Abstract
The causes of death in patients with gastric adenocarcinoma have not been well characterized. This nationwide population‐based cohort study included 56 240 patients diagnosed with gastric adenocarcinoma in 1970‐2014 in Sweden. We used competing‐risks regression to compare cause‐specific risks of death in patients with different characteristics and a multiple‐cause approach to assess proportions of deaths attributable to each cause. Among 53 049 deaths, gastric cancer was the main (77.7% of all deaths) underlying cause. Other major underlying causes were nongastric malignancies (8.0%), ischemic heart disease or cerebrovascular disease (6.5%), and respiratory diseases (1.4%). Risk of death from gastric cancer steadily decreased in patients with cardia adenocarcinoma over the study period, but remained relatively stable in patients with noncardia adenocarcinoma since the 1980s. Risk of death from other malignancies increased during later calendar periods (subhazard ratio [SHR] = 2.16, 95% confidence interval [CI] 1.97‐2.38, comparing 2001‐2014 with 1970‐1980). Compared with men, the risk of death in women with cardia adenocarcinoma was higher from gastric cancer (SHR = 1.18, 95% CI 1.10‐1.27), but lower from other malignancies (SHR = 0.80, 95% CI 0.71‐0.91). In multiple‐cause models, 60.4%‐71.2% of all deaths were attributable to gastric cancer and 9.5%‐12.1% to other malignancies. The temporal trends of cause‐specific risks from multiple‐cause models were similar to those of underlying causes. Our findings suggest that although most deaths in patients with gastric adenocarcinoma are due to gastric cancer, other causes of death are common. Patients with cardia adenocarcinoma face considerable increasing risk of death from other causes over time, particularly from other malignancies.
Keywords: cause of death, mortality, prognosis, stomach neoplasm, survival
Causes of death in patients with gastric adenocarcinoma have not been well described. In this nationwide population‐based cohort study of 56,240 patients, we found that most deaths in patients with gastric adenocarcinoma were due to gastric cancer, but other causes of death were common, and increasingly so in patients with cardia adenocarcinoma. Attention to common comorbidities may benefit the follow‐up and survival of patients with gastric adenocarcinoma.
1. INTRODUCTION
Gastric cancer is the sixth most common type of cancer worldwide, and gastric adenocarcinoma comprises over 90% of the total number of cases. 1 , 2 , 3 Gastric adenocarcinoma carries a poor prognosis with an overall 5‐year survival rate below 30% in European countries, after slight improvements in recent years. 4 , 5 , 6 Characterizing the specific causes of death in patients with gastric adenocarcinoma is essential for improving the treatment and follow‐up. However, to our knowledge, this has not been done previously in a population‐based setting.
Analyzing causes of death data is methodologically complex, particularly because of 2 main challenges. First, competing risks are often present in the analysis of survival data, ie the occurrence of the primary cause of death of interest is precluded by death from other causes. 7 Therefore, the analyses should ideally assess the influence of competing risks. Second, although the reporting of cause‐specific mortality is usually based on the so‐called underlying cause of death, it is debatable if it is appropriate to attribute each death to a single disease, particularly in an era of high prevalence of chronic diseases. 7 , 8 Furthermore, designating the underlying cause of death is often a subjective judgment of certifying physicians, which could lead to considerable rates of errors. 9 , 10 Therefore, the contribution of other causes recorded on a death certificate, in addition to the underlying causes, should also be evaluated.
Using data from nationwide health data registries with complete coverage and follow‐up, we assessed the causes of death in patients diagnosed with gastric adenocarcinoma in Sweden during more than 4 decades. To avoid bias, we used a competing‐risks model comparing cause‐specific risks of death in patients with different characteristics and applied a multiple‐cause approach assessing both underlying and contributing causes of death.
2. METHODS
2.1. Study design and participants
This was a population‐based cohort study including all patients diagnosed with gastric adenocarcinoma (International Classification of Diseases, 7th edition [ICD‐7] code 151; and histology code 096 according to WHO/HS/CANC/24.1 classification) in Sweden between 1 January 1970 and 31 December 2014. These patients were identified from the Swedish Cancer Registry, which has 98% national coverage of both cardia and noncardia gastric cancers and a rate of histological confirmation of virtually 100%. 11 , 12 We obtained information on causes of death of the cohort members through linkage to the Swedish Cause of Death Register, which covers all deaths of residents in Sweden with a nonreporting rate lower than 1%. 13 All participants were followed up from the date of gastric adenocarcinoma diagnosis until the date of death or the end of the study (31 December 2014), whichever occurred first. The unique personal identity number assigned to all Swedish residents enabled linkages of the individual participants’ data between the registers. The study was approved by the Regional Ethical Review Board in Stockholm, Sweden (Protocol Number: 2015/1916‐31/1).
