Restricted prognostic value of tumor size in gastric cancer: A retrospective cohort study based on SEER database

doi:10.21203/rs.3.rs-2084764/v1

Download PDF

Research Article

Restricted prognostic value of tumor size in gastric cancer: A retrospective cohort study based on SEER database

https://doi.org/10.21203/rs.3.rs-2084764/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Although tumor size (Ts) is regarded as the “T” stage of the Tumor-Node-Metastasis (TNM) staging system for many solid tumors, the prognostic impact in gastric cancer (GC) remains uncertain and conflicting.

Methods: We enrolled 6960 eligible cases from the Surveillance, Epidemiology, and End Results (SEER) database. The X-tile program was used to select the best cutoff value of Ts. Then the Kaplan-Meier method and the Cox proportional hazards model were applied to examine the efficacy of Ts on prognostic prediction for overall survival (OS) and gastric cancer-specific survival (GCSS). The presence of nonlinear association was determined by restricted cubic splines (RCS).

Results: Ts was divided into three groups: small size (≤ 2.5 cm), medium size (2.6-5.2 cm), and large size (≥ 5.3 cm). After adjusting by covariates, Ts larger than 2.5 cm predicted a worse prognosis; however, no survival difference on OS was suggested between the medium and large groups. Similarly, according to the adjusted RCS models, for tumors larger than 3.9 cm, large tumor size showed no significant differences on OS and GCSS with those smaller. However, the stratified analyses proposed the usage of the 3-way cut of Ts in prognostic prediction for patients with both inadequate lymphadenectomy and negative lymph node metastasis.

Conclusions: Ts as a prognostic predictor may not have good clinical applicability in GC. Otherwise, it is recommended for patients with both insufficient examination of lymph nodes and stage N0 disease.

Stomach neoplasms

Tumor size

Prognosis

Lymphatic metastasis

The Tumor-Node-Metastasis (TNM) classification for gastric cancer (GC) is considered a practical and credible category for postoperative follow-up and treatment [1, 2]. However, the T stage, determined by the depth of tumor invasion in gastric cancer, is defined as the maximum diameter of the tumor in other cancers such as breast cancer, lung cancer, and thyroid cancer [3-5]. In contrast, the prognostic impact of tumor size (Ts) in GC remains uncertain. In recent years, a variety of researches suggested that Ts is a simple but significant factor in the prognostic prediction of GC [6-9]. Despite this, it is still difficult to accurately estimate the clinical applicability and superiority of Ts. Conflicting results under different subgroups of GC, inconsistent cutoff value, and small sample size all restrain the effectiveness of Ts in evaluating the prognosis of patients with GC. Furthermore, the above studies usually divided Ts into different experimental groups, showing only linear relationship of Ts with prognosis. Hence, given the continuous nature of Ts, exploring whether there is a nonlinear relationship between Ts and prognosis is essential and valuable [10].

Ts is a concrete reflection of tumor progression and burden, partially determining whether the tumor manifests any detectable clinical symptoms (dysphagia, abdominal pains, gastric outlet obstruction, and others) [11]. Pathologically, Ts is considered a possible indicator of the depth of invasion and lymph node metastasis [12-14]. However, the clear relationship between Ts and lymph node metastasis (LNM) is still unclear. In contrast, the numbers of lymph nodes examined have been proved critical for nodal metastatic status determination, which insufficient numbers of examined lymph nodes (< 16) may underestimate the nodal stage and increase the chance of residual malignancy [15, 16]. Thus, in the circumstance of inadequate examination of lymph nodes, whether the evaluation of Ts can improve the accuracy of prognostic assessment, especially for patients with node-negative gastric cancer?

To address these unresolved issues, we enrolled a large population of 6960 eligible cases from the SEER database to confirm the relationship of Ts with long-term survival and LNM.

2.1 Selection of Patients

Patients were selected from the SEER database (2004-2018) using the SEER*Stat software (version 8.3.9.2). In general, patients were initially enrolled with the following inclusion criteria: 1) primary site codes C16.0-C16.9 for histologically proved gastric cancer, 2) histologic type codes were 8140/3, 8144/3, 8211/3, 8255/3, 8260/3, 8263/3, 8480/3, 8490/3. Meanwhile, information on grade, histological type, T stage, N stage, TNM stage, tumor size, regional nodes examined, regional nodes positive, tumor location, history of malignancy, and demographic characteristics was collected. Of note, according to the 8th edition of the American Joint Committee on Cancer (AJCC) Staging Manual, the TNM status of each case was re-evaluated and updated. In the present study, we aimed to evaluate the association of Ts with overall survival (OS) and gastric cancer-specific survival (GCSS), so the information regarding the cause of death, survival time, and survival status was also gathered. In this study, the follow-up cutoff date was December 31, 2018. OS was defined as the time from surgery until death from any cause, and GCSS was defined as the interval from surgery until death from GC. Finally, we identified a total of 73,738 cases during the initial screening process.

