Association of Nilotinib With Cardiovascular Diseases in Patients With Chronic Myelogenous Leukemia: A National Population-Based Cohort Study

Abstract

Background

Tyrosine kinase inhibitor (TKI) treatment has been identified to be a risk factor for metabolic syndrome and cardiovascular diseases (CVDs) in patients diagnosed with chronic myeloid leukemia (CML). However, the specific contribution of post-TKI metabolic syndrome and the individual TKIs, including imatinib, nilotinib, and dasatinib, contribute to the development of CVDs remains unclear.

Methods

We conducted a nationwide database to investigate the incidence of post-TKI metabolic syndrome, including diabetes, hyperlipidemia, and hypertension, as well as their association with CVDs. To compare the risk of post-TKI comorbidities and CVDs among TKIs, we utilized the incidence rate ratio (IRR), and subdistribution hazard ratio (SHR) calculated from multiple Fine-Gray models.

Results

A total of 1211 patients without diabetes, 1235 patients without hyperlipidemia, and 1074 patients without hypertension were enrolled in the study. The incidence rate of post-TKI diabetes and hyperlipidemia was the highest in patients treated with nilotinib compared to imatinib and dasatinib (IRRs ≥ 3.15, Ps ≤ .047). After adjusting for confounders, nilotinib remained a significant risk factor for post-TKI diabetes and hyperlipidemia at an SHR of 3.83 (P < .001) and 5.15 (P < .001), respectively. Regarding the occurrence of CVDs, patients treated with nilotinib were more likely to develop CVDs than those treated with imatinib in non-hyperlipidemic group (IRR = 3.21, P = .020). Pre-existing and post-TKI hyperlipidemia were found to have a stronger association with CVDs, with SHR values of 5.81 (P = .034) and 13.21 (P = .001), respectively.

Conclusion

The findings of this study indicate that nilotinib treatment is associated with increased risks of diabetes and hyperlipidemia, with hyperlipidemia being the most significant risk for CVDs. Therefore, we recommend that CML patients receiving nilotinib should undergo screening for diabetes and hyperlipidemia prior to initiating TKI treatment. Additionally, regular monitoring of lipid profiles during TKI therapy and implementing effective management strategies to control hyperlipidemia are crucial.

Introduction

Chronic myeloid leukemia (CML) is a myeloproliferative malignancy associated with the presence of the fusion gene BCR-ABL.¹ The development of tyrosine kinase inhibitors (TKIs) has revolutionized the clinical management of patients with CML. Imatinib mesylate, the first-generation TKI for CML, has shown promising efficacy, with an 82.8% complete cytogenetic response and an 83.3% overall survival rate at 10-year follow-up.^2,3 Second-generation TKIs, nilotinib and dasatinib, have shown an even deeper molecular response (MR^4.5) rate than imatinib at 5-year follow-up (53.5% vs. 31.4%, and 42% vs. 33%, respectively).^4,5

After decades of TKI use for the treatment of CML, treatment-free trials have demonstrated the possibility of drug discontinuation. However, treatment-free remission rates ranged from 43% to 59% in patients with chronic-phase CML under restricted criteria of achieving a deep molecular response (MR^4.0) or better.^6-8 Consequently, a large proportion of patients with CML still require lifelong TKI therapy. Moreover, to achieve a deeper molecular response rate and undergo treatment-free remission, nilotinib and dasatinib are sometimes favored over imatinib due to their higher potency, especially in patients anticipating discontinuation of the drug.⁹ Therefore, when more patients receive second-generation TKIs, concerns regarding their long-term adverse effects arise.

The profile of adverse events varies among the 3 TKIs. For example, peripheral edema is associated with imatinib, peripheral arterial occlusion with nilotinib, and pleural effusion with dasatinib.^3-5 Prospective trials for second-generation TKIs have reported cardiovascular diseases (CVDs), including pulmonary hypertension, peripheral artery occlusion, myocardial infarction, and ischemic stroke.^10,11) Some real-world studies suggest that CVD incidence is significantly higher in the nilotinib group compared to the imatinib and dasatinib groups.^12-14 Furthermore, patients treated with nilotinib have higher incidences of hyperglycemia, hypercholesterolemia, and hypertension compared to those receiving imatinib.⁴ Since diabetes, hyperlipidemia, and hypertension are also known risk factors for CVDs,¹⁵ it is unclear whether the increased risk of CVDs observed in the nilotinib group is primarily due to TKI-induced metabolic comorbidities or nilotinib itself. To address this issue, we conducted this population-based study to explore the incidences of post-TKI diabetes, hyperlipidemia, and hypertension across different TKIs and further investigate the association between these comorbidities and the risk of CVDs.

