LASSO Cox Regression Model Research Articles

Prognostic value of a gene signature in clear cell renal cell carcinoma.

Renal cancer is acommon urogenital system malignance. Novel biomarkers could provide more and more critical information ontumor features and patients' prognosis. Here, we performed an integrated analysis on the discovery set and established a three-gene signature to predict the prognosis for clear cell renal cell carcinoma (ccRCC). By constructing a LASSO Cox regression model, a 3-messenger RNA (3-mRNA) signature was identified. Based on the 3-mRNA signature, we divided patients into high- and low-risk groups, and validated thisby using three other data sets. In the discovery set, this signature could successfully distinguish between the high- and low-risk patients (hazard ratio (HR), 2.152; 95% confidence interval (CI),1.509-3.069; p < 0.0001). Analysis ofinternal and two external validation sets yielded consistent results (internal: HR, 2.824; 95% CI, 1.601-4.98; p < 0.001; GSE29609: HR, 3.002; 95% CI, 1.113-8.094; p = 0.031; E-MTAB-3267: HR, 2.357; 95% CI, 1.243-4.468; p = 0.006). Time-dependent receiver operating characteristic (ROC) analysis indicated that the area under the ROC curve at 5 years was 0.66 both in the discovery and internal validation set, while the two external validation sets also suggested good performance of the 3-mRNA signature. Besides that, a nomogram was built and the calibration plots and decision curve analysis indicated the good performance and clinical utility of the nomogram. In conclusion, this 3-mRNA classifier proved to be a useful tool for prognostic evaluation and could facilitate personalized management of ccRCC patients.

A Novel Messenger RNA Signature as a Biomarker for Predicting Early Relapse in Non-Small Cell Lung Cancer

Background: High throughput gene expression profiling has showed great promise in providing insight into molecular mechanisms. Recurrence-related mRNAs may potentially enrich genes with the ability to predict cancer recurrence and survival, therefore we attempted to build an early recurrence-associated gene signature to improve prognostic prediction of lung cancer. Methods: Propensity score matching was conducted between patients in early relapse group and long-term survival group from TCGA training series (N=579) and patients were matched 1:1. Global transcriptome analysis was then performed between the paired groups to identify tumor specific mRNAs. Finally, using LASSO Cox regression model, we built a multi-gene early relapse classifier incorporating forty mRNAs. The prognostic and predictive accuracy of the signature was internally validated in another 193 lung cancer patients. Results: Forty mRNAs were finally identified to build an early relapse classifier. With specific risk score formula, patients were classified into a high-risk group and a low-risk group. Relapse free survival was significantly different between the two groups in both discovery (HR: 3.126, 95% CI: 2.249-4.346, P<0.001) and internal validation (HR 1.806, 95% CI 1.077-3.030, P=0.025). Further analysis revealed that the prognostic value of this signature was independent of tumor stage, histotype and EGFR mutation (P<0.05). Receiver operating characteristic (ROC) analysis showed that the area under ROC curve of this signature was higher than TNM stage alone (0.771 vs 0.686, P<0.05). Conclusions: Our forty-mRNA-based classifier provides a reliable model for predicting early recurrence in non-small cell lung cancer after surgery. This model may facilitate personalized therapy-decision making for these patients. Funding Statement: This research was supported by the National Science Foundation of China (No. 81802374). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. Declaration of Interests: The authors have declared no conflicts of interests. Ethics Approval Statement: The study has been approved by the Institutional Review Boards of Huashan Hospital, Fudan University.

Radiomics signature of computed tomography imaging for prediction of survival and chemotherapeutic benefits in gastric cancer.