2.2. Outcome ascertainment
We analyzed both underlying and contributing causes of deaths in patients with gastric adenocarcinoma. According to the definition from the WHO, the underlying cause of death refers to any disease or injury initiating the train of events leading directly to death, or the circumstances of the accident or violence producing the fatal injury; contributing causes are other significant conditions that have unfavorably influenced the course of the morbid process and thus contributed to the fatal outcome. 14 The specific ICD codes used for defining causes of death are listed in Tables S1 and S2. Causes of death were categorized into the following 5 major groups: (i) gastric cancer; (ii) malignancies other than gastric cancer; (iii) ischemic heart disease or cerebrovascular disease; (iv) nonmalignant diseases of the respiratory tract; and (v) other causes.
2.3. Statistical analysis
We assessed the distribution of underlying causes of death in patients with gastric adenocarcinoma by categories of duration of follow‐up (1 year or less, 1‐5 years, or more than 5 years), which was calculated from the date of gastric adenocarcinoma diagnosis to the end of follow‐up. Competing‐risks regression was used to compare the cause‐specific risks of death in patients with different characteristics. This method was preferred over the standard Cox proportional hazard regression because, in the latter method, competing events (deaths from other causes) are simply treated as censored observations. The competing‐risks regression is based on the method of Fine and Gray and estimates the cause‐specific cumulative incidence functions of death in the presence of competing risks. 15 , 16 The Fine and Gray competing‐risks model also provides subhazard ratios (SHRs), which describe the relative effect of covariates on the subdistribution hazard function. 17 Variables included in the model were sex (male or female), age at diagnosis (less than 50, 50‐59, 60‐69, 70‐79 years, or 80 years or older), and calendar period of diagnosis (1970‐1980, 1981‐1990, 1991‐2000, or 2001‐2014). Cumulative incidence functions of death from the 5 major groups of underlying causes within 5 years of diagnosis were calculated by calendar period.
In a multiple‐cause model, we quantitatively assessed the total number and proportion of deaths attributable to each cause according to different values of weight (ω) assigned to the underlying cause of death. 7 For each death event with a total of k recorded causes (including underlying and contributing causes), a given larger weight ω (eg 0.75) was assigned to the underlying cause, and the rest of the weight (1 − ω) was equally shared by other causes, ie the fraction of death attributable to each contributing cause being (1‐ω)/(k‐1). We then calculated the sum of fractions of death attributable to each cause from all deceased individuals and the corresponding proportion out of all deaths. In the absence of recorded contributing cause of death, the fraction for the underlying cause of death was given the value of 1. Three weighting approaches (ω = 0.75, 0.5, and ω = 1/k) were used separately. When ω = 1/k, it was assumed that all causes had contributed equally to the death. 7 , 18
We further separately analyzed the data in patients with cardia adenocarcinoma (ICD‐7 code 151.1) and those with noncardia adenocarcinomas. All statistical analyses were undertaken using SAS 9.4 for Windows (SAS Institute). A 2‐sided P value below 0.05 was considered statistically significant.
3. RESULTS
3.1. Participants
This study comprised 56 240 patients diagnosed with gastric adenocarcinoma, including 7542 (13.4%) patients with cardia adenocarcinoma and 48 698 (86.6%) patients with noncardia adenocarcinoma. A majority of patients were men (61.6%) and the male predominance was stronger for cardia adenocarcinoma (76.3%). The mean age at diagnosis was 71.4 years (SD, 11.6 years). More detailed information about the participants is presented in Table 1.
TABLE 1.
