Expanding the Lauren Classification: A New Gastric Cancer Subtype?

Abstract

See “Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil,” by Lei Z, Tan IB, Das K, et al, on page 554.Adenocarcinoma of the stomach remains a major public health issue. Although the incidence of this cancer in the United States has been falling, with only 21,600 new cases expected in 2013,1Siegel R. Naishadham D. Jemal A. Cancer statistics, 2013.CA Cancer J Clin. 2013; 63: 11-30Crossref PubMed Scopus (11432) Google Scholar the prognosis for patients remains grim. Despite improvements in treatment approaches, 5-year survival remains at <30%.1Siegel R. Naishadham D. Jemal A. Cancer statistics, 2013.CA Cancer J Clin. 2013; 63: 11-30Crossref PubMed Scopus (11432) Google Scholar In addition to the typically advanced stage of diagnosis for most gastric adenocarcinomas, these dismal survival statistics reflect the lack of effective treatment options. Identifying specific signaling pathways in individual patients might improve treatment outcomes, but only limited data are available. In this issue of Gastroenterology, Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar describe an approach to overcoming this obstacle. Their results identify 3 molecular signatures of gastric adenocarcinoma (Table 1). Patient survival data indicate that one of these subtypes may best be treated using 5-fluorouracil. In vitro data also suggest that another subtype may be particularly sensitive to phosphatidyl-inositol-3-kinase inhibitors.Table 1Gastric Adenocarcinoma ClassificationsLauren (1965)Lei et al. (2013)DiffuseIntestinal typeMesenchymalProliferativeMetabolicIntestinal type morphology0%aClassification based on criteria of Lauren (1965).11100%aClassification based on criteria of Lauren (1965).1130%aClassification based on criteria of Lauren (1965).11(7%)aClassification based on criteria of Lauren (1965).1174%aClassification based on criteria of Lauren (1965).11(71%)bClassification based on criteria of Tan et al. (2011).754%aClassification based on criteria of Lauren (1965).11(84%)bClassification based on criteria of Tan et al. (2011).7Diffuse morphology100%aClassification based on criteria of Lauren (1965).110%aClassification based on criteria of Lauren (1965).1159%aClassification based on criteria of Lauren (1965).11(93%)bClassification based on criteria of Tan et al. (2011).717%aClassification based on criteria of Lauren (1965).11(29%)bClassification based on criteria of Tan et al. (2011).741%aClassification based on criteria of Lauren (1965).11(16%)bClassification based on criteria of Tan et al. (2011).7Intestinal metaplasia55%91%Chronic gastritis45%88%Copy number alterationLowHighAmplified genesCCNE1, MYC, ERBB2, KRASAberrant methylationHypermethylationHypomethylationTP53 mutationsLowHighLowa Classification based on criteria of Lauren (1965).11Lauren P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type carcinoma. An attempt at a histo-clinical classification.Acta Pathol Microbiol Scand. 1965; 64: 31-49Crossref PubMed Scopus (4973) Google Scholarb Classification based on criteria of Tan et al. (2011).7Tan I.B. Ivanova T. Lim K.H. et al.Intrinsic subtypes of gastric cancer, based on gene expression pattern, predict survival and respond differently to chemotherapy.Gastroenterology. 2011; 141: 476-485Abstract Full Text Full Text PDF PubMed Scopus (261) Google Scholar Open table in a new tab This study builds on previous mRNA analyses of gastric cancers by this group and others.3Cho J.Y. Lim J.Y. Cheong J.H. et al.Gene expression signature-based prognostic risk score in gastric cancer.Clin Cancer Res. 2011; 17: 1850-1857Crossref PubMed Scopus (246) Google Scholar, 4Zouridis H. Deng N. Ivanova T. et al.Methylation subtypes and large-scale epigenetic alterations in gastric cancer.Sci Transl Med. 2012; 4: 156ra140Crossref PubMed Scopus (131) Google Scholar, 5Mueller A. Bachmann E. Linnig M. et al.Selective PI3K inhibition by BKM120 and BEZ235 alone or in combination with chemotherapy in wild-type and mutated human gastrointestinal cancer cell lines.Cancer Chemother Pharmacol. 2012; 69: 1601-1615Crossref PubMed Scopus (62) Google Scholar, 6Deng N. Goh L.K. Wang H. et al.A comprehensive survey of genomic alterations in gastric cancer reveals systematic patterns of molecular exclusivity and co-occurrence among distinct therapeutic targets.Gut. 2012; 61: 673-684Crossref PubMed Scopus (493) Google Scholar, 7Tan I.B. Ivanova T. Lim K.H. et al.Intrinsic subtypes of gastric cancer, based on gene expression pattern, predict survival and respond differently to chemotherapy.Gastroenterology. 