Immune checkpoint inhibitors and venous thromboembolism in patients with head and neck cancer undergoing surgery

ABSTRACT

Objective

Venous thromboembolism (VTE) is associated with significant morbidity. Although this risk is multifactorial, recent studies suggest immune checkpoint inhibitors (ICIs) may also contribute to increased VTE risk. The aim of this study is to evaluate VTE risk in a cohort of patients with head and neck cancer treated with surgery and ICIs.

Methods

De-identified data from the TriNetX Global Collaborative Network database was used to identify adult surgical patients (≥18 years) using International Classification of Diseases 10^th Revision and Common Procedural Terminology codes and were further refined by use of nivolumab, pembrolizumab, or cemiplimab within 1-year before or up to 3-months after surgery. Cohorts were propensity score matched, and the primary study outcome was the 3-month composite rate of VTE.

Results

After propensity score matching, there were 1,471 patients in each cohort and they were well balanced according to demographics, body mass index, comorbidities, medication use, and radiation history up to 1-year before the index event. The composite rate of VTE in the study population was 3.7% and was higher in patients treated with ICIs (4.6% versus 2.9%; OR, 1.6; 95% CI, 1.1, 2.5).

Conclusions

This study highlights the importance of risk stratification and risk reduction in the setting of increasing ICI use for patients with HNC undergoing surgery.

Plain Language Summary

People with cancer, including head and neck cancer (HNC), are at a higher risk of developing blood clots in their veins after surgery – known as venous thromboembolism (VTE). These clots can cause serious health problems. New treatments called immune checkpoint inhibitors (ICIs), such as nivolumab and pembrolizumab, help the immune system fight cancer but may also increase the risk of blood clots. In this study, we used a large international health database to look at adults with HNC who had surgery. We compared two groups: those who received ICIs around the time of surgery and those who did not. After making sure the two groups were similar in age, health conditions, and other factors, we found that patients who received ICIs had a higher rate of blood clots within three months after surgery (4.6%) compared to those who did not (2.9%). Our findings suggest that using ICIs around the time of surgery may increase the risk of blood clots in patients with HNC. As ICIs become more common in cancer treatment, it is important to consider this risk. Doctors should carefully assess each patient’s risk and consider steps to prevent blood clots when using ICIs in surgical cancer care.

1. Introduction

Patients with cancer have an increased risk of deep vein thrombosis (DVT) and pulmonary embolism (PE) [1,2]. This increased risk is multifactorial and has been attributed to factors such as tumor biology, oncologic surgery, and use of chemotherapeutic agents [3]. Newer studies, however, suggest immune checkpoint inhibitors (ICIs) may also contribute to increased VTE risk by triggering a robust inflammatory response, similar to that seen in immunothrombosis and thromboinflammation [4–8].

The use of ICIs in head and neck cancer (HNC) has expanded since the FDA approved nivolumab and pembrolizumab in 2016 as second-line treatment for recurrent/metastatic disease. Recently, results from clinical trials such as KEYNOTE − 689 have demonstrated a statistically significant improvement in event-free survival with the addition of neoadjuvant pembrolizumab followed by surgery, compared to upfront surgery followed by adjuvant therapies in patients with locally advanced HNC [9]. These findings, along with ongoing studies such as NRG – HN014, suggest a paradigm shift in the management of HNC toward incorporating ICIs in both the neoadjuvant and adjuvant settings [9–14]. Despite favorable outcomes and few adverse effects, neither recurrent/metastatic nor neoadjuvant HNC trials were designed to detect a difference in VTE events among patients receiving ICIs [10–13]. Therefore, VTE events may be underreported in these patients. This is particularly relevant because immunotherapy is the most rapidly growing area of HNC treatment.

The purpose of this study is to evaluate the epidemiology of VTE in a cohort of patients with HNC treated with surgery and ICIs.

2. Methods

This study was exempt by the State University of New York Upstate Medical University institutional review board. We performed a propensity score-matched cohort study using the TriNetX Global Collaborative Network database, a global research network that provides access to deidentified electronic medical records from more 155 million patients at 131 participating healthcare organizations [15]. TriNetX is compliant with the Health Insurance Portability and Accountability Act (HIPAA) and contains health information on diagnoses, procedures, medications, oncologic staging, and lab values. Data were collected on 26 January 2025.