According to the exclusion criteria, patients with Ts larger than 20 cm, distant metastasis, missing information, or radiotherapy were then excluded. Ultimately, we included 6960 patients with tumor size (0.1 to 20 cm) and active follow-up in this study. This study was in line with the STROCSS criteria [17]. The SEER database is a public database with de-identified information of the patients, hence, institutional ethical approval and informed consent were waived (www.seer.cancer.gov).

2.2 Statistical analysis

Using the X-tile program, version 3.6.1, Ts was divided into three groups according to the χ² value of the log-rank test [18]. However, variables such as Ts and examined lymph nodes (ELNs) were also tested as continuous variables at prognostic analyses. Generally, we adopted the chi-squared test to explore the relationship between clinicopathological variables and tumor size categories. The Kaplan-Meier method was used to generate survival curves for OS and GCSS, while the log-rank test was applied to analyze the differences between curves. Hazard ratios (HR) and 95% confidence intervals (95% CI) were calculated using the Cox proportional hazards model. In addition, we carried out stratum analysis to evaluate the influence of the clinicopathological parameters on the efficiency of Ts in prognostic prediction.

To flexibly model the relationship with prognosis and LNM, Ts and ELNs were also subjected to restricted cubic spline (RCS) analysis by taking five knots at the 5th, 35th, 50th, 65th, and 95th centiles. In the RCS models, we defined the reference of Ts as the median size (3.9 cm). In terms of ELNs, we took 15 as the reference [19]. Statistical significance was accepted as a p-value < .05, and all statistical analyses were performed on R, version 4.0.4 (https://www.rproject.org/).

3.1 Patient characteristics were stratified by tumor size

In this study, we included 6960 eligible cases (Fig. 1A). The median Ts and ELNs were 3.9 cm and 16, respectively. The median follow-up period of all patients was 27 months (range 1-179 months). Then, by selecting the highest χ² value from the analysis of survival data with X-tile, the studied population was divided into three groups with small (≤ 2.5 cm), medium (2.6-5.2 cm), and large (≥ 5.3 cm) tumor size (Fig. 1B, Fig. S1A). Small tumor group was correlated with better OS than medium or large tumor group (the median OS was 126 months, 38 months, and 21 months for patients with small, medium, and large tumor group, respectively, P <.001) (Fig. S1B). In the graphical presentation, diffuse distribution of risk across the plot was discerned, suggesting a continuous indirect association of Ts with patient’s survival (Fig. 1B). Subsequently, the patient characteristics stratified by these three Ts categories are summarized in Table 1. Differences were detected for all variables except for sex (P = 0.203 for sex). Large tumors generally contained an undifferentiated type, showed deeper invasion and extended lymph node metastasis, and were tented to occupy the entire stomach.

3.2 Effects of tumor size on OS and GCSS

The Kaplan-Meier curves for OS and GCSS based on different Ts groups were presented (Fig. 1C, D). For tumors with size of ≤ 2.5 cm, 2.6-5.2 cm, or ≥ 5.3 cm, the five-year OS rates were 66.8%, 41.0%, 28.1% (P <.0001), respectively, and the corresponding five-year GCSS rates were 83.2%, 59.8%, 39.6% (P <.0001). Next, in the adjusted Cox models, continuous variables such as ELNs and Ts (OS for Ts: HR = 1.02, 95% CI = 1.01-1.04, GCSS for Ts: HR = 1.03, 95% CI = 1.02- 1.05) were found to be an independent prognostic factor for OS and GCSS, as were the following categorical variables: age, race, T stage, N stage, and location (Supplementary Table 1). When Ts was included as a categorical variable in the multivariate Cox regression analysis, the HRs for GCSS gradually raised with increasing Ts (≤ 2.5 cm vs. 2.6-5.2 cm, HR = 0.82, 95% CI = 0.71-0.94, P = 0.005, ≥ 5.3 cm vs. 2.6-5.2 cm, HR = 1.13, 95% CI = 1.03-1.24, P = 0.012) (Table 2). However, the HRs for OS did not show significant difference between the medium and large group (≥ 5.3 cm vs. 2.6-5.2 cm, HR = 1.02, 95% CI = 0.95-1.10, P = 0.564) (Table 2, Supplementary Table 2).

