Survival Outcomes and Patterns of Care for Stage II or III Resected Gastric Cancer by Race and Ethnicity

Key Points Question Among patients with stage II or III gastric cancer who underwent a curative surgical procedure, are there outcome differences by race and ethnicity? Findings In this cohort study of 6938 patients with clinical stage II or III noncardia gastric adenocarcinoma, Asian and Hispanic patients had statistically significantly better overall survival (OS) compared with White patients. Black patients were observed to have similar OS with White patients overall, but Black patients did have statistically significantly better OS among those who received neoadjuvant therapy. Meaning Among patients with stage II or III gastric adenocarcinoma undergoing a curative surgical procedure, there were variations in OS associated with race and ethnicity.


Introduction
Gastric cancer is the fifth most common cancer in the world with more than 26 500 new diagnoses and 11 130 deaths annually in the United States. 1 Although the incidence has declined, some studies suggest that there is increasing incidence and mortality in younger patients, 2 and 5-year survival rate is still poor at 30%. 3 Surgical resection is potentially curative; however, variation in extent of lymph node dissection and type of perioperative chemotherapy and radiation therapy have led to numerous paradigms for therapeutic sequencing in advanced resectable gastric cancer.
The updated analysis of the Intergroup 0116 trial demonstrated a strong persistent overall survival benefit from adjuvant chemoradiation compared with surgery alone (hazard ratio [HR], 1.32 [95% CI, 1.10-1.60]). 4However, most patients who entered in this trial underwent less than a D2 lymphadenectomy, and many have argued that the chemoradiation compensated for inadequate node dissection.This argument was strengthened with the ARTIST 2 trial 5 indicating no benefit of adding radiotherapy to adjuvant chemotherapy after D2 lymphadenectomy in patients with stage II or III node-positive gastric cancer.Perioperative chemotherapy became an established treatment regimen based on the MAGIC trial, 6 which showed a significant improvement in 5-year overall and progression-free survival compared with surgery alone.This was followed by the FLOT4 trial, 7 which established perioperative FLOT (fluorouracil plus leucovorin, oxaliplatin, and docetaxel) as the new preoperative standard chemotherapy over MAGIC's ECF (fluorouracil or capecitabine plus cisplatin and epirubicin) regimen.These studies have changed our practice pattern for locally advanced gastric cancer over the past decade but variabilities in use of neoadjuvant or adjuvant treatment protocols have existed across the United States due to different rates of adoption of these studies.
Studies have also found significant disparities in gastric cancer outcomes across racial and ethnic groups in the United States.One study hypothesized that a substantial driver of disparity is the fact that Asian, Black, Hawaiian or Pacific Islander, and Hispanic patients are more likely than non-Hispanic White patients to develop noncardia, diffuse-type gastric cancer, which is associated with treatment resistance and worse outcomes. 8There are data to suggest population differences in molecular subtypes of gastric cancer according to race, affecting therapeutic response and prognosis. 9Minoritized racial and ethnic populations may also face systems-level differences in care leading to worse outcomes. 10,11Others have questioned whether response to neoadjuvant or adjuvant chemotherapy differs across racial and ethnic groups. 12,13Our current study was designed to review the recent patterns of care in resectable gastric cancer in the United States, compared with survival outcomes of patients by race and ethnicity, and analyze differences in outcomes after different multimodality treatment regimens stratified by race and ethnicity.

Data
This was a retrospective cohort study using the National Cancer Database (NCDB), which is a hospital-based registry of cancer outcomes produced by the American Cancer Society and the American College of Surgeon's Commission on Cancer (CoC).The deidentified database includes data collected from teaching hospitals, community cancer centers, and other cancer centers including Veterans Affairs hospitals located in 49 states and Puerto Rico.Only CoC-approved hospitals are

