Journal of Surgical Oncology

The Decision Criterion of Histological Mixed Type in ‘‘T1/T2’’ Gastric

Carcinoma—Comparison Between TNM Classification and Japanese

Classification of Gastric Cancer

HIROKI SHIMIZU, MD,1 DAISUKE ICHIKAWA, MD,1* SHUHEI KOMATSU, MD,1 KAZUMA OKAMOTO, MD,1

ATSUSHI SHIOZAKI, MD,1 HITOSHI FUJIWARA, MD,1 YASUTOSHI MURAYAMA, MD,1

YOSHIAKI KURIU, MD,1 HISASHI IKOMA, MD,1 MASAYOSHI NAKANISHI, MD,1 TOSHIYA OCHIAI, MD,1

YUKIHITO KOKUBA, MD,1 MITSUO KISHIMOTO, MD,2 AKIO YANAGISAWA, MD,2 AND EIGO OTSUJI, MD1

1Division of Digestive Surgery, Department of Surgery, Kyoto Prefectural University of Medicine, Kawaramachihirokoji, Kamigyo-ku,Kyoto, Japan

2Department of Pathology, Kyoto Prefectural University of Medicine, Kawaramachihirokoji, Kamigyo-ku, Kyoto, Japan

Background: This study was designed to evaluate the clinical significance of undifferentiated component in differentiated T1/T2 gastric

adenocarcinoma.

Methods: Two hundred thirty-one patients who underwent curative gastrectomy were diagnosed pathologically as differentiated type T1/T2

gastric cancer according to Japanese Classification of Gastric Carcinoma (JCGC). The patients were divided into subgroups, pure differentiat-

ed type (pure D group, 181 patients) and differentiated-predominant mixed type (D > U group, 51 patients). The clinicopathological features

of D > U group were compared with those of pure D group, and also those of undifferentiated-predominant type (U > D group).

Results: Patients in D > U group were more likely to have larger and deeper tumors with lymphatic invasion and metastases than pure D

group. However, there was no significant difference in clinicopathological factors between D > U and U > D groups, except for depth of

tumor invasion. The postoperative 5-year survival rate of D > U group was significantly poorer than that of pure D group (88% and 98%,

P ¼ 0.011). Multivariate analysis revealed the presence of undifferentiated component was an independent prognostic factor.

Conclusions: The presence of undifferentiated component in differentiated T1/T2 gastric cancer is associated with tumor progression. There-

fore, the decision criterion of histological mixed type in TNM classification is better suited than JCGC in T1/T2 gastric cancer.

J. Surg. Oncol. � 2011 Wiley Periodicals, Inc.

KEY WORDS: classification; mixed histological type; prognostic factor

INTRODUCTION

Gastric cancer is known as one of the most aggressive malignan-

cies and is the second most lethal cancer globally [1]. Recent advan-

ces in diagnostic tools and routine endoscopic screening, however,

have led to early detection in East Asia, especially Japan and Korea,

and therefore, less invasive treatments, such as endoscopic treatments

and laparoscopic surgery, have been treatment options for this lethal

disease in these areas. The histological type has been recognized as

one of the predictive factors for lymph node metastasis [2–4] and

prognosis of gastric cancer patients [5–7]. Indeed, the histological

type is defined as one of the indicative factors for endoscopic treat-

ments in the treatment guidelines for gastric cancer in Japan [8].

Therefore, accurate diagnosis of histological type is indispensable in

determining treatment options.

Tumor tissue does not necessarily consist of a single histological

component, and sometimes contains a mixture of several different

histological components to some degree. Little, however, is known

about the clinical significance of the histological mixture; as such,

there has been a small discrepancy in the classifications of histologi-

cal mixed type between two widely used staging systems, the Japa-

nese Classification of Gastric Carcinoma (JCGC) [9] and the TNM

classification [10]. The mixed histological type has been classified

on the basis of the predominant histological component in JCGC,

but on the basis of the poorest histological component in TNM.

