Gastric cancer is the fourth most common malignancy in the world and accounts for the second-highest number of cancer-related deaths after lung cancer.¹ It generally occurs between the fifth and seventh decades of life, and 70% of cases occur in developing countries.² The prognosis of advanced-stage gastric cancer is poor, and the 5-year overall survival (OS) rate is less than 30%. With developments in new surgical techniques, radiotherapy (RT), and chemotherapy (CT), the 5-year OS rates of early-stage disease have exceeded 95%.³ In addition to TNM staging, new guiding clinical and pathological markers are required to accurately evaluate patients who would benefit from personalized treatment after surgery and to facilitate clinical management.⁴

Tumor budding (TB), which is thought to constitute the basis for invasion and metastasis, is defined as the presence of single malignant cells or the presence of undifferentiated malignant cell clusters consisting of 2–4 cells, called tumor buds, in the tumor stroma on the invasive surface of the tumor. An important step in the invasion process is the loss of intercellular connections and stromal invasion. Therefore, a close relationship is thought to exist between TB and the epithelial-mesenchymal transition, where adhesion connections between cells disappear, with the loss of membranous E-cadherin and β-catenin expression of epithelial cells.⁵

TB was first described by Imai in the Japanese medical literature in 1949.⁶ Subsequent studies have been conducted on the association of TB with the prognosis in many solid malignancies such as breast, gastric, colon, rectal, and cervical cancers. In 1989, Morodomi et al⁶ found that TB increased the risk of lymphatic invasion and lymph node metastasis in rectal cancers, and following studies showed it to be associated with recurrence and distant metastasis in colorectal cancers.^7,8 Currently, high TB in colorectal cancer is accepted as a well-known poor prognostic factor, and routine evaluation of TB has been recommended.⁹ The International Association for Cancer Control (UICC) has defined TB as an additional independent prognostic factor in colorectal. The UICC emphasized the necessity of studies to determine the most appropriate assessment method before TB is used as a routine marker of metastasis and lymph node invasion risk, especially in T1 tumors.¹⁰

The prognostic value of TB in gastric cancer is not well known, and there is a limited number of relevant clinical studies. Therefore, we investigated the relationship between patients’ clinicopathological characteristics and TB, as well as the effect of the budding level on prognosis.

Study design

This study had a retrospective and cross-sectional design, and it was organized in accordance with the STROBE statement.¹¹

Participants

In our institutional database, the records of patients who were diagnosed with gastric cancer and underwent surgery at our hospital between 2012 and 2020 were retrospectively reviewed. In total, 107 patients aged between 18 and 85 were included.

Patients with gastric lymphomas or neuroendocrine tumors, metastatic disease at diagnosis, mucinous adenocarcinomas, and signet ring cell carcinomas (since they were non-cohesive at diagnosis), and patients who had another primary cancer were not included.

Data collection tools

Patients’ age, sex, blood type, height, and weight were collected, as well as the date of the first diagnosis, tumor size, tumor location, operation type, number of lymph nodes removed, number of lymph nodes involved, surgical margin, lymphovascular invasion (LVI), perineural invasion (PNI), tumor stage, histological grade, whether the patient received adjuvant CT or RT, recurrence status, and the time from diagnosis to recurrence were.

Pathological slides stained with hematoxylin and eosin were retrospectively evaluated with light microscopy to assess TB. The tumor invasion surface was found by ×10 magnification on light microscopy during scoring, and the area where budding was densely observed was called a “hotspot.” In this region, the cell groups separated from the tumor were counted under ×20 magnification of the objective. The number of buds was counted per 0.785 mm² area, which is the sample area diameter suitable for a lens diameter of 20 mm in accordance with the Colorectal Carcinoma Reporting Protocol that was published in 2016. TB was graded as follows according to the Colorectal Carcinoma Reporting Protocol:

- No tumor budding

- Low grade: 0–4 tumor buds

- Moderate grade: 5–9 tumor buds

- High grade: ≥ 10 tumor buds

Three patients without TB were not included in the analysis. The patients were divided into two groups according to the degree of TB: high (tumor bud count ≥ 10) and low (bud count < 10) (Fig. 1, 2).

Figure 1. High level of tumor budding on invasive tumor surface (H&E stain). (A) High grade tumor budding density (×20 magnification) (asterisk). (B) High grade tumor budding density (×10 magnification) (asterisk).

