Trends in Incidence, Survival and Mortality of Gastric Cancer in the United States: A Population-Based Study, 2001-2015

Background: Gastric cancer remains one of the leading causes of death, and a burden of public health in the United States (US). The aim of the study was to provide updates to gastric cancer estimates, and analyzed the long-term trends in incidence, survival, and mortality of gastric cancer in the US, which was helpful for the monitoring of the screening program and the prevention strategy. Methods: The incidence, and long-term trends in incidence, survival, and mortality of gastric cancer in the US from 2001 to 2015 were analyzed. The data were obtained from the Surveillance, Epidemiology, and End Results (SEER) Database. Age-adjusted incidence rates were calculated, joinpoint regression and age-period-cohort analyses were conducted. All statistical tests were 2-sided. Results: The overall age-adjusted incidence of gastric cancer decreased over the study period, with an annual percentage change (APC) of -1.4% (95% confidence interval [CI] = -1.1 to 13.3; P < 0.001). The incidence rates leveled off at an earlier age (< 45 years) and creased obviously with age. The age rate deviations increased sharply before the age of 47.5 years (age rate deviation = 0.92; 95% CI = 0.71 to 1.13). The 5-year mortality attributed to gastric cancer declined from 65.98% to 56.29% over the study period. The trend of 5-year mortality attributed to gastric cancer showed no significant fluctuation. The hazard ratio for 5-year all-cause death increased with cancer stage from 1.22 (95% CI = 1.13 to 1.33; P < 0.001) to 4.71 (95% CI = 4.40 to 5.06; P < 0.001). Conclusion: During the study period, the incidence decreased, while the survival rate increased slightly. Specially, the trend of 5-year mortality attributed to gastric cancer did not vary significantly. The data demonstrated that the prognosis of gastric cancer in the US remained challenging.


Introduction
Results (SEER) Database. We highlighted the trends that can help to inform global and local interventions to lower the disease burden and, perhaps, curtail the increasing number of incident cases.

Data Source and Variables
Population-based data of cancer was available for the SEER program, which was publicly and freely accessible (https://seer.cancer.gov/). The data of patients diagnosed with gastric cancer from 2001 and 2015 were obtained from the SEER-9 Registries. The SEER-9 Registries included nine areas (Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, and Utah) and three races (white, black, and other), covering approximately 9.4% of the US population (Scott et al., 2020). The age-adjusted incidence, survival and mortality of gastric cancer were obtained via SEER*Stat software. Due to the absence of direct access to the survival and mortality of gastric cancer, we exported the individual patient data, which included survival outcomes, and calculated the survival and mortality rates of gastric cancer. The long-term trends of the incidence, survival and mortality of gastric cancer and the differences between subgroups [including sex, age, race (white, black, and other), stage, the year of diagnosis, and survival outcome (alive, dead attributed to gastric cancer)] were evaluated in this study.

Statistical Analysis
Age-adjusted gastric cancer incidence rates could be directly calculated by SEER*Stat software (version 8.3.9). Age-adjusted incidence rates were standardized according to the US standard population in 2000. To identify changes in gastric cancer incidence rate trends, joinpoint regression was conducted for every age, sex, stage, and race by using Joinpoint software (version 4.9.0). The software calculated the annual percentage change (APC) in incidence rates between trend-change points and average annual percentage change (AAPC) in the whole period studied. t test and Z test were respectively used to test whether the APC and AAPC were statistically different from zero. All statistical tests were 2-sided. The P-value was considered statistically significant when it was less than 0.05.
To help understand the differences in gastric cancer incidence rates and trends by age, age-period-cohort analysis was conducted by using a web tool for age-periodcohort analysis (Rosenberg et al., 2014). We divided gastric cancer incidence cases extracted from the SEER-9 Registries into 5-year age groups (ages 5-9 through ≥ 85 years), 5-year calendar periods (2001-2005, 2006-2010, 2011-2015), and 5-year birth cohorts.
Cox proportional hazards models were used to estimate the effect of different predefined factors on 5-year survival, including age, sex, race, stage, tumor site, and histologic grade, by R software (version 4.0.0).

Baseline Characteristics
Demographic and clinical characteristics of gastric cancer diagnosed from 2001 to 2015 in SEER-9 Registries are shown in Table 1. 32,029 patients have been diagnosed with gastric cancer during the study period. Among all cases, patients diagnosed with gastric cancer were mainly over the age of 45, and the number of patients in each age group increased with age. Patients were mainly male (61.4%) and white (68.49%). Except for the unknown stages, more than one-quarter (28.56%) of the patients were diagnosed at stage IV. 30.30% (9719/32029) of the gastric carcinomas occurred in gastric cardia. The average tumor size was 44 mm (interquartile range = 25 to 68).

