Is the short-term lethal effect of NO 2 ignored?Focus on the relationship between short-term exposure to NO 2 and tumor death in Shenyang, China

Although related studies have shown that the short-term exposure of NO 2 is associated with all-cause and respiratory mortality, there is still no clear evidence for the effect of NO 2 on tumor death. This study aimed to understand whether the lethal effect of NO 2 air pollution was underestimated in China. We comprehensively estimated the association between mortality and NO 2 concentration and also the population susceptibility differences for whole tumors(C00-D48), lung cancer(C34), gastric cancer(C16), liver cancer(C22), colorectal cancer(C18-C20), breast cancer(C50), thyroid cancer(C73), prostate cancer(C61), bladder cancer(C67) and leukemia (C90.1, C91-C95) in a typical old industrial city in the northeast of China. The generalized linear model showed that the increase of NO 2 concentration was positively correlated with an increase of 0.95% (95% condence interval, 0.33-1.57%), 1.86% (0.79-2.95%), 2.33% (0.38-4.32%), 2.55% (0.68-4.45%), 16.3% (4.01-30.04%), 11.29% (5.9-16.94%), respectively, in the mortality of whole tumors, lung cancer, liver cancer, colorectal cancer, thyroid cancer, and leukemia in the city of Shenyang. Population susceptibility differences were also signicant for the short-term effects of NO 2 on tumor mortality. Young and old persons suffering from leukemia were at more death risk from increased concentration of NO 2 and most of this risk of death can be observed on lag0. However, young adults aged 15 to 65 with all tumors and thyroid cancer, were more susceptible to their effects. Women with total tumors, colorectal cancer, were more vulnerable to die from NO 2 air pollution, and signicant risk effects can be observed in lag0-lag6 and lag0/lag4/lag5, respectively. Since WHO has recommended a short-term (24-hour) nitrogen dioxide AQGs level of 25 μg/m 3 , which is more stringent than that of 80 μg/m 3 in the Ambient Air Quality Standard of China. We suggest that the local government should seriously consider guiding legislation and policies of air pollution based on the new WHO Global Air Quality Guidelines, to protect public health with an adequate margin of safety. positive between trac-related NO 2 and lung cancer mortality The results of our with those yielded by the It was found that there was a signicant correlation between the death of lung cancer patients and an increased level of NO 2 concentration. We found that males were more vulnerable to the NO 2 exposure than females on lung cancer mortality, and the delayed effects for 3days (lag0-2) were observed for male lung cancer patients. The lag patterns were similar in the elderly, however, for the young and middle-aged adults, only effect estimates for lag2 were signicant. In the analysis of different age groups in our study, we observed the signicant risk effect of NO 2 on the death of elderly patients with lung cancer; In different genders, NO 2 has a risk effect on both male and female lung cancer patients, but the effect on male patients is mostly on Lag1 and for women is mostly on Lag3 Epidemiologic air pollutants as a risk factor in six cohorts, it 10µg/m 3 the association was varying across age groups. The signicant relationship was only observed in the young and middle-aged population and the strongest effect of NO 2 was observed at 4 and 5 lag days. Associations between NO 2 and mortality of liver cancer were intensied, in two-pollutant models, with PM 2.5 or other gaseous pollutants such as CO or O 3 . The effect of NO 2 pollution on liver cancer death requires further investigation and conrmation. In our study, we found that NO 2 had a signicant risk effect on the death of young and middle-aged patients with liver cancer; It also had a signicant risk of liver cancer in men and women, which occurred on lag4 and lag6 respectively. in NO 2 (Raaschou-Nielsen et al. 2016). Our study also found a signicant risk effect of NO 2 on death in leukemia patients. The risk effect of NO 2 on the death of leukemia children concentrated on lag4, while for young adults and elderly, the risk effect of NO 2 concentrated on the day of pollution; In male and female leukemia patients, NO 2 showed a signicant risk effect. For men, the risk effect concentrated on lag0, but for women, this signicant risk effect could still be observed not only on lag0, Lag1, lag2, and lag5.


