Introduction

Gastric cancer is an aggressive malignancy that carries a poor prognosis despite treatment advances, with a 5-year survival rate of 31% [1]. Its incidence continues to decline in the United States, with a decrease of 1.7% for men and 0.8% for women annually from 1992 to 2010 [2]. However, the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute estimates that there will be 28,000 new cases and 10,960 deaths from gastric cancer in 2017. Gastric cancer typically afflicts older adults, with a median age at diagnosis of 68 years in the United States; >95% of all new cases are diagnosed in patients > 40 years of age [3].

Though young adults are less commonly affected by gastric cancer, there have been mixed reports about the prognosis of younger patients from studies around the world. Some have suggested that young patients more often have diffuse-type and signet-ring histologies, higher stage disease, more frequent nodal involvement, and higher postoperative mortality [4,5,6,7,8,9,10,11]; and others have explored the idea of gastric cancer in the young as a different clinical entity, raising questions for the role of differential management [12, 13]. Most studies are reports from single institutions or use the Surveillance, Epidemiology, and End Results (SEER) database, which lacks detailed treatment data unless combined with data from the Medicare-linked database. The Medicare data, however, lack information on younger patients. In addition, while key trials in resectable [14,15,16,17] and metastatic [18,19,20,21,22] and gastric adenocarcinoma have included patients under the age of 40, they have not reported specifically on their presentation or survival. Among young adults with gastric adenocarcinoma, the geographic and ethnic distributions, receipt of treatment, and outcomes have not been fully characterized.

Young adult patients with gastric cancer face unique challenges, including biologic variation in tumors, differences in treatment effectiveness, tolerance, and adherence, issues related to fertility preservation, and psychosocial considerations associated with the early death [4, 23]. Some variation exists in the specific cutoffs used to define young adult patients, but large groups including the National Cancer Institute [24] and the Adolescent and Young Adult Oncology Progress Review Group (AYAO PRG) [25] have used age 39 as the upper limit to define young adults. In addition to studies examining gastric cancer in young adults [26, 27], survivorship studies across cancer types have similarly used 39 as the upper limit to define young adults [28, 29]. The AYAO PRG notes that patients < 40 years old have had limited improvement in survival, thus warranting further study [25].

The National Cancer Database (NCDB) is maintained as a joint project of the American College of Surgeons Commission on Cancer (CoC) and the American Cancer Society, and captures registry data that include approximately 70% of newly diagnosed cancer cases nationwide each year, compared to just 28% of all cancer cases in the SEER database [30]. The NCDB provides data on a variety of clinical and demographic patient data as well as initial management and survival. Studies examining patterns of care and survival outcomes in young patients using the NCDB have been conducted for several other areas, including lung, testicular, and breast cancer [31,32,33]. Here, we report our descriptive analysis of the NCDB with respect to differences across patient, tumor, and treatment characteristics between younger and older adults with gastric adenocarcinoma to elucidate factors affecting outcomes.

Materials and methods

Patient selection

The Institutional Review Board at our institution deemed analysis of the NCDB Participant User File to have exempt status. Adults initially diagnosed between 2004 and 2013 with International Classification of Diseases for Oncology, 3rd edition (ICD-O-3) topographical codes C16.0-C16.6, C16.8, or C16.9 were defined as having gastric cancer. Patients were identified ICD-O-3 morphological codes for adenocarcinoma histology; other histologies were excluded. Only patients with invasive gastric adenocarcinoma without prior history of malignancy were included. Patients with cancers originating at other sites (e.g., gastroesophageal junction) and extending into the stomach were excluded. Patients with missing clinical staging, treatment, follow-up, or vital status information were excluded. Inclusion and exclusion criteria are summarized in the CONSORT diagram in Fig. 1.

