The impact of implementing current treatment modalities and female sex on gastric cancer outcomes, 2000–2016: a longitudinal nationwide cohort study

Abstract

Background

The implementation of current treatment modalities and their impact on nationwide gastric cancer outcomes remain poorly understood. Biological differences between females and males could impact survival. We aimed to analyze rates of gastric surgery, chemotherapy, and radiotherapy as well as changes in overall survival among gastric cancer patients diagnosed between 2000–2008 and 2009–2016, respectively, in Finland.

Material and methods

Data on gastric cancer patients were collected from national registries. Cox regression analysis and the Kaplan–Meier method were used to analyze differences in survival.

Results

We identified 9223 histologically confirmed gastric cancer patients. The rate of gastric surgery decreased from 44% (n = 2282) to 34% (n = 1368; p < 0.001). The proportion of gastric surgery patients who underwent preoperative oncological treatment increased from 0.5% (n = 12) to 16.2% (n = 222) between the calendar periods (p < 0.001) and stood at 30% in 2016. The median overall survival (OS) improved from 30 months [95% confidence interval (CI) 28–33] to 38 months (95%CI 33–42; p = 0.006) and the period 2009–2016 independently associated with a lower risk of death [hazard ratio (HR) 0.78, 95%CI 0.70–0.87] among patients who underwent gastric surgery. Females exhibited a lower risk of death (HR 0.88, 95%CI 0.81–0.97) among patients who underwent gastric surgery.

Conclusion

Preoperative oncological treatment was gradually introduced into clinical practice and OS among gastric surgery patients improved. Moreover, female surgical patients exhibited a better survival than male patients.

Keywords:

• Gastric cancer
• epidemiology
• survival

Background

Gastric cancer is one of the most common cancers and among the leading causes of cancer-related mortality globally. Specifically, gastric cancer carries a wide variability in incidence and prognosis, particularly when comparing Eastern and Western countries [1–5]. A declining trend in overall gastric cancer incidence has been observed in many countries [6]. For instance, the incidence of non-cardia gastric cancer has declined in the US and some European countries, while the incidence of the cardia subtype has increased [7,8]. According to the Finnish Cancer Registry (FCR) between 2000 and 2019, the incidence of gastric cancer in Finland decreased from 18.6 per 100 000 to 10.2 per 100 000 [5]. While screening for gastric cancer is common in high-incidence countries such as Japan and South Korea, no population-wide screening protocol exists in Finland [9,10]. In the CONCORD-3 study, which encompasses population-based registries from 71 countries, the five-year survival rate for gastric cancer ranged from 20% to 40% in most countries, holding steady at 25.5% in Finland for the period 2010–2014 [3]. Patient characteristics including cancer stage at diagnosis serve as potential explanations for country-specific differences [11,12]. Another factor influencing regional variation is the prevalence of Helicobacter pylori infection, as well as lifestyle and environmental factors [1,13]. In addition, previous research has shown that the prognosis of gastric cancer in females differs from that in males [14,15].

Treatment and outcome reports from nationwide cohorts in Western countries remain rare. In the last two decades, the treatment of gastric cancer has evolved as new oncological treatment protocols have proven beneficial on outcomes, including reports from the INT‐0116 (2001) and MAGIC (2006) trials [16–20]. Currently, the European Society for Medical Oncology recommends perioperative chemotherapy for stage IB or more advanced resectable gastric cancer [21]. Furthermore, surgery has evolved, given that evidence favors D2 lymphadenectomy in Western populations given its association with lower locoregional recurrence and gastric cancer–related death rates compared with D1 surgery as shown in the Dutch D1D2 trial [22]. Reports from the Netherlands and France have shown that the centralization of surgical treatment has increased resection rates, reduced postoperative mortality, and improved survival [23,24]. In Finland, the extent of utilization of modern oncological treatment protocols as well as D2 lymphadenectomies and minimally invasive surgery for gastric cancer are unknown [25,26]. Finland, with a population of 5.5 million, has a comprehensive public healthcare system. High-quality national registers in Finland provide excellent nationwide longitudinal data for research [27,28]. Here, we aimed to analyze the rates of gastric surgery, chemotherapy, and radiotherapy as well as changes in overall survival among gastric cancer patients diagnosed between 2000–2008 and 2009–2016, respectively, in Finland.

Material and methods

Patients diagnosed with gastric cancer (ICD-10 code C16) from 2000 through 2016 were identified from the FCR. All healthcare visits were gathered from the National Institute for Health and Welfare’s Care Register for Health Care (HILMO), which provides nationwide data on patients discharged from inpatient care, day surgeries, and specialized outpatient care facilities, as well as avoHILMO, a register of primary healthcare visits. Surgery, radiotherapy (RT), and chemotherapy (CT) were identified based on the Nordic Medico-Statistical Committee Classification of Surgical Procedures [29]. In addition, oncological treatment data from the Helsinki and Uusimaa Hospital District were collected from local registers. Datasets were compared to determine the completeness of the HILMO data and were then combined. Data on residence as well as time and cause of death data were collected from Statistics Finland. Follow-up data were available until the end of 2020 and patients still living were censored in the survival analyses.