Figure 2. Low level of tumor budding on invasive tumor surface (H&E stain). (A) Low grade tumor budding density (×10 magnification) (asterisk). (B) Low grade tumor budding density (×20 magnification) (asterisk).

Ethical aspects of the research

Before the study, institutional permission was acquired from the hospital where the data would be collected, as well as ethics committee approval (IRB No: 05.11.2019/21/387). Because of the retrospective nature of this study, the requirement for informed consent was waived. The Helsinki Declaration was followed in the research.

Data analysis

Data were analyzed in SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Patients diagnosed with gastric cancer were staged according to TNM staging system, in which the characteristics of tumor (T), lymph node (N) and metastasis (M) were evaluated according to the recommendations of the American Joint Committee on Cancer.

OS was defined as the time from first diagnosis to the patient’s last visit or death. The TB score was calculated in all patients. The patients were evaluated in terms of OS and recurrence during their follow-up period, and the relationships of the degree of TB with OS, recurrence, and patients’ demographic and histopathological features were investigated.

Statistical analysis

The mean, median, standard deviation, lowest, highest, frequency and ratio values were used to present descriptive statistics. The distribution of variables was measured with Kolmogorov-Smirnov test. The Mann–Whitney U test was used to analyze quantitative independent data. The chi-squared test was used to analyze qualitative independent data, and the Fisher exact test was used when the conditions of the chi-squared test were not met. Cox regression (univariate and multivariate) analysis and the Kaplan-Meier method were used for survival analysis. A P-value of less than 0.05 was considered significant.

Of the 104 patients included, 35 (33.7%) were women, 69 (66.3%) were men, and the mean age at diagnosis was 62 ± 11 years. The mean weight was 69 ± 16 kg, and the mean body mass index (BMI) was 25.3 ± 5.3 kg/m². The most common blood type was group A (43.2%). The histopathological characteristics of the patients are shown in Table 1. The mean number of lymph nodes removed during surgery was 35 ± 14 (range, 11–81), and the mean tumor size was 5.4 ± 2.5 cm.

Table 1 . Histopathological Characteristics.

Characteristic	Value
Number of dissected lymph nodes	32 (11–81)
Size of tumor (cm)	5.5 (1–12)
Location of tumor
Cardia	24 (23.1)
Antrum	34 (32.7)
Corpus	33 (31.7)
Incisura angularis	3 (2.9)
Diffuse	7 (6.7)
Fundus	3 (2.9)
Extent of lymph node dissection
D1	11 (10.6)
D2	93 (89.4)
Resection type
Subtotal gastrectomy	35 (33.7)
Total gastrectomy	69 (66.3)
Surgical margins
Positive	12 (11.5)
Negative	92 (88.5)
T stage
I	10 (9.6)
II	14 (13.5)
III	47 (45.2)
IV	33 (31.7)
N stage
0	28 (26.9)
I	15 (14.4)
II	27 (26.0)
III	34 (32.7)
Grade
I	10 (9.6)
II	57 (54.8)
III	37 (35.6)
Lymphovascular invasion
Present	72 (69.2)
Absent	32 (30.8)
Perineural invasion
Present	66 (63.5)
Absent	38 (36.5)

Values are presented as mean (range) or number (%).

The clinical characteristics of the patients are presented in Table 2. Of the deaths, 19 were due to causes other than cancer (e.g., cerebrovascular event, myocardial infarction, and pneumonia). Patients’ age, sex distribution, weight, and BMI at the time of diagnosis did not differ significantly between the groups with low or high budding.

Table 2 . Clinical Characteristics of the Patients.

Characteristic	Value
Total follow-up (mo)	23 (1.5–96)
Number of received CT cycles	6 (2–8)
Follow up until recurrence (mo)	14 (4–49)
Adjuvant CT
Present	71 (68.3)
Absent	33 (31.7)
Adjuvant CT protocol
Folfox	33 (31.7)
FLOT	5 (4.8)
Capox	12 (11.5)
Cisplatin-Capecitabine	9 (8.7)
Other	12 (11.5)
Adjuvant RT
Present	39 (37.5)
Absent	65 (62.5)
Recurrence
Present	23 (22.1)
Absent	81 (77.9)
Location of recurrence
Lungs	3 (2.9)
Brain	1 (1.0)
Liver	6 (5.8)
Bones	1 (1.0)
Local	3 (2.9)
Ovaries	2 (1.9)
Peritoneum	4 (3.8)
Diffuse	3 (2.9)
Survival status
Death	34 (32.7)
Survival	70 (67.3)

Values are presented as mean (range) or number (%).

