Background

Tumors of the gastroesophageal junction and gastric cardia present many challenges to the clinician, the first of which is accurate classification. Gastric cardia malignancies are considered separate entities from both esophageal and distal gastric cancers.1 The methods of characterizing proximal gastric malignancies include the American Join Committee on Cancer (AJCC) and Siewert classifications which both emphasize commonalities with esophageal cancer.2–4 Although gastric cancer incidence and mortality have decreased over the previous several decades, gastric cancer remains the fourth most common cancer and second leading cause of cancer-related death worldwide.5 In the USA, gastric cancer is the seventh most common cause of cancer-related death.5 In addition, the incidences of proximal gastric adenocarcinoma and esophageal adenocarcinoma have been observed to increase in recent years despite a decrease in the incidence of all gastric adenocarcinomas.6–8

Several different operative approaches are commonly employed for resection of proximal gastric cancers, including abdominal, transthoracic, and thoraco-abdominal. For tumors judged to be more distal and sharing characteristics with other gastric adenocarcinomas, total gastrectomy is often employed as first-line surgical therapy. A recent American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) analysis showed that total gastrectomy had increased rates of major morbidity and mortality when compared to partial gastrectomy; however, this analysis looked at all gastric malignancies and did not differentiate based on disease site.9 Likewise, another recent ACS-NSQIP analysis compared gastrectomy to esophagectomy for proximal gastric cancer, but did not parse the surgical approach.10

More proximal gastroesophageal cancers are typically treated in a manner similar to distal esophageal malignancies requiring gastroesophageal resection for adequate treatment. A recent meta-analysis of transthoracic approaches to esophageal resection shows increased postoperative morbidity but equivalent short-term mortality when compared to abdominal approaches.11 Approach-associated morbidity rates are important, as complications from surgical resection of proximal gastric cancer can lead to delays in adjuvant therapy, which adversely affect oncologic outcomes.12–15

The aim of this study was to characterize the short-term morbidity and mortality associated with various surgical approaches utilized to resect malignancies of the gastric cardia. Additionally, this study sought to characterize risk factors for morbidity and mortality that are common to patients undergoing resection of a gastric cardia malignancy.

Methods

The ACS-NSQIP is a multi-center, prospectively gathered, and risk-adjusted database. The ACS-NSQIP database includes more than 150 perioperative risk factor and outcomes variables. Details regarding ACS-NSQIP data collection, variables, and analysis have been previously published.16

The ACS-NSQIP Participant Use File (PUF) was interrogated to identify patients who underwent an operation for malignant neoplasm predominantly located at the gastric cardia (ICD-9 code 151.0 only) at participating NSQIP hospitals from 2005 to 2012. Current Procedure Terminology (CPT) codes were used to group patients by surgical approach. Patients who underwent total gastrectomy (TG) (CPT code 43620, 43621, and 43622), transhiatal esophagogastrectomy (THE) (CPT code 43107), or thoraco-abdominal esophagogastrectomy (TAE) (CPT code 43112 and 43117) were identified. Ancillary PUF fields were used to exclude esophagectomies associated with an intestinal conduit and emergency operations from analysis. There were 204 patients who underwent TG, 271 patients underwent THE, and 507 patients underwent TAE. These patients’ clinical characteristics are detailed in Table 1. The surgical approach was also recoded into a binary categorical variable with the two abdominal (non-thoracotomy) approaches (TG + THE) combined and compared to TAE.

Table 1 Demographics of patients with gastric cardia malignancy by approach
Full size table

Major morbidity was defined as the occurrence of at least one of the following complications: organ space infection, pneumonia, unplanned intubation, pulmonary embolism, ventilator requirement greater than 48 h, progressive renal insufficiency, acute renal failure, cerebrovascular accident, coma, cardiac arrest, myocardial infarction, deep venous thrombosis, sepsis, septic shock, and return to the operating room.17 Cardiopulmonary morbidity was defined as the occurrence of at least one of the following complications: pneumonia, unplanned intubation, pulmonary embolism, ventilator requirement greater than 48 h, cardiac arrest, or myocardial infarction. Mortality was defined as death from any cause within 30 days of surgery or death occurring over 30 days from surgery if the operation and mortality were within a single prolonged hospitalization.