At present, only patients with clinically T1 (cT1) tumors can be

subjects for the limited treatments, such as endoscopic treatment and

the limited surgery with modified lymphadenectomy. Subgroup of

patients with T2 tumors showed better prognoses, who might be next

candidates for the modified surgery [11]. For the reason, we selected

patients with T1/T2 gastric cancer as subjects of this study for possi-

ble targets of the modified treatments. In this study, we examined the

clinicopathological features of the differentiated-predominant type in

T1/T2 gastric cancer by comparison with the pure differentiated type

and also undifferentiated-predominant type, and decided which of

the staging systems would be more suitable for determination of

treatment options in gastric cancer patients.

MATERIALS AND METHODS

A consecutive series of 423 patients with diagnosis of T1/T2

gastric cancer according to the criteria of JCGC was studied. All

patients underwent curative gastrectomy with lymph node dissection

Abbreviations: JCGC, Japanese Classification of Gastric Carcinoma;ESD, endoscopic submucosal dissection; HR, hazard ratio; CI, confidenceinterval.

Hiroki Shimizu and Shuhei Komatsu contributed equally to this work.

*Correspondence to: Dr. Daisuke Ichikawa, MD, Division of DigestiveSurgery, Department of Surgery, Kyoto Prefectural University ofMedicine, 465 Kajii-cho, Kawaramachihirokoji, Kamigyo-ku, Kyoto 602-8566, Japan. Fax: 81-75-251-5522. E-mail: ichikawa@koto.kpu-m.ac.jp

Received 19 April 2011; Accepted 24 November 2011

DOI 10.1002/jso.23010

Published online in Wiley Online Library(wileyonlinelibrary.com).

� 2011 Wiley Periodicals, Inc.

in the Division of Digestive Surgery, Kyoto Prefectural University of

Medicine, between 1999 and 2008. Patients who were treated by

chemotherapy before surgery were excluded from this study. Two

hundreds thirty-one of the 423 patients were diagnosed as differenti-

ated gastric cancer according to the criteria of JCGC, whereas 192

patients as undifferentiated. The clinicopathological factors of these

patients were obtained from hospital records.

The resected stomach was opened and was placed on a flat board

with the mucosal side up, and fixed in 10% buffered formalin solu-

tion. After fixation, the neoplasms were sectioned on the maximum

cross-sectional plane parallel to the lesser curvature line based on

the general rules of the JCGC [9]. In general, the tumor had been

sectioned in its entirety parallel to the reference line at intervals of

5 mm. The resected specimens were embedded in paraffin, and

stained by hematoxylin and eosin. The histological quantitative

predominance were recorded by examining all stained specimens by

at least two pathologists. The histological type of tumor was divided

into two major histological categories: (1) intestinal, expanding, or

differentiated type and (2) diffuse, infiltrative, or undifferentiated

type [12,13]. The JCGC definition states that differentiated type

includes tubular and papillary adenocarcinomas, which arise from

gastric mucosa with intestinal metaplasia, and that undifferentiated

type includes poorly differentiated adenocarcinoma, signet ring cell

carcinoma, and mucinous adenocarcinoma, which arise from ordi-

nary gastric mucosa without intestinal metaplasia [14]. The staging

of gastric cancer was performed using the histopathological findings

of tumor according to the criteria of JCGC. Tumor invasions to mu-

cosa, submucosa, and muscle layer were classified as T1a, T1b, and

T2, respectively, and lymph node metastases numbering 0, from 1 to

2, from 3 to 6, and more than 7 nodes were classified as N0, N1, N2,

and N3, respectively, on the basis of the 14th JCGC, the same as the

7th TNM classification. The differentiated type of gastric cancer, di-

agnosed according to the criteria of JCGC, consist of two subgroups:

(1) pure differentiated component with no undifferentiated compo-

nent (pure D group) and (2) differentiated-predominant mixed type

with less than 50% undifferentiated component (D > U group). The

clinicopathological features of the D > U group were examined by

comparison with those of the pure D group, and also those of the

undifferentiated-predominant type (U > D group; 60 patients, who

were subset of 192 gastric cancer patients diagnosed as undifferenti-

ated according to the criteria of JCGC except for those with the pure

undifferentiated), which an undifferentiated component accounted

for more than 50% and diagnosed as the undifferentiated type of

gastric cancer according to the criteria of JCGC.