CT, chemotherapy; RT, radiotherapy; Folfox, folinic acid (leucovorin, FOL), fluorouracil (5-FU, F), oxaliplatin (Eloxatin, OX); FLOT, fluorouracil (5-FU, F), leucovorin (L), oxaliplatin (O), docetaxel (D); Capox, capecitabine and oxaliplatin.

The pathological parameters of the patients in the high-TB and low-TB groups are demonstrated in Table 3. As the tumor size increased, the proportion of patients with high TB was significantly higher (P = 0.018). As the T stage (P = 0.003) and N stage (P = 0.001) of the tumor increased, high TB was observed significantly more frequently. Grade III tumors were more common in the high-TB group (P = 0.03) (Table 3).

Table 3 . Pathological Parameters and Tumor Budding.

Characteristic	Low budding (n = 62)	High budding (n = 42)	P-value
Tumor size (cm)	4.9 (2.5–7.3)	6.1 (3.7–8.5)	0.018*
T stage
I	9 (14.5)	1 (2.4)	0.003***
II	11(17.7)	3 (7.1)
III	30 (48.4)	17 (40.5)
IV	12 (19.4)	21 (50.0)
N stage
I	25 (40.3)	3 (7.1)	0.001***
II	12 (19.4)	3 (7.1)
III	11 (17.7)	16 (38.1)
IV	14 (22.6)	20 (47.6)
Grade
I	9 (14.5)	1 (2.4)	0.030***
II	36 (58.1)	21 (50.0)
III	17 (27.4)	20 (47.6)
Lymphovascular invasion
Present	40 (64.5)	32 (76.2)	0.206***
Absent	22 (35.5)	10 (23.8)
Perineural invasion
Present	34 (54.8)	32 (76.2)	0.027***
Absent	28 (45.2)	10 (23.8)
Location of tumor
Cardia	14 (22.6)	10 (23.8)	0.988**
Antrum	19 (30.6)	15 (35.7)
Corpus	21 (33.9)	12 (28.6)
Incisura angularis	2 (3.2)	1(2.4)
Diffuse	4 (6.5)	3 (7.1)
Fundus	2 (3.2)	1 (2.4)

Values are presented as mean (range) or number (%).

*t test; **Fisher exact test; ***Chi-square test (P < 0.05).

In clinical practice, patients with positive surgical margins or with T3-4 tumors and lymph node involvement receive RT. Consistent with the fact that TB increased as the T and N stages increased, high TB was found to be significantly more common in patients who received adjuvant RT (P = 0.03).

In the OS analysis performed in 85 patients, excluding the patients who died due to non-disease-related causes during follow-up (19 patients), the median OS was 58 months (95% confidence interval [CI], 5.9–110.5 months) in the high-TB (Fig. 1), while the median OS could not be calculated in the low-TB group (Fig. 2). OS was significantly longer in the low-TB group than in the high-TB group (P = 0.038). The median disease-free survival (DFS) time predicted in the analysis of 85 patients was not reached in the group with low TB, while the median DFS was 49 months in the group with high TB (95% CI, 21.7–76.3 months). However, the difference did not reach statistical significance due to the small number of patients and the small number of events (P = 0.135).

In the univariate survival analysis, high TB, age at diagnosis over 60 years, low BMI, high N stage, the presence of LVI, and recurrence were found to have significant effects on OS. High TB was identified as an independent risk factor predicting mortality in the multivariate model, and age at diagnosis over 60 years and the presence of LVI had significant associations with OS and recurrence (Fig. 3).

Figure 3. Cumulative survival analysis between the high and low budding groups. CI, confidence interval.

The first step in the development of metastasis is the detachment of tumor cells from the primary lesion and invasion of stromal tissue. TB, which is thought to constitute the basis for invasion and metastasis, has been defined as the presence of single malignant cells on the invasive surface of the tumor or the presence of undifferentiated malignant cell clusters consisting of 2–4 cells, called tumor buds, in the tumor stroma.¹² Various methods of evaluating TB have been described. The most common method used in the past was scanning several areas and calculating the average number of buds.^13,14 In our study, to count tumor buds, we used an area of 0.785 mm² in the region where TB was most intense in hematoxylin and eosin-dyed preparations (i.e., the “hotspot”). TB has not yet become a standard prognostic marker in gastric cancer due to variations in definitions and evaluation methods.¹⁵