To develop and validate a radiomics signature for the prediction of gastric cancer (GC) survival and chemotherapeutic benefits. In this multicenter retrospective analysis, we analyzed the radiomics features of portal venous-phase computed tomography in 1591 consecutive patients. A radiomics signature was generated by using the Lasso-Cox regression model in 228 patients and validated in internal and external validation cohorts. Radiomics nomograms integrating the radiomics signature were constructed, demonstrating the incremental value of the radiomics signature to the traditional staging system for individualized survival estimation. The performance of the nomograms was assessed with respect to calibration, discrimination, and clinical usefulness. The radiomics signature consisted of 19 selected features and was significantly associated with DFS (disease-free survival) and OS (overall survival). Multivariate analysis demonstrated that the radiomics signature was an independent prognostic factor. Incorporating the radiomics signature into the radiomics-based nomograms resulted in better performance for the estimation of DFS and OS than the clinicopathological nomograms and TNM staging system, with improved accuracy of the classification of survival outcomes. Further analysis showed that stage II and III patients with higher radiomics scores exhibited a favorable response to chemotherapy. In conclusion, the newly developed radiomics signature is a powerful predictor of DFS and OS, and it may predict which patients with stage II and III GC benefit from chemotherapy.

Clinical Significance of Organic Anion Transporting Polypeptide Gene Expression in High-Grade Serous Ovarian Cancer.

High-grade serous ovarian cancer (HGSOC) is considered the most deadly and frequently occurring type of ovarian cancer and is associated with various molecular compositions and growth patterns. Evaluating the mRNA expression pattern of the organic anion transporters (OATPs) encoded by SLCO genes may allow for improved stratification of HGSOC patients for targeted invention. The expression of SLCO mRNA and genes coding for putative functionally related ABC-efflux pumps, enzymes, pregnane-X-receptor, ESR1 and ESR2 (coding for estrogen receptors ERα and ERß) and HER-2 were assessed using RT-qPCR. The expression levels were assessed in a cohort of 135 HGSOC patients to elucidate the independent impact of the expression pattern on the overall survival (OS). For identification of putative regulatory networks, Graphical Gaussian Models were constructed from the expression data with a tuning parameter K varying between meaningful borders (Pils et al., 2012; Auer et al., 2015, 2017; Kurman and Shih Ie, 2016; Karam et al., 2017; Labidi-Galy et al., 2017; Salomon-Perzynski et al., 2017; Sukhbaatar et al., 2017). The final value used (K = 4) was determined by maximizing the proportion of explained variation of the corresponding LASSO Cox regression model for OS. The following two networks of directly correlated genes were identified: (i) SLCO2B1 with ABCC3 implicated in estrogen homeostasis; and (ii) two ABC-efflux pumps in the immune regulation (ABCB2/ABCB3) with ABCC3 and HER-2. Combining LASSO Cox regression and univariate Cox regression analyses, SLCO5A1 coding for OATP5A1, an estrogen metabolite transporter located in the cytoplasm and plasma membranes of ovarian cancer cells, was identified as significant and independent prognostic factor for OS (HR = 0.68, CI 0.49–0.93; p = 0.031). Furthermore, results indicated the benefits of patients with high expression by adding 5.1% to the 12.8% of the proportion of explained variation (PEV) for clinicopathological parameters known for prognostic significance (FIGO stage, age and residual tumor after debulking). Additionally, overlap with previously described signatures that indicated a more favorable prognosis for ovarian cancer patients was shown for SLCO5A1, the network ABCB2/ABCB3/ABCC4/HER2 as well as ESR1. Furthermore, expression of SLCO2A1 and PGDH, which are important for PGE2 degradation, was associated with the non-miliary peritoneal tumor spreading. In conclusion, the present findings suggested that SLCOs and the related molecules identified as potential biomarkers in HGSOC may be useful for the development of novel therapeutic strategies.

Transcriptome profiling reveals an integrated mRNA-lncRNA signature with predictive value of early relapse in colon cancer.

The purpose of our study was to develop a multigene signature based on transcriptome profiles of both mRNAs and lncRNAs to identify a group of patients who are at high risk of early relapse in stages II-III colon cancer. Firstly, propensity score matching was conducted between patients in early relapse group and long-term survival group from GSE39582 training series (N = 359) and patients were matched 1:1. Global transcriptome analysis was then performed between the paired groups to identify tumor specific mRNAs and lncRNAs. Finally, using LASSO Cox regression model, we built a multigene early relapse classifier incorporating 15 mRNAs and three lncRNAs. The prognostic and predictive accuracy of the signature was internally validated in 102 colon cancer patients and externally validated in other 241 patients. In the training set, patients with high risk score were more likely to suffer from relapse than those with low risk score (HR: 2.67, 95% CI: 2.07-3.46, P < 0.001). The results were validated in the internal validation set (HR: 2.23, 95% CI: 1.23-3.78, P = 0.003) and external validation (HR 1.88, 95% CI 1.42-2.48; P < 0.001) set. Time-dependent receiver operating curve at 1 year showed that the integrated mRNA-lncRNA signature [area under curve (AUC) = 0.742] had better prognostic accuracy than AJCC TNM stage (AUC = 0.615) in the entire 702 patients. In addition, survival decision curve analyses at 12 months revealed a good clinical usefulness of the integrated mRNA-lncRNA signature. In conclusion, we successfully developed an integrated mRNA-lncRNA signature that can accurately predict early relapse.

