Clinicopathological Characteristics, Survival Outcomes, and Genetic Alterations of Younger Patient With Gastric Cancer: Results From China National Cancer Center and cBioPortal

Background Survival outcomes of younger patients with gastric cancer (GC) remains controversial. The present study sought to explore clinicopathological characteristics, survival outcomes, and genetic alterations in the younger and older patients with GC. Patients with GC were identied from the China National Cancer Center Gastric Cancer Database (NCCGCDB) during 1998–2018. Survival analysis was conducted via Kaplan-Meier estimates and Cox proportional hazards models. Sequencing data were enrolled from the China National Cancer Center, TCGA, and MSKCC databases. A total of 1146 younger and 16988 older cases were included. Patients in the younger group were predominant in poor differentiation (53.7% versus 33.8%, P<0.0001), and pTNM stage IV (19.5% versus 11.8%, P<0.001). The 5-year overall survival (OS) of patients in NCCGCDB has noticeable increased from 1998 to 2018. Younger age was an independent prognostic factor for GC patients in pTNM stage III (P=0.014), while other stages showed no difference. Compared to the younger, older patients had a higher mutation frequency in LRP1B, GNAS, APC, KMT2D genes (all P< 0.05). In addition, although no signicant difference, results of the China National Cancer Center, TCGA, and MSKCC cohorts indicated that younger patients dominated in CDH1, RHOA, and CTNNB1 mutations. Stable proportion of younger cases and noticeable survival improvements were reported in the China National Cancer Center. Younger patients with pTNM stage III had a worse survival than older, while other tumor stages showed no difference. Furthermore, distinctive molecular characteristics were identied in younger GC patients, which might partly explain histopathological behaviors and prognosis of in this subpopulation. next-generation sequencing; single lipoprotein cyclic

DNA extraction and sequencing DNA from tumor samples and peripheral blood was extracted via the QIAamp DNA Mini Kit (QIAGEN, Valencia, CA) according to the manufacturer's instructions. QubitdsDNA assay was performed to measure the DNA concentration. A minimum of 50 ng of DNA is required for construction of an NGS library.
For GC patients with available DNA, targeted DNA sequencing was performed. DNA was pro led by using a capture-based targeted sequencing panel (Burning Rock Biotech, Guangzhou, People's Republic of China), targeting 520 genes and spanning 1.67M of Human genomic regions.

Sequencing data analysis
Sequencing data were mapped to the human genome (hg19) using BWA aligner 0.7.10. Local alignment optimization and variant calling were performed using GATK v3.2-2, while DNA translocation analysis was performed using both Tophat2 and Factera 1.4.3. Different mutation calling thresholds were applied on samples with different DNA qualities to avoid false-positive mutation calls due to DNA damage. Single nucleotide variants (SNVs) and insertion-deletions (INDELs) indels identi ed were annotated using the dbNSFP(v30a), COSMIC (v69), and dbSNP (snp138) database. Variants with a global minor allele frequency greater than 1.0% in 1000Genome Project (Phase3, http://www.1000genomes. org/data) were considered as common SNPs and removed. Tumor mutational burden (TMB) was de ned as the number of somatic, coding, base substation, and indels per megabase of genome examined. Synonymous mutations were counted to reduce sampling noise. White blood cells were used to lter germline mutations.

Statistical Analysis
The bar chart was plotted to evaluate the variation in tumor stage for younger and older patients from 1998 to 2018. The Chi-squared test was performed to compare categorical variables between the two groups, while continuous variables were evaluated using Student's t-test. OS and PFS for the younger and older group were calculated with the Kaplan-Meier method, while the log-rank test estimated the relevant survival discrepancy. Associations between risk factors and OS were investigated by univariate and multivariate Cox proportional hazard regression analysis, while the corresponding hazard ratio (HR) and 95% CI were generated. The covariates included in the nal models were determined by using stepwise selection with minimized AIC. The signi cant level for adding variables was 0.05, and the signi cance level for removing variables was 0.10. Statistical signi cance was set at two-sided P values less than 0.05, and all analyses in the study were conducted using SAS software v9.4 (SAS Institute, Inc., Cary, NC).