We excluded patients under the age of 18 at diagnosis (n = 3) and patients for whom the diagnosis relied only on the death certificate (n = 16), an autopsy (n = 886), or clinical examination (no histology; n = 124). Based on histology, we excluded patients with neuroendocrine tumors (n = 556), gastrointestinal stromal tumors (n = 139), unspecified neoplasms (n = 496), and other tumors with a rare histology including mesenchymal tumors (n = 81).

Data on total and partial gastrectomies as well as endoscopic mucosal or submucosal resections were collected, including the procedure type and time of surgery as well as emergency and elective surgeries. Open and minimally invasive surgeries were analyzed together. In this study, we grouped patients who underwent a partial or total gastrectomy together under gastric surgery. Postoperative mortality was calculated at 30 and 90 days following surgery.

Patients who underwent elective exploratory surgery, either open or laparoscopic, within 100 days of diagnosis, but who underwent no gastric resection were also identified. This represented a patient group initially treated with a radical intent, but found to be nonresectable. We labeled this group diagnostic surgery/nonresectable.

Data were analyzed in three groups based on surgical treatment: gastric surgery, diagnostic surgery/nonresectable, and no surgery. By comparing the timing of CT and/or RT and surgery, we further stratified patients into the following oncological treatment subgroups. In the gastric surgery group, the subgroups were perioperative (CT and/or RT before and after surgery), preoperative only (CT and/or RT only before surgery), postoperative only (CT and/or RT only after surgery), and none (no CT or RT). In the diagnostic surgery/nonresectable group, the subgroups were preoperative (CT and/or RT before diagnostic surgery), postoperative (CT and/or RT only after diagnostic surgery), and none (no CT or RT). In the no surgery group, the subgroups were palliative (any CT and/or RT) and none (no CT or RT). Information about specific CT agents was unavailable. In addition, there was no information available regarding whether CT and/or RT was intended as curative, downstaging, or palliative.

The cancer stage classification used by the FCR differs from the TNM classification [30]. Patients were reclassified into the following categories: local, locally advanced, metastatic, and unknown (Supplementary Table 1).

We used the Charlson comorbidity index (CCI) to weight comorbidities [31–33]. CCI was determined based on the healthcare visits preceding the gastric cancer diagnosis at any point during the study period.

To examine changes in treatment protocols and survival among patients, we compared groups based on the year of diagnosis using two time periods: 2000–2008 versus 2009–2016. The presumed clinical implementation of perioperative treatment was one of the factors used to select the cut-off time. In addition, it divided the research period approximately in half. However, many changes in treatment have occurred gradually over time and overlapping across districts. We calculated overall survival (OS), censoring patients who were living as of 31 December 2020.

All statistical analyses were performed using IBM’s SPSS Statistics for Windows (version 27.0.1; IBM Corporation, Armonk, NY, USA) or R (version 4.0.0; Foundation for Statistical Computing, Vienna, Austria). Changes in treatment group proportions were evaluated using the linear-by-linear association, Fisher’s exact test or the chi-square test. The Mann-Whitney U test was used to analyze nonparametric continuous variables. Survival among different groups was calculated using the Kaplan–Meier method and differences were compared using the log-rank test. The Bonferroni correction was applied to multiple comparisons to adjust the p values.

Cox’s proportional hazards model was used for analyses of prognostic factors among patients who underwent gastric surgery. For the preoperative only group, classification occurred on the day of surgery and, for the postoperative only and perioperative groups, classification was based on the beginning of postoperative CT and/or RT. Follow-up began at the first treatment. In 151 cases, preoperative treatment began more than 12 months before gastric surgery or the delay in initiating postoperative treatment following gastric surgery exceeded six months. Given that our focus was specifically on analyzing the impact of primary treatment, these patients were excluded from the model. Additionally, patients with an unknown stage were excluded, resulting in a total of 2482 patients included in the multivariable model. The immortal time bias was corrected for oncological treatments using time-dependent variables to classify patients into different oncological treatment groups in a timely manner. The Cox regression assumption of constant hazard ratios (HRs) over time (proportional hazards) was assessed using the Schoenfeld residuals plotted over time and testing for a trend. Time-dependent variables were used for age, stage, and the CCI score to account for small deviations from the proportional hazard assumption. We also considered interactions in the multivariate models.

For all analyses, we considered p < 0.05 as statistically significant and used two-tailed tests. This study was approved by the National Institute for Health and Welfare, Statistics Finland, and Helsinki University Hospital.

