AIM: To investigate the difference in long-term outcomes between gastric cancer patients with and without a primary symptom of overt bleeding (OB).

METHODS: Consecutive patients between January 1, 2007 and March 1, 2012 were identified retrospectively by reviewing a gastric cancer database at Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. A follow-up examination was performed on patients who underwent a radical gastrectomy. OB due to gastric cancer included hematemesis, melena or hematochezia, and gastric cancer was confirmed as the source of bleeding by endoscopy. Patients without OB were defined as cases with occult bleeding and those with other initial presentations, including epigastric pain, weakness, weight loss and obstruction. The 3-year overall survival (OS) rate, age, gender, AJCC T stage, AJCC N stage, overall AJCC stage, tumor size, histological type, macroscopic (Borrmann) type, lymphovascular invasion and R status were compared between patients with and without OB. Moreover, we carried out a subgroup analysis based on tumor location (upper, middle and lower).

RESULTS: We identified 939 patients. Of these, 695 (74.0%) were hospitalized for potential radical gastrectomy and another 244 received palliative resection, rerouting of the gastrointestinal tract, chemotherapy, radiotherapy or no treatment due to the presence of unresectable tumors. Notably, there was no significant difference in the percentage of OB patients between resectable cases and unresectable cases (20.3% vs 22.1%, P = 0.541). Follow-up examination was performed on 653 patients (94%) who underwent radical gastrectomy. We found no significant difference in 3-year OS rate (68.2% vs 61.2%, P = 0.143) or clinicopathological characteristics (P > 0.05) between these patients with and without OB. Subgroup analysis based on tumor location showed that the 3-year OS rate of upper gastric cancer was significantly higher in patients with OB (84.6%) than in those without OB (48.1%, P < 0.01) and that AJCC stages I-II (56.4% vs 35.1%, P = 0.017) and T1-T2 category tumors (30.8% vs 13%, P = 0.010) were more frequent in patients with OB than in those without OB. There was no significant difference in 3-year OS rate or clinicopathological characteristics between patients with and without OB (P > 0.05) for middle or lower gastric cancer.

CONCLUSION: Upper gastric cancer patients with OB exhibited tumors at less advanced pathological stages and had a better prognosis than upper gastric cancer patients without OB.

Gastrointestinal bleeding is classified as either overt bleeding (OB) or occult bleeding depending on the presence or absence of visible bleeding[1]. OB is one of the most frequent complications in patients with gastric cancer[2], as 1%-10% of hospitalized gastric cancer patients initially present with OB[3-7].

Numerous studies on gastric cancer bleeding have focused on hemostasis and rebleeding[2,4,5,7-9]. However, data regarding the clinicopathological characteristics and long-term outcomes of cases presenting with OB due to gastric cancer are extremely limited, and there are no reports comparing the clinical outcomes of patients with OB based on tumor location. The present study aimed to answer these questions, with a particular focus on OB in upper gastric cancer.

During the study period, 939 consecutive gastric cancer patients were identified, including 195 (20.8%) with OB and 744 without OB. Of these, 695 (74.0%) were hospitalized for potential radical gastrectomy and another 244 received palliative resection, rerouting of the gastrointestinal tract, chemotherapy, radiotherapy or no treatment due to the presence of unresectable tumors accompanied by distant metastasis, extensive abdominal metastasis or poor performance status. Notably, there was no significant difference in the percentage of OB patients between resectable cases and unresectable cases (20.3% vs 22.1%, P = 0.541) (Table 1).

A follow-up examination was performed on 653 (94.0%) of the 695 patients who underwent a radical gastrectomy. The patients who underwent follow-up were aged 63.0 ± 12.0 years (range, 16-93 years). The male-to-female ratio among the 653 enrolled patients was 2.05 (439 males and 214 females). Of these patients, 132 (20.2%) were hospitalized with a primary symptom of OB. Additionally, 128 patients who achieved hemostasis underwent elective surgery, whereas 4 patients who failed to achieve hemostasis underwent emergency surgery. The characteristics of the 132 patients with OB and the 521 patients without OB were comparable (Table 2). There was no significant difference in the 3-year overall survival rate between patients with OB (68.2%, 319/521) and those without OB (61.2%, 90/132, P = 0.143) (Figure 1A).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Three-year overall survival. A: The whole patient group; B: The upper gastric cancer; C: The middle gastric cancer; D: The lower gastric cancer. OB: Patients with a primary symptom of overt bleeding; Non-OB: Patients without a primary symptom of OB.

