Chronological improvement of survival in patients with advanced gastric cancer over 15 years

Abstract

Background:

Recent trials have reported a median overall survival (OS) of 11–17 months in patients with advanced gastric cancer (AGC). However, it is unclear how recently approved drugs contribute to patient prognosis.

Objectives:

We aimed to evaluate the characteristics and survival in patients with AGC over the past 15 years.

Design:

Retrospective study.

Methods:

We evaluated data of 1355 patients with AGC who received first-line chemotherapy between January 2005 and March 2019 at a single institution. We compared the characteristics and survival rates across four periods: January 2005–December 2007 (period A), January 2008–February 2011 (period B), March 2011–May 2015 (period C), and June 2015–March 2019 (period D). The median follow-up duration was 13.1 months, with 312, 333, 393, and 317 patients in periods A, B, C, and D, respectively.

Results:

There were no significant differences in patient characteristics between the four periods, except for the proportion of patients who underwent prior gastrectomy and human epidermal growth factor receptor 2 (HER2) testing. Patients in period D had significantly longer OS than those in period A [median: 15.7 versus 12.4 months; adjusted hazard ratio (aHR): 0.79; p = 0.02]. The mean OS in patients with liver metastasis (LM) in period D was remarkably longer than that in patients in period A (median: 19.3 versus 12.4 months; aHR: 0.61; p < 0.01), while that in patients with peritoneal metastasis showed limited improvement.

Conclusion:

Clinical strategy changes, including gastrectomy, HER2 testing, and approval of new drugs, may be associated with improved OS in patients with AGC. In the last 4 years, a remarkable improvement has been observed in patients with LM.

Introduction

Gastric cancer is the sixth most prevalent cancer and the third cause of death globally. ¹ Systemic chemotherapy is used to treat unresectable gastric cancer, with ongoing advancements in chemotherapeutic strategies being reported every year. Until 2007, no established standard of care for first-line treatment existed. However, according to the SPIRITS study, fluoropyrimidine and platinum agents were established as the standard first-line chemotherapy in 2008.^2,3 Trastuzumab was approved for human epidermal growth factor receptor 2 (HER2)-positive gastric cancer according to the ToGA study. ⁴ The RAINBOW study demonstrated significantly higher overall survival (OS) in patients with advanced gastric cancer (AGC) treated with ramucirumab and paclitaxel than in those treated with paclitaxel alone for platinum-refractory AGC [9.6 versus 7.4 months, respectively, hazard ratio (HR): 0.807, p = 0.017]. ¹³ The ATTRACTION-2 and TAGS studies showed significantly higher OS in AGC patients treated with nivolumab and trifluridine/tipiracil (FTD/TPI) than in those treated with placebo (nivolumab: 5.26 versus 4.14 months, respectively, HR: 0.63, p < 0.0001; FTD/TPI: 5.7 versus 3.6 months, respectively, HR: 0.69, p = 0.00058).^9,10 The DESTINY-Gastric01 study showed a significantly higher response rate to trastuzumab and deruxtecan than to the physician’s treatment choice for HER2-positive gastric cancer (51% versus 14%, p < 0.001). ¹¹ As demonstrated, several drugs have shown efficacy in pivotal studies on metastatic gastric cancer.

Despite the efficacy of approved drugs in clinical trials, it is unclear how these drugs contribute to the overall prognosis in gastric cancer patients. A previous retrospective study reported that AGC patients involved in randomized clinical trials had better Eastern Cooperative Oncology Group Performance Status (ECOG PS) and higher chance of receiving subsequent chemotherapy than those treated in clinical settings. ⁵ Thus, this study aimed to evaluate clinicopathological features and survival in patients with AGC in the past 15 years in clinical practice.

Materials and methods

Patients

This retrospective study included patients with AGC who received first-line chemotherapy between January 2005 and March 2019 at the Aichi Cancer Center Hospital in Japan. The data collection cut-off date was December 2020. The inclusion criteria were as follows: presence of histologically or cytologically proven gastric or esophagogastric junction adenocarcinoma, unresectable or recurrent gastric cancer, and receiving first-line treatment at our institution (adjuvant chemotherapy was defined as first-line treatment in patients with recurrence while under treatment or at <24 weeks from the last administered treatment).

The following patient data were collected from medical records: sex, age, ECOG PS, HER2 status, histological findings, history of gastrectomy, metastatic sites, all regimens received, and laboratory data. Data on exposure to the following key drugs were also collected: fluoropyrimidines, platinum agents (cisplatin or oxaliplatin), taxanes (paclitaxel, docetaxel, and nab-paclitaxel), irinotecan, FTD/TPI, trastuzumab, ramucirumab, and nivolumab. The proportion of patients receiving sequential treatment was determined from the data of patients who experienced OS events.