Demographic and Clinicopathological features
From 1998 to 2018, a total of 1146 younger cases (6.3%) and 16988 older cases (93.7%) were nally included. Among all GC patients, 2035 were diagnosed in period 1, 3859 in period 2, 6054 in period 3, and 6150 in period 4 (As shown in Table 1). There were signi cant differences in the distribution of gender, smoking, alcohol history, BMI, primary tumor location, differentiation, pTNM stage between the younger group and older group (all P < 0.01). Distinctive demographic disparities among age groups were founded in the China National Cancer Center. Younger patients were predominant in females (50.1% versus 21.6%, P < 0.0001). Conversely, relatively higher percentages of smokers (21.1% versus 42.4%, P < 0.0001), alcohol drinkers (20.9% versus 34.5%, P < 0.0001), and overweight/obesity (BMI ≥ 23) (36.1% versus 53.4%, P < 0.0001) were revealed in older. Table 1 Demographic and clinicopathological characteristics between younger group and older group in 4 consecutive periods (bidirectional cohort 1998-2018 Trends of clinicopathological features of younger patients with GC over the past 20 years were investigated. The median proportion of patients in the younger group over time was 6.2% (range 5.5-8.2%). No trend for the proportion of young patients from period 2 to period 4 was noted on linear regression (P = 0.053).

OS and PFS for younger patients with GC
Changing trends of OS and PFS between younger and older groups were shown in Table 2 Table 2 The 5-year overall and progression-free survival rates by cancer stages (bidirectional cohort 1998-2018).    Figure 3 revealed the Kaplan-Meier curves for OS between the younger and older groups (Fig. 3). The analysis showed younger patients had a worse survival outcome in stage III (P = 0.0095) but a better prognosis in stage I (P = 0.03). However, no signi cant difference was found in stage II and stage IV (P = 0.60, P = 0.37, resp.).

Prognostic Factors in Univariate and Multivariate Analyses
To investigate the signi cant factors impacting survival outcomes in young patients with GC, univariate and multivariate analyses were performed. The results revealed signi cantly different survival of GC based on the following parameters (  The nal model was builded by using stepwise selection with minimized AIC,and the covariates included in the nal models were selected by the stepwise selection method, with a signifcant level for adding variables of 0.05 and a signi cant level of removing variables of 0.10. In the univariate analysis, older patients with GC had better survival outcomes than the younger (HR = 0.80, 95% CI: 0.71-0.90, P = 0.0001). After strati cation by pTNM stage, compared to the older, younger group showed a better prognosis in stage I (P = 0.04) but worse in stage III (P < 0.01) (Supplementary Table 2). However, the multivariate analysis demonstrated that younger age was not an independent factor for poor survival outcomes (P = 0.20). In subdivided pTNM stages, a comparison presented that younger group was associated with worse survival outcomes in tumor stage III (P = 0.014), but no statistical signi cance in stages I, II, and IV (P = 0.074, 0.59 and 0.76, resp.).