OB in patients with gastric cancer is a rare but serious condition with potentially dangerous effects. Although numerous studies have been conducted on the risk factors associated with gastric cancer bleeding and rebleeding, data regarding long-term outcomes following surgery and the pathological characteristics of gastric cancer patients with OB are extremely limited. The results of the current study provide information for clinical physicians to use when evaluating gastric cancer patients with OB. These findings also provide the following new perspectives relevant to our understanding of this group of patients: gastric cancer with OB is not synonymous with advanced gastric cancer, and its prognosis is no worse than the prognosis for gastric cancer without OB. In fact, patients with proximal gastric cancer with OB exhibit less advanced pathological stages and a better prognosis than patients with proximal gastric cancer without OB.

To investigate why the 3-year overall survival rate following surgery was significantly higher for upper gastric cancer patients presenting with OB than without OB, we compared clinicopathological characteristics between these two subgroups of upper gastric cancer patients. Our results revealed that compared to proximal gastric cancer patients without OB, those with OB exhibited not only less advanced tumors with respect to both overall AJCC stage and AJCC T stage but also smaller tumors. Further multivariate survival analysis demonstrated that the overall AJCC stage, tumor size and OB were independent factors affecting prognosis. An increment of one stage in the overall AJCC classification was associated with a 1.638-fold increased risk of death within 3 years of surgery. This result may explain why the former type of cancer exhibits a more positive prognosis. Fox et al[16] demonstrated that 62% of gastric cancer patients with bleeding were classified as overall AJCC stages I-II. Similarly, Kodama[17] and colleagues and Moreno-Otero et al[6] found that 72.2% of these patients exhibited early stage tumors and an intraluminal growth pattern with irregularly distributed erosions or shallow, but not deep, ulcerations. These previous results support the findings of the current study. Gertsch et al[18] compared the prognosis of gastric cancer patients with bleeding and serosal invasion with the prognosis of gastric cancer patients without complications, but found no statistically significant difference in clinicopathological characteristics or prognosis between the two groups. The following two considerations could explain why the results of Gertsch et al[18] were not consistent with the findings of the current study. On one hand, Gertsch et al[18] did not perform a stratified analysis that accounted for the differences in the locations of gastric cancer. On the other hand, the two studies assessed different subject populations: in particular, Gertsch et al[18] examined gastric carcinoma patients with serosal invasion, whereas we examined all gastric cancer patients at tumor stages T1-T4b.

