Prognostic factors and efficacy of postoperative chemotherapy in patients with gastric cancer with positive peritoneal cytology after gastrectomy

Abstract

BACKGROUND

Peritoneal lavage cytology-positive (CY1) gastric cancer (stage IV) has a poor prognosis, though some cases fare better. Therefore, identifying prognostic factors and an optimal treatment strategy is crucial.

AIM

To investigate prognostic factors in patients with gastric cancer who underwent gastrectomy with CY1, and to evaluate the optimal postoperative chemotherapy regimen.

METHODS

This multicenter retrospective cohort study analyzed prognostic factors and postoperative chemotherapy in patients with CY1 gastric cancer who underwent gastrectomy, excluding those with macroscopic peritoneal dissemination. Data from 13 institutions (2015-2019) were reviewed.

RESULTS

Overall, 82 patients met the inclusion criteria. The median overall survival was 22.8 months, and diffuse-type histology and the absence of postoperative chemotherapy were identified as independent poor prognostic factors. The 5-year survival rate was 82.4% for those receiving fluoropyrimidine plus docetaxel/oxaliplatin vs 21.8% for those with S-1 monotherapy or a cisplatin-based regimen. Median overall survival was not reached in the fluoropyrimidine + docetaxel/oxaliplatin group but was 22.9 months in the S-1/cisplatin group. Chemotherapy regimen was an independent prognostic factor (hazard ratio = 5.47, P = 0.004). The fluoropyrimidine plus docetaxel/oxaliplatin group had an average relative dose intensity of 82.1%, with significantly more patients achieving a relative dose intensity ≥ 80% than in the S-1 monotherapy or cisplatin-based group (P = 0.001).

CONCLUSION

Diffuse-type histology and the absence of postoperative chemotherapy influence the prognosis of patients with CY1 gastric cancer. Combination therapy with oxaliplatin or docetaxel may enhance the treatment intensity and improve survival outcomes after gastrectomy.

Key Words: Gastric cancer; Peritoneal lavage cytology-positive; Chemotherapy; Prognostic factor; Treatment strategy; Cancer statistics; Survival

Core Tip: This multicenter retrospective cohort study identifies key prognostic factors in cytology-positive gastric cancer patients undergoing gastrectomy. Diffuse-type histology and absence of postoperative chemotherapy were independent poor prognostic indicators. Critically, combination chemotherapy with fluoropyrimidine plus docetaxel or oxaliplatin significantly improved survival, with a 5-year survival rate of 82.4% vs 21.8% for S-1 or cisplatin-based regimens. The chemotherapy regimen was an independent predictor of overall survival. These findings highlight the importance of intensive postoperative chemotherapy in improving outcomes for this high-risk population.

INTRODUCTION

Gastric cancer is the fifth most common malignancy worldwide, accounting for 23% of all gastric cancers in Eastern Asia[1]. In Japan, although the incidence rate has been declining recently, it remains the third and fourth most common cancers in men and women, respectively. Additionally, its mortality rate ranked third. Therefore, developing effective gastric cancer treatment strategies is crucial[2]. Although the mortality rate has decreased recently[3], this can be attributed to advancements in endoscopic techniques, which have led to an increase in early diagnosis. The development of novel anticancer agents and immune checkpoint inhibitors has improved treatment outcomes for unresectable or recurrent advanced gastric cancer. However, the reported median survival periods remained relatively poor, ranging from 12.9 to 13.8 months in the Checkmate649, Keynote849, and Toga trials[4-6].

Peritoneal lavage cytology-positive (CY1) gastric cancer frequently results in peritoneal dissemination recurrence and is defined as M1 (metastasis to distant organs) in the major tumor-node-metastasis classifications (Japanese Classification of Gastric Carcinoma, Union for International Cancer Control, American Joint Committee on Cancer), categorizing it as a stage IV disease with a poor prognosis. The reported frequency of positive peritoneal cytology results during surgery ranges from 4% to 41%[3], with higher positivity rates observed in advanced tumors and approximately 10% in T3 or deeper tumors[4]. CY1 is a stronger prognostic factor than tumor depth or lymph node metastasis and has been reported to be the most significant prognostic determinant[5].

Palliative chemotherapy remains the primary treatment for stage IV gastric cancer in patients with peritoneal dissemination. However, the treatment strategies for CY1 gastric cancer differ significantly. Recently, when peritoneal dissemination is suspected based on imaging studies, diagnostic laparoscopy has been increasingly performed to confirm peritoneal dissemination or CY1 status. If either is confirmed, chemotherapy is initiated, and surgery is considered if therapeutic effects are achieved. However, when surgery is performed first, CY1 can be identified intraoperatively or postoperatively. In such cases, gastrectomy is frequently performed in clinical practice. This approach is weakly recommended in the Japanese “Gastric Cancer Treatment Guidelines, 5^th Edition”, when no other non-curative factors are present and is combined with postoperative chemotherapy using S-1. In contrast to other stage IV gastric cancers, CY1 gastric cancer can achieve treatment outcomes comparable to those of R0-resected gastric cancer through gastrectomy; however, sufficient evidence supporting this approach has not yet been established. Despite the use of combination regimens incorporating S-1 with oxaliplatin or docetaxel as postoperative adjuvant therapy for stage III gastric cancer following curative resection, a discrepancy remains in current clinical guidelines. These guidelines recommend S-1 monotherapy for postoperative chemotherapy in CY1 gastric cancer, which is classified as stage IV. Although combination therapy with these agents is also expected to improve treatment outcomes in CY1 gastric cancer, there is currently no supporting evidence. This study aimed to investigate the prognostic factors in patients with gastric cancer who underwent gastrectomy with CY1 as the sole non-curative factor and evaluate the optimal postoperative chemotherapy regimen in these individuals. This multicenter retrospective cohort study was conducted by collecting data from institutions affiliated with the Hiroshima Surgical Study of Clinical Oncology (HiSCO).

