PDF
ePub
Citation
Print
INTRODUCTION
In the United States, gastric cancer is a devastating disease, with a 5-year overall survival of only 30.6% [1]. Surgical resection with adequate oncologic margins and removal of regional lymph nodes (LNs) offers the best hope for long-term survival. LN status is an important prognostic indicator in gastric cancer, with positive LNs suggesting a poor prognosis [23]. However, patients with node-negative disease still have a 17% chance of disease recurrence, and a 5-year overall survival of only 53% [4].
Extent of lymphadenectomy remains a controversial topic in surgical management of gastric cancer. In Japan, extended lymphadenectomy, referring to a D2 LN dissection, is standard of care [5]. However, initial data from randomized controlled trials in British and Dutch populations failed to find a significant survival benefit for D2 dissections over D1 dissections [67]. Long-term follow-up in the Dutch study found improved disease specific survival with D2 dissections [8]. In the United States, the National Comprehensive Cancer Network (NCCN) guidelines currently recommend gastrectomy with D1 or modified D2 LN dissection, with preservation of the distal pancreas and spleen; the surgeon should examine at least LNs [9]. Recent evidence from Asian populations demonstrates a survival benefit to increasing the number of examined lymph nodes (eLNs), even in node-negative disease [10111213]. However, this has not been explored in African, European, or North and South American populations. Given differences in gastric cancer between Asian and other populations, the results of these studies are not necessarily applicable to African, European, or North and South American populations. This study sought to determine whether number of eLN, a proxy for lymphadenectomy, effects survival in US patients with node-negative gastric cancer.
MATERIALS AND METHODS
Data
This was a retrospective cohort study using data from the National Cancer Data Base (NCDB). This clinical oncology database, jointly sponsored by the American College of Surgeons and the American Cancer Society, is sourced from hospital registry data collected from over 1,500 Commission on Cancer (CoC) accredited facilities. The NCDB captures over 70% of newly diagnosed cancer cases in the United States. The NCDB contains readily available de-identified data, and therefore this study was not subject to institutional review board approval or oversight.
Patient selection
The NCDB (2003–2011) was reviewed for patients diagnosed with clinical stages I–III gastric cancer, who underwent surgical resection, with or without systemic therapy. Patients with clinical stage IV disease or unknown stage were excluded. Clinical stage is coded in the NCDB according to standard practice at each individual institution. Patients who did not undergo surgical resection were excluded. Patients were categorized by receipt of neoadjuvant therapy (NAT) vs. initial resection, and further stratified by number of eLN: ≤10, 11–15, 16–20, 21–30, and >30.
Outcomes and covariates
The primary variable assessed was overall survival. Analyses controlled for patient and disease characteristics including age, sex, race, insurance type (private, Medicare, Medicaid and other government programs, unknown, and not insured), and the Charlson/Deyo comorbidity index (CCI), an index of 15 comorbidities (myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, rheumatologic disease, peptic ulcer disease, mild liver disease, diabetes, diabetes with chronic complications, hemiplegia or paraplegia, renal disease, moderate or severe liver disease, and acquired immunodeficiency syndrome) [1415]. Median income of the patient's zip code was used as a proxy for socioeconomic status. Treatment facilities were characterized by type (community, comprehensive community, academic or research institution, other) and US geographic region (northeast, south, mid-west, west). Disease was characterized by the American Joint Committee on Cancer (AJCC) clinical stage, surgery type (proximal gastrectomy, total gastrectomy, distal gastrectomy, and surgery not otherwise specified), number of regional LNs removed, number of positive regional LNs, surgical margins (no residual tumor, residual tumor not otherwise specified, microscopic residual tumor, macroscopic residual tumor, and indeterminate and unknown margins), pathologic stage, and adjuvant therapy. The main covariate of interest was eLN, which was subdivided into 5 groups: ≤10, 11–15, 16–20, 21–30, and >30 LNs.
Statistical analysis
Statistical analyses were performed with Stata software (version 12.1; StataCorp., College Station, TX, USA). Patient, disease, and facility characteristics were compared within each cohort with analysis of variance for continuous variables and χ2 tests for categorical variables. Kaplan-Meier analyses were performed for each clinical stage and treatment, and stratified by eLN. The proportional hazards assumption was violated, and thus multivariable survival analyses were performed for each initial treatment using a Weibull model, controlling for covariates described above.