A four-microRNA classifier as a novel prognostic marker for tumor recurrence in stage II colon cancer

About 20 percent of TNM-stage II colon cancer patients who are treated by surgical resection develop recurrence, and adjuvant chemotherapy in this group is still debated among researchers and clinicians. Currently, adverse histopathological and clinical factors are used to select patients for adjuvant chemotherapy following surgery. However, additional biomarkers to classify patients at risk of recurrence are needed. We have conducted a study using fresh frozen tumor tissue from 54 TNM-stage II colon cancer patients and performed microRNA profiling using next-generation sequencing. For the selection of the prognostic microRNAs, a LASSO Cox Regression model was employed. For the validation, we used the publically available TCGA-COAD cohort (n = 122). A prognostic panel of four micorRNAs (hsa-miR-5010-3p, hsa-miR-5100, hsa-miR-656-3p and hsa-miR-671-3p) was identified in the study cohort and validated in the TCGA-COAD cohort. The four-microRNA classifier successfully identified high-risk patients in the study cohort (P < 0.001) and the validation cohort (P = 0.005). Additionally, a number of established risk factors and the four-miRNA classifier were used to construct a nomogram to evaluate risk of recurrence. We identified a four-microRNA classifier in patients with TNM-stage II colon cancer that can be used to discriminate between patients at low- and high risk of recurrence.

Radiomic Signature of Computed Tomography Imaging for Prediction of Survival and Chemotherapeutic Benefits in Gastric Cancer

To develop and validate a radiomics signature for prediction of gastric cancer (GC) survival and chemotherapeutic benefits. In this multicenter retrospective analysis, we analyzed radiomics features of portal venous-phase computed tomography in 1,591 consecutive patients. A radiomics signature was generated by using the Lasso-Cox regression model in 228 patients and validated in internal and external validation cohorts. Radiomics nomograms integrating the radiomics signature were constructed, demonstrating the incremental value of the radiomics signature to the traditional staging system for individualized survival estimation. The performance of the nomograms was assessed with respect to calibration, discrimination, and clinical usefulness. The radiomics signature consisted of 19 selected features and was significantly associated with DFS and OS. Multivariate analysis demonstrated that the radiomics signature was an independent prognostic factor. Incorporating the radiomics signature into the radiomics-based nomograms resulted in better performance for the estimation of DFS than the clinicopathologic nomograms and the TNM staging system, with improved accuracy of the classification of survival outcomes. Further analysis showed stage II and III patients with higher radiomics scores exhibited a favorable response to chemotherapy. In conclusion, the newly developed radiomics signature was a powerful predictor of OS and DFS. Moreover, the radiomics signature could predict which patients with stage II and III GC benefit from chemotherapy. Funding Statement: This work was supported by grants from: National Natural Science Foundation of China 81672446, 81600510, 81370575, 81570593. Natural Science Foundation of Guangdong Province, 2014A030313131. Science and Technology Planning Project of Guangzhou, 2014B020228003, 2014B030301041, 2015A030312013. Science and Technology Program of Guangzhou, 158100076, 201400000001-3. Public welfare in Health Industry, National Health and Family Planning Commission of China (201402015, 201502039). Key Clinical Specialty Discipline Construction Program. Declaration of Interests: The author(s) indicated no potential conflicts of interest. Ethics Approval Statement: Ethical approval was obtained for this retrospective analysis at every participating center, and the informed consent requirement was waved.

ImmunoScore Signature: A Prognostic and Predictive Tool in Gastric Cancer.