MATERIALS AND METHODS

Study design

We analyzed data from patients with gastric cancer with CY1 as the only non-curative factor who underwent gastrectomy at 13 HiSCO-affiliated institutions from 2015 to 2019. Among them, we targeted patients who met the following selection criteria: (1) Age ≥ 20 years; (2) Eastern Cooperative Oncology Group performance status of 0-2; (3) Histologically confirmed gastric adenocarcinoma and underwent standard surgery, including lymph node dissection, with D1+ dissection allowed; (4) CY1 status confirmed by peritoneal cytology; and (5) No evidence of distant metastasis. Although a central review of peritoneal lavage cytology was not conducted, but, all participating institutions evaluated the cytological specimens using Papanicolaou staining by the 15^th edition of the Japanese Classification of Gastric Carcinoma. In addition, cases with suspected false-positive findings were uniformly categorized as CY0. Patients with localized peritoneal metastasis (P1a) who underwent complete macroscopic resection were excluded from this study.

Patients who did not undergo lymph node dissection or receive preoperative chemotherapy were excluded. Moreover, at the time of the data cut-off, no patients in this study had received immune checkpoint inhibitors as part of postoperative chemotherapy, nor had any received hyperthermia therapy. In the subgroup analysis of patients who underwent postoperative chemotherapy, individuals for whom an initial efficacy assessment could not be performed were excluded from this study. This study protocol was approved by the Ethics Committee of JA Hiroshima General Hospital, No. 23-27, and the Ethical Guidelines for Life Science and Medical Research Involving Human Subjects. Data were collected from the HiSCO database.

Treatment and procedure

The regimens and durations of postoperative chemotherapy were determined according to the protocols established at each institution. For S-1 monotherapy, the dosage and schedule were determined based on the ACTS-GC trial. Similarly, for S1 + cisplatin therapy, the SPIRIT trial served as a reference. The new combination chemotherapy (NC) groups received S-1 plus oxaliplatin (SOX), capecitabine plus oxaliplatin, or S-1 plus docetaxel. In cases of human epidermal growth factor receptor 2 (HER2)-positive tumors, combination therapy with trastuzumab was permitted for any of the doublet therapies. Patients who could not continue the doublet therapy for at least two courses and switched to S-1 monotherapy were allocated to the S-1 monotherapy group. The administration period continued until disease progression, the occurrence of adverse events, or the patient requested to discontinue treatment.

Statistical analysis

We analyzed the overall survival (OS) and progression-free survival (PFS) based on the presence or absence of chemotherapy and conducted a subgroup analysis comparing the conventional chemotherapy (CC) and NC groups. OS was defined as the time from the date of surgery to death from any cause, and the last follow-up date was set as the cutoff. PFS was defined as the time from the date of surgery to the earlier occurrence of recurrence or death. Patients who experienced no recurrence and were still alive were censored at the last follow-up date. The data cutoff date was December 31, 2024. OS and PFS were calculated using the Kaplan-Meier method, and between-group differences were assessed using the log-rank test. The median observation periods and corresponding 95% confidence intervals (CI) were calculated using the inverse Kaplan-Meier method.

To minimize intergroup bias, survival rates were evaluated using multivariate analysis with the Cox proportional hazards regression model, and hazard ratios (HR) and 95%CIs were reported. The covariates for multivariate analysis were examined using factors that showed significant differences in the univariate analysis and those considered clinically relevant to prognosis. The variables included in the multivariate analysis were analyzed using Schoenfeld residual plots to assess the time-varying effects of covariates, confirming that the proportional hazards assumption was satisfied. Categorical data comparisons were performed using Fisher’s exact test or the χ² test. To address potential confounding between treatment groups, propensity scores were estimated using a logistic regression model. Inverse probability of treatment weighting (IPTW) was applied to balance covariates across treatment groups, and stabilized weights were used to avoid extreme values. A Cox proportional hazards model weighted by IPTW was used to evaluate differences in recurrence-free and OS rates. Kaplan-Meier survival curves were also generated using IPTW weights. Propensity score analysis was performed using R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria), and other statistical analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, United States).

