The Role of the Cancer Center When Using Lymph Node Count as a Quality Measure for Gastric Cancer Surgery

Importance  Cancer center recognition, offered as accreditation by the American College of Surgeons Commission on Cancer or the National Cancer Institute, and quality measure reporting purport to improve the quality of cancer care. For surgically resectable gastric cancer, removal of 15 or more lymph nodes has been associated with improved outcomes and has been endorsed as a gastric cancer quality measure.

Objectives  To determine whether cancer center classification is associated with compliance with the lymph node–count quality measure and the effect of compliance with the measure on overall survival.

Design, Setting, and Participants  A retrospective review of prospectively collected population-based data from the Surveillance, Epidemiology, and End Results Cancer Registry of Greater California and California Cancer Registry was conducted. Participants included patients who underwent surgery for stage I to III gastric adenocarcinoma between January 1, 2004, and December 31, 2010.

Main Outcomes and Measures  Compliance with removal of 15 or more lymph nodes and overall survival.

Results  Of 3321 gastric cancer cases, 42.3% had a minimum of 15 lymph nodes removed. Hospitals with cancer program recognition treated 69.9% of the cases. In hospitals without cancer program approval, 34.8% of the patients had 15 or more lymph nodes removed compared with 45.5% in the facilities with cancer program approval. Logistic regression analysis demonstrated that patients undergoing gastrectomy had significantly higher odds of having 15 or more lymph nodes removed if they were younger (trend P < .001), Asian/other race/ethnicity (adjusted odds ratio [AOR], 1.24; 95% CI, 1.03-1.50), or non-Hispanic black (AOR, 1.37; 95% CI, 1.03-1.82) compared with non-Hispanic white, received diagnosis at a progressively higher stage (trend P < .001), or received diagnosis in a more recent year (trend P < .001). Removal of 15 or more lymph nodes was associated with cancer program recognition (vs no recognition) (odds ratio, 1.48; 95% CI, 1.25-1.74). Cox proportional hazards regression showed that improved survival was predicted by removal of 15 or more lymph nodes (hazard ratio [HR], 0.70; 95% CI, 0.63-0.78) but not by cancer program recognition (HR, 1.03; 95% CI, 0.92-1.15).

Conclusions and Relevance  Although adequate lymph node retrieval is more likely in hospitals with a recognized cancer program, survival outcome is associated with the lymph node count rather than with cancer program classification. Less than half of the cases reviewed in this study met the minimum lymph node–count guideline, indicating the need for process improvement for all hospitals.

Although it has been 15 years since the Institute of Medicine¹ released a landmark report addressing the quality of cancer care in the United States, their most recent publication, Delivering High-Quality Cancer Care, admonishes barriers to quality-care delivery and concludes that “the cancer care delivery system is in crisis.”² Among updated recommendations, the Institute of Medicine supports developing a national quality-reporting program with the support of professional societies using quality measures that are directly linked to outcomes. Established in 1922 by the American College of Surgeons, the Commission on Cancer (CoC) accredits hospitals with cancer programs that demonstrate adherence to continuously updated quality standards, mandated data reporting to the National Cancer Database, and periodic self-assessment and site visits. Institutions accredited by the CoC include more than 1500 US cancer programs that provide care to approximately 70% of the patients with newly diagnosed cancer in the nation.^3,4 Likewise, the National Cancer Institute (NCI) developed the framework for its network of cancer centers after passage of the National Cancer Act in 1971. The 68 cancer centers in the United States that currently carry NCI designation of a cancer center or comprehensive cancer center have met stringent criteria for basic science–, clinical-, and population-based research, training, and outreach designed to provide the public with efficient access to translational medicine for cancer care.⁵ Inherent in the mission of the CoC and NCI programs is an imperative to improve the quality of cancer care.