We postulated that the ImmunoScore (IS) could markedly improve the prediction of postsurgical survival and chemotherapeutic benefits in gastric cancer (GC). A prediction model for GC patients was developed using data from 879 consecutive patients. The expression of 27 immune features was detected in 251 specimens by using immunohistochemistry, and a 5-feature-based ISGC was then constructed using the LASSO Cox regression model. Testing and validation cohorts were included to validate the model. Using the LASSO model, we established an ISGC classifier based on 5 features: CD3invasive margin (IM), CD3center of tumor (CT), CD8IM, CD45ROCT, and CD66bIM. Significant differences were found between the high-ISGC and low-ISGC patients in the training cohort in 5-year disease-free survival (45.0% vs. 4.4%, respectively; P <0.001) and 5-year overall survival (48.8% vs. 6.7%, respectively; P <0.001). Multivariate analysis revealed that the ISGC classifier was an independent prognostic factor. A combination of ISGC and tumor, node, and metastasis (TNM) had better prognostic value than TNM stage alone. Further analysis revealed that stage II and III GC patients with high-ISGC exhibited a favorable response to adjuvant chemotherapy. Finally, we constructed 2 nomograms to predict which patients with stages II and III GC might benefit from adjuvant chemotherapy after surgery. The ISGC classifier could effectively predict recurrence and survival of GC, and complemented the prognostic value of the TNM staging system. Moreover, the ISGC might be a useful predictive tool to identify stage II and III GC patients who would benefit from adjuvant chemotherapy.

A novel Immunoscore signature to predict survival in patients with gastric cancer: Implications for immunotherapy.

84 Background: Immunotherapies are transforming the treatment of gastric cancer (GC). Unfortunately, the majority of patients do not respond to immunotherapy. Thus, a novel risk score is needed to estimate tumor-infiltrating immune cells in predicting survival of GC and guide of more effective immunotherapy strategies. Methods: In total, 700 patients with GC from Gene Expression Ominus were sorted by training (490 patients) and validation (210 patients) cohorts.Immune risk score (IRS) signature based on the association between the expression of every immune cell and the duration of individual patients' survival was constructed using the LASSO Cox regression model. The relation between immunomodulatory genes, microsatellite instability and IRS value in GC was explored. Results: We established an IRS signature consisting of 11 types of immune cell fractions selected by the LASSO model, an IRS signature for GC was significantly different developed to classify patients into high- and low-risk groups in three cohorts. In the initial training cohort of patients, an IRS signature for GC was developed to classify patients into high- and low-risk groups in 8-year overall survival (hazard ratio [HR] = 2.93, P &lt; 0.0001). The performance of IRS was validated in the validation cohort (HR = 1.77, P = 0.011) and in the entire cohort (HR = 2.54, P &lt; 0.0001). In an ad-hoc analysis, we found the IRS value was substantially correlated with the expression of immunomodulatory genes including PD-1, PD-L1 and PD-L2, a prominent lower IRS was observed in microsatellite-instable (MSI) tumors compare with the microsatellite-stable (MSS) tumors. We developed nomograms for clinical use that integrated IRS signature and four clinicopathological risk factors was generated to predict which patients might have better survival with GC, and it performed well in the three cohorts (concordance index: 0.775, 0.682, and 0.752, respectively). Conclusions: These findings indicate that the assessment of the immune status via Immunoscore provides a potent predictor of survival in patients with GC and this resource may help facilitate the development of precision immunotherapy.

A novel messenger RNA signature as a prognostic biomarker for predicting relapse in pancreatic ductal adenocarcinoma.

Pancreatic ductal adenocarcinoma (PDAC) death rate and recurrence rate have remained obstinately high. Current methods can not satisfy the need of predicting cancer relapse accurately. Utilizing expression profiles of 10 GEO datasets (N = 774), we identified 154 differentially expressed genes (DEGs) between PDAC and normal pancreas tissue or paracancerous tissue. Next we built a 16-mRNA-based signature by means of the LASSO COX regression model. We also validated the prognostic value of the signature. Patients were classified into high-risk and low-risk group according to the signature risk score; 1 year RFS was 45% (95% CI: 31.6%–63.9%) for high-risk group in contrast to 92.5% (95% CI: 86.3%–99.1%) for low-risk group. Moreover, it could predict RFS well in cases with the receipt of different treatment modalities. The 16-mRNA-based signature was an independent and powerful prognostic biomarker for RFS for PDAC patients (HR = 7.74, 95% CI: 3.25–18.45, p < 0.0001).