Results

Cohort characteristics

The proportion of postmortem diagnoses decreased from 9.3% (n = 609/5922) in 2000–2008 to 5.9% (n = 293/4700) in 2009–2016, p < 0.001. A total of 9223 histologically confirmed gastric cancer patients were included in further analyses. The follow-up period lasted until the end of 2020, at which point 8219 patients (89.1%) had died, while 1004 patients (10.9%) remained living. The median follow-up was 32.5 (interquartile range [IQR] 12.7–92.8) months among gastric surgery patients, 6.3 (IQR 3.4–11.9) months among diagnostic surgery/nonresectable patients, and 4.8 (IQR 1.8–13.5) months among no surgery patients. Table 1 summarizes the baseline patient characteristics. Diagnosis was based on the histology of the primary tumor in 8895 patients (96.4%) and the histology of a metastasis in 328 patients (3.6%). Overall, 5291 patients were male (57.3%) and 3932 were female (42.6%).

Table 1. Baseline characteristics of gastric cancer patients diagnosed in 2000–2008 and 2009–2016 in Finland (n = 9223).

	2000–2008		2009–2016
	n (%)		n (%)		p
Total cases	5174		4049
Median age	72	IQR 61–79	71	IQR 63–80	0.001
Female	2280	(44)	1652	(41)	0.004
Catchment area					0.005
Helsinki	1306	(25)	1132	(28)
Tampere	1154	(22)	883	(22)
Turku	1021	(20)	737	(18)
Kuopio	955	(19)	692	(17)
Oulu	707	(14)	581	(14)
Charlson comorbidity index					<0.001
0	2899	(56)	1786	(44)
1	1181	(23)	904	(22)
2	643	(12)	647	(16)
3+	451	(9)	712	(18)
Tumor localization					<0.001
Cardia	940	(18)	845	(21)
Non-cardia	525	(10)	1609	(40)
Unknown	3709	(72)	1595	(39)
Histology^1					<0.001
Adenocarcinoma, diffuse	1468	(28)	1360	(34)
Adenocarcinoma, intestinal	664	(13)	1006	(25)
Adenocarcinoma, unspecified	3042	(59)	1683	(42)
Stage					<0.001
Local	869	(17)	342	(8)
Locally advanced	728	(14)	410	(10)
Metastatic	2243	(43)	1387	(34)
Unknown	1334	(26)	1910	(47)
Surgery
Gastric surgery	2282	(44)	1368	(34)	<0.001
Elective^2	1957	(86)	1159	(85)
Emergency^2	316	(14)	201	(15)	0.464
Diagnostic surgery/nonresectable	453	(9)	202	(5)	<0.001
No surgery	2439	(47)	2480	(61)	<0.001

IQR: interquartile range.

¹According to Lauren et al. [47]. ²The frequency within the gastric surgery group.

The proportion of patients with metastatic disease increased significantly from 58.4% (n = 2243, in 2000–2008) to 64.8% (n = 1387, in 2009–2016) considering only patients with a known stage of disease (p < 0.001). Comparing the stage groups between 2000–2008 and 2009–2016, there was no difference in the median age among local (74 vs 73, p = 0.493), locally advanced (70 vs 69, p = 0.832) and metastatic stages (69 vs 69, p = 0.103), however patients with an unknown stage were older (75 vs 73, p = 0.019). Furthermore, analyzing the mean CCI score between 2000–2008 and 2009–2016, scores were higher in the later period across all stage groups: local (1.12 vs 1.45, p = 0.003), locally advanced (0.76 vs 1.08, p = 0.002), metastatic (0.61 vs 1.01, p < 0.001) and unknown stage (0.91 vs 1.42, p < 0.001).

Surgeries performed and postoperative mortality

We identified 1187 (32.5%) partial gastrectomies and 2398 (65.7%) total gastrectomies. Comparing the periods 2000–2008 and 2009–2016, the proportion of laparoscopic subtotal gastrectomies increased from 1.1% (n = 8) to 11.5% (n = 57; p < 0.001). In 2016, 24% (n = 12) of subtotal gastrectomies were laparoscopic. Three patients underwent endoscopic mucosal resection, with two of these patients subsequently undergoing surgery.

Among patients who underwent gastric surgery, 30-day postoperative mortality was 2.0% (n = 46) in 2000–2008 versus 2.6% (n = 35) in 2009–2016 (p = 0.297). In addition, 90-day postoperative mortality was 6.0% (n = 136) in 2000–2008 and 6.5% (n = 89) in 2009–2016 (p = 0.552).

Furthermore, the 30-day postoperative mortality was 1.6% (n = 31) in 2000–2008 and 1.6% (n = 19) in 2009–2016 among patients who underwent elective gastric surgery (p = 0.906), and 4.7% (n = 15) vs. 8.0% (n = 16) among patients who underwent emergency gastric surgery during the two time periods (p = 0.134). Moreover, among patients who underwent elective gastric surgery, the 90-day postoperative mortality remained consistent at 4.7% (n = 92) in 2000–2008 and 4.7% (n = 54) in 2009–2016 (p = 0.957). In contrast, among patients who underwent emergency gastric surgery, the 90-day postoperative mortality was 13.9% (n = 44) in 2000–2008 and 17.4% (n = 35) in 2009–2016 (p = 0.282).