With the release of the 1997⁶ and 2002⁷ editions of the American Joint Committee on Cancer (AJCC) staging for gastric cancer, the recommendation that a minimum of 15 lymph nodes should be removed during surgical treatment for gastric cancer has been exhaustively studied. It is unclear whether the benefits are the results of stage migration and appropriately directed adjuvant therapy or any direct therapeutic effect; however, removal of 15 or more lymph nodes at the time of gastrectomy for cancer predicts improved survival outcomes^8-10 and is recommended in the National Comprehensive Cancer Network guidelines for surgical treatment of resectable gastric cancer.¹¹ Easily measureable and directly linked to outcomes, evaluation of at least 15 lymph nodes in gastrectomy specimens obtained during stomach cancer–directed operations is well suited for adoption as a quality standard. In May 2013, the Quality Integration Committee of the CoC formally endorsed guidelines specifying removal of a minimum of 15 lymph nodes as a quality measure for gastrectomies performed for nonmetastatic gastric adenocarcinomas.¹² We sought to determine whether cancer center recognition by either the CoC or NCI is associated with compliance with the 15 or more lymph node–count quality measure and whether compliance with the lymph node–count quality measure or cancer center classification predicts long-term survival outcomes.

This study was approved by the institutional review board at Loma Linda University Medical Center. The need for informed consent was waived.

We conducted a nonconcurrent, population-based cohort study using statewide data from the California Cancer Registry (CCR). The CCR is the state-mandated cancer surveillance system that consists of the 3 most populated Surveillance, Epidemiology, and End Results (SEER) program registries. Since 1988, information about the statewide occurrence of cancer, including precise anatomic and histologic definitions of the cancer, stage at diagnosis, types of cancer-directed therapies, treatment hospitals, variables necessary to compute hospital surgical volume, demographic information for patients, and vital status, has been confidentially reported to the CCR.¹³ As defined by the CoC, the lymph node–count quality measure for gastric cancer mandates that 15 or more lymph nodes are removed during gastric resection for gastric cancer. Cases meeting the designated criteria for analysis include patients with nonproximal stomach adenocarcinomas (International Classification of Diseases for Oncology [Third Edition] codes M-8120 to M-8240 and M-8255 to M-8576),¹⁴ clinical or pathologic stages I to III (nonstage IV or metastatic) who received gastrectomy (partial, subtotal, hemigastrectomy, near total, total, or not otherwise specified) at the reporting facility (Facility Oncology Registry Data Standards site-specific surgery codes 30-80) (Figure).¹⁵ Evaluable cases included those performed between January 1, 2004, and December 31, 2010, that were reported to the CCR by February 28, 2013. The hospital where gastrectomy was performed was verified for each study patient. California hospitals with current or prior CoC accreditation or NCI-designated cancer programs were classified as having a recognized cancer program. Other California hospitals that had never received CoC accreditation or NCI designation were classified as not having a recognized cancer program.

Compliance with the CoC minimum lymph node–count quality measure was assessed according to cancer program status (recognized or not recognized), age group at diagnosis (18-49, 50-69, and ≥70 years), sex, race/ethnicity (Asian/other, non-Hispanic black, Hispanic, and non-Hispanic white), cancer stage at diagnosis (IA, IB, II, IIIA, and IIIB), hospital gastrectomy volume, and year of diagnosis. Race/ethnicity was identified by medical records or surname. The AJCC sixth edition⁷ staging was used during the period of data entry in the CCR. A binary hospital surgical volume variable was defined using the 50th percentile cutoff as 1 to 20 and 21 or more (21-104) gastrectomies during the 7-year study period.

Multivariable logistic regression was used to assess the odds ratio for compliance (≥15 lymph nodes) vs noncompliance (1-14 lymph nodes) with the CoC minimum lymph node–count quality measure. To exclude mortality attributed to postoperative morbidity, long-term survival was calculated from 90 days after surgery through death or to the end of the survival period (July 31, 2012). Using the same gastrectomy cases and independent variables listed above, we performed a Cox proportional hazards model to evaluate the effects of all covariables, including the cancer program status and lymph node count variables, to assess whether cancer program classification and compliance with the guideline for removal of 15 or more lymph nodes independently predicted survival following gastrectomy. P value trends were determined by a test for linear trend by treating ordinal categories of dependent variables as continuous in the logistic regression and Cox proportional hazards ratio models. Data analyses were conducted using SEER*Stat, version 8.0.4 (Surveillance Research Program, National Cancer Institute SEER*Stat software) and SAS, version 9.3 (Base SAS, version 9.3 utilities; SAS Institute Inc).