Distribution of basic characteristics of all patients with gastric adenocarcinoma diagnosed in Sweden, 1970‐2014
| Characteristics | Total gastric adenocarcinoma | Cardia adenocarcinoma | Noncardia adenocarcinoma | |||
|---|---|---|---|---|---|---|
| Number (%) | Person‐years | Number (%) | Person‐years | Number (%) | Person‐years | |
| Total | 56 240 (100) | 120 261 | 7542 (100) | 12 331 | 48 698 (100) | 107 930 |
| Gender | ||||||
| Male | 34 649 (61.6) | 71 574 | 5756 (76.3) | 9519 | 28 893 (59.3) | 62 055 |
| Female | 21 591(38.4) | 48 687 | 1786 (23.7) | 2812 | 19 805 (40.7) | 45 875 |
| Age at diagnosis, y | ||||||
| <50 | 2753 (4.9) | 11 774 | 428 (5.7) | 1040 | 2325 (4.8) | 10 734 |
| 50‐59 | 5556 (9.9) | 20 552 | 1023 (13.6) | 2617 | 4533 (9.3) | 17 934 |
| 60‐69 | 13 112 (23.3) | 36 444 | 2107 (27.9) | 4176 | 11 005 (22.6) | 32 268 |
| 70‐79 | 20 365 (36.2) | 37 821 | 2487 (33.0) | 3432 | 17 878 (36.7) | 34 389 |
| ≥80 | 14 454 (25.7) | 13 670 | 1497 (19.9) | 1066 | 12 957 (26.6) | 12 604 |
| Mean ± SD | 71.4 ± 11.6 | 69.3 ± 11.6 | 71.7 ± 11.5 | |||
| Calendar period at diagnosis | ||||||
| 1970‐1980 | 18 486 (32.9) | 37 934 | 1071 (14.2) | 1265 | 17 415 (35.8) | 36 669 |
| 1981‐1990 | 15 516 (27.6) | 37 590 | 1642 (21.8) | 2923 | 13 874 (28.5) | 34 667 |
| 1991‐2000 | 11 175 (19.9) | 26 589 | 1983 (26.3) | 3793 | 9192 (18.9) | 22 797 |
| 2001‐2014 | 11 063 (19.7) | 18 147 | 2846 (37.7) | 4350 | 8217 (16.9) | 13 796 |
| Length of follow‐up, y | ||||||
| ≤1 | 37 637 (66.9) | 10 861 | 4925 (65.3) | 1794 | 32 712 (67.2) | 9068 |
| 1‐5 | 12 425 (22.1) | 26 640 | 2052 (27.2) | 4186 | 10 373 (21.3) | 22 453 |
| >5 | 6178 (11.0) | 82 760 | 565 (7.5) | 6351 | 5613 (11.5) | 76 409 |
3.2. Underlying causes of death
A total of 53 049 (94.3%) patients with gastric adenocarcinoma died during follow‐up. Among these deaths, 36 940 (69.6%) occurred within 1 year of diagnosis and 48 333 (91.1%) occurred within 5 years of diagnosis. Table 2 shows the distribution of the underlying causes of death by duration of follow‐up. Gastric cancer was the underlying cause of death in 41 226 (77.7%) of all deceased patients. Malignancies other than gastric cancer were the second most common group (n = 4256, 8.0%). Other relatively common underlying causes of death were ischemic heart disease (n = 2595, 4.9%), cerebrovascular disease (n = 863, 1.6%), and nonmalignant respiratory diseases (n = 762, 1.4%). Among the 4716 patients who died after having survived for at least 5 years after gastric adenocarcinoma diagnosis, gastric cancer was a less common underlying cause of death (n = 976, 20.7%), while the frequencies of other common underlying causes of death were relatively higher, ie other malignancies (n = 770, 16.3%), ischemic heart disease (n = 816, 17.3%), cerebrovascular disease (n = 384, 8.1%), and nonmalignant respiratory diseases (n = 343, 7.3%). Separate analyses showed a lower proportion of deaths from gastric cancer and a higher proportion of other malignancies as the underlying causes of death in patients with cardia adenocarcinoma than in noncardia adenocarcinoma patients (Tables S3 and S4).
TABLE 2.
Distribution of underlying causes of death by duration of follow‐up in patients with gastric adenocarcinoma diagnosed in Sweden, 1971‐2014
| Mortality status | Duration of follow‐up, y | All | ||
|---|---|---|---|---|
| ≤1 | 1‐5 | >5 | ||
| Total a | 56 240 (100) | 18 603 (100) | 6178 (100) | 56 240 (100) |
| Alive throughout follow‐up b , c | 697 (1.2) | 1032 (5.6) | 1462 (23.7) | 3191 (5.7) |
| All deaths c | 36 940 (65.7) | 11 393 (61.2) | 4716 (76.3) | 53 049 (94.3) |
| Cause‐specific death d | ||||
| Gastric cancer | 31 134 (84.3) | 9116 (80.0) | 976 (20.7) | 41 226 (77.7) |
| Malignancies other than gastric cancer | 2518 (6.8) | 968 (8.5) | 770 (16.3) | 4256 (8.0) |
| Infectious and parasitic diseases | 29 (0.1) | 26 (0.2) | 59 (1.3) | 114 (0.2) |
| Nonmalignant blood diseases | 15 (0.04) | 10 (0.1) | 13 (0.3) | 38 (0.1) |
| Endocrine, nutritional, and metabolic diseases | 34 (0.1) | 26 (0.2) | 81 (1.7) | 141 (0.3) |
| Diabetes | 24 (0.1) | 20 (0.2) | 62 (1.3) | 106 (0.2) |
| Dementia and Alzheimer’s disease | 4 (0.01) | 16 (0.1) | 141 (3.0) | 161 (0.3) |
| Ischemic heart disease | 1337 (3.6) | 442 (3.9) | 816 (17.3) | 2595 (4.9) |
| Cerebrovascular disease | 314 (0.9) | 165 (1.5) | 384 (8.1) | 863 (1.6) |
| Hypertensive disease | 23 (0.1) | 7 (0.1) | 30 (0.6) | 60 (0.1) |
| Nonmalignant diseases of respiratory system | 292 (0.8) | 127 (1.1) | 343 (7.3) | 762 (1.4) |
| Chronic obstructive pulmonary disease | 87 (0.2) | 30 (0.3) | 116 (2.5) | 233 (0.4) |
| Influenza and pneumonia | 167 (0.5) | 73 (0.6) | 180 (3.8) | 420 (0.8) |
| Nonmalignant diseases of digestive system | 314 (0.9) | 65 (0.6) | 131 (2.8) | 510 (1.0) |
| Nonmalignant diseases of urinary system | 35 (0.1) | 29 (0.3) | 60 (1.3) | 124 (0.2) |
| External causes, including suicide | 33 (0.1) | 32 (0.3) | 97 (2.1) | 162 (0.3) |
| Suicide | 14 (0.04) | 5 (0.04) | 13 (0.3) | 32 (0.1) |
| Other specified | 828 (2.2) | 359 (3.2) | 811 (17.2) | 1998 (3.8) |
Data are shown as n (%).