Incidence Trends of Gastric Cancer
The overall and subgroups age-adjusted incidence rates of gastric cancer are shown in Supplemental Table 1. The overall age-adjusted incidence of gastric cancer decreased over the study period with an APC of -1.4% [95% CI = -1.1% to 13.3%; P < 0.001]. We further estimated the trends in different subgroups, including age, sex, stage, and race, presented in Figure 1 and Table 2.
The age-adjusted incidence rates in the 55 -64, 65 -74, > 75 age groups were all above the overall level and higher than those in the 15 -44 and 45 -64 age groups ( Figure 1A). There were significant decreases in the 65 -74, > 75 age groups (APC = -1.7% and -2.0%, respectively; P < 0.001 for both). In the > 75 age group, the incidence rate was the highest, while the decreased trend in incidence was the most obvious. The trends in other age groups leveled off during the study period (P > 0.05 for each group).
The age-adjusted incidence rates in both males and females decreased, but with a different incidence rate ( Figure 1B). The incidence rate in males decreased with an APC of -1.7% per year (95% CI = -1.9% to -1.4%; P < 0.001), which was higher than that in females (APC = -0.8%; 95% CI = -1.3% to -0.3%; P = 0.004). The incidence rates in males were all about one time higher than those in females in each year.
The trend of incidence was the most evident in stage IV compared with other stage groups ( Figure 1C). The incidence of the stage IV decreased with an APC of -1.6% (95% CI = -2.2% to -1.1%; P < 0.001). And the incidence rates of the stage IV were the highest from 2004 to 2015, compared with other stage groups. The incidence decreased significantly in white, black, and other population (APC = -1.2%, -1.7%, and -2.8%, respectively; P < 0.001 for each group). The incidence rates in white population were far lower than those in black and other population ( Figure 1D).
To explore the age distribution of gastric cancer diagnosed varied across sex, stage, and race groups, we also calculated age-specific incidence, presented in Figure  2 and Supplemental Table 2. The incidence rates leveled off at an earlier age (< 45 years) and creased obviously with age. Although gastric cancer incidence rates generally increased with age, the trends of incidence also varied by sex (Figure 2A), race ( Figure 2B), and stage ( Figure 2C) groups for patients aged 80 -84 years and ≥ 85 years. Compared with black and other population, the increase of incidence was smaller in white population. Except for the stage I A, the increases of incidence in other stage groups were smaller than before. Figure 2. D illustrated the results of age-period-cohort analysis: the age rate deviations and 95% CIs for gastric cancer incidence. The age rate deviations increased sharply before the age of 47.5 years (age rate deviation = 0.92; 95% CI = 0.71 to 1.13). The age rate deviations of incidence peaked at 47.5 years old and then decreased.

Trends of 5-year Survival of Gastric Cancer
The trends of 5-year survival were analyzed by age, sex, stage, and race (as shown in Figure 3 and Supplemental Table 3). The trends of 5-year survival varied similarly across groups, including age, sex, stage, and race, that is, the 5-year survival improved slightly before 2012 when the patients diagnosed and the peaked in 2012. The 5-year survival rate of patients diagnosed between 2014 and 2015 sharply declined to 0%. Among all age groups, the 5-year survival rate of the ≥ 75 years age group was the lowest, and those for other age groups were all higher than overall 5-year survival rate ( Figure  3A). For the long-term trends, the 5-year survival rates of each stage group from the lowest to the highest were IV, III B, III A, II, I B, and I A, respectively ( Figure 3C). The 5-year survival by sex and race increased continuously with little fluctuation, during 2001 through 2013 diagnosed ( Figure 3B and 3D). Additionally, the overall survival rates by different characteristics were shown in Supplemental Figure 1.