After applying case selection criteria according to the CoC quality measure definition,¹² there were 3321 patients available for study (Figure). Overall, 42.3% of patients who underwent gastrectomy had 15 or more lymph nodes removed. Most (69.9%) gastrectomies were performed in hospitals with cancer program recognition. In hospitals without cancer program approval, 34.8% of the patients had 15 or more lymph nodes removed compared with 45.5% in institutions with cancer program approval. The mean age of patients having 1 to 14 lymph nodes removed was 70.4 years compared with 66.9 years for those with 15 or more lymph nodes removed (P < .001). The proportion of patients with at least 1 positive lymph node was not significantly different between those who received treatment in hospitals with cancer program recognition (52.1%) compared with those without cancer program recognition (54.2%) (P = .07, t test). Logistic regression analysis (Table 1) indicated that patients undergoing gastrectomy had significantly higher odds of having 15 or more lymph nodes removed if they were younger (trend, P < .001), were Asian/other or non-Hispanic black, had the cancer diagnosed at progressively higher stage (trend, P < .001), received treatment in a higher volume hospital, or received the cancer diagnosis in a more recent year (trend, P < .001). Patients who received treatment in a facility with cancer program recognition had 48% higher odds of having 15 or more lymph nodes removed compared with those who received treatment in a facility without cancer program recognition (OR, 1.48; 95% CI, 1.25-1.74). There was no significant difference in lymph node retrieval among women vs men or Hispanics vs non-Hispanic whites. There were upward linear trends in the annual percentages of patients meeting the 15 or more lymph node–count guideline when they received treatment in hospitals with (P = .001) or without (P = .13) cancer program recognition. However, a test for parallelism between the 2 trend lines showed significant divergence (P < .001): programs with cancer center recognition had a more rapid annual increase in the proportion of patients with 15 or more lymph nodes removed.

Adjusting for compliance with the 15 or more lymph node–removal guideline, the Cox proportional hazards regression model (Table 2) demonstrated significantly worse long-term survival after gastrectomy for older patients and those presenting with a more advanced stage of cancer and improved survival for Asian/other patients compared with non-Hispanic whites and those who received treatment in higher-volume hospitals. There were no significant survival differences between women and men, by year of diagnosis, between non-Hispanic blacks or Hispanics vs non-Hispanic whites, or between hospitals with and without cancer program approval. The mortality hazard ratios for patients in whom 15 or more lymph nodes were resected showed an independent survival benefit (hazard ratio, 0.70; 95% CI, 0.63-0.78) compared with those who had 1 to 14 lymph nodes removed.

The present study adds to the wealth of literature^8-10 demonstrating that removal of 15 or more lymph nodes improves the outcomes for patients undergoing definitive surgical therapy for gastric cancer and further supports the endorsement of the minimum lymph node–count gastric cancer quality measure by the CoC. Findings from this report show that when patients undergo gastrectomy for nonmetastatic gastric cancer, their risk of death is reduced by 30% when 15 or more lymph nodes are resected compared with removal of 1 to 14 lymph nodes. The National Comprehensive Cancer Network guidelines recommend that patients undergoing surgery for a curative intent of gastric cancer should undergo D1 and D2 lymphadenectomy¹⁶ with the goal of examining 15 or more lymph nodes.¹¹ Although a direct therapeutic benefit from extended lymph node dissection has been argued,⁹ removal of 15 or more lymph nodes has been associated with improved staging, regardless of whether a D1 or D2 lymphadenectomy is performed.⁸ The effect of site vs extent of lymphadenectomy deserves further exploration; however, CCR data do not distinguish between D1 and D2 dissections in gastrectomy specimens. The CCR data cannot determine whether differences in the number of lymph nodes retrieved are the result of better quality surgery or pathologic specimen lymph node identification.