Patients alive at the beginning of the follow‐up window.
Patients alive at the beginning of the follow‐up window who did not die during the specific period of follow‐up.
Percentage of all patients alive at the beginning of the follow‐up window.
Percentage of all deaths occurring during the specific period of follow‐up.
Among all deaths with nongastric malignancies as underlying causes (n = 4256), 2586 (60.8%) deaths were due to cancers of digestive organs, 262 (6.2%) to lung cancer, and 256 (6.0%) to prostate cancer (Table 3). The distribution of nongastric malignancy‐related deaths varied between patients with cardia adenocarcinoma and those with noncardia adenocarcinoma. Esophageal cancer was the most commonly recorded underlying cause of death from nongastric malignancies (n = 1406, 20.9% of all deaths) in patients with cardia adenocarcinoma. In patients with noncardia adenocarcinoma, common nongastric malignant underlying causes of death were colorectal cancer (13.6% of all deaths from nongastric malignancies), pancreatic cancer (11.2%), lung cancer (9.3%), prostate cancer (8.7%), and esophageal cancer (7.3%; Table 3).
TABLE 3.
Number (%) of deaths with malignancies other than gastric cancer as underlying causes in patients diagnosed with gastric adenocarcinoma in Sweden, 1970‐2014
| Anatomic site | Total gastric adenocarcinoma | Cardia adenocarcinoma | Noncardia adenocarcinoma |
|---|---|---|---|
| Total | 4256 (100) | 1600 (100) | 2656 (100) |
| Lip, oral cavity, and pharynx | 25 (0.6) | 2 (0.1) | 23 (0.9) |
| Digestive organs | 2586 (60.8) | 1481 (92.6) | 1105 (41.6) |
| Esophagus | 1601 (37.6) | 1406 (87.9) | 195 (7.3) |
| Colon, rectum, and anus | 383 (9.0) | 23 (1.4) | 360 (13.6) |
| Liver and biliary passages | 177 (4.2) | 12 (0.8) | 165 (6.2) |
| Pancreas | 316 (7.4) | 19 (1.2) | 297 (11.2) |
| Lung (including trachea and bronchus) | 262 (6.2) | 15 (0.9) | 247 (9.3) |
| Breast | 124 (2.9) | 4 (0.3) | 120 (4.5) |
| Uterus | 29 (0.7) | 1 (0.1) | 28 (1.1) |
| Ovary | 48 (1.1) | 2 (0.1) | 46 (1.7) |
| Prostate | 256 (6.0) | 24 (1.5) | 232 (8.7) |
| Bladder | 73 (1.7) | 10 (0.6) | 63 (2.4) |
| Kidney | 47 (1.1) | 3 (0.2) | 44 (1.7) |
| Thyroid | 8 (0.2) | 0 (0) | 8 (0.3) |
| Hodgkin lymphoma | 4 (0.1) | 0 (0) | 4( 0.2) |
| Non‐Hodgkin lymphoma | 63 (1.5) | 3 (0.2) | 60 (2.3) |
| Leukemia | 60 (1.4) | 0 (0) | 60 (2.3) |
| Others and unspecified | 671 (15.8) | 55 (3.4) | 616 (23.2) |
3.3. Competing‐risks regression
The estimated SHRs and their 95% confidence intervals (CIs) of cause‐specific death within 5 years of gastric adenocarcinoma diagnosis in the competing‐risks regression are presented in Table S5. Compared to male patients, female patients had slightly higher risk of death from gastric cancer (SHR 1.08; 95% CI, 1.05‐1.10), but lower risks of death from other malignancies (SHR 0.75; 95% CI, 0.70‐0.81) and ischemic heart disease or cerebrovascular disease (SHR 0.72; 95% CI, 0.65‐0.81). Separate analyses for cardia and noncardia adenocarcinomas (Tables S6 and S7) showed that the sex difference was limited to patients with cardia adenocarcinomas in deaths from gastric cancer (SHR 1.18; 95% CI, 1.10‐1.27) and other malignancies (SHR 0.80; 95% CI, 0.71‐0.91). The reduced risk of death from respiratory diseases in female patients was limited to non‐cardia adenocarcinoma patients (SHR 0.67, 95% CI 0.51‐0.87). Risks of death from ischemic heart disease or cerebrovascular disease and respiratory diseases increased with age, whereas no such age pattern was observed for death from gastric cancer or from other malignancies (Table S5).
