Introduction
Improved quality and safety in cancer treatment, including the development of targeted anticancer agents, is improving survival rates in cancer patients. In many cases, patients die due to treatment complications instead of the cancer per se [
1]. According to Khorana AA [
2], cancer per se was the most common cause of death in cancer patients, followed by thrombosis. There are two types of thrombosis, venous thromboembolism (VTE) which includes pulmonary thromboembolism and deep vein thrombosis, and arterial thromboembolism (ATE) which includes cerebral infarction, myocardial infarction, and peripheral embolism. When thrombosis develops in cancer patients, it is referred to as cancer-associated thrombosis (CAT). Compared with patients without cancer, the onset rate of VTE or ATE in cancer patients is 4–7 times higher for VTE [
3] and two times higher for ATE [
4]. In addition, the onset rate of myocardial infarction has been reported to be three times higher in cancer patients than in patients without cancer [
4].
Various risk factors such as age, sex, performance status (PS), and cancer type are reportedly associated with CAT onset. The type of anticancer agent may be an additional risk factor [
5]. Previous studies have investigated only VTE onset rate, which is reportedly 1.92% with the use of regimens containing cisplatin (CDDP) [
6], 3.8% with ramucirumab (Ram) alone, 6.1% with Ram + paclitaxel (PTX) combination therapy [
7‐
11] and 11.9% with bevacizumab (Bev) alone [
12]. In addition, a study that compared VTE onset rate in a CDDP-based treatment group (CDDP group) and a non-CDDP group showed that it was 1.67 times higher in the CDDP group [
6]. On comparing standard antineoplastic therapy used with or without Bev, VTE onset rate was reported to be 1.33 times higher in the Bev group than in the non-Bev group [
12]. In the case of nivolumab administration, only VTE incidence [
13‐
20] has been reported and no reports on risk factors for CAT onset including ATE are present. Moreover, no reports on VTE or ATE onset rate during pembrolizumab administration are available.
It has been reported that the mechanism of CAT development in tumor-bearing patients involves microvesicles rich in tissue factors being released from neoplastic cells, promoting fibrin formation and platelet aggregation. Meanwhile, circulating mucins and P-selectins act to form platelet-rich microthrombi [
21]. In a case report, Kunimasa et al. [
22] suggested that the sudden increase of reactivated T cells immediately after pembrolizumab administration could be associated with CAT onset. Conventionally, although immune checkpoint inhibitors (ICIs) are considered less cardiotoxic, Johnson DB et al. [
23] reported the cases of two patients with melanoma in whom fatal myocarditis developed following treatment with ipilimumab and nivolumab. A meta-analysis conducted by Wang DY et al. investigating the fatal toxic effects associated with ICI [
24] showed that mortality in patients who developed cardiac or neurological disorders was 43%, which was compared with other side effects. Although the incidence of CAT leads to cardiovascular events, cardiac function disturbance can trigger blood clot development. In other words, the mechanism of CAT development during ICI administration to cancer patients may differ from that of CAT development in tumor-bearing patients proposed by Varki [
21]. Therefore, understanding the risk of CAT incidence induced by ICI administration including risk factors for CAT onset is meaningful in assessing points of caution during clinical ICI administration.
Materials and methods
Subjects
Patients who underwent treatment at Fujita Health University Hospital from 1st April, 2017, to 31st March, 2018 with nivolumab or pembrolizumab for lung cancer, kidney cancer, stomach cancer, urothelial carcinoma, or malignant melanoma were enrolled in the study. Patients who developed thrombosis at the start of ICI administration were excluded.
Investigations
This was a retrospective study to identify patients and associated variables from electronic medical records of Fujita Health University Hospital. CAT-positive patients who developed new blood clots within the period from the start of ICI administration to 3 months after it ended were classified as a P group, whereas CAT-negative patients were classified as an N group. Data on the following variables that may be the risk factor for CAT onset were extracted: pre-treatment age, sex, cancer type [
5], blood type [
25], BMI, thromboembolic disease history, heart disease history, diabetes history, blood transfusion history, surgery history, and radiation history. For blood biomarkers, pre-treatment white blood cell (WBC), hemoglobin (Hb), platelet (Plt) and D-dimer [
26] values were collected. The use of antiplatelet therapy, anticoagulants, steroids, and erythropoietin drugs during ICI administration were included in the analysis. The pre-treatment Khorana VTE risk assessment score in the P group patients was calculated.
Assessment
Blood clots were confirmed from the interpretation of CT and ultrasound imaging as well as medical charts. Khorana VTE risk assessment score was calculated based on cancer type, pre-treatment Plt count, Hb level, erythropoietin drug use, WBC count, and BMI.
Statistical analysis
Non-normally distributed variables were described by the median and interquartile range. Non-parametric pairwise comparisons were performed using the Mann–Whitney U test. Univariate analysis was performed as an exploratory analysis of the risk factors. Subsequently, factors with hazard rates of <20% were incorporated in the multivariate logistic regression model. Hosmer–Lemeshow statistical test was used to verify the goodness of fit. The statistical analysis software used was SPSS Ver.22.0 (IBM Corporation, Armonk, NY, USA), and the significance level was set at less than 5%.
Ethics
The study is in compliance with the Ethical Guidelines for Clinical Research and was approved by the Medical Research Institutional Review Board of Fujita Health University.