The prognostic impact of a mixture of undifferentiated component

was evaluated by univariate and multivariate analyses. There were 11

and 2 patients died from irrelevant other disease during their follow-

up periods in pure D and D > U groups, respectively. In order to

examine the oncological outcome of histological predominance, we

used cause-specific overall survival rate in this study. Chi-squared

and Fisher’s exact probability tests for categorical variables and the

Mann–Whitney U-test for unpaired data for continuous variables

were performed to compare the clinicopathological characteristics

between two groups. The cumulative survival rates were calculated

using the Kaplan–Meier method, and the log-rank test was used for

assessment of differences between prognostic factors. Multivariate

analysis of factors influencing cause-specific overall survival rate

was performed using the Cox proportional hazards model. A P-value

of less than 0.05 was considered to be statistically significant.

RESULTS

The mean age of patients was 66 years (range: 32–91), and the

male:female ratio was 3.9. Of all 231 patients diagnosed as differen-

tiated gastric cancer according to the criteria of JCGC, 180 patients

(78%) were classified in the pure D group, and 51 patients (22%) in

the D > U group. The median follow-up period was 1,049 days

(range: 8–3618).

The clinicopathological characteristics of the two groups and also

the U > D group are shown in Table I. Patients in the D > U group

had significantly larger (P ¼ 0.049) and deeper tumors (P < 0.001)

with more advanced lymphatic invasion (P ¼ 0.022) and metastases

(P < 0.005) than the pure D group. There was no significant differ-

ence in gender, age, tumor location, tumor macroscopic type, and

venous invasion between the two types. On the other hand, there was

no significant difference between D > U and U > D groups except

for depth of tumor invasion.

Regarding the prognoses, recurrences developed in 6 (2.6%) of

231 patients, 3 (1.7%) in the pure D group and 3 (5.9%) in the

D > U group. In the pure D group, peritoneal metastasis developed

in 2 patients and lymph node metastasis in 1 patient, whereas lymph

node metastasis developed in 2 patients and liver metastasis in 1

patient in the D > U group.

The postoperative cause-specific 5-year survival rate of patients in

the D > U group was significantly poorer than that of the pure D

group (88% and 98%, P ¼ 0.011; Fig. 1). Because the prognosis

was also affected by other factors in which there was a significant

difference between D > U and pure D groups (Table I), and then

these significant prognostic factors and the presence of undifferenti-

ated component were analyzed by multivariate analysis. The multi-

variate analysis revealed that the presence of undifferentiated

component and also venous invasion were independent prognostic

factors (Table II).

DISCUSSION

Gastric cancer has various histological types, and each of them

exhibits different characteristics. It is well known that histological

type is one of the most important factors to predict the prognosis and

recurrence patterns in patients with gastric cancer. In early gastric

cancer, histological type also affects the extent of lymph node metas-

tasis. In particular, the undifferentiated type is one of the indepen-

dent risk factors of lymph node metastasis [2–4]. In advanced gastric

cancer, the histological type is one of the independent prognostic

factors in addition to the depth of invasion and lymph node metasta-

sis [5–7]. Therefore, the histology of gastric carcinoma has been

regarded as an important prognostic factor and is widely used in

making decisions on treatment strategy. There, however, have been

few studies about the histological mixture of differentiated and

undifferentiated components in gastric cancer, and little is known

about the clinical significance of a histological mixture such as an

undifferentiated component in differentiated-predominant gastric

cancer. In this regard, there has been a small discrepancy between

JCGC and TNM classifications. In detail, the mixed histological type

of gastric cancer has been classified on the basis of the predominant

type in JCGC; however, it has been classified on the basis of the

poorest differentiated component in TNM classification. There

has been no universal standard for definition of a histological mix-

ture in gastric cancer. Therefore, the importance of evaluating the

significance of quantitative or qualitative predominance should be

emphasized.