Trends of 5-year Mortality Attributed to Gastric cancer
The trends of 5-year mortality attributed to gastric cancer were analyzed by age, sex, stage, and race (as shown in Figure 4 and Supplemental Table 4). The trends of 5-year mortality attributed to gastric cancer were similar in each group and declined slowly over the period by     age, sex, stage, and race. The 5-year mortality for the 75 years age group was the highest among all age groups ( Figure 4A). The 5-year mortality in males was a little higher than that in females ( Figure 4B). The distribution of 5-year mortality in stage was obviously related to the stage ( Figure 4C). The 5-year mortality rates from the highest to the lowest were in stage IV, III B, III A,II, I B, and I A, respectively. The 5-year mortality in white population was relatively higher than those in black and other population ( Figure 4D).     (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003). To ensure consistent standards and reliable results, the data from 2004-2015 were used for analysis.
The effect of different characteristics on 5-year all-cause mortality was analyzed by Cox regression ( Table  3). The hazard ratio (HR) of 5-year all-cause mortality increased with age, but not statistically significant. The 5-year all-cause mortality in females was lower than that in males (HR = 0.91; 95% CI = 0.88 to 0.95; P < 0.001). Among stage groups, taking stage I A as the reference, the HR for 5-year all-cause mortality increased with stage from 1.22 (95% CI = 1.13 to 1.33; P < 0.001) to 4.71 (95% CI = 4.40 to 5.06; P < 0.001). Over the study period, the HR of death leveled off with the year. From 2004 to 2013, the HR of death decreased slightly, ranging from 0.58 to 0.68 (P < 0.001 for each group). Compared with the white population, the HR of the black population (HR = 1.12; 95% CI = 1.06 to 1.18; P < 0.001) was higher, while the HR of other population (HR = 0.86; 95% CI = 0.82 to 0.91; P < 0.001) was lower. Compared with tumors in gastric cardia, the HR of tumors at other sites varied similarly, ranging from 0.74 to 0.91 (P < 0.05 for each group). Taking grade I as the reference, the HR for other histologic grades ranged from 1.29 to 1.59 (P < 0.001 for each group).

