Improvement in the Prognosis of Esophageal Cancer Patients Treated by Surgery Alone (the Sweet Approach) in a High Volume Center Between 1970 and 2008 - A Retrospective Analysis of 5349 Cases


 Backgrounds: Worldwide the largest number of esophageal cancer surgery is performed in China and left thoractomy with limited lymphadenectomy (the Sweet approach) used to be the main surgical procedure before 2010. However the prognosis associated with this procedure had improved or not remains unanswered, duo to frequent lost to follow-up of patients operated since 1990. Methods: Using data of 5349 patients treated by surgery alone (Sweet) in a high volume center in 26 years (for which lost to follow-up <10%), we analyzed 5-year survival rates according to 1970-1979, 1980-1989, 1990-1994, and 2008 in relation to clinicopathological characteristics and technique development.Results: Both the rate of postoperative complication and of 3-month mortality decreased significantly (11.4%, 9.7%, 8.9% , 5.2%, P<0.01; 6.0%, 5.4%, 3.3%, 3.1%, P<0.05), but the death rate of surgical complications remained unchanged (52.9%, 55.2%, 56.9% , 59.4%, P>0.38). The proportion of patients without lymph node metastasis increased significantly ( 30.9%, 31.0%, 35.5% to 52.9%, P<0.001). The 5-year survival rate increased significantly (32%, 35%, 38%, 43%, P<0.000) albeit a limited number of lymph node retrieved (median: 4.0, 4.0, 4.0, 5.0, P >0.18 ). Multivariate Cox regression analysis identified pTNM stage (P<0.001), postoperative complications (P<0.001), lymph node metastasis (P<0.001), surgery year (P P<0.01), and age (P<0.05) as independent signiﬁcant prognostic factors.Conclusion: The prognosis of esophageal carcinoma patients treated by surgery alone (Sweet approach) had improved from 1970-2008. Our data suggests that this improvement is due mainly to early diagnosis and reduced postoperative complication by experienced hands at high volume centers.


Introduction
According to GLOBACAN 2018, 572 034 incident cases and 508 585 deaths of esophageal cancer were estimated as having occurred in 2018 in China, constituting 53.7% and 55.7% of the world's total. [1] Surgery is the main treatment modality. This is especially true in northern central China where both the incidence and mortality rate of esophageal cancer have been estimated to be the world's highest. [2][3][4]In the early 1950s, a public cancer center with high volume surgical capacity was established in each of the four provinces of Henan, Hebei, Shanxi, and Shandong to provide surgical treatment. Regarding the surgical approach used, before esophagectomy through the right thoracic approach with three eld lymphadenectomy became popular during the late 2000s as supported by the Chinese Society of Esophageal Cancer, Chinese Anti-Cancer Association, [5] most esophagectomy performed at high volume centers had been through the left thoracic approach with incomplete two eld lymphadenectomy (Sweet).
[6] The department of Thoracic Surgery at the 4 th Hospital of Hebei Medical University (Hebei Provincial Cancer Center) alone had performed 24653 cases of esophagectomy/esophagogastrectomy by this procedure with curative intent from 1952 to 2009, exclaiming the largest number found with any single institute. [3,4] A series of large numbered studies have been published by these high volume centers reporting increasing rates of resectability and decreasing rates of postoperative complication associated with the Sweet approach. [2][3][4][5] However doubt remains as to whether the long term survival rate has improved. [5] A main reason for this is that starting from 1990, follow-up of patients after discharge has become more and more di culty as a result of growing doctor-patient distrust caused by the increased pro t-seeking behavior in the medical sector. The high rate of lost to follow-up has reduced the credibility of survival rates estimated. [7] To investigate if there is any improvement in long term survival associated with patients treated by surgery alone (Sweet), we actively re-contacted family of patients operated since 1990 and succeeded in obtaining complete survival data for the patients operated from 1990-1994 and in 2008. Then together with previous cohort data for patients resected from 1970-1989, we analyzed ve year survival rates according to the four time periods of 1970-1979, 1980-1989, 1990-1994, and 2008 in relation to the trend of clinicopathological factors and technique development.

