Long-term outcome of esophageal mucosal squamous cell carcinoma without lymphovascular involvement after endoscopic resection


  • Informed consent was obtained from all patients before endoscopic resection was performed.



Esophageal cancer is an aggressive cancer with a reported 3-year survival of 20%. However, early-stage esophageal cancer can be cured by endoscopic resection (ER). The long-term survival of esophageal mucosal squamous cell carcinoma after ER was investigated by calculating the standard mortality rate (SMR).


From January 1995 to December 2004, 110 patients with 138 esophageal mucosal squamous cell carcinomas without lymphovascular involvement were treated by ER. Long-term survival after ER was compared with that in the general population by calculating SMR. Subgroup analysis of patients without second primary cancer diagnosed within 1 year before ER (subgroup A) was also performed.


A total of 108 patients (98.2%) were followed-up completely, with a mean observation period of 4.7 (0.4–11.3) years. The cumulative 5-year survival rate of all patients and subgroup A was 79.5% and 86.6%, respectively. Overall mortality (SMR, 1.68; 95% confidence interval [CI], 1.05–2.55) and mortality from malignant tumor (SMR, 3.14; 95% CI, 1.79–5.09) was significantly higher than that in the general population. SMR of esophageal cancer was high, although it was not significantly different from that in the general population (SMR, 4.82; 95% CI, 0.06–26.81). In subgroup A overall mortality (SMR, 0.86; 95% CI, 0.41–1.57) was similar to that in the general population.


High overall mortality in patients with esophageal mucosal cancer after ER was mainly due to elevated mortality from second primary cancer. Favorable mortality in subgroup A indicates the efficiency of ER as a curative treatment for esophageal mucosal cancer. Cancer 2008. © 2008 American Cancer Society.

Esophageal cancer is the fifth leading cause of death from cancer.1 Squamous cell carcinoma is the prevailing tumor type in Asia, whereas this has now been overtaken by esophageal adenocarcinoma in Europe and North America.2 Localized esophageal cancer, regardless of its histologic type, is commonly treated with surgical resection or chemoradiation, or their combination.2–4 Even after aggressive therapy the 3-year survival rate of patients with local- regional esophageal cancer is 20%.5, 6 Moreover, surgical resection and chemoradiation are associated with high treatment-related morbidity and mortality.2, 4 Thus, advanced-stage esophageal cancer is a treatable but rarely curable disease.

However, this cancer, when diagnosed at an early stage, can be cured by endoscopic resection (ER).2, 7–9 Since the 1990s, many early esophageal cancers have been treated by ER in Japan. More recently, in Western countries ER of gastrointestinal lesions, especially Barrett esophagus, with high-grade dysplasia or cancer has been increasingly used.10, 11 The techniques of ER are designed to completely remove the diseased mucosa by resection through the middle or deeper part of the submucosa. Then the resected specimen is used for pathologic staging to stratify the patient's risk of developing metastasis. Although this method is widely used in Japan, there are relatively few reports on long-term survival after ER.7–9 For further development of this method, investigation of long-term survival and confirming its adequacy is important.

In the assessment of survival, comparison of mortality rates between the study subjects and the general population are sometimes used to assess the excess mortality risk in the study subjects. The ratio of actual observed mortality in the study group and the expected mortality calculated from the general population is known as the standardized mortality ratio (SMR). In this study we assessed the efficacy of ER for mucosal esophageal cancer by obtaining detailed follow-up data from patients after ER and calculating SMR.



From January 1995 to December 2004, all patients with esophageal cancer were registered, along with details including therapeutic modalities, with the hospital cancer registry in Osaka Medical Center for Cancer and Cardiovascular Diseases, Japan. Data about cancer spread and treatment strategy were obtained from this database.

Detailed clinicopathologic data of patients who underwent ER were collected from medical records and included the location, size, and type of tumor and the histologic findings of the resected specimens. From these datasets we identified 110 patients with 138 esophageal cancers who met our inclusion criteria: 1) histologically proven squamous cell carcinoma; 2) mucosal cancer, including in situ lesions, without lymphovascular involvement; 3) no lymph node or distant metastasis; and 4) no previous history of radiotherapy, surgery, or ER for esophageal cancer.