Discussion
Gastric cancer is an important cause of morbidity and mortality in many parts or the world. According to the study, the decrease in incidence was observed in the US from 2001 to 2015, which were in accordance with the global trend. The overall incidence of gastric cancer decreased from 7.8% to 6.6% with an APC of -1.4%.
The APC of the decrease in the US was lower than in most countries of the world (Scott et al., 2020). The epidemiological characteristics of incidence over the study period were in accordance with previous global reports (Karimi et al., 2014;Colquhoun et al., 2015;Scott et al., 2020): the incidence rates increased continuously with age before 84 years old, either in overall, sex, stage, and race. The incidence rates in males were about one time higher than those in females, and the decrease of incidence in males was also higher than in females. Most of the gastric cancer patients were in advanced stages when diagnosed. The incidence rates of stage IV were higher than those in other stages. The incidence rates of black and other population were 1.5 -2 folds of the white. Additionally, 30.3% of the gastric carcinomas were in cardia, while 69.7% were non-cardia located.
The decreasing trend of gastric cancer incidence and mortality in most populations was due to the falling rates of non-cardia gastric cancer, which has been linked to the decline of H. pylori infection rates. (Etemadi et al., 2020) H. pylori was a known carcinogen for non-cardia gastric cancer, which most adults probably once infected during their life course (de Martel et al., 2013). From 1990 to 2020, gastric cancer dropped from the fourth leading incident cancer worldwide to the fifth, and from the second leading cause of cancer deaths to the fourth (following lung, colorectal, and breast cancers) (Sung et al., 2021). As a result, gastric cancer accounted for the fourth highest mortality cancer, however, this decline in burden relative to other cancers and the dramatic decline in age-standardized rates might not necessarily led to a lower burden of gastric cancer on the health systems in high-risk countries. That is due to the changes in the age structure and growth of the population, which means that numbers of incident cases and deaths of gastric cancer have continued to increase in many locations (Etemadi et al., 2020). Thus, a further decrease in the absolute number of cases and deaths could be possible if the rates in east Asia, where almost half of the incident cases and deaths occur, are further reduced (Etemadi et al., 2020). However, the increasing trend of incidence occurred in the younger population (age ≤ 44 years). Obesity has been confirmed to be a risk factor and related to the increasing incidence rates in western population (Petryszyn et al., 2020). The prevalence rates of H. pylori infection have decreased notably in the younger populations in most parts of the world. The increase in obesity among the younger population may be one of the main causes of the increasing trend in the younger population.
In previous literature (Smith et al., 2013;Li, 2020;Scott et al., 2020;Wong et al., 2021), the researchers used the age of 40 or 50 years as the cutoff point to analyze the incidence risk of gastric cancer in subgroups. They also used this cutoff point to design the screening strategy for early gastric cancer. In this study, the results of ageperiod-cohort analyses may support to take the age of 47.5 years as the exact age cutoff point adjusting period and cohort. Our findings were generally in accordance with those recommendation for cutoff point and it can help to provide guidance for more accurate screening and prevention for gastric cancer.
The prognosis of gastric cancer in the US remained challenge. The 5-year survival ranged from 20% to 30%, with a little increase in recent years. Compared with the 5-year survival of 20% in most regions of the world (Kim et al., 2016), there was no gap in 5-year survival in the US. On the contrary, the 5-year survival was higher in highincidence regions (for example, the survival in Japan and Korea approximately achieved 70%) (Nashimoto et al., 2013;Jung et al., 2018). The prognosis in high-incidence countries have improved significantly due to the adoption of population-based mass screening, which makes it possible for more and more patients to be diagnosed and treated at an early stage. In contrast, the incidence of gastric cancer when diagnosed at stage I A or I B ranged from 1.4% to 1.8% in the US.
The prognosis of gastric cancer strongly depended on the stage when diagnosed. The 5-year survival and mortality rates distributed differently among stage groups. The prognosis of gastric cancer could be worse when diagnosed in advanced stage, comparing with in early stage. In all stage groups, the incidence in stage IV was the highest, ranging from 2.3% to 2.8%, while the 5-year survival in stage IV was the lowest, ranging from 2.09% to 4.06%. On one hand, the early stage of gastric cancer generally present with minimal or no symptoms; on the other hand, the incidence of H. pylori infection and gastric cancer was stably low, the awareness of population was insufficient, and a national population-based screening program has not been established in the US. Those reasons resulted in the fact that the patients with gastric cancer in the US were more likely to be diagnosed at an advanced stage. That may provide one possible explanation for the current condition of little improvement in the prognosis in the US, although treatment has been greatly improved. Application of three-stage prevention was an effective strategy for reducing mortality and improving survival rate. The primary prevention was the test and eradication of H. pylori. The secondary prevention was endoscopic screening in at-risk populations. The population-based screening has been proved to be cost-effective in highincidence countries (i.e., Korea and Japan), but in the US, which was a low-incidence country, it was thought costly. (Casamayor et al., 2018) The cost-effective of populationbased screening in the US was controversial and needed to be further studied.
Particularly, it was found that the 5-year survival of cases diagnosed in 2014 and 2015 was zero. However, the 5-year mortality rates attributed to gastric cancer of cases diagnosed in 2014 and 2015 decreased lightly, which were 61.46% and 58.69%, respectively. It indicated that most of the deaths of the cases were caused by other reasons, instead of gastric cancer. In 2019 and 2020, the event with the greatest risk to health and life was the outbreak of COVID-19 pandemic. It was assumed that the condition of zero 5-year survival was related to the COVID-19 pandemic. Cancer patients with COVID-19 co-infection were at an increased risk of mortality. Data showed that gastrointestinal cancer were with the top three pooled COVID-19 in-hospital all-cause mortality rate (Venkatesulu et al., 2021).
For resectable gastric cancer, guidelines recommended endoscopic resection at an early stage. Additionally, guidelines recommended chemotherapy as the first-line therapy for advanced gastric cancer. For patients with stage II or III gastric cancer, guidelines recommended tumor resection/lymphadenectomy D2, and adjuvant chemotherapy. For resectable gastric cancer at stage IV, perioperative chemotherapy was recommended (Casamayor et al., 2018). Our study demonstrated that the decrease in mortality occurred in gastric cancer patients diagnosed at an early stage, which could likely reflect the improvements in curative treatment modalities, while the 5-year mortality attributed to gastric cancer remained at high levels (ranging from 56.29% to 65.98%) and declined slowly. Therefore, it was necessary to enhance the monitoring to improve the prognosis of gastric cancer, and increase investments to seek for better treatment methods.
In conclusion, the incidence of gastric cancer has been decreasing from 2001 to 2015. However, most of the patients with gastric cancer were still diagnosed at an advanced stage. Even though slight improvements of the prognosis of gastric cancer have been made, it was still far from satisfaction. In order to improve the prognosis of the gastric cancer and lower the burden of disease, it is necessary to establish a national, populationbased screening program and prevention strategy. Future iterations of the research by providing annual updates to regional and country-level gastric cancer estimates, would be helpful for the monitoring of the screening program and the prevention strategy.