Opt-out recruitment was employed according to the Japanese government policy since this clinical research was conducted using only retrospective clinical data without any intervention. This study was approved by the Institutional Review Board of the Aichi Cancer Center Hospital (No. 2019-1-179).

Definition of periods

The study was divided into four periods: period A, January 2005–December 2007; period B, January 2008–February 2011; period C, March 2011–May 2015; and period D, June 2015–March 2019 (Supplemental Figure 1).

Period A refers to the time before the establishment of the fluoropyrimidine plus a platinum agent as the standard first-line chemotherapy. Period B denotes the period after fluoropyrimidine and platinum agents were established as the first-line chemotherapy. Period C represents the period after the approval of trastuzumab in Japan. Period D represents the period after the approval of ramucirumab in Japan.

Statistical analysis

OS was defined as the time from the date of the first chemotherapy administration to the date of death from any cause or date of the last follow-up. Progression-free survival (PFS) was defined as the time from the date of the first chemotherapy administration to the date of disease progression based on imaging findings, clinical progression, or death due to any cause. Post-progression survival (PPS) was defined as the time from the date of disease progression, as determined by imaging findings or clinical progression, to the date of death from any cause or date of the last follow-up. The main analysis involved the comparison of survival rates among the four periods. OS, PFS, and PPS were estimated using the Kaplan–Meier method and compared among periods using the log-rank test. HRs and confidence intervals (CIs) were estimated using Cox proportional hazard models in both the univariate and multivariate analyses. OS evaluation was based on a trend test for prolonged prognosis over different time periods. Covariates with p values <0.05 in the univariate analysis were included without an intermediate variable in the multivariate analysis. ECOG PS, history of gastrectomy, histology findings, serum alkaline phosphatase (ALP), number of metastatic sites, and periods were incorporated in the multivariate analysis as adjustment factors. The computed tomography (CT) scans or the other tests including HER2 and microsatellite instability (MSI) were managed according to the local practice by the physicians.

All statistical analyses were performed using the R software (R Development Core Team, Vienna, Austria). Statistical significance was set at p < 0.05 in all analyses.

Results

Overall, 1355 patients were treated with first-line chemotherapy. These patients were distributed across periods A, B, C, and D as follows: 312, 333, 393, and 317 patients, respectively. Patient characteristics are summarized in Table 1. The number of patients aged 65 years and above and of those with an ECOG PS of 0 was gradually increased between the periods. However, the number of patients who underwent gastrectomy was decreased. The HER2 status in approximately half of the patients in periods A and B was unknown, whereas it was determined in almost all patients in periods C and D. Sex, serum ALP level, histological type, and number of metastatic sites did not differ among the periods.

Table 1.

Patient characteristics.

Variables	Period A, N = 312	Period B, N = 333	Period C, N = 393	Period D, N = 317	p Value
Age – no. (%)					0.008
⩽65 years	184 (59)	179 (54)	195 (50)	146 (46)
>65 years	128 (41)	154 (46)	198 (50)	171 (54)
Sex – no. (%)					0.549
Male	204 (65)	212 (64)	267 (68)	201 (63)
Female	108 (35)	121 (36)	126 (32)	116 (37)
ECOG PS – no. (%)					<0.001
0	92 (29)	112 (34)	183 (47)	168 (53)
⩾1	220 (71)	221 (66)	208 (53)	149 (47)
Prior gastrectomy – no. (%)					<0.001
Yes	153 (49)	131 (39)	147 (38)	206 (34)
No	159 (51)	202 (61)	241 (62)	209 (66)
Histology – no. (%)					0.800
tub/pap	92 (29)	105 (32)	118 (31)	89 (28)
por/sig/unknown	220 (71)	228 (68)	266 (69)	227 (72)
HER2 status – no. (%)					0.481 a
Positive	19 (6)	28 (8)	76 (19)	62 (20)
Negative	83 (27)	155 (47)	289 (74)	253 (80)
Unknown	210 (67)	150 (45)	28 (7)	2 (1)
Serum ALP – no. (%)
⩽ULN	232 (75)	250 (77)	288 (74)	237 (75)
⩾ULN	76 (25)	77 (23)	101 (26)	79 (25)
Liver metastasis – no. (%)
Yes	92 (29)	90 (27)	99 (25)	76 (24)
No	220 (71)	243 (73)	292 (75)	241 (76)
Peritoneal metastasis – no. (%)
Yes	181 (58)	188 (55)	226 (58)	196 (62)
No	131 (42)	145 (45)	165 (42)	121 (38)
Number of metastatic site – no. (%)
⩽1	182 (58)	200 (60)	189 (48)	168 (53)
⩾2	130 (42)	133 (40)	202 (52)	149 (47)

^aOnly cases with HER2 testing were analyzed.