2011; 141: 476-485Abstract Full Text Full Text PDF PubMed Scopus (261) Google Scholar, 8Lee H.J. Nam K.T. Park H.S. et al.Gene expression profiling of metaplastic lineages identifies CDH17 as a prognostic marker in early stage gastric cancer.Gastroenterology. 2010; 139: 213-225.e3Abstract Full Text Full Text PDF PubMed Scopus (115) Google Scholar, 9Ooi C.H. Ivanova T. Wu J. et al.Oncogenic pathway combinations predict clinical prognosis in gastric cancer.PLoS Genet. 2009; 5: e1000676Crossref PubMed Scopus (294) Google Scholar, 10Suzuki K. Suzuki I. Leodolter A. et al.Global DNA demethylation in gastrointestinal cancer is age dependent and precedes genomic damage.Cancer Cell. 2006; 9: 199-207Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar By combining 192 previously reported9Ooi C.H. Ivanova T. Wu J. et al.Oncogenic pathway combinations predict clinical prognosis in gastric cancer.PLoS Genet. 2009; 5: e1000676Crossref PubMed Scopus (294) Google Scholar and 56 new microarray expression profiles, Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar created a database describing 248 Singaporean primary gastric cancers. These were broken into 3 unique groups using hierarchical clustering with iterative feature selection. The 201 gastric cancers that best represented the 3 groups were used to develop classification algorithms. The algorithms were then validated using a separate set of 70 Australian primary gastric cancers. The data show that at least for gastric cancers arising in Singapore and Australia, subclassification into the 3 groups is reproducible.To better understand the biological significance of the distinct groups, Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar examined the differentially expressed genes. The clusters were designated as mesenchymal, to reflect expression of genes in the epithelial-mesenchymal transition pathway; proliferative, because growth promoting oncogenic pathways were activated; and metabolic, owing to expression of genes associated with metabolic pathways.Further analyses of each group showed that mesenchymal tumors lost expression of E-cadherin, the epithelial cadherin isoform. Consistent with the frequent loss of E-cadherin expression in signet ring cell gastric cancers, nearly 60% of mesenchymal tumors were classified as Lauren's diffuse type (Figure 1A), and 70% had at least some features of diffuse gastric cancer.11Lauren P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type carcinoma. An attempt at a histo-clinical classification.Acta Pathol Microbiol Scand. 1965; 64: 31-49Crossref PubMed Scopus (4973) Google Scholar Further, consistent with reports that diffuse gastric cancers tend to be hypermethylated,12Oue N. Oshimo Y. Nakayama H. et al.DNA methylation of multiple genes in gastric carcinoma: association with histological type and CpG island methylator phenotype.Cancer Sci. 2003; 94: 901-905Crossref PubMed Scopus (92) Google Scholar mesenchymal tumors displayed genomic hypermethylation. Transforming growth factor-β, vascular endothelial growth factor, nuclear factor-κB, mammalian target of rapamycin, sonic hedgehog, and cancer stem cell pathways were also activated in mesenchymal gastric cancers. In contrast, p53 mutations and DNA copy number variations were limited.In contrast with mesenchymal gastric adenocarcinomas, nearly 75% of proliferative tumors were Lauren's intestinal type (Figure 1B). These tumors tended to have activation of E2F, MYC, and RAS pathways and mutations of CCNE1, MYC, ERBB2, and KRAS. Proliferative cancers were hypomethylated and often harbored p53 mutations. These characteristics suggest that the proliferative cluster overlaps with Lauren's intestinal type and that this group includes tumors responsive to therapies targeting epithelial growth factor receptor (EGFR), ERBB2, and growth factor or stem cell pathways.The real surprise of this paper is the presence of a third gastric adenocarcinoma cluster. Why did this study identify 3 groups when previous molecular analyses, including those by this same group using many of the same gastric cancer specimens, found only 2? One possibility is the hierarchical clustering analysis employed here. Although powerful, this approach requires the user to predefine the number of clusters in the data set. Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar determined the correct number of clusters empirically. Increasing the cluster number from 2 to 3 improved the statistical accuracy of subclassification, whereas further cluster number increases provided only minimal benefit. This led Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar to postulate that 3 subtypes of gastric cancer were present in the set. The differences between the 3 groups identified support this hypothesis.The mesenchymal and proliferative subtypes largely coincide with Lauren's diffuse and intestinal subtypes. So what is the third type of gastric cancer? Histologically, these tumors are nearly evenly split between diffuse and intestinal types. There are no characteristic gene amplifications, and p53 mutations are rare. It is therefore understandable that the so-called metabolic type tumors were lost within the diversity of diffuse and intestinal type tumors in previous analyses. However, it is notable that the metabolic tumor group includes features of spasmolytic polypeptide-expressing metaplasia.8Lee H.J. Nam K.T. Park H.S. et al.Gene expression profiling of metaplastic lineages identifies CDH17 as a prognostic marker in early stage gastric cancer.Gastroenterology. 2010; 139: 213-225.e3Abstract Full Text Full Text PDF PubMed Scopus (115) Google Scholar, 13Schmidt P.H. Lee J.R. Joshi V. et al.Identification of a metaplastic cell lineage associated with human gastric adenocarcinoma.Lab Invest. 1999; 79: 639-646PubMed Google ScholarArmed with this new information, Lei et al classified gastric cancer cell lines as mesenchymal, proliferative, or metabolic and tested their in vitro sensitivity to chemotherapeutic agents. Remarkably, metabolic tumors were highly sensitive to 5-fluorouracil. This seems to also be true in vivo, because 5-fluorouracil treatment was associated with 100% 5-year survival among Singaporean patients with metabolic type gastric cancers. Cancer-specific, disease-free survival was also improved and, after adjusting for TNM stage, overall survival was also enhanced by 5-fluorouracil treatment in Australian metabolic tumors. Lei et al also found that mesenchymal-type gastric cancer cell lines were more sensitive to phosphatidyl-inositol-3-kinase inhibitors in vitro, although in vivo patient data are not available.While these data are striking, one might question the dependence on retrospective review of pathology reports, which are recognized to be unreliable owing to variation in classification criteria over time. How could review of the actual histopathology have added to the hierarchical clustering analysis? Although long forgotten, it is worth noting that Lauren's landmark study was not simply a description of intestinal-type and diffuse gastric adenocarcinomas. This 1965 work was a detailed study that analyzed morphologic data in a manner similar to that in which molecular data are now used. Multiple tumor characteristics were collected and clusters were combined when they were found to overlap significantly. Ultimately, Lauren was left with 2 primary groups: Intestinal type and diffuse. In addition to the distinctive architecture and cytology, these subtypes also differed in that intestinal type cancers presented a decade later than diffuse adenocarcinomas. Lauren also found that intestinal-type tumors were associated with chronic gastritis and intestinal metaplasia in 90% of cases, whereas only half of diffuse cancers had these features. Thus, intestinal-type, or proliferative, cancers likely develop as a result of chronic gastritis and intestinal metaplasia, whereas diffuse, or mesenchymal, adenocarcinomas do not. The new molecular data further suggest that the metabolic subtype defined by Lei et al may originate within spasmolytic polypeptide-expressing metaplasia.13Schmidt P.H. Lee J.R. Joshi V. et al.Identification of a metaplastic cell lineage associated with human gastric adenocarcinoma.Lab Invest. 