Annals of Surgical Oncology

, Volume 21, Issue 4, pp 1123–1129

Proposal of the Surgical Options for Primary Tumor Control During Sentinel Node Navigation Surgery Based on the Discrepancy Between Preoperative and Postoperative Early Gastric Cancer Diagnoses

Authors

  • Ji Yeon Park
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Bang Wool Eom
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Hong Man Yoon
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Soo Jin Kim
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Soo-Jeong Cho
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Jong Yeul Lee
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Chan Gyoo Kim
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Myeong-Cherl Kook
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Il Ju Choi
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
  • Byung Ho Nam
    • Cancer Biostatistics Branch, Research Institute for National Cancer Control & EvaluationNational Cancer Center
  • Young-Woo Kim
    • Gastric Cancer Branch, Research Institute & HospitalNational Cancer Center
Gastrointestinal Oncology

DOI: 10.1245/s10434-013-3427-2

Cite this article as:
Park, J.Y., Ryu, K.W., Eom, B.W. et al. Ann Surg Oncol (2014) 21: 1123. doi:10.1245/s10434-013-3427-2
  • 286 Views

Abstract

Background

There is no consensus on the optimal method of primary tumor control, determined by preoperative clinical factors, during sentinel node (SN) navigation surgery for early gastric cancer (EGC). In this study, we investigated the accuracy of clinical diagnosis based on preoperative examination in patients with EGC and proposed surgical options for primary tumor control during SN navigation surgery.

Methods

We analyzed 815 patients with clinical stage IA gastric cancer who underwent gastrectomy at the National Cancer Center in Korea between March 2001 and February 2011. The clinical stage was determined by endoscopy, endoscopic ultrasonography, and abdominal computed tomography.

Results

The preoperative assessment of tumor depth and tumor size was accurate in 57.5 and 70.8 % of patients, respectively. Tumor depth and size were underestimated in 8 and 25.3 % of patients. The overall accuracy of histologic diagnosis by endoscopic biopsy was 87.2 %. Of those tumors diagnosed preoperatively as differentiated, 20.5 % revealed mixed histology of undifferentiated type.

Conclusions

The recommendation for SN biopsy may be limited to tumors sized 3 cm or smaller to avoid positive lateral margins and to minimize the risk of skip metastases. Endoscopic resection may safely be applied to small mucosal cancers, but other surgical options should be employed for undifferentiated large mucosal lesions, given their tendency for diffuse invasion. Full-thickness resection is preferable for submucosal cancers, to secure clear vertical margins.

Recently, the Japanese Gastric Cancer Association released guidelines for treating gastric cancer, recommending surgery customized to the disease stage, with reduced lymphadenectomy in appropriate patients.1 Several alternative measures for patients with EGC, such as endoscopic resection, function-preserving gastrectomy, and laparoscopic gastrectomy, have been introduced in an attempt to preserve function and to maintain quality of life while treating the malignancy. Among these alternative options for EGC treatment, sentinel node (SN) navigation surgery has been consistently investigated for more than a decade; it has not yet been put into clinical practice because of inconclusive results and the unsatisfactory sensitivity of SN biopsy. Results from several recent studies, however, show that the procedure is feasible, with acceptable short-term outcomes.2,3 Based on these results, the Korean Surgical Society is preparing a multicenter, phase III trial comparing conventional laparoscopic gastrectomy to laparoscopic SN biopsy with organ-preserving surgery in patients with clinical stage IA gastric cancer; the quality control study is currently underway (NCT01544413).4

The optimal EGC treatment strategy during SN navigation surgery largely depends on the detection rate of SN and its sensitivity; selecting a strategy also depends on the available options for primary tumor control, which in turn depend on the accuracy of the preoperative diagnosis. Adequate preoperative evaluation and accurate assessment of the primary tumor and of the lymph node (LN) status are therefore crucial in making the best therapeutic decision for each patient. Underestimation of the tumor or LN status during the diagnostic workup could lead to insufficient treatment and subsequent disease recurrence. Given the importance of clinical staging in determining treatment plans, it is paramount to critically review current clinical staging modalities.