Nomogram for Predicting Cardiovascular Mortality in Incident Peritoneal Dialysis Patients: An Observational Study

Cardiovascular mortality risk is high for peritoneal dialysis (PD) patients but it varies considerably among individuals. There is no clinical tool to predict cardiovascular mortality for PD patients yet. Therefore, we developed a cardiovascular mortality risk nomogram in a PD patient cohort. We derived and internally validated the nomogram in incident adult PD patients randomly assigned to a training (N = 918) or a validation (N = 460) dataset. The nomogram was built using the LASSO Cox regression model. Increasing age, history of cardiovascular disease or diabetes were consistent predictors of cardiovascular mortality. Low hemoglobin and serum albumin, high hypersensitive C-reactive protein and decreasing 24 hours urine output were identified as non-traditional cardiovascular risk predictors. In the validation dataset, the above nomogram performed good discrimination (1 year c-statistic = 0.83; 3 year c-statistic = 0.78) and calibration. This tool can classify patients between those at high risk of cardiovascular mortality (high-risk group) and those of low risk (low-risk group). Cardiovascular mortality was significantly different in the internal validation set of patients for the high-risk group compared to the low-risk group (HR 3.77, 2.14–6.64; p < 0.001). This novel nomogram can accurately predict cardiovascular mortality risk in incident PD patients.

Recurrence-associated gene signature optimizes recurrence-free survival prediction of colorectal cancer.

High throughput gene expression profiling has showed great promise in providing insight into molecular mechanisms. Metastasis‐related mRNAs may potentially enrich genes with the ability to predict cancer recurrence, therefore we attempted to build a recurrence‐associated gene signature to improve prognostic prediction of colorectal cancer (CRC). We identified 2848 differentially expressed mRNAs by analyzing CRC tissues with or without metastasis. For the selection of prognostic genes, a LASSO Cox regression model (least absolute shrinkage and selection operator method) was employed. Using this method, a 13‐mRNA signature was identified and then validated in two independent Gene Expression Omnibus cohorts. This classifier could successfully discriminate the high‐risk patients in discovery cohort [hazard ratio (HR) = 5.27, 95% confidence interval (CI) 2.30–12.08, P < 0.0001). Analysis in two independent cohorts yielded consistent results (GSE14333: HR = 4.55, 95% CI 2.18–9.508, P < 0.0001; GSE33113: HR = 3.26, 95% CI 2.16–9.16, P = 0.0176). Further analysis revealed that the prognostic value of this signature was independent of tumor stage, postoperative chemotherapy and somatic mutation. Receiver operating characteristic (ROC) analysis showed that the area under ROC curve of this signature was 0.8861 and 0.8157 in the discovery and validation cohort, respectively. A nomogram was constructed for clinicians, and did well in the calibration plots. Furthermore, this 13‐mRNA signature outperformed other known gene signatures, including oncotypeDX colon cancer assay. Single‐sample gene‐set enrichment analysis revealed that a group of pathways related to drug resistance, cancer metastasis and stemness were significantly enriched in the high‐risk patients. In conclusion, this 13‐mRNA signature may be a useful tool for prognostic evaluation and will facilitate personalized management of CRC patients.

