New-Onset Arrhythmias in Advanced Lung Cancer Patients Undergoing Anlotinib Treatment.

Introduction
Lung cancer is recognized as the most prevalent cancer, and advanced lung cancer (ALC) remains one of the major cause of cancer-related death worldwide [1]. As molecular targeted therapy developed rapidly, the efficacy of tyrosine kinase inhibitors (TKI) against ALC has been acknowledged [2]. Anlotinib is a third-line anti-cancer agent approved for ALC [3–7], and exerts anti-tumor effects by targeting multiple molecules, such as vascular endothelial growth factor receptor (VEGFR) [3]. Anti-angiogenesis is the primary anti-tumor mechanism of anlotinib and could induce a series of cardiovascular adverse events, such as hypertension, heart failure, and thromboembolism [8, 9]. Previous studies have observed a non-negligible occurrence of adverse events when applying anlotinib for NSCLC as a third-line treatment, of which Grade 3 to 4 adverse events reached 21.67%, such as elevated thyroid-stimulating hormone, increased total cholesterol, and hypertension [4, 7]. Yet current knowledge remains lacking on the proarrhythmic effects of anlotinib. Therefore by focusing on ALC patients, the present study explored the characteristics and risk factors of new-onset arrhythmias (NOA) during anlotinib treatment, and its impact on patients’ outcome.

Materials and Methods
The study population of this single-center retrospective observational study was obtained from the institutional database. The study protocol was approved by Shanghai Chest Hospital Ethics Committee (IS22023). And the requirement for written informed consent was waived. This study was conducted in accordance with the guidelines of the Declaration of Helsinki (as revised in 2013).

Study Design and Patients
We reviewed the medical records of 1127 lung cancer patients taking anlotinib between May 2017 and November 2021 at Shanghai Chest Hospital. And ALC patients were enrolled who received anlotinib either as monotherapy or concurrently with other anti-cancer agents, such as chemotherapy, targeted therapy, and immune checkpoint inhibitor therapy. The inclusion criteria were: (1) Stage III-IV lung cancer (as per the 8th edition of the TNM classification). (2) Age ≥ 18. (3) Disease progression was observed after first-line treatment of chemotherapy and TKI for all patients with EGFR or ALK mutation, and at least second-line treatment of chemotherapy for patients without driver alterations. (4) In a cycle of 21 days, anlotinib was administrated 12 mg per day from the 1 st to 14th day, and patients received anlotinib treatment for at least one cycle of treatment. (5) Electrocardiogram (ECG) was obtained once every month from the initiation of anlotinib treatment to one month after the last dose of anlotinib. (6) patients were scored 0 to 1, as per the Eastern Cooperative Oncology Group Performance. Exclusion criteria were as follows: (I) age < 18; (II) no ECG at the time of diagnosis; (III) NYHA class III-IV; (IV) incomplete baseline medical records; (V) missing follow-up ECG data. The detailed process of patient selection is illustrated in Fig. 1.

Acquisition of ECG and Holter, the Definition of NOA, Occurrence, and Outcome
Before the initiation of antlotinib treatment, baseline ECG was obtained in all patients. All ECG and Holter files during follow-up were obtained once every month since the initiation of anlotinib treatment till one month after the last dose of anlotinib. Other than regular follow-up, patients were instructed to obtain an ECG if symptoms occurred such as palpitation and chest tightness. The standard 12-lead ECG was obtained for each patient with tracing speed at 25 mm/second and amplitude of 0.1mV/mm. The parameters of ECG were acquired automatically from the institutional ECG system (Cardiac Science ECG system), such as RR-interval, PR-interval, and QRS duration. Manual verification was performed by an experienced cardiologist who was blinded to patients’ group.
NOA was defined as new-onset arrhythmia events that occurred between the initiation of anlotinib treatment and one month after the last dose of anlotinib, which had not been previously recorded. And the following arrhythmias were considered NOA if either no arrhythmia was observed in baseline ECG, and NOA occurred during anlotinib treatment or follow-up ECG yielded new types of arrhythmias relative to baseline ECG.
NOA was the primary endpoint of this study, and all-cause mortality was the secondary endpoint. The time to onset of NOA was defined as the duration from the first application of anlotinib to the first NOA occurrence. NOAs included (1) supraventricular arrhythmias (SVA), including atrial fibrillation (AF), atrial flutter and atrial tachycardia; (2) conduction disorders, including atrioventricular block (AVB), intraventricular block (IVB), and non-specific intraventricular block; (3) QT prolongation (QTP); (4) premature atrial/ventricular contraction (premature contractions; Table S1 in Supplementary file 1). The grade of NOA was assessed based on the National Cancer Institute Common Terminology Criteria for AEs (version 5.0; Supplementary file 2).