Materials and Methods

Data Sources

This retrospective study utilized data from the Taiwan Cancer Registry (TCR) database, the National Health Insurance Research Database (NHIRD), and the Death Registry (DR) database. The TCR database is a population-based cancer registry that undergoes a rigorous review process by a panel of specialists who evaluate medical records, imaging, and pathology reports. Therefore, the TCR database is a reliable data source for identifying patients with CML and other malignant carcinomas, as it contains data that is directly relevant to the exemption of copayments for these vulnerable beneficiaries. The NHIRD is a nationwide database containing longitudinal medical records of beneficiaries enrolled in the National Health Insurance (NHI) program, which provides comprehensive healthcare coverage for over 99% of the Taiwanese population. The NHIRD provides detailed information on comorbidities and treatment modalities from 1995 to 2017 obtained from inpatient and outpatient databases. This study received approval from the Institutional Review Board of Chang Gung Medical Foundation (approval number: 202100255B1).

Study Population

The study cohort consisted of patients who were diagnosed with CML (ICD-O-3 codes: 98753, 98633) between July 1, 2004 and December 31, 2017, as identified from TCR database. Originally, there were 2260 patients with CML. However, 722 patients were excluded if they meet any one of the following criteria: being under the age of 20 at the time of CML diagnosis, development of other cancers before or after the CML diagnosis, not receiving any of the 3 studied TKIs (imatinib, nilotinib, and dasatinib), using the first TKI for <1 month, having received TKI earlier than 1 month before CML diagnosis date, having previous CVDs consistent with the definition of the secondary endpoint in this study, and having a history of atrial fibrillation or chronic kidney disease. Among the remaining 1538 patients, 3 groups were identified: non-diabetes group (n = 1211), non-hyperlipidemia group (n = 1235), and non-hypertensive group (n = 1074). For example, the subjects in the non-diabetes group had to meet a criterion of having no inpatient or outpatient diagnosis of diabetes mellitus before the index date, which was defined as the 30th day after the initiation of the first TKI use. Furthermore, all eligible patients in each group were further classified by the first TKI they received, namely imatinib, nilotinib, and dasatinib. The inclusion and exclusion criteria for each study group are described in Fig. 1.

Outcomes of Interest

The primary outcomes in this study were the occurrences of post-TKI comorbidities, such as newly diagnosed diabetes, hyperlipidemia, and hypertension after TKI use in non-diabetes, non-hyperlipidemia, and non-hypertensive groups, respectively. For each group, events were defined as the primary diagnosis observed during at least 3 hospital visits, including both inpatient and outpatient visits, during the follow-up period. The survival time for each group was defined as the time from the first day of TKI use to the new diagnosis of the respective comorbidity, or to the censored date during TKI follow-up. The TKI follow-up period was defined as the time from the index date to the earliest date of the following: switching to another treatment, TKI discontinuation, death, or the end of the study (December 31, 2017). Switching treatment was considered as switching to another TKI or other therapy, and patients were censored on the date of the next treatment initiation. Discontinuation was defined as a gap in TKI coverage of more than 60 consecutive days, and patients were censored on the day of the last prescription plus a 30-day lag period to account for any possible delayed drug effects.

The secondary outcomes of interest were CVDs, specifically ischemic stroke, coronary heart disease (CHD), and peripheral artery disease (PAD). The occurrence of CVDs was defined as any one of the following primary diagnoses: (1) hospitalization with a major or minor diagnosis of ischemic stroke, confirmed by computed tomography or magnetic resonance imaging; (2) hospitalization with a diagnosis of CHD, and receipt of cardiac catheterization, coronary artery bypass graft, or percutaneous transluminal coronary angioplasty; (3) at least one hospital admission with a diagnosis of PAD. All CML patients in the 3 groups were followed from the first date of initial TKI use until a diagnosis of CVD or any censoring event occurred during TKI follow-up. These events were defined by the ICD-9-CM and ICD-10-CM codes, which are summarized in Supplementary Table S1.