Statistics

All continuous variables are reported as median value with interquartile range (IQR) and/or range, or mean value ± standard deviation. Continuous variables were compared using the Kruskal-Wallis test or the Mann–Whitney U test. Categorical variables were compared using the chi-squared test or Fisher’s exact test. A 2-tailed p value less than 0.05 was considered significant. All appropriate risk factors found to have a univariate p value less than 0.10 were included in multivariate analysis. Multivariate analysis was done using binary logistic regression and the backwards stepwise method. All statistical analysis was performed using SPSS Version 22.0.0.0 (SPSS, Inc., Chicago, IL, USA).

Results

Overall, there were 982 patients identified in the database who met inclusion criteria. The median age was 65 years (range 20–88) and 807 (82.2 %) were male. The number of patients allocated to each approach was 204 TGs (20.8 %), 271 THE (27.6 %), and 507 TAE (51.6 %). Patients undergoing TG were older (p < 0.001), had lower ASA scores (p = 0.030), and were less likely to be male (p < 0.001) than those undergoing THE or TAE. Patients in all three groups had similar rates of functional independence, body mass index (BMI), smoking, and alcohol use (Table 1).

The most frequently present preoperative comorbidity for all three approaches was hypertension. There were few differences in preoperative comorbidity rates; however, patients with CHF or renal failure were found exclusively in the TG group, all be it at low levels. Patients undergoing a THE and TAE resection were more likely to receive preoperative radiotherapy than those undergoing gastric only resection (TG 9.8 %, THE 24.7 %, and TAE 23.9 %, p < 0.001). Patients undergoing TG were also more likely to have had an operation in the previous 30 days (p = 0.016) (Table 2).

Table 2 Comorbidities of patients with gastric cardia malignancy by surgical approach
Full size table

Regarding perioperative outcomes, operative times were increased for TAE (391.1 ± 121.2 min) vs. TG (321.1 ± 130.3 min) or THE (294.1 ± 101.1 min) (p < 0.001). Hospital length of stay was near 14 days for all patients in this cohort, with no differences based on surgical approach. Morbidity (TG 40.7 %, THE 43.5 %, and TAE 49.9 %, p = 0.048) and major morbidity (TG 32.8 %, THE 32.1 %, and TAE 36.7 %, p = 0.368) were considerable across all three approaches with TAE being the most morbid procedure (Fig. 1). Although cardiopulmonary morbidity rates were equivalent across approaches, postoperative myocardial infarction was more common after TAE (2.0 %) compared to TG (0 %) and THE (0.4 %) (p = 0.035). There was no difference in 30-day mortality by operative approach (Table 3). Additionally, TAE had higher rates of perioperative transfusion (p = 0.035) when compared to abdominal approaches (Table 4).

Fig. 1
figure 1

Comparison of mortality, cardiopulmonary morbidity, major morbidity, and all morbidity by approach

Full size image
Table 3 Univariate analysis of outcome measures by surgical approach
Full size table
Table 4 Univariate analysis of outcome measures by surgical approach
Full size table

For all patients with proximal gastric cancer undergoing surgery, the factors associated with major morbidity by univariate analysis included smoking (p = 0.042), alcohol use (p = 0.013), dyspnea (p = 0.031), diabetes (p = 0.033), COPD (p = <0.001), history of myocardial infarction (p = 0.003), hypertension treated with medication (p = 0.011), perioperative transfusion (p = 0.034), prior cardiothoracic surgery (p = 0.003), and preoperative chemotherapy (p = 0.016). On multivariate analysis, alcohol use (p = 0.023, OR = 2.30), COPD (p = 0.001, OR = 2.69), history of myocardial infarction (p = 0.041, OR = 9.26), hypertension treated with medication (p = 0.034, OR = 1.35), perioperative transfusion (p = 0.034, OR = 1.45), prior cardiothoracic surgery (p = 0.050, OR = 2.54), and operation in the previous 30 days (p = 0.049, OR = 2.94) were independent factors associated with major morbidity. The only independent factor protective for major morbidity was preoperative chemotherapy (p = 0.005, OR = 0.50).

The factors which were associated with mortality included age greater than 65 years old (p = 0.003), dyspnea (p = 0.009), diabetes (p = 0.022), COPD (p = 0.019), history of myocardial infarction (p = 0.015), hypertension (p = 0.031), renal failure (p = 0.048), operation within the previous 30 days (p = 0.044), any morbidity (p < 0.001), major morbidity (p < 0.001), and major morbidity (p < 0.001). On multivariate analysis, the independent predictors of 30-day mortality were age greater than 65 years (p = 0.003, OR = 8.34), history of myocardial infarction (p = 0.004, OR = 21.89), and cardiopulmonary morbidity (p < 0.001, OR = 27.03) (Table 5).