According to the UICC consensus report, TB is considered an independent negative prognostic factor in colorectal cancer. In addition, TB was reported to be associated with higher TNM stage and higher tumor grade, and it was recommended that TB should be routinely taken into consideration for treatment plans of colorectal cancer and should be included in current guidelines.⁹ TB has also shown associations with a poor prognosis in various epithelial cancer types such as breast cancer, laryngeal squamous cell carcinoma, lung adenocarcinoma, esophageal cancer, pancreatic adenocarcinoma, endometrial carcinoma, and gallbladder cancer.^5,16–21

In their systematic review, Guo et al²⁰ analyzed seven studies including 2,187 patients. According to this review, high TB is associated with unfavorable prognosis, advanced stage, a more aggressive course, more LVI, and more lymph node involvement.²² In our study, which supports these data, as the tumor size increased, the proportion of high TB was found to be significantly higher (P = 0.018). High TB was also significantly more common in tumors with a higher T stage (P = 0.003), N stage (P < 0.001), and grade (P = 0.03).

Ulase et al²¹ found that TB in intestinal-type gastric cancer was also correlated with histological grade, T and N stage, disease stage, LVI, and PNI.²³ In our study, a similar relationship was found between TB and T stage, N stage, grade, and PNI. LVI was seen at a higher rate in the high-TB group (76%) than in the low-TB group, but the difference was not statistically significant (P = 0.206).

Although the 5-year OS rate exceeds 95% in early-stage gastric cancers, some patients experience recurrence after curative treatment. Additional predictive and directive prognostic parameters such as TB are essential, especially in T1N0 and T2N0 tumors, for which the role of adjuvant CT is not clear.²⁴

LVI plays an important role among the known risk factors for lymph node metastasis in early-stage gastric cancers.^25,26 In their study on early-stage gastric cancers, Du et al²⁵ found that the association between LVI and TB was significantly higher in patients with lymph node metastasis than in those without lymph node metastasis. The authors reported that LVI and TB were the most important risk factors in submucosal lesions. With the results of more comprehensive studies in the future, additional surgical resection, including lymph node dissection, may be considered if these risk factors are detected after the endoscopic resection of early-stage cancers.²⁷ Similarly, Olsen et al²⁶ found statistically significant associations between TB and recurrence, differentiation, LVI, and the T and N stages of tumors.

Gabbert et al²⁷ concluded that TB was a significant negative prognostic factor for 5-year OS, especially in patients with intestinal-type adenocarcinoma. Likewise, in a study by Kemi et al,²⁸ the 5-year OS rates were found to be significantly lower in the high-TB group (23.0%) than in the low-TB group (41.0%). In our study, the relapse rate and time to recurrence did not differ significantly between the low-TB and high-TB groups (P = 0.732). However, the predicted OS was significantly shorter in the high-TB group than in the low-TB (P = 0.038). Similarly, the predicted DFS of the high-TB group was shorter than that of the low-TB group. However, this difference did not reach statistical significance due to the small number of patients and events (P = 0.135).

In their univariate analysis, Zhang et al²⁹ found that high TB was associated with poor prognosis (e.g., larger tumor size and lymph node involvement). In their multivariate analysis, they found that TB and positive nodal metastases were independent negative prognostic factors in gastric adenocarcinoma.³⁰ In contrast to this study, Ulase et al²¹ found no effect of TB on OS in multivariate analysis. In the univariate analysis of our study, a high TB score, being over 60 years old, low BMI, high N stage, presence of LVI, and recurrence had significant effects on OS.

TB has a similar mechanism to the epithelial-mesenchymal transition. Loss of expression of E-cadherin and β-catenin, which are known as adhesion molecules, has been observed in cells with TB.¹⁵ Therefore, agents targeting the epithelial-mesenchymal transition pathway in patients with high TB may have a place in their treatment in the future.

The limitations of the study

The limitations of our study are the relatively small number of patients and the single-center design. Due to the short follow-up time and few events, a longer follow-up period would be required to obtain more robust findings.

Conclusion

In our study, a high TB rate was found to be associated with tumor size, T stage, N stage, and high tumor grade, supporting data available in the literature. In addition, PNI was found more frequently in the high-TB group. The predicted OS was also significantly shorter in the high-TB group than in the low-TB group. These findings suggest that TB may be a poor prognostic marker in gastric cancer, as well as in colorectal cancer. If more comprehensive studies are conducted and a standard method is established for the detection of TB, we believe that TB will be used to guide treatment decisions and predict patients’ prognoses.

None.

No potential conflict of interest relevant to this article was reported.