JAMA Network Open | Equity, Diversity, and Inclusion
Outcomes of Gastric Cancer by Race and Ethnicity allowed to report data to the NCDB and an estimated 70% of all new diagnoses of cancer in the United States are captured by the database.The NCDB Participant User Data File (PUF) is a Health Insurance Portability and Accountability Act-compliant data file containing cases submitted to the CoC's NCDB and waives informed consent process as the PUF only contains deidentified patient level data.This analysis used data available to CoC-accredited facilities that lacked personal identifiers and were exempt from institution review board approval owing to use of publicly available, deidentified data, in accordance with 45 CFR §46.We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Cohort
The NCDB for gastric cancer contained 255 165 patients.The NCDB was queried for patients treated in 2006 to 2019 with American Joint Committee on Cancer clinical stage IIA to IIIC gastric adenocarcinoma who underwent partial or total gastrectomy, excluding patients with cancers of the gastric cardia due to overlap with Siewert II and III malignant neoplasms of the gastroesophageal junction.For the purpose of this study, we excluded those with previous history of cancer, those who did not undergo gastrectomy, and those who had pathology different from adenocarcinoma.We also excluded those without information on sequencing of radiation, chemotherapy, and surgical resection procedure.Lastly, because we were interested in association with neoadjuvant or adjuvant therapy, we excluded those with stage 0 to I and stage IV gastric cancers as well as those with unknown clinical stage.This left us with 7545 patients, of which 6938 patients had complete follow-up with survival data.

Variables
Variables available for analyses included age, sex, race and ethnicity, Charlson Deyo score, primary site, histology, grade, stage, extent of lymph node dissection, number of lymph nodes positive, treatment facility type, distance from treatment facility to zip code, insurance status, treatment regimen and sequence, type of surgical procedure, and residual tumor resection classification.
Patient's race and ethnicity were determined from predefined NCDB data based on assignment by a CoC registrar according to fixed categories.For the race and ethnicity variable, we created 4 large categories: Asian, Black, Hispanic, and White.We grouped the Asian race category to include East Asian, South Asian, and Pacific Islander.A small portion of those who did not fall into the 4 categories of race and ethnicity were grouped as having other race and ethnicity.The multimodality treatment modality variable was categorized as neoadjuvant chemotherapy only, perioperative chemotherapy, neoadjuvant chemoradiation only, perioperative chemotherapy with radiation (either adjuvant or neoadjuvant), adjuvant chemotherapy only, adjuvant chemoradiation only, and surgical procedure only.In comparing neoadjuvant vs no neoadjuvant chemotherapy groups, we grouped the neoadjuvant chemotherapy group to contain any form of chemotherapy prior to surgical procedure (ie, neoadjuvant chemotherapy only, perioperative chemotherapy, neoadjuvant chemoradiation, perioperative chemotherapy with radiation).Among those who received neoadjuvant therapy, we compared the clinical stage and the final pathological stage, and then categorized response to neoadjuvant therapy.