Although treating institutions in this study that held current or prior CoC accreditation or NCI cancer center designation were 48% more likely to achieve the 15 lymph node threshold than were institutions without a recognized cancer program, overall, fewer than half of all patients with surgically resectable, nonmetastatic gastric cancer had 15 or more lymph nodes removed between 2004 and 2010 in California. In the present report, 45.5% of recognized cancer programs met the 15 or more lymph node–count benchmark, and only 34.8% of treating facilities without a recognized cancer program met the benchmark. Our findings are in keeping with those of previous studies^10,17-19 evaluating the variation in lymph node retrieval by hospital type, with a range or 23.2% to 44.1% demonstrating removal of 15 or more lymph nodes. Hospital factors, such as type of CoC accreditation, NCI designation, hospital surgical volume, teaching status, and geographic location, have been associated with the number of lymph nodes retrieved.

An increasing likelihood of compliance with the 15 or more lymph node–count benchmark was associated with younger age, Asian/other or non-Hispanic black race/ethnicity, increasing stage at diagnosis, higher hospital gastrectomy volume, and more recent year of diagnosis. The association of younger age, increasing stage, and higher hospital volume with greater lymph node retrieval is expected and corroborates published reports.^10,18,20 The association of increased lymph node retrieval in nonwhite patients has also been noted^18,20 previously and attributed to the care of patients of minority race/ethnicity in higher volume hospitals or tertiary-care institutions with surgical specialists or host factors such as obesity. The Hawaii Cancer Registry^10,18 has repeatedly shown the highest compliance (56% of patients) with the benchmark of 15 or more lymph nodes retrieved. According to the US Census Bureau, 57.4% of the population in Hawaii self-report as Asian alone or as Asian in combination with another race/ethnicity compared with 5.6% of the overall US population. The higher proportion of Asians in Hawaii could affect regional differences in the lymph node yield in SEER reporting.²¹ Our data set included the most current available data from the CCR, and it is notable that our final year for analysis, 2010, was the first year in which more than 50% of the cases met the 15 or more lymph node–count benchmark.

Although cancer program recognition by the CoC or NCI was significantly associated with higher compliance with the 15 or more lymph node–count quality measure, treatment at a facility with a recognized cancer program was not associated with long-term survival outcomes. Our Cox proportional hazards regression model showed that survival after gastrectomy was predicted, independently, by retrieval of 15 or more lymph nodes, younger age, Asian/other race/ethnicity, higher hospital volume, and, as expected, a lower stage at diagnosis. Birkmeyer et al²² found that, although operative mortality was lower at NCI-designated institutions, overall survival was not affected by cancer center designation. Gastrectomy at an NCI-designated facility showed a modest survival benefit only after adjustment for patient characteristics and residual hospital volume differences (hazard ratio, 0.91; 95% CI, 0.83-0.99). Hospital specialization did not affect survival outcomes in a study²³ from the Netherlands that found 5-year survival after gastrectomy to be equivalent among teaching universities, teaching institutions other than universities, and nonteaching hospitals.

Treatment in a facility with a recognized cancer program is purported to reflect greater surgical specialization, multidisciplinary care, access to novel therapy and clinical trials, and standardized survivorship protocols to monitor for early recurrence. Although CCR data provide information about processes of care that have been linked to outcomes, such as lymph node counts, details regarding the receipt of multimodal adjuvant therapy in the outpatient setting after surgery may be reported less robustly. Using National Cancer Database data, Dudeja et al¹⁹ showed that, among CoC-accredited facilities, receipt of recommended multimodality therapy after gastrectomy occurred in less than one-third of the patients. The absence of a survival benefit among patients receiving gastrectomies in hospitals with a recognized cancer program is often proposed to be the result of a tendency for patients having more comorbid conditions to undergo the operation in specialized facilities. We found no significant difference in nodal positivity rates among hospitals with and those without cancer program recognition. To evaluate whether programs with cancer center recognition were more likely to care for patients with higher T-stage cancer that could facilitate node retrieval, we replaced the AJCC stage variable with a T-stage variable. Findings for variables other than staging were nearly identical using the 2 staging schemes, although the AJCC staging variable was a more robust predictor of compliance with the 15 or more lymph node–count guideline and mortality hazards than was the T-stage variable (eTable in the Supplement). Other studies^22,24 have shown no differences in Charlson comorbidity scores, Atlas scores, and severity of illness based on a facility’s NCI designation or CoC accreditation. Information on comorbidities is not available in the CCR data, preventing further evaluation of this hypothesis.