Recently, limited gastrectomy with laparoscopic surgery [15,16]

and endoscopic submucosal dissection (ESD) with narrow band im-

aging magnified endoscopy [17] have emerged as new less invasive

technologies in early gastric cancer treatments. Gotoda et al. [18]

reported that, in 3,016 patients with T1a gastric cancer, none of the

1,230 differentiated cases less than 30 mm in size were associated

with lymph node metastases (95% CI: 0–0.4%); in contrast, 18

(2.2%) of 821 undifferentiated cases less than 30 mm in size were

associated with lymph node metastases. These results have changed

2 Shimizu et al.

Journal of Surgical Oncology

our conception of early gastric cancer treatments. Consequently, the

criteria defined in the guidelines have been widely accepted in treat-

ments for early gastric cancer and histological type has been

regarded as one of the most important factors to make a decision of

indication for ESD and limited surgical lymph node dissection in

laparoscopic gastrectomy.

In this study, we showed a significant difference between pure

differentiated type and differentiated-predominant mixed type in

several clinicopathological factors, such as tumor size, lymphatic

invasion, depth of tumor invasion, lymph node metastasis, although

these two types have been treated as the same histological type in

JCGC. On the other hand, there was no significant difference be-

tween D > U and U > D groups except for depth of tumor invasion,

although these two types have been treated as the different histologi-

cal type in JCGC but the same histological type in TNM classifica-

tion. These results clearly demonstrated that the D > U group was

more like the U > D group than the pure D group, especially in the

features of lymphatic spreading. Some authors have reported the

clinical significance of histological mixed type in gastric cancer

[19–24]. Tajima et al. [19] indicated that, in patients with T1b gastric

cancer, the risk of lymph node metastasis was significantly higher in

differentiated-predominant mixed type than in pure differentiated

type, which was consistent with our present result. Moreover, we

demonstrated that the prognosis of patients in the D > U group was

significantly poorer than that in the pure D group in the present

study. Because the D > U group tended to include more advanced

cases than the pure D group, we conducted the multivariate analysis

with other prognostic factors. As a result, the presence of undifferen-

tiated component was found to be one of the independent prognostic

factors in patients with differentiated-type T1/T2 gastric cancer.

Therefore, additional treatment options and/or intensive follow-up

might be recommended for patients who histological examination

Fig. 1. Comparison survival curves according to histological groupsin T1/T2 gastric cancer patients. Postoperative cause-specific survivalof the D > U group was significantly poorer than that of the pure Dgroup (P ¼ 0.011).

TABLE I. Correlation Between Histological Groups and Clinicopathological Characteristics

Variable n

Histological group

Pure D D > U U > D

Total 291 180 51 60

Sex n.s. n.s.

Male 220 145 (81%) 39 (77%) 36 (60%)

Female 71 35 (19%) 12 (24%) 24 (40%)

Age n.s.

<65 128 77 (43%) 24 (47%) 27 (45%)

�65 163 103 (57%) 27 (53%) 33 (55%)

Size (mm) � n.s.

<25 123 88 (49%) 17 (33%) 18 (30%)

�25 168 92 (51%) 34 (67%) 42 (70%)

Location n.s. n.s.

Lower, Middle 221 139 (77%) 40 (78%) 42 (70%)

Upper 70 41 (23%) 11 (22%) 18 (30%)

Macroscopic type n.s. n.s.

Type 0 246 158 (88%) 43 (84%) 45 (75%)

Non-type 0 45 22 (12%) 8 (16%) 15 (25%)

Lymphatic invasion � n.s.

Negative 219 147 (82%) 34 (67%) 38 (63%)

Positive 72 33 (18%) 17 (33%) 22 (37%)

Venous invasion n.s. n.s.

Negative 260 163 (91%) 46 (90%) 51 (85%)

Positive 31 17 (9%) 5 (10%) 9 (15%)

T stagea �� ��T1a 125 97 (54%) 11 (22%) 17 (28%)

T1b 120 65 (36%) 34 (67%) 21 (35%)

T2 46 18 (10%) 6 (12%) 22 (33%)

Lymph node metastasis �� n.s.

Negative 252 163 (91%) 38(75%) 51 (85%)

Positive 39 17 (9%) 13 (26%) 9 (15%)

aJapanese Classification of Gastric Cancer.�P < 0.05.��P < 0.01.

