Tracking the molecular features of nonpolypoid colorectal neoplasms: a systematic review and meta-analysis.

OBJECTIVES: Nonpolypoid colorectal neoplasms (NP-CRNs) are proposed as a major contributor to the occurrence of interval cancers, but their underlying biology remains controversial. We conducted a systematic review and meta-analysis to clarify the major biological events in NP-CRNs. METHODS: We systematically searched for studies examining molecular characteristics of NP-CRNs. We performed random effect meta-analyses. We measured the heterogeneity among studies using I(2) and possible publication bias using funnel plots. RESULTS: Fifty-three studies on KRAS, APC, or BRAF mutations, microsatellite instability (MSI), CpG island methylator phenotype (CIMP), or DNA promoter hypermethylation were included. We observed less KRAS mutations (summary odds ratio (OR) 0.30, confidence interval (CI)=0.19-0.46, I(2)=77.4%, CI=70.1-82.9) and APC mutations (summary OR 0.42, CI=0.24-0.72, I(2)=22.6%, CI=0.0-66.7) in NP-CRNs vs. protruded CRNs, whereas BRAF mutations were more frequent (summary OR 2.20, CI=1.01-4.81, I(2)=0%, CI=0-70.8), albeit all with large heterogeneity. Less KRAS mutations were especially found in NP-CRNs subtypes: depressed CRNs (summary OR 0.12, CI=0.05-0.29, I(2)=0%, CI=0-67.6), non-granular lateral spreading tumors (LSTs-NG) (summary OR 0.61, CI=0.37-1.0, I(2)=0%, CI=0-74.6), and early nonpolypoid carcinomas (summary OR 0.11, CI=0.06-0.19, I(2)=0%, CI=0-58.3). MSI frequency was similar in NP-CRNs and protruded CRNs (summary OR 0.99, CI=0.21-4.71, I(2)=70.3%, CI=38.4-85.7). Data for promoter hypermethylation and CIMP were inconsistent, precluding meaningful conclusions. CONCLUSIONS: This meta-analysis provides indications that NP-CRNs are molecularly different from protruded CRNs. In particular, some subtypes of NP-CRNs, the depressed and LST-NG, are featured by less KRAS mutations than polypoid CRNs. Prospective, multicenter studies are needed to clarify the molecular pathways underlying nonpolypoid colorectal carcinogenesis and potential implications for surveillance intervals.