Clinical Data Collection
The clinical data was retrieved from the hospital database. The baseline variables included demographic characteristics, medical history, laboratory examination results, anti-tumor therapeutic regimen, and characteristics of lung cancer. In addition, during anlotinib treatment, cardiovascular agents and medications with potential proarrhythmic effects were also recorded, including setron, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, beta-receptor blockers, statins, and anticancer drugs.

Statistical Analysis
For continuous variables consistent with normal distribution, data were presented as mean ± standard deviation, and compared with independent-samples t-test. For continuous variables that didn’t conform to normal distribution, data were presented as median (interquartile range, IQR), and compared with Mann-Whitney-U test. Categorical variables were presented as percentages and Chi-square test was performed for comparison. Multivariate logistic regression analysis was performed to identify the independent risk factors for NOA, and odds ratios (OR) and 95% confidence intervals (CI) were calculated. Overall survival was estimated with Kaplan-Meier method and log-rank test. To avoid immortal time bias when evaluating the impact of NOA on overall survival, a 3-month landmark analysis for overall survival included only patients surviving at 3 months after the initiation of anlotinib treatment. SPSS version 28 was used for data analysis, and the rms package of R software 4.1.2 was used to construct the nomogram model. Statical significance was defined if the P value was < 0.05.

Results

Patient Characteristics
Figure 1 illustrated the detailed process of patient selection. Briefly, 922 patients were enrolled and further categorized into NOA group (n = 208) and control group (n = 714). As summarized in Table 1, the overall population had a mean age of 60.7 ± 9.6 years, mean BMI of 23.15 ± 3.10 Kg/m2 and 72.5% patients were male (n = 668). NSCLC was diagnosed in 750 patients (81.3%), and 673 patients were diagnosed Stage IV (673/922, 73%). The overall smoking rate was 63.0% (n = 581). The overall incidence of diabetes (all patients with type 2 diabetes mellitus) was 12.40% (n = 114). Concomitant platinum-based agents and PD-1 inhibitors were prescribed in 153 patients (16.6%) and 249 patients (27.0%), respectively.

Compared with control group, the NOA group was associated with a senior age, (62.68 ± 8.69 vs. 60.07 ± 9.71, P = 0.001), a lower level of serum sodium (137.68 ± 3.54 vs. 138.15 ± 3.29, P = 0.049), a higher incidence of diabetes (18.8% vs. 10.5%, P = 0.001) and higher prevalence of concomitant platinum-based agents (25.0% vs. 14.1%, P < 0.001), etoposide (11.1% vs. 6.7%, P = 0.039), and PD-1 inhibitors (37.0% vs. 24.1%, P < 0.001).

The Incidence and Grade of NOA
NOA events occurred in 208/922 (22.6%) patients, of which the majority experienced a single type of NOA (174/208, 84%), a smaller portion experienced two types of NOA (32/208, 15%), and only two patients experienced three types of NOA (2/208, 1%). The detailed incidence of each type of NOA is illustrated in Fig. 2A. In patients with a single type of NOA, QTP was the most common type (92/174, 52.9%), followed by premature contractions (42/174, 24.1%) and conduction disorders (33/174, 19.0%). And SVA was the least common type (7/174, 4.0%). In patients with two types of NOA, the most common type was QTP plus conduction disorders (12/32, 37.5%). We described the overlap between QTP (120/208, 57.7%), premature contractions (59/208, 28.4%), conduction disorders (49/208, 23.6%), and SVA (16/208, 7.69%) (Fig. 2B). Combined therapy had an increased risk of NOA than anlotinib monotherapy (169/565, 29.9% vs. 39/357, 10.9%; P < 0.001) (Fig. 3).
As shown in Fig. 4A, among 208 patients with NOA, 177 patients underwent Grade 1 NOA (177/208, 85%), while 22 patients underwent Grade 3 NOA (22/208, 11%). Notably, all Grade 3 NOA events were QTP and occurred at least one month after the initiation of anlotinib treatment (Fig. 4B).

The Occurrence Time of NOA
The time to onset of NOA was defined as the duration from the first application of anlotinib to the first NOA occurrence. The time to onset of each type of NOA is shown in Fig. 5. The time to onset ranged from 7 to 356 days after anlotinib treatment, with a median time to onset of 61.9 days (IQR 30.0–120.8.0.8 days). The time to onset was within five months in 85.1% patients. And the median time to onset was shortest in QTP (60.1 days; IQR, 31.0–112.5.0.5 days), and longest in SVA (92.0 days; IQR, 40.0–139.2.0.2 days). The median time to onset of conduction disorders and premature contractions were 64.6 days (IQR 27.4–135.0 days) and 91.0 days (IQR 38.7–159.0 days), respectively.

