High sensitivity c-reactive protein and circulating biomarkers of endothelial dysfunction in patients with chronic myeloid leukemia receiving tyrosine kinase inhibitors

Abstract Tyrosine kinase inhibitors (TKIs) have revolutionized the management of patients with chronic myelogenous leukemia (CML); however, they may cause cardiovascular (CV) toxicities. In this cross-sectional study, we explored whether high-sensitivity C-reactive protein (hsCRP) and novel markers of vascular dysfunction were associated with exposure to specific TKIs, in 262 CML patients. Hs-CRP level was not associated with CML disease activity or treatment with a specific TKI. Body mass index (OR: 1.15, 95% CI: 1.108–1.246; p < 0.001) and CML duration (OR: 1.004, 95% CI: 1.001–1.008; p = 0.024) were independently associated with higher hs-CRP. In exploratory analyses, novel endothelial-centric markers (e.g. ET-1 and VCAM-1) were differential across the various TKIs, particularly amongst nilotinib- and ponatinib-treated patients. While Levels of hs-CRP do not appear to be correlated with specific TKIs, circulating markers of vascular dysfunction were altered in patients treated with specific TKIs and should be explored as potential markers of TKI-associated CV risk.

Despite the availability of several validated risk prediction models to predict the risk of future CVEs in the general population (e.g.Framingham risk score; FRS) [22][23][24], accumulating evidence suggests that the unique interactions of cardiovascular (CV) risk factors and TKI-specific effect on the vasculature may limit the applicability of these risk prediction models in the CML population [4,20,21].Of the commonly used biomarkers, high-sensitivity C-reactive protein (hsCRP) has been the most widely studied biomarker in risk stratification in the healthy general population.However, it is unclear whether hsCRP can be used in cancer patients for CV risk prediction because it is believed that hsCRP levels are elevated in cancer patients [25].Additionally, as most CVEs appear to be caused by arterial events in CML patients [11][12][13][14], attention has been drawn to the functioning of the vascular endothelium [26].Therefore, more sensitive and specific novel biomarkers of endothelial dysfunction, particularly biomarkers involved in vascular integrity, tone, and coagulative capacity, could serve as more relevant metrics of CV risk [27].
Emerging in vitro and in vivo data support the notion that vasculotoxic TKIs uniquely damage vascular integrity, resulting in a pro-inflammatory and adhesiveactivated phenotype [2,28].However, there is a stark lack of fundamental CV biomarker data in the CML population and a limited understanding of how vascular dysfunction, particularly markers of activation of the endothelium, differ amongst the various TKIs used in CML patients.To address the current knowledge gap, we conducted a cross-sectional study to determine the distribution of hsCRP and the prevalence of CV risk factors in CML patients.Additionally, this study aimed to robustly define TKI-specific CV risk by examining an established inflammatory marker (i.e.hsCRP) and to explore how TKI treatment impacts novel endothelialcentric biomarkers [e.g.vascular cell adhesion protein-1 (VCAM-1), E-Selectin, Endothelin-1, etc.].

Study population
Three hundred and seventeen consecutive adult patients who attended the outpatient CML clinic at Princess Margaret Cancer Center in Toronto, Canada were enrolled from January 2017 to January 2020 in a prospective study (CML prospective-CV) to measure the incidence and predictors of future CVEs.The current analysis is a cross-sectional study to determine the prevalence of CV risk factors and the distribution of hsCRP and endothelial cell markers in enrolled CML patients.All participants provided written informed consent.The study was conducted in accordance with protocols approved by the Research Ethics Board (REB) of the University Health Network (REB#: 16-6229.0).Exclusion criteria for the study included: (i) Unwilling to provide consent, (ii) Physician identified evidence of active infection or recent trauma at the time of venipuncture, (iii) change or discontinuing the TKI less than three months before biomarker measurement, (iv) History of rheumatologic or inflammatory diseases, (v) Previous history of cardiovascular disease (CVD).After excluding patients who did not meet the inclusion criteria, 262 patients were included in this study.