Patients who were diagnosed in more recent calendar periods had lower risk of death from gastric cancer than those diagnosed in earlier periods; patients diagnosed in 2001‐2014 had decreased risk of death from gastric cancer than those diagnosed in 1970‐1980 (SHR 0.72; 95% CI, 0.70‐0.74). Patients diagnosed in later calendar periods had higher risk of death from other malignancies (Table S5 and Figure 1). The decreased risk of death from gastric cancer and the increased risk of death from other malignancies were more pronounced in patients with cardia adenocarcinoma than in those with noncardia adenocarcinoma, showing steady changes over the whole study period (Table S6). Patients diagnosed with noncardia adenocarcinomas in 1981‐1990 had lower risk of death from gastric cancer than those diagnosed in 1970‐1980 (SHR 0.86; 95% CI, 0.84‐0.88), but the risk showed little reduction in patients diagnosed after the 1980s. The risk of death from malignancies other than gastric cancer in patients with noncardia adenocarcinoma was stable before the year 2000, but increased in 2001‐2014 (SHR 1.27; 95% CI, 1.11‐1.46, compared with 1970‐1980; Table S7). Risks of death from ischemic heart disease or cerebrovascular disease and respiratory disease were highest in patients diagnosed in 1981‐1990 and decreased over time thereafter (Table S5 and Figure 1).
FIGURE 1.
Cumulative incidence functions and subhazard ratios (SHRs) with 95% confidence intervals (CIs) for major causes of death by calendar period at diagnosis of gastric adenocarcinoma in Sweden, 1970‐2014, from competing‐risks regressions
3.4. Multiple‐cause model
Gastric cancer was recorded as either an underlying or a contributing cause of death in 83.2% of all deaths among patients with gastric adenocarcinoma; other malignancies were recorded as causes of death in 27.7%, respiratory diseases in 10.5%, ischemic heart disease in 10.3%, and cerebrovascular disease in 4.2% of all deaths (Table S8).
Table 4 presents the numbers and percentages of deaths attributable to each cause according to different values of weight (ω) assigned to the underlying cause of death. When ω = 0.75, 71.2% of all deaths were attributable to gastric cancer, 9.5% to other malignancies, 6.6% to ischemic heart disease or cerebrovascular disease, and 2.4% were attributable to respiratory diseases. With the increasing value of ω, the number of deaths attributable to gastric cancer decreased, and that of other causes increased. Under the “equal weights” approach, ie ω = 1/k, gastric cancer accounted for 60.4% of all deaths, and other malignancies accounted for 12.1%. Separate analyses for cardia and noncardia adenocarcinomas showed similar patterns, except for substantially lower proportions of deaths attributable to gastric cancer (49.5%‐60.3%) but higher proportions attributable to other malignancies (24.8%‐26.4%) in patients with cardia adenocarcinoma (Tables S9 and S10).
TABLE 4.