1999; 79: 639-646PubMed Google ScholarInclusion of classic histologic data, as well as degree of differentiation, mitotic rate, and presence of spasmolytic polypeptide-expressing metaplasia, may have greatly enriched Lei's analysis and minimized the redundancy apparent in the tendency of mesenchymal gastric cancers to be diffuse and of proliferative tumors to be intestinal type. Adding histopathology to the cluster analysis might have also provided tools for development of a simpler means to identify 5-fluorouracil–sensitive tumors. This approach has been successful in identifying subgroups of other tumors. For example, histologic separation of desmoplastic small cell tumors from a larger group of small, blue, round cell tumors made it possible to recognize that these lesions had a characteristic chromosomal translocation resulting in a EWS–WT1 gene fusion.14Gerald W.L. Rosai J. Ladanyi M. Characterization of the genomic breakpoint and chimeric transcripts in the EWS-WT1 gene fusion of desmoplastic small round cell tumor.Proc Natl Acad Sci U S A. 1995; 92: 1028-1032Crossref PubMed Scopus (257) Google Scholar, 15Rosai J. Why microscopy will remain a cornerstone of surgical pathology.Lab Invest. 2007; 87: 403-408Crossref PubMed Scopus (64) Google Scholar Detection of this fusion protein is now part of the standard diagnostic workup of small, blue, round cell tumors and is used to identify desmoplastic small cell tumors within that group. Similarly, histology and assay of a select set of genes and proteins might provide an effective means to subclassify gastric adenocarcinoma. The latter might include simple immunohistochemistry or molecular analyses of p53, MYC, EGFR, ERBB2/Her2, K-RAS, Ki67, mucin genes, stem cell markers, thymidylate synthase, and dihydropyrimidine dehydrogenase (low expression of thymidylate synthase and dihydropyrimidine dehydrogenase are associated with favorable 5-fluorouracil responses in colorectal cancer16Salonga D. Danenberg K.D. Johnson M. et al.Colorectal tumors responding to 5-fluorouracil have low gene expression levels of dihydropyrimidine dehydrogenase, thymidylate synthase, and thymidine phosphorylase.Clin Cancer Res. 2000; 6: 1322-1327PubMed Google Scholar). Notably, recent reports have used either a simple 6-gene panel or the combination of well-differentiated intestinal-type and microsatellite instability as markers of 5-fluorouracil–sensitive gastric adenocarcinomas.3Cho J.Y. Lim J.Y. Cheong J.H. et al.Gene expression signature-based prognostic risk score in gastric cancer.Clin Cancer Res. 2011; 17: 1850-1857Crossref PubMed Scopus (246) Google Scholar, 17An J.Y. Kim H. Cheong J.H. et al.Microsatellite instability in sporadic gastric cancer: its prognostic role and guidance for 5-FU based chemotherapy after R0 resection.Int J Cancer. 2012; 131: 505-511Crossref PubMed Scopus (109) Google ScholarOverall, the study by Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar and similar reports3Cho J.Y. Lim J.Y. Cheong J.H. et al.Gene expression signature-based prognostic risk score in gastric cancer.Clin Cancer Res. 2011; 17: 1850-1857Crossref PubMed Scopus (246) Google Scholar, 4Zouridis H. Deng N. Ivanova T. et al.Methylation subtypes and large-scale epigenetic alterations in gastric cancer.Sci Transl Med. 2012; 4: 156ra140Crossref PubMed Scopus (131) Google Scholar, 5Mueller A. Bachmann E. Linnig M. et al.Selective PI3K inhibition by BKM120 and BEZ235 alone or in combination with chemotherapy in wild-type and mutated human gastrointestinal cancer cell lines.Cancer Chemother Pharmacol. 2012; 69: 1601-1615Crossref PubMed Scopus (62) Google Scholar, 6Deng N. Goh L.K. Wang H. et al.A comprehensive survey of genomic alterations in gastric cancer reveals systematic patterns of molecular exclusivity and co-occurrence among distinct therapeutic targets.Gut. 2012; 61: 673-684Crossref PubMed Scopus (493) Google Scholar, 7Tan I.B. Ivanova T. Lim K.H. et al.Intrinsic subtypes of gastric cancer, based on gene expression pattern, predict survival and respond differently to chemotherapy.Gastroenterology. 2011; 141: 476-485Abstract Full Text Full Text PDF PubMed Scopus (261) Google Scholar, 8Lee H.J. Nam K.T. Park H.S. et al.Gene expression profiling of metaplastic lineages identifies CDH17 as a prognostic marker in early stage gastric cancer.Gastroenterology. 2010; 139: 213-225.e3Abstract Full Text Full Text PDF PubMed Scopus (115) Google Scholar, 9Ooi C.H. Ivanova T. Wu J. et al.Oncogenic pathway combinations predict clinical prognosis in gastric cancer.PLoS Genet. 