This study was conducted to analyze the accuracy of the clinical diagnosis made using preoperative information and to propose appropriate surgical options for primary tumor control during SN navigation surgery for the upcoming phase III trial.

MATERIALS AND METHODS

Patients

A retrospective review of a prospectively established database at the National Cancer Center in Korea identified all patients with pathologically proven gastric adenocarcinoma who underwent surgery between March 2001 and February 2011. We included in the study those who were clinically diagnosed with stage IA disease, according to the American Joint Committee on Cancer (AJCC) seventh edition criteria, as determined by three diagnostic modalities: esophagogastroduodenoscopy (EGD), endoscopic ultrasonography (EUS), and abdominal computed tomography (CT). Patients who had not undergone all three preoperative diagnostic examinations were excluded. Patients found to have multiple gastric cancer lesions during the preoperative workup also were excluded. Approval for this review of hospital records was obtained from the institutional review board (NCC NCS 12-580); the need for patient informed consent was waived.

Preoperative Diagnosis and Staging

Preoperative diagnoses were based on the findings of EGD, EUS, and abdominal CT. The depth of tumor invasion was determined from the results of all three modalities, and the longest tumor diameter measured on EGD was used to define the tumor size. All EGD procedures were performed by experienced endoscopic specialists. Findings from EUS combined with those of abdominal CT were used to assess the presence of LN metastasis. The CT results were reviewed by well-trained gastrointestinal radiology staff. When the preoperative diagnosis differed between the three diagnostic modalities, we selected the worst T and N stages. A dichotomous classification was used for histologic grading of lesions: the differentiated type comprised well- or moderately differentiated tubular adenocarcinoma or papillary adenocarcinoma; the undifferentiated type comprised poorly differentiated tubular adenocarcinoma, signet ring cell carcinoma, or mucinous adenocarcinoma. The tumor was designated as undifferentiated when endoscopic biopsy samples revealed mixed components of differentiated and undifferentiated histology types.

Pathologic Staging

The postoperative pathologic stage was determined using the criteria set forth in the seventh edition of the AJCC classification system.5 An expert on gastric lesions performed the pathologic examinations. The depth of submucosal invasion in surgically resected specimens was classified as sm1 (upper third), sm2 (middle third), or sm3 (lower third).

Definition of Underestimation

The preoperative staging, determined by the results of the preoperative evaluations, was compared with the postoperative pathologic staging. Underestimation of the depth of invasion was determined when the pathologic depth of the primary tumor was deeper than that predicted by the preoperative evaluation. The tumor size was considered accurate when the size difference between the endoscopic measurement and the pathologic result was 1 cm or less; the tumor size was considered underestimated if the pathologic tumor size was more than 1 cm larger than the endoscopic measurement. Histologic underestimation was determined when a tumor, defined preoperatively as differentiated, was proven to be undifferentiated type on final pathology.

RESULTS

Patient Characteristics by Preoperative Evaluation

Of 1,219 patients diagnosed with clinical stage IA gastric cancer, 67 with multiple lesions were excluded from analysis. Another 337 patients were missing one or more of the three diagnostic modalities and also were excluded, leaving 815 patients eligible for the study. This population included 32 patients who had undergone previous endoscopic tumor resection and were subsequently referred for surgery due to a risk of LN metastasis or residual tumor. The patients’ demographic and clinical characteristics are shown in Table 1. All characteristics were assessed from the results of preoperative evaluation, including the pathologic reports from endoscopic biopsy.
Table 1

Clinical characteristics of patients preoperatively diagnosed with T1N0M0 gastric cancer (total n = 815)

Characteristics

No. (%)

Age (mean ± SD, year)

57.6 ± 12.1

Sex

 Male

488 (59.9)

 Female

327 (40.1)

Depth

 

 Mucosa

260 (31.9)

 Submucosa

555 (69.1)