Covariates

Baseline characteristics included age at CML diagnosis, gender, and pre-existing comorbidities. The comorbidities examined in this study were diabetes, hyperlipidemia, and hypertension. Atrial fibrillation and chronic kidney disease were not included in the analysis due to the small number of subjects with these conditions and their strong association with CVDs, which could have confounded the results. The presence of the 3 comorbidities was identified using ICD-9-CM and ICD-10-CM diagnostic codes, as specified in Supplementary Table S1. Confirmation of these comorbidities required at least 3 outpatient visits or at least 1 hospitalization within the 12 months preceding the index date.

Statistical Analysis

Baseline characteristics among the 3 TKIs were compared using either one-way ANOVA or Chi-square test, whichever was appropriate. The incidence rates of post-TKI comorbidities and CVD were calculated by dividing the number of events by the total follow-up person-years, and were presented as the number of events per 1000 person-years. The rate ratio was used to compare the incidence of events between any 2 TKI treatments. The time from the first day of TKI initiation to the occurrence of post-TKI comorbidity event was analyzed using the subdistribution hazard model, proposed by Fine and Gray, which accounts for death as a competing risk.¹⁶ In addition, the Fine-Gray model with a time-varying definition of post-TKI comorbidity, which allowed patients to transfer from a status of comorbidity-free to comorbidity within TKI follow-up, was used to compare the time to CVD between patients with and without the occurrence of comorbidity. The results are presented as subdistribution hazard ratios (SHRs) with corresponding P values. All data management and statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA). All statistical tests were 2-sided, and a significance level of .05 was used.

Results

The Incidence of Post-TKI Diabetes and Hyperlipidemia Was More Common in Nilotinib Compared to Dasatinib and Imatinib

The study enrolled a total of 1211 patients without diabetes, 1235 patients without hyperlipidemia, and 1074 patients without hypertension. Overall, the demographic data regarding sex and age were not significantly different among the 3 TKIs in each comorbidity group, except for the non-hypertensive group where the gender distributions differed (P = .049). Additionally, the pre-existing comorbidities, types of comorbidities, and numbers of comorbidities were similar among 3 TKI therapy cohorts in all groups, with the majority of patients in each group having no other comorbidities (Table 1).

The study evaluated the association between different TKIs and the risk of developing new-onset diabetes, hyperlipidemia, and hypertension post-TKI treatment. This was achieved by calculating the incidence rate ratio (IRR) based on the discrepancy of crude incidence rates (Table 1). In the non-diabetes group, patients treated with imatinib, nilotinib, and dasatinib had a median follow-up time of 3.73, 2.08, and 2.02 years, respectively. The nilotinib group had a higher incidence (25.79 per 1000 person-years) of post-TKI new-onset diabetes than imatinib (8.20 per 1000 person-years) and dasatinib (5.18 per 1000 person-years), resulting in an IRR of 3.15 (nilotinib vs. imatinib, P < .001) and an IRR of 4.97 (nilotinib vs. dasatinib, P = .033). As for the risk of post-TKI new-onset hyperlipidemia in the non-hyperlipidemia group, patients treated with imatinib, nilotinib, and dasatinib had a median follow-up time of 4.06, 2.02, and 1.92 years, respectively. The patients in the nilotinib group were at a higher risk of post-TKI new-onset hyperlipidemia than the imatinib group (IRR = 4.47 with P < .001) and the dasatinib group (IRR = 3.50 with P = .047, respectively). In contrast, the risk of post-TKI hypertension was similar among the 3 TKI cohorts.

Nilotinib was the Only Significant Risk Factor for Post-TKI Newly Occurred Diabetes and Hyperlipidemia

In addition to comparing the crude incidence rates of each post-TKI comorbidity between different TKIs, we further examined the impact of TKI choice and common risk factors by using a multiple Fine-Gray model (Table 2). Notably, nilotinib was the only significant risk factor for newly occurring post-TKI diabetes and hyperlipidemia with a SHR of 3.83 (P < .001) and 5.15 (P < .001), respectively. In contrast, the occurrence of hypertension was significantly associated with older age (SHR = 3.51 with P < .001) and pre-existing DM (SHR = 2.54 with P = .048), irrespective of the choice of TKIs.

Nilotinib Increased the Risk of CVDs Regardless of Prior Comorbidities

The analysis of the time from the first date of TKI treatment to the occurrence of CVDs revealed the crude incidence rates of CVDs in patients without metabolic comorbidities prior to TKI use (Table 3). Although most of IRRs did not reach statistical significance due to small numbers of CVDs, certain trends were observed. In the non-diabetes group, patients treated with nilotinib had the highest incidence of CVDs (8.25 per 1000 person-years), followed by imatinib (3.27 per 1000 person-years) and dasatinib (2.57 per 1000 person-years). The IRR of CVDs in patients receiving nilotinib was 2.53 times higher (P = .082) compared to imatinib and 3.21 times higher (P = .287) compared to dasatinib. In the non-hyperlipidemia group, nilotinib was associated with a significantly higher risk of CVDs (IRR = 3.21, P = .020) compared to imatinib.