Prognostic and predictive value of immunoscore signature in gastric cancer.

e15594 Background: TNM staging system is not adequate to define the prognosis of patients with gastric cancer (GC). This system is also unable to predict whether the GC patients are likely to benefit from adjuvant chemotherapy. We postulated that ImmunoScore of GC (ISGC) could markedly improve the prediction of postsurgical survival and adjuvant chemotherapeutic benefits. Methods: 125 GC patients were enrolled as a training cohort to detect the expression of 27 immune features using immunohistochemistry and then constructed a five-feature-based ISGC using the LASSO Cox regression model. Internal validation cohort (126 specimens) and two external validation cohorts (628 specimens) were utilized to validate the prognostic and predictive value of ISGC. Results: We established the ISGC classifier based on the five features: CD3invasive margin (IM), CD3center of tumor (CT), CD8IM, CD45ROCT, and CD66bIM. The ISGC classifier could distinguish GC patients with high-ISGC from those with low-ISGC with significant differences in 5-year disease-free survival (45.0% v.s. 4.4%, p &lt; 0.001) and 5-year overall survival (48.8% v.s. 6.7%, p &lt; 0.001). According to the multivariate analysis, the ISGC classifier was proved to be an independent prognostic factor. A combination of ISGC and TNM had better prognostic value than TNM stage alone. In a further analysis, stage II and III GC patients with high-ISGC exhibited a favorable response to adjuvant chemotherapy. To provide a quantitative method to predict stage II and III GC patients’ probability of 3- and 5-year overall survival, we constructed two nomograms that integrated the ISGC and clinicopathological risk factors. Calibration plots showed that the nomograms performed well compared with an ideal model. The predictive accuracy and clinical usefulness of the nomograms were also demonstrated. Conclusions: The ISGC classifier could effectively predict recurrence and survival of GC, and complemented the prognostic value to TNM system. Moreover, the classifier might be a useful predictive tool to identify candidates with stage II and III GC who would benefit from adjuvant chemotherapy. Therefore, the ISGC might facilitate the counseling and personalize the postoperative management of GC patients.

Abstract 4791: Prognostic model of lower grade gliomas

Abstract Background Lower grade gliomas (LGGs WHO grade II/III glioma) account for one third of all gliomas. Most LGGs generally show a slow progression, but some show a more aggressive clinical course, where several clinical/genetic factors, such as tumor size, presence of neurologic deficit before surgery and 1p19q LOH, have been reported to correlate patients’ survival. However, no large-scale studies prevent establishment of a reliable prognostication system. Methods Status of somatic mutations and copy number variations (CNVs) were investigated for 335 Japanese patients with LGG using whole exome/targeted sequencing and single nucleotide polymorphism-array karyotyping, respectively. Corresponding data were also publically available for 425 patients from the Cancer Genome Atlas (URL: http://cancergenome.nih.gov/). Correlation of genetic lesions and other parameters with overall survival (OS) was analyzed for combined 538 patients, which were divided into two sets, 269 training and 269 validation sets. First, the 269 patients of the training set were classified into 3 types according to the characteristic mutations and CNVs: Type 1 (mutated IDH with 1p/19q LOH), Type 2 (mutated IDH without 1p19q LOH), Type 3 (IDH wild type) patients. Using the LASSO Cox regression model, we built a classifier based on gene mutations, CNVs (over 5% of cases), sex, age, pathology, WHO grade and operation type (gross total resection or not). We validated the accuracy of this classifier in terms of prediction of OS in the independent group of 269 patients. Results Using the LASSO model, the patients in Type 1 and 3 were grouped into low- and high-risk groups, whereas Type 2 was not because no significant risk factors were extracted for Type 2 patients. Combining high-risk Type 1 and Type 2, between which OS was not significantly different, the entire training set was divided into 4 groups showing significantly different OS, low- and intermediate-risk groups and high- and very high-risk groups, respectively. In the training set, 5-year OS for low-, intermediate-, high-, and very high-risk groups was 100%, 82%, 49%, and 0%, which were 94%, 70%, 43%, and 14%, in the validation set, respectively. The performance of the new model was evaluated by receiver operator characteristic (ROC) analysis, which showed significantly higher accuracy compared to other models based solely on clinical/histological parameters, including pathology, WHO grade and Karnofsky performance status (KPS). Conclusion We established a new classifier based both on genetic and clinical parameters, which provides a reliable tool for predicting OS in LGGs patients and should be useful to guide therapy. Citation Format: Kosuke Aoki, Hiromichi Suzuki, Hideo Nakamura, Masahiro Mizoguchi, Tetsuya Abe, Satoru Miyano, Ichiro Takeuchi, Toshihiko Wakabayashi, Seishi Ogawa, Atsushi Natsume. Prognostic model of lower grade gliomas. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4791. doi:10.1158/1538-7445.AM2015-4791