RESULTS

Patient characteristics

Overall, 125 cases were collected from 13 institutions in the Hiroshima Prefecture. Forty-three patients were excluded as they did not meet the eligibility criteria (macroscopically resected P1a cases, n = 31; R2 [residual tumor (gross/pathological) with peritoneal dissemination or positive margins in locally advanced disease] resection cases; macroscopic and/or pathological residual tumor due to peritoneal dissemination (P1a–c) or locally advanced disease with positive margins, n = 7; cases with other distant metastases, n = 2; CY0 cases, n = 1; cases without lymphadenectomy, n = 1; and cases without follow-up, n = 1). Among eligible patients, 42 underwent postoperative chemotherapy, whereas 40 did not. A CONSORT diagram, including the chemotherapy regimens, is shown in Figure 1.

Figure 1 Flow diagram of gastric cancer patients with peritoneal lavage cytology-positive who underwent gastrectomy. CY1: Peritoneal lavage cytology-positive; P1a: Localized peritoneal metastasis; R2: Residual tumor (gross/pathological) with peritoneal dissemination or positive margins in locally advanced disease; M1: Metastasis to distant organs; CY0: Peritoneal lavage cytology-suspected false positive; D0 LN: Lymph node (D0); CDDP: Cisplatin; DOC: Docetaxel.

The patient backgrounds and clinical characteristics are summarized in Table 1. The mean age was 74.3 years, and six patients had a performance status of 2. Macroscopic type 4 was observed in 22 (26.8%) patients. All patients had tumor invasion of T3 or deeper, and all but one case had lymph node metastasis. The most common surgical procedure was distal gastrectomy (43 patients), followed by total gastrectomy (37 patients, 45.1%). Regarding lymph node dissection, 21 (25.6%) patients underwent reduced D2 dissection, whereas 13 (15.9%) underwent extended dissection. Postoperative complications occurred in 22 (24.4%) patients. Approximately half of the patients did not receive postoperative chemotherapy. While these patients underwent surgical resection, adjuvant chemotherapy was not initiated owing to various factors including poor performance status, presence of comorbidities, impaired organ function, or individual preference of the patients.

Table 1 Characteristics of patients with peritoneal lavage cytology-positive who underwent gastrectomy.

Variables	All (n = 82)
Age	74.3 (42-96)
Sex
Male	56
Female	26
Performance status (ECOG)
0	56
1	20
2	6
Macroscopic type
Non-type 4	60
Type 4	22
Preoperative CEA
> ULN	21
≤ ULN	60
Unknown	1
Preoperative CA19-9	1
> ULN	25
≤ ULN	55
Unknown	2
Pathological T factor
T3	16
T4a	59
T4b	7
Pathological N factor
N0	1
N1-2	35
N3	46
Histology
Intestinal	31
Diffuse	51
HER2 status
Positive	14
Negative	46
Unknown	22
Surgery
Distal gastrectomy	43
Total gastrectomy	37
Proximal gastrectomy	2
Lymph node dissection
< D2	21
D2	48
D2+	13
Postoperative complication1
Yes (grade III)	22 (5)
No	60
Postoperative CTx
Yes	42
No	40

¹Clavien-Dindo classification.

ECOG: Eastern Cooperative Oncology Group; CEA: Carcinoembryonic antigen; ULN: Upper limit of normal; CA19-9: Carbohydrate antigen 19-9; HER2: Human epidermal growth factor receptor 2; CTx: Chemotherapy.

Prognosis

During the follow-up period until the analysis cutoff point, recurrence was observed in 50 (40%) patients. Peritoneal dissemination was the most common recurrence type, occurring in 33 (68.8%) patients, followed by hematogenous in nine patients and lymphatic metastases in seven patients, and other types in one patient. The median PFS was 13.6 months, with a 5-year PFS rate of 26.5%. The median observation period was 71.6 months. Based on the analysis cutoff, 51 deaths were observed, including 42 cancer-related and 9 from other causes. There were 31 survivors, with a median OS (MST) of 22.8 months and a 5-year survival rate of 31.2%.

Univariate analysis of OS, age (< 75 years vs ≥ 75 years), gender (male vs female), histology (intestinal vs diffuse), and the presence or absence of postoperative chemotherapy were significantly associated with prognosis. We conducted a multivariate analysis using these factors. The results showed that histology and the presence or absence of postoperative chemotherapy were independent prognostic factors (Table 2). Similar findings were observed in the PFS analysis (data not shown).

Table 2 Univariate and multivariate analysis of overall survival in patients with peritoneal lavage cytology-positive who underwent gastrectomy.