Number (%) of deaths attributable to each cause according to different values of weight (ω) assigned to the underlying cause of death, Sweden, 1970‐2014
| Cause of death | ω = 0.75 | ω = 0.50 | ω = 1/k a |
|---|---|---|---|
| Gastric cancer | 37 763 (71.2) | 34 301 (64.7) | 32 012 (60.4) |
| Malignancies other than gastric cancer | 5021 (9.5) | 5786 (10.9) | 6409 (12.1) |
| Infectious and parasitic diseases | 156 (0.3) | 197 (0.4) | 220 (0.4) |
| Nonmalignant blood diseases | 105 (0.2) | 172 (0.3) | 216 (0.4) |
| Endocrine, nutritional, and metabolic diseases | 256 (0.5) | 371 (0.7) | 448 (0.8) |
| Diabetes | 195 (0.4) | 284 (0.5) | 337 (0.6) |
| Dementia and Alzheimer’s disease | 171 (0.3) | 181 (0.3) | 185 (0.3) |
| Ischemic heart disease | 2592 (4.9) | 2589 (4.9) | 2439 (4.6) |
| Cerebrovascular disease | 885 (1.7) | 907 (1.7) | 889 (1.7) |
| Hypertensive disease | 102 (0.2) | 144 (0.3) | 177 (0.3) |
| Nonmalignant diseases of respiratory system | 1266 (2.4) | 1769 (3.3) | 1988 (3.8) |
| Chronic obstructive pulmonary disease | 279 (0.5) | 324 (0.6) | 349 (0.7) |
| Influenza and pneumonia | 775 (1.5) | 1130 (2.1) | 1243 (2.3) |
| Nonmalignant diseases of digestive system | 813 (1.5) | 1117 (2.1) | 1289 (2.4) |
| Nonmalignant diseases of urinary system | 180 (0.3) | 237 (0.4) | 278 (0.5) |
| External causes, including suicide | 365 (0.7) | 568 (1.1) | 786 (1.5) |
| Suicide | 31 (0.1) | 31 (0.1) | 30 (0.1) |
| Other specified | 3335 (6.3) | 4672 (8.8) | 5674 (10.7) |
This “equal weights” approach assumes that all causes have contributed equally to each death, where k is the total number of causes (including underlying and contributing causes).
The temporal trends in the proportions of deaths attributable to groups of causes in multiple‐cause models had similar patterns to those observed for underlying causes in the competing‐risks regressions. The proportion of deaths attributable to gastric cancer decreased over time and ranged from 67% to 75% in 1970‐1980 compared with 55% to 71% in 2001‐2014, depending on the value of weight (ω) assigned to the underlying cause. The proportion attributable to other causes, particularly nongastric malignancies, increased over time (Figure 2).
FIGURE 2.
Proportions of deaths attributable to major groups of causes according to different values of weight (ω) assigned to the underlying cause by calendar period at diagnosis of gastric adenocarcinoma in Sweden, 1970‐2014. The “equal weights” approach (ωnonbreakingspace= 1/k) assumes that all causes contribute equally to each death, where k is the total number of causes, including underlying and contributing causes
4. DISCUSSION
This study found that gastric cancer was the main cause of death in short‐term survivors of gastric adenocarcinoma, but accounted for a smaller proportion of deaths in long‐term survivors. Other major causes of death were nongastric malignancies, particularly cancers of the digestive organs, as well as ischemic heart disease, cerebrovascular disease, and respiratory diseases. The cause‐specific risks of death varied between patients with cardia and noncardia adenocarcinoma. Patients diagnosed in more recent calendar periods had a decreased risk of death from gastric cancer and were more likely to die from other causes, particularly from other malignancies, compared with those diagnosed in earlier periods.
To the best of our knowledge, this study is the largest and the first population‐based study examining the causes of death in patients with gastric adenocarcinoma. The complete nationwide coverage and high quality of data from Swedish health data registers lend validity to the findings. The comprehensive survival analyses used in this study, including the competing‐risks model and multiple‐cause model, took into account both underlying and contributing causes of death. Nevertheless, these methods have some limitations. The competing‐risks model considers only the underlying causes of death, whereas the multiple‐cause model does not solve the “competing risks” problem. 7 The magnitude of the SHRs obtained in the competing‐risks model should be interpreted with caution because the SHRs cannot directly quantify the strengths of the associations between predictors and cause‐specific risk of mortality. 17 In the multiple‐cause model, the choice of weights assigned to the underlying cause of death was arbitrary. Finally, this study had limited statistical power in some subgroup analyses of patients with the sublocation cardia adenocarcinoma.
Few studies have investigated the causes of death in patients diagnosed with gastric cancer. A hospital‐based study of 172 selected postoperative patients with gastric cancer in Italy and a similar study of 323 postoperative patients in China both showed that tumor recurrence was the most common cause of death, and other common causes of death were other primary cancers, cardiovascular disease, cerebrovascular disease, and respiratory diseases. 19 , 20 These observations are consistent with the findings of the present study, although the specific proportions of deaths due to each cause were not comparable because of substantial differences in patient characteristics, analytic methods, and sample sizes. No previous study has assessed the temporal trends in the cause‐specific deaths or had adequate statistical power to undertake robust subgroup analyses.
The present study suggests that gastric cancer remains the predominant cause of death in patients with gastric adenocarcinoma, particularly within the first 5 years of diagnosis. During the last 3‐4 decades, a steadily decreased risk of death from gastric cancer over time was seen in patients with cardia adenocarcinoma, whereas no such strong reduction was found for noncardia adenocarcinoma. These findings are consistent with a recent Swedish population‐based analysis, which found that the relative survival improved for cardia adenocarcinoma, but not for noncardia adenocarcinoma, during 1990‐2017. 4 The decreased disease‐specific mortality in patients diagnosed with cardia adenocarcinoma in recent decades could be attributable to a better selection of patients for curative treatment, increased use of neoadjuvant therapies, introduction of multidisciplinary team meetings, and centralization of services. 21 , 22 , 23 The lack of improved disease‐specific survival after noncardia adenocarcinoma in recent years in Sweden warrants closer investigation, but inadequate centralization of treatment could be one explanation.