Discussion
Our study provided a comprehensive analysis of younger GC patients at clinical and molecular levels. A primary nding was the stable proportion and signi cant survival improvements of younger cases in the China National Cancer Center from 1988 to 2018. Younger age was proven to be an independent prognostic factor for GC patients in pTNM stage III, while patients in stage I, stage II, and stage IV had similar survival among different age groups. Compare to the younger, a higher mutation frequency of LRP1B, KMT2D, APC, and GNAS genes in older patients was demonstrated in the current sequencing analysis, which might partly explain the histopathological behaviors and prognosis between the young and old groups.
Consistent with the previous study [10-15, 23, 24], our ndings revealed unique clinical features in younger patients, including female dominance, higher proportions of poor differentiation, and advanced tumor stage. Although the mechanisms for female predominance in younger patients were not clear, some studies indicated that hormonal factors might account for the sex difference [25,26]. Apart from H. pylori infection [27] and some gene distinctions [28], the vague symptoms and delayed diagnosis were also considerable reasons for the advanced stage of younger patients [29,30].
With an increase of 41.6% for the younger group and 45.9% for the older group, a signi cant 5-year survival improvement of GC was revealed in our center, which was in line with the increasing trend of pTNM stages I and II, surgery, and multimodality treatment. In China, survival trends of GC could be attributed to an improvement of quality of clinical services, including the improved access to primary healthcare, greater availability of diagnostic facilities, as well as enhanced effectiveness of the treatment [4]. Until 2015, the screening and early detection programs including GC have expanded to 31 provinces [31].
Encouragingly, the detection rate of GC has increased steadily bene t from the large-scale application of upper gastrointestinal endoscopy [32]. In addition, the widespread decision-making by the multi-disciplinary team (MDT) [33] and the emergence of individually multimodal therapies [34] were also closely related to a better prognosis for GC.
To date, survival outcomes of younger patients were still controversial. Previous studies suggested younger patients had a worse survival [14,15], while several analyses showed on difference [12,13]. Current data demonstrated a worse prognosis for younger patients in pTNM stage III, whereas other stages showed similar survival. Therefore, variations of prognosis between the younger and older group might be related to stage distributions. Compare to older patients, the higher proportion of multimodality treatment in younger re ected a better tolerance for GC, while more advanced stage and aggressive behaviors might be related to a worse survival in this subpopulation [14]. Besides, the reduced organ functions and rising risk of comorbidities also potentially effect GC prognosis that occurred with age [35]. In addition, although genomic pro les of younger patients were not yet fully ascertained, several studies demonstrated a strong association between certain gene mutations and the prognosis of GC among age groups [16,17].
The present study observed the differences in the prevalence of several gene mutations between younger and older patients, indicating a speci c mutational spectrum in younger patients with GC. Signi cantly, a higher prevalence of LRP1B mutations were identi ed in old patients. LRP1B encoded a unique lowdensity lipoprotein receptor (LDLR) that functioned in the binding and internalization of ligands [36]. Until now, the concrete functions of LRP1B in carcinogenesis were controversial [37][38][39][40]. Considered as a tumor suppressor, several ndings indicated the impaired LRP1B expression could promote the proliferation, migration, and invasion of carcinoma cells [38,39]. However, the latest update of the Network of Cancer Genes (NCG 6.0) listed LRP1B as a potentially false-positive tumor suppressor [39]. Zhou et al. [40] also demonstrated LRP1B was one of the recurrently mutated driver genes, which was strongly associated with the development of GC. Further analysis was required to investigate the pathobiological property of LRP1B in GC.
We also observed that older patients had a higher likelihood of GNAS mutations than younger. GNAS was situated on chromosome 20q13.3 and encoded alpha subunit of the stimulatory G-protein [41]. It was suggested that mutations of GNAS caused the activation of Wnt and ERK1/2 MAPK pathways via autonomous synthesis of cyclic adenosine monophosphate (cAMP), thus promoting intestinal tumorigenesis [42]. After that, several studies demonstrated the frequent GNAS mutations in gastric adenocarcinoma [43,44]. Consistent to current study, Esser et al. [41] also indicated the expression of GNAS was more prevalent in well-and moderately-differentiated GC, which was potentially correlated with older patients. Although the underlying mechanism was currently unclear, GNAS mutation might potentially affect the progression of GC via the protein kinase A (PKA), MAPK, and Wnt signaling pathways [45].
As in the past reports [46, 47], our analysis showed older patients had a higher prevalence of APC mutation than younger. Long considered a canonical tumor suppressor, APC could inhibit the excessive proliferation of tumor cells via regulating Wnt signaling pathway [48]. Mutation of APC was common in GC, particularly in well-differentiated adenocarcinoma [49,50]. Zhan et al. [51] proposed a strong synergy between APC alterations and MEK inhibitors in enhancing the signaling output of the Wnt cascades. Gerner et al. [52] also observed mutant APC was associated with the elevation of nitric oxide synthase 2 and the dysregulation of polyamine metabolism. These data suggested that APC mutation might contribute to the oncogenesis of GC. In addition, a recent study also demonstrated a signi cant association target high expression of APC and adverse prognosis of GC patients [53]. The observed prognostic alteration could be the result of the disruption of Wnt cascades, metabolic dysregulation, or some other unknown mechanisms.
KMT2D was among the most frequently mutated genes in some types of cancer including GC [54][55][56][57] , as KMT2D could activate some speci c pro-apoptotic genes or repress certain genes related to cell growth and survival pathways. The above results revealed that the KMT2D gene had different functions and biological effects depending on the type of cancer. Furthermore, several studies indicated that mutations of KMT2D were substantially higher in older patients than in younger [61, 62], and the age disparity was also demonstrated in our analysis. More experimental studies were necessary to elucidate the underlying functions of KMT2D in GC among age groups.
Our study has several strengths. First, clinicopathological characteristics and survival trends of the younger patients from 1998 to 2018 were comprehensively described based on NCCGCDB, which might serve as a reference for a large population-based study in China. Second, our study, for the rst time, suggested distinctive mutation disparities including LRP1B, GNAS, APC, and KMT2D genes in younger and older patients, which might partly explain the progression and survival of GC among age groups in molecular levels. The major limitation of our study was the insu cient sequencing sample size, particularly for pTNM subtype analysis. As such, possibility might not be ruled out for some of the signi cant ndings. Based on the TCGA and MSKCC, we further analyze a largervolume sequencing sample to compare genetic alterations among age groups. In addition, regional and racial disparity of younger patients from China and western counties might also exist, which could affect the clinicopathological and molecular property of GC potentially. Therefore, parts of ndings from TCGA and MSKCC might not be strongly generalizable to Chinese patients. Finally, some variables including tumor markers, neoadjuvant therapy, and adjuvant therapy were also needed to compare between the younger and older patients.

Consent for publication
Informed consent was obtained from all participants for publication.

Availability of data and materials
All data generated during this study are included in this published article and its supplementary les.

Competing interests
No potential con icts of interest were disclosed.