Variables	n	Univariate analysis		Multivariate analysis
		HR (95%CI)	P value	HR (95%CI)	P value
Age (years)
≤ 75	41	Reference		Reference
> 75	41	2.20 (1.25-3.87)	0.006	1.81 (0.98-3.34)	0.057
Gender
Male	56	Reference		Reference
Female	26	1.79 (1.01-3.21)	0.048	1.45 (0.79-2.69)	0.233
Performance status (ECOG)
0, 1	76	Reference
2	6	1.03 (0.32-3.33)	0.955
BMI (kg/m2)
< 25	73	Reference
≥ 25	8	1.06 (0.42-2.67)	0.900
Preoperative CEA
≤ ULN	60	Reference
> ULN	21	1.50 (0.81-2.80)	0.201
Preoperative CA19-9
≤ ULN	55	Reference
> ULN	25	0.81 (0.42-1.56)	0.527
Surgery
DG, PG	45	Reference
TG	37	1.21 (0.70-2.10)	0.504
Lymph node dissection
≥ D2	61	Reference
< D2	21	1.36 (0.71-2.62)	0.354
Postoperative complication
No	60	Reference
Yes	22	1.47 (0.82-2.63)	0.200
Macroscopic type
Non-type 4	60	Reference
Type 4	22	1.63 (0.90-2.95)	0.107
Histology
Intestinal	31	Reference		Reference
Diffuse	51	1.89 (1.05-3.39)	0.034	2.65 (1.43-4.92)	0.002
pT
T3	16	Reference
T4	66	1.14 (0.53-2.42)	0.743
pN
N0-2	36	Reference
N3	46	1.08 (0.62-1.88)	0.785
Postoperative CTx
Yes	42	Reference		Reference
No	40	3.24 (1.82-5.78)	0.001	3.26 (1.75-6.08)	0.001

HR: Hazard ratio; CI: Confidence interval; ECOG: Eastern Cooperative Oncology Group; BMI: Body mass index; CEA: Carcinoembryonic antigen; ULN: Upper limit of normal; CA19-9: Carbohydrate antigen 19-9; DG: Distal gastrectomy; PG: Proximal gastrectomy; TG: Total gastrectomy; CTx: Chemotherapy.

Subgroup analysis of cases undergoing postoperative chemotherapy

A subgroup analysis was conducted on 42 patients with CY1 gastric cancer who underwent gastrectomy followed by postoperative chemotherapy. The chemotherapy regimens administered included S-1 monotherapy, fluoropyrimidine plus cisplatin, fluoropyrimidine plus oxaliplatin, and S-1 plus docetaxel in 19, 6, 13, and 4 patients, respectively (Figure 1). The S-1 monotherapy and fluoropyrimidine plus cisplatin regimens were classified as the CC group, while the S-1 plus docetaxel and fluoropyrimidine plus oxaliplatin regimens were classified as the NC group. A comparative analysis was performed between the two groups, and their characteristics are summarized in Table 3. Although the mean age was ≤ 75 years in both groups, the CC group tended to be slightly older. No significant differences were observed between the two groups regarding clinical factors. Changes in chemotherapy regimens over time are shown in Figure 2. Initially, CC was predominantly used; however, from 2016 onward, NC was introduced, and since 2018, most patients have been treated with NC.

Figure 2 Transition of chemotherapy. CTx: Chemotherapy.

Table 3 Characteristics of patients with peritoneal lavage cytology-positive who undergo gastrectomy followed by postoperative chemotherapy, n (%).

	Conventional chemotherapy1 (n = 25)	New combination chemotherapy2 (n = 17)	P value
Age (years)	72 (42-87)	66.2 (45-76)	0.073
Gender			0.589
Male	20 (80)	14 (82)
Female	5 (20)	3 (18)
Performance status (ECOG)			0.595
0, 1	24 (96)	17 (100)
2	1 (4)	0 (0)
Preoperative BMI (kg/m2)	21.1 (17.5-27.7)	21.7 (17.3-27.5)	0.531
CEA			0.437
≤ ULN	20 (83)	13 (76)
> ULN	4 (7)	4 (24)
CA19-9			0.565
≤ ULN	19 (79)	13 (76)
> ULN	5 (21)	4 (24)
Surgery			0.175
DG, PG	17 (68)	8 (47)
TG	8 (32)	9 (53)
Lymph node dissection			0.397
≥ D2	20 (80)	15 (88)
< D2	5 (20)	2 (12)
Postoperative complication			0.589
No	20 (80)	14 (82)
Yes	5 (20)	3 (18)
Macroscopic type			0.173
Non-type 4	9 (36)	3 (18)
Type 4	16 (64)	14 (82)
Histology			0.542
Intestinal	8 (32)	7 (41)
Diffuse	17 (68)	10 (59)
pT			0.463
T3	6 (24)	3 (18)
T4	19 (76)	14 (82)
pN			0.845
N0-2	11 (44)	8 (47)
N3	14 (56)	9 (53)
HER2 status			0.346
Positive	16	8
Negative	6	4
Unknown	3	5
BW decrease rate3			0.465
< 10%	18 (81)	15 (88)
≥ 10%	4 (19)	2 (12)
Pre-chemotherapy BMI (kg/m2)	19.9 (15.1-26.3)	20.7 (16.0-25.5)	0.377
Ccr (mL/minute)			0.174
< 60	14 (56)	13 (82)
≥ 60	11 (44)	4 (18)
Pre-chemotherapy CEA			0.534
≤ ULN	21 (84)	15 (88)
> ULN	4 (16)	2 (12)
Pre-chemotherapy CA19-9			0.630
≤ ULN	22 (92)	16 (94)
> ULN	2 (8)	1 (6)

¹Conventional chemotherapy includes S1 monotherapy or combination therapy with fluoropyrimidine + cisplatin.