Our report has certain limitations. We did not adjust for adjuvant chemotherapy or radiotherapy as independent variables. Survival differences may result from selection bias in adjuvant treatment recommendations. Another study²⁵ evaluated data on neoadjuvant and adjuvant chemotherapy for gastric cancer in the CCR and found that less than 10% of the cases eligible for chemotherapy had sufficient data available for analysis; thus, incorporating a chemotherapy variable would diminish the size of the study population. It is reasonable to surmise that discovery of sufficient and representative counts of lymph nodes in surgical specimens and the quality of slide review by pathology services may be associated with the quality of other services in the same hospitals. The effect of and interactions between these quality measures, which could arguably be expected to affect 3-month postgastrectomy survival, were not assessed in the present study, but lend support to the use of the 15 or more lymph node–count quality measure as a surrogate marker for the overall quality of gastric cancer care at the hospital level. This study did not formally evaluate differences in the counts of lymph nodes obtained from D1 vs D2 lymph node dissections, since the site of lymph node retrieval is not reported in the CCR. Additional limitations of the present study include the use of AJCC sixth edition staging rather than seventh edition staging, owing to the reporting criteria during this time period. However, several studies^26-28 have shown that AJCC seventh edition staging does not improve survival prognostication for gastric cancer over sixth edition staging. Finally, our assumption that institutions previously, but not currently, accredited by the CoC (eg, Kaiser-Permanente hospitals) have maintained some of the CoC-mandated quality standards may not be valid because of funding or organizational challenges.

We found that only 42% of the gastrectomies performed in California hospitals during the study period met the proposed minimum CoC lymph node–count guideline. The survival benefit from removal of 15 or more lymph nodes during gastrectomy for cancer, which is independent of an institution’s cancer program status, suggests that compliance with the current CoC minimum lymph node–count guideline can benefit all patients receiving cancer-directed gastrectomies. Identification of 15 or more lymph nodes in a gastrectomy specimen may be a global marker for quality of care. The survival benefit attributed to resection of 15 or more lymph nodes could result from improved adjuvant multimodal therapy augmented by better staging that was possible with larger and more representative lymph node counts owing to higher quality surgery or pathology services, regardless of the cancer program designation of the hospital where the surgery was performed. Collectively, these findings suggest the opportunity for further improvement of gastric cancer outcomes through better compliance with the CoC minimum lymph node–count guideline.

Accepted for Publication: April 14, 2014.

Corresponding Author: Sharon S. Lum, MD, Department of Surgery, Division of Surgical Oncology, School of Medicine, Loma Linda University, 11175 Campus St, CP 21111, Loma Linda, CA 92350 (slum@llu.edu).

Published Online: November 26, 2014. doi:10.1001/jamasurg.2014.678.

Author Contributions: Drs Morgan and Lum had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Morgan, Friedman, Senthil, Lum.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Morgan, Friedman, Senthil, Lum.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Morgan, Ji, Friedman, Lum.

Obtained funding: Morgan.

Administrative, technical, or material support: Morgan, Ji, Dyke, Lum.

Study supervision: Morgan, Senthil, Lum.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported in part by National Cancer Institute/Surveillance, Epidemiology, and End Results contracts N01-PC-35136, N01-PC-35139, and N02-PC-15105; Centers for Disease Control and Prevention/National Program for Cancer Registries contract U58DP000807-01; and the California Department of Public Health, Cancer Surveillance Branch.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Previous Presentation: The study was presented at the Pacific Coast Surgical Association Meeting; February 15, 2014; Dana Point, California.