Discussion
The status of CAT incidence following ICI administration remains unclear, with insufficient current evidence. The present study showed that CAT incidence with nivolumab and pembrolizumab treatment was 7.1% and 10.8%, respectively, in clinical practice. These values were not lower than the CAT onset rates with Ram or Bev (VEGF inhibitors) treatment, which were 3.8% [
7,
8] and 11.9% [
13], respectively. Therefore, ICI administration and CAT onset may be related. In addition, to explore risk factors for CAT onset during ICI administration, we performed univariate analysis to determine the significantly different factors; significantly different variables in Table
1 include drugs closely related to blood clot formation and variables previously reported as risk factors. Candidate factors were identified and multivariate logistic regression analysis was performed. Our analysis showed that thromboembolic disease history was a risk factor for blood clot formation and heart disease history a potential risk factor. Khorana AA et al. investigated VTE risk in cancer patients [
5] and showed that clinical risk factors such as cancer-related, treatment-related, and patient-related factors as well as candidate laboratory biomarkers were associated with for cancer-related VTE. They hypothesized that patient-related factors including ATE-related complications and VTE history were risk factors for VTE in cancer patients. Our findings in the present study are similar to the above previously reported data. Although many of the risk factors for VTE described by Khorana AA et al. [
5] were not identified in the present study, we found that thromboembolic disease history and heart disease history were associated with CAT onset during ICI administration. This suggests the need for more frequent examinations related to blood clots when administering ICI to patients with thromboembolic disease history and heart disease history. In addition, Khorana AA et al. [
5] reported that D-dimer was a risk factor for VTE in cancer patients. It has been reported that if the level increases by >1.44 μg/mL compared to that before treatment initiation, VTE onset is more likely [
27]. Among our patients that developed CAT, D-dimer was measured in 7 patients. The levels increased in these patients in the period between ICI treatment initiation to CAT occurrence with a median increment level of 2.2, a minimum of 0.7, and a maximum of 30.8. Among these seven patients, only 4 showed an increase of >1.44 μg/mL in the D-dimer level. According to Stein PD et al. [
28], D-dimer is inadequate to determine positive blood clot formation. Even if D-dimer was measured in routine medical examination, it would not be a predictive factor for blood clot formation.
ATE incidence in the present study following ICI administration was 4.9%. This was higher than the ATE incidence following Bev treatment (3.8%) [
29]. Therefore, it appears crucial to focus on ATE onset when administering ICIs, as is the case for VEGF inhibitors. Because ATE has a high mortality risk when treatment is delayed, we recommend advanced collaboration with Departments of Strokology and Cardiology when administering ICIs to high-risk patients.
Further, we investigated the effects of platelets, which are related to blood clot formation, using the pre-treatment Plt count; levels in the P and N groups were compared; the Plt count in the N group was significantly higher than that in the P group. Khorana AA et al. studied VTE risk in cancer patients [
5] and listed a pre-chemotherapy platelet count of ≥350,000/μL as one of the risk factors. Therefore, the difference in Plt count between the two groups in the present study may be less likely to be associated with the outcome. The lower Plt count in the P group compared with that in the N group could be attributed to the proportion of patients using antiplatelet drugs in the P group (60%), which was considerably higher than that in the N group (13.4%). Even if platelet aggregation had been controlled at treatment initiation, ICI administration itself could have led to CAT incidence in patients with thromboembolic disease history and heart disease history.
For CAT onset prediction, the Khorana score in the P group was calculated. We found that 5 out of 10 cases scored 2 points, whereas the other cases scored 1 point. Khorana score quantifies prediction of VTE risk before administration of anticancer agents in tumor-bearing patients. The score is calculated based on cancer type, Plt count, Hb level, erythropoietin drug use, WBC count, and BMI [
30]. Score 0 is classified as low-risk, score 1–2 is classified as intermediate-risk, and score ≥ 3 is classified as high-risk. VTE incidence in the derivation and validation cohorts, respectively, was 1.8% and 2% in the intermediate-risk category. The patients in the present study were determined as having intermediate risk. These findings demonstrate that it is difficult to predict VTE onset using the Khorana score alone. Recent research has suggested risk prediction using the Vienna score, which combines D-dimer and P-selectin levels with the Khorana score [
31,
32]. Therefore, we included D-dimer and P-selectin levels although they are not typically measured in routine medical examination. A limitation in the research design of the present retrospective study is the measured rates of D-dimer and P-selectin of 52.5% and 0%, respectively. It is necessary to investigate whether the Vienna score is useful in predicting VTE onset following ICI administration, in a prospective study.
In the present study, the rate of CAT onset during ICI administration and the risk factors for CAT onset were analyzed. These data will be useful in preventing CAT-associated diseases from becoming severe during ICI administration. However, because the present study was a retrospective medical record survey, there were several limitations; information considered related to VTE onset such as PS, histology, advanced stage (metastatic), and central venous catheters [
33] could not be sufficiently extracted; because the study investigated five cancer types (lung cancer, kidney cancer, stomach cancer, urothelial carcinoma, and malignant melanoma), the effects of pre-treatment could not be determined as pre-treatment differed in each cancer type; and although we attempted to investigate the possibility of a sudden increase in reactivated T cells immediately after ICI administration, this could not be performed because T cell count was measured in extremely few patients. In addition, in the present study, the number of CAT onset cases was small, and a sufficient number of cases could not be identified. In future studies, it will be essential to include a larger number of cases.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.