2009; 5: e1000676Crossref PubMed Scopus (294) Google Scholar, 10Suzuki K. Suzuki I. Leodolter A. et al.Global DNA demethylation in gastrointestinal cancer is age dependent and precedes genomic damage.Cancer Cell. 2006; 9: 199-207Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar, 17An J.Y. Kim H. Cheong J.H. et al.Microsatellite instability in sporadic gastric cancer: its prognostic role and guidance for 5-FU based chemotherapy after R0 resection.Int J Cancer. 2012; 131: 505-511Crossref PubMed Scopus (109) Google Scholar represent new hope for personalized therapy of gastric adenocarcinoma. Finding ways to apply this information to identify tumor subsets and develop molecularly tailored, individualized therapies will require creative thinking in this era of evidence-based, cost-effective medicine. See “Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil,” by Lei Z, Tan IB, Das K, et al, on page 554. See “Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil,” by Lei Z, Tan IB, Das K, et al, on page 554. See “Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil,” by Lei Z, Tan IB, Das K, et al, on page 554. Adenocarcinoma of the stomach remains a major public health issue. Although the incidence of this cancer in the United States has been falling, with only 21,600 new cases expected in 2013,1Siegel R. Naishadham D. Jemal A. Cancer statistics, 2013.CA Cancer J Clin. 2013; 63: 11-30Crossref PubMed Scopus (11432) Google Scholar the prognosis for patients remains grim. Despite improvements in treatment approaches, 5-year survival remains at <30%.1Siegel R. Naishadham D. Jemal A. Cancer statistics, 2013.CA Cancer J Clin. 2013; 63: 11-30Crossref PubMed Scopus (11432) Google Scholar In addition to the typically advanced stage of diagnosis for most gastric adenocarcinomas, these dismal survival statistics reflect the lack of effective treatment options. Identifying specific signaling pathways in individual patients might improve treatment outcomes, but only limited data are available. In this issue of Gastroenterology, Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar describe an approach to overcoming this obstacle. Their results identify 3 molecular signatures of gastric adenocarcinoma (Table 1). Patient survival data indicate that one of these subtypes may best be treated using 5-fluorouracil. In vitro data also suggest that another subtype may be particularly sensitive to phosphatidyl-inositol-3-kinase inhibitors. This study builds on previous mRNA analyses of gastric cancers by this group and others.3Cho J.Y. Lim J.Y. Cheong J.H. et al.Gene expression signature-based prognostic risk score in gastric cancer.Clin Cancer Res. 2011; 17: 1850-1857Crossref PubMed Scopus (246) Google Scholar, 4Zouridis H. Deng N. Ivanova T. et al.Methylation subtypes and large-scale epigenetic alterations in gastric cancer.Sci Transl Med. 2012; 4: 156ra140Crossref PubMed Scopus (131) Google Scholar, 5Mueller A. Bachmann E. Linnig M. et al.Selective PI3K inhibition by BKM120 and BEZ235 alone or in combination with chemotherapy in wild-type and mutated human gastrointestinal cancer cell lines.Cancer Chemother Pharmacol. 2012; 69: 1601-1615Crossref PubMed Scopus (62) Google Scholar, 6Deng N. Goh L.K. Wang H. et al.A comprehensive survey of genomic alterations in gastric cancer reveals systematic patterns of molecular exclusivity and co-occurrence among distinct therapeutic targets.Gut. 2012; 61: 673-684Crossref PubMed Scopus (493) Google Scholar, 7Tan I.B. Ivanova T. Lim K.H. et al.Intrinsic subtypes of gastric cancer, based on gene expression pattern, predict survival and respond differently to chemotherapy.Gastroenterology. 2011; 141: 476-485Abstract Full Text Full Text PDF PubMed Scopus (261) Google Scholar, 8Lee H.J. Nam K.T. Park H.S. et al.Gene expression profiling of metaplastic lineages identifies CDH17 as a prognostic marker in early stage gastric cancer.Gastroenterology. 2010; 139: 213-225.e3Abstract Full Text Full Text PDF PubMed Scopus (115) Google Scholar, 9Ooi C.H. Ivanova T. Wu J. et al.Oncogenic pathway combinations predict clinical prognosis in gastric cancer.PLoS Genet. 2009; 5: e1000676Crossref PubMed Scopus (294) Google Scholar, 10Suzuki K. Suzuki I. Leodolter A. et al.Global DNA demethylation in gastrointestinal cancer is age dependent and precedes genomic damage.Cancer Cell. 2006; 9: 199-207Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar By combining 192 previously reported9Ooi C.H. Ivanova T. Wu J. et al.Oncogenic pathway combinations predict clinical prognosis in gastric cancer.PLoS Genet. 