The 30-day mortality rates among gastric surgery patients further stratified by other oncological treatment administered (Table 2) were 0% (n = 0) in the perioperative, 1.8% (n = 1) in the preoperative-only, 0% (n = 0) in the postoperative-only, and 2.8% (n = 80) in the none subgroups. Similarly, the 90-day mortality rates were 0.6% (n = 1) in the perioperative, 12.3% (n = 7) in the preoperative-only, 1.0% (n = 6) in the postoperative-only, and 7.4% (n = 211) in the none subgroups.

Table 2. Surgical and other oncological treatments for gastric cancer among patients diagnosed in 2000–2008 and 2009–2016 (n = 9223).

	Total		2000–2008		2009–2016
Surgery and other oncological treatments^1	n	(%)	n	(%)	n	(%)	p^2
Gastric surgery	3650	(40)	2282	(44)	1368	(34)	<0.001
Perioperative	177	(5)	7	(0)	170	(12)	<0.001
Preoperative only	57	(2)	5	(0)	52	(4)	<0.001
Postoperative only	575	(16)	275	(12)	300	(22)	<0.001
None	2841	(78)	1995	(87)	846	(62)	<0.001
Diagnostic surgery / nonresectable	655	(7)	453	(9)	202	(5)	<0.001
Preoperative	11	(2)	1	(0)	10	(5)	<0.001
Postoperative	98	(15)	45	(10)	53	(26)	<0.001
None	545	(83)	407	(90)	138	(68)	<0.001
No surgery	4918	(53)	2439	(47)	2479	(61)	<0.001
Palliative	824	(17)	165	(7)	659	(27)	<0.001
None	4095	(83)	2274	(93)	1821	(74)	<0.001

¹Chemotherapy and/or radiotherapy. ²Bonferroni-adjusted p values.

All treatment modalities and changes over the study period

Table 2 summarizes the gastric surgery, diagnostic surgery/nonresectable, and no surgery groups, further stratifying them into subgroups based on the oncological treatment received. The rate of gastric surgery (44% vs 34%, p < 0.001) and diagnostic surgery/nonresectable (9% vs 5%, p < 0.001) decreased substantially between the two time periods. Comparing the time periods, the median age remained unchanged among the gastric surgery (71 vs 70, p = 0.251), diagnostic surgery/nonresectable (68 vs 69, p = 0.078) and no surgery (74 vs 72, p = 0.519) groups. Furthermore, the proportion of gastric surgery patients aged 80 years or older was 18% (n = 416) in 2000–2008 and 21% (n = 283) in 2009–2016 (p = 0.068). However, the mean CCI scores were higher in the later period among gastric surgery (0.82 vs 1.22, p < 0.001), diagnostic surgery/nonresectable (0.44 vs 0.86 p < 0.001) and no surgery (0.84 vs 1.29, p < 0.001) groups.

The rate of preoperative oncological treatment among gastric surgery patients increased from 0.5% (n = 12) to 16.2% (n = 222) between the two time periods (p < 0.001). In addition, the annual rate of preoperative treatment rose from 7.3% (n = 14) in 2009 to 30% (n = 136) in 2016. The proportion of gastric surgery patients receiving preoperative oncological treatment in 2006–2016 stratified by university hospital districts appears in Supplementary Figure 1. In Helsinki, the rate of preoperative treatment among gastric surgery patients was high throughout 2009–2016 (range 25–50%), subsequently increasing in other districts.

The proportion of patients who underwent RT increased from 2.0% (n = 104) in 2000–2008 to 6.7% (n = 270) in 2009–2016 (p < 0.001). More specifically, among patients with cardia tumors, RT increased from 2.6% (n = 24) to 8.3% (n = 70) (p < 0.001) across the time periods. Among patients with non-cardia tumors, RT increased from 1.5% (n = 57) to 6.5% (n = 103) (p = 0.190) when comparing the time periods.

Furthermore, the proportion of patients who underwent CT increased from 6.8% (n = 352) in 2000–2008 to 27.8% (n = 1124) in 2009–2016 (p < 0.001). When oncological treatments were compared between local patient register data and HILMO data, 79.6% of patients who received such treatment according to local register data also received that treatment according to the HILMO data. The patients who underwent chemotherapy were older (median age 66 vs 64, p < 0.001) and had more comorbidities (mean CCI 1.37 vs 0.82, p < 0.001) in the later time period.

Survival and predictors

For all patients, median survival was 10.4 months [95% confidence interval (CI) 9.8–11.0] in 2000–2008 and 10.9 months (95% CI 10.2–11.5) in 2009–2016 (p = 0.725). Overall survival (OS) among the gastric surgery, diagnostic surgery/nonresectable, and no surgery groups appears in Figure 1, while Table 3 summarizes the comparison of OS stratified by treatment (Table 3A) and stage of disease (Table 3B).

Figure 1. Kaplan–Meier for overall survival of gastric cancer patients in Finland among the following treatment groups: (A) gastric surgery, (B) diagnostic surgery/nonresectable, and (C) no surgery, comparing the time periods 2000–2008 and 2009–2016. The Bonferroni correction was used to adjust the log-rank test p values.