Before ER, endoscopic examination with chromoendoscopy after direct instillation of 1.5% iodine solution was performed in all patients to evaluate the depth and lateral spread of cancer. Computed tomography of the chest and abdomen was also carried out to identify lymph node metastasis. Lymph nodes ≥10 mm in diameter were defined as metastatic.

ER was indicated for esophageal cancer without ulcer or obvious protrusion that suggested invasion to a deeper part of the submucosal layer. ER was limited to lesions ≤50 mm in diameter. Lesions >50 mm were resected only upon patient request. Circumferential resections were avoided because of the risk of stricture formation. ER was performed under intravenous sedation with midazolam and pentazocine.

ER was performed by the endoscopic mucosal resection (EMR) method or the endoscopic submucosal dissection method. EMR was mainly performed using the standard strip biopsy method,12 with a 2-channel endoscope (GIF-2T200 or 2T-240; Olympus, Tokyo, Japan). Saline was injected into the submucosa beneath the lesion with an injection needle (Varixor 23G; Top, Tokyo, Japan). A high-frequency electrosurgical snare and grasping forceps were passed through the channel. An area near the lesion was grasped with the forceps to capture the lesion into the snare and the lesion was strangulated. Forced coagulation current was then applied to resect the lesion.

Endoscopic submucosal dissection was mainly performed using a Hook knife as described by Oyama et al.13 A forward-viewing endoscope (GIF-Q240Z; Olympus) with an attachment (D-201-11,804; Olympus) on its tip was introduced. Marking dots were made 2–3 mm outside the margins of the lesion with a Hook knife. A 200 mL 10% glycerin solution with 1 mL 0.1% adrenaline was injected into the submucosa and the mucosa was incised outside the marking dots with the Hook knife using the Endocut mode of an electrosurgical generator (ICC 200; Erbe, Tubingen, Germany). The diseased mucosa was then separated from the muscularis propria by the injection of glycerin solution into the submucosa. The submucosal connective tissue just beneath the lesion was dissected from the muscle layer using the Hook knife in the forced coagulation mode. Injection was repeated as needed, and further resection was carried out to ensure total removal of the lesion.

Complete local remission was defined as the disappearance of iodine-unstained lesions on endoscopic examination and no evidence of residual cancer on biopsy specimens at first follow-up. Informed consent was obtained from all patients before ER was performed.

Histologic Evaluation and Further Treatment

All specimens were cut into 2-mm slices, embedded in paraffin, and subjected to hematoxylin and eosin staining. Detailed histologic assessment was performed according to the Japanese Classification of Esophageal Carcinoma.14 Mucosal cancer without lymphovascular involvement was followed up without any additional treatment immediately after ER. In patients with lymphovascular involvement and/or submucosal invasion the risk of metastasis was estimated based on the histologic findings. Further treatment was decided on depending on the risk of metastasis and the patient's condition.


After ER all patients were included in a follow-up program. Upper gastrointestinal endoscopy with iodine staining was conducted at 3 and 9 months and annually thereafter. Computed tomography of the neck, chest, and abdomen was performed at least once a year to detect lymph node or distant metastases.

Follow-up patient data were obtained by referral to their medical records. Causes of death were confirmed by death certificates documented by the patients' physicians. For the outcome of patients who moved away from our hospital we attempted to obtain details by questionnaires or telephone conversations with their families and referring physicians. The International Classification of Diseases, Ninth Revision, Clinical Modification was used to code and classify mortality data. The starting date of the follow-up was defined as the date of ER and the end of the follow-up was either the date of death or the end of December 2006, whichever occurred first. Patients with unknown vital status were regarded as censored at the time when they were last known to be alive.

Subgroup Analysis

This study cohort included many patients with multiple primary cancers. We defined synchronous or metachronous (including previous) cancers other than esophageal as second primary cancers. In patients with multiple primary cancers, second primary cancers largely influenced the clinical course after ER. In a patient receiving ER, we usually obtained information about second primary cancer from medical records and interviews with the patient and a referring physician. To reduce the impact of second primary cancer on mortality after ER, we excluded patients with recently diagnosed second primary cancer (other than in situ carcinoma of stomach and colon). Ninety patients without second primary cancer diagnosed within 1 year before ER were defined as subgroup A.