Fig. 1
figure 1

CONSORT diagram of inclusion criteria and age category

Full size image

Data elements

Adult patients were divided into two age groups: < 40 years (young adult) and ≥ 40 years. The rationale for young adult group was to remain consistent with the National Cancer Institute definition and with the previous studies examining considerations uniquely related to survivorship in young adults. Baseline patient characteristics were compared across age groups. Surgery was defined as any surgical procedure, including local excision/ablation, gastrectomy, or surgical procedure not otherwise specified. Radiation treatment was defined as any external beam radiation administered to the esophagus, stomach, abdominal site not otherwise specified, and/or lymph nodes. Chemotherapy was defined as any type and number of agents as part of the first treatment course. Patients were divided into several treatment groups: surgery only, chemotherapy alone, surgery + chemotherapy, surgery + chemotherapy + radiation, chemotherapy + radiation, no treatment, and other. Any sequence or timing of these combinations was considered in these mutually exclusive groups (see Table 1).

Table 1 Patient characteristics
Full size table

Statistical analysis

Chi-squared tests were used to compare baseline characteristics between the two age groups. Least-squares regression was used to identify a potential trend towards a changing proportion of young adult patients over time. Survival analysis was performed using the Kaplan–Meier method and log-rank test. A descriptive nomogram was constructed using the rms package (https://cran.r-project.org/web/packages/rms/rms.pdf). An accelerated failure time survival model was fitted and the overall survival time was assumed to follow a Weibull distribution. All covariates for which > 80% of data were available were selected based on clinical and/or statistical significance. Included covariates were sex, race, Charlson–Deyo Comorbidity score, grade, treatment, and clinical stage. All covariates were treated as discrete variables. All statistical analyses were performed using Stata Version 13.0 (StataCorp, College Station, TX, USA) or R statistical software version 3.4.3 (R Development Core Team, 2017). Hypothesis testing was two-sided and a p value of ≤ 0.05 was used to indicate significance for all comparisons with Bonferroni corrections applied for multiple comparison.

Results

Patient population

A total of 70,084 patients meeting inclusion criteria were identified, of which 2615 (4%) were aged < 40 and 67,469 (96%) were aged ≥ 40. The median proportion of young adults over time was 3.7% (range 3.7–4.0%); no trend for proportion of young adults over time was noted on linear regression (p = 0.655). Baseline patient characteristics are presented in Table 1. Young adults were more likely than adults aged ≥ 40 to be female, non-white, Hispanic, residing in a metropolitan area, treated in a northeastern US state, or treated at a facility in the top decile of gastric cancer patient volume. They were also more likely to present with more advanced, distal tumors with poor risk features, have a Charlson–Deyo comorbidity score of 0, and have either Medicaid, private, or no insurance. Among patients with stage IV disease, young adults were more likely to present with distant organ ± lymph node metastases as compared to older patients. Young adults were also significantly more likely to present with bone metastases as compared to older patients. Young adults with Stage IV disease were equally likely as adults ≥ 40 years old (20 vs. 21%, p = 0.664) to have organ metastases in 2 + sites.

Treatment differences

Stage-specific treatments separated by age group are shown in Fig. 2. Despite higher stage disease, younger patients were more likely to receive treatment than adults ≥ 40 years (87 vs. 81%, p < 0.001). For stage I disease, fewer young adult patients received surgery alone as compared to older patients (47 vs. 58%, p < 0.001). For stage II disease, young adults more commonly received surgery + chemotherapy + radiotherapy (RT) than older patients (51 vs. 34%, p < 0.001). For stage III patients, young adults also more commonly received surgery + chemotherapy + RT as compared to older patients (48 vs. 34%, p < 0.001). For stage IV disease, young adults were more likely than older patients (55 vs. 38%, p < 0.001) to receive chemotherapy alone. Among stage IV disease patients, a similar proportion of patients received multimodal treatment (21 vs. 19%, p = 0.02).