Mixed Histology in Gastric Cancer 3

Journal of Surgical Oncology

prove the presence of undifferentiated component in specimens

resected by limited treatments, such as ESD or local resections, even

they were diagnosed preoperatively as differentiated-type of gastric

cancer. If treated by standard operations, the clinical significance of

histological mixed type in T2 gastric cancer seems less critical than

that in T1 tumor. However, subgroups of patients with T2 tumors

might be next candidates for modified operations.

Concerning recurrence, recurrence patterns of gastric cancer were

reported to be influenced by the histological type [5,25]. Sano et al.

[25] reported the recurrence pattern according to histological findings

in patients with early gastric cancer, in which hematogenous metas-

tasis was dominant in the differentiated type and local or peritoneal

metastasis was in the undifferentiated type. In the present study, the

number of recurrent patients was only 6 of 231 patients (2.6%).

Therefore, we could not show a tendency of recurrence pattern

according to the histology. However, the recurrence rate in differenti-

ated-predominant mixed type was indeed high compared with that in

the pure differentiated type: 5.9% (3/51) versus 1.7% (3/180). Close

follow-up should be recommended for patients with undifferentiated

component even in differentiated gastric cancer.

CONCLUSIONS

The presence of an undifferentiated component in differentiated

T1/T2 gastric cancer is associated with tumor progression, especially

lymphatic spreading. Therefore, the decision criterion of histological

mixed type in TNM classification is better suited than JCGC for

determination of treatment options in T1/T2 gastric cancer patients.

REFERENCES

1. Ferlay J, Shin HR, Bray F, et al.: Estimates of worldwide bur-den of cancer in 2008: GLOBOCAN 2008. Int J Cancer2010;127:2893–2917.

2. Popiela T, Kulig J, Kolodziejczyk P, et al.: Long-term results ofsurgery for early gastric cancer. Br J Surg 2002;89:1035–1042.

3. Gotoda T, Sasako M, Ono H, et al.: Evaluation of the necessityfor gastrectomy with lymph node dissection for patients withsubmucosal invasive gastric cancer. Br J Surg 2001;88:444–449.

4. Folli S, Morgagni P, Roviello F, et al.: Risk factors for lymphnode metastases and their prognostic significance in early gas-tric cancer (EGC) for the Italian Research Group for GastricCancer (IRGGC). Jpn J Clin Oncol 2001;31:495–499.

5. Adachi Y, Yasuda K, Inomata M, et al.: Pathology and progno-sis of gastric carcinoma: Well versus poorly differentiated type.Cancer 2000;89:1418–1424.

6. Noda S, Soejima K, Inokuchi K: Clinicopathological analysis ofthe intestinal type and diffuse type of gastric carcinoma. Jpn JSurg 1980;10:277–283.

7. Riberio MM, Sarmento JA, Simoes MAS, et al.: Prognostic sig-nificance of Lauren and Ming classifications and other patholog-ic parameters in gastric carcinoma. Cancer 1981;47:780–784.

TABLE II. Univariate and Multivariate Analyses of Prognostic Factors of Differentiated Type

Factor

Number

(n ¼ 231)

5-Year survival

rate (%)a

Univariateb Multivariatec

P-value HR 95% CI P-value

Sex

Male 184 95.0 0.2599 —

Female 47 100.0

Age

<65 101 96.2 0.7220

�65 130 95.9

Size (mm)

<25 105 96.6 0.7271

�25 126 95.6

Location

Lower, Middle 179 97.0 0.6088 —

Upper 52 94.0

Macroscopic type

Type 0 201 96.1 0.7007

Non-type 0 30 95.0

Lymphatic invasion

Negative 181 98.9 0.0014 2.423 0.216–27.142 0.473

Positive 50 86.1

Venous invasion

Negative 209 99.3 <0.0001 23.653 2.025–276.311 0.012

Positive 22 75.3

T stage

T1 207 96.3 0.4871 2.755 0.270–27.778 0.392

T2 24 93.8

Lymph node metastasis

Negative 201 97.8 0.0003 5.038 0.628–40.391 0.128

Positive 30 83.1

Histological group

Pure D 180 97.8 0.0110 12.146 1.375–107.317 0.025

D > U 51 88.4

HR, hazard ratio; CI, confidence interval.aSurvival rate was calculated based on cause-specific death.bKaplan–Meier method, and the statistical significance was determined by log-rank test.cMultivariate survival analysis was performed using Cox’s proportional hazard model.