Tumor size on EGD (cm)

2.34 ± 1.24

 ≤1.0

108 (13.3)

 1.1–2.0

356 (43.7)

 2.1–3.0

201 (24.7)

 3.1–4.0

93 (11.4)

 >4.0

57 (7)

EGC gross typea

 Elevated (I, IIa)

211 (25.9)

 Flat + depressed (IIb, IIc, III)

604 (74.1)

Ulcer findings

 Absent

681 (83.6)

 Present

134 (16.4)

Longitudinal location of tumor

 Upper third

63 (7.7)

 Middle third

329 (40.4)

 Lower third

423 (51.9)

Circular location of tumor

 Lesser curvature

312 (38.3)

 Greater curvature

156 (19.1)

 Anterior wall

138 (16.9)

 Posterior wall

177 (21.7)

Histologyb

 Differentiated

391 (48)

 Undifferentiated

424 (52)

Total

815 (100)

Results are from preoperative evaluation, including pathologic reports of endoscopic biopsy

Values are presented as mean ± standard deviation or number of patient (%)

EGD esophagogastroduodenoscopy, EGC early gastric cancer

aElevated, I, IIa; flat+depressed, IIb, IIc, III

bDichotomous classification; differentiated type includes well or moderately differentiated adenocarcinoma or papillary adenocarcinoma; undifferentiated type includes poorly differentiated adenocarcinoma, signet-ring cell carcinoma, or mucinous adenocarcinoma; mixed type histology on endoscopic biopsy was classified as undifferentiated type

Diagnostic Accuracy

Table 2 shows the correlation between the preoperative evaluation results and the postoperative pathologic tumor depth. The preoperative assessment of tumor depth by the combination of EGD, EUS, and CT was accurate in 57.5 % of patients. The accuracy was lower in submucosal cancers (49.2 %) than in mucosal cancers (75.4 %). A large proportion (281/555; 50.6 %) of preoperatively diagnosed submucosal tumors turned out to be mucosal cancers on pathologic examination. Tumor depth was underestimated preoperatively in 8 % of patients (65/815) by a combination of all three examinations.
Table 2

Accuracy of preoperative diagnosis of wall invasion in EGC patients

N = 815

pT1

pT2

Total

Accuracy (%)

pM

pSM1

pSM2

pSM3

pPM

Differentiated

       

 cT1

       

  cM

61

10

14

6

0

91

67

  cSM

150

43

43

64

0

300

50

 

54

Undifferentiated

       

 cT1

       

  cM

135

11

8

15

0

169

79.9

  cSM

131

26

39

58

1

255

48.2

 

60.8

Total

       

 cT1

       

  cM

196

21

22

21

0

260

75.4

  cSM

281

69

82

122

1

555

49.2

Overall accuracy of EGD, EUS and CT for the depth of invasion (%)

57.5

Overestimation (%)

34.5

Underestimation (%)

8

M mucosa, SM submucosa, MP proper muscle, cT preoperative diagnosis of wall invasion, pT pathologic diagnosis of wall invasion

The correlation between preoperative and postoperative tumor size is shown in Table 3. The overall accuracy of endoscopy in predicting tumor size was 70.8 %. The proportion of underestimation (25.3 %) was much higher than the proportion of overestimation (3.9 %).
Table 3

Accuracy of preoperative diagnosis of size in EGC patients (cm)

Clinical size

Size difference

≤1.0

>1.0

Total

Accuracya (%)

Differentiated type

    

 ≤1

28

9

37

75.7

 1.1–2.0

133

35

168

79.2

 2.1–3.0

80

27

107

74.8

 3.1–4.0

36

16

52

69.2

 >4

15

12

27

55.6

Total

292

99

391

74.7

Undifferentiated type

 ≤1

47

24

71

66.2

 1.1–2.0

129

59

188

68.6

 2.1–3.0

67

27

94

71.3

 3.1–4.0

25

16

41

61.0

 >4

17

13

30

56.7

 Total

285

139

424

67.2

Overall accuracy of EGD for tumor size

577 (70.8 %)