²Novel chemotherapy include fluoropyrimidine + oxaliplatin or fluoropyrimidine + docetaxel.

³BW decrease rate represents the rate of weight loss from the time of surgery to the initiation of chemotherapy.

ECOG: Eastern Cooperative Oncology Group; BMI: Body mass index; CEA: Carcinoembryonic antigen; ULN: Upper limit of normal; CA19-9: Carbohydrate antigen 19-9; DG: Distal gastrectomy; PG: Proximal gastrectomy; TG: Total gastrectomy; HER2: Human epidermal growth factor receptor 2; BW: Body weight; Ccr: Creatineine clearance.

PFS and OS of cases undergoing postoperative chemotherapy

The median observation period for PFS and OS was 77.5 and 68.0 months, respectively. During the observation period, 20 (80%) cases of recurrence were noted in the CC group, with 20 (80%) cases of peritoneal recurrence. In the NC group, there were five (29.4%) cases of recurrence, with three (17.6%) cases of peritoneal recurrence.

The 5-year PFS rate was 16.0% in the CC group, whereas it was 76.5% in the NC group. The median PFS was 10.8 months in the CC group and 90.4 months in the NC group, demonstrating a significant improved outcomes in the NC group (HR = 4.30, 95%CI: 1.71-10.82, P = 0.002; Figure 3A. Moreover, early postoperative recurrence within one year occurred in 15 of 25 patients in the CC group, compared to only 2 patients in the NC group. The MST in the CC group was 22.9 months with a 5-year survival rate of 21.8%. In contrast, in the NC group, the MST was not reached, and the 5-year survival rate was 82.4%, demonstrating significantly improved outcomes in the NC group (HR = 5.34, 95%CI: 1.77-16.14, P = 0.003; Figure 3B). Incorporating baseline covariates, including: Age at treatment initiation, performance status, presence of type 4 disease, and the year of postoperative treatment initiation, we estimated propensity scores using a logistic regression model. After IPTW adjustment, the new treatment group demonstrated significantly better PFS and OS rates than the conventional treatment group (HR = 0.10, 95%CI: 0.03-0.31, P < 0.001; HR = 0.09, 95%CI: 0.03-0.31, P < 0.002, respectively). The concordance index (C-index) of the IPTW-weighted model was 0.75, indicating its good discriminative ability. The weighted Kaplan-Meier curve also revealed a marked survival advantage in the new treatment group (Figure 3C and D). The concordance index of the IPTW-weighted Cox model was 0.75, indicating good predictive discrimination. In the multivariate analysis of OS, macroscopic type 4, presence of postoperative NC, and duration of postoperative chemotherapy were independent prognostic factors (Table 4).

Figure 3 Kaplan-Meier curves. A: Kaplan-Meier curves of survival in the two postoperative chemotherapy groups for progression-free survival; B: Kaplan-Meier curves of survival in the two postoperative chemotherapy groups for overall survival; C: Inverse probability of treatment weighting-adjusted Kaplan-Meier curves for progression-free survival; D: Inverse probability of treatment weighting-adjusted Kaplan-Meier curves for overall survival. CTx: Chemotherapy; PFS: Progression-free survival.

Table 4 Univariate and multivariate analysis of overall survival in patients with peritoneal lavage cytology-positive who underwent postoperative chemotherapy.

Variables	n	Univariate analysis		Multivariate analysis
		HR (95%CI)	P value	HR (95%CI)	P value
Age (years)
≤ 75	29	Reference
> 75	13	1.39 (0.58-3.33)	0.460
Gender
Male	34	Reference
Female	8	1.55 (0.57-4.21)	0.395
Performance status (ECOG)
0, 1	41	Reference
2	1	1.82 (0.24-13.69)	0.562
Pre-chemotherapy BMI (kg/m2)
≥ 25	2	Reference
< 25	37	1.21 (0.16-9.04)	0.854
Pre-chemotherapy CEA
≤ ULN	36	Reference
> ULN	6	1.46 (0.49-4.30)	0.498
Pre-chemotherapy CA19-9
≤ ULN	38	Reference
> ULN	3	1.53 (0.36-6.60)	0.566
Surgery
DG, PG	25	Reference
TG	17	1.21 (0.52-2.82)	0.667
Lymph node dissection
≥ D2	35	Reference
< D2	7	1.09(0.37-3.22)	0.883
Postoperative complication
No	34	Reference
Yes	8	1.18 (0.43-3.21)	0.746
Macroscopic type
Non-type 4	14	Reference		Reference
Type 4	28	2.17 (0.92-5.09)	0.076	5.00 (1.76-14.25)	0.003
Histology
Intestinal	15	Reference
Diffuse	27	1.78 (0.72-4.38)	0.210
pT
T3	9	Reference
T4	33	1.86 (0.55-6.33)	0.318
pN
N0-2	19	Reference
N3	23	0.98 (0.42-2.27)	0.964
HER2 status
Positive	10	Reference
Negative	24	2.09 (0.69-6.39)	0.194
Postoperative CTx
New combination chemotherapy	17	Reference		Reference
Conventional chemotherapy	25	5.34 (1.77-16.14)	0.003	9.53 (2.54-35.79)	< 0.001
Duration of postoperative CTx
≥ 1 year	16	Reference		Reference
< 1 year	26	12.75 (2.98-55.55)	< 0.001	18.98 (3.68-97.93)	< 0.001
Ccr (mL/minute)
≥ 60	27	Reference
< 60	15	1.08 (0.45-2.61)	0.862
Grade III adverse event1
No	22	Reference
Yes	20	0.64 (0.27-1.55)	0.326

¹Common Terminology Criteria for Adverse Event v5.0.