Patients eligibility
Only esophageal cancer patients treated by surgery alone using the Sweet approach at the Department of Thoracic surgery at the Hebei Province Cancer Center were eligible for analysis. Since complete follow-up data were obtained only for patients resected from 1970-1994 and in 2008 (with lost to follow-up rate below 10%), patients were recruited on the basis that they were operated in these 26 calendar years. Because in these years the Sweet approach had been the dominating surgical procedure used at the Department, and multidisplinary therapy (surgery plus neoadjuvant or adjuvant chemotherapy, radiotherapy, chemoradiotherapy ,or immunotherapy) was not routinely offered at that time, 82.9% of the operated patients (5349/6453) were eligible for analysis. Patients excluded included: (1). 43 patients with P T is tumor (UICC/AJCC 7 th stage 0) and 143 patients with p T 1a (Tumor invading the lamina propria, UICC/AJCC 7 th stage I a )for reasons as stated in literature [8]; (2). 809 patients who had been offered neoadjuvant or adjuvant chemotherapy, radiotherapy, or immunotherapy in addition to surgery for reasons to eliminate confounding by advances in these multimodality therapy; (3). 13 patients accepted transhiatal esophagectomy, and 90 patients undergone esophagectomy through a right thoractomy were also excluded. (4). 6 patients with cervical esophageal cancer were excluded because these tumors are biologically different from their thoracic counterpart and were operated in collaboration with surgeons of the laryngology Department.
Finally a total of 5349 patients with carcinoma of the thoracic esophagus undergoing esophagectomy through left thoractomy with curative intent were included (1970-1979 (N=1505), 1980-1989(N=1840), 1990-1994 (N=1387), and 2008 (N=617)). Surgery was performed by 60 chief thoracic surgeons spanning 3-4 consecutive generations. By the end of 2017 any chief surgeon in the department has performed at least a minimum of 100 esophagectomies. Surgical techniques used were introduced in detail by a previous study. [4] 2. year afterwards for surveillance of surgical results and recurrence. During visit, esophagography, CT, laboratory examination were prescribed and necessary intervention procedure was performed to treat complication. In addition to period clinic visit by the patient personally, the patient or his/her family is also contacted every half a year through letter or telephone by a special follow-up group set up by the Department of Thoracic Surgery at the cancer center. This communication with the patient or his/her family is continued year after year. If response from the patient stopped, an active visit or call was made by the follow-up group to the family to con rm the survival status. From 1970From -1989, every year the lost to follow-up rate was less than 10%. But since 1990, the rate began to exceed 10%. To reduce lost to follow-up, we actively visited the family of each non-response patient and if again failed, we checked household record at the Room of Vital Statistics in the Local Police Station to verify the survival status. By this way, we collected complete survival data for patients operated in the year from 1990-1994 and in 2008.

Statistical Analysis
The end point for follow-up was to 31, December 2017. The minimum time of follow-up for censored patient had been 9 years. The SPSS statistical package, version 20.0 for Windows (IBM, Armonk, NY, USA) was used to construct data sets and perform statistical analyses. The signi cant level of difference between means of normally distributed continuing variables was tested by T Test, that between distribution unknown variables was tested by the Mann-Whitney Test, and that between rates of qualitative variables was tested by the Chi-square Test. One-way analysis of variance (ANOVA) was employed to test the difference in continuous variables over the four time periods and the linear trend. The increasing trend in the rates of resectability and postoperative survival and the decreasing trend in the rates of postoperative complication and operative mortality according to the four time periods were examined by linear-by-linear association of the Pearson chi-square test. Five-year survival rates for large sample of patients (N>=100) were calculated by Life-table methods and the differences between time period were tested according to Gehan Test. For small sample of patients in subgrouped analysis, 5-year survival rates were calculated by the Kapalan-Meier method and the differences between time periods were tested by the Log-rank test. P<0.05 was considered as statistically signi cant. When calculating survival rates, patients died of surgical complications were included.
To evaluate the effect of prognostic factors on survival of patients with surgically resectable esophageal cancer (degree of residual tumor: R0 , R1 , R2), multivariate analysis was performed using Cox regression model. Variables found to be signi cant in univariate analysis were included, such as sex, age, decade of surgery (1970s vs. 1980s vs. 1990-94 vs. 2008), pTNM stage ( B , A, B , III A , III B ,  IIIC (excluding T4bN3)), postoperative complication (absent vs. present), averaged number of lymph nodes harvested, percent of lymph nodes metastasized. For variables found to be related to prognosis by multivariate Cox regression, their trend over the four decades were assessed.

Ethics approval and consent to participate
The study was approved by the Institutional Ethics Review Board of 4th Hospital of Hebei Medical University (20160001). Informed consent to participate was obtained from study participants and/or their family.

Consent to publication
There are no details/images/videos relating to an individual person reported within the manuscript. Consent to publication nonapplicable.

General demographic and clinicopathologic Characteristics
Of the 5349 esophageal cancer patients treated by surgery alone with the Sweet approach, the male to female sex ratio was 2.1 :1 and the mean age at diagnosis was 53.3 years old. Regarding tumor location of esophageal cancer, the tumor was in the upper third in 225 of 5349 cases (4.2%), in the middle third in 3969 (74.2%), and in the lower third in 1155 (21.6%). Histologically 98.6% of esophageal cancer is squamous cell carcinoma, of which 57.1% were poorly differentiated.