We also used RCS to visualize the association between Ts and prognosis. First, unadjusted Cox models with RCS showed a nonlinear positive correlation of Ts with the HRs for OS and GCSS (P for nonlinearity was < .0001 in OS, P for nonlinearity was < .0001 in GCSS) (Fig. S2A, B). After adjusting by covariates, the HRs for OS increased rapidly within the size of 3.9 cm, but then a U-shaped distribution was manifested and started to increase slowly (P for total was < .0001, P for nonlinearity was < .0001) (Fig. 1E). Meanwhile, the HRs for GCSS also raised quickly, reaching a plateau when Ts was approximately 4 cm (P for total was < .0001, P for nonlinearity was < .002) (Fig. 1F). However, when Ts exceeds the turning point at 3.9 cm, no statistical significance was achieved regarding the 95% CI of HR for both OS and GCSS (Fig. 1E, F). Therefore, despite the continuous nature of Ts, it should be treated as a categorical marker when assessing survival in GC, with statistically valid cut-points.

According to the results of multivariate analysis, we found the large group (≥ 5.3 cm) did not show a worse OS than the medium group as expected. Hence, we conducted the stratified analysis to determine the underlying effectors. Significant survival differences were demonstrated for all the subgroups (age, sex, race, grade, location, and N stage, P < 0.05 for OS) except for patients with stage T2 disease (P = 0.086 for OS) (Fig. 2A, B, Fig. S3A-E). Meanwhile, overlapping survival curves were also found for subgroups based on the T and N stage, reflecting the instability of this 3-way split of Ts. Regarding the performance of Ts on GCSS, survival differences were detected for all the subgroups (all P <.01) (Fig. 2C, Fig. S4A-F). However, in a manner similar to OS, the curves of GCSS also intersected under the subgroups of T2 stage, T4 stage, and N3 stage (Fig. 2C, Fig. S4A-F). Together, these results may suggest Ts is a more effective predictor of prognosis for patients with stage T1 and N0 disease.

3.3 The effect of tumor size on lymph node metastasis

Ts is recognized as an indicator of tumor growth, tumor invasion, and lymph node metastasis. However, the nonlinear association between Ts and LNM has not been assessed in GC. After adjusting by covariates, a nonlinear positive relationship was indicated, that the ORs for LNM steadily increased when Ts got larger (P for nonlinearity was < .0001) (Fig. 3A). As expected, extended lymph nodes dissection also showed an increased number of positive lymph nodes (P for nonlinearity was < .0001) (Fig. 3B). However, the risk of LNM began to flatten with ELNs higher than 15. It is reasonable because the number of positive lymph nodes is not truly determined by the number of harvested lymph nodes but by the biological behavior of the tumor.

3.4 Association of tumor size with prognosis based on examined lymph nodes

After adjusting by covariates, the RCS models showed a nonlinear negative correlation of ELNs with the HRs for OS and GCSS (P for total was < .0001, P for nonlinearity was < .0001 in OS, P for total was < .0001, P for nonlinearity was < .0001 in CSS) (Fig. S5A, B). The HRs rapidly declined but became sluggish when ELNs reached 15. However, a direct correlation of ELNs with a better prognosis could be concluded.

An insufficient number of harvested lymph nodes for diagnosis may result in the underestimation of nodal stages [20]. Considering the effect of Ts on LNM and prognosis, we then evaluated the value of Ts in prognostic prediction for patients with insufficient ELNs. Notably, as presented in Figures 3C and D, statistical significances of survival difference were obtained for all subgroups (group 1: ELNs < 16 with no LNM, group 2: ELNs < 16 with LNM, group 3: ELNs ≥ 16 with no LNM, group 4: ELNs ≥ 16 with LNM, all p-value < .0001). Specifically, this 3-way cut of Ts was more suitable for the evaluation of GCSS (Fig. 3D).