Table 5 Analysis of predictors of major morbidity and mortality
Full size table

Discussion

Analysis of the ACS NSQIP data shows that there is significant morbidity associated with all surgical approaches to resection of proximal gastric cancer. TAE proved to be the most morbid of procedures likely because of the additional thoracic field. It is important to note that the study population was contained to patients diagnosed with proximal gastric cancer, excluding patients coded as esophageal cancer. Although the utilization of TAE could indicate appropriate surgical management of a more aggressive gastroesophageal junction lesion that may engender more complications, in this dataset, there were identical rates of neoadjuvant chemoradiotherapy between the THE and TAE groups, suggesting that choice of surgical technique was more surgeon specific than patient or cancer specific. These data indicate that when oncologic equipoise is present, the least morbid approach should be considered.

The ACS-NSQIP dataset does not collect information regarding the use of laparoscopy prior to 2010 and therefore was not analyzed in this study. One possible explanation for the differences in morbidity, besides an additional operative field, is the difference in laparoscopy utilization and diffusion to practice for each procedure. The first laparoscopic partial gastrectomy for cancer was reported in the early 1990s and laparoscopic total gastrectomy in the early 2000s. Laparoscopic gastrectomy for malignancy has proved to be a safe alternative to open surgery in a randomized controlled trial published in 2005 with at least equivalent long-term outcomes.18 The first minimally invasive esophagectomy was performed much later with the first randomized controlled trial showing equivalent improved short-term outcomes published in 2012.19 In the future, increased utilization of minimally invasive esophagectomy may help to decrease the gap in morbidity and length of stay as compared to open abdominal approaches.

In selecting an appropriate operative intervention for gastric cardia malignancy, there are still many variables to consider. The ability to perform a complete oncologic resection remains the cornerstone of operative resection; however, additional factors including patient age, comorbidity status, extent of gastroesophageal involvement, histologic type, and the risks inherent to each operation should also the factors into surgical decision making.20 Multi-modality treatment of proximal gastric cancer has been shown to improve outcomes but delays from surgical complications have been shown to lead to delays in adjuvant therapy as well as increased rates of early recurrence.12,13,21 In patients with significant comorbidities who are more likely to experience surgical complications, selection of the least morbid approach that will achieve an R0 resection is paramount to improving outcomes.

This analysis also demonstrates the importance of considering patient comorbidities when deciding to operate for proximal gastric cancer. A history of advanced age combined with prior myocardial infarction, renal failure, or a recent operation was associated with the highest risk and should trigger a specific informed discussion with the patient prior to proceeding with resection. The analysis did show modifiable risk factors associated with major morbidity. Patient’s alcohol consumption of greater than two drinks a day within 2 weeks of operation was independently associated with major morbidity. This indicates that patients should be counseled regarding alcohol cessation in the preoperative period. Additionally, having a history of another operation performed within 30 days was associated with major morbidity. While it’s not always possible to delay definitive cancer treatments due to a previous operation, clinicians should reconsider scheduling of multiple invasive procedures in temporal proximity.

This study has several limitations. In order to investigate proximal gastric cancer, gastric cardia malignancy was used as a surrogate diagnosis code. Gastric cardia malignancy is not a perfect surrogate for proximal gastric cancer and represents a weakness in the current ICD-9 code classification that is a component of the ACS-NSQIP database. In future iterations of diagnostic coding, it may be beneficial to redefine terminology used to report tumor location. The ACS NSQIP database does not collect cancer-specific variables including cancer staging, histologic grading, margin status, and resection status. The database also does not collect oncologic outcomes. Unfortunately, this limits the ability of the analysis to account for stage-specific differences in the utilization of neoadjuvant therapy and/or radical surgical resection that may be more risky, but simultaneously more likely provide optimal long-term oncologic outcomes. Finally, ACS-NSQIP data does not supply information regarding hospital or operative surgeon procedural volumes and specialty training that is likely to influence patient outcomes.22–24

Conclusion

In proximal gastric cancer, all three operative approaches were associated with clinically significant rates of major morbidity. Abdominal approaches are utilized with less frequency but are associated with significantly decreased morbidity. The ability to perform a complete oncologic resection, morbidity by approach, tumor location, margin status, clinical staging, and other patient risk factors all need to be considered in concert when selecting an operative approach for gastric cardia cancer.