Impact of Post-TKI Diabetes, Hyperlipidemia, and Hypertension on the Risk of CVDs

To examine the risk factors associated with development of CVDs in patients undergoing TKI treatment, we built a multiple Fine-Gray model for each comorbidity group (Table 4). We treated the occurrence of newly diagnosed comorbidities during TKI therapy as a time-dependent covariate, enabling us to evaluate the impact of post-TKI comorbidities on CVD risk, along with different TKI drugs and pre-existing comorbidities. In the non-diabetes group, patients with pre-existing hyperlipidemia were more likely to develop CVDs (SHR = 5.81, P = .034) after adjusting for sex and age. Although we did not find significant associations between TKI drugs and post-TKI diabetes with CVD risk, there were trends suggesting that patients treated with nilotinib had the highest hazard (nilotinib vs. imatinib SHR = 2.08 and nilotinib vs. dasatinib SHR = 3.02, with P = .232 and .318, respectively), and that post-TKI diabetes increased CVD risk (SHR = 4.96, P = .109).

In the non-hyperlipidemia group, both pre-existing diabetes and new diagnosed post-TKI hyperlipidemia greatly elevated the risk of developing CVDs (SHR = 3.33 with P = .019, and SHR = 13.21 with P = .001, respectively). Although patients treated with nilotinib appeared to have a higher hazard of CVDs than patients treated with imatinib and dasatinib, this difference did not reach statistical significance (SHR = 2.15 with P = .165 and SHR = 1.20 with P = .793, respectively).

For patients without pre-existing hypertension before TKI treatment, those who developed hypertension during TKI treatment had a significantly higher chance of developing CVDs than those who did not (SHR = 7.68 with P = .015). In addition, pre-existing hyperlipidemia was positively associated with CVD in this non-hypertension group (SHR = 11.81 with P = .011).

Discussion

Many studies have reported an increased risk of CVDs associated with second-generation TKIs, particularly nilotinib being the most concerning drug.^10,17 The long-term follow-up of the phase III ENESTnd trial showed that within 5-year and 10-year follow-ups, 7.5% and 20% of patients on nilotinib experienced CVDs, respectively.^4,18 Our cohort study indicated that nilotinib was the most significant risk factor for post-TKI newly occurring diabetes and hyperlipidemia compared to imatinib and dasatinib. These results are consistent with the real-world data from the United States, which demonstrate a higher risk of developing type 2 diabetes and hyperlipidemia associated with nilotinib compared to imatinib or dasatinib,¹⁹ indicating that there are no racial differences.

Although we found only a small number of CVD cases in CML patients, our findings highlight nilotinib as a significant risk factor for CVDs. Compared to the first-generation imatinib, the incidence rate of CVDs was over 3-fold higher with statistical significance in the non-hyperlipidemia group for patients treated with nilotinib. In addition, nilotinib consistently demonstrated a higher but not significant IRR of CVDs compared to dasatinib. Since metabolic syndrome, namely diabetes mellitus, hyperlipidemia, and hypertension, is also a risk factor for CVDs,^20,21 it is difficult to discern whether the increased CVD risk associated with nilotinib is due to TKI-induced metabolic changes or direct effects of nilotinib. To address this issue, we attempted to use the subdistribution hazard model allowing patients to transition from a comorbidity-free state to developing comorbidities during TKI treatment, to examine the risk of CVDs in patients with and without comorbidities. Our analysis revealed that nilotinib was associated with a non-significantly higher risk for CVDs compared to the other 2 TKIs, whereas pre-existing and post-TKI hyperlipidemia emerged as the most significant risk factors for CVDs. It is important to note that nilotinib itself is a strong risk factor for hyperlipidemia, which may lead to an underestimation of the risk associated with nilotinib in the analysis of multiple Fine-Gray models. Consequently, efforts to prevent diabetes and hyperlipidemia in CML patients receiving nilotinib may help mitigate the associated risk of CVDs.