Gastric cancer is one of the most frequent and deadly cancers [21]. The completeness of surgical debulking is decisive for the outcome of patients with GC. Therefore, surgeons usually pay more attention to Ts than to depth of invasion to secure credible surgical management and outcomes [22]. However, Ts is seldom considered in the prognostic evaluation for patients who received gastrectomy. Thus, it is necessary to conduct a study to evaluate the true value of Ts in prognostic prediction for GC.

In our study, we grouped Ts into three categories using the X-tile software. As expected, a large tumor size was associated with a worse prognosis. However, this 3-way split did not distinguish well the difference in prognosis between large and medium groups during the multivariable analysis for OS. To further confirm these results, Ts was analyzed as a continuous variable in the RCS model. Notably, with Ts larger than 3.9 cm, the risk of OS or GCSS started to flatten, and the 95% CI of HRs became insignificant. Hence, these results may suggest that Ts should be treated as a category variable with a 2-way cut. However, a 2-way split inevitably restricts its application because of the reduction of efficacy in evaluating the prognosis of patients. For example, in breast cancer, according to the 8th edition of the AJCC staging system, tumors are categorized as T1, T2, T3 stages with a tumor size of ≤ 2 cm, > 2cm but ≤ 5cm, > 5cm, respectively [3]. On the other hand, the above results may suggest the better performance of this 3-way cut in specific subgroups of GC.

In the stratum analyses, an insignificant result was detected for stage T2 disease, suggesting the influences of the T stage on the prognostic significance of Ts. Meanwhile, overlapping curves between medium and large groups were discerned at the advanced T/N stages. Thus, we thought the 3-way cut of Ts as prognostic indicator should be confined in early-stage diseases. Ts is given as the maximum diameter of the tumor, while the T stage, in some way, is the vertical diameter along with the stomach. Hence, to some extent, the depth of tumor invasion may determine the tumor size since the cancer cells disseminate horizontally and vertically [10]. In addition, there may be some reasons to explain the weakened negative effect on prognosis for larger tumor size. First, a large tumor size is prone to present with symptoms and therefore with earlier treatment and diagnosis. Second, a large tumor size suggests a more sufficient surgical margin and more comprehensive treatment. Third, a larger tumor means that the patient has a greater tolerance to the tumor burden. However, more well-designed studies are needed to prove these hypotheses. Of note, Zu et al. suggested that Ts was not a significant predictor of survival in patients with early-stage tumors by classifying the cases into small size groups (≤ 5 cm) and large size groups (≥ 5 cm) [23]. However, this study comprised only 0.8% (n = 6) of patients with stage T1 tumors larger than 5 cm.

Some studies have revealed the association between Ts and LNM as well as the prognostic impact of Ts for node-negative GC. A study that enrolled 1577 patients with early-stage gastric adenocarcinomas from the national SEER database suggested that Ts was an independent predictor of nodal metastasis [24]. Liang et al. pointed out that Ts also served as an independent prognostic factor of node-negative GC [25]. However, the sample size of this study was relatively small (n = 436). In line with the aforementioned studies, our study further showed a nonlinear positive association between Ts and LNM. In this regard, we then evaluated whether Ts improves the accuracy of the prognostic prediction for patients with insufficient ELNs. Accordingly, our study indicated that Ts is a valuable prognostic indicator for patients with ELNs less than 16. Thus, we specifically recommend the assessment of Ts for patients with both insufficient ELNs and stage N0 diseases.

Several limitations should not be neglected in this study. First, this study was conducted based on the retrospective information from the SEER database, which might introduce selection biases and influence the results. Second, despite the enrollment of a large, multicenter population, the sample size of individual subgroups got small when stratified by Ts. Third, the information on chemotherapy was not included due to incomplete chemotherapy information in the SEER database, resulting in a possible deviation of our survival analysis to the real world. Despite these limitations, based on the large scale of patients and significant follow-up from the SEER database as well as rigorous statistics, our results are more convincing.

In conclusion, our study clearly illustrated the relationship between Ts and prognosis. Ts larger than 2.5 cm predicts poorer prognosis, however, when Ts reaches 3.9 cm or more, larger tumor size does not show an additively negative effect on prognosis. However, a nonlinear positive relationship between Ts and LNM was suggested. Thus, we think Ts as a prognostic predictor may not have good clinical applicability in GC.

Acknowledgments

We are very grateful to the SEER Program of United States National Cancer Institute for providing the data.

Funding support

This work was supported by the National Natural Science Foundation of China (Grant numbers: 81874219).