Introduction
A report by World Health Organization (WHO) showed that, in 2012, poor air quality was responsible for the loss of seven million lives around the world, with more than one-third of those deaths occurring in the fast-developing nations of Asia. Being the largest developing country, China has been facing some of the worst air quality in recent decades mainly due to its rapid economic development (Li et al. 2015). In 2015, outdoor air pollution was estimated to contribute annually to 4.2 million premature deaths, with 1.1 million premature deaths in China alone (Cohen et al. 2017). Nitrogen dioxide mainly comes from the release of high-temperature combustion processes, such as motor vehicle exhaust and imperfect combustion of coal caused by the boiler. Shenyang is a representative industrial city in northern China. Its winter lasts from November to March every year, accompanied by heating. Our previous studies have also shown that air pollution in northern China, Shenyang, was associated with increased mortality, mainly due to the equipment of heating system (Liu et  Many epidemiological studies have found that exposure to air pollutants not only has a risk effect on lung cancer but also many other systemic tumors of the human, such as urinary system tumors, colorectal cancer, and childhood leukemia (Filippini et Yorifuji and Kashima. 2020). And it can also be found that the correlation between NO 2 exposure and incidence/mortality of tumor was more signi cant, and the effect seemed to be more stable and long-lasting than that of other pollutants, compared with other particulates and gaseous pollutants. However, the evidence that short-term exposure to ambient NO 2 could lead to tumor death is still inconclusive.
Since burning coal for heating in winter will lead to the increased concentration of ambient NO 2 . We suspected that the short-term exposure effect of nitrogen dioxide may be more serious in Shenyang based on the previous studies. Although some studies focused on the short-term effects of air pollutants in China (Chen et al. 2012a, Chen et al. 2012b), due to the geographical and demographic differences (Lin et al. 2017), the evidence from other cities in China may not be suitable enough to apply to Shenyang city. Here, we provided a critical appraisal of the effect of exposure to NO 2 on tumor death.

Data Collection
The mortalities of tumors in Shenyang urban area from 2013 to 2018 were obtained from the death registration system of Shenyang Center for Disease Control and Prevention (the causes of death were registered in the system according to the Tenth edition of the international classi cation of diseases, ICD-10). The daily death data of all tumors (C00-D48), lung cancer (C34), gastric cancer (C16), liver cancer (C22), colorectal cancer (C18-C20), breast cancer (C50), thyroid cancer (C73), prostate cancer (C61), bladder cancer (C67) and leukemia (C90.1, C91-C95) were screened out from death registration system. Data was strati ed according to age and gender.
The daily average concentration of six pollutants (PM 2.5 , PM 10 , SO 2 , NO 2 , CO, and O 3 ) in Shenyang were obtained from 13 air quality monitoring stations of the national environmental monitoring center in Shenyang. Meteorological data (including daily average temperature, wind speed, humidity, and air pressure) was downloaded from the website of the Chinese National Meteorological Administration.

Data Analysis
Page 3/18 The number of tumor deaths, pollutant concentration, and meteorological data were statistically described to understand the changing trend of tumor mortality and pollutant distribution characteristics, separately. The Spearman correlation coe cients of air pollutants were calculated to detect the relationship among air pollutants, and the deviance of daily death toll was measured to verify the distribution characteristics. The generalized linear model was established by using well-sorted data. In the model, the degree of freedom of spline function controlling covariates in the model referred to previous literature and uses the Akaike information criterion (AIC criterion), to obtain the best-t model. According to other similar articles, took 7 days as a period to observe the lag effect of NO 2 short-term exposure. Firstly, a single pollutant model was established to observe the effects of single-day Lag (denoted by Lag1, Lag2……Lagk)both and cumulative Lag(denoted by Lag01, Lag02……Lag0k), and then the stability of NO 2 -induced impact on tumor disease risk effect was compared according to the results of two-pollutant models and threepollutant models. Finally, the above methods were also used to analyze the threshold and lag effect of NO 2 in the analysis of strati ed populations, based on gender and age. There were three age groups: < 15 years old: children; 15-65 years old: middle-aged and young/adults; > 65 years old: old population. The different cancer types, including lung cancer, gastric cancer, liver cancer, colorectal cancer, breast cancer, thyroid cancer, prostate cancer, bladder cancer, and leukemia, were also analyzed separately.
All statistical analyses were implemented by R(4.0.2) software. A p-value less than 0.05 is considered statistically signi cant.