HR: Hazard ratio; CI: Confidence interval; ECOG: Eastern Cooperative Oncology Group; BMI: Body mass index; CEA: Carcinoembryonic antigen; ULN: Upper limit of normal; CA19-9: Carbohydrate antigen 19-9; DG: Distal gastrectomy; PG: Proximal gastrectomy; TG: Total gastrectomy; HER2: Human epidermal growth factor receptor 2; CTx: Chemotherapy; Ccr: Creatineine clearance.

The details of chemotherapy are outlined in Table 5. The average time from surgery to chemotherapy initiation was within 6 weeks in both groups. The time to treatment failure was longer in the NC group (17.0 months) than in the CC group (12.6 months), although this difference was not statistically significant. Regarding adverse events, grade III or higher events were observed significantly more frequently in the NC group (71%) than in the CC group (32%). The reasons for discontinuing chemotherapy in the CC group included recurrence and adverse events in 15 and 4 patients, respectively. In the NC group, recurrence was the reason for discontinuation in four patients and adverse events in another four. Post-recurrence treatments were administered to 10 of the 15 patients in the CC group and two of the four patients in the NC group, with no significant difference between the groups. Regarding the relative dose intensity (RDI) of the infusions, the NC group maintained a higher intensity (82.1%) than the CC group (74.5%), with an approximately 8% difference. Furthermore, the NC group had a significantly higher proportion of patients maintaining over 80% RDI (58.8%) than the CC group (33.3%).

Table 5 Details of chemotherapy.

	Conventional chemotherapy (n = 25)	New combination chemotherapy (n = 17)	P value
TI until the postoperative CTx (week)1	5.1 (1.9-9.1)	5.1 (1.4-10.6)	0.908
TTF (months)
Oral drug	12.6(1.6-49.0)	17.0 (3.0-45.8)	0.129
Intravenous	5.3 (1.2-8.6)	5.4 (2.1-11.4)	0.946
Course: Median (range)
Intravenous	3.6 (1-5)	6.2 (2-15)	0.098
RDI: Median (%)
Intravenous	74.5 (54.5-100)	82.1 (60.9-100)	0.276
< 80%: > 80%	4:2	7:10	0.001
Adverse event (grade III)2	8 (32%)	17 (71%)	0.014
Reason for CTx discontinuation
Recurrence: Adverse events, etc.3	15:4:6	4:4:9	0.051
Subsequent CTx after recurrence	10/15 (67%)	2/4 (50%)	0.475

¹Time interval until the postoperative chemotherapy refers to the period between surgery and the initiation of chemotherapy.

²Common Terminology Criteria for Adverse Event v5.0.

³Includes planned cessation, long-term administration, and ongoing treatment.

TI: Time interval; CTx: Chemotherapy; TTF: Time to failure; RDI: Relative dose intensity.

DISCUSSION

In our analysis of prognostic factors in patients with gastric cancer who underwent gastrectomy, where CY1 was the only non-curative factor, histological type, and the presence or absence of postoperative chemotherapy were identified as independent prognostic determinants. Furthermore, in patients receiving postoperative chemotherapy, the addition of oxaliplatin or docetaxel to fluoropyrimidine, may enhance treatment intensity and improve survival outcomes after gastrectomy; however, prospective validation is needed.

CY1 is an independent non-curative factor and is classified as stage IV gastric cancer. However, in real-world clinical practice, gastrectomy is frequently performed, even when CY1 is detected intraoperatively. The MST in patients undergoing gastrectomy alone is only 12.1 months; however, the administration of postoperative S-1 monotherapy has been reported to improve the 5-year OS and PFS rates to 26% and 21%, respectively, demonstrating more favorable outcomes than other stage IV gastric cancers. Based on these findings, postoperative chemotherapy following gastrectomy for CY1 gastric cancer may yield treatment outcomes comparable to those for R0-resected gastric cancer. Therefore, gastrectomy, followed by postoperative chemotherapy, is currently considered the standard treatment. However, patients with concomitant peritoneal dissemination or a high number of cancer cell clusters in the peritoneal fluid have been reported to have a poor prognosis, similar to those with R2 resection[7]. Consequently, we excluded macroscopically resected P1a cases and focused on those where CY1 was the sole non-curative factor.

Independent prognostic factors for CY1 gastric cancer after gastrectomy have been reported to include signet ring cell carcinoma, lymph node metastasis, type 4 tumors, and the presence or absence of postoperative chemotherapy[7,8]. In our study, the histological subtype and receipt of postoperative chemotherapy were identified as prognostic factors, whereas type 4 tumors and lymph node metastasis were not.