Fig. 2
figure 2

Stage-specific treatment across age strata. S surgery, C chemotherapy, RT radiotherapy, tx treatment

Full size image

Overall survival

Five-year overall survival across all stages for adults aged < 40 and those aged ≥ 40 was 21.1% [95% confidence interval (CI), 19.2–23.0%] and 22.1% (95% CI, 21.8–22.5%), respectively. Five-year stage-specific survival by age group is shown in Fig. 3. For young patients with stage I disease, median survival was not reached at last follow-up. Median survival for younger patients with stage II, III, and IV disease was 43.9 [95% confidence interval (CI), 34.7–66.5], 25.7 (95% CI, 21.7–31.0), and 7.9 (95% CI, 7.5–8.5) months, respectively. For adults > 40 years, the median survival for stage I, II, III, and IV disease was 61.5 (95% CI, 58.7–64.3), 29.9 (95% CI, 28.8–31.1), 16.9 (95% CI, 16.6–17.3), and 5.7 (95% CI, 5.6–5.8) months. On univariate log-rank analysis, young adults had significantly better 5-year OS than older patients for stages II, III, and IV disease (p < 0.001). When comparing stage IV patients receiving multimodal therapy on log-rank analysis, young adults showed no difference as compared to older patients (p = 0.56).

Fig. 3
figure 3

Kaplan–Meier curves displaying 5-year overall survival for each age group, separated by clinical stage. Estimates for 5-year overall survival and log-rank survival comparisons are noted on each graph. Significance was assessed at the 0.05 level

Full size image

Nomogram for young adult patients

The nomogram identified clinical stage as the largest contributor to overall survival, followed by treatment, grade, race, Charlson–Deyo comorbidity score, and sex. Each category within these variables was assigned a score on the point scale. The 5-year overall survival probability can be determined by computing the sum of scores, locating it on the total point scale, and drawing a straight line down on the survival scale (Fig. 4). The corresponding accelerated failure time model data is reported in Table 2.

Fig. 4
figure 4

Nomogram for overall survival for young adults with gastric adenocarcinoma

Full size image
Table 2 Accelerated failure time model results for young adult patients with gastric adenocarcinoma
Full size table

Discussion

This hospital-based analysis capturing 70% of all cancers diagnosed in the United States demonstrated that young adult patients with gastric adenocarcinoma are more likely to present with metastatic, high-grade disease. To our knowledge, this study is one of the first to look specifically at young adult patients in the era of modern trials after 2004 in localized and metastatic disease that have shaped gastric cancer treatment. The existing literature in this area is largely limited to older retrospective studies with small sample sizes, univariate analysis, significant confounding, and lack of comparison to older patients. A small number of larger analyses have been conducted but have limited or no data from the last 10 years [4, 10, 34].

Several findings regarding patient presentation in this hospital-based analysis are consistent with past reports, including higher rates of female sex, non-white race, signet-ring or diffuse histology, advanced clinical stage disease, higher grade, advanced T-stage, and nodal involvement [4,5,6,7,8,9,10,11,12,13]. Among young adult patients, findings that have not been previously reported including a larger proportion of patients treated in metropolitan areas, in the northeast, at facilities in the top decile of gastric cancer patient volume, and with private, Medicaid, or no insurance and higher frequency of bone metastases. To date, there is no clear explanation for the larger proportion of young adult females with gastric cancer, though some studies have hypothesized that hormonal factors such as a higher percentage of estrogen receptor-positive cells may be responsible [34,35,36]. A higher frequency of bone metastases in young adult patients may also be explained by estrogen receptor positivity, which has been shown to be associated with osseous metastasis in tumors at other sites [37, 38]. Markers of more advanced, aggressive disease in young adults have been reported by several series. While the NCDB does not capture patient history prior to the initial presentation, other series have reported shorter symptom duration, delays in diagnosis, and hereditary factors may be responsible for presentation with advanced disease [6, 12, 34, 39]. Despite an overall decrease in the incidence of gastric cancer in the United States, the proportion of young adults with gastric cancer in this analysis has remained around 4% with no obvious trend between 2004 and 2013.