Interestingly, the present study shows increasing risk of death from other malignancies over time in gastric adenocarcinoma patients. These malignancies accounted for 14%‐19% of all deaths in the latest period, 2001‐2014, in the multiple‐cause analyses. A large proportion of such deaths was due to cancers of other digestive organs, the lung and prostate. These findings could be explained by shared risk factors for gastric adenocarcinoma and other malignancies, eg tobacco smoking, dietary factors, and possibly treatments given to cancer patients, eg irradiative effects from radiotherapy. 2 , 3 , 24 The evidence concerning the role of Helicobacter pylori infection, the strongest risk factor for noncardia gastric adenocarcinoma, in the development of other cancers remains controversial. 25 , 26 Notably, esophageal cancer was recorded as the underlying cause for around one‐fifth of deaths in patients with cardia adenocarcinoma, which is likely, to some extent, to be due to misclassification of these tumors. 11 , 12 However, the increasing risk of death from nongastric malignancies over time remained when analyses were restricted to patients with gastric noncardia adenocarcinoma, where tumor misclassification is not a major issue.
The higher risk of death from gastric cancer but lower risk of death from other malignancies in female patients, compared with male patients, is less likely to be explained by any misclassification in causes of death, which would have affected both sexes in the same manner. Thus, there should be an underlying mechanism, but this remains to be identified. The observed sex differences and age‐dependent patterns in risks of death from ischemic heart disease or cerebrovascular disease and respiratory disease are in line with corresponding cause‐specific mortality patterns in the general population. 27
Despite all efforts to improve diagnostic procedures and therapy, the prognosis following diagnosis in patients with gastric adenocarcinoma has not improved much in European countries, with an overall 5‐year survival still below 30% (18% in Sweden). 4 , 5 , 6 One possible explanation might be the increasing risks of death from other causes in addition to gastric adenocarcinoma recurrence, especially in an era when chronic diseases are increasingly common in an ageing population. Patients with gastric adenocarcinoma might be more vulnerable to death from other coexisting conditions, eg due to increased rates of immunodeficiency, malnutrition, and systemic inflammation. 28 Our multiple‐cause analyses revealed that 29%‐45% of all deaths were due to causes other than gastric cancer in patients diagnosed after the year 2000. These findings highlight the need to consider comorbidities in the clinical management of gastric adenocarcinoma patients for an improved overall prognosis.
In summary, this nationwide and population‐based Swedish study reveals that, although patients diagnosed with gastric adenocarcinoma mainly die from this cancer, the disease‐specific mortality has steadily decreased during the last 3‐4 decades in patients with cardia adenocarcinoma. Patients diagnosed with gastric adenocarcinoma face considerable increasing risk of death from other causes over time, particularly from other malignancies.
DISCLOSURE
The authors declare that they have no conflict of interest.
Supporting information
Supplementary Material
ACKNOWLEDGMENTS
This study was supported by the Swedish Research Council (grant number 521‐2014‐2536) and Swedish Cancer Society (grant numbers 190043 and 180684). The funding bodies had no role in the study design, the collection, analysis, and interpretation of data, or the writing of the article and the decision to submit it for publication.