Clinical data collection
Information on CV risk factors was obtained during the visit through interviews and chart review.Habitual smoking across the previous 12 months was used to define 'current smoker' status.Height (meters) and weight (kg) were measured to calculate the body mass index (BMI; kg/m 2 ).History of excessive drinking was defined as drinking more than 10 drinks per week for women or more than 15 drinks per week for men in the last year.Patients were defined as physically active if they exercised at least three times a week, with each session lasting at least 30 minutes.Family history of premature CVD was defined as CVD in a male parent or sibling before age 55 years or a female parent or sibling before age 65.Framingham risk score (FRS) was calculated for eligible patients (i.e.current age between 30-75 years with no previous history of CVD).Diabetes mellitus was defined as fasting serum glucose ≥ 7 mmol/L or hemoglobin A1c ≥ 6.5% on two previous occasions, using blood glucose-lowering medications or a documented history of diabetes.Dyslipidemia was defined as abnormal low-density lipoprotein (LDL) requiring treatment based on Canadian Cardiovascular Society guidelines or documented medication use for dyslipidemia [29].Each participant underwent a physical examination and blood pressure determination.Past medical/medication history was obtained through interviews and subsequently validated through chart review.Blood pressure measurements were obtained at the time of clinical assessment in the seated position with an automatic machine.Hypertension was defined as a systolic blood pressure ≥140 mmHg and/ or diastolic blood pressure ≥90 mmHg measured during two separate clinic visits or the documented use of antihypertensive medications.
Laboratory parameters were collected as reported by the center, with standard international reference ranges applied to decide the cutoff point for abnormal levels.Clinical bloodwork, including lipid profiles, hemoglobin A1c, hsCRP, and molecular testing for BCR-ABL, were measured post-enrollment using standard-of-care techniques.High-sensitivity CRP values ≥10 mg/L were repeated after three months, and the second measurement was considered for analysis.

Preparation of human plasma samples
Peripheral blood samples were collected with standard-of-care BCR-ABL bloodwork at the time of study enrollment.Ten milliliters of whole blood were drawn from the cubital vein (second in draw order) into BD Vacutainer Blood Collection Tubes (BD Biosciences, San Jose, CA, USA) containing K 2 EDTA.Plasma was separated from whole blood through centrifugation (2,500 × g avg , room temperature, 20 min), with the platelet-rich plasma being carefully separated from other blood components.Prior to freezing, platelet-rich plasma samples were centrifuged (2x; 2,500 × g avg , 4 °C, 15 min) to facilitate platelet reduction.All samples were stored at −80 °C until downstream utilization.At no time during the process was the blood or derivatives subjected to temperatures below 4 °C or above 25 °C.

Data visualization and Statistical analysis
Clinical Data -Categorical variables have been summarized with counts and percentages.Continuous variables have been summarized with Mean, Standard Deviation (SD) or medians and ranges.The chi-squared test (χ 2 test) was used to assess the association between categorical variables and two predetermined levels of hsCRP.Logistic regression analysis was used to determine the impact of potential covariates of interest on the two predetermined levels of hsCRP.All p-values were 2-sided, with a p < 0.05 considered the threshold for a statistically significant difference.Statistical analyses were performed using SAS version 9.4 of the SAS system for Windows (Copyright 2002-2012 SAS Institute, Inc., Cary, NC).Vascular Biomarker Data -Data were analyzed using GraphPad Prism 9.0.0 for MacOS (GraphPad Software, Inc., La Jolla, CA, USA).Experiments comparing more than two groups were analyzed using one-or two-way ANOVA testing with a Tukey post-hoc test.p-values of <0.05 were considered statistically significant and indicated in the graphs as reported by the analysis software with significance thresholds of p < 0.05, p < 0.01, p < 0.001, and p < 0.0001 indicated as *, **, ***, and **** respectively.Although many hypotheses were tested throughout the manuscript, no experiment-wide multiple-test correction was applied.Unless indicated otherwise, graphs depict averaged values of independent data points with technical replicates and have error bars displayed as mean ± standard deviation (±S.D.).

Current tyrosine kinase inhibitor exposure, chronic myeloid leukemia activity and highsensitivity c-reactive protein levels
Table 3 shows the current TKI, molecular response, and distribution of hsCRP in enrolled patients.Figure 1 shows the distribution of hsCRP levels in this cohort considering recent TKI exposure.There was no significant association between molecular response and two categories of hsCRP (p = 0.415).Patients who had hsCRP more than 2 mg/L had a longer duration of CML (p = 0.028).hsCRP > 2 mg/L or hsCRP ≤ 2 mg/L was not more frequent in patients receiving specific TKIs (p = 0.126).In other words, treatment with specific TKIs or even being off TKI was not associated with increased chances of having higher or lower hsCRP in this population.This analysis was repeated in patients with low FRS (<10%) or with an age less than 30 years to reduce the confounding effect of CV risk factors on hsCRP level (n = 152), and there was no association between specific TKI and level of hsCRP (p = 0.09).There was no significant association between the median duration of treatment with TKI and two categories of hsCRP (hsCRP ≤2: TKI exposure 71.51 months versus hsCRP >2: TKI exposure 79.53 months, p = 0.424).