4 Shimizu et al.

Journal of Surgical Oncology

8. Japanese Gastric Cancer Association. Japanese gastric cancertreatment guidelines 2010 (ver. 3). Gastric Cancer 2011;14:113–123.

9. Japanese Gastric Cancer Association. Japanese classification ofgastric carcinoma: 3rd English edition. Gastric Cancer 2011;14:101–112.

10. Sobin L, Gospodarowicz M, Wittekind C, editors. Internationalunion against cancer. TNM classification of malignant tumours,7th edition. New York: Wiley-Blackwell; 2010.

11. Tokunaga M, Hiki N, Fukunaga T, et al.: Better prognosis of T2gastric cancer with preoperative diagnosis of early gastric can-cer. Ann Surg Oncol 2009;16:1514–1519.

12. Nakamura K, Sugano H, Takagi K: Carcinoma of the stomachin incipient phase: Its histogenesis and histologic appearances.Gann 1968;59:251–258.

13. Lauren P: The two histologic main types of gastric carcinoma:Diffuse and so-called intestinal type carcinoma. Acta PatholMicrobiol Scand 1965;64:31–49.

14. Tahara E, Semba S, Tahara H: Molecular biological observationin gastric cancer. Semin Oncol 1996;23:307–315.

15. Huscher CG, Mingoli A, Sgarzini G, et al.: Laparoscopic versusopen subtotal gastrectomy for distal gastric cancer. Ann Surg2005;241:232–237.

16. Kitano S, Shiraishi N, Uyama I, et al.: A multicenter study ononcologic outcome of laparoscopic gastrectomy for early gastriccancer in Japan. Ann Surg 2007;245:68–72.

17. Oyama T, Tomori A, Hotta K, et al.: Clinicopathologicalfeature of the superficial well differentiated gastric adenocar-cinoma with poorly differentiated component [in Japanesewith English abstract]. Stomach Intestine 2007;42:1625–1634.

18. Gotoda T, Yanagisawa A, Sasako M, et al.: Incidence of lymphnode metastasis from early gastric cancer: Estimation with alarge number of cases at two large centers. Gastric Cancer2000;3:219–225.

19. Tajima Y, Murakami M, Yamazaki K, et al.: Risk factors forlymph node metastasis from gastric cancers with submucosalinvasion. Ann Surg Oncol 2010;17:1597–1604.

20. Hanaoka N, Tanabe S, Mikami T, et al.: Mixed-histologic-typesubmucosal invasive gastric cancer as a risk factor for lymphnode metastasis: Feasibility of endoscopic submucosal dissec-tion. Endoscopy 2009;41:427–432.

21. Iwamoto J, Mizokami Y, Ito M, et al.: Clinicopathological fea-tures of undifferentiated mixed type early gastric cancer treatedwith endoscopic submucosal dissection. Hepatogastroenterology2010;57:185–190.

22. Watanabe G, Ajioka Y, Kato T, et al.: Pathological characteris-tics of differentiated-type early gastric carcinoma mixed withundifferentiated-type—Status of lymph node metastasis andmacroscopic features [in Japanese with English abstract]. Stom-ach Intestine 2007;42:1577–1587.

23. Tanabe H, Iwashita A, Haraoka S, et al.: Clinicopathologicalcharacteristics of differentiated mixed-type early gastric carcino-ma with lymph node metastasis [in Japanese with English ab-stract]. Stomach Intestine 2007;42:1561–1576.

24. Fujisaki J, Yamamoto Y, Yamamoto N, et al.: ESD treatment forhistologically mixed-type early gastric cancer (including recur-rence and metastasis to the lymph node) [in Japanese with En-glish abstract]. Stomach Intestine 2007;42:1659–1669.

25. Sano T, Sasako M, Kinoshita T, et al.: Recurrence of earlygastric cancer. Follow-up of 1475 patients and review of theJapanese literature. Cancer 1993;72:3174–3178.

Mixed Histology in Gastric Cancer 5

Journal of Surgical Oncology