Young adults as a group displayed similar survival, with a 5-year OS of 21% as compared to 22% for older patients, possibly due to a greater percentage of young patients presenting with metastatic, high-grade disease. Among stage IV patients, young adults were more likely to present with organ involvement, suggestive of a potentially greater burden of disease. When stage-specific survival was examined, young adults appeared to do as well as or better than older patients.

Within each disease stage, choice of treatment differed substantially across age groups. For stage I disease, the most commonly chosen treatment for each age group was surgery only, but young adult patients were more likely to additionally receive chemotherapy or RT. Young adults with stage II and III disease were more likely to receive surgery + chemotherapy + RT than older patients. For stage IV disease, younger patients were less likely to receive no treatment than older patients. A trend towards more aggressive treatment in younger patients may reflect a perceived ability for patients to tolerate treatment given younger age and fewer comorbidities. Nonetheless, more aggressive treatment with trimodality therapy was not used more often in patients < 40 years of age, suggesting that age alone may not change physicians’ perceived optimal treatment. Studies have indicated that younger patients are more likely to suffer postoperative mortality, though these data are not captured in the NCDB [5, 12]. Although staging in gastric adenocarcinoma [40] serves to guide treatment and prognosis, these results demonstrate that other factors, including age, may strongly impact how clinicians choose therapies for patients.

The nomogram for young adult patients showed stage, treatment, and grade to be most predictive of 5-year overall survival, whereas Charlson–Deyo comorbidity score, race, and sex were found to be less predictive. The demonstrated effect of treatment is consistent with other gastric cancer studies including the MAGIC trial [14], which showed improved OS for patients receiving perioperative chemotherapy, Intergroup Study 0116 [15], which showed improved OS for patients receiving adjuvant chemoradiotherapy, and RTOG 9904 [41], which showed an improved pathologic complete response rate with preoperative chemoradiation. The nomogram findings support the use of multimodality therapy in young adult patients as appropriate, particularly since these patients are more likely to present without significant comorbidities or impairment of functional status.

This study has several limitations typical of large database analysis. The patient demographic attributes in the NCDB are limited; there may be confounders within the analysis that are unable to be accounted for. There is no recorded information on cancer outcomes beyond OS, including locoregional control, distant control, salvage treatment, cancer-specific survival, or quality of life. The database also lacks metrics of response to treatment, types of systemic therapy received, and molecular data. Toxicities, such as malabsorption, dumping syndrome, cytopenias, ototoxicity, and secondary malignancies, are known to have a profound impact on patients but are not recorded in the NCDB. While the age categories chosen in this study reflect divisions studied in the previous studies, they are inherently arbitrary, and there may be differences in outcomes for smaller age subgroups. Several attributes in the NCDB, such as histology, grade, and sites of metastatic involvement are incompletely coded, which also limits our analysis. Therapeutic approaches have also evolved between 2004 and 2013, the timeframe of this study; D2 surgery, pancreas and spleen preserving gastrectomy, minimally invasive surgery, perioperative chemoradiation, and targeted therapy have been increasingly offered during these years [42]. Though the NCDB offers detailed information on treatment, the impact of these changes in treatment approach cannot be granularly assessed in the present study.

Despite these limitations, this study is one of the largest hospital-based analyses in the United States examining the presentation, treatment, and outcomes of young gastric adenocarcinoma patients. While these data do not provide any information on the benefit of surveillance, there appears to be a strong advantage to the early diagnosis in younger patients. Educating primary care physicians and patients about the early signs of gastric cancer might be beneficial in reducing mortality, particularly among Asian/Pacific Islander and Hispanic patients with a family history or known genetic predisposition. Given the paucity of randomized evidence for younger patients with gastric adenocarcinoma, large database analyses like this represent the best resource to potentially answer clinical questions surrounding younger patients.