Department of Surgery,

s Archer SY, Johnson JJ, Hodin RA. p21 gene regulation during enterocyte differentiation: identification of activator and repressor elements. Association for Academic Surgery, November, 2000. Archer SY, Kim HJ, Johnson JJ, Hodin RA. Cyclin B1 transcriptional repression in colon cancer cells: dual pathways of regulation. Gastroenterology 2001; 120(5): A292. Ayabe T, Satchell DP, Tanabe H, Hagen SJ, Wilson CL, Ouellette AJ. Processing and activation of -defensins in mouse paneth cells. 2001, Gastroenterology 120(5): 968A. Badrichani AZ, Ferran C. A20 and BCL proteins exert a broad and complementary cytoprotective effect in endothelial cells via blockade of NF-kappaB and NFAT. Transplant Proc. 2001;33:450. Bhagat L, Agrawal S, Singh VP, Song AM, van Acker G, Mykoniatis A, Steer ML, Saluja AK. HSP 70 antisense oligonucleotide administration prevents the thermal stress-induced expression of HSP70 in rat pancreas and abolishes the thermal stress-induced protection against caeruleininduced pancreatitis. Gastroenterology 2001;120: A537. Bhagat L, Hietaranta AJ, Singh VP, Song AM, Mykoniatis A, van Acker GJD, Pan A, Steer ML, Saluja AK. Water immersion stress prevents trypsinogen activation in the caerulein model of pancreatitis by altering intracellular calcium. Pancreas 2000;21: A432. Bhagat L, Saluja AK, Singh VP., Song AM, van Acker G, Mykoniatis A, Steer Ml. Role of actin cytoskeleton in caerulein-induced intra-acinar cell activation of trypsinogen in an in vitro model of pancreatitis. Gastroenterology 2001;120: A236. Bhagat L, Saluja AK, van Acker G, Singh VP,. Song A, Mykoniatis A, Steer ML. Lysosomal hydrolase/digestive zymogen co-localization and intra-acinar cell activation of trypsinogen: which is the horse and which is the cart? Gastroenterology 2001;120: A540.


Introduction
Regional incidences of gastric cancer are variable, and the prevalence of the disease shows a declining tendency.However, gastric cancer remains the fourth most common malignancy worldwide, and there is a higher incidence of the disease in Far Eastern Asian countries, such as Korea and Japan.(1) This is in contrast to Western countries where the incidence is much lower.However, the West Clinicopathological Features of Upper Third Gastric Cancer during a 21-Year Period (Single Center Analysis) has relatively higher proportions of UTG compared with Asian countries.(2)Some recent studies have reported an increasing trend in the incidence of UTG in Japan, China, and Korea.(3)(4)(5)(6) The 2004 Nationwide Gastric Cancer Registry also reported similar findings.
(7) Such increases could be related to recent changes in lifestyles of the Korean population.Some studies in both Western and Eastern countries have reported a poorer prognosis with UTG.(8,9) Others, however, found that the prognosis of patients with UTG was no poorer than that of patients with MLG in each equal TNM stage.

Materials and Methods
We retrospectively analyzed the medical records of

Results
Among the 12,300 study group, 1,111 patients were excluded due to pre-operative chemotherapy (N=39), remnant gastric can- When TNM stages were compared, the UTG group showed a significantly larger tumor size than the MLG group in all TNM stages except stage IV.There were some missing values in the cell differentiation and Lauren's classification categories due to limitations in medical records but the UTG group exhibited a significantly higher proportion of diffuse Lauren's type and a higher proportion of undifferentiated cell types compared to the MLG group (Table 1).
Chronologically, the proportion of UTG was 2.6% in 1986 and rapidly increased to 12.5% in 1992.Subsequently, the disease  showed a slow increase, rising to 14.2% in 2006.The increase in the rate of UTG was estimated to be about 1.6% per year before 1992 (P＜0.001); however, this dropped to 0.21% per year after 1992 (P=0.028) (Fig. 1).The median age for UTG was in the sixth decade of life from 1986 to 1995.This shifted to the seventh decade of life after 1996, a finding that was also observed in the MLG group.A chronological review of the TNM stage showed that stage I and II cancers were more prevalent in both the UTG and MLG groups than in the past (Fig. 2).Comparison of the individual stratified TNM stages showed that the 5-year survival rate of the UTG group was significantly poorer than that of the MLG in stage I-III (Fig. 3).
Tumor size, tumor location, Lauren's classification, cell differentiation, radicality, and TNM stage were identified as prognostic factors in univariate analysis.However, in multivariate analysis, sex, tumor size, tumor location, radicality, and TNM stage were found to be independent prognostic indicators (Table 2).