Bleeding occurs as a result of gastric cancer for many reasons, including an increase in the size of the tumor body, insufficient blood supply to the central portion of the tumor, the softening, necrosis, and ulceration of tumor tissues, bleeding from the surface of the tumor ulcer, and the rupture of small and medium blood vessels[6]. An interesting phenomenon discovered in this study was that in cases of upper gastric cancer, tumors accompanied by OB tended to be classified as an earlier pathological stage than tumors without OB; however, lower gastric adenocarcinomas did not follow this trend. Why did gastric adenocarcinomas with OB at different locations exhibit such different pathological presentations? Koh et al[2] examined the effects of different hemostasis treatment methods on bleeding resulting from gastric cancer and found that when transarterial embolization was performed after endoscopic hemostasis had failed, the left gastric artery was the predominant problematic artery. These results suggested that it would be challenging to stop gastric cancer bleeding from the left gastric artery. The current study also found that in 82.1% of upper gastric cancer patients with OB, the tumor was located in the cardia and the lesser curvature, for which the left gastric artery is the main feeding artery. Based on the combination of the results of the present study and the findings of Koh et al[2], we propose the following hypothesis. The left gastric artery and its branches are adjacent to the celiac artery and have high pressure and rich blood flow. Therefore, tumors in the cardia and the lesser curvature grow rapidly and readily display invasion and necrosis. When tumor tissue infiltration destroys branches of the left gastric artery, any resulting bleeding may be resistant to self-coagulation because these branches have high vascular pressure; thus, OB can easily occur in these cases. Therefore, OB provides an opportunity for the early diagnosis and treatment of upper gastric cancer patients. Lower gastric adenocarcinomas typically arise at the gastric antrum. The blood supply to this location is primarily provided by the right gastric artery and the right gastroepiploic artery, which originate from the proper hepatic artery and the gastroduodenal artery, respectively. Both of these arteries are far from the celiac artery and have relatively low pressure. When tumor tissues invade the branches of these arteries, the resulting bleeding can relatively rapidly coagulate, and the subsequent recurrence and non-OB cause fibrosis and thrombosis near the blood vessels. Thus, OB does not readily occur in these cases. This hypothesis requires additional validation by future studies. Another process could cause OB resulting from a proximal gastric adenocarcinoma. The cardia is narrower than other parts of the stomach, and tumor tissue infiltration increases vascular fragility at this location. When food passes the cardia, it is in a rough, undigested state. Therefore, mechanical stimuli can easily cause the rupture of blood vessels at this site.

Bleeding resulting from gastric cancer is closely associated with the vascular composition of the tumor. In 1971, Folkman[19] proposed that tumor growth is dependent on angiogenesis. Furthermore, he suggested that tumor cells and blood vessels constitute a highly integrated environment. Angiogenesis plays an important role in the process of cancer development[20,21]. High intratumor microvessel density correlates with high AJCC stage and lymph node metastasis. Intratumor microvessel density was found to have independent prognostic significance to cancer[22], including gastric cancer[23], compared to traditional prognostic markers based on multivariate analysis. Studies in Japan and China[24] showed that the vascular composition of gastric cancer can be categorized into hypovascular and hypervascular tumors, corresponding to Borrmann types II-III and Borrmann types I and IV gastric adenocarcinomas, respectively. In the current study, among upper gastric cancer cases, OB was observed more frequently in cases of early gastric cancer and Borrmann type I tumors. There was no difference in the percentage of Borrmann types II-IV tumors between upper gastric cancer patients with and without OB. Is the microvessel density or the Borrmann type of gastric adenocarcinomas associated with hemorrhaging from these tumors? Additional large-scale studies are warranted to precisely determine the impact of intratumor microvessel density on hemorrhaging from gastric cancer.

To determine whether selection bias may have influenced the conclusions of the current study, we investigated all patients receiving all types of therapies from our database and found that the rate of OB was similar between resectable and unresectable tumors. Therefore, selection bias was not possible for the whole patient group and upper gastric cancer cohort. But in the lower gastric cancer cases, the percentage of patients with OB in patients who underwent radical gastrectomy was lower than the percentage of patients with OB in patients who did not undergo radical gastrectomy. Including patients who did not undergo radical gastrectomy, the overall 3-year survival rate may be lower in patients with OB than in patients without OB.

Cases of proximal gastric cancer, particularly cancer of the gastric cardia, only manifest as significant symptoms such as epigastric pain, weakness, weight loss and obstruction after the tumor has deeply infiltrated the gastric tissue. Therefore, most patients already have mid- to late-stage cancer by the time that they seek treatment to relieve the symptoms of eating difficulties and abdominal pain. As a result, these patients exhibit a poor prognosis. A subset of patients without OB have occult bleeding; however, this type of bleeding is generally not discovered in a timely fashion, resulting in the delay of disease treatment. Because there is currently no reliable tumor screening system in China, early diagnoses are difficult to obtain. Thus, OB provides an opportunity to diagnose proximal gastric cancer at an earlier pathological stage, thereby significantly improving patient prognosis.