Background:

To investigate the standardized incidence ratios (SIR) of stroke in patients with head and neck cancer and their relationship to radiotherapy.

Methods:

Patients with head and neck cancer ages 20–85 years were enrolled from 2007 to 2016 using the Taiwan Cancer Registry. The study endpoint was fatal and non-fatal ischemic stroke, ascertained by the National Health Insurance Research Database. Age- and sex-adjusted SIRs, categorized by 10-year age standardization, were used to compare the patients with head and neck cancer with a randomly selected 2,000,000 general population. We compared the risk of stroke in patients with head and neck cancer who received radiotherapy or surgery alone. Multivariable adjusted hazard ratios (HR) and 95% confidence intervals (CI) were obtained from Cox regression analysis with competing risk.

Results:

Among 41,266 patients (mean age, 54.1 years; men, 90.6%) in the median follow-up period of 3.9 years, 1,407 strokes occurred. Compared with the general population, the overall SIR of stroke was 1.37 (95% CI, 1.30–1.44) in patients with head and neck cancer. In patients with head and neck cancer, the fully adjusted HR of stroke in those who received radiotherapy was 0.96 (95% CI, 0.83–1.10), compared with those who received surgery alone.

Conclusions:

Patients with head and neck cancer had a higher risk of fatal or non-fatal ischemic stroke. The risk of stroke was not higher in patients initially treated with radiotherapy.

Impact:

Oncologists should emphasize stroke prevention in all patients with head and neck cancer, not only in those who received radiotherapy.

Globally, stroke is the second most common cause of mortality and disability (1). Stroke has an enormous impact on countries’ socio-economic development. The global burden is still increasing (2) and approximately 80% of strokes are due to ischemic cerebral infarction (3). With the progressive advances in cancer screening, molecular diagnostics, and breakthrough new treatments, patients with cancer are regarded as patients living with a chronic disease, and long-term quality of life in patients with prolonged survival has been emphasized. Attributed to the biological factors such as hypercoagulability, adverse effects of treatment and stress, diagnosis-related factors such as defection bias and reduced antithrombotic use in patients with cancer, and shared risk factors such as atrial fibrillation, obesity and smoking, stroke and cancer were highly associated (4). Almost all cancer survivors are at a higher risk of a fatal stroke than the general population and the risk of stroke increases with time (5). The identification of ischemic stroke in patients with cancer has become important in the increasing numbers of cancer survivors, regarding the mortality, morbidity, and long-term quality of life.

Head and neck cancers begin in the mucosal surfaces of the upper aerodigestive tract, usually involve oral cavity, nasal cavity, oropharynx, hypopharynx, and larynx (6). Previous studies have shown that patients with oral cavity and pharynx cancer had a 3- to 4-folds standardized mortality ratio, the relative risk of stroke death, as compared with the general population in the United States (5). An unproportionally high incidence of head and neck cancers in Asians has been reported. Head and neck cancers only account for 3% of malignancies in the United States (7), but were the 5th leading cause of cancer-related death in Taiwan (8), reflecting the prevalence of risk factors like tobacco, alcohol consumption, betel nut chewing, viral infections, and genetic differences (9). However, the standardized incidence ratio (SIR) of stroke in Asian head and neck cancer survivors is lacking.

In addition, previous studies revealed that the risk of ischemic stroke was higher in patients with head and neck cancer who received radiotherapy (10), implying that stroke could be a complication of neck irradiation (11, 12). Patients with localized head and neck carcinomas are generally managed with either surgery or radiotherapy alone; patients with more advanced disease are typically managed using a multimodal approach (12). Furthermore, radiotherapy-induced complications were based on DNA damage and were highly associated with tumor characteristics, treatment anatomic area, and treatment factors such as cumulative dose, technique, and other cancer treatments (13). However, these important confounders were not available in previous studies (14–17). Thus, our study aimed to investigate the SIR of fatal and non-fatal ischemic stroke in patients with head and neck cancer in Taiwan, and the relationship between ischemic stroke and radiotherapy in patients with head and neck cancer receiving radiotherapy, based on comprehensive cancer information.

Study design and study participants

The Taiwan Cancer Registry (TCR), a population-based nationwide cancer registration system in Taiwan, is organized and funded by the Ministry of Health and Welfare. Because the Cancer Control Act was issued in 2003, more than 97% of newly diagnosed cancer cases in Taiwan were registered in the TCR (18). The TCR established a long-form database to record detailed items about the cancer stage, grade, differentiation, histology, size at diagnosis, and initial treatment details such as the date of the first surgical procedure, dosage, and radiotherapy techniques. Over 96% of head and neck cancers in Taiwan have been recorded in the long-form TCR database (8). Information about smoking, alcohol use, chewing of betel nuts, body weight, and body height has been recorded since 2011 in the long-form TCR database (19).

Participants with head and neck cancer were included (n = 78,053). Using the long-form TCR, the primary diagnosis made between 2007 and 2016 and the diagnosed ages 20 to less than 85 years old were identified. Tumors of the oral cavity, oropharynx, hypopharynx, larynx, and salivary glands, based on the International Classification of Diseases for Oncology diagnosis criteria (detailed codes in Supplementary Table S1), were included. We excluded participants with the following: Double cancer (if the head and neck cancer patient had another primary cancer diagnosis rather than head and neck cancer), duplicated data, carcinoma in situ, received further radiotherapy in the initial surgery alone group, missing data, received other treatments and an established history of stroke. This study was conducted in accordance with the Declaration of Helsinki and was approved under exempt review procedures from the Institutional Review Board of the Hsinchu MacKay Memorial Hospital (20MMHIs402e).

Exposures

The initial treatment modalities were obtained from the TCR, which had an excellent data quality (19). We categorized the initial treatment modalities into two groups: The surgery alone reference group and the exposed group. Individuals initially treated with any radiotherapy were included in the exposed group, referred to as curative or palliative radiotherapy with or without any other therapies, including radiotherapy alone, concurrent chemoradiotherapy, surgery and radiotherapy, and surgery with concurrent chemoradiotherapy.