Table 3. Overall survival among gastric cancer patients diagnosed in 2000–2008 and 2009–2016 stratified by (A) surgery and other oncological treatment and (B) stage (n = 9223).

	2000–2008					2009–2016
Surgery and other oncological treatment^1	1 y	3 y	5 y	Mdn	(95% CI)	1 y	3 y	5 y	Mdn	(95% CI)	P^2
A
Gastric surgery	0.74	0.46	0.36	30	(28–33)	0.81	0.51	0.40	38	(33–42)	0.006
Perioperative	0.86	0.43	0.29	29	(0–69)	0.94	0.64	0.49	56	(36–75)	0.486
Preoperative only	0.80	0.40	0.20	28	(0–61)	0.71	0.50	0.42	31	(3–59)	1.000
Postoperative only	0.87	0.47	0.32	33	(28–37)	0.84	0.42	0.29	30	(26–33)	1.000
None	0.73	0.46	0.36	30	(27–33)	0.77	0.51	0.39	39	(33–45)	0.180
Diagnostic surgery/nonresectable	0.23	0.01	0.00	6	(6–7)	0.27	0.05	0.03	7	(6–7)	0.098
Preoperative	0.00	0.00	0.00	11	NC	0.60	0.00	0.00	12	(11–13)	1.000
Postoperative	0.33	0.04	0.00	10	(9–12)	0.42	0.13	0.07	9	(6–13)	1.000
None	0.22	0.01	0.00	6	(5–6)	0.20	0.01	0.01	6	(4–7)	1.000
No surgery	0.24	0.09	0.06	4	(4–5)	0.29	0.11	0.08	6	(5–6)	<0.001
Palliative	0.49	0.10	0.06	12	(9–14)	0.47	0.14	0.09	12	(11–13)	1.000
None	0.23	0.09	0.06	4	(4–4)	0.23	0.10	0.08	4	(3–4)	1.000

Table

	2000–2008					2009–2016
Stage	1 y	3 y	5 y	Mdn	(95% CI)	1 y	3 y	5 y	Mdn	(95% CI)	P^2
B
Local	0.82	0.61	0.52	68	(57–79)	0.80	0.63	0.55	77	(56–99)	0.904
Locally advanced	0.77	0.42	0.30	29	(25–32)	0.74	0.40	0.27	25	(21–29)	1.000
Metastatic	0.26	0.07	0.04	6	(5–6)	0.27	0.09	0.07	5	(5–6)	1.000
Unknown	0.40	0.21	0.16	8	(7–9)	0.49	0.25	0.19	12	(11–12)	<0.001

1 y: 1 year; 3 y: 3 years; 5 y: 5 years; Mdn: median in months; CI: confidence interval; NC: not calculated. ¹Chemotherapy and/or radiotherapy. ²Bonferroni-adjusted p values.

Comparing 2000–2008 to 2009–2016, patients with a diffuse-type histology had an OS of 10.6 (95% CI 9.5–11.6) versus 11.1 (95% CI 10.1–12.2) months (p = 0.260). Similarly, patients with an intestinal-type histology had an OS of 15.0 (95% CI 12.3–17.7) versus 13.7 (95% CI 12.2–15.3) months (p = 0.361).

The Cox multivariate analysis of survival among patients who underwent gastric surgery appears in Table 4. We adjusted the multivariate analysis for age, sex, stage, CCI, and catchment area. We observed a possible interaction between age and stage as well as sex and stage. After adjustment, the period 2009–2016 associated with a lower risk of death (HR 0.78) compared with the period 2000–2008. Postoperative treatment (CT and/or RT) increased OS compared with no CT or RT. Patients who underwent only preoperative treatment exhibited a worse OS (HR 2.23) compared with patients who did not receive any CT or RT. We detected no significant differences in survival between catchment areas.

Table 4. Cox regression analysis of the risk of death among patients who underwent gastric surgery. A hazard Ratio (HR) > 1 indicates a higher risk of death (n = 2482)^a.