Patients were identified using International Classification of Diseases 10^th Revision (ICD-10) and Common Procedural Terminology (CPT) codes. Adult patients between 18 and 90 years with HNC who underwent ablative and/or reconstructive surgery were identified according to ICD-10 topography and CPT codes and were further refined by ICI use. The experimental cohort consisted of patients treated with nivolumab, pembrolizumab, or cemiplimab within 1-year before and/or up to 3-months after surgery, while the control cohort did not receive ICIs during this time period (Supplementary Tables S1 and S2).

Patient characteristics were collected up to 1 year before the index event and included demographics, comorbidities associated with increased risk of VTE [16], use of anticoagulant medications, relevant radiation and/or chemotherapeutic history, and tumor, node, metastasis (TNM) staging per the American Joint Committee on Cancer [17]. The index event was defined as the day a patient satisfied the criteria for head and neck surgical codes, with or without related ICIs, and a diagnosis of HNC.

The primary study outcome was the 3-month composite rate of VTE (PE and DVT). Secondary outcomes included the 3-month rates of DVT and PE (Supplementary Table 3(a,b)). Patients with the outcome prior to the time window were excluded from final analyses.

2.1. Statistical analysis

Cohorts were propensity score matched in a 1:1 ratio by performing logistic regression to generate a propensity score for each patient and utilizing a greedy, nearest neighbor matching with a caliper of 0.1 pooled standard deviations. Baseline characteristics were then compared between cohorts with standardized mean differences (SMDs) using chi-squared and independent t-tests for categorical and continuous variables, respectively. An SMD less than 0.1 suggests covariates are adequately balanced [18]. Results are reported as numbers and percentages or means ± standard deviations. The association between ICIs and risk of VTE was estimated using logistic regression for the outcomes of interest. Results are reported as effect sizes with odds ratios (OR) and 95% confidence intervals (CI). Statistical analysis was performed in January 2025 using the TriNetX Analytics Network.

3. Results

3.1. Demographics and baseline characteristics

We identified 65,836 eligible patients, of whom 1,484 received ICIs within 1-year before surgery or up to 3-months after surgery, and 64,352 patients who did not receive ICI-therapy. After propensity score matching, there were 1,471 patients in each cohort and they were well balanced according to demographics, comorbidities, medication use, and TNM staging up to 1-year before the index event (Table 1). Mean ages for the experimental and control cohorts were 63.9 ± 13.3 and 64.1 ± 13.1 years, respectively, and both cohorts had higher percentages of male (71.0% vs 71.3%, SMD = 0.008) and Caucasian patients (81.2% vs 82.9%, SMD = 0.04). Mean follow-up time was 83.3 ± 18.2 days for the experimental cohort and 84.4 ± 18.2 days for the control cohort.

Table 1.

Patient characteristics before and after propensity score matching.