2009; 5: e1000676Crossref PubMed Scopus (294) Google Scholar and 56 new microarray expression profiles, Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar created a database describing 248 Singaporean primary gastric cancers. These were broken into 3 unique groups using hierarchical clustering with iterative feature selection. The 201 gastric cancers that best represented the 3 groups were used to develop classification algorithms. The algorithms were then validated using a separate set of 70 Australian primary gastric cancers. The data show that at least for gastric cancers arising in Singapore and Australia, subclassification into the 3 groups is reproducible. To better understand the biological significance of the distinct groups, Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar examined the differentially expressed genes. The clusters were designated as mesenchymal, to reflect expression of genes in the epithelial-mesenchymal transition pathway; proliferative, because growth promoting oncogenic pathways were activated; and metabolic, owing to expression of genes associated with metabolic pathways. Further analyses of each group showed that mesenchymal tumors lost expression of E-cadherin, the epithelial cadherin isoform. Consistent with the frequent loss of E-cadherin expression in signet ring cell gastric cancers, nearly 60% of mesenchymal tumors were classified as Lauren's diffuse type (Figure 1A), and 70% had at least some features of diffuse gastric cancer.11Lauren P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal-type carcinoma. An attempt at a histo-clinical classification.Acta Pathol Microbiol Scand. 1965; 64: 31-49Crossref PubMed Scopus (4973) Google Scholar Further, consistent with reports that diffuse gastric cancers tend to be hypermethylated,12Oue N. Oshimo Y. Nakayama H. et al.DNA methylation of multiple genes in gastric carcinoma: association with histological type and CpG island methylator phenotype.Cancer Sci. 2003; 94: 901-905Crossref PubMed Scopus (92) Google Scholar mesenchymal tumors displayed genomic hypermethylation. Transforming growth factor-β, vascular endothelial growth factor, nuclear factor-κB, mammalian target of rapamycin, sonic hedgehog, and cancer stem cell pathways were also activated in mesenchymal gastric cancers. In contrast, p53 mutations and DNA copy number variations were limited. In contrast with mesenchymal gastric adenocarcinomas, nearly 75% of proliferative tumors were Lauren's intestinal type (Figure 1B). These tumors tended to have activation of E2F, MYC, and RAS pathways and mutations of CCNE1, MYC, ERBB2, and KRAS. Proliferative cancers were hypomethylated and often harbored p53 mutations. These characteristics suggest that the proliferative cluster overlaps with Lauren's intestinal type and that this group includes tumors responsive to therapies targeting epithelial growth factor receptor (EGFR), ERBB2, and growth factor or stem cell pathways. The real surprise of this paper is the presence of a third gastric adenocarcinoma cluster. Why did this study identify 3 groups when previous molecular analyses, including those by this same group using many of the same gastric cancer specimens, found only 2? One possibility is the hierarchical clustering analysis employed here. Although powerful, this approach requires the user to predefine the number of clusters in the data set. Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar determined the correct number of clusters empirically. Increasing the cluster number from 2 to 3 improved the statistical accuracy of subclassification, whereas further cluster number increases provided only minimal benefit. This led Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar to postulate that 3 subtypes of gastric cancer were present in the set. The differences between the 3 groups identified support this hypothesis. The mesenchymal and proliferative subtypes largely coincide with Lauren's diffuse and intestinal subtypes. So what is the third type of gastric cancer? Histologically, these tumors are nearly evenly split between diffuse and intestinal types. There are no characteristic gene amplifications, and p53 mutations are rare. It is therefore understandable that the so-called metabolic type tumors were lost within the diversity of diffuse and intestinal type tumors in previous analyses. However, it is notable that the metabolic tumor group includes features of spasmolytic polypeptide-expressing metaplasia.8Lee H.J. Nam K.T. Park H.S. et al.Gene expression profiling of metaplastic lineages identifies CDH17 as a prognostic marker in early stage gastric cancer.Gastroenterology. 