	Univariate				Multivariable
			95% CI				95% CI
	p	HR	Lower	Upper	p	HR	Lower	Upper
Age at diagnosis^2
		1.00				1.00
≥72 years	<0.001	1.34	1.20	1.50	<0.001	1.40	1.25	1.59
Sex
Male		1.00				1.00
Female	0.004	0.91	0.84	1.00	0.007	0.88	0.81	0.97
Year of diagnosis
2000–2008		1.00				1.00
2009–2016	0.051	0.90	0.82	1.00	<0.001	0.78	0.70	0.87
Catchment area
Helsinki		1.00				1.00
Turku	0.084	0.89	0.78	1.02	0.316	0.93	0.82	1.07
Tampere	0.555	0.96	0.85	1.09	0.554	0.96	0.84	1.10
Kuopio	0.578	0.96	0.84	1.11	0.690	0.97	0.84	1.12
Oulu	0.131	0.90	0.77	1.03	0.324	0.93	0.80	1.08
Stage^1
Localized		1.00				1.00
Regional	<0.001	2.90	2.44	3.44	<0.001	3.02	2.53	3.59
Distant	<0.001	7.33	6.14	8.74	<0.001	7.35	6.13	8.80
Charlson comorbidity index^2
0	<0.001	1.59	1.39	1.83	<0.001	1.44	1.25	1.65
1		1.00				1.00
2	<0.001	1.33	1.15	1.53	0.001	1.27	1.10	1.47
≥3	<0.001	1.38	1.17	1.62	<0.001	1.37	1.16	1.62
Surgery
Open subtotal gastrectomy		1.00				1.00
Laparoscopic subtotal gastrectomy	0.108	0.66	0.39	1.10	0.343	0.77	0.45	1.32
Total gastrectomy	<0.001	0.85	0.78	0.93	0.986	1.00	0.91	1.10
Elective surgery		1.00				1.00
Emergency surgery	<0.001	2.23	1.99	2.51	<0.001	1.89	1.67	2.13
Other oncological treatment^3
None		1.00				1.00
Perioperative	0.007	0.63	0.45	0.88	0.369	0.85	0.59	1.22
Preoperative only	0.806	1.07	0.62	1.85	0.005	2.23	1.28	3.89
Postoperative only	0.814	1.02	0.88	1.17	0.012	0.82	0.71	0.96

HR: hazard ratio; CI: confidence interval. ¹In 151 cases, the preoperative treatment began more than 12 months before gastric surgery or the delay in initiating postoperative treatment following gastric surgery exceeded six months. These patients as well as patients with an unknown stage were excluded from the model. ²Age at diagnosis, stage, and Charlson comorbidity index exhibited minor deviations from the Cox regression assumption of a constant hazard over time. A time-dependent correction term (HR 1.08/year for age, 0.92/year for regional stage, 0.81/year for distant stage, and 0.92/year for CCI = 0) was used to account for this deviation. Possible interactions between age and stage as well as sex and stage, respectively, were detected in the multivariable model, but these were ignored.

^aChemotherapy and/or radiotherapy.

Sex-specific differences

A significantly higher proportion of females (41.3%, n = 1625) underwent gastric surgery than males (38.3%, n = 2025; p = 0.003). The proportion of females was 44.5% (n = 1625) among gastric surgery patients, 44.3% (n = 290) among diagnostic surgery patients and 41.0% (n = 2017) among no surgery patients (p = 0.003). The stages of disease among males were as follows; 13.3% (n = 703) local, 12.4% (n = 654) locally advanced, 39.6% (n = 2093) metastatic, and 34.8% (n = 1841) unknown. Among females, the composition of the stages of disease were as follows: 12.9% (n = 508) local, 12.3% (n = 484) locally advanced, 39.1% (n = 1537) metastatic, and 35.7% (n = 1403) unknown (p = 0.834). The 30-day postoperative mortality was statistically similar between female (1.8%, n = 30) and male patients (2.5%, n = 51) (p = 0.177). Moreover, we detected no difference in 90-day postoperative mortality between female (5.9%, n = 96) and male patients (6.4%, n = 129) (p = 0.580). Comparing patients with a known tumor location, cardia was a significantly more common tumor location among male (52.8%, n = 1297) than among female patients (33.4%, n = 488) (p < 0.001). A diffuse histology was also more common among females (37.6%, n = 1478) than among male patients (25.5%, n = 1350) (p < 0.001), while there was no difference in the proportion of intestinal histology between female (17.5%, n = 689) and male patients (18.5%, n = 981) (p = 0.219). Finally, female patients (HR 0.88) exhibited a lower risk of death (HR 0.88, 95% CI 0.81–0.97) compared with male patients in the multivariable analysis (Table 4).

Discussion

The centralization of gastric cancer surgery has reportedly increased resection rates in the Netherlands, rising slightly from 37.6% in 2009–2011 to 39.6% in 2013–2015 [23]. However, our nationwide register study reveals a different trend: the proportion of patients undergoing diagnostic surgery (nonresectable) or gastric surgery decreased from the period 2000–2008 to the period 2009–2016. One potential explanation for this decrease lies in the observed shift towards more advanced stage of disease at diagnosis. However, the rate of an unknown stage among patients was remarkably high, especially during the 2009–2016 period. Moreover, CCI scores were higher in the later period, as well as in stage-specific comparisons, indicating that patients had experienced a greater burden of chronic illness upon presentation, possibly preventing curative-intent treatment. A similar trend in higher CCI scores was observed across all treatment groups. It is plausible that improved staging, along with a reduction in palliative surgeries, may have contributed to the lower rate of surgery and the observed shift towards more advanced stages. We observed no significant change in the proportion of emergency procedures among gastric surgeries (Table 1). Stricter patient selection may have impacted the increase in overall survival among those who underwent gastric surgery. Furthermore, overall survival increased across all treatment groups (Figure 1), but remained unchanged across all patients, suggesting that more patients were ineligible for treatment with a curative intent.