No significant difference was observed in lymph node metastasis, as there was only one N0 case, and the treatment outcomes were comparable between N1 and N3 cases. Although type 4 tumors showed no significant difference, the MST was 12.9 and 24.4 months for type 4 and non-type 4 tumors, respectively, indicating a favorable trend for non-type 4 tumors. Moreover, in patients who underwent postoperative chemotherapy, non-type 4 cases demonstrated significantly better outcomes, indicating a potentially greater prognostic benefit of postoperative chemotherapy in this subgroup.

The following are two approaches to postoperative chemotherapy for CY1 gastric cancer after gastrectomy: One considering it as adjuvant chemotherapy for recurrence prevention and the first-line chemotherapy for unresectable or recurrent gastric cancer. One year of postoperative adjuvant S-1 monotherapy has long been the standard treatment for stage II and III gastric cancer in Japan[9]. However, comparative trials between combination therapies, including oxaliplatin or docetaxel, and S-1 monotherapy in patients with pStage III have demonstrated improved prognosis[10,11]. Therefore, oxaliplatin- or docetaxel-based combination therapies are currently recommended as adjuvant therapies for stage III gastric cancer. In contrast, the SPIRIT trial previously established S-1 plus cisplatin as the standard regimen for first-line treatment of unresectable or recurrent gastric cancer[12]. However, the non-inferiority of SOX to S-1 plus cisplatin has been demonstrated. Furthermore, owing to the lower incidence of adverse events, such as myelosuppression and renal dysfunction, as well as the advantages of administration methods, oxaliplatin has become the preferred platinum-based agent. Additionally, subgroup analysis of the G-SOX trial reported favorable therapeutic effects against peritoneal dissemination and a high disease control rate[13]. Consequently, SOX is currently considered one of the most promising first-line regimens for gastric cancer.

Based on the development of these treatment strategies in clinical practice, postoperative chemotherapy for CY1 gastric cancer after gastrectomy generally comprises mainly S1 monotherapy or combination therapy with fluoropyrimidine and cisplatin. However, combination regimens incorporating oxaliplatin or docetaxel have become more prevalent recently. Nevertheless, the Japanese guidelines weakly recommend postoperative chemotherapy with S-1 monotherapy for CY1 gastric cancer[14], with reported treatment outcomes including a 5-year survival rate and MST of 26%-27.1% and 23.2-29.5 months, respectively[15,16]. In contrast, combination therapy with cisplatin demonstrated no additional survival benefit, as the MST remained at 24.4 months, similar to that of S-1 monotherapy. One potential contributing factor is the high incidence of adverse events associated with cisplatin administration after gastrectomy, which prevents the maintenance of an adequate RDI. Although no definitive threshold for effective RDI in cisplatin combination therapy has been established, the SPIRIT trial reported that maintaining a cisplatin RDI of ≥ 90% resulted in an additional survival benefit over S1 monotherapy[12]. In chemotherapy for gastric cancer, an RDI of approximately 80% is generally considered the cutoff value for efficacy[17]. In that study, tumor shrinkage was observed in patients maintaining an RDI of ≥ 80% compared to those with a RDI of < 80%; however, no significant contribution to OS was observed. This lack of an OS benefit may be attributed to factors such as sample size limitations and the ability to adjust the dosage to reduce adverse events in patients with unresectable gastric cancer, which differs in intent from adjuvant therapy for CY1 gastric cancer after gastrectomy.

Regarding RDI in adjuvant chemotherapy, studies in breast cancer have reported better PFS and OS when an RDI of ≥ 85% was maintained[18]. In postoperative adjuvant chemotherapy for gastric cancer, an analysis using an RDI cutoff of 76% for S-1 showed that patients maintaining an RDI of ≥ 76% had significantly better PFS, identifying RDI as an independent prognostic factor[19]. Since the 5-year survival rate for CY1 gastric cancer is better than that for other stage IV gastric cancers, some cases may achieve curative outcomes. Therefore, maintaining a high RDI could contribute to achieving curative intent in this population. In this study, the RDI of cisplatin was approximately 72%. Similarly, Nakayama et al[20] reported that the RDI for cisplatin in combination with S1 for CY1 gastric cancer after gastrectomy was approximately 70%; however, no additional benefit from combination therapy was observed. Given that these two studies reported nearly identical RDIs, the real-world postoperative RDI of cisplatin for CY1 gastric cancer was estimated to be approximately 70%. At this RDI level, no significant advantage over S-1 monotherapy was observed.