ALP, alkaline phosphatase; ECOG PS, Eastern Cooperative Oncology Group Performance Status; HER2, human epidermal growth factor receptor 2; ULN, upper limit of normal.

Changes in key drug exposure and proportion of patients receiving sequential treatment

Almost all patients received fluoropyrimidines throughout the study period. Platinum-based agents have been widely used since period B when platinum combination therapy was established as the standard of care. However, the proportion of patients who received irinotecan was gradually decreased. For HER2-positive gastric cancer, trastuzumab, ramucirumab, and nivolumab have been widely used since their approval [Table 2(a)].

Table 2.

Exposure to key drugs and patients receiving sequential treatment.

(a) Frequency of exposure to key drugs.

Key drugs	Period A		Period B		Period C		Period D		p Value
	N = 312	%	N = 333	%	N = 393	%	N = 317	%
FPs	288	92.3	300	90.1	369	93.9	317	100	<0.001
Platinum	133	42.6	247	74.2	337	85.8	298	94.0	<0.001
Taxane	207	66.3	240	72.1	286	72.8	238	75.1	0.0968
IRI	123	39.4	162	48.6	154	39.2	82	25.9	<0.001
Ram	0	0	12	3.6	40	10.2	203	64.0	<0.001
Nivo	0	0	2	0.6	30	7.6	136	42.9	<0.001
FTD/TPI	0	0	0	0	1	0.3	17	5.4	<0.001
Tmab	15	4.8	26	7.8	73	23.4	57	18.0
(only HER2+)	15/19	(78.9)	26/28	(92.9)	73/76	(96.1)	57/62	(91.9)	0.0414 a

The proportion of patients receiving second-line treatment was approximately 80% [Table 2(b)]. The proportion of patients receiving third-line treatment was the highest in period D.

Survival changes among periods

The median follow-up duration was 13.1 months. Of a total 1355 patients, 1288 patients (95.1%) died at the analysis. The median OS durations in periods A, B, C, and D were 12.4, 12.7, 13.8, and 15.7 months, respectively [Figure 1(a)]. The OS in period D was significantly longer than that in period A (HR: 0.63, 95% CI: 0.54–0.75, p < 0.001). Multivariate analysis also demonstrated significantly longer OS in period D than in period A [adjusted HR (aHR): 0.64, 95% CI: 0.53–0.76, p < 0.001] [Table 3(a) and Supplemental Table 1]. The trend test showed a gradual improvement in prognosis over time (p < 0.001).

Figure 1.

Kaplan–Meier curves of OS, PFS, and PPS after first-line chemotherapy. (a) Kaplan–Meier curve of OS. (b) Kaplan–Meier curve of PFS. (c) Kaplan–Meier curve of PPS.

CI, confidence interval; HR, hazard ratio; PFS, progression-free survival; PPS, post-progression survival; 1yOS, 1-year overall survival; 2yOS, 2-year overall survival; 3yOS, 3-year overall survival.

Table 3.

Univariate and multivariate analyses of OS, PFS, and PPS.

	Univariate analysis			Multivariate analysis
	HR	95% CI	p Value	HR	95% CI	p Value
(a) OS
Period A	1.00	–	–	1.00	–	–
Period B	0.87	0.74–1.02	0.089	0.89	0.76–1.04	0.142
Period C	0.75	0.64–0.87	<0.001	0.77	0.66–0.91	0.002
Period D	0.63	0.54–0.75	<0.001	0.64	0.53–0.76	<0.001
(b) PFS for first-line chemotherapy
Period A	1.00	–	–	1.00	–	–
Period B	0.95	0.81–1.12	0.089	0.96	0.81–1.13	0.621
Period C	0.75	0.64–0.88	<0.001	0.76	0.64–0.89	<0.001
Period D	0.68	0.58–0.80	<0.001	0.68	0.58–0.81	<0.001
(c) PPS for first-line chemotherapy
Period A	1.00	–	–	1.00	–	–
Period B	0.83	0.70–0.97	0.023	0.84	0.71–1.00	0.044
Period C	0.81	0.69–0.95	0.008	0.87	0.74–1.03	0.100
Period D	0.74	0.63–0.88	<0.001	0.76	0.64–0.91	0.003

CI, confidence interval; HR, hazard ratio; OS, overall survival; PFS, progression-free survival; PPS, post-progression survival.