Statistical Analysis

Cumulative survival curves were determined with the Kaplan-Meier method. The significance of differences in survival were assessed by the log-rank test. Cancer patients usually have excess risk in SMR from unity because a certain proportion of cancer patients will die from the underlying cancer, in addition to the expected mortality calculated from the general population. SMR of patients with esophageal cancer is thought to be high because most patients die from esophageal or second primary cancer. Because ER is performed as a curative operation for early esophageal cancer the procedure can be regarded as successful when the SMR is near unity. Therefore, we calculated SMR for the assessment of ER as a treatment for esophageal cancer.

For the calculation of SMR the observed number of deaths was compared with the expected number, which was calculated by applying sex, 5-year age, 5-year calendar time, and cause-specific mortality rates for the general population in the Osaka Prefecture, as prepared by the Statistics and Information Department, Japan Ministry of Health and Welfare.15 The SMR was expressed by dividing the observed number of deaths in the study group by the expected number of deaths calculated from the general population. The standard error and 95% confidence interval (CI) of SMR were estimated by assuming a Poisson distribution and differences in mortality between the study cohort and the general population were considered significant if the CI did not include unity. Data analysis was performed with SAS/PC statistical package (SAS Institute, Cary, NC).


Patient Characteristics at Entry

A flow chart of patient recruitment is shown in Figure 1. From January 1995 to December 2004, 1049 patients with esophageal cancer visited our hospital. Among these, 159 patients received ER as their first treatment. In all, 110 of 159 patients who received ER fulfilled our inclusion criteria. The characteristics of the study population and lesions are shown in Table 1. The breakdown of second primary cancer was as follows: 14 hypopharyngeal, 3 mesopharyngeal, 11 laryngeal, 5 tongue, 13 gastric, 4 colon, and 3 lung cancers.

Figure 1.

Flow chart of patient recruitment.

Table 1. Characteristics of Patients and Lesions at Entry
Characteristics of patientsTotal n = 110Subgroup A n = 90
  1. ER indicates endoscopic resection; EMR, endoscopic mucosal resection; ESD, endoscopic submucosal dissection.

Mean age, y, (range)65 (44–84)65 (44–84)
Patients with a history of second primary cancer4929
Patients with second primary cancer diagnosed within 1 year before ER200
Characteristics of lesions
 No. of lesions138111
 Mean tumor size, mm (range)13 (3–70)13 (3–60)
 In situ cancer/invasive cancer82/5666/45
 Poorly differentiated squamous cell carcinoma11
Method of ER
 En bloc/piecemeal89/4972/39


A total of 138 ERs were carried out in 110 patients. Detailed data of ER are presented in Table 1. ER was performed in accordance with our inclusion criteria, except for 4 large lesions (>50 mm diameter) that were treated upon the patients' request. Complete local remission was obtained in all lesions after 1 session of ER. The only major complication observed during ER was 1 case of perforation.

Follow-up Data

Vital status was unknown in 2 patients who were lost to follow-up at 3 and 16 months after ER. A total of 108 patients (98.2%) were completely followed up, with a mean observation period of 4.7 (range, 0.4–11.3) years. Detailed data of local recurrence, metachronous esophageal cancer, and second primary cancer are shown in Table 2.

Table 2. Number of Patients With Local Recurrence and Metachronous Cancer After ER
 Total n = 110Subgroup A n = 90
  1. ER indicates endoscopic resection.

Complete follow-up10888
No. of patients with local recurrence87
No. of patients with metachronous esophageal cancer after ER1211
No. of patients with metachronous second primary cancer after ER1514

Local recurrence was observed in 8 lesions in 8 patients. Seven lesions were successfully treated by additional ER or endoscopic ablation, whereas 1 lesion could not be cured by 2 additional ERs and was successfully treated with chemoradiation.

Eighteen metachronous esophageal cancers in 12 patients were found after ER. Sixteen lesions in 10 patients were treated by ER or endoscopic ablation therapy, and 1 patient was treated by surgical resection. In the other patient, submucosal invasive cancer developed in the cervical esophagus with cervical lymph node metastasis 14 months after ER. The lesions could not be cured by chemoradiation and he died of esophageal cancer 40 months after ER.

