analyses, using blue rectangles in a plot generated by RevMan.

14We performed a subgroup analysis according to the type of control group (active or placebo). The impact of follow-up time / drug exposure in the risks of hypertension or atrial fibrillation was also ascertained through Hartung-Knapp method for meta-regression using follow-up time (months) as a covariate.

15Reporting/Publication bias tests for funnel plot asymmetry were only used if a minimum of 10 studies were included in the meta-analysis.

16As recommended by the Grading of Recommendations Assessment, Development and Evaluation (GRADE) Working Group methodology, two reviewers independently assessed all the critical outcomes in the following domains: risk of bias, inconsistency, indirectness, imprecision and publication bias.

17We found 8 eligible RCTs with 2580 patients (54.7% treated with ibrutinib) (Fig 1). The trial with the smaller sample size had 150 patients (iNNOVATE) and the largest had 578 patients (HELIOS). All the trials, with exception of the RAY trial had an open-label design (Fig 2). There were 2 placebo-controlled trials and 6 active controlled studies. None of the trials was designed for systematically search for hypertension or atrial fibrillation (high risk of selective reporting bias) (Fig 2). All the trials allowed crossovers from control to ibrutinib arm, with rates reaching 40% as occurred in iNNOVATE study. Only the RESONATE-2 trial did not allow crossovers during the study (low risk of bias in ‘Other bias’ in Fig 2). The main characteristics of the included studies are depicted in Table 1 and the risk of bias plot in Fig 2 and Figure A in S1 File. Blood pressure measurements frequency was reported in half of the studies, only 3 studies reported the when ECG were performed (Table 1). About 75% of the studies reported the adverse events using CTCAE (Table 1).

18The pooled analysis included 8 studies. Half of the studies, individually, did not reach the statistical significance.

19Overall the pooled estimate showed that ibrutinib increased significantly the risk of hypertension with a RR of 2.82 (95%CI 1.52–5.23; p-value <0.001) (Fig 3 and Fig 4) and substantial statistical heterogeneity was observed (tau2 = 0.461; I2 = 66%). The 95% prediction interval ranged from 0.45 to 17.60 (Fig 4).

20The small number of studies found (<10) preclude the use of reporting/publication bias tests and funnel plot interpretation (Figure B in S1 File).

21The pooled analysis included 8 studies. Four of them showing significant increases in the risk of atrial fibrillation. The pooled risk ratio showed more than 4-fold increase in the risk of AF with ibrutinib (RR = 4.69, 95%CI 2.17–7.64; p-value <0.001) (Fig 4 and Fig 5). The 95% prediction interval ranged from 1.13 to 19.43 (Fig 4).

22The analysis was unremarkable for substantial statistical heterogeneity (tau2 = 0.216; I2 = 22.8%). The Figure C in S1 File shows the funnel plot for this outcome, but the reporting/publication bias tests and plot interpretation were not formally performed due to small number of studies (<10).

23The subgroup analyses according with the control group (active control / placebo) did not show any significant interactions with risk estimates (Figure D in S1 File).

24The meta-regression using follow-time as a covariate did not show a significant relationship between hypertension risk (p = 0.93) nor atrial fibrillation risk (p = 0.158) with ibrutinib exposure time (Figure E and Figure F in S1 File). The meta-analyses using the Mantel-Haenszel method and random effects models is shown in Figure G in S1 File, and the results are globally similar to the primary analyses.

25We used the GRADE approach to ascertain the confidence of the obtained pooled results. Both hypertension and atrial fibrillation were analysed through the same trials. Overall, we considered that the results may have been influenced by performance bias and selective reporting bias (Fig 2). Even though the results were consistent, and the estimates were large for hypertension and very large for atrial fibrillation. This determined a moderate and high certainty in the pooled evidence for hypertension and atrial fibrillation risk with ibrutinib, respectively. The GRADE evaluation with absolute and relative risk increases are detailed in Table 2.

26In this systematic review with meta-analysis, we showed that ibrutinib was significantly associated with an increased risk of hypertension in randomized controlled trials. Hypertension was considered a common adverse event of ibrutinib in the Summary of Product Characteristics of ibrutinib, but the data was solely derived from the rates of reporting in clinical studies or during post marketing surveillance without a formal comparison with a control arm. In this systematic review we also confirmed with new published trials that ibrutinib was associated with atrial fibrillation.