Outcomes

Our primary outcome was fatal and non-fatal cerebrovascular events, including transient ischemic attacks and ischemic strokes. The diagnosis of patients hospitalized for cerebrovascular events was ascertained by reviewing the International Classification of Diseases, Clinical Modification (ICD-CM) codes (details in Supplementary Table S2) retrieved from the National Health Insurance Research Database (NHIRD), which covers 99.9% of the Taiwanese population (20). The cause of death was confirmed by reviewing Taiwan's National Death Registry.

Covariates

Patient characteristics, including age, sex, geographic region of residence, and tumor characteristics (including tumor stage, grades, and histology) were obtained from the long-form TCR database. Medication-based and diagnosis-based comorbidities were retrieved from the NHIRD using the drug codes and ICD-CM codes (Supplementary Table S3).

Statistical analyses

To ensure an unbiased estimated SIR, two million Taiwanese were randomly selected from individuals of age 20 to 85 years in the Department of Household Registration, Ministry of the Interior database on January 1, 2007. Rather than a matched cohort, the large number of two million is more representative of the general population. We did not exclude patients with head and neck cancer in the reference group because the incidence was relative low in the general population. The index date for the head and neck cancer group was the date of cancer diagnosis; the index date for the reference group was January 1, 2007, the initial date in our database. Person-years were calculated for each participant from the index date until the date of occurrence of an event, death, year since diagnosis or on December 31, 2018, whichever came first. The incidence rates of ischemic stroke were the number of cases divided by the number of 1,000 person-years of follow-up. We calculated the observed events to be divided on the basis of the expected events. The SIRs were adjusted by age and sex and categorized by 10-year-age standardization, and adjacent 10-year-age groups were merged if sparse events were noted (21, 22). An additional sensitivity analysis of the SIR for nasopharyngeal cancer was performed. To evaluate the time trend, we calculated the SIR for each year since both cancers were diagnosed.

For objective two, categorical variables were analyzed using the χ2 test, and continuous variables were analyzed using the t test. We calculated person-years for each participant from the index date of initial treatment in the long-form database until the date of occurrence of an event, death, or the first date of recurrence or on December 31, 2018, whichever came first. Events or deaths that occurred after the first recurrence date were not counted. Kaplan–Meier survival curves were used to demonstrate ischemic stroke between the two groups over time. The log-rank test was used to compare the differences. Cox proportional hazards regression analysis, with and without the competing risk of death, was performed to estimate multivariable adjusted hazard ratios (HR) and 95% confidence intervals (CI). The parallel lines of the negative logarithm plot against the logarithm of the follow-up time indicate that the proportional hazards assumption was satisfied (Supplementary Fig. S1). A patient was considered to have a competing risk if he/she died of non-ischemic stroke-related causes. Sub-distribution hazards competing risk models proposed by Gray (23), specific to prognosis, were performed. The following models were applied: Adjusted for sex and age (model 1); further adjusted for urbanization of residence, sub-site, histology, staging, and grading (model 2); further adjusted for hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, aspirin use, antiplatelet agent use, and anticoagulant agent use (model 3). To avoid confounding by indication and sample size decrease, we used the propensity score adjusting rather than matching. A quintile propensity score using covariates in model 3 were applied in model 4.

Potential effect modifiers—sex and age (cutoff, 55 years)—were assessed using the likelihood ratio test to compare the goodness-of-fit of the models with and without the interaction terms in the fully adjusted model with the competing risk. We also compared the risk of stroke in different type and dosage of the radiotherapy. Two sensitivity analyses with the competing risk of non-stroke death were performed to test the robustness of our results. We explored the risk of stroke in those who were treated with surgery alone and radiotherapy only. Additional adjustments for smoking habits, alcohol use, chewing of betel nuts, and body mass index from the long-form TCR database since 2011 were applied for model 5. A quintile propensity score using covariates in model 5 were applied in model 6.

All statistical tests were two-tailed with a type I error of 0.05 and a P value of 0.05 was considered statistically significant. Analyses were performed using SAS software (version 9.4; SAS Institute) and Stata version 15 (Stata Corporation).

A total of 78,053 patients with head and neck cancer were identified in the 2007–2016 TCR database. We excluded participants with the following: double cancer (n = 7,572), duplicated data (n = 9,378), carcinoma in situ (n = 502), <20 or ≥85 years old (n = 961), received further radiotherapy in the initial surgery alone group (n = 7,916), missing data (n = 3,919), received other treatments (n = 5,191), and an established history of stroke (n = 1,348). Finally, 41,266 patients were included. The flow diagram is shown in Supplementary Fig. S2. The mean (standard deviation) age of the participants was 54.1 ± 11.1 years, and 90.6% were men. Compared with the 1,026 stroke events expected in the two million general population, we observed 1,407 stroke events in 185,303.3 person-years during a median follow-up time of 3.9 years. The incidence rate of fatal and non-fatal stroke was 7.59/thousand person-years. Generally, patients with head and neck cancer had a significantly higher risk of ischemic stroke than the general population, and the age- and sex-adjusted SIR was 1.37 (95% CI, 1.30–1.44). The SIRs of ischemic stroke in the first five years after head and neck cancer diagnosis were 1.61 (95% CI, 1.44–1.78), 1.42 (95% CI, 1.30–1.53), 1.32 (95% CI, 1.22–1.42), 1.33 (95% CI, 1.24–1.42), and 1.35 (95% CI, 1.27–1.44), respectively; the SIR over time is shown in Fig. 1 and Supplementary Table S4. Men ages 20–49, 50–59, 60–69, and 70–84 had an attenuated increasing risk of strokes of 4.00 (95% CI, 3.51–4.49), 1.50 (95% CI, 1.37–1.64), 1.16 (95% CI, 1.04–1.27), and 0.93 (95% CI, 0.80–1.06), respectively. The age-specific SIR of stroke in women was not significantly higher in each group (Table 1).