Results
Data Description Table 1 shows the distribution of air pollutants, meteorological data, and the number of cancer deaths during 2013-2018. In the study period, the daily average concentrations of PM 2.5 , PM 10  It can be observed that in Table1 there were signi cant differences in temperature among all the observed meteorological factors in Shenyang. From 2013 to 2018, the highest temperature in Shenyang urban area was 32.4°C, while the lowest temperature was -22.8°C. This also indicates the differences in climate between the north and south of China. Among all kinds of tumors, the average daily death toll of all tumors, lung cancer, gastric cancer, liver cancer, colorectal cancer, bladder cancer, leukemia, breast cancer, prostate cancer, and thyroid cancer were 23.23,7.79, 1.76, 2.30, 2.59, 0.11, 0.46, 0.85, 0.29 and 0.06 respectively, and the average daily death of lung cancer was much higher than that of others.
Table3 shows both total and tiered data about the number of deaths caused by tumors and other nine tumors in each year during the ve years. During 2013-2018, the number of deaths of tumors was 8229, 8392, 8456, 8661, 8610, and 8545 respectively. In addition, in Table3, a few children died of cancer. The change in the number of tumor deaths is shown in Figure 1. From this picture, it is easy to see that during the study period, the death toll of the whole tumors, lung cancer, liver cancer, colorectal cancer, pancreatic cancer, breast cancer, bladder cancer, and leukemia increased. The death toll of gastric cancer decreased. The number of deaths from thyroid cancer remained basically unchanged after uctuations.Table4 shows the Spearman correlation coe cients between pollutants and meteorological data. NO 2 was signi cantly correlated with other air pollutants.
It had the strongest correlation with PM 2.5 (r=0.7), and a signi cant negative correlation with ozone, with a correlation coe cient of -0.46. The correlations between NO 2 and meteorological factors were also signi cant. The correlation coe cient between NO2 and temperature is negative, which means that NO 2 concentration will increase with the decrease of temperature.
Single-pollutant model According to the established single-pollutant model (Fig. 2), it could be observed that the increase of NO 2 concentration was related to the death toll of the whole tumors, lung cancer, liver cancer, colorectal cancer, leukemia, and thyroid cancer.
The effect of NO 2 on tumor death reached peak values on Lag1. The mortality of all tumors increased by 0.95% (95% CI: 0.33-1.57%) for per 10µg/m 3 increment in daily average concentration. The maximum effect on lung cancer death also occurred on Lag1, with a 1.86% (95%CI: 0.79-2.95%) increase per 10µg/m 3 . The signi cant effect for liver cancer was observed only on Lag4 with an increment of 2.33% (95% CI: 0.38 -4.32%).
For the cumulative lag effect, in the whole tumors and lung cancer death population, statistically signi cant effect values could be observed on lag01 to lag07, and reach the maximum in lag03 (RR=1.01; 95% CI: 1.00-1.02) and lag07 (RR=1.03, 95% CI: 1.01-1.05) respectively; On lag05 to lag07, NO 2 had a signi cant effect on the death of liver cancer. Signi cant risk effects were observed on Lag01 and Lag04 to Lag07 for colorectal cancer death toll; In the death population of thyroid cancer, signi cant risk effects could be observed in Lag01 to Lag05, and reach the maximum on Lag03 (RR=1.25; 95% CI: 1.06-1.48). A signi cant risk effect of NO 2 could be observed on Lag01 to Lag03 in leukemia deaths.

Multi-pollutant model
As shown in Fig. 3, the risk effects of NO 2 on the tumors death were very robust in the multi-pollutant models, except for breast and prostate cancer, although the respective RR values varied slightly.
The risk effect of NO 2 mostly persisted from lag0 to lag5 in all tumor deaths; In lung cancer, the danger effect was concentrated on lag1-lag3; For colorectal cancer, the hazard effect focused on lag5; For leukemia, the danger effect was observed almost exclusively on lag0. While NO 2 showed incoherent harmful effects for liver and thyroid cancer, for liver cancer the harmful effects were mostly observed on lag0, lag2, lag4, and lag6; For thyroid cancer, the hazard effects were more frequently found on lag0, lag2, and Lag3.