Fifteen second primary cancers in 15 patients were found after ER. Ten lesions received curative treatment: surgery for 4 oropharyngeal cancers and 1 lung cancer; radiotherapy for 3 oropharyngeal cancers; and ER for 2 stomach cancers. Two hepatocellular carcinomas developed in 2 patients with cirrhosis and were treated by radiofrequency ablation therapy and transcatheter arterial embolization therapy, respectively. Three patients died of gallbladder cancer, cholangiocarcinoma, and hepatocellular carcinoma, respectively.

We identified 22 deaths during follow-up (Table 3). Death from malignant tumor was identified in 16 (73%) of these. The origin of the malignant tumors was 6 oropharyngeal, 3 lung, 3 liver, 1 laryngeal, 1 esophageal, 1 rectal, and 1 gallbladder. Other causes of death were cerebrovascular disease in 2 patients, heart disease in 2, and cirrhosis in 2 patients (Table 3). In the analysis of mortality data, cerebrovascular disease and heart disease were classified into circulatory disease (International Classification of Diseases). There was no significant difference in survival between lesions of different sizes (≤20 mm vs >20 mm) or depth of invasion (in situ vs invasive lesion).

Table 3. SMR in All Patients and Subgroup A
 All patients, n = 110Subgroup A, n = 90
Mean follow-up, y [SD]: 4.7 [2.7]Mean follow-up, y [SD]: 5.1 [2.6]
ObservedExpectedSMR (95% CI)ObservedExpectedSMR (95% CI)
  1. SMR indicates standardized mortality ratio; CI, confidence interval.

  2. Difference from the expected number of deaths was considered significant if 95% CI of SMR did not include unity.

  3. Circular disease includes cerebrovascular vascular diseases and cardiovascular diseases.

All deaths2213.071.68 (1.05–2.55)1011.690.86 (0.41–1.57)
Malignant tumor165.103.14 (1.79–5.09)54.621.08 (0.35–2.52)
Oropharynx60.0968.40 (24.98–148.88)00.080.00 (0.00–46.88)
Liver30.903.34 (0.67–9.76)30.813.72 (0.75–10.88)
Esophagus10.214.82 (0.06–26.81)10.185.41 (0.07–30.10)
Lung31.172.56 (0.52–7.49)00.070.00 (0.00–3.43)
Liver cirrhosis20.219.73 (1.09–35.12)20.1811.01 (1.24–39.73)
Circulatory diseases43.381.18 (0.32–3.03)33.020.99 (0.20–0.90)

Survival Curves

Kaplan-Meier curves of all patients and subgroup A were constructed (Fig. 2). The cumulative 5-year survival rate (mean observation period) of all patients and subgroup A was 79.5% (4.7 years) and 86.6% (5.1 years), respectively.

Figure 2.

Overall survival of all patients and subgroup A.


Differences in mortality in patients from that of the general population were further assessed by calculating SMR. Overall mortality (SMR, 1.68; 95% CI, 1.05–2.55) for patients treated by ER was significantly higher than that of the general population (Table 3). Mortality from malignant tumors (SMR, 3.14; 95% CI, 1.79–5.09), especially from oropharyngeal cancer (SMR, 68.40; 95% CI, 24.98–148.88), was much higher than that in the general population. SMR of esophageal cancer was high, although it was not significantly different from that in the general population (SMR, 4.82; 95% CI, 0.06–26.81).

In subgroup A overall mortality (SMR, 0.86; 95% CI, 0.41–1.57) and mortality from malignant tumors (SMR, 1.08; 95% CI, 0.35–2.52) were similar to those of the general population (Table 3).


The present study showed that high overall mortality in patients with esophageal cancer after ER was mainly due to elevated mortality from second primary cancer. In subgroup A patients had a similar survival rate to that of the general population, which indicates complete control of esophageal cancer in almost all patients, and the invasiveness of ER did not increase the risk of serious side effects that might trigger death.