RESULTS
From 2003 to 2011, the median number of eLN in node negative resected gastric cancer has steadily risen from 8 to 14 (Fig. 1).
Fig. 1
Trends in median number of LN examined from 2003 to 2011.
LN = lymph node; eLN = examined lymph node.
Patient characteristics of NAT cohort
Of the 1,036 (27%) who received NAT, 40.5% (n=420) had ≤10, 21.8% (n=226) had 11–15, 16.8% (n=174) had 16–20, 13.4% (n=139) had 21–30, and 7.4% (n=77) had >30 eLN (Fig. 2). Of those who received NAT, 58.0% underwent a suboptimal lymphadenectomy as defined according to NCCN guidelines (<15 eLN).
Patient, disease, and treatment characteristics of patients who received NAT, stratified by eLN, are presented in Table 1. Patients who received a suboptimal lymphadenectomy (eLN ≤10) tended to be male (P<0.001), were more likely to be treated at a comprehensive community center (P<0.001), and were more likely to undergo a proximal gastrectomy (P<0.001). Age (P=0.915), race (P=0.395), comorbidities (P=0.785), surgical margins (P=0.225), and clinical and pathological stages (P=0.901 and P=0.124, respectively) did not significantly differ among eLN groups.
Table 1
Patient, disease, and treatment characteristics of NAT cohort
Survival of NAT Cohort
Kaplan-Meier analyses of patients who received NAT are stratified by eLN and presented in Fig. 3. Inadequate lymphadenectomy (eLN ≤10) was associated with worse survival in clinical stage II and III disease; however, this association was only significant in stage III disease (P=0.020).
Results of a Weibull survival model of patients who received NAT are presented in Table 2. A greater number of eLN was associated with improved hazards of mortality (eLN 16–20: HR, 0.71; P=0.039, eLN 21–30: HR, 0.55; P=0.001). Treatment at an academic center was also associated with a reduction in mortality (HR, 0.52, P=0.005). Greater age (80–90: HR, 3.52; P<0.001) and coverage by Medicaid (HR, 1.59, P=0.019) were associated with increased hazards of mortality.
Table 2
Factors impacting survival in NAT cohort
Patient characteristics of initial surgery cohort
Of the 2,795 patients who underwent initial surgery, 42.5% (n=1,187) had ≤10, 19.8% (n=553) had 11–15, 14.5% (n=404) had 16–20, 15.5% (n=432) had 21–30, and 7.8% (n=219) had >30 eLN (Fig. 4). Of those who underwent initial surgery, 58.6% underwent a suboptimal lymphadenectomy (<15 eLN).
Patient, disease, and treatment characteristics of patients who underwent initial surgery are stratified by eLN and presented in Table 3. Patients who received a suboptimal lymphadenectomy (eLN ≤10) tended to be older (P<0.001), white (P<0.001), and treated at a comprehensive community center (P<0.001). They tended to undergo a proximal gastrectomy (P<0.001), and have a more advanced pathological stage (P=0.024). Clinical stage (P=0.192), surgical margins (P=0.161), and adjuvant therapy (P=0.061) did not significantly differ among eLN groups.
Table 3
Patient, disease, and treatment characteristics of initial surgery cohort
Survival of initial surgery cohort
Kaplan-Meier analyses of patients who underwent initial surgery are stratified by eLN and presented in Fig. 5. Inadequate lymphadenectomy with eLN ≤10 was associated with worse survival in clinical stage I–III diseases; however, this association was only significant in stages I and II (P<0.001 and P=0.002, respectively).
Results of a Weibull survival model of patients who underwent initial surgery are presented in Table 4. A greater number of eLN was associated with improved hazards of mortality (eLN 11–15: HR, 0.81; P=0.021, eLN 16–20: HR, 0.73; P=0.004, eLN 21–30: HR, 0.62; P<0.001, and eLN >30: HR, 0.58; P=0.001). Female sex, Hispanic or other race, greater median income, treatment at an academic or research center, and receipt of adjuvant therapy were also associated with a reduction in mortality. Greater age, coverage by Medicaid or Medicare, receipt of total gastrectomy, positive surgical margins, and advanced pathological stage were associated with increased hazards of mortality.