Statistical Analysis
Univariate Cox regression analyses were performed to assess for variables that were associated with overall survival (OS).Variables with P < .05 on the univariate model and had more than 20 events per variable were included in the multivariate model.Median follow-up was estimated by the method of Schemper and Smith. 14
Black and White patients had significantly more comorbidities compared with Asian and Hispanic patients (Table 1).Higher proportions of Asian patients received treatment in academic tertiary centers (51.7% [541 of 1046]) with facilities being less than 10 miles from their home address (68.5% [717 of 1046]) compared with other racial and ethnic groups (eTable 1 in Supplement 1).In terms of insurance status, lower proportions of White patients were uninsured (  1).It should also be noted that even though certain minor differences between different racial and ethnic groups achieved statistical significance in the current sample size, they may not reflect clinical relevance.When looking at trends over time, the use of perioperative chemotherapy increased over time  2).
When looking at unadjusted OS of patients with clinical stage II or III cancer by race and ethnicity, we found that Asian and Hispanic patients had significantly better survival than White patients (P < .001for both Asian and Hispanic patients) (Figure 1).Furthermore, when comparing OS of these patients by receipt of some form of neoadjuvant chemotherapy in their treatment regimen sequence, we found that those who received neoadjuvant chemotherapy treatment did significantly better than those who did not receive neoadjuvant chemotherapy (eFigure in Supplement 1).In the multivariate analyses, perioperative chemotherapy was associated with improved OS (HR, 0.79 [95% CI, 0.69-0.90])(Table 3).In addition, after adjusting for sociodemographic and clinical covariates, including type of multimodality treatment regimen, race and ethnicity was a significant factor in OS.The adjusted OS was significantly higher in Asian patients with 36% lower risk of death compared with White patients (HR, 0.64 [95% CI, 0.58-0.72])(Table 3).The OS for Hispanic patients as compared with White patients also persisted with multivariate adjustment (HR, 0.77 [95% CI, 0.69-0.85]).Other variables associated with differences in OS included age, comorbidities, facility type, insurance status, histology, lymphovascular invasion, surgical margins, clinical stage, number of  3).
To explore whether there was also an association of race and ethnicity and neoadjuvant therapy with OS, we compared the OS of our cohort who received some form of neoadjuvant chemotherapy by race and ethnicity (Figure 2A) and those who did not receive any form of neoadjuvant chemotherapy by race and ethnicity (Figure 2B).Black patients who received neoadjuvant chemotherapy had OS that appeared similar to Hispanic patients (Figure 2A), whereas Black patients who did not receive neoadjuvant chemotherapy had OS falling to be similar to that of White patients (Figure 2B).Similarly, although Black patients had a similar HR with White patients when looking at the whole cohort (HR, 0.96 [95% CI, 0.88-1.04]),Black race was associated with improved OS when only looking at the group of patients who received neoadjuvant therapy (HR, 0.78 [95% CI, 0.67-0.90])(Table 3).Interestingly, Black patients had the highest proportion of being downstaged or achieving pathological complete response after neoadjuvant chemotherapy (34.4%) along with Asian patients (35.3%) when compared with White (28.4%) and Hispanic (30.8%) patients.

Discussion
In this study of 6938 patients in the NCDB with stage II or III gastric adenocarcinoma undergoing surgical resections, there are wide variations and statistically significant changes in the use of neoadjuvant and adjuvant therapy over time, and differential outcomes in treatment response and OS associated with race and ethnicity.Among all multimodality regimens, perioperative chemotherapy was associated with improved OS, whereas the number of positive lymph nodes and positive surgical margin were associated with the lowest OS.Among all patients in our cohort, Asian and Hispanic patients had improved OS compared with Black and White patients.Among those who received neoadjuvant therapy, Black patients were associated with higher OS compared with White patients.Asian and Black patients had higher proportions responding favorably to neoadjuvant therapy, with downstaging or pathological complete response, compared with White patients.
This retrospective cohort study of the NCDB found that race and ethnicity were independently associated with gastric cancer outcomes, even when adjusted for sociodemographic and clinical factors.Specifically, Asian and Hispanic patients had improved survival outcomes compared with Black and White patients.Other studies have reported similar results; an earlier NCDB study of 47 217 patients between 2000 to 2015 found that patients who self-identified as Asian had improved survival outcomes, followed by Black patients, then White patients. 15Various factors may drive this Perioperative chemotherapy with radiation The National Cancer Database lists races and ethnicities as other if they cannot be categorized as American Indian/Aleutian/Eskimo, Asian Indian, Black, Chamorran, Chinese, Fiji Islander, Filipino, Japanese, Guamanian, Hawaiian, Hmong, Kampuchean, Korean, Laotian, Melanesian, Micronesian, New Guinean, Other Asian, Pacific Islander, Pakistani, Polynesian, Samoan, Spanish/Hispanic, Tahitian, Thai, Tongan, Vietnamese, or White.
difference, such as differences in tumor response to chemotherapy or lymph node metastasis rate.
6][17] Our study, however, did not find that Asian and Hispanic patients had higher lymph node metastasis compared with Black and White patients with clinical stage II or III gastric cancer.Some studies have also shown that Asian patients were more likely to receive adequate lymphadenectomy. 18We also found that Asian and Hispanic patients had a higher number of regional lymph nodes examined.However, this may not be the sole driver of improved outcomes across different racial and ethnic groups.A study using the SEER database of more than 12 000 patients found that the survival benefit persisted after adjustment for age, gender, tumor grade, and number of examined and positive lymph nodes. 19Similarly, a study of more than 47 000 patients from the California Cancer Registry found that Asian and Hispanic race and ethnicity were strong independent factors of survival. 20Our study also found that race and ethnicity is an important independent factor of survival in multivariate analysis, with Asian patients having the lowest HR, aligned with the results of prior studies.
This study found that better survival outcomes were associated with patients with stage II or III gastric adenocarcinoma treated with neoadjuvant therapy, particularly perioperative chemotherapy, which has been found in a meta-analysis of randomized clinical trials. 21  insurance and low education level, which were both independently associated with lower rates of receiving neoadjuvant treatment. 11In the present study, although lack of insurance or Medicaid insurance was more common in Asian, Black, and Hispanic patient populations compared with White patients, we did not find disparities in receipt of multimodality treatment with White patients having the highest proportion of patients who underwent surgical resection only without multimodality treatment.This may have been due to White patients presenting at older age as we observed significantly higher proportions of those at least 65 years of age.Correspondingly, White patients also had the worst OS outcomes.
As it is still unclear what is most strongly contributing to race and ethnicity-based differences in survival outcomes, some have speculated that individuals may differ in degree of response to chemotherapy.Various genetic sequence variants have been correlated with increased or decreased response to neoadjuvant chemotherapy for gastric cancer, 13 and it is possible that these sequence variants have different frequencies across racial and ethnic groups.A small prospective study the from University of California San Diego found that Asian race was independently associated with increased likelihood to respond to neoadjuvant therapy and achieve a pathologic complete response when compared with Black or White patients. 12The authors suggested possible reasons such as variations in molecular subtype frequencies with differences in chemotherapy response and a difference in host immune response. 12In our study, the use of neoadjuvant treatment was found to be associated with better OS, and with some similarity to results from Rajabnejad et al, 12 the rate of downstaging or complete pathological response was significantly higher in the Asian and Black patient groups as compared with the White patient group.Interestingly, although Black patients overall had similar OS compared with White patients, Black patients had higher OS compared with White patients only among those who received neoadjuvant therapy.