Expression of endothelial-centric biomarkers amongst TKI therapies
The vascular endothelium acts as an interface between the blood and tissues and maintains an antiinflammatory and antithrombotic microenvironment in a healthy state [30,31].Emerging data support the notion that vasculotoxic TKIs directly damage endothelial cells in a dose-dependent manner [20,32].A custom simple plex assay for markers robustly associated with endothelial dysfunction (i.kinase receptor Tie-2 [Tie-2], and endoglin-1 was used to test if vascular dysfunction/endothelial activation markers may better reflect the underlying effects of TKIs on the CV system. Using the first-generation TKI imatinib as a reference, there were notable TKI-dependent effects.Amongst markers commonly associated with vasomodulatory and angiogenic function (i.e.Ang-2, Tie-2, ET-1, and Endoglin-1), patients receiving imatinib therapy displayed higher levels of Ang-2 compared to those receiving dasatinib (p = 0.002), nilotinib (p = 0.002), and ponatinib (p = 0.003, Figure 2(A)); this contrasted the higher levels of Ang-2 amongst patients currently off therapy (p < 0.0001).Additionally, patients receiving imatinib displayed lower levels of ET-1 than those receiving bosutinib (p = 0.0202), while having higher levels when compared to those receiving nilotinib (p < 0.001), ponatinib (p = 0.0409), and those currently off therapy (p = 0.0105, Figure 2(B)).Levels of Tie-2 were significantly higher in patients receiving nilotinib (p = 0.009) while significantly lower amongst ponatinib patients (p = 0.001, Figure 2(C)).Finally, significantly higher levels of endoglin-1 were present amongst those receiving ponatinib (p < 0.001, Figure 2(D)).