15-65-year-old
In all tumors, liver cancer, thyroid cancer, and leukemia, NO 2 not only showed independent hazard effects but also remained signi cant and stable in most multi-pollutant models, as shown in Fig. 5. There were also great differences in the time and persistence of NO 2 harmful effects. For all tumors, lung cancer, and leukemia, the risk effect of NO 2 was mostly observed on lag0 and Lag1, while for other malignant tumors, it was mostly observed after lag2. In addition, for young and middle-aged patients with all tumors, liver cancer, and thyroid cancer, the risk effect of NO 2 lasted for a long time, about 3-4 days.

>65-year-old
It could be observed from the Fig. 6 that the signi cant and lasting independent harmful effect of NO 2 could be observed only in elderly patients with lung cancer, and this harmful effect could last for three days from lag0. After the adjustment of multi-pollutant models, the signi cant and stable risk effect of NO 2 could be observed in all cancers, lung cancer, colorectal cancer, thyroid cancer, bladder cancer, and leukemia in the elderly.
For all tumors and lung cancer, the harmful effect of NO 2 lasted a long time. For colorectal cancer, thyroid cancer, bladder cancer, and leukemia, the harm of NO 2 could only take 1 day, on lag5, Lag1, and lag0 respectively.

MALE
In Fig. 7, it can be seen that the independent harmful effect of NO 2 was very signi cant in male patients with lung cancer, which lasted for three days from lag0. In the multi-pollutant models, only stable risk effects for male patients with lung cancer and leukemia could be found. Among them, for lung cancer, the duration of risk effect was mostly 2 days, while for leukemia, the harmful effect could be observed only on lag0. FEMALE For female cancer patients, the independent risk effect of NO 2 could be observed in the whole tumors, colorectal cancer, thyroid cancer, and leukemia. For all tumors, liver cancer, colorectal cancer, and leukemia, the risk effect of NO 2 could be observed in almost all multi-pollutant models.
As shown in Fig. 9, the duration of NO 2 harmful effect on female tumor patients was very different. For the whole tumors, lasting and signi cant harmful effects could be observed almost from lag0 to lag6. For liver cancer, signi cant relative risks almost appeared on lag6. The effect of NO 2 on colorectal cancer was mostly concentrated on lag0, lag4, and lag5. For leukemia, the effect of NO 2 could continue from lag0 to lag2 and mostly focused on lag0 and lag2. Besides, it could be found that the effect of NO 2 concentrated on Lag3 for lung cancer and lag0 for thyroid cancer.