Currently, there is no national guideline for the treatment of gastric cancer in Finland, and the extent to which modern treatment modalities have been implemented across different regions in Finland is unknown. A recent Finnish, medium-volume single-center study found that multimodal therapy combined with minimally invasive surgery improved treatment outcomes in operable gastric cancer patients during the period from 2005 to 2021 [26]. In that study, the proportion of gastric cancer patients who underwent perioperative treatment was 7.7% (n = 5/65) in 2005–2010, 20.7% (n = 12/58) in 2011–2015, and 63.8% (n = 37/58) in 2016–2021. By contrast, in Norway, preoperative chemotherapy was recommended in the national guidelines for patients with resectable gastric cancer as early as 2006, and, in 2007–2008, 43% of the operated patients received neoadjuvant treatment [34]. According to their data, most patients with resectable disease were deemed unsuitable for perioperative treatment due to factors such as an advanced age and co-existing medical conditions. We, however, observed a gradual increase in the implementation of perioperative treatment over time with encouraging outcomes among patients who underwent perioperative treatment in addition to surgery. However, preoperative combined with postoperative treatment in addition to surgery did not associate to a significantly different risk of death when compared with undergoing surgery alone in the multivariable analysis. Furthermore, patients in the preoperative only group exhibited an increased risk of death (HR 2.23), possibly due to complications preventing adjuvant treatment or due to aggressive disease progression. As such, we observed a high 90-day postoperative mortality among gastric surgery patients in the preoperative only and none (no CT or RT) subgroups. The results of the multivariable model should however be carefully interpreted as both curative-intent and noncurative patients were included and there is most likely mixing with the diagnostic surgery/non-resectable group. In the MAGIC trial, five-year survival was 36.3% for patients undergoing perioperative ECF (epirubicin, cisplatin, and 5-fluorouracil) CT [18]. Recent reports have indicated that FLOT (docetaxel, oxaliplatin, leucovorin, and 5-fluorouracil) appears superior to perioperative ECF/ECX CT, where median OS was 50 versus 35 months, respectively [20,35,36]. The benefit derived from the neoadjuvant and adjuvant components remains to be established.

In the Dutch D1D2 trial, while five-year survival was 45% and 47% for D1 and D2 lymphadenectomy groups, respectively, after a follow-up of 15 years, a D2 lymphadenectomy associated with a lower rate of locoregional recurrence and gastric cancer–related death rates than D1 surgery [22,37]. Unfortunately, in our study, the extent of lymphadenectomy performed could not be determined. We anticipate that the FINEGO study will provide further insight into this specific issue [38]. In our data, however, the five-year survival rates were 36% for the period 2000–2008 and 40% for 2009–2016 among patients who underwent gastric surgery. In a Japanese institution–based nationwide study from 2001–2007, the five-year OS rates among resected patients reached 71%; however, that study consisted of younger patients, primarily diagnosed at a less advanced stage of disease [39]. Interestingly, in a retrospective Japanese study, elderly patients less frequently underwent adjuvant therapy and exhibited poorer outcomes in stage III disease [40]. This agrees with our study, given that patients >72 years of age who underwent gastric surgery exhibited a significantly higher risk of death during the follow-up period. Notably, considerable variation exists in the treatment of elderly gastric cancer patients, as demonstrated by a study comparing treatment in Belgium, Denmark, the Netherlands, Norway, and Sweden [41].

The levels of postoperative mortality remained unchanged across the time periods with a 30-day mortality of 1.6% and a 90-day mortality of 4.7% among elective surgical patients. Postoperative mortality was higher following emergency surgeries but we detected no significant difference across the time periods. In the Netherlands, 30-day postoperative mortality rates appeared to decrease significantly from 6.5% in 2009–2011 to 4.1% in 2013–2015, while 90-day mortality fell from 10.6% to 7.2% for the same time periods [23]. In a nationwide study from France, 30-day postoperative mortality was 4.9%, while 90-day postoperative mortality climbed to 8.6% in 2010–2012 [24].

The proportion of patients receiving any CT or RT remained low between 2000 and 2008. Similar trends were observed in a US register study from 1991 to 2009, when the majority of stage IB–IV patients underwent surgery only from 2006 through 2009 [42]. The improvement in survival in the no surgery group lies in the increased proportion of patients receiving systemic palliative treatment. In a Korean single-institution study, Koo et al. reported an improvement in survival from 9.6 months in 2000–2003 to 11.7 months in 2008–2011 among patients with metastasized and recurrent gastric cancer who received CT [43].