Discussion
Our study aimed to determine the prevalence of CV risk factors and explore the association between hsCRP levels with CV risk factors, CML disease activity (measured by molecular response), and treatment with specific TKIs.Additionally, we examined the levels of circulating endothelial-centric markers to assess if they vary with the type of TKI used for treatment.We showed that CV risk factors are prevalent in CML patients.We demonstrated that hsCRP, a marker of systemic inflammation, was associated with certain CV risk factors (e.g.BMI) and the duration of CML.Surprisingly, we found no association between hsCRP level and CML treatment outcomes (based on molecular response) or exposure to specific TKIs.Our data demonstrated that a substantial proportion of patients with CML had low hsCRP (40% less than 1 mg/L and 60% less than 2 mg/L).To our knowledge, this is the first comprehensive report of hsCRP levels in CML patients exploring its association with CV risk factors and cancer treatment outcomes.It is important to mention that in a prospective study, Sicuranza et al. showed that hsCRP level increased over time in nilotinib-treated patients compared with imatinib or dasatinib-treated patients; however, hsCRP level remained below 2 mg/L [33].We could not explore temporal changes in our study due to the cross-sectional design.Additionally, we found that endothelial-centric marker expression differed amongst the various TKIs, suggesting that they may more accurately reflect   effects than generalized markers of inflammation (i.e.hsCRP).This provides an impetus to further pursue these markers for associations with CVEs in the future.
CV risk factors were highly prevalent in this study, and 92% of patients had at least one CV risk factor.This is likely because CML is a disease of older adults and CV risk factors are common in cancer patients due to shared risk factors between CVD and cancer [34].Similar to the general population, CV risk factors can explain some of the excess CV risk in the CML population [35].An emerging body of evidence shows that CV risk factors are major contributors to CVEs in CML patients receiving TKIs [34,35].Considering the high prevalence of CV risk factors and the uncertainty regarding CV safety of BCR-ABL TKIs, careful CV risk modification should be an integral part of managing patients with CML.
Previous studies have shown that CML patients are at high risk of CVEs, and specific TKIs such as nilotinib and ponatinib are associated with a higher risk of CVEs [7].Low-grade chronic inflammation is present in all stages of vascular dysfunction.Thus, one would expect higher inflammation burden and high hsCRP levels in patients receiving specific TKIs such as nilotinib and ponatinib that are associated with an increased risk of CVEs [7].However, no association was found between hsCRP level and exposure to specific TKIs in this study.While the low number of patients receiving nilotinib or ponatinib in this study makes it difficult to draw definitive conclusions, the low hsCRP level in a subset of patients receiving nilotinib is intriguing.Additionally, in patients on ponatinib, 6 out of 7 had elevated hsCRP and the lack of statistically significant association in this group could be due to the small number of patients in the ponatinib group.
Chronic systemic inflammation leading to endothelial dysfunction has been demonstrated to contribute to the pathogenesis and progression of atherosclerosis in the general population [36] and in patients with high systemic inflammation such as rheumatoid arthritis [37].data suggest that systemic inflammation, as reflected by elevated hsCRP, might be present in patients with a longer duration of CML, which may result in increased CV risk in these patients.Additionally, comparable to the general population, we found a strong and independent association between high hsCRP and high BMI.Indeed, hsCRP is a strong predictor of future CVEs, and is independent of CV risk factors in the general population [38], and statin treatment in individuals with hsCRP of >2 mg/l and LDL-C < 1.55 mmol/L reduces the risk of CVEs by 40% in the general population [39].Thus, there is strong evidence that patients with residual inflammatory risk nonetheless benefit from statins in the general population.Given that hsCRP in the CML population was not associated with CML treatment outcomes measured as a molecular response, using hsCRP to risk stratify patients is an alternative strategy that might be used to aggressively modify traditional CV risk factors in CML patients with persistent systemic inflammation.Future prospective studies are required to investigate the association between hsCRP and subsequent risk of CVEs in CML patients so that this biomarker can be used for CV risk stratification.
Although the underlying mechanisms of TKI-induced CVEs remain unclear, a wealth of evidence is emerging that supports the notion that low-grade chronic inflammation coupled with vascular dysfunction can enhance the risk of CVEs similar to the general population [40,41].Although not reflected in the classical CV metrics examined, TKI-induced bioactive factors released into the circulation, including peptides and nucleic acids secreted by immune cells, the vasculature, and skeletal muscle, have been implicated in adverse effects [42].As systemically administered medications circulate in immediate contact with the vascular endothelium, it remains a critical nidus for initiating vascular dysfunction.Chronic systemic inflammation and endothelial activation occur during atherosclerosis in general population [43,44] and in patients with rheumatoid arthritis [45].Less is known about the CML population.
Several lines of in vitro and in vivo evidence support the hypothesis that endothelial function is dramatically affected by exposure to specific TKIs [32,46,47].In particular, studies have shown endothelial derangements (i.e.modulation of cytoskeletal proteins), increased endothelial apoptosis, and enhanced expression of pro-inflammatory proteins after exposure to TKIs [32,46,47].Contrasting the lack of association between hsCRP and specific TKI exposure, markers of vascular dysfunction and endothelial activation displayed stark differences amongst the different TKIs.Of particular interest, there was a consistent upregulation of pro-thrombotic factors (i.e.sVCAM-1 and sICAM-1) while significant decreases in angiogenic factors (i.e.Ang-2 and ET-1) amongst individuals receiving nilotinib and ponatinib -two of the most vasculotoxic therapies.These observations are in line with clinical and in vitro studies examining nilotinib and ponatinib, which have highlighted pro-thrombotic angiopathies, decreased endothelial proliferation, and enhanced leukocyte adhesion [32,[46][47][48].Intriguingly, patients on imatinib therapy displayed consistently lower levels of pro-inflammatory and pro-adhesion markers when compared to those on TKIs as well as those off-TKIs.This supports the notion that imatinib may have cardioprotective effects, particularly when it comes to maintaining vascular homeostasis and endothelial quiescence [4].As a whole, this data highlights the potential of endothelium to influence long-term CV risk and suggests the utility of endothelial-centric biomarkers for precision medicine.Future prospective studies are required to explore the utility of endothelial-centric biomarkers and the association with future CVEs in the CML population.
The present study has limitations that should be considered when interpreting the data.Although our data suggest no association between hsCRP and specific TKIs, the number of patients in each TKI group precludes drawing definitive conclusions.Furthermore, as the study was designed as cross-sectional, it excludes any assumptions of causality.While in vitro studies have highlighted the capacity of TKIs to affect endothelial homeostasis, causal relations need to be further determined in vivo.

Conclusions
This study shows that hsCRP, a marker of systematic inflammation, is associated with high BMI and CML duration but not with CML disease activity or specific TKI exposure.Prospective studies are needed to explore the association between hsCRP levels and the risk of future CVEs.Additionally, endothelial-centric markers display TKI-specific profiles that may relate to the risk for CVEs.However, future research is needed to examine the expression of markers associated with endothelial activation and more broadly, vascular dysfunction, both upon diagnosis of CML and longitudinally throughout therapy.

Table 1 .
patient demographics and clinical characteristics of patients at baseline.

Table 2 .
Distribution of high-sensitivity c-reactive protein and cardiovascular risk factors.

Table 3 .
Distribution of categories of high-Sensitivity C-reactive protein considering recent Tyrosine kinase inhibitor exposure and Chronic myelogenous Leukemia variables.