| Background and aim:Colorectal neuroendocrine neoplasms(NENs)are a group of rare diseases with increasing incidence in recent decades,and G1 and G2 NENs constitute the majority of all colorectal NENs.Although G1 and G2 colorectal NENs are characterized with low malignant potential,slow growth,good differentiation and favorable prognosis,they are still somewhat invasive,and some patients can present nodal metastasis,even distant metastasis.However,at present,clinicians have insufficient understanding of their potential malignancy and aggressiveness,and the existing guidelines fail to reach consensus on their diagnosis and treatment strategies,which makes it hard for these patients to receive standardized diagnosis and treatment.Our study aimed to construct prediction models for nodal and distant metastasis in G1 and G2 colorectal NENs to provide a simple and effective reference tool for clinical diagnosis and treatment.Material and methods:A training set and validation set were obtained by including G1 and G2 colorectal NEN patients who had received resection of their primary tumors from the Surveillance,Epidemiology,and End Results(SEER)database and the National Cancer Center/Cancer Hospital Chinese Academy of Medical Science,respectively.We used the training set to develop nomograms for nodal and distant metastasis,respectively.And we used the validation set to perform external validation for the established nomograms.Univariate and multivariate logistic regression analysis were used to identify independent risk factors associated with the target outcomes,and the screened independent risk factors were incorporated into the final Logistic regression models to construct prediction models and draw nomograms.The receiver operating characteristic(ROC)curves were drawn and the area under the curves(AUC)were calculated to evaluate the discriminability.The calibration plots were drawn and the Hosmer-Lemeshow tests were performed to assess the calibration ability.The Brier scores were used to assess the overall performance of the models.Results:Our study included 3690 and 172 consecutive patients with G1 and G2 colorectal NENs who underwent surgical resection of primary tumors from the SEER database and the Cancer Hospital Chinese Academy of Medical Science as training set and validation set,respectively.Tumor size,tumor location and T stage were incorporated into the prediction model for nodal metastasis.For the training set,the AUC value was 0.972[95%confidence interval(CI):0.964-0.980],and the P value was 0.999 for Hosmer-Lemeshow test.For the validation set,the AUC value was 0.897(95%CI:0.846-0.948),the P value was 0.537 for Hosmer-Lemeshow test,which indicated that the model had satisfactory discriminability and calibration in predicting nodal metastasis.The age,tumor size,T stage and N stage were enrolled into the prediction model for distant metastasis.The AUC values were 0.942(95%CI:0.928-0.957)and 0.939(95%CI:0.890-0.987)in the training set and validation set,respectively.The P values for Hosmer-Lemeshow tests were 1 and 0.597 in the training set and validation set,respectively.This signified that the model had favorable discriminability and calibration in predicting distant metastasis.Conclusions:The prediction models have accepted discriminability and calibration in predicting nodal and distant metastasis in G1 and G2 colorectal NENs.They are simple and effective and can offer reliable reference for clinical practice.For clinical problems that are vague in current guidelines and can not get common consensus,our models can provide clear recommendations and help to standardize clinical diagnosis and treatment.Background and aim:G1 and G2 colorectal neuroendocrine neoplasms(NENs)are the most common colorectal NENs,and most of them are confined in the mucosa and submucosa,only a few of them can show metastatic disease.Owing to their low incidence,the relevant basic research has not been fully performed,which leads to the insufficient understanding of their genetic characteristics and metastatic mechanisms.Our study aimed to delineate their genetic characteristics and explore their underlying metastatic mechanisms.Material and methods:Our study collected specimens from patients with G1 and G2 colorectal NENs diagnosed at the National Cancer Center/Cancer Hospital Chinese Academy of Medical Sciences between November 2018 and December 2022 and used next-generation sequencing technology for targeted sequencing of 808 cancer-related genes.We delineated the somatic mutation characteristics,copy number variation(CNV),mutation spectrum,mutation signature,potential targetable gene alterations for targeted therapy and then we further compared the genetic characteristics between patient with metastatic disease and those without metastatic disease.We utilized KEGG enrichment analysis to explore the abnormal pathways of G1 and G2 colorectal NENs and study their potential metastatic mechanisms.Results:Our study collected 54 specimens of G1 and G2 colorectal NENs,including 23 cases with nodal or distant metastasis and 31 cases without metastatic disease.In the whole cohort,38(70.4%)patients were detected with somatic mutation and a total of 209 genes with 355 mutation sites were found,including 306(86.2%)missense mutations and 17(4.8%)frameshift mutations,etc.The common mutation genes were NCOR2(24.1%),BRD4(11.1%),MDC1(11.1%),ARID1A(9.3%),AXIN2(9.3%),etc.In the whole group,19(35.2%)patients were found with CNV and a total of 45 genes with 57 CNVs were observed.The common CNVs included amplification of HIST1H3D(5.6%),amplification of HIST1H3E(5.6%)and loss of PTEN(5.6%).The mutation spectrum analysis showed that C>T and T>C transitions were the major fractions of gene mutations,base transition was far more frequency than base transversion.The mutation signature analysis indicated that Signature 1 and Signature 17 were the main mutation signatures.The KEGG enrichment analysis revealed that PI3K-Akt(P<0.001),MAPK(P<0.001),and Rap1(P<0.001)were the major abnormal pathways.The survival analysis showed that KMT2D mutation was the independent risk factor of tumor progression.A total of 24 patients were detected with potential targetable gene alterations for targeted therapy and 26 potential targetable genes were screened.The common potential targetable genes were ARID1A(9.3%),CHEK2(7.4%),NF1(7.4%),TSC2(7.4%).There were significant different genetic features between patients with nodal or distant metastasis and patients without metastatic disease.The metastatic patients share only 47(22.5%)mutation genes,20(6.6%)mutation sites and 6(13.3%)CNVs with patients without metastatic disease.NCOR2(21.7%),BRD4(13%),CDKN1B(13%),CYP3A5(13%),EIF1AX(13%)were the common mutation genes in patients with metastatic disease,while NCOR2(25.8%),MDC1(19.4%),AXIN2(12.9%),PIK3C2G(12.9%),PTPRT(12.9%)were the common mutation genes in patients without metastatic disease.For patients with metastatic disease,the most common abnormal pathway was cell senescence and it was detected in 56.5%of patients.For patients without metastatic disease,the most common abnormal pathway was PI3K-AKT signal pathway and it was detected in 45.2%of patients.The patients with metastatic disease presented significant higher proportion of abnormal pathway of cell senescence(56.5%vs 25.8%,P=0.022)and lysine degradation(43.5%vs 16.1%,P=0.027)than patients without metastatic disease.Conclusions:G1 and G2 colorectal NENs presented significant different gene characteristics compared with colorectal adenocarcinomas and G3 neuroendocrine carcinomas.There might be a small tumor subgroup in G1 and G2 colorectal NEN,which might obtain stronger ability of invasiveness than most other G1 and G2 colorectal NEN through aberrant cell senescence and lysine degradation pathway,resulting in regional lymph node metastasis or distant metastasis. |
PDF Full Text Request