The incidence of gastric cancer is higher in male patients, with increasing evidence suggesting a sexual dimorphism to cancer biology [6]. Notably, a meta-analysis of the effects of estrogen on gastric cancer found that longer exposure to the effects of estrogen, either ovarian or exogenous, may reduce the risk of gastric cancer [44]. In our study, females who underwent gastric surgery had a lower risk of death (HR 0.88) compared with males. In addition, a significantly higher proportion of females underwent gastric surgery than males. Interestingly, we detected no difference in postoperative mortality or stage of disease, although female patients had more often non-cardia tumors and diffuse histology. A recent Dutch study found that female patients had a higher rate of poorly differentiated and diffuse-type tumors and a lower postoperative morbidity compared with male patients [45]. Similar to our findings, a single-center report from South Korea found that, among gastric cancer patients who underwent surgical treatment, female patients experienced a better OS than male patients [15]. In a single-center study, Kim et al. reported that young females (≤45 years) exhibited a worse survival compared with male patients, particularly in cases of signet ring cell carcinoma [14]. They also identified gender differences in the Lauren histological type and tumor location, consistent with our findings since female patients had a higher proportion of diffuse-type and gastric body tumors.

Diffuse-type gastric adenocarcinoma exhibits distinct characteristics that are not yet fully understood [46]. The reduction of Helicobacter pylori infection, known to be effective for intestinal-type gastric cancer, does not seem to have the same effectiveness for diffuse-type tumors. Although a large proportion of cases with a non-specific adenocarcinoma histology interfered with our analysis, it was found that a diffuse histology became more prevalent in the later time period. A study from the Netherlands reported a disproportionate decrease in the incidences of histological subtypes as the incidence of intestinal-type gastric cancer decreased from 4.81 to 1.83 per 100 000, with diffuse type falling from 3.12 to 1.99 per 100 000 in the period 1989–2015 [47].

The strengths of this study include its reliance on a comprehensive long-term nationwide dataset for gastric cancer patients. Furthermore, we relied on a large sample size with excellent coverage across the entire Finnish population. In addition, our data were collected from high-quality registers [27,28]. According to a quality assessment by Leinonen et al. FCR had an estimated completeness of 96% for solid malignant tumors diagnosed in 2009–2013 [28]. In another study, the estimated completeness of FCR was 91.8% for esophageal cancer in 1990–2014 when compared with HILMO [48]. In stomach cancer, FCR's completeness rate was 87%, whereas HILMO's was 92.7% [49]. Furthermore, the completeness of HILMO has been evaluated as good, especially in relation to inpatient data [27].

The limitations of this study include the secondary nature of register data. The accuracy of stage data in FCR has been reported as suboptimal and the stage of disease is difficult to accurately record in FCR [28,50]. The high proportion of unknown stage diagnoses in 2009–2016 implies a possible downward trend in the quality of FCR staging data specifically. In addition, a substantial proportion of patients were excluded due to postmortem diagnoses, including diagnoses based on autopsy reports and death certificates although the proportion of post-mortem diagnosis decreased over time. For comparison, the corresponding proportion in the SURVMARK-2 population-based study was 1.2% for the period 2012–2014 [12]. During 2000–2008, the utilization of CT remained notably low. Particularly among diagnostic surgery/non-resectable cases, controversy surrounds the reality that patients who were eligible for diagnostic surgical interventions did not receive subsequent oncological treatment. This raises concerns regarding the possibility of incomplete oncological data. FCR includes unstructured crude oncological treatment data, but this was ignored due to its reported poor accuracy [50]. Patients who received oncological therapy but who never underwent exploratory or gastric surgery could not be distinguished from patients receiving palliative care in the no surgery group. We also acknowledge that gastric surgeries include both operations with radical intent and palliative resections. We may have missed endoscopic radical-intent operations due to register code ambiguity of endoscopic procedures. However, the detection of early gastric cancer reportedly remains rare in Finland [25]. CCI 0 (vs CCI 1) was associated with a higher risk of death, which could be attributed to undiagnosed illnesses or incomplete data.

Conclusion

To conclude, patients who underwent perioperative treatment and gastric surgery exhibited encouraging outcomes. However, a larger proportion of patients presented with metastatic disease upon diagnosis and a diminishing proportion of patients underwent diagnostic or gastric surgery when comparing the time periods 2000–2008 and 2009–2016. The Charlson comorbidity index scores were higher in the later time period indicating that patients had more chronic illnesses at presentation. In addition, female patients had a lower risk of death among patients who underwent gastric surgery. Greater efforts are needed to improve the early detection of gastric cancer.

Abbreviations
OS	=	overall survival
CI	=	confidence interval
FCR	=	Finnish Cancer Registry
HILMO	=	Care Register for Health Care
CCI	=	Charlson comorbidity index
IQR	=	interquartile range
HR	=	hazard ratio
Mdn	=	median
y	=	year

Acknowledgments

We thank Vanessa Fuller, from Language Services at the University of Helsinki, for proofreading our manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Due to legal restrictions, the data underlying this study cannot be shared.

Additional information

Funding

This work was supported by the Finnish Cancer Society (HS and CH), the Helsinki University Hospital Research Fund (CH, PP, MS, and HS), the Sigrid Juselius Foundation (CH and HS), Finska Läkaresällskapet (MS), Medicinska Understödsföreningen Liv och Hälsa (MS), and the Georg C and Mary Ehrnrooths Foundation (HS). None of the funders participated in the study.