Discussion
The aim of this study was to evaluate chronological changes in the proportion of UTG during the past 21 years and to determine the prognosis of patients with UTG.In contrast to previous reports of an increasing trend in UTG worldwide, we found that the UTG proportions are nearly not changing these days.Prior to 1992, the prevalence of UTG showed a relatively rapid rise, with rates increasing up to 1.69% annually.However, this rate has decreased to 0.21% per year since 1992 at SNUH (Fig. 1).Those changes are not likely the result in real incidence but may be attributable to increasing rates of screening for early gastric cancer (EGC); there was a simultaneous increase in the proportion of EGC identified during the same period.The changes may also be due to patient characteristics, with specific types of patients being referred to the SNUH.According to a nationwide survey of gastric cancer in Korea in 2004, the prevalence of UTG had increased from 11.2% to  Many studies of UTG cite late detection as an important factor in the poor prognosis of the disease; the most common symptoms of UTG, such as weight loss and gastroesophageal reflux are usually very insidious and do not occur until the disease is at an advanced stage.(8,21,25) Some studies have argued that early detection can improve the prognosis if radical resection can be performed.(5,10,11) In our study, early stage UTG showed nearly a 90% 5-year survival rate, suggesting that more effort should be expended on the early detection and treatment of upper third gastric lesions.As noted previously by Yokoi et al., (26) there are some limitations associated with UTG screening, such as the presence of trivial mucosal lesions, simple color changes, and technical difficulties.To ensure that UTG lesions are not overlooked, the endoscopist should pay special attention when examining lesions in the upper third stomach.
As this was a retrospective analysis and patients who did not undergo surgery because of advanced disease were not included, our study does not represent all the clinical aspects of UTG in Korea.However, from a surgical perspective, it may offer useful information.
In conclusion, the prevalence of UTG showed a temporary increasing trend prior to the 1990s, with an annual rise of 1.69%.
However, since then, the annual rate of increase has been just 0.21% at SNUH.The UTG group showed significantly poorer 5-year survival rates compared with the MLG group in each TNM stage except stage IV.Tumor size, tumor location, radicality, and TNM stage were independent prognostic indicators in both univariate and multivariate analysis.

( 10 , 11 )
It remains unclear whether the disease prognosis is due to different biologic characteristics or late detection.In this study, we evaluated chronological changes in the prevalence of UTG, the prognosis of UTG patients who underwent gastric resection at SNUH during the last 21 years, and potential prognostic indicators of UTG.

Fig. 2 .Fig. 3 .
Fig. 2. Chronological changes in gastric cancer stages based on the tumor location.Th e proportion of stage I and II cancer increased in both groups, whereas the proportion of stage III and IV decreased; it is not clear whether this represents a real increase in the prevalence of EGC or is simply the result of greater diagnoses due to increased health screening.(A) Annual TNM stage distribution in UTG.(B) Annual TNM stage distribution in MLG.EGC = early gastric cancer; MLG = middle or lower third gastric cancer.