Abstract P6-08-28: Prognostic and predictive value of an integrated mRNA-lncRNA signature in triple-negative breast cancer: A comprehensive transcriptome analysis

Abstract Purpose Triple-negative breast cancer (TNBC) is a highly diverse group of disease, and clinical outcome of patients with TNBC is highly variable. Due to the heterogeneity of TNBCs, there is limited molecular signature routinely used for predicting the risk of disease recurrence and benefit of adjuvant chemotherapy. Our study aims to develop and validate a RNA signature, integrating messenger RNAs (mRNAs) and long non-coding RNAs (lncRNAs) together, for TNBC patients to improve risk stratification and avoid unnecessary adjuvant therapy. Methods Using transcriptome microarrays, we analyzed 33 paired TNBC and adjacent normal breast tissues, and identified 1,644 mRNAs and 1,047 lncRNAs which were differentially expressed between tumors and normal tissues. We further determined the expression of these mRNAs and lncRNAs in an additional 134 TNBC samples using transcriptome microarrays, and confirmed their associations with patients' recurrence-free survival (RFS). Using the LASSO Cox regression model, we built an integrated mRNA-lncRNA signature incorporating seven mRNAs and three lncRNAs. Prognostic and predictive accuracy of the signature was tested in the training set of 167 TNBC patients and further validated in an independent validation set of 143 TNBC patients. Results In the training set, we identified 36 mRNAs and 32 lncRNAs which were tumor-specific and significantly associated with patients' RFS. Using the LASSO Cox regression model, an integrated mRNA-lncRNA signature based on seven mRNAs (CCR4, CTSB, ERO1L, HIF1A, IGFR1R, SRD5A1 and TFF1) and three lncRNAs (n381928, n333541 and TCONS_I2_00013109-XLOC_I2_007048) was developed in the training set and subsequently validated in the validation set. Patients were classified to the high-risk (high risk of recurrence) and low-risk (low risk of recurrence) groups according to their scores in the signature. In the training set, multivariate analysis showed that the predicted high-risk group had higher risk of developing recurrent disease within five years of surgery than the low-risk patients (hazard ratio [HR] = 6.12, 95% confidence interval [CI] 2.76-13.56, P&amp;lt;0.001). In the validation set, the predicted high-risk group also had poorer RFS in multivariate analysis (HR = 5.09, 95% CI 1.82-10.96, P&amp;lt;0.001). Furthermore, analyzing the areas under the time-dependent receiver operating curve for five-year RFS, we proved the integrated mRNA-lncRNA signature had better prognostic value than the seven-mRNA-only signature and clinicopathological risk factors in both the training and validation sets. Finally, Cox proportional hazards models were utilized to test the interaction between adjuvant chemotherapy and different risk groups. Patients in the low-risk group had a more favorable response to adjuvant chemotherapy both in the training and validation sets. Conclusion The integrated mRNA-lncRNA signature is a reliable tool for predicting disease recurrence and benefit of adjuvant chemotherapy in TNBC patients. If further validated in larger population, it could facilitate patient counseling and individualize treatment of TNBC. Citation Format: Yizhou Jiang, Yi-Rong Liu, Ke-Da Yu, Xin Hu, Xiao-En Xu, Ling Yao, Zhi-Ming Shao. Prognostic and predictive value of an integrated mRNA-lncRNA signature in triple-negative breast cancer: A comprehensive transcriptome analysis [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P6-08-28.

Somatic Mutations of ASXL1, RUNX1 and SETBP1 Improve Prognostic Stratification of Patients with Chronic Myelomonocytic Leukemia