Overestimationb

32 (3.9 %)

Underestimationc

206 (25.3 %)

aProportion of patients with size discrepancy less than 1.0 cm between clinical and pathologic findings

bEndoscopic tumor size–pathologic tumor size >1.0 cm

cPathologic tumor size–endoscopic tumor size >1.0 cm

The histologic discrepancy between the endoscopic biopsy specimens and the resection specimens is shown in Table 4. Mixed histology of undifferentiated type was found in 80 of 391 (20.5 %) tumors that were classified as differentiated by preoperative endoscopic biopsy. Preoperative underestimation of histology occurred in 9.8 % of patients.
Table 4

Accuracy of pathologic examination results of endoscopic biopsy specimens

Pathologic result of endoscopic biopsy

Final pathologic report after surgery

Total

Accuracy (%)

Differentiated

Undifferentiateda

Differentiated

311

80

391

79.5

Undifferentiateda

24

400

424

94.3

Overall accuracy (%)

87.2

Overestimationb (%)

2.9

Underestimationc (%)

9.8

aMixture of differentiated and undifferentiated histology was designated to undifferentiated type

bOverestimation was calculated by dividing the number of patients with the differentiated type of tumor in the final pathology by the number of patients who had undifferentiated cancer from endoscopic biopsy results

cUnderestimation was calculated by dividing the number of patients with the undifferentiated type of tumor in the final pathology by the number of patients who had differentiated cancer from endoscopic biopsy results

Incidence of LN Metastasis Based on Preoperative Findings

LN metastases were found in 63 patients (7.7 %) who underwent curative surgery, making the negative predictive value of the preoperative imaging studies 92.3 % (752/815). Table 5 shows the correlation between the clinical findings and the frequency of LN metastasis. Seven of 91 (7.7 %) differentiated mucosal cancers were associated with LN metastasis regardless of the presence of ulceration. The size of these lesions on endoscopic measurement ranged from 2.0 to 5.0 cm. In differentiated submucosal cancer, the rate of LN metastasis was 4.7 % (14/300). Nine of 169 (5.3 %) undifferentiated mucosal cancers had nodal metastasis; six of these were smaller than 2.0 cm on endoscopic measurement.
Table 5

Incidence of lymph node metastasis according to preoperative clinical depth and size of cancer lesions in patients with cT1N0 gastric cancer

 

≤1.0 cm

1.1–2.0 cm

2.1–3.0 cm

3.1–4.0 cm

>4.0 cm

Total

LNM

N

%

LNM

N

%

LNM

N

%

LNM

N

%

LNM

N

%

LNM

N

%

Differentiated

                  

 Mucosa

0

11

0.0

3

41

7.3

2

26

7.7

1

5

20.0

1

8

12.5

7

91

7.7

 Submucosa

1

26

3.8

7

127

5.5

4

81

4.9

2

47

4.3

0

19

0.0

14

300

4.7

 Total

1

37

2.7

10

168

6.0

6

107

5.6

3

52

5.8

1

27

3.7

21

391

5.4

Undifferentiated

 Mucosa

0

43

0

6

75

8

0

27

0

3

15

20

0

9

0

9

169

5.3

 Submucosa

3

28

10.7

10

113

8.8

10

67

14.9

9

26

34.6

1

21

4.8

33

255

12.9

 Total

3

71

4.2

16

188

8.5

10

94

10.6

12

41

29.3

1

30

3.3

42

424

9.9

LNM lymph node metastasis

DISCUSSION

Endoscopy, EUS, and abdominal CT are widely used to evaluate preoperative tumor depth of invasion and the degree of nodal metastasis in EGC patients. Most previous reports have evaluated the accuracy of each diagnostic measure separately, but in practice the therapeutic method decision depends on the combined results of several examination modalities. In addition, few studies have evaluated the frequency of LN metastasis predicted by preoperative findings that actually determines the therapeutic plan. The present study investigated the overall accuracy of combined diagnostic measures in the assessment of primary tumors and LN status.