Discussion
In the present study, all people who died from the tumor in the old industrial city of Northeast China were analyzed. We sought to comprehensively estimate the impacts of NO 2 on the mortality of cancers. Ambient NO 2 exposure may increase the risk of tumor death. And the harmful impacts were almost still signi cant after controlling for the other pollutants in our two-pollutant models or multi-pollutant models. Sex and age are effect modi ers of the association between NO 2 exposure and daily cancer mortality risk.
The results from our study were consistent with several prior studies. A meta-analysis showed that all-cancer mortality signi cantly correlated with long-term exposure to NO 2 (RR: 1.06, 95% CI: 1.02-1.10) (Kim et al. 2018). Similarly, a study that studied the relationship between childhood cancer risk and acute air pollution in Tehran, Iran, observed a positive association between NO 2 exposure and childhood cancer incidence (Sei et al. 2019).
Another study in Great Britain also con rmed that exposure to NO 2 increased the risk of childhood cancer (Knox. 2005). In our study, NO 2 also showed a strong risk for middle-aged and elderly cancer patients. Besides, the risk effect of NO 2 on male cancer patients was not signi cant. On the contrary, for female patients, the risk effect of NO 2 on death was signi cant and lasted for a long time China is a big country in coal consumption, especially in the north of China with prolonged heating times (Xing et al. 2019). Since the respiratory tract is the rst organ that contacts and affected ambient air pollutants. The correlation between NO 2 exposure and mortality from respiratory diseases, especially lung cancer is well-established. A Japanese study showed that there was a statistically signi cant association between NO 2 and lung cancer death (HR = 1.20, 95% CI: 1.03-1.40) (Yorifuji et al. 2013). In addition, a cohort study of 70000 people in California observed a statistically signi cant positive correlation between NO 2 and lung cancer death (Jerrett et al. 2013). Besides, a meta-analysis reported that there was a positive correlation between tra c-related NO 2 and lung cancer mortality (Hamra et al. 2015). The results of our study were consistent with those yielded by the above studies. It was found that there was a signi cant correlation between the death of lung cancer patients and an increased level of NO 2 concentration. We found that males were more vulnerable to the NO 2 exposure than females on lung cancer mortality, and the delayed effects for 3days (lag0-2) were observed for male lung cancer patients. The lag patterns were similar in the elderly, however, for the young and middle-aged adults, only effect estimates for lag2 were signi cant. In the analysis of different age groups in our study, we observed the signi cant risk effect of NO 2 on the death of elderly patients with lung cancer; In different genders, NO 2 has a risk effect on both male and female lung cancer patients, but the effect on male patients is mostly on Lag1 and for women is mostly on Lag3 Epidemiologic studies suggest air pollutants as a risk factor in liver cancer. In a study with six European cohorts, it was estimated per a 10µg/m 3 increase in NO 2 concentrations was associated with a 17% (95% CI: 2%-35%) increase in liver cancer incidence in the fully adjusted model (So et al. 2021). However, the relationship between NO 2 and liver cancer was still inconclusive. A cohort study conducted in Danish and Austria did not demonstrate a signi cant association between the increase of NO 2 concentration and the high incidence rate of liver cancer (Pedersen et al. 2017).
In our present study, although it could be found that liver cancer death is related to NO 2 concentration, Moreover, the association was varying across age groups. The signi cant relationship was only observed in the young and middle-aged population and the strongest effect of NO 2 was observed at 4 and 5 lag days. Associations between NO 2 and mortality of liver cancer were intensi ed, in two-pollutant models, with PM 2.5 or other gaseous pollutants such as CO or O 3 . The effect of NO 2 pollution on liver cancer death requires further investigation and con rmation. In our study, we found that NO 2 had a signi cant risk effect on the death of young and middle-aged patients with liver cancer; It also had a signi cant risk of liver cancer in men and women, which occurred on lag4 and lag6 respectively.
In a cancer prevention study -(CPS -) in the United States, it was found that there was a signi cant positive correlation between NO 2 exposure and the mortality of stomach cancer, colorectal cancer, breast cancer, and bladder cancer. However, no association between NO 2 and gastric cancer death had been found in our study. There was a positive but weak association between NO 2 exposure and colorectal cancer mortality, and the HR was 1.06 (95% CI: 1.02-1.10, per 6.5 ppb) (Turner et al. 2017). In our study, we also found a signi cant association between colorectal cancer death and NO 2 exposure. In the analysis of different age groups, we found that NO 2 had a signi cant harmful effect on elderly patients with colon cancer, and it concentrated on lag5; For people of different genders, NO 2 was only harmful to female patients with colorectal cancer, and mostly appears on lag0, lag4, and lag5.