2010; 139: 213-225.e3Abstract Full Text Full Text PDF PubMed Scopus (115) Google Scholar, 13Schmidt P.H. Lee J.R. Joshi V. et al.Identification of a metaplastic cell lineage associated with human gastric adenocarcinoma.Lab Invest. 1999; 79: 639-646PubMed Google Scholar Armed with this new information, Lei et al classified gastric cancer cell lines as mesenchymal, proliferative, or metabolic and tested their in vitro sensitivity to chemotherapeutic agents. Remarkably, metabolic tumors were highly sensitive to 5-fluorouracil. This seems to also be true in vivo, because 5-fluorouracil treatment was associated with 100% 5-year survival among Singaporean patients with metabolic type gastric cancers. Cancer-specific, disease-free survival was also improved and, after adjusting for TNM stage, overall survival was also enhanced by 5-fluorouracil treatment in Australian metabolic tumors. Lei et al also found that mesenchymal-type gastric cancer cell lines were more sensitive to phosphatidyl-inositol-3-kinase inhibitors in vitro, although in vivo patient data are not available. While these data are striking, one might question the dependence on retrospective review of pathology reports, which are recognized to be unreliable owing to variation in classification criteria over time. How could review of the actual histopathology have added to the hierarchical clustering analysis? Although long forgotten, it is worth noting that Lauren's landmark study was not simply a description of intestinal-type and diffuse gastric adenocarcinomas. This 1965 work was a detailed study that analyzed morphologic data in a manner similar to that in which molecular data are now used. Multiple tumor characteristics were collected and clusters were combined when they were found to overlap significantly. Ultimately, Lauren was left with 2 primary groups: Intestinal type and diffuse. In addition to the distinctive architecture and cytology, these subtypes also differed in that intestinal type cancers presented a decade later than diffuse adenocarcinomas. Lauren also found that intestinal-type tumors were associated with chronic gastritis and intestinal metaplasia in 90% of cases, whereas only half of diffuse cancers had these features. Thus, intestinal-type, or proliferative, cancers likely develop as a result of chronic gastritis and intestinal metaplasia, whereas diffuse, or mesenchymal, adenocarcinomas do not. The new molecular data further suggest that the metabolic subtype defined by Lei et al may originate within spasmolytic polypeptide-expressing metaplasia.13Schmidt P.H. Lee J.R. Joshi V. et al.Identification of a metaplastic cell lineage associated with human gastric adenocarcinoma.Lab Invest. 1999; 79: 639-646PubMed Google Scholar Inclusion of classic histologic data, as well as degree of differentiation, mitotic rate, and presence of spasmolytic polypeptide-expressing metaplasia, may have greatly enriched Lei's analysis and minimized the redundancy apparent in the tendency of mesenchymal gastric cancers to be diffuse and of proliferative tumors to be intestinal type. Adding histopathology to the cluster analysis might have also provided tools for development of a simpler means to identify 5-fluorouracil–sensitive tumors. This approach has been successful in identifying subgroups of other tumors. For example, histologic separation of desmoplastic small cell tumors from a larger group of small, blue, round cell tumors made it possible to recognize that these lesions had a characteristic chromosomal translocation resulting in a EWS–WT1 gene fusion.14Gerald W.L. Rosai J. Ladanyi M. Characterization of the genomic breakpoint and chimeric transcripts in the EWS-WT1 gene fusion of desmoplastic small round cell tumor.Proc Natl Acad Sci U S A. 1995; 92: 1028-1032Crossref PubMed Scopus (257) Google Scholar, 15Rosai J. Why microscopy will remain a cornerstone of surgical pathology.Lab Invest. 