The reported overall accuracy of endoscopy in predicting tumor depth in patients with EGC is 63.0–83.4 %, and the accuracy of EUS is 64.8–90.1 %.69 Abdominal CT seems to have a limited role in differentiating tumor depth in EGC but may play an important role in excluding advanced cancer lesions (those deeper than T2).10 In the present study, the overall accuracy of the combination of three imaging modalities in predicting tumor depth was only 57.5 %; this is lower than figures given in previous reports. This low accuracy rate seems to be due to the fact that we used the worst result of the three diagnostic studies to determine the clinical stage, causing tumor overestimation in a considerable number of patients. True submucosal cancers could have been overestimated preoperatively and excluded from the analysis of the present study from the first, resulting in markedly lower underestimation rate in clinical submucosal cancers (1/555, 0.2 %) than in clinical mucosal cancers (64/260, 24.6 %). Nonetheless, 8 % of the total enrolled patients with clinical stage IA were preoperatively underestimated by tumor depth. Previous studies have consistently reported that undifferentiated tumors tend to be underestimated when it comes to tumor depth.8,11 However, our results showed that underestimation rate was higher in differentiated tumor (30/91, 33 %) than in undifferentiated tumors (34/169, 20.1 %). It is possibly because the present study did not include patients who had gastric cancers satisfying the absolute indications for endoscopic resection, and who therefore completed their treatment endoscopically; this population would have had no need to undergo further therapeutic intervention based on the accurate preoperative diagnosis. The exclusion of these patients may be responsible for the relatively low accuracy rate as well as the increased risk of underestimation in patients with differentiated mucosal cancer.

The preoperative measurement of tumor size, performed solely by endoscopy, was accurate in 70.8 % of patients when we allowed a difference of less than 1 cm between the endoscopic measurement and the pathologic result. We postulated that surgeons or endoscopists may easily compensate for the difference of less than 1 cm by securing a resection margin of more than 2 cm. Interestingly, more than a quarter of tumors were underestimated in terms of size, the opposite of what we saw with tumor depth. Margulies et al.12 report that endoscopists frequently underestimate the size of objects viewed through the scope. Kim et al.13 purport that endoscopic estimation of gross lesion margins may underestimate the true histopathologic margins in signet ring cell EGC. Because the treatment plan differs according to the tumor size in patients with EGC, patients with underestimated tumor size are at risk for insufficient treatment. Poor demarcation of undifferentiated tumors could lead to underestimation of the actual tumor size, and the method of size measurement may be another reason for the size underestimation observed in this study. An endoscopic ruler or a standard-size disc patch can be used for size determination, but in most cases the lesion is generally measured by the naked eye, using an opened grasp of endoscopic forceps for comparison. Thus, a standard reliable measurement method is required before the category of tumor size can be included in the treatment guideline criteria. Some investigators have suggested chromoendoscopy and fluorescein electronic endoscopy could improve the accuracy of tumor size determination.1416 Histologic discrepancy between the endoscopic forceps biopsy and the resected specimens causes some tumors to be falsely categorized preoperatively, which consequently results in under- or overestimating the risk of LN metastasis.1719 In the present study, one-fifth of patients diagnosed with differentiated cancer on endoscopic biopsy turned out to have undifferentiated histology on final pathology and most of them had a tumor with mixed type of histology. This seems to be the major cause of the histologic underestimation. The heterogeneity of gastric cancer might be another factor creating a discrepancy in pre- and postoperative histologic diagnosis.