In contrast, regarding combination therapy with oxaliplatin or docetaxel, the ARTIST2 trial reported a high completion rate of 85% for the planned eight cycles, with no significant differences in adverse events compared to S-1 monotherapy, except for an increase in grade 3 peripheral neuropathy[11]. In the JACCRO GC-07 trial, the completion rate for six planned cycles of docetaxel was 71%, with an increase in grade 3 Leukopenia and neutropenia; however, these toxicities were reported to be manageable[10]. Similarly, in our study, the mean intravenous RDI in the NC group exceeded 80%, indicating that chemotherapy intensity was well maintained. Moreover, while 15 cases of recurrence within 1 year were observed in the CC group, early recurrence within 1 year occurred in only approximately 10% of the cases in the NC group. Therefore, maintaining high-intensity chemotherapy during the early postoperative period likely contributes to the prevention of early recurrence, ultimately leading to improved PFS and OS. However, caution is warranted given the high incidence (70%) of grade III or higher adverse events in the NC group. Such severe events may necessitate further dose reductions or discontinuation of treatment. To sustain an RDI of ≥ 80%, it is important to limit dose reductions to a single step, which requires appropriate management of treatment-related toxicities.

Other treatment strategies for CY1 gastric cancer include neoadjuvant and intraperitoneal chemotherapy. Recently, staging laparoscopy has enabled preoperative identification of P1a or CY1 cases when peritoneal dissemination is suspected on imaging. A multicenter retrospective study conducted by the JCOG compared the outcomes of patients with P0CY1 or P1a disease who received neoadjuvant chemotherapy and those who underwent upfront surgery. The 5-year survival rates were 22.3% and 21.5%, respectively, with no significant difference between the two groups[21]. However, more than 90% of the chemotherapy regimens used in this study were cisplatin-based, with only a few patients receiving oxaliplatin- or docetaxel-based regimens. Notably, 25% of the patients received a triplet regimen that included a taxane-based agent, which demonstrated a high rate of CY0 conversion and improved prognosis. These findings suggest that intensifying chemotherapy with taxane-based regimens improves the treatment outcomes for CY1 gastric cancer.

Furthermore, in cases where chemotherapy induced CY1, conversion surgery resulted in remarkably favorable survival outcomes, with an MST of 42.4-108.5 months[22,23]. A particularly noteworthy finding was that the rate of peritoneal recurrence postoperatively was approximately 20%, which was significantly lower than the 55% rate reported in patients with CY1 gastric cancer who received S-1 monotherapy after gastrectomy. This suggests that maintaining sufficient chemotherapy intensity, even after gastrectomy, helps suppress peritoneal recurrence. The chemotherapy regimens most commonly used in conversion surgery cases were doublet or triplet regimens, indicating that perioperative administration of surgery combined with high-intensity chemotherapy involving at least two agents may contribute to an improved prognosis in CY1 gastric cancer.

Regarding intraperitoneal chemotherapy, Kuramoto et al[24] reported that peritoneal lavage combined with intraperitoneal cisplatin administration achieved a 5-year survival rate of 43.8%. However, peritoneal lavage alone yielded negative results in previous studies[25,26], highlighting the importance of intraperitoneal chemotherapy. Currently, a phase III randomized controlled trial is underway to compare standard systemic chemotherapy alone with systemic chemotherapy combined with intraperitoneal chemotherapy for CY1 type 4 gastric cancer diagnosed via staging laparoscopy[27]. These treatments are designed to directly reduce or eradicate free-floating cancer cells in the peritoneal cavity, aiming to suppress peritoneal recurrence.

This study had some limitations. First, as this was a retrospective cohort study with a small sample size, the results should be carefully interpreted owing to potential issues such as selection bias, random error, and reduced statistical power. Second, the postoperative chemotherapy regimens varied across institutions because they were selected at the discretion of individual clinicians at each facility. Some institutions adhered strictly to these guidelines and administered S-1 monotherapy; thus, potential selection bias due to patient background characteristics could not be fully eliminated. Third, the postoperative chemotherapy regimens were influenced by temporal trends. Since the efficacy of combination therapy for adjuvant treatment of stage III gastric cancer has been demonstrated, its use has increased. Consequently, the selection of the TC and NC groups was influenced by their different historical backgrounds. Fourth, data on HER2 status were missing for some patients. In the analysis of cases with available HER2 status, no significant impact on prognosis was observed in the present study. However, given that HER2 overexpression can influence prognosis and guide the use of targeted therapies such as trastuzumab; the absence of this information may have introduced unmeasured confounding. Fifth, the sample size was small, and multiple comparison corrections were not performed in subgroup analysis; thus, the interpretation of statistical significance should be approached with caution. As previously mentioned, recently, staging laparoscopy has been increasingly performed in cases suspected of peritoneal dissemination or malignant ascites, thereby substantially limiting the number of cases that met the criteria for this study. Therefore, drawing definitive conclusions from this study is difficult. To increase the sample size, aggregating data through the utilization of clinical registries or conducting a meta-analysis is essential.

CONCLUSION

In conclusion, prognostic factors for gastric cancer with CY1 after gastrectomy include host-related factors such as diffuse-type histology, which is associated with poor prognosis. Additionally, the introduction of postoperative chemotherapy may have contributed to improved prognosis. Combination regimens involving oxaliplatin or docetaxel are potentially particularly effective, as they are thought to help maintain treatment intensity even after gastrectomy.

ACKNOWLEDGEMENTS

We would like to express our gratitude to all patients who participated in this study and to the co-investigators affiliated with Hiroshima Surgical Study of Clinical Oncology for their valuable contributions.