2007; 87: 403-408Crossref PubMed Scopus (64) Google Scholar Detection of this fusion protein is now part of the standard diagnostic workup of small, blue, round cell tumors and is used to identify desmoplastic small cell tumors within that group. Similarly, histology and assay of a select set of genes and proteins might provide an effective means to subclassify gastric adenocarcinoma. The latter might include simple immunohistochemistry or molecular analyses of p53, MYC, EGFR, ERBB2/Her2, K-RAS, Ki67, mucin genes, stem cell markers, thymidylate synthase, and dihydropyrimidine dehydrogenase (low expression of thymidylate synthase and dihydropyrimidine dehydrogenase are associated with favorable 5-fluorouracil responses in colorectal cancer16Salonga D. Danenberg K.D. Johnson M. et al.Colorectal tumors responding to 5-fluorouracil have low gene expression levels of dihydropyrimidine dehydrogenase, thymidylate synthase, and thymidine phosphorylase.Clin Cancer Res. 2000; 6: 1322-1327PubMed Google Scholar). Notably, recent reports have used either a simple 6-gene panel or the combination of well-differentiated intestinal-type and microsatellite instability as markers of 5-fluorouracil–sensitive gastric adenocarcinomas.3Cho J.Y. Lim J.Y. Cheong J.H. et al.Gene expression signature-based prognostic risk score in gastric cancer.Clin Cancer Res. 2011; 17: 1850-1857Crossref PubMed Scopus (246) Google Scholar, 17An J.Y. Kim H. Cheong J.H. et al.Microsatellite instability in sporadic gastric cancer: its prognostic role and guidance for 5-FU based chemotherapy after R0 resection.Int J Cancer. 2012; 131: 505-511Crossref PubMed Scopus (109) Google Scholar Overall, the study by Lei et al2Lei Z. Tan I.B. Das K. et al.Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil.Gastroenterology. 2013; 145: 554-565Abstract Full Text Full Text PDF PubMed Scopus (304) Google Scholar and similar reports3Cho J.Y. Lim J.Y. Cheong J.H. et al.Gene expression signature-based prognostic risk score in gastric cancer.Clin Cancer Res. 2011; 17: 1850-1857Crossref PubMed Scopus (246) Google Scholar, 4Zouridis H. Deng N. Ivanova T. et al.Methylation subtypes and large-scale epigenetic alterations in gastric cancer.Sci Transl Med. 2012; 4: 156ra140Crossref PubMed Scopus (131) Google Scholar, 5Mueller A. Bachmann E. Linnig M. et al.Selective PI3K inhibition by BKM120 and BEZ235 alone or in combination with chemotherapy in wild-type and mutated human gastrointestinal cancer cell lines.Cancer Chemother Pharmacol. 2012; 69: 1601-1615Crossref PubMed Scopus (62) Google Scholar, 6Deng N. Goh L.K. Wang H. et al.A comprehensive survey of genomic alterations in gastric cancer reveals systematic patterns of molecular exclusivity and co-occurrence among distinct therapeutic targets.Gut. 2012; 61: 673-684Crossref PubMed Scopus (493) Google Scholar, 7Tan I.B. Ivanova T. Lim K.H. et al.Intrinsic subtypes of gastric cancer, based on gene expression pattern, predict survival and respond differently to chemotherapy.Gastroenterology. 2011; 141: 476-485Abstract Full Text Full Text PDF PubMed Scopus (261) Google Scholar, 8Lee H.J. Nam K.T. Park H.S. et al.Gene expression profiling of metaplastic lineages identifies CDH17 as a prognostic marker in early stage gastric cancer.Gastroenterology. 2010; 139: 213-225.e3Abstract Full Text Full Text PDF PubMed Scopus (115) Google Scholar, 9Ooi C.H. Ivanova T. Wu J. et al.Oncogenic pathway combinations predict clinical prognosis in gastric cancer.PLoS Genet. 2009; 5: e1000676Crossref PubMed Scopus (294) Google Scholar, 10Suzuki K. Suzuki I. Leodolter A. et al.Global DNA demethylation in gastrointestinal cancer is age dependent and precedes genomic damage.Cancer Cell. 2006; 9: 199-207Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar, 17An J.Y. Kim H. Cheong J.H. et al.Microsatellite instability in sporadic gastric cancer: its prognostic role and guidance for 5-FU based chemotherapy after R0 resection.Int J Cancer. 2012; 131: 505-511Crossref PubMed Scopus (109) Google Scholar represent new hope for personalized therapy of gastric adenocarcinoma. Finding ways to apply this information to identify tumor subsets and develop molecularly tailored, individualized therapies will require creative thinking in this era of evidence-based, cost-effective medicine. Identification of Molecular Subtypes of Gastric Cancer With Different Responses to PI3-Kinase Inhibitors and 5-FluorouracilGastroenterologyVol. 145Issue 3PreviewAlmost all gastric cancers are adenocarcinomas, which have considerable heterogeneity among patients. We sought to identify subtypes of gastric adenocarcinomas with particular biological properties and responses to chemotherapy and targeted agents. Full-Text PDF