Figure 1.

The age- and sex-adjusted standardized incident ratios of ischemic stroke over time.

Figure 1.

The age- and sex-adjusted standardized incident ratios of ischemic stroke over time.

Close modal
Table 1.

The age and sex adjusted standardized incident ratios in patients with head and neck cancer.

WomenObserved eventsExpected eventsPerson-yearsStandardized incident ratioLower limitUpper limit
20–49 6,192 2.84 0.87 4.80 
50–59 21 15 5,907 1.37 0.78 1.95 
60–69 30 31 4,079 0.97 0.63 1.32 
70–84 44 58 3,431 0.75 0.53 0.98 
Men       
20–49 257 64 63,959 4.00 3.51 4.49 
50–59 465 309 60,531 1.50 1.37 1.64 
60–69 384 332 30,171 1.16 1.04 1.27 
70–84 198 213 11,033 0.93 0.80 1.06 
Total 1,407 1,026 185,303 1.37 1.30 1.44 
WomenObserved eventsExpected eventsPerson-yearsStandardized incident ratioLower limitUpper limit
20–49 6,192 2.84 0.87 4.80 
50–59 21 15 5,907 1.37 0.78 1.95 
60–69 30 31 4,079 0.97 0.63 1.32 
70–84 44 58 3,431 0.75 0.53 0.98 
Men       
20–49 257 64 63,959 4.00 3.51 4.49 
50–59 465 309 60,531 1.50 1.37 1.64 
60–69 384 332 30,171 1.16 1.04 1.27 
70–84 198 213 11,033 0.93 0.80 1.06 
Total 1,407 1,026 185,303 1.37 1.30 1.44 

Note: Bold font indicates a significant risk.

Sensitivity analysis of the SIR for nasopharyngeal cancer showed similar results. A total of 11,710 patients with nasopharyngeal cancer were identified in the database. After excluding participants with the following: double cancer (n = 571), duplicated data (n = 76), carcinoma in situ (n = 2), <20 or ≥85 years old (n = 151), received further radiotherapy in the initial surgery alone group (n = 0), missing data (n = 813), received other treatments (n = 790), and an established history of stroke (n = 159), a total of 9,148 patients were included. The flow diagram is shown in Supplementary Fig. S3. The mean (standard deviation) age of the participants was 50.3 ± 11.8 years, and 76.2% were men. Compared with the 237 stroke events expected in the two million general population, we observed 342 stroke events in 53,429.9 person-years during a median follow-up time of 5.9 years. The incidence rate of fatal and non-fatal stroke was 7.59/thousand person-years. Similarly, patients with nasopharyngeal cancer had a significantly higher risk of ischemic stroke than the general population, the age- and sex-adjusted SIR was 1.44 (95% CI, 1.29–1.60). The SIRs of ischemic stroke in the first five years after nasopharyngeal cancer diagnosis were 1.24 (95% CI, 0.88–1.60), 1.10 (95% CI, 0.86–1.34), 1.09 (95% CI, 0.90–1.29), 1.25 (95% CI, 1.07–1.44), and 1.31 (95% CI, 1.13–1.48), also shown in Fig. 1 and Supplementary Table S4. Men ages 20–49, 50–59, 60–69, and 70–84 had an attenuated increasing risk of strokes of 4.17 (95% CI, 3.28–5.07), 1.60 (95% CI, 1.29–1.91), 1.20 (95% CI, 0.92–1.48), and 0.72 (95% CI, 0.49–0.95), respectively (Supplementary Table S5).

For objective two, a total of 78,053 patients with head and neck cancer were identified. We excluded participants with the following: Double cancer (n = 7,572), duplicated data (n = 9,378), carcinoma in situ (n = 502), <20 or ≥85 years old (n = 961), received further radiotherapy in the initial surgery alone group (n = 7,916), missing data (n = 973), received other treatments (n = 5,020), and an established history of stroke (n = 1,519). Finally, 44,212 patients were included. The flow diagram is shown in Supplementary Fig. S4. The baseline characteristics of the 44,212 individuals included are listed in Table 2. The mean (standard deviation) age of the participants was 54.1 ± 11.1 years, and 90.6% were men. Patients in the radiotherapy group tended to have advanced, poorly differentiated or undifferentiated cancer, less comorbidities, less medication use, more underweight or normal weight, had a habit of smoking, drinking, or using betel nuts, received updated radiotherapy techniques, and a therapeutic dosage of more than 6,000 centigray. Most covariates were significantly different between the two groups, except for anti-coagulant agent use (P = 0.40). The missing covariate rates were generally less than 3%. 17% were missing for smoking habit, 28.5% were missing for betel nut chewing habit, and 37.7% were missing for radiotherapy dosage.

Table 2.

Baseline characteristics of participants according to the initial treatment modalities.