Xie S‐H, Chen H, Lagergren J. Causes of death in patients diagnosed with gastric adenocarcinoma in Sweden, 1970‐2014: A population‐based study. Cancer Sci. 2020;111:2451–2459. 10.1111/cas.14441
REFERENCES
- 1. Global Burden of Disease Cancer Collaboration , Fitzmaurice C, Akinyemiju TF, et al. Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability‐adjusted life‐years for 29 cancer groups, 1990 to 2016: a systematic analysis for the global burden of disease study. JAMA Oncol. 2018;4(11):1553‐1568. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Karimi P, Islami F, Anandasabapathy S, Freedman ND, Kamangar F. Gastric cancer: descriptive epidemiology, risk factors, screening, and prevention. Cancer Epidemiol Biomarkers Prev. 2014;23:700‐713. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Rugge M, Genta RM, Di Mario F, et al. Gastric cancer as preventable disease. Clin Gastroenterol Hepatol. 2017;15(12):1833‐1843. [DOI] [PubMed] [Google Scholar]
- 4. Asplund J, Kauppila JH, Mattsson F, Lagergren J. Survival trends in gastric adenocarcinoma: a population‐based study in Sweden. Ann Surg Oncol. 2018;25:2693‐2702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Dassen AE, Lemmens V, van de Poll‐Franse LV, et al. Trends in incidence, treatment and survival of gastric adenocarcinoma between 1990 and 2007: a population‐based study in the Netherlands. Eur J Cancer. 2010;46:1101‐1110. [DOI] [PubMed] [Google Scholar]
- 6. Lagergren J, Mattsson F. Diverging trends in recent population‐based survival rates in oesophageal and gastric cancer. PLoS ONE. 2012;7:e41352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Moreno‐Betancur M, Sadaoui H, Piffaretti C, Rey G. Survival analysis with multiple causes of death: extending the competing risks model. Epidemiology. 2017;28:12‐19. [DOI] [PubMed] [Google Scholar]
- 8. Omran AR. The epidemiologic transition: a theory of the epidemiology of population change. Milbank Q. 1971;2005(83):731‐757. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. McGivern L, Shulman L, Carney JK, Shapiro S, Bundock E. Death certification errors and the effect on mortality statistics. Public Health Rep. 2017;132:669‐675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Johansson LA, Westerling R. Comparing Swedish hospital discharge records with death certificates: implications for mortality statistics. Int J Epidemiol. 2000;29:495‐502. [PubMed] [Google Scholar]
- 11. Lindblad M, Ye W, Lindgren A, Lagergren J. Disparities in the classification of esophageal and cardia adenocarcinomas and their influence on reported incidence rates. Ann Surg. 2006;243:479‐485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Ekstrom AM, Signorello LB, Hansson LE, Bergstrom R, Lindgren A, Nyren O. Evaluating gastric cancer misclassification: a potential explanation for the rise in cardia cancer incidence. J Natl Cancer Inst. 1999;91:786‐790. [DOI] [PubMed] [Google Scholar]
- 13. Brooke HL, Talbäck M, Hörnblad J, et al. The Swedish cause of death register. Eur J Epidemiol. 2017;32:765‐773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. World Health Organization . International Statistical Classification of Diseases And Related Health Problomes. 10th revision. Volume 2 Instruction manual. Geneva: World Health Organization, 2010. [Google Scholar]
- 15. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496‐509. [Google Scholar]
- 16. Dignam JJ, Zhang Q, Kocherginsky M. The use and interpretation of competing risks regression models. Clin Cancer Res. 2012;18:2301‐2308. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Austin PC, Fine JP. Practical recommendations for reporting fine‐gray model analyses for competing risk data. Stat Med. 2017;36:4391‐4400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Piffaretti C, Moreno‐Betancur M, Lamarche‐Vadel A, Rey G. Quantifying cause‐related mortality by weighting multiple causes of death. Bull World Health Organ. 2016;94:870‐879. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Guadagni S, Catarci M, Kinoshita T, Valenti M, De Bernardinis G, Carboni M. Causes of death and recurrence after surgery for early gastric cancer. World J Surg. 1997;21:434‐439. [DOI] [PubMed] [Google Scholar]
- 20. Huang B, Wang Z, Xing C, Sun Z, Zhao B, Xu H. Long‐term survival results and prognostic factors of early gastric cancer. Exp Ther Med. 2011;2:1059‐1064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Young JA, Shimi SM, Kerr L, McPhillips G, Thompson AM. Reduction in gastric cancer surgical mortality over 10 years: an adverse events analysis. Ann Med Surg (Lond). 2014;3:26‐30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. van de Poll‐Franse LV, Lemmens VE, Roukema JA, Coebergh JW, Nieuwenhuijzen GA. Impact of concentration of oesophageal and gastric cardia cancer surgery on long‐term population‐based survival. Br J Surg. 2011;98:956‐963. [DOI] [PubMed] [Google Scholar]
- 23. Cunningham D, Allum WH, Stenning SP, et al. Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med. 2006;355:11‐20. [DOI] [PubMed] [Google Scholar]
- 24. Rapiti E, Fioretta G, Verkooijen HM, et al. Increased risk of colon cancer after external radiation therapy for prostate cancer. Int J Cancer. 2008;123:1141‐1145. [DOI] [PubMed] [Google Scholar]
- 25. Franceschi F, Zuccala G, Roccarina D, Gasbarrini A. Clinical effects of Helicobacter pylori outside the stomach. Nat Rev Gastroenterol Hepatol. 2014;11:234‐242. [DOI] [PubMed] [Google Scholar]
- 26. Chen Y, Segers S, Blaser MJ. Association between Helicobacter pylori and mortality in the NHANES III study. Gut. 2013;62:1262‐1269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Weires M, Bermejo JL, Sundquist K, Sundquist J, Hemminki K. Socio‐economic status and overall and cause‐specific mortality in Sweden. BMC Public Health. 2008;8:340. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Ockenga J, Valentini L. Review article: anorexia and cachexia in gastrointestinal cancer. Aliment Pharmacol Ther. 2005;22:583‐594. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material