Table 1 .
Sociodemographic and Clinical Variables by Race and Ethnicity for Clinical Stage II or III Gastric Cancer

Table 1 .
Sociodemographic and Clinical Variables by Race and Ethnicity for Clinical Stage II or III Gastric Cancer (continued) The National Cancer Database lists races and ethnicities as other if they cannot be categorized as American Indian/Aleutian/Eskimo, Asian Indian, Black, Chamorran, Chinese, Fiji Islander, Filipino, Japanese, Guamanian, Hawaiian, Hmong, Kampuchean, Korean, Laotian, Melanesian, Micronesian, New Guinean, Other Asian, Pacific Islander, Pakistani, Polynesian, Samoan, Spanish/Hispanic, Tahitian, Thai, Tongan, Vietnamese, or White. a

Table 2 .
Sociodemographic and Clinical Variables Over Time Among Patients With Clinical Stage II or III Gastric Cancer 11rtunately, we are seeing an increasing proportion of patients receiving some form of neoadjuvant therapy over time and declining use of postoperative chemoradiotherapy, indicating appropriate adoption of treatment standards established by clinical trials over time.However, disparities in receipt of multimodality therapy in the United States have been shown.Using the NCDB, Al-Refaie et al10found that between 1998 and 2006, when use of neoadjuvant chemotherapy for gastric adenocarcinoma was not standard practice, Black and Hispanic patients were less likely to receive multimodal therapy with adjuvant chemotherapy compared with White patients.10Morerecently,astudy of more than 16 000 patients using NCDB from 2006 to 2014 found that non-Hispanic White patients were more likely to receive neoadjuvant treatment than Asian or Pacific Islander, Black, or Hispanic patients.11Theyfound that race and ethnicity other than non-Hispanic White was also associated with no

Table 3 .
Multivariate Analysis of Hazard Ratios for Sociodemographic and Clinical Variables Outcomes of Gastric Cancer by Race and Ethnicity