Surgery aloneTreatments with any radiotherapy
n = 15,168n = 29,044
Mean (SD)Mean (SD)P
Age 54.7 (11.9) 53.8 (10.6) <0.001 
Radiotherapy dosage (centigray) — 6,088.9 (1,784.1) <0.001 
Body mass indexa 25.5 (4.2) 23.9 (4.2) <0.001 
 n (%) n (%)  
Age, y   <0.001 
 20–39 1,521 (10.0) 2,306 (7.9)  
 40–64 10,563 (69.6) 22,231 (76.5)  
 65–85 3,084 (20.3) 4,507 (15.5)  
Men 13,370 (88.2) 26,691 (91.9) <0.001 
Urbanization 8,001 (52.8) 14,852 (51.1) 0.001 
Sub-site   <0.001 
 Oral 14,354 (94.7) 26,797 (92.6)  
 Larynx 429 (2.8) 1,123 (3.9)  
 Salivary gland 376 (2.5) 1,010 (3.5)  
Squamous cell carcinoma 14,443 (95.2) 27,481 (94.6) 0.007 
Stage   <0.001 
 I 8,988 (60.2) 2,242 (7.8)  
 II 4,055 (27.2) 3,220 (11.2)  
 III 957 (6.4) 4,249 (14.8)  
 IV 935 (6.3) 19,094 (66.3)  
Grade   <0.001 
 Well or moderately differentiated 12,863 (84.8) 19,512 (67.2)  
 Poorly differentiated or undifferentiated 626 (4.1) 4,129 (14.2)  
 Unknown 1,679 (11.1) 5,403 (18.6)  
Radiotherapy technique   — 
 Traditional 2D, 3D RT — 1,943 (7.0)  
 IMRT or VMAT with/without IGRT — 24,344 (87.8)  
 More than one technique — 1,455 (5.2)  
Radiotherapy dosage ≥6,000 centigray — 24,133 (87.7) — 
Comorbidities 
 Hypertension 6,764 (44.6) 11,624 (40) <0.001 
 Diabetes mellitus 3,193 (21.1) 5,007 (17.2) <0.001 
 Dyslipidemia 3,766 (24.8) 5,441 (18.7) <0.001 
 Atrial fibrillation 234 (1.5) 345 (1.2) <0.001 
Medication 
 Aspirin 2,605 (17.2) 3,731 (12.9) <0.001 
 Anti-platelet agents 1,655 (10.9) 2,563 (8.8) <0.001 
 Anti-coagulant agents 173 (1.1) 3,06 (1.1) 0.4 
Body mass indexa   <0.001 
 Underweight 287 (2.9) 1,406 (8.0)  
 Normal weight 3,451 (35.3) 8,110 (46.1)  
 Overweight 2,887 (29.5) 4,504 (25.6)  
 Obesity 3,162 (32.3) 3,584 (20.4)  
Personal habitsa 
 Smoker 6,039 (71.9) 11,812 (78.2) <0.001 
 Alcohol user 4,357 (43.5) 9,134 (51.2) <0.001 
 Betel nuts user 3,396 (47.7) 7,048 (53.3) <0.001 
Surgery aloneTreatments with any radiotherapy
n = 15,168n = 29,044
Mean (SD)Mean (SD)P
Age 54.7 (11.9) 53.8 (10.6) <0.001 
Radiotherapy dosage (centigray) — 6,088.9 (1,784.1) <0.001 
Body mass indexa 25.5 (4.2) 23.9 (4.2) <0.001 
 n (%) n (%)  
Age, y   <0.001 
 20–39 1,521 (10.0) 2,306 (7.9)  
 40–64 10,563 (69.6) 22,231 (76.5)  
 65–85 3,084 (20.3) 4,507 (15.5)  
Men 13,370 (88.2) 26,691 (91.9) <0.001 
Urbanization 8,001 (52.8) 14,852 (51.1) 0.001 
Sub-site   <0.001 
 Oral 14,354 (94.7) 26,797 (92.6)  
 Larynx 429 (2.8) 1,123 (3.9)  
 Salivary gland 376 (2.5) 1,010 (3.5)  
Squamous cell carcinoma 14,443 (95.2) 27,481 (94.6) 0.007 
Stage   <0.001 
 I 8,988 (60.2) 2,242 (7.8)  
 II 4,055 (27.2) 3,220 (11.2)  
 III 957 (6.4) 4,249 (14.8)  
 IV 935 (6.3) 19,094 (66.3)  
Grade   <0.001 
 Well or moderately differentiated 12,863 (84.8) 19,512 (67.2)  
 Poorly differentiated or undifferentiated 626 (4.1) 4,129 (14.2)  
 Unknown 1,679 (11.1) 5,403 (18.6)  
Radiotherapy technique   — 
 Traditional 2D, 3D RT — 1,943 (7.0)  
 IMRT or VMAT with/without IGRT — 24,344 (87.8)  
 More than one technique — 1,455 (5.2)  
Radiotherapy dosage ≥6,000 centigray — 24,133 (87.7) — 
Comorbidities 
 Hypertension 6,764 (44.6) 11,624 (40) <0.001 
 Diabetes mellitus 3,193 (21.1) 5,007 (17.2) <0.001 
 Dyslipidemia 3,766 (24.8) 5,441 (18.7) <0.001 
 Atrial fibrillation 234 (1.5) 345 (1.2) <0.001 
Medication 
 Aspirin 2,605 (17.2) 3,731 (12.9) <0.001 
 Anti-platelet agents 1,655 (10.9) 2,563 (8.8) <0.001 
 Anti-coagulant agents 173 (1.1) 3,06 (1.1) 0.4 
Body mass indexa   <0.001 
 Underweight 287 (2.9) 1,406 (8.0)  
 Normal weight 3,451 (35.3) 8,110 (46.1)  
 Overweight 2,887 (29.5) 4,504 (25.6)  
 Obesity 3,162 (32.3) 3,584 (20.4)  
Personal habitsa 
 Smoker 6,039 (71.9) 11,812 (78.2) <0.001 
 Alcohol user 4,357 (43.5) 9,134 (51.2) <0.001 
 Betel nuts user 3,396 (47.7) 7,048 (53.3) <0.001 

Note: Bold font indicates a significant risk; treatments with radiotherapy referred to curative or palliative radiotherapy with/without any other therapies in the initial treatment.

Abbreviations: 2D, two-dimensional radiotherapy; 3D, three-dimensional radiotherapy; IGRT, image-guided radiotherapy; IMRT, intensity-modulated radiotherapy; RT, radiotherapy; VMAT, volumetric modulated arc therapy. Smoker: A person not specifically designated as a non-smoker or who has undocumented medical records, unknown smoking status, or has not quit smoking for at least 15 years shall be defined as a smoker. Alcohol user: A person not specifically designated as a non-alcoholic or who has undocumented medical records or unknown drinking status shall be defined as an alcohol user. Betel nuts user: A person not specifically designated as a non–betel nut user or who has undocumented medical records or unknown betel nut–use status shall be defined as a betel nut user.

aVariables from database since 2011.