Characteristics before propensity score matching Cohort 1 (N = 1,484) and Cohort 2 (N = 64,352)
Cohort			Mean ± SD	Patients	% of Cohort	P-Value	Std diff.
Demographics
12	AI	Age at Index	63.8 +/- 13.261.7 +/- 14.5	1,48460,848	100%100%	<0.001	0.155
12	2106-3	White		1,20547,425	81.2%77.9%	0.003	0.081
12	UNK	Unknown Race		864,674	5.8%7.7%	0.007	0.075
12	F	Female		37321,103	25.1%34.7%	<0.001	0.210
12	2054-5	Black or African American		804,535	5.4%7.5%	0.003	0.084
12	M	Male		1,05538,810	71.1%63.8%	<0.001	0.156
12	2028-9	Asian		522,044	3.5%3.4%	0.760	0.008
Diagnosis
12	J40-J4A	Chronic lower respiratory diseases		1676,618	11.3%10.9%	0.645	0.012
12	K50-K52	Noninfective enteritis and colitis		19456	1.3%0.7%	0.020	0.053
12	I26	Pulmonary embolism		32451	2.2%0.7%	<0.001	0.119
12	I82	Other venous embolism and thrombosis		58944	3.9%1.6%	<0.001	0.145
12	I83	Varicose veins of lower extremities		10107	0.7%0.2%	<0.001	0.077
12	I21	Acute myocardial infarction		22647	1.5%1.1%	0.122	0.037
12	D68.51	Activated protein C resistance		075	0%0.1%	0.176	0.050
Procedure
12	1001	Radiation		2121,959	14.3%3.2%	<0.001	0.399
Medication
12	1364430	apixaban		37692	2.5%1.1%	<0.001	0.102
12	1114195	rivaroxaban		15340	1.0%0.6%	0.022	0.051
12	11289	warfarin		10603	0.7%1.0%	0.221	0.035
12	67108	enoxaparin		2133,051	14.4%5.0%	<0.001	0.320
12	235473	heparin, porcine (deprecated 2017)		64789	4.3%1.3%	<0.001	0.183
12	32968	clopidogrel		181,017	1.2%1.7%	0.172	0.038
12	40048	carboplatin		6379	4.2%0.1%	<0.001	0.284
12	2555	cisplatin		15117	1.0%0.2%	<0.001	0.106
12	56946	paclitaxel		4778	3.2%0.1%	<0.001	0.240
Laboratory
12	9083	BMI	26.3 +/- 6.527.4 +/- 6.6	1,09039,419	73.5%64.8%	<0.001	0.169
12		At least 25.1 kg/m2		61124,859	41.2%40.9%	0.805	0.006
Characteristics after propensity score matching Cohort 1 (N = 1,471) and Cohort 2 (N = 1,471)
Demographics
12	AI	Age at Index	63.9 +/- 13.364.1 +/- 13.1	1,4711,471	100%100%	0.566	0.021
12	2106-3	White		1,1951,219	81.2%82.9%	0.249	0.043
12	UNK	Unknown Race		8684	5.8%5.7%	0.874	0.006
12	F	Female		371370	25.2%25.2%	0.966	0.002
12	2054-5	Black or African American		7977	5.4%5.2%	0.869	0.006
12	M	Male		1,0441,049	71.0%71.3%	0.839	0.008
12	2028-9	Asian		5036	3.4%2.4%	0.125	0.057
Diagnosis
12	J40-J4A	Chronic lower respiratory diseases		165178	11.2%12.1%	0.455	0.028
12	K50-K52	Noninfective enteritis and colitis		1923	1.3%1.6%	0.534	0.023
12	I26	Pulmonary embolism		3233	2.2%2.2%	0.900	0.005
12	I82	Other venous embolism and thrombosis		5552	3.7%3.5%	0.768	0.011
12	I83	Varicose veins of lower extremities		1010	0.7%0.7%	1	<0.001
12	I21	Acute myocardial infarction		2120	1.4%1.4%	0.875	0.006
12	D68.51	Activated protein C resistance		00	0%0%	--	--
Procedure
12	1001	Radiation		205238	13.9%16.2%	0.089	0.063
Medication
12	1364430	apixaban		3637	2.4%2.5%	0.906	0.004
12	1114195	rivaroxaban		1315	0.9%1.0%	0.704	0.014
12	11289	warfarin		1011	0.7%0.7%	0.827	0.008
12	67108	enoxaparin		208225	14.1%15.3%	0.376	0.033
12	235473	heparin, porcine (deprecated 2017)		5962	4.0%4.2%	0.781	0.010
12	32968	clopidogrel		1810	1.2%0.7%	0.129	0.056
12	40048	carboplatin		5026	3.4%1.8%	0.005	0.103
12	2555	cisplatin		1515	1.0%1.0%	1	<0.001
12	56946	paclitaxel		3821	2.6%1.4%	0.025	0.083
Laboratory
12	9083	BMI	26.3 +/- 6.527.1 +/- 6.8	1,080969	73.4%65.9%	0.005	0.123
12		At least 25.1 kg/m2		605596	41.1%40.5%	0.736	0.012

3.2. Study outcomes (Table 2)

Table 2.

Primary and secondary study outcomes.

	Experimental cohort			Control cohort
Outcomes	Patients in cohort	Patients with outcome	Risk	Patients in cohort	Patients with outcome	Risk	Odds ratio	95% Confidenceinterval
Composite VTE Ratea	1,297	60	4.6%	1,350	39	2.9%	1.6	1.1, 2.5
PEb	1,397	24	1.7%	1,420	10	0.7%	2.5	1.2, 5.2
DVTc	1,339	50	3.7%	1,372	29	2.1%	1.8	1.1, 2.9

^a174 patients in Cohort 1 and 121 patients in Cohort 2 were excluded from results because they had the outcome prior to the time window.

^b74 patients in Cohort 1 and 51 patients in Cohort 2 were excluded from results because they had the outcome prior to the time window.

^c132 patients in Cohort 1 and 99 patients in Cohort 2 were excluded from results because they had the outcome prior to the time window.