Authorship contributions

J.X. Conceptualization, Investigation, Formal analysis, Visualization, Writing-Original Draft. K.S. Methodology, Validation, Data Curation, Visualization, Writing-Review & Editing, Formal analysis. H.F. Investigation, Validation, Data Curation, Visualization, Writing-Review & Editing, Formal analysis. G.W. Software, Data Curation, Writing-Review & Editing. K.L. Methodology, Validation. Y.W. Software, Validation. X.M. Validation, Writing-Review & Editing. P.N. Software, Validation. Z.X. Supervision, Resources. L.Y. Conceptualization, Supervision, Resources, Funding acquisition.

Conflict of interest

The authors have no conflicts of interest to declare.

Ethical approval

The SEER database is a public database with de-identified information of the patients, hence, institutional ethical approval and informed consent were waived (www.seer.cancer.gov).

Data availability statement

The data could be downloaded at SEER program (www.seer.cancer.gov), and the code used during the current study are available from the corresponding author on reasonable request.

H. In, I. Solsky, B. Palis, et al. Validation of the 8th Edition of the AJCC TNM Staging System for Gastric Cancer using the National Cancer Database. Ann Surg Oncol. 2017,24 (12): 3683-3691. https://doi.org/10.1245/s10434-017-6078-x.
X. Ji, Z. D. Bu, Y. Yan, et al. The 8th edition of the American Joint Committee on Cancer tumor-node-metastasis staging system for gastric cancer is superior to the 7th edition: results from a Chinese mono-institutional study of 1663 patients. Gastric Cancer. 2018,21 (4): 643-652. https://doi.org/10.1007/s10120-017-0779-5.
A. E. Giuliano, J. L. Connolly, S. B. Edge, et al. Breast Cancer-Major changes in the American Joint Committee on Cancer eighth edition cancer staging manual. CA Cancer J Clin. 2017,67 (4): 290-303. https://doi.org/10.3322/caac.21393.
R. Rami-Porta, H. Asamura, W. D. Travis, et al. Lung cancer - major changes in the American Joint Committee on Cancer eighth edition cancer staging manual. CA Cancer J Clin. 2017,67 (2): 138-155. https://doi.org/10.3322/caac.21390.
O. A. Manzardo, M. Cellini, R. Indirli, et al. TNM 8th edition in thyroid cancer staging: is there an improvement in predicting recurrence? Endocr Relat Cancer. 2020,27 (6): 325-336. https://doi.org/10.1530/erc-19-0412.
J. Deng, R. Zhang, Y. Pan, et al. Tumor size as a recommendable variable for accuracy of the prognostic prediction of gastric cancer: a retrospective analysis of 1,521 patients. Ann Surg Oncol. 2015,22 (2): 565-72. https://doi.org/10.1245/s10434-014-4014-x.
X. Liu, Y. Xu, Z. Long, et al. Prognostic significance of tumor size in T3 gastric cancer. Ann Surg Oncol. 2009,16 (7): 1875-82. https://doi.org/10.1245/s10434-009-0449-x.
X. Wang, F. Wan, J. Pan, et al. Tumor size: a non-neglectable independent prognostic factor for gastric cancer. J Surg Oncol. 2008,97 (3): 236-40. https://doi.org/10.1002/jso.20951.
H. M. Wang, C. M. Huang, C. H. Zheng, et al. Tumor size as a prognostic factor in patients with advanced gastric cancer in the lower third of the stomach. World J Gastroenterol. 2012,18 (38): 5470-5. https://doi.org/10.3748/wjg.v18.i38.5470.
H. Feng, Z. Lyu, J. Zheng, et al. Association of tumor size with prognosis in colon cancer: A Surveillance, Epidemiology, and End Results (SEER) database analysis. Surgery. 2021,169 (5): 1116-1123. https://doi.org/10.1016/j.surg.2020.11.011.
G. Maconi, G. Manes ,G. B. Porro. Role of symptoms in diagnosis and outcome of gastric cancer. World J Gastroenterol. 2008,14 (8): 1149-55. https://doi.org/10.3748/wjg.14.1149.
T. Ohashi, S. Komatsu, D. Ichikawa, et al. Tumor Index as a Combined Indicator of Tumor Depth and Size in Gastric Cancer. Anticancer Res. 2016,36 (4): 1895-900.
R. M. Kwee ,T. C. Kwee. Predicting lymph node status in early gastric cancer. Gastric Cancer. 2008,11 (3): 134-48. https://doi.org/10.1007/s10120-008-0476-5.
H. Feng, J. Zheng, C. Zheng, et al. The probability of Lymph node metastasis with a tumor size larger than and smaller than 4 cm is different in stages T1-T3 of Siewert type II adenocarcinoma of esophagogastric junction: A Population-Based Study. J Cancer. 2021,12 (22): 6873-6882. https://doi.org/10.7150/jca.63392.
J. Deng, H. Yamashita, Y. Seto, et al. Increasing the Number of Examined Lymph Nodes is a Prerequisite for Improvement in the Accurate Evaluation of Overall Survival of Node-Negative Gastric Cancer Patients. Ann Surg Oncol. 2017,24 (3): 745-753. https://doi.org/10.1245/s10434-016-5513-8.
D. D. Smith, R. R. Schwarz ,R. E. Schwarz. Impact of total lymph node count on staging and survival after gastrectomy for gastric cancer: data from a large US-population database. J Clin Oncol. 2005,23 (28): 7114-24. https://doi.org/10.1200/JCO.2005.14.621.
R. Agha, A. Abdall-Razak, E. Crossley, et al. STROCSS 2019 Guideline: Strengthening the reporting of cohort studies in surgery. Int J Surg. 2019,72: 156-165. https://doi.org/10.1016/j.ijsu.2019.11.002.
R. L. Camp, M. Dolled-Filhart ,D. L. Rimm. X-tile: a new bio-informatics tool for biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res. 2004,10 (21): 7252-9. https://doi.org/10.1158/1078-0432.ccr-04-0713.
A. Biondi, D. D'Ugo, F. C. Cananzi, et al. Does a minimum number of 16 retrieved nodes affect survival in curatively resected gastric cancer? Eur J Surg Oncol. 2015,41 (6): 779-86. https://doi.org/10.1016/j.ejso.2015.03.227.
T. Son, W. J. Hyung, J. H. Lee, et al. Clinical implication of an insufficient number of examined lymph nodes after curative resection for gastric cancer. Cancer. 2012,118 (19): 4687-93. https://doi.org/10.1002/cncr.27426.
H. Sung, J. Ferlay, R. L. Siegel, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021, https://doi.org/10.3322/caac.21660.
S. Sato, C. Kunisaki, Y. Tanaka, et al. Curative-Intent Surgery for Stage IV Advanced Gastric Cancer: Who Can Undergo Surgery and What Are the Prognostic Factors for Long-Term Survival? Ann Surg Oncol. 2019,26 (13): 4452-4463. https://doi.org/10.1245/s10434-019-07790-1.
H. Zu, F. Wang, Y. Ma, et al. Stage-stratified analysis of prognostic significance of tumor size in patients with gastric cancer. PLoS One. 2013,8 (1): e54502. https://doi.org/10.1371/journal.pone.0054502.
S. K. Pokala, C. Zhang, Z. Chen, et al. Lymph node metastasis in early gastric adenocarcinoma in the United States of America. Endoscopy. 2018,50 (5): 479-486. https://doi.org/10.1055/s-0043-122379.
Y. Liang, L. Liu, X. Xie, et al. Tumor Size Improves the Accuracy of the Prognostic Prediction of Lymph Node-Negative Gastric Cancer. J Surg Res. 2019,240: 89-96. https://doi.org/10.1016/j.jss.2019.02.037.