14 .
2%.(17)However, another nationwide survey of the disease in 2009 reported a UTG prevalence of just 13.4% in Korea and concluded that the increasing trend of UTG in Western countries has not been seen in Korea, a finding consistent with our study.(18)We observed a shift in the peak age distribution of UTG from the sixth to the seventh decade of life, which may be explained by the aging population in Korea.The decrease that we observed in TNM stage III and IV patients and the increase in stage I and II compared to the prevalence in the past could be attributable to technical advances in endoscopy and increased diagnosis of patients with early stage disease (Fig.2).

Table 1 .
The number of patients in the UTG group was 1,260 (11.3%), and the mean age was 54.7 years.The male to female ratio was about 2 : 1, which is not different from the MLG group.The proportion of patients with advanced gastric cancer was significantly higher in the UTG group compared with the MLG group (77.7% vs. 60.4%).Total gastrectomy was the most frequent type of operation performed in the UTG group; proximal gastrectomy was performed in 11.8% of the patients.The proportion of R0 resection was slightly lower in the UTG group compared with the MLG group (91.4% vs. 94%).The distribution of the TNM stage showed more advanced stage cancer in UTG than in MLG patients, and the mean number of cancer-positive lymph nodes was greater in the UTG group.The mean tumor size in the UTG group was larger than that in the MLG group (5.2 cm vs. 4.4 cm).

Table 1 . Continued
(12)= upper third gastric cancer; MLG = middle or lower third gastric cancer; TG = total gastrectomy; PG = proximal gastrectomy; DSG = distal subtotal gastrectomy; EGC = early gastric cancer; AGC = advanced gastric cancer; LN = lymph node; WD = well diff erentiated; MD = moderately diff erentiated; PD = poorly diff erentiated.*Others=gastrojejunostomy(1),wedgeresection(1), pylorus preserving gastrectomy (1) in UTG and gastrojejunostomy(16), wedge resection(12), pylorus preserving gastrectomy (40), exploration (2) in MLG; † The total number of subjects in cell differentiation and Lauren's classification category is less than that stated enrolled number of patient due to missing value; ‡ P<0.05, statistically signifi cant.Fig. 1. Chronological changes in gastric cancer patient numbers based on tumor location.(A)Annualnumber of UTG and MLG at SNUH.UTG was 2.6% (9 patients) in 1986 and rapidly increased to 12.5% (66 patients) in 1992.It reached a peak of 16.2% (96 patients) in 2004.A temporary drop in the rate to 12.5% (40 patients) was observed in 2000, which may be related to a medical strike at the time in Korea for separating medical practice and pharmaceutical dispensing system.(B) Estimated increase in the rates of UTG by linear regression analysis.Before 1992, UTG had been increasing at the rate of 1.69% per year (P<0.001,R 2 =0.913); since 1992, the rate has reduced at 0.21% per year (P=0.028,R 2 =0.321).UTG = upper third gastric cancer; MLG = middle or lower third gastric cancer; SNUH = Seoul National University Hospital.

Table 2 . Univariate analysis of prognostic factors in gastric cancer by Kaplan-Meier method and multivariate analysis according to Cox proportional hazards model. Size, TNM stage, radicality, and tumor location were analyzed for significant prognostic factors by both univariate and multivariate analysis
(22)he prognosis analysis, patients with UTG showed a poorer prognosis than those with MLG, with significantly lower 5-year survival rates in each TNM stage, except stage IV.The poor prognosis with UTG compared with MLG may be due to a number of UTG showed a higher proportion of diffuse type cells, whereas MLG exhibited more intestinal type cells (Table1), all of which may be related to the differences in the pathophysiology of UTG and MLG.KcColl et al.(22)proposed that at least two differ- (21) confidence interval; UTG = Upper third gastric cancer; MLG = Middle or lower third gastric cancer.*Exp(B)=Hazard ratio; † P<0.05, statistically signifi cant.UTG showed a significantly higher incidence of undifferentiated cell types than MLG, a finding consistent with other studies.Maeda et al.(21)reported that UTG had a more aggressive disease course and poorer prognosis due to undifferentiated or poorly differentiated cancer cells leading to advanced gastric cancer.In Lauren's clas-sifications,