Death rate of complication and anastomotic leakage
In contrast to the decreasing rates of postoperative complication and operative mortality, the death rate of complication remained the same as 52.9%, 55.2%, 36.9%, and 59.4% over the four time periods (P>0.05) however and so did the death rate of anastomotic leakage(77.3%, 85.7%, 96.9%, 75.0%, P>0.05) ( Table 2).

. Improvement in ve year survival rates
The overall 5-year survival rate was 36% and the 5-year survival rate progressed signi cantly from 32%, 35%, 38% to 43% (P<0.001) (  Fig 2). When the signi cance of improvement in 5-year survival rate between consecutive decades was tested, that from 1970s to 1980s and from 1980s to 1990-1994 were non-signi cant, but the progression from 1990-1994 to 2008 was signi cant (Table 3, Fig 2).
When improvement in 5-year survival over the four time periods was analyzed according to pTNM stage, signi cant improvement was observed only among patients in the stage of B / A / B /III A /III B /III C (excluding T 4b N 3 ) (Fig 3), not among patients in the stage of III C (T 4b N 3 )/IV (Fig 4).

Multivariate Cox regression analyses
Of the 7 demographic or clinicopathologic characteristics found to be signi cant by univariate Cox regression analysis, ve were identi ed as signi cant independent prognostic factors by the backwards multivariate model. These were pTNM stage (P<0.001), postoperative complications (P<0.001), the percent of lymph node metastasized (P<0.001), decade of surgery (P<0.01), and the average age of the patients(P<0.05).However, sex and the median number of lymph nodes harvested did not reached signi cant level as independent prognostic factor. Of the ve signi cant prognostic factors, three had changed signi cant over the four decades; e g., pTNM stage and rate of postoperative complications decreased signi cantly (both P<0.001), but the average age of patients resected increased signi cantly (P<0.05). Of note, the median number of lymph node retrieved was eliminated in step 2 backwards regression from signi cant prognostic factors (Table 4).