The composite risk of VTE (all DVT and PE) in the study population was 3.7% (99 of 2,647 patients) and was higher in patients treated with ICIs (4.6% versus 2.9%; OR, 1.6; 95% CI, 1.1, 2.5). The overall risk of PE and DVT were 1.2% (34 of 2,817 patients) and 2.9% (79 of 2,711 patients), respectively. The risk of PE in the experimental cohort was 1.7% as compared to 0.7% in the control group (OR, 2.5; 95% CI, 1.2, 5.2). The risk of DVT was higher in the experimental cohort (3.7% versus 2.1%; OR, 1.8; 95% CI, 1.1, 2.9).

4. Discussion

Although it’s well established that factors such as surgery, cancer, and use of chemotherapeutic agents are associated with increased VTE risk, little is known about the risk of VTE in HNC patients treated with ICIs [1–3]. This propensity score-matched cohort study found that ICIs were associated with an increased risk of composite VTE when administered within 12-months prior to and/or up to 3-months after surgery for HNC. While the odds of both PE and DVT were significantly higher in surgical patients treated with ICIs, the overall increase in composite VTE risk was predominantly driven by a higher rate of DVT in the ICI-treated group. Nevertheless, the statistically significant increase in PE events is clinically relevant because they are associated with higher rates of mortality compared to DVT.

To our knowledge, this study is the first to report an association between perioperative ICI use and elevated VTE risk among patients with HNC, adding to recent evidence linking immunotherapy to thromboembolic events [4–6,8]. These findings are especially noteworthy in patients undergoing surgery for HNC because these patients have a higher risk of perioperative VTE at baseline [19–21]. In the evolving era of immunotherapy, these findings highlight the need for adequate risk reduction during the perioperative period.

4.1. Limitations

This study has several limitations inherent to its observational, retrospective design. First, the possibility of residual confounding cannot be excluded despite propensity score matching. The lack of detailed data on the timing, duration, dose, and adherence to both anticoagulant and ICI therapies contribute to cohort heterogeneity and may result in unbalanced cohorts due to unmeasured variables. Less than half of the study population received anticoagulant prophylaxis in the year preceding surgery, and granular medication data were not uniformly available. Additionally, we compared comorbidities and other factors known to be associated with increased perioperative VTE [16]. Therefore, we cannot exclude the possibility of unequal patient or procedure-specific factors – such as additional comorbidities, duration of cancer diagnosis, active infections, or hospital length of stay – between cohorts.

Second, VTE events may be underreported due to misclassification, misdiagnosis, and/or other reporting errors within the electronic health record. Furthermore, we relied on ICD-10 and CPT codes to define exposures and outcomes, which introduces the possibility of coding or misclassification bias. Lastly, a structural limitation of the TriNetX platform is that it only allows a time-specific relationship to exist between two groups. Therefore, the diagnosis of HNC may have been recorded after surgery or ICI initiation, which would result in an index date corresponding to the date of cancer diagnosis rather than the date of surgery, leading to potential inaccuracies in cohort index dates.

5. Conclusions

Immunotherapy is the most rapidly growing area of HNC treatment. Despite considerable advances in cancer-related outcomes, this study adds to a rising concern that ICIs may be associated with increased VTE risk, especially when used in the neoadjuvant and adjuvant setting [9–13]. This is particularly relevant in otolaryngology because surgery remains the mainstay of primary and salvage treatment for many HNCs and there are no society-specific guidelines for the prevention of perioperative VTE. Although further studies are needed to validate these findings, our study highlights the importance of individualized risk stratification and effective thromboprophylaxis to reduce the risk of perioperative VTE in patients with HNC.

Funding Statement

This paper was not funded.

Article highlights

• Immune checkpoint inhibitors (ICIs) are increasingly utilized in the management of HNC.

• A retrospective cohort study using the TriNetX global health database identified over 1,400 patients in each matched cohort (surgery + ICI versus surgery alone).

• The 3-month composite VTE rate was significantly higher in the ICI group (4.6%) compared to the control group (2.9%; OR 1.6, 95% CI 1.1–2.5).

• Both pulmonary embolism (PE) and deep vein thrombosis (DVT) rates were increased in patients treated with ICIs, with PE having a notably higher odds ratio.

• These findings support the need for individualized VTE risk assessment and prophylaxis strategies in patients receiving ICIs, especially in the perioperative setting.

Author contributions

Kiranya Arnold MD: study design, manuscript writing, editing, and review.

Febronia Mansour BS: study validation, manuscript editing and review.

Komal Akhtar MD: study design, manuscript writing, editing, and review.

Disclosure statement

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

Reviewer disclosure

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/1750743X.2025.2536460