The median PFS durations in periods A, B, C, and D were 4.3, 5.1, 6.0, and 6.6 months, respectively [Figure 1(b)]. The PFS in period D was significantly longer than that in period A (HR: 0.68, 95% CI: 0.58–0.80, p < 0.001). Multivariate analysis also demonstrated that the PFS in period D was significantly longer than that in period A (aHR: 0.68, 95% CI: 0.58–0.81, p < 0.001) [Table 3(b)].

(b) Proportion of patients receiving sequential treatment.

Treatment line	Period A		Period B		Period C		Period D		p Value
	N = 302	%	N = 314	%	N = 340	%	N = 244	%
First-line	302	100	314	100	340	100	244	100	1.000
Second-line	234	77.5	240	76.4	262	77.1	206	84.4	0.080
Third-line	114	37.7	164	52.2	148	43.5	139	57.0	<0.001
Fourth-line	48	15.9	83	26.4	67	19.7	56	23.0	0.012
Fifth-line	5	1.7	29	9.2	20	5.9	24	9.8	<0.001
Sixth-line	1	0.3	9	2.9	7	2.1	5	2.0	0.076
Seventh-line	0	0	2	0.6	2	0.6	0	0	0.383

^aOnly cases with HER2 testing were analyzed.

FPs, fluoropyrimidines; FTD/TPI, trifluridine/tipiracil; IRI, irinotecan; Nivo, nivolumab; Ram, ramucirumab; Tmab, trastuzumab.

The median PPS durations in periods A, B, C, and D were 6.2, 6.6, 6.6, and 7.5 months, respectively [Figure 1(c)]. The PPS in period D was significantly longer than that in period A (HR: 0.74, 95% CI: 0.63–0.88, p < 0.001). Multivariate analysis also revealed that the PPS in period D was significantly longer than that in period A (aHR: 0.76, 95% CI: 0.64–0.91, p = 0.003) [Table 3(c)].

Survival by metastatic sites

The characteristics of patients, frequency of exposure to key drugs, and proportion of patients receiving sequential treatment based on metastatic sites are summarized in Supplemental Tables 2–4.

In patients with liver metastases, the median OS durations were 12.4, 11.0, 14.3, and 19.3 months in periods A, B, C, and D, respectively [Figure 2(a)]. The OS in period D was significantly longer than that in periods A (HR: 0.45, 95% CI: 0.33–0.66, p < 0.001) and C (HR: 0.63, 95% CI: 0.45–0.90, p = 0.009). Multivariate analysis also revealed that the OS in period D was significantly longer than that in period A (aHR: 0.45, 95% CI: 0.31–0.65, p < 0.001) (Supplemental Table 5).

Figure 2.

Kaplan–Meier curve of OS in patients with liver or peritoneal metastases. (a) Patients with liver metastasis. (b) Patients with peritoneal metastasis.

CI, confidence interval; HR, hazard ratio; OS, overall survival.

In patients with peritoneal metastasis, the median OS durations in periods A, B, C, and D were 11.0, 11.4, 12.1, and 13.0 months, respectively [Figure 2(b)]. The OS in period D was longer than that in period A (HR: 0.71, 95% CI: 0.58–0.88, p = 0.002) (Supplemental Table 5). However, the OS in period D was not longer than that in period C (HR: 0.95, 95% CI: 0.77–1.16, p = 0.601).

Discussion

We evaluated the data of 1355 patients with AGC over the past 15 years to determine patient characteristics and survival by specific time periods. Survival was improved during the four periods of clinical practice. Although most prognosis-related patient characteristics did not differ over time, the number of patients with ECOG PS 0 or older patients was gradually increased, and the number of those with a history of gastrectomy decreased. Exposure to key drugs varied over time, with new drugs being widely used after approval. Notably, the survival in patients with liver metastasis (LM) was improved substantially, whereas that in patients with peritoneal metastasis saw only minor improvements over the last 4 years.