Discussion
It has been claimed that in China the long term survival of esophageal cancer patients treated by surgery alone has not improved during the past 60 years. [5]Reasons for this include late stage at diagnosis, relaxation of surgical indication, exclusive use of esophagectomy through the left thoracic approach with incomplete two eld lymphadenectomy (Sweet). [2][3][4][5]However, lack of reliable follow-up data in high volume centers for patients operated during the 1990s and 2000s, as a result of growing hospital-patient distrust, has also contributed in causing the uncertainty. [7] After excluding esophageal cancer patients who had undergone multimodality treatment, our present retrospective study found that from 1970s, 1980s, 1990-1994, to 2008, the 5-year survival rate of those patients treated by surgery alone had increased signi cantly from 32%, 35%, 38% and 43% (P<0.001). Together with a 5-year survival rate of 23.5% previously reported for 1290 esophageal cancer cases resected between 1952-1973 (when adjuvant therapy was scarce) at the center, [9] our results suggest that signi cant improvement in long term survival has been associated with the Sweet approach from 1952-2008, just as that reported in Japan associated with the right thoractomy with three eld lymphadenectomy. [10][11][12][13][14][15][16][17] Since we excluded those patients who had undergone multimodality treatment, survival improvement associated with the development of chemo/radio/immuno therapy was not the focus of present study.
Not only an increasing trend, the 5-year survival rates ( 35%-38%) of patients resected during the 1980s and 1990s associated with the Sweet approach in our center are also comparable to the rate reported by Ando (40%) for 419 esophageal cancer cases resected from 1981-1995 at Keio University Medical School Hospital of Japan by the right thoracic approach with three eld lymphadenectomy, [14] or are identical to that of 34.3% as reported by Isono K for a total of 1740 cases of esophageal cancer resected similarly at 35 Japanese centers between 1983-1989, [10] although in Isono's study hospital deaths due to severe surgical complication were excluded when calculating 5-year survival rates.
Esophagectomy through the right thoracic approach with three eld lymphadenectomy was rst introduced in 1983 at the Cancer Center of Ciba University Medical School of Japan. [10] Although this surgical approach carries a higher risk of operative complications such as injury of recurrent laryngeal nerves and is time-consuming, it is preferred over the Sweet approach because it provides excellent surgical exposure to the upper right mediastinal lymph nodes and ensures higher harvesting of suspected metastatic lymph nodes. [15] In Japan the approach was reported to have increased the 5-year survival rate by 10% than the traditional Sweet approach. [10][11][12][13][14][15][16][17] It is convincing that a complete lymphadenectomy would have contributed signi cantly more than a limited one to the long term survival. In expectation of breakthrough in long term survival of esophageal surgery, right thoractomy with three eld LND was rapidly accepted since it was introduced to China in the late 2000s. [15] On the other hand, the Sweet procedure has been widely used in high volume cancer centers in China for more than 80 years.  [14] In present study, although postoperative complication decreased signi cantly, death rate among patients with complication unchanged, suggesting prevention of surgical complication is very important.
Our Center in collaboration with the second Department of Surgery of Shinshu University Medical School of Japan performed a comparative study between esophagectomy through the Sweet approach with limited two eld lymph node dissection (LND) vs. esophagectomy through the right thoracic approach with extensive two or three eld LND in 1994. The results indicate that although lymph node harvesting is increased and more metastatic lymph nodes are found by extensive LND, it also carries a higher rate of surgical complications and a high operative mortality rate, and the stage-speci c 5-year survival rates showed no signi cant difference ( Table 5). The authors concluded that extended lymphadenectomy should be balanced to surgical safety to improve the long term survival. [23] In a randomized prospective clinic trail by Li B et al of Shanghai Cancer Center of Fudan University published in 2017, disease-free survival (DFS) and overall survival (OS) were compared between the right thoractomy with extensive two eld lymphadenectomy (n=146) and left thoractomy with limited two eld lymphadenectomy (n=140). The number of median lymph node retrieved was signi cantly different (22 (17-33) vs. 18 (13-26), P<0.001)). The 3-year DFS rates were 62% vs. 52% (P= 0.047) and the 3-year OS rates were 74% vs. 60% ( P= 0.029) respectively. [24]However, a recent study in a high volume center (Henan Cancer Center) comparing the right (n=202) and left thoracic approach esophagectomy (n=235) found no signi cant difference in 5-year survival rate. [20] The Henan colleagues argued that the left approach as described in the Fudan trial might have been unrepresentative of a standard Sweet approach by experienced hands in high volume centers because the usual advantages in operation time, hospital stay and surgical complication are not apparent as compared to that in other Chinese studies. [18][19][20][21]25]Interestingly however, Henan colleagues not only demonstrated the usual advantages concerning surgical safety for the Sweet approach, the median number of lymph node retrieved by them with the left approach was no less than that with the right approach (median 21 (8-64) vs.26 (8-60), P=0.708). [20] The discrepancy between these two studies seem to suggest that both a minimum complication and a complete lymphadenectomy are mandatory for the improvement of long term survival, and currently in China regarding both there is room for improvement.
In Cox regression analyses, we found pTNM stage, postoperative complication, percent of lymph node metastasis and year of surgery were signi cant independent prognostic factors, but the median number of lymph node retrieved has not reached signi cant level. The result is understandable because the median number of lymph node retrieved at our center has not improved from 1970-2008. we consider the improvement in long term survival in present study is related to but not limited to following factors: 1. The proportion of patients with early cancer increased signi cantly over the four decades. This is possibly due to free population based screening programs organized by the government in endemic areas. Balloon cytology had been used between 1970-1990, and in recent years endoscopy with iodine staining and biopsy has become the main modality. 4. Improvement in surgical techniques such as choosing the neck as the site of esophagogastrostomy to increase the length of resection, [2][3][4] the increased use of stapled anastomoses since the late 1980s to reduce the rate of anastomotic leakage, [2][3][4] and development of advanced peri-surgical management by targeted nursing and nutritional support etc. [2][3][4] The strength of present study lies in the large number of patients with complete follow-up data in 26 years after excluding those who undergone multimodality treatment in a high-volume thoracic cancer center, and the surgical experience built up since 1952. In addition, patients died of surgical complications were included when calculating the ve year survival rates to re ect the in uence of surgical complication.
A weak point of this study lies in the limited number of lymph node retrieved, which might have prevented accurate TNM staging. However, since we are focused on studying the increasing trend of long term survival without the contribution of a signi cantly increased lymph node dissection, the inadequate lymph node staging may not have biased the trend of overall long term survival.
In summary, by analyzing survival data of 5349 cases of esophageal cancer treated by surgery alone with the Sweet approach in 26 calendar years in a high volume center, we found signi cant improvement in the short and long term results from 1970-2008. Because the number of median lymph node retrieved at our center had remained 4-5 over the four periods and which is far from being adequate according to current standard, [26] improvement in survival from 1970-2008 may have been derived largely from early diagnosis and reduced surgical complication by experienced hands with a high volume center. While the competence should be maintained, effort must be made to improve the extent of lymphadenectomy.

Declarations
Ethics approval and consent to participate The study was approved by the Institutional Ethics Review Board of 4th Hospital of Hebei Medical University (20160001). Informed consent to participate was obtained from study participants and/or their family.

Consent to publication
There are no details/images/videos relating to an individual person reported within the manuscript. Consent to publication nonapplicable.   Stage speci c progress in survival with surgically resected esophageal cancer from 1970 to 2008