Table 4
Factors impacting survival in initial surgery cohort
DISCUSSION
Despite a near linear improvement in median eLN from gastrectomy over the past decade, most US patients with node-negative gastric cancer received operations that failed to meet NCCN guidelines of retrieving ≥15 LNs. This concerning statistic suggests that surgeons nationwide may not understand the correlation between the systemic potential of gastric cancer and clearance of regional LNs, thus failing to perform the meticulous LN dissection necessary to achieve an optimal LN yield. University facilities may adhere more strictly to NCCN guidelines, as a study on 7 US universities reported a mean of 16 eLN, and our current study found that academic centers were more frequently associated with adequate oncologic lymphadenectomy [4].
The effect of eLN on western patients was explored by Smith et al. [16] with a retrospective analysis of the Surveillance, Epidemiology, and End Results (SEER) database from 1973 to 1999. They reported an association between increased number of eLN and improved survival in patients with T1-3, N0-1 gastric cancer [16]. However, given the increased utilization of NAT, as well as improvement in lymphadenectomy techniques, this analysis also includes antiquated treatment protocols [17]. Jin et al. [4] evaluated factors associated with recurrence and survival in 317 patients with node-negative gastric cancer from 2000–2012, and reported that eLN ≥15 was associated with improvements in overall survival, but not disease recurrence. However, given the small sample size and inclusion of exclusively university facilities, the authors conceded that the analysis may have been insufficiently powered [4]. To our knowledge, this is the first study to evaluate national treatment patterns and outcomes of node-negative gastric cancer after the formation of the Meta-Analysis Group in Cancer (MAGIC) though it did include data from years prior to the publication of the trial by Cunningham et al. [17] in 2006 which established NAT as standard of care. It also includes the largest number of western patients to undergo NAT or surgery for node-negative gastric cancer.
In our study, both in patients who received NAT and in those who underwent initial surgery, a greater number of eLN was associated with improved survival. A study by Deng et al. [18] of 112 Chinese patients with node-negative gastric cancer reported that eLN >20 was associated with improved survival. Another study on 600 Chinese patients with node negative gastric cancer recently reported that eLN was the strongest independent prognostic predictor and urged eLN to be considered a mandatory requirement for improving prognostic evaluations [10]. An Italian study, pre-dating the MAGIC trial, of 301 node-negative patients from 1992–2002 reported an association between eLN >25 and improved survival [19]. These studies all advocate for meticulous LN dissection and optimizing the number of harvested nodes.
To our knowledge, this is the largest and most contemporary study on western patients with node-negative gastric cancer treated at a wide spectrum of CoC-accredited facilities. However, there are some important limitations which should be considered when interpreting the results. The NCDB is a database sourced from hospital registry data from diverse institutions across the United States, and data recording may vary slightly from facility to facility. Furthermore, though the NCDB was designed to collect oncologic data, it lacks chemotherapy regimen, disease recurrence, disease-specific death, and complications data. Given the retrospective nature of this study, there likely exists a selection bias, with patients with more severe presentations treated more aggressively. Additionally, node-negative disease was defined as patients with AJCC pathologic N0 disease. The database does not specify why patients received NAT, but it is possible that patients with clinically node positive disease converted to pathologic node negative disease following NAT and would thus be included in this study. While the NCCN recommends examination of at least 15 LNs, this study included patients with suboptimal lymphadenectomies to provide a more complete overview of US gastric cancer care. However, sub-optimal lymphadenectomy was controlled for in the multivariable Weibull survival analyses. Most patients in this study underwent proximal gastrectomy, which could include patients with gastroesophageal junction cancer, which current NCCN guidelines classify as esophageal tumors. However, this study used the NCDB gastric participant user file (PUF), not the esophagus PUF, to reduce possible misclassification. To further account for this, the multivariable Weibull analyses controlled for tumor location. Finally, it is important to consider the differences in pathologic specimen evaluation between Eastern countries and the United States [20]. While Eastern surgeons dissect out each LN station, western surgeons typically submit specimens en bloc [21].
In conclusion, most US patients with node-negative gastric cancer receive a suboptimal lymphadenectomy. Even in node-negative disease, increasing the number of retrieved lymph nodes appears to have therapeutic and prognostic value, irrespective of initial treatment. This suggests a survival benefit to meticulous lymphadenectomy in western patients with node-negative gastric cancer.
XML Download