We documented 1,427 incident cases of fatal and non-fatal stroke in 195,568.4 person-years during a median (interquartile range) follow-up time of 3.8 (1.7–6.7) years. The incidence rate of fatal and non-fatal stroke was 7.30/thousand person-years. A person may have several events, and we used the first event to calculate the person-years. The fatal and non-fatal stroke-free survival time was significantly different between the two groups (log-rank test, P < 0.001). The Kaplan–Meier survival curves are shown in Fig. 2, and the number of individuals at risk in each group over time are shown in Supplementary Table S6. Without competing risk, the risk of fatal and non-fatal stroke in patients with head and neck cancer treated with radiotherapy were significant higher than those who treated with surgery alone, the model 3 and fully adjusted HRs were 1.73 (95% CI, 1.49–2.00) and 1.31 (95% CI, 1.14–1.49). The risks were attenuated considering the competing risk of non-stroke death. The model 3 adjusted HR was 1.36 (95% CI, 1.17–1.56) and turned to insignificant after adjusting for propensity score, the adjusted HR was 0.96 (95% CI, 0.83–1.10; Table 3).

Figure 2.

The Kaplan–Meier survival curves of fatal and non-fatal ischemic stroke according to the initial treatment modalities.

Figure 2.

The Kaplan–Meier survival curves of fatal and non-fatal ischemic stroke according to the initial treatment modalities.

Close modal
Table 3.

The risk of fatal and non-fatal ischemic stroke according to the initial treatment modalities.

Treatments with any radiotherapy
Surgery aloneWithout competing riskWith competing risk of death (sub-distribution hazards)
Participants 15,168 29,044 
Fatal and non-fatal strokes 465 962 
Person-years 87,387.2 108,181.2 
Incidence rate (per 1,000 person-years) 5.32 8.89 
Model 1 1.77 (1.58–1.98) 1.09 (0.97–1.21) 
Model 2 1.69 (1.46–1.96) 1.33 (1.15–1.53) 
Model 3 1.73 (1.49–2.00) 1.36 (1.17–1.56) 
Model 4 1.31 (1.14–1.49) 0.96 (0.83–1.10) 
Treatments with any radiotherapy
Surgery aloneWithout competing riskWith competing risk of death (sub-distribution hazards)
Participants 15,168 29,044 
Fatal and non-fatal strokes 465 962 
Person-years 87,387.2 108,181.2 
Incidence rate (per 1,000 person-years) 5.32 8.89 
Model 1 1.77 (1.58–1.98) 1.09 (0.97–1.21) 
Model 2 1.69 (1.46–1.96) 1.33 (1.15–1.53) 
Model 3 1.73 (1.49–2.00) 1.36 (1.17–1.56) 
Model 4 1.31 (1.14–1.49) 0.96 (0.83–1.10) 

Note: Bold font indicates a significant risk, presented with hazard ratio with 95% confidence interval. Treatments with radiotherapy referred to curative or palliative radiotherapy with/without any other therapies in the initial treatment. Model 1, Adjusted for age (20–39, 40–64, and ≥65 years) and sex. Model 2, Additional adjustment for urbanization of residence (six urban areas or not), sub-site (oral, larynx, salivary gland), histology (squamous cell carcinoma or not), staging (I, II, III, IV), grading (well or moderately, poorly or undifferentiated, unknown). Model 3, Additionally adjusted for hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, aspirin use, antiplatelet agent use, anticoagulant agent use. Model 4, Adjusted for quintile propensity score.

Subgroup analysis (Table 4) revealed that the risk of stroke in head and neck cancer was not significant interaction with sex; the fully adjusted HR was 0.95 (95% CI, 0.82–1.09) in women and 1.05 (95% CI, 0.65–1.70) in men, Pinteraction was 0.13. The risk in the group of patients who received any radiotherapy was significantly higher, especially in patients less than 55 years, than in the counterpart; the adjusted HR was 1.72 (95% CI, 1.33–2.23) versus 0.89 (95% CI, 0.72–1.11), Pinteraction <0.001. Compared with the traditional radiotherapy technique, advanced technique did not change the point estimate significantly, the adjusted HR was 1.24 (95% CI, 0.95–1.62). Compared with the lower radiotherapy dosage (0–6,000 centigray), higher dosage (6,000–8000 centigray) had a borderline protective trend, the adjusted HR was 0.80 (95% CI, 0.64–1.00; Supplementary Table S7).

Table 4.

Subgroup analyses for the risk of fatal and non-fatal ischemic stroke according to the initial treatment modalities with the competing risk.

VariablesSurgery aloneTreatments with any radiotherapyPinteraction
Sex   0.13 
 Women 0.95 (0.82–1.09)  
 Men 1.05 (0.65–1.70)  
Age   <0.001 
 <55 years 1.72 (1.332.23)  
 ≥55 years 0.89 (0.72–1.11)  
VariablesSurgery aloneTreatments with any radiotherapyPinteraction
Sex   0.13 
 Women 0.95 (0.82–1.09)  
 Men 1.05 (0.65–1.70)  
Age   <0.001 
 <55 years 1.72 (1.332.23)  
 ≥55 years 0.89 (0.72–1.11)  

Note: Bold font indicates a significant risk, presented with hazard ratio with 95% confidence interval. Treatments with radiotherapy referred to curative or palliative radiotherapy with/without any other therapies in the initial treatment. Adjusted for model 4, the quintile propensity score.