Multivariable Analysis of Risk Factors for NOA
Univariate logistic regression analysis yielded univariate risk factors of NOA including diabetes, senior age, concomitant PD-1 inhibitors, etoposide and platinum-based agents. After adjusting variables which could be potentially associated with cardiovascular adverse events, including paclitaxel, beta-blockers, and serum electrolyte level (sodium, potassium, calcium), multivariate logistic regression analysis still yielded that senior age, diabetes, concomitant PD-1 inhibitors, and platinum-based agents were independently associated with the occurrence of NOA (Fig. 6).

Anlotinib Monotherapy Subgroup Analysis
NOA was observed in 39 of 357 (10.92%) patients who received anlotinib monotherapy, including QTP (n = 28, 7.8%), SVA (n = 5, 1.4%), conduction disorders (n = 7, 2.0%) and premature contractions (n = 7, 2.0%). Univariate logistic regression analysis yield that diabetes and age were the risk factors for NOA. After adjusting variables which may potentially associated with cardiovascular adverse events, including levels of serum sodium, potassium, calcium, Beta blockers. Multivariate logistic regression analysis still yielded that age (OR, 1.056; 95% CI, 1.013–1.101; P = 0.01) and diabetes (OR, 2.572; 95%CI, 1.067–6.196; P = 0.035) were independently associated with NOA occurrence in anlotinib monotherapy subgroup (Table 2).

The Impact of NOA on the Outcome
The median overall survival of the overall population, NOA group, and control group was 10.1 months (95% CI, 9.0–11.1.0.1 months), 10.6 months (95% CI, 8.37–15.1 months) and 10.0 months (95% CI, 8.87–11.2 months), respectively. And the overall survival rate was 26.7%, 23.2%, and 27.2% for overall patients, NOA group and control group, respectively (Fig. 7). The overall survival rate of NOA group was similar with that of control group (23.2% vs. 27.2%, P = 0.75). In the 2-year survival analysis, it seems that NOA occurrence had little impact on patients’ overall survival, which is consistent with the low incidence of Grade 3 NOA (2.39%, 22/922). ECG follow-up is from the initiation of anlotinib treatment till one month after the last dose of anlotinib. During ECG follow-up, no malignant arrhythmia was observed, including symptomatic sinus node dysfunction, complete AVB, and ventricular tachycardia/flutter/fibrillation.

Discussions
The present study investigated the clinical characteristics and risk factors of NOA and its impact on the outcome of ALC patients undergoing anlotinib treatment. The main findings are as follows: (1) NOA occurred in 22.6% patients, of which QTP was the most common type (120/208, 57.7%). QTP was the only type of Grade 3 NOA. And the NOA incidence was significantly higher after combined therapy than that after anlotinib monotherapy (29.9% VS. 10.9%, P < 0.001). (2) The median TTO for NOA was 61.9 days (IQR 30.0–120.8.0.8 days). And 85.1% NOA occurred within the first five months of anlotinib treatment. (3) The independent predictors of NOA included age, diabetes, and concomitant PD-1 inhibitors and platinum-based agents. (4) In the 2-year survival analysis, NOA had little impact on patients’ overall survival. It seemed that the NOA had a non-negligible occurrence and a generally benign impact on the outcome of patients with advanced NSCLC undergoing anlotinib treatment.

Overview of NOA
To the best of our knowledge, our study is the first to report NOA after anlotinib treatment. And landmark analysis was employed to balance immortal time bias. The primary finding of our study was that the occurrence of NOA was 22.6%, and the occurrence of NOA above Grade 3 was 11%. QTP was the most common type, which was consistent with frequently reported cardiovascular toxicity of QT prolongation in patients prescribed anlotinib [9].
Although all Grade 3 NOA in our study were QTP, which did not require medical treatment, no malignant arrhythmias or deaths were observed. In the ALTER0303 trial, QTP occurred in 77 (26.2%) patients in the anlotinib arm and 7 (2.4%) patients in the anlotinib arm experienced Grade 3 QTP without any clinical symptoms [9]. Previous studies have also shown that life-threatening arrhythmias are rare during treatment with anti-cancer agents, with an occurrence below 1% [10]. Grade 4 and Grade 5 arrhythmias were absent in the present study, and none of NOA progressed to malignant arrhythmias, including symptomatic sinus node dysfunction, complete AVB, and ventricular tachycardia/flutter/fibrillation. And it seems that the impact of NOA on patients’ ouctome after anlotinib treatment are generally benign without requirement for medical intervention. However, the malignant arrhythmic risk increases significantly when QTc is > 500 ms or increased for > 60 ms from baseline. In clinical practice, routine electrocardiogram (ECG) monitoring should be conducted for patients receiving anlotinib [10], especially since the 2nd month after initiation of anlotinib treatment, considering that most NOA events occurred 30 days after anlotinib treatment.