In this study, 49 of 110 patients had a present or past history of second primary cancer. Among these, 22 patients had oropharyngeal cancer. Multiple developments of squamous cell carcinoma in the esophagus and oropharynx,16, 17 frequently seen in patients with esophageal cancer, are explained by the field cancerization theory.18 The prognosis of double primary cancers is usually influenced by oropharyngeal as well as esophageal cancer.19, 20

Other than oropharyngeal cancer, mortality from cirrhosis (SMR, 9.73; 95% CI, 1.09–35.12) was significantly higher than that in the general population, whereas that from liver cancer (SMR, 3.34; 95% CI, 0.67–9.76) and lung cancer (SMR, 2.56; 95% CI, 0.52–7.49) was high, without being statistically significant. In the present study 5 patients died of liver-related diseases (3 with hepatic carcinoma and 2 with cirrhosis). Four of these 5 patients had hepatitis C virus (HCV) and cirrhosis at the time of ER and 3 were heavy alcohol drinkers (consumption >50 g ethanol/day). It is well known that the risk for esophageal cancer increases in proportion to the amount of alcohol consumed. A mutant allele in the ALDH gene, which is prevalent in Asians, enhances the risk of cancer.21 Similarly, alcohol consumption is an important risk factor in the progression of HCV-related liver diseases,22 and is regarded as a major risk factor for the rise in liver cancer mortality.23

Furthermore, patients with cirrhosis usually receive annual endoscopic examination for evaluation of esophageal varices. An increased opportunity to receive endoscopic examination enhances the discovery rate of esophageal cancer. Higher mortality from liver diseases after ER of esophageal cancer must be due to the close relation between the 2 diseases, rather than any effect of ER. A similar explanation is possible for the relatively higher mortality from lung cancer, because smoking is a major risk factor for both esophageal and lung cancer,24 and patients with esophageal cancer have more opportunity to receive chest computed tomography.

Sato et al.19 reported that, in patients with double primary cancers, second primary cancer is the major cause of death, as long as esophageal cancer does not have lymph node involvement. In agreement with this study, our cohort had a higher mortality rate from malignant tumors (SMR, 3.14, 95% CI, 1.79–5.09), especially from oropharyngeal cancer (SMR, 68.40; 95% CI, 24.98–148.88), and this resulted in a higher overall mortality rate. This indicates the importance of treatment of second primary cancer in patients receiving ER.

The SMR of esophageal cancer was high, although it was not significantly different (SMR, 4.82; 95% CI, 0.06–26.81) from that in the general population. The difference may have been significant if we had studied a larger number of patients. However, in this study we wanted to focus on the finding that esophageal cancer mortality, observed in only 1 case, did not negatively impact the favorable overall survival after ER.

The risk of lymph node metastasis in patients with esophageal mucosal cancer is reported to be 0% to 11%.25–27 The higher potential for insidious metastasis in patients treated by ER results in higher recurrence and mortality. However, mucosal cancer patients without lymphovascular involvement had minimal risk of developing lymph node metastasis.26, 27 In the present study, patients with esophageal mucosal cancer and no lymphovascular involvement received no treatment after ER, whereas 3 patients with lymphovascular involvement received further treatment. Our excellent cause-specific survival after ER was achieved because patients with minimal risk for metastatic spread were accurately selected using the pathological specimens from ER.

As for long-term survival after ER, Takeshita et al.7 have shown that there was no cause-specific death after 3 years follow-up in 43 patients with esophageal mucosal cancer treated with endoscopy. Kodama and Kakegawa8 reported a favorable 5-year survival (>90%) after ER of esophageal mucosal cancer. Their study, although it included much information about ER, was a result of responses from 143 institutions to questionnaires on superficial cancer of the esophagus in Japan.

Makuuchi9 reviewed the results of ER of 378 lesions in 249 patients. Lymph node metastasis and subsequent death from cancer were observed in patients who had submucosal involvement. The 5-year disease-specific survival rate of EMR was 97.9% for all patients. His report included many patients and resulted in an excellent outcome. However, detailed follow-up and cause of death were not mentioned. Sufficient follow-up is essential for survival analysis, because lower follow-up rates result in overestimation of survival.

In the present study detailed analysis of second primary cancer and follow-up data from more than 100 patients were collected with a high follow-up rate (98.2%). By analyzing SMR, overall mortality after ER was higher than that in the general population, mainly due to second primary cancer. In the subgroup analysis (patients without second primary cancer diagnosed within 1 year before ER), overall mortality after ER was similar to that in the general population, with a mean follow-up period of 5.1 years, which indicates the efficiency of ER as a curative treatment for esophageal mucosal cancer without lymphovascular involvement.