The sensitivity analysis results are presented in Supplementary Table S8. Compared with those who were treated with surgery alone, those who were treated with radiotherapy only, had a significant higher risk of stroke in model 3, the adjusted HR was 1.42 (95% CI, 1.23–1.64) and turned to be in-significant of adjusted HR of 0.87 (95% CI, 0.60–1.26) in model 4. Similarly, patients with head and neck cancer who received radiotherapy had a higher risk of stroke after further adjustment for personal habits and body mass index in the 2011 database in model 5, the adjusted HR was 1.40 (95% CI, 1.06–1.85) and turned to be insignificant in model 6, the adjusted HR was 1.14 (95% CI, 0.96–1.36).

Our study confirmed that patients with head and neck cancer had a significantly higher SIR of fatal and non-fatal ischemic stroke than the general population, especially in the first year of cancer. Younger, male patients with head and neck cancer were more prone to stroke than the general population of the same age. Nasopharyngeal cancers had similar descending and ascending trends of SIR over time. However, the risk of ischemic stroke was not higher in those initially treated with radiotherapy than those who received surgery alone, considering the indication of radiotherapy.

Taiwan has one of the highest incidences of head and neck cancers in the world (24), and head and neck cancer survivors have steadily increased due to advances in smoking cessation, early screening, and treatment improvement (13). Head and neck cancer survivors have a significantly higher risk of stroke (5, 25, 26), especially in men (27) and in Asia (10). An U-shaped risk of stroke over time were hypothesized on the basis of the review evidence (4), which similar to those of other studies, another meta-analysis results of an inverse U-shaped risk of stroke-like migraine attacks after radiotherapy over time in pediatric patients (28). Further stroke trends over time in cohort study could be investigated. Although the elderly had a high risk of stroke (14, 15, 29, 30), younger patients with head and neck cancer had a higher risk of stroke compared with cancer-free individuals of the same age (10, 16, 31). Our additional analysis of nasopharyngeal cancer yielded results similar to those of previous studies (32), which showed that younger patients of both sexes had a higher stroke risk. The relative lower risk in the elder group, especially in women, could be explained by other risk factors of stroke other than cancer in the elderly.

Traditionally, patients with head and neck cancer treated with radiotherapy alone or combined with other treatments were reported to have a significantly higher risk of ischemic stroke (14, 15, 17, 27, 29, 31–33). The risk of stroke in our study was attenuated after applying the Cox model incorporated with competing risk non-stroke death, which was reasonable, considering that our enrolled population was predominantly in stage IV cancer, and mostly died of cancer itself rather than stroke. In addition, the non-significant results in the age- and sex-adjusted model may be explained by the underlying high-risk reference group, which had more comorbidities and used more medications. Treatment for head and neck cancer requires consideration of the patient-specific comorbidities, the tumor-specific factors and the treatment-specific complications. Previous studies did not adjust theses covariates fully (14, 15, 17, 27, 29–33). Considering all the above factors adjusted in our models, the stroke risk in patients who received radiotherapy was still high. However, the risk of stroke turned to be in-significant after adjusted for propensity score, indicating that the risk of stroke of radiotherapy was highly confounded by indication, which was also supported by our sensitivity analyses and the borderline protective trend of higher radiotherapy dosage rather than the lower dosage. However, possible bias may exist due to the high missing rate of radiotherapy dosage and techniques in our database. To answer the question of radiotherapy-related risk of stroke in patients with head and neck cancer, more studies considered thoroughly confounders and details about radiotherapy should be proceed.

Multiple factors contribute to this increased risk of stroke in patients with head and neck cancer, including carotid artery stenosis (25), increased deposition of plaque (34), and other preexisting risk factors for stroke, particularly smoking. The risk of ischemic stroke increased immediately before and in the year after cancer diagnosis, when cancer-related hypercoagulability was at its peak, followed by a reduction in risk for several years. This was followed again years later by a progressive increase in risk due to the long-term effects of cancer treatments, particularly radiotherapy vasculopathy (4). Theoretically, the oxidative pressure of radiotherapy and inflammation accelerated atherosclerosis predispose the development of ischemic stroke (35). However, the role of radiotherapy was tangled with the underlying advanced patients. Patients with stage IV cancer had a 10-fold risk of stroke compared with stage I patients under the mechanism of high tumor burden–related hyperviscosity and vasculitis (4).

Our findings helped researchers to recognize the long-term health risks in head and neck cancer survivors, who were usually middle-aged men with a habit of smoking, drinking, or chewing betel nuts, and had a relatively low socioeconomic status (36). Neurological complications are one of the most common reasons why patients with cancer are admitted to the hospital (37). Surviving from head and neck cancer but later disabled by a stroke could have a huge impact on the family. We provided evidence to clinicians that stroke prevention is a major issue in this specific population regardless of cancer therapies, especially in younger men. Smoking cessation, lipid lowering, and control of risk factors should be advocated. We expect neuro-oncology multidisciplinary teams to participate in therapeutic decisions and manage neurological complications of cancer (38).

To the best of our knowledge, this is the first study that enrolled large numbers of representative head and neck cancer participants in Taiwan, calculated age- and sex-adjusted SIRs, applied a competing risk model, and controlled for tumor characteristics. The precise methodology with adequate statistical power in the population without selection bias made our results highly convincing. We acknowledge the limitations of this study. First, we focused on ischemic stroke. Further studies focusing on hemorrhagic stroke or cardiac complications should be conducted. Second, we did not control for socioeconomic status, a possible confounder. We did our best to control habitation urbanization. Third, our data could not provide the information of human papillomavirus status, which had an important role in tumor staging and treatment strategies. Further studies that consider social economic status and focus on human papillomavirus status are warranted.

In conclusion, patients with head and neck cancer had a higher risk of ischemic stroke than the general population, especially the younger men. The risk of stroke was not higher in patients treated with radiotherapy, considered the indication of treatment and competing risk.