Anlotinib and NOA
As a multi-targeting TKI, anlotinib exerts its anti-cancer effect via inhibition of VEGFR, c-Kit, FGFR, and PDGFR-α [3]. Therefore by inhibiting VEGFR, anlotinib shares similar mechanisms of toxicity as other TKIs, possibly due to the inhibition of AMPK and K channels, mitochondrial disorders, apoptosis, and thyrotoxicosis, and anti-angiogenesis [10, 11]. Moreover, The binding of VEGF to VEGFR leads to the activation of downstream PI3K-Akt pathway, which in turn induces upregulation of INa and ICa-L and downregulation of IKr and IKs [12, 13], impairing cardiac conduction function, and subsequently leading to cardiac conduction block.

PD-1 Inhibitors and NOA
Another important finding is that the incidence of NOA increased significantly in patients undergoing combined therapy of anlotinib plus ICIs, compared with those undergoing anlotinib monotherapy (29.74% vs. 10.9%, P < 0.001), which is consistent with previously reported increased risk of major adverse cardiovascular events after combined therapy of VEGF inhibitors plus ICIs [8]. Meanwhile, the ICI-induced cardiac arrhythmias have been well-recognized in previous studies, including various degrees of cardiac conduction block, bradycardia, atrial fibrillation, ventricular arrhythmias, and even sudden cardiac death [14–19], possibly due to (1) increased T-cell clone expansion results in the appearance of new self-recognizing effector cells; (2) the release from inhibition of preexisting self-recognizing effector T cells; (3) tumor damage exposure of autoantigens, resulting in expansion of auto-recognizing clones; (4) antibody-mediated autoimmunity; (5) a shift toward a more active and less selective innate immune response; (6) bystander damage from increased levels of cytokines produced by the active antitumor immune response; (7) direct damage resulting from specific molecule expression on target cells [20].
Moreover, concomitant ICIs and VEGF inhibitors may further increase the risk of cardiac arrhythmias, possibly because (1) VEGF could upregulate the expression of PD-L1 [21], and (2) hence by inhibition of VEGFR, anlotinib could aggravate autoimmune myocardial injury via downregulation of PD-L1 and ensuing myocardial injury via infiltration of CD8 + T cell [22].

Other Predictors of NOA
Senior age is associated with an increased incidence of NOA, probably associated with the electrophysiologic characteristics of aged hearts including (1) a higher variability of the activation pattern; (2) an increased dispersion of the epicardial potential duration; (3) a prolongation of the AV-conduction time and of the duration of the epicardial activation signal, which was fractionated in aged hearts. Histological findings showed extensive incorporation of fat cells and connective tissue in ventricular and AV-node tissues, which may explain the prolonged conduction time. These changes help to explain the enhanced susceptibility to arrhythmogenic stimuli with age [23].
The proarrhythmic profile of platinum-based anti-cancer agents has been well recognized. The possible mechanisms included [24, 25] (1) direct damage to cardiomyocytes including cardiac conduction system; (2) inducing ischemia via coronary vasospasm and micro-thrombosis; (3) secondary hypertrophy, fibrosis, and interstitial edema as revealed by histopathological studies.
Diabetes could increase the susceptibility of the myocardium to arrhythmias, with possible mechanisms as follows [26]. Connective tissue septa normally present within individual muscle bundles in an intermittent distribution are considered to be associated with localized dissociation of excitation [27]. Accordingly, the increase of collagen content in the diabetic myocardium in a nonhomogeneous distribution may exaggerate this phenomenon. This study supports the view that glucose intolerance in humans can be associated with an increase in cardiac events.

Limitations
The present study is a single-center study with limited study population. This work was a retrospective cohort study possibly influenced by potential selection bias, and needed further validation with a prospective study. Patients without timely ECG were excluded, which may alter the frequency of NOA. The potential interactions between PD-1 inhibitors and platinum-based agents and the dose dependencies of both drugs remains to be explored in further studies.

Conclusions
The occurrence of NOA is non-negligible in advanced NSCLC patients taking anlotinib, with QTP as the most common type. The independent predictors of NOA included age, diabetes, and concomitant PD-1 inhibitors and platinum-based agents. NOA incidence increased as concomitant anti-cancer agents were prescribed. The impact of NOA on patients’ outcome seemed generally benign, and further studies are warranted for verification.

Supplementary Information
Below is the link to the electronic supplementary material.