A study in South Korea found that NO 2 was signi cantly associated with breast cancer incidence, a district with 10 ppb higher NO 2 concentration suffered from higher OR of breast cancer incidence by 1.14 (95% CI: 1.12-1.16) in the single-pollutant model, and NO 2 remained associated with the breast cancer incidence rate when additionally adjusted with the other three pollutants in multi-pollutant models (Hwang et al. 2020). A Canadian study also found that the incidence of breast cancer is associated with long-term exposure to NO 2 (Hystad et al. 2015). However, we did not observe any signi cant association between NO 2 exposure and the mortality of breast cancer.
The association between NO 2 exposure and bladder cancer was observed in 15 European cohort studies and a study of bladder cancer in Spain, but with no statistical signi cance (Pedersen et al. 2018;Turner et al. 2019a). Another research in Canadian found that, for each increase of 5 parts per billion of NO 2 , the adjusted OR was 1.44 (95% CI: 1.21-1.73) (Parent et al. 2013). In our study, we found an increase of 10µg/m 3 in NO 2 average concentration corresponded to an increase of 5.4% in bladder cancer mortality in the elderly. In our analysis of bladder cancer, NO 2 only had a signi cant risk effect on elderly patients with bladder cancer in different age groups.
In addition, in our study, we also found a signi cant association between NO 2 and leukemia death. The association between NO 2 and leukemia mortality was evident on the day of the event (lag0) for children but for adults, it was present for up to two days (lag2) before the event. Also, the effects of NO2 on leukemia mortality appeared immediately for males and lasted for 2 days for females. Prior epidemiological studies on the relationship between air pollutants and leukemia mostly focused on children, and relatively few focused on adults. A meta-analysis found that, for all children with leukemia, NO 2 exposure caused an OR of 1.21 (95% CI: 0.97-1.52), and for acute lymphoblastic leukemia was 1.21 (95% CI: 1.04-1.41) (Filippini et al. 2015). But the association between NO 2 and childhood leukemia was found to be not signi cant in another meta-analysis of the relationship between tra c pollutants and childhood leukemia (Gong et al. 2019). In an epidemiological study of adult leukemia, it was found that the odds ratio for acute myeloid leukemia was 1.31(95% CI:1.02-1.68) per 10mg/m 3 increase in NO 2 (Raaschou-Nielsen et al. 2016). Our study also found a signi cant risk effect of NO 2 on death in leukemia patients. The risk effect of NO 2 on the death of leukemia children concentrated on lag4, while for young adults and elderly, the risk effect of NO 2 concentrated on the day of pollution; In male and female leukemia patients, NO 2 showed a signi cant risk effect. For men, the risk effect concentrated on lag0, but for women, this signi cant risk effect could still be observed not only on lag0, Lag1, lag2, and lag5.
No association between NO 2 and gastric cancer death has been found in our study, and no difference has been observed among patients in different age or gender groups. Few relevant studies found a signi cant association between them. The research on this aspect needs further indepth discussion. Indeed, a total of 2163 days of NO 2 concentration data were obtained during our study period. Only on 55 days, the 24-hour average NO2 concentration exceeded the level of 80µg/m 3 according to the Chinese air quality standard. However, following WHO AQG2021, the number of days that exceeded the standard was 1832. The gap between these two numbers is surprising. And from the results, we could nd a signi cant association between NO 2 and cancer patient death. This led us to suggest that, under the current policy, the survival of cancer patients would be affected even if the concentration of NO 2 drop to the so-called healthy levels. So, this study not only reminded us of the existence of the short-term lethal effect of NO 2 but also might have implications for local governments to make policies for NO 2 pollutant controls.

Limitations
Due to the accumulation of air pollutants, it was di cult to avoid that the short-term exposure effect we studied was not affected by the long-term cumulative effect. In addition, due to the strong linear relationship among air pollutants, although multi-pollutant models with strong collinearity have been removed, it was still di cult to completely remove the in uence of collinearity in the results.

Conclusions
In this study, we found the signi cant associations of ambient NO 2 concentration with the increase of mortality in patients with all tumors, lung cancer, liver cancer, colorectal cancer, thyroid cancer, and leukemia in Shenyang. Short-term exposure to NO 2 may induce a signi cantly increased risk of death in children with all tumors and leukemia. The higher risk of death in the patients with tumors and thyroid cancer in the young and middle-aged population should also be especially concerned due to the signi cant and stable correlation with the concentration of NO 2 ; In the elderly, NO 2 showed the signi cant risk for all tumors, lung cancer, colorectal cancer, bladder cancer, and leukemia. The mortality of male patients with lung cancer and leukemia was signi cantly correlated with the concentration of NO 2 , while for females, the death of all patients with tumor, liver cancer, colorectal cancer, and leukemia were signi cantly positively correlated with the increased short-term exposure of NO 2 . However, more researches are needed to better understand potential age and gender differences.