Table 1 Characteristics of patients were arranged by tumor size.

Variables	Overall	Tumor size groups (cm)			P^†
Variables	Overall	≤ 2.5	2.6-5.2	≥ 5.3	P^†
Age (year)					< 0.001
10-49	615	220	219	176
50-64	1778	643	616	519
65-85	3934	1220	1473	1241
≥ 85	633	135	253	245
Sex					0.203
Male	4183	1347	1559	1277
Female	2777	871	1002	904
Race					< 0.001
White	4279	1377	1571	1331
Black	1112	278	443	391
Other^‡	1569	563	547	459
Grade					< 0.001
Well	447	267	109	71
Moderate	2219	791	853	575
Poorly, undifferentiated	4294	1160	1599	1535
Histology					< 0.001
Adenocarcinoma	5447	1713	2081	1653
Other ^§	1513	505	480	528
T stage					< 0.001
T1	2104	1422	571	111
T2	954	320	439	195
T3	2344	321	1013	1010
T4	1558	155	538	865
N stage					< 0.001
N0	3635	1785	1191	659
N1	1110	230	516	364
N2	972	126	435	411
N3a	840	57	332	451
N3b	403	20	87	296
Location					< 0.001
Cardia/Fundus	1607	589	609	409
Body	761	257	272	232
Antrum/Pylorus	2405	769	904	732
Lesser/Greater curvature	1182	351	464	367
Other ^¶	1005	252	312	441
Examined lymph nodes (Median [IQR])	16 [9-24]	14 [8-22]	15 [9-23]	19 [11-27]