In our analysis of time to post-TKI comorbidities, we observed that CML patients treated with nilotinib showed significantly higher hazards of developing post-TKI diabetes and hyperlipidemia compared to the other 2 drugs. These results are consistent with the conclusions from a previous study conducted by Franklin et al¹⁹ The association between diabetes and hyperlipidemia has been well-documented.²² However, our data revealed that the correlations between nilotinib and diabetes, as well as between nilotinib and hyperlipidemia, were stronger than the correlation between diabetes and hyperlipidemia themselves. These findings suggest that nilotinib may be involved in the metabolism of glucose and lipid. Although limited in number, some studies have provided evidence that nilotinib may have a direct impact on glucose and lipid metabolism.^23-25 One proposed mechanism is the development of insulin resistance accompanied by compensated hyperinsulinemia, which can contribute to the onset of diabetes and hyperlipidemia. In addition, genetic factors also contributed to the association between nilotinib and the metabolism of glucose,²⁶ which aligns with our findings. Furthermore, when considering other risk factors simultaneously in a multiple Fine-Gray model, we did not observe a significant association between TKIs and the occurrence of hypertension. Conversely, we found that advanced age was the greatest risk factor for hypertension, which is consistent with the findings of Rodgers et al²⁷

The mechanism underlying nilotinib-induced CVDs is complex. Previous reports have highlighted direct interference of endothelial cell integrity and platelet function.²⁸ Nilotinib has also been associated with atherosclerotic peripheral arterial occlusion.¹⁸ According to our results, we highly speculate that nilotinib may play a role in the metabolism of glucose and lipid, leading to the development of post-TKI diabetes and hyperlipidemia, and may indirectly contribute to an increased risk for CVDs. Therefore, many guidelines emphasize the importance of carefully considering the impacts of metabolic syndrome when contemplating nilotinib therapy for CML patients.^9,29,30 Currently, there is no justification for implementing preventive measures solely based on nilotinib treatment. However, an effective strategy for selecting TKIs involves avoiding nilotinib in CML patients with poorly controlled diabetes and hyperlipidemia to reduce the risk of CVDs. Furthermore, maintaining a high level of awareness during follow-up is essential, as comorbidities may not be present at the initiation of nilotinib therapy and may develop over time with long-term treatment.

This study had several limitations. First, the relatively small number of cases of post-TKI diabetes, hypertension, hyperlipidemia, and CVDs in each comorbidity group, due to stringent definitions compared to other studies, may have led to non-significant results, despite the presence of some observed trends. Second, the risk of CVDs in CML patients is influenced by factors such as the specific TKI drug used, TKI-associated adverse events, and baseline comorbidities. Although we used a multiple Fine-Gray model to examine the newly diagnosed post-TKI metabolic syndrome, it remains difficult to fully elucidate their complex impacts on the development of CVDs. In addition, the severity of comorbidities could not be evaluated since laboratory data and blood pressure were not available in NHIRD. Last, this retrospective study aimed to simulate real-world scenarios and examine the possible effectiveness of treatment strategies. Further prospective studies will be valuable in clarifying the relationship between TKIs, pre-existing and newly diagnosed comorbidities, and CVDs.

Conclusion

The present study represents the first report demonstrating a significant association between nilotinib, metabolic syndrome, and CVDs. Our findings highlight the importance of screening for diabetes and hyperlipidemia prior to initiating TKI treatment, regular monitoring of lipid profiles during TKI use, and effective management of hyperlipidemia in CML patients receiving nilotinib.

Acknowledgments

We thank the Health and Welfare Data Science Center, Ministry of Health and Welfare, Taiwan for making available the database used in this study.

Funding

This work was supported by grants to M.C. Chen from Chang Gung Medical Foundation, Taiwan (CMRPD1J0101-0102 and CMRPD1L0371).

Conflict of Interest

The authors indicated no financial relationships.

Author Contributions

Conception/design: C.-E.H., K.-D.L., J.-J.C., H.-E.T., L.H.-L.Y., M.-C.C. Data analysis and interpretation: All authors. Manuscript writing: C.-E.H., K.-D.L., L.H.-L., M.-C.C. Final approval of manuscript: All authors.

Data Availability

The data underlying this article cannot be shared publicly due to legal restrictions imposed by the government of Taiwan in relation to the “Personal Information Protection Act.” Requests for data can be sent as a formal proposal to the Health and Welfare Data Science Center, Ministry of Health and Welfare, Taiwan (https://dep.mohw.gov.tw/DOS/cp-5119-59201-113.html).

References

Author notes