The incidence of nodal metastasis based on clinical depth of invasion and tumor size was somewhat different in this study than that reported in previous studies based on final pathology.20 Even in differentiated intramucosal cancer, nodal metastasis occurred in 7 of 91 (7.7 %) patients; 3 had lesions smaller than 2.0 cm without ulcerative findings that met the absolute indication for endoscopic resection according to the Japanese guidelines.1 All three of these lesions were underestimated for tumor depth, tumor size, and histologic type; they were revealed to be undifferentiated submucosal cancers larger than 2 cm. This suggests that there is a considerable discrepancy between the preoperative diagnostic evaluation of the primary tumor and the postoperative pathologic result. The incidence of LN metastasis in patients with differentiated tumors was higher in mucosal cancer than in submucosal cancer; clinically underestimating a considerable number of submucosal tumors as mucosal lesions is a possible explanation for this result. Furthermore, the previously mentioned patient population, who finished treatment with endoscopic resection and therefore were not included in the present study, should have been less likely to demonstrate LN metastasis. This exclusion might account for the relatively high incidence of LN metastasis in patients with differentiated mucosal cancers.

When a patient is diagnosed with EGC, surgeons and gastroenterologists should always keep in mind that cancer lesions can be underestimated preoperatively. For this reason, expanding the indications for endoscopic resection based on pretreatment clinical findings seems to carry a not insignificant risk of leaving residual tumor and metastatic LNs, and thus should be carefully considered. Primary tumors should be resected with a margin adequate to obtain clear vertical and lateral margins, taking into consideration the possible discrepancy between the tumor depth and size based on the initial endoscopic evaluation and the final resected specimen. Additional procedures such as SN biopsy during surgery might serve as an alternative, compensating for the discrepancy in pre- and postoperative diagnosis and decreasing the incidence of missed metastatic nodes in patients with EGC; this also would serve to preserve patients’ quality of life.

Based on the results of this study, we suggest several aspects to be considered for determining the method of primary tumor resection during SN navigation surgery for the upcoming phase III clinical trial as follows:
  1. (1)

    Endoscopic resection of submucosal tumors may confer a considerable risk of vertical-margin involvement. Because clinically diagnosed mucosal cancers of both differentiated and undifferentiated histology rarely involve the proper muscle layer and have a very low possibility of SM3 layer invasion, endoscopic resection can be safely applied without concern for vertical margin involvement. Submucosal cancers, on the other hand, seem to carry a substantial risk of residual tumor when endoscopic resection is applied; full-thickness resection, such as wedge resection or segmental resection, would be more appropriate treatment.

     
  2. (2)

    The accuracy in tumor size measurement generally decreases as the tumor size increases, and undifferentiated EGCs often invade more diffusely than predicted on initial endoscopic examination. A sufficient resection margin should be ensured when planning gastrectomy with curative intent, with a gross resection margin of 2 cm in T1 tumors.1 As undifferentiated mucosal cancers larger than 2 cm seem to have a strong chance of size underestimation, surgical options other than endoscopic resection should be exercised to secure clear lateral margins. Furthermore, we suggest that SN biopsy should be limited to patients with tumors 3 cm or smaller, because the incidence of LN metastasis dramatically increases if the tumor is larger than 3 cm preoperatively, particularly in undifferentiated cancer. In addition, the risk of skip metastasis markedly increases as tumors become larger and it is not feasible to secure margins greater than 2 cm on every side in limited gastrectomy for larger tumors.21

     

In conclusion, the use of current diagnostic methods to make a definite treatment plan has many shortcomings; the accuracy of tumor status prediction leaves much to be desired. A decision on minimally invasive treatment for EGC should be carefully made after consideration of the possible preoperative underestimation of the primary tumor and of the LN status.

ACKNOWLEDGMENT

This work was supported by a Grant (NCC 1110550-2) from the National Cancer Center, Republic of Korea.

DISCLOSURES

Drs. Ji Yeon Park, Keun Won Ryu, Bang Wool Eom, Hong Man Yoon, Soo Jin Kim, Soo-Jeong Cho, Jong Yeul Lee, Chan Gyoo Kim, Myeong-Cherl Kook, Il Ju Choi, Byung Ho Nam, and Young-Woo, Kim have no conflicts of interest or financial ties to disclose.

Copyright information

© Society of Surgical Oncology 2013