T.-L. Yeh reports grants from Health Promotion Administration, Ministry of Health and Welfare and Ministry of Science and Technology in Taiwan during the conduct of the study. C.-C. Wang reports grants from Ministry of Science and Technology, Taiwan, and Ministry of Health and Welfare, Taiwan, during the conduct of the study. No disclosures were reported by the other authors.

The content of this research may not represent the opinion of the Health Promotion Administration, Ministry of Health and Welfare in Taiwan. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the article.

T.-L. Yeh: Conceptualization, validation, investigation, writing–original draft, writing–review and editing. C.-T. Hsieh: Software, formal analysis, methodology, project administration. H.-Y. Hsu: Conceptualization, methodology. M.-C. Tsai: Conceptualization, methodology. C.-C. Wang: Conceptualization, validation. C.-Y. Lin: Conceptualization. B.-Y. Hsiao: Software, methodology. J.-R. Jhuang: Methodology. C.-J. Chiang: Resources, methodology, project administration. W.-C. Lee: Resources, supervision, methodology. K.-L. Chien: Conceptualization, resources, supervision.

The datasets generated and/or analyzed during the current study are not publicly available due to the terms of consent to which the participants agreed; data are, however, available from the authors upon reasonable request and with permission from the Health Promotion Administration at the Ministry of Health and Welfare in Taiwan. The authors appreciate the cooperation of Taiwanese National Health Insurance Research Database of Ministry of Health and Welfare (NHIRD_MOHW) and Taiwan Cancer Registry Center for supporting this study. We would like to thank Dr. Y.-C. Lor (Hsinchu MacKay Memorial Hospital, Family Medicine Department) for editing the manuscript. W.-C. Lee received grants from the Health Promotion Administration, Ministry of Health and Welfare (A1091115 Tobacco Health and Welfare Taxation). M.-C. Tsai received grants from Ministry of Science and Technology in Taiwan (MOST 110-2314-B-195-004).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Collaborators GBDN
.
Global, regional, and national burden of neurological disorders, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016
.
Lancet Neurol
2019
;
18
:
459
80
.
2.
Collaborators GBDS
.
Global, regional, and national burden of stroke, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016
.
Lancet Neurol
2019
;
18
:
439
58
.
3.
Krishnamurthi
RV
,
Feigin
VL
,
Forouzanfar
MH
,
Mensah
GA
,
Connor
M
,
Bennett
DA
, et al
.
Global and regional burden of first-ever ischaemic and haemorrhagic stroke during 1990–2010: findings from the Global Burden of Disease Study 2010
.
Lancet Glob Health
2013
;
1
:
e259
81
.
4.
Navi
BB
,
Iadecola
C
.
Ischemic stroke in cancer patients: a review of an underappreciated pathology
.
Ann Neurol
2018
;
83
:
873
83
.
5.
Zaorsky
NG
,
Zhang
Y
,
Tchelebi
LT
,
Mackley
HB
,
Chinchilli
VM
,
BE
Z
.
Stroke among cancer patients
.
Nat Commun
2019
;
10
:
5172
.
6.
Lambert
R
,
Sauvaget
C
,
de Camargo Cancela
M
,
Sankaranarayanan
R
.
Epidemiology of cancer from the oral cavity and oropharynx
.
Eur J Gastroenterol Hepatol
2011
;
23
:
633
41
.
7.
Siegel
RL
,
Miller
KD
,
Jemal
A
.
Cancer statistics, 2020
.
CA Cancer J Clin
2020
;
70
:
7
30
.
8.
2017 Taiwan Cancer Registry Report
.
Health Promotion Administration, Ministry of Health and Welfare
.
Available from
: https://www.hpa.gov.tw/Pages/Detail.aspx?nodeid=269&pid=12235.
9.
Sankaranarayanan
R
,
Masuyer
E
,
Swaminathan
R
,
Ferlay
J
,
Whelan
S
.
Head and neck cancer: a global perspective on epidemiology and prognosis
.
Anticancer Res
1998
;
18
:
4779
86
.
10.
Huang
R
,
Zhou
Y
,
Hu
S
,
Ren
G
,
Cui
F
,
Zhou
PK
.
Radiotherapy exposure in cancer patients and subsequent risk of stroke: a systematic review and meta-analysis
.
Front Neurol
2019
;
10
:
233
.
11.
Gujral
DM
,
Chahal
N
,
Senior
R
,
Harrington
KJ
,
Nutting
CM
.
Radiation-induced carotid artery atherosclerosis
.
Radiother Oncol
2014
;
110
:
31
8
.
12.
Furness
S
,
Glenny
AM
,
Worthington
HV
,
Pavitt
S
,
Oliver
R
,
Clarkson
JE
, et al
.
Interventions for the treatment of oral cavity and oropharyngeal cancer: chemotherapy
.
Cochrane Database Syst Rev
2011
:
4
:
CD006386
.
13.
Haddad
RI
,
Shin
DM
.
Recent advances in head and neck cancer
.
N Engl J Med
2008
;
359
:
1143
54
.
14.
Chen
MC
,
Kuan
FC
,
Huang
SF
,
Lu
CH
,
Chen
PT
,
Huang
CE
, et al
.
Accelerated risk of premature ischemic stroke in 5-year survivors of nasopharyngeal carcinoma
.
Oncologist
2019