27The major contributions for the current knowledge are: 1) the confirmation of hypertension as adverse drug reaction of ibrutinib; 2) the updated meta-analysis showing the association between ibrutinib and atrial fibrillation with more trials (8 RCTs) and more precision and confidence on the estimate compared with other published meta-analyses; 3) The use of prediction intervals to determine that atrial fibrillation risk is likely to be increased but hypertension risk may not be increased in new studies, taking into account the dispersion of data; 4) the use of GRADE approach to include all aspects of the trials and analyses to ascertain the confidence on the new data supports that the confidence of AF increased risk is high and moderate for increase of hypertension risk.

28As a secondary effect of ibrutinib, hypertension needs to be treated to mitigate cardiovascular consequences, such as atrial fibrillation, myocardial infarction or stroke.

29One of the reasons for this study was the known link between hypertension and atrial fibrillation. Despite the heterogeneous results of studies evaluating the association of hypertension and AF in patients treated with ibrutinib, the link of hypertension and atrial fibrillation is well established. Being hypertension a risk factor for AF, an adequate diagnosis and treatment of this risk factor may be beneficial as hypertension and AF are linked to stroke. The risk ratios of hypertension and atrial fibrillation associated with ibrutinib were different (RR 2.82 hypertension, RR 4.69 for atrial fibrillation) and preclude an absence of direct association. The absolute risk (Table 2) shows that ibrutinib increases risk of hypertension in 7.8% (12 new cases for 100 patients treated) and atrial fibrillation in 4.1% (5 new cases for 100 patients treated). This paradoxus (lower RR and higher absolute risk for hypertension compared with AF) is explained by the higher prevalence and reported adverse event of hypertension compared with atrial fibrillation.

30Beyond the putative role of hypertension-linked AF, the inhibition of the PI3K-Akt pathway by ibrutinib may also play a pathogenic role in the development of AF, but it is still controversial.

31One of the most important limitations is that none of the included studies was designed to detect / adjudicate incident or worsening of hypertension and atrial fibrillation. The confidence on the current evidence is at least moderate, and the future trials with ibrutinib will surely acknowledge the blood pressure and atrial fibrillation issues. At outcome level, the subjective nature of hypertension reporting by investigators should be noticed as at high-risk of selective reporting bias. The diagnosis of AF is more objective, but it also depends on clinical manifestations, and it is known that AF can be clinically silent (until the embolic consequences) and/or paroxysmal, which may impair an accurate reporting.

32The concomitant increase in the risks of AF and hypertension, together with the increased bleeding risk of ibrutinib is a clinical challenge. In patients with AF and one thromboembolic risk factor (such as hypertension), the stroke risk is considered non-negligible and the guidelines say that anticoagulation should be considered. Thus, AF with ibrutinib, thromboembolic risk factors (hypertension) and bleeding risk factors (antiplatelet effect of ibrutinib and anticoagulation) may pose difficulties to the clinicians.

33This systematic review was not able to capture differences the clinical consequences and severity of hypertension and atrial fibrillation, that vary widely. An unanswered relevant question for the clinical practice is the degree of blood pressure increase induced by ibrutinib and its relevance for the development of other consequences.

34Our results are important for clinicians and investigators as they emphasize that other cardiovascular events / risk factors associated with ibrutinib, such as arterial hypertension, should be targeted in further investigations.

35The efficacy of ibrutinib poses new questions as these data are derived from the trials reveal an increase of patients’ survival. Studies with longer follow-up are required to better characterize the risk at longer term, as well as other clinical aspects such as the carry-over effect, the durability of the adverse events, and the management of hypertension and/or atrial fibrillation with different strategies.

36Results and conclusion here presented are weakened by limitations inherent to meta-analysis and individual studies.

37The higher risk of bias was found for potential selective reporting of the outcomes, i.e. the outcomes were not primary and actively searched, and were reported at the discretion of the investigator. The results might have been also influenced by the risk of performance bias (lack of blinding) which limits the association of the adverse events with ibrutinib. A limitation is that there is not a single RCT primarily designed to evaluate hypertension and atrial fibrillation as primary outcomes. Nonetheless, the degree of confidence on the current evidence is moderate/high.

38The non-negligible rates of crossovers from control arms to ibrutinib (when disease progression occurred in control arm) may limit our analysis. However, the analysis of the data as intention-to-treat increases the confidence in our results as some control patients could have developed hypertension and/or atrial fibrillation after crossing to the ibrutinib arm and were analysed as control. Thus, despite this conservative approach, our results were significant.

39Our results suggest that ibrutinib increases significantly the risks of hypertension and atrial fibrillation. The pooled data for hypertension had moderate quality evidence, and high-quality evidence for atrial fibrillation. Whether hypertension is a crucial factor for ibrutinib-induced atrial fibrillation is still unknown.