In our study, survival improved over the past 15 years. ECOG PS, history of gastrectomy, serum ALP levels, and metastatic sites are the known prognostic factors for AGC.^6,7 Multivariate analysis revealed strong associations between known prognostic factors and OS, and the period was found to be a new factor associated with OS. We consider this to be due to the approval of the new regimen and drugs. Notably, this study presented that the frequency of exposure to key drugs and the proportion of patients receiving sequential treatment varied across the periods. According to the SPIRITS study, platinum agents were more widely used following period B of this study.^2,3 Oxaliplatin, approved for AGC in 2014 in Japan, based on the G-SOX study, showed favorable efficacy and safety when combined with S-1, compared with a combination of S-1 plus cisplatin in older patients. ⁸ The number of older patients receiving platinum agents, particularly oxaliplatin, may increase in clinical practice (data not shown). Trastuzumab, ramucirumab, nivolumab, and FTD/TPI have also been extensively used after approval in this study.^4,9–11 Although paclitaxel and irinotecan were used as second- or third-line treatments during periods A and B, the frequency of irinotecan use was decreased in period C due to paclitaxel’s prevalence in second-line chemotherapy based on the results of the WJOG4007 and RAINBOW studies. In contrast, nivolumab and FTD/TPI replaced irinotecan in third-line chemotherapy.^12,13 In our study, nivolumab was frequently administered after approval. We believe that the appropriate use of new drugs and regimens contributes to prolonged survival.

The number of patients with an ECOG PS score of 0 gradually was increased over the study periods. Although it was difficult to determine the cause, increased health awareness could be one of the reasons. According to the Cancer Registry and Statistics (Cancer Information Service, National Cancer Center, Japan), an increasing number of individuals have recently undergone screening for gastric cancer. ¹⁴ We believe that in recent years, with increased screenings, more individuals are being diagnosed while in good general condition. Another reason may be that improvements in imaging have made it easier to detect liver or peritoneal metastases.

We observed changes in clinical strategies, including HER2 testing and gastrectomy, during this study. Most patients underwent HER2 testing during periods B, C, and D after trastuzumab approval. Moreover, the number of patients with a history of gastrectomy declined and the proportion of patients who underwent primary resection decreased. Several reasons can be attributed to this observation. First, adjuvant chemotherapy was developed, and the recurrence rates reduced following curative resection according to data from the ACTS-GC, CLASSIC, and START-2 trials.^15–17 The results of these clinical trials were reflected in clinical practice during periods B, C, and D. Second, the REGATTA study showed no benefit of primary resection before systemic chemotherapy. ¹⁸ Because the history of gastrectomy had been known as a good prognostic factor in previous studies, the decrease in patients with gastrectomy and the improved OS seems to be conflicting. Referring to the result of REGATTA study, the good prognosis in patients with the history of gastrectomy reported in the previous studies may have been due to selection bias.

The OS in patients with LM was notably improved over the last 4 years. We believe that ramucirumab largely contributed to this improvement. The RAINBOW study for LM showed that paclitaxel plus ramucirumab improved OS and PFS, with greater improvements observed in patients with LM than in those without (OS: HR: 0.71 versus 0.88; PFS: HR: 0.47 versus 0.76). ¹⁹ Preclinical data showed an association between anti-vascular endothelial growth factor (VEGF) discontinuation and enhanced LM, indicating a strong correlation between VEGF and LM. ²⁰ However, the OS improvement in patients with peritoneal metastasis was minimal. One reason for these differing improvements was the variation in the proportion of patients who received sequential treatment. Liver metastases are easily detected in imaging examinations, whereas peritoneal metastases are challenging to detect, complicating the decision to switch treatments. New methods need to be developed to replace RECIST in determining tumor progression in peritoneal metastasis.

This study had some limitations. First, this study was conducted at a single institution. Since our institution specializes in cancer treatment in Japan, we believe that our institution provided appropriate treatment, and we could accurately evaluate the efficacy of drugs using data from our institution as opposed to using data from other institutions. Second, we did not evaluate adverse events. With the approval of new drugs, including immune checkpoint inhibitors, various adverse events have likely occurred in recent years. However, the TAGS study reported better performance status maintenance in patients who received FTD/TPI than in those who received placebo, ⁹ suggesting that with effective treatment, adverse events are manageable. Third, the follow-up period was inadequate, particularly in period D. The OS was longer in period D, and a longer follow-up time was required. It is necessary to report long-term follow-up data in the future.

Conclusion

Changes in clinical strategies, including gastrectomy, HER2 testing, and approval of new drugs, may be associated with improved OS in patients with AGC. Over the last 4 years, remarkable improvement has been observed in patients with LM. Therefore, new drugs need to be approved in the future. Comprehensive genomic profiling has become widespread and is expected to further improve the prognosis of gastric cancer.