;
24
:
e891
e7
.
15.
Wu
YT
,
Chen
CY
,
Lai
WT
,
Kuo
CC
,
Huang
YB
.
Increasing risks of ischemic stroke in oral cancer patients treated with radiotherapy or chemotherapy: a nationwide cohort study
.
Int J Neurosci
2015
;
125
:
808
16
.
16.
Lee
CC
,
Su
YC
,
Ho
HC
,
Hung
SK
,
Lee
MS
,
Chiou
WY
, et al
.
Increased risk of ischemic stroke in young nasopharyngeal carcinoma patients
.
Int J Radiat Oncol Biol Phys
2011
;
81
:
e833
8
.
17.
Smith
GL
,
Smith
BD
,
Buchholz
TA
,
Giordano
SH
,
Garden
AS
,
Woodward
WA
, et al
.
Cerebrovascular disease risk in older head and neck cancer patients after radiotherapy
.
J Clin Oncol
2008
;
26
:
5119
25
.
18.
Chiang
CJ
,
Wang
YW
,
Lee
WC
.
Taiwan's nationwide cancer registry system of 40 years: past, present, and future
.
J Formos Med Assoc
2019
;
118
:
856
8
.
19.
Chiang
CJ
,
You
SL
,
Chen
CJ
,
Yang
YW
,
Lo
WC
,
Lai
MS
.
Quality assessment and improvement of nationwide cancer registration system in Taiwan: a review
.
Jpn J Clin Oncol
2015
;
45
:
291
6
.
20.
Ho Chan
WS
.
Taiwan's healthcare report 2010
.
EPMA J
2010
;
1
:
563
85
.
21.
Cole
SR
,
Lopez-Gatell
H
.
Epidemiologic methods: studying the occurrence of illness
.
Am J Epidemiol
2004
;
159
:
1106
-.
22.
Ury
HK
,
Wiggins
AD
.
Another shortcut method for calculating the confidence interval of a Poisson variable (or of a standardized mortality ratio)
.
Am J Epidemiol
1985
;
122
:
197
8
.
23.
Gray
RJ
.
A class of K-sample tests for comparing the cumulative incidence of a competing risk
.
Ann Stat
1988
;
16
:
1141
54
.
24.
Hsu
W-L
,
Yu
K
,
Chiang
C-J
,
Chen
T-C
,
Wang
C-P
.
Head and Neck Cancer Incidence Trends in Taiwan, 1980–2014
.
Int J Head Neck Sci
2017
;
1
:
180
9
.
25.
Dorresteijn
LD
,
Kappelle
AC
,
Boogerd
W
,
Klokman
WJ
,
Balm
AJ
,
Keus
RB
, et al
.
Increased risk of ischemic stroke after radiotherapy on the neck in patients younger than 60 years
.
J Clin Oncol
2002
;
20
:
282
8
.
26.
Haynes
JC
,
Machtay
M
,
Weber
RS
,
Weinstein
GS
,
Chalian
AA
,
Rosenthal
DI
.
Relative risk of stroke in head and neck carcinoma patients treated with external cervical irradiation
.
Laryngoscope
2002
;
112
:
1883
7
.
27.
Kuan
FC
,
Lee
KD
,
Huang
SF
,
Chen
PT
,
Huang
CE
,
Wang
TY
, et al
.
Radiotherapy is associated with an accelerated risk of ischemic stroke in oral cavity cancer survivors after primary surgery
.
Cancers
2020
;
12
:
616
.
28.
Zhang
F
,
Wang
K
,
Du
P
,
Yang
W
,
He
Y
,
Li
T
, et al
.
Risk of stroke in cancer survivors: a meta-analysis of population-based cohort studies
.
Neurology
2021
;
96
:
e513
e26
.
29.
Arthurs
E
,
Hanna
TP
,
Zaza
K
,
Peng
Y
,
Hall
SF
.
Stroke after radiation therapy for head and neck cancer: what is the risk?
Int J Radiat Oncol Biol Phys
2016
;
96
:
589
96
.
30.
Hong
JC
,
Kruser
TJ
,
Gondi
V
,
Mohindra
P
,
Cannon
DM
,
Harari
PM
, et al
.
Risk of cerebrovascular events in elderly patients after radiation therapy versus surgery for early-stage glottic cancer
.
Int J Radiat Oncol Biol Phys
2013
;
87
:
290
6
.
31.
Huang
YS
,
Lee
CC
,
Chang
TS
,
Ho
HC
,
Su
YC
,
Hung
SK
, et al
.
Increased risk of stroke in young head and neck cancer patients treated with radiotherapy or chemotherapy
.
Oral Oncol
2011
;
47
:
1092
7
.
32.
Chu
CN
,
Chen
PC
,
Bai
LY
,
Muo
CH
,
Sung
FC
,
Chen
SW
.
Young nasopharyngeal cancer patients with radiotherapy and chemotherapy are most prone to ischaemic risk of stroke: a national database, controlled cohort study
.
Clin Otolaryngol
2013
;
38
:
39
47
.
33.
Chu
CN
,
Chen
SW
,
Bai
LY
,
Mou
CH
,
Hsu
CY
,
Sung
FC
.
Increase in stroke risk in patients with head and neck cancer: a retrospective cohort study
.
Br J Cancer
2011
;
105
:
1419
23
.
34.
Brown
PD
,
Foote
RL
,
McLaughlin
MP
,
Halyard
MY
,
Ballman
KV
,
Collie
AC
, et al
.
A historical prospective cohort study of carotid artery stenosis after radiotherapy for head and neck malignancies
.
Int J Radiat Oncol Biol Phys
2005
;
63
:
1361
7
.
35.
Weintraub
NL
,
Jones
WK
,
Manka
D
.
Understanding radiation-induced vascular disease
.
J Am Coll Cardiol
2010
;
55
:
1237
9
.
36.
Vučičević Boras
V
,
Fučić
A
,
Baranović
S
,
Blivajs
I
,
Milenović
M
,
Bišof
V
, et al
.
Environmental and behavioural head and neck cancer risk factors
.
Cent Eur J Public Health
2019
;
27
:
106
9
.
37.
Numico
G
,
Cristofano
A
,
Mozzicafreddo
A
,
Cursio
OE
,
Franco
P
,
Courthod
G
, et al
.
Hospital admission of cancer patients: avoidable practice or necessary care?
PLoS ONE
2015
;
10
:
e0120827
.
38.
Hurford
R
,
Strowd
RE
III
.
Cancer and stroke: a relationship thicker than blood
.
Neurology
2021
;
96
:
143
4
.

Supplementary data