IQR, interquartile range. ^†p-value for the chi-square test. ^‡Including others (Asian or Pacific Islander, American Indian/Alaska Native) and unknown. ^§Including tumors categorized as cystic, mucinous, and serous neoplasms. ^¶Including subgroups of overlapping lesions of stomach and Stomach, NOS.

Table 2 Associations of tumor size with OS and GCSS in the SEER database.

Tumor size	Five-year survival rate	Univariate analysis		Multivariate analysis ^†
Groups	Five-year OS (%)	HR (95% CI)	P	HR (95% CI)	P
≤ 2.5 cm	66.8%	0.48 (0.44-0.52)	< 0.001	0.81 (0.73- 0.89)	< 0.001
2.6-5.2 cm	41.0%	1 (Reference)		1 (Reference)
≥ 5.3 cm	28.1%	1.41 (1.32-1.52)	< 0.001	1.02 (0.95- 1.10)	0.564
Groups	Five-year GCSS (%)	HR (95% CI)^‡	P^‡	HR (95% CI) ^‡	P^‡
≤ 2.5 cm	83.2%	0.37 (0.32-0.42)	< 0.001	0.82 (0.71- 0.94)	0.005
2.6-5.2 cm	59.8%	1 (Reference)		1 (Reference)
≥ 5.3 cm	39.6%	1.79 (1.63-1.96)	< 0.001	1.13 (1.03- 1.24)	0.012

^†Adjusted by age, race, grade, histology, T stage, N stage, ELNs, and tumor location. ^‡Outcome variable is GCSS.

The significant results are in bold.

No competing interests reported.

SupplementaryFigures.doc
Supplementary figure 1 Tumor size was categorized into three groups by the X-tile program. A Histogram presentation of the grouped cohort with two cut-points (2.5 and 5.2 cm), producing small, medium, and large subsets. B The Kaplan-Meier plot shows the survival difference between individual subsets (small subset = blue, medium subset = gray, large subset = magenta). Supplementary figure 2 Association between tumor size and survival in unadjusted Cox models with RCS. Red lines represent estimated HRs of Ts (3.9 cm as reference), shadow area shows 95% CI. A Unadjusted RCS model for OS (P for nonlinearity was < .0001). B Unadjusted RCS model for GCSS (P for nonlinearity was < .0001). Supplementary figure 3. Survive curves for OS in the stratified analysis of tumor size. Stratified analysis based on age (A), sex (B), race (C), grade (D), and tumor location (E). P for log-rank test. Supplementary figure 4. Survive curves for GCSS in the stratified analysis of tumor size. Stratified analysis based on age (A), sex (B), race (C), grade (D), tumor location (E), and N stage (F). P for log-rank test. Supplementary figure 5 Prognostic impact of ELNs using Cox models with RCS. Adjusted Cox models with RCS for OS (A) and GCSS (B). P for total was < .0001, P for nonlinearity was < .0001 in OS, P for total was < .0001, P for nonlinearity was < .0001 in GCSS. Red lines represent estimated HRs (15 as reference), shadow area shows 95% CI. Relative risk was adjusted by age, race, histology, grade, T stage, N stage, tumor location, and tumor size.
SupplementaryTables.doc
Supplementary Table 1 Cox regression models for univariate and multivariate survival analysis. Supplementary Table 2 Cox regression models for tumor size in the SEER database.

Download PDF

Version 1

posted

You are reading this latest preprint version

Restricted prognostic value of tumor size in gastric cancer: A retrospective cohort study based on SEER database

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials And Methods

2.1 Selection of Patients

2.2 Statistical analysis

3. Results

3.1 Patient characteristics were stratified by tumor size

3.2 Effects of tumor size on OS and GCSS

3.3 The effect of tumor size on lymph node metastasis

3.4 Association of tumor size with prognosis based on examined lymph nodes

4. Discussion

5. Conclusions

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1