Chronic myelomonocytic leukemia (CMML) is a myelodysplastic/myeloproliferative neoplasm characterized by a highly variable clinical course. Based on clinical, hematologic and cytogenetic parameters, we previously developed a CMML-specific Prognostic Scoring System (CPSS) that stratifies patients into 4 different risk groups [Blood 2013;121:3005-15]. Recently, recurrent somatic mutations have been identified in CMML, and preliminary evidence suggests that selected mutated genes may provide useful prognostic information.In this study, we performed a comprehensive mutation analysis of genes implicated in myeloid malignancies in a large and well characterized cohort of CMML patients with the aim to dissect relationships between genotype and disease phenotype and to integrate somatic mutations into a clinical/molecular prognostic model.Thirty-eight genes were analyzed by high throughput sequencing (Illumina MiSeq, San Diego, CA) in a cohort of 199 CMML patients (pts) diagnosed according to WHO classification, and in 12 pts with monocytosis not fulfilling WHO diagnostic criteria. Myelodysplastic and myeloproliferative subtypes (CMML-MD and CMML-MP, respectively) were defined according to FAB criteria. Least absolute shrinkage and selection operator (Lasso) and Cox proportional hazards methods were adopted to select and weight variables for prognostic scoring.Ninety-three percent of pts showed at least one somatic mutation (median number per patient: 2, range 0-6). The most frequently mutated genes were TET2 (44%), SRSF2 (43%), ASXL1 (34%), KRAS (11%), NRAS (10%), CUX1 (10%), CBL (9%), RUNX1 (7%), SETBP1 (7%), JAK2 (6%), SF3B1 (6%), and U2AF1 (5%). A significant association was found between mutations in TET2 and RNA splicing factors (P=.037), 42 of 199 CMML pts (21%) showing co-occurrence of TET2 and SRSF2 mutations. Mutations in genes involved in signaling were significantly associated with CMML-MP (P=.002), whereas SF3B1 mutations were associated with CMML-MD (P=.024).The number of mutations per patient inversely correlated with overall survival (OS) (HR=1.32, P=.021). In univariate analysis, mutations in ASXL1 (HR=2.31, P=.026) , RUNX1 (HR=3.53, P=.02) and SETBP1 (HR=3.85, P=.005) significantly affected OS. Focusing the analysis on disease subtype, ASXL1 mutations significantly affected survival in CMML-MD (HR=3.45, P=.025), whereas CUX1 and SETBP1 had a significant prognostic value in CMML-MP (HR=4.33, P=.013 and HR=4.4, P=.025, respectively).In order to investigate the additive value of somatic mutations to current prognostic assessment, we first fitted a Lasso Cox regression model for genetic variable selection. The selected variables were then included in an unpenalized Cox regression in order to obtain unbiased coefficients. The statistically significant variables were CPSS-specific cytogenetic risk groups (HR=2.49, P=.001), mutations in ASXL1 (HR=2.77, P=.018), RUNX1 (HR=5.39, P=.009) and SETBP1 (HR=3.96, P=.013). Based on regression coefficients, we defined a CMML-specific genetic risk score that was able to identify three different groups (Low risk: normal karyotype and –Y; Intermediate: other abnormalities, mutations in ASXL1; High: trisomy 8, complex karyotype, mutations in RUNX1 or SETBP1), with significantly different OS (HR=2.24, P<.001). The Akaike information criterion showed that this genetic risk score performed better than the original CPSS cytogenetic risk classification (AIC 274 vs. 282, respectively). According to the CMML-specific genetic risk score, 36% of patients had a shift toward a higher risk category compared with cytogenetic risk classification. Then, the new genetic risk categories were integrated in the CPSS (CPSS-Mol), which was able to identify 4 risk groups with different OS (HR=2.29, P<.001) and risk of disease progression (HR=2.62, P<.001). As a robustness analysis, we also fitted a Lasso Cox regression model including clinical and genetic variables, and all variables included in the clinical/molecular score were confirmed. The Akaike information criterion showed that the CPSS-Mol performed better than the original CPSS (AIC 251 vs. 253, respectively).In conclusion, this study showed that mutation pattern in CMML significantly correlates with disease phenotype. The integration of somatic mutations in current scoring systems significantly improves prognostic stratification of patients. DisclosuresNo relevant conflicts of interest to declare.

Prognostic and predictive value of a microRNA signature in stage II colon cancer: a microRNA expression analysis

Current staging methods do not accurately predict the risk of disease recurrence and benefit of adjuvant chemotherapy for patients who have had surgery for stage II colon cancer. We postulated that expression patterns of multiple microRNAs (miRNAs) could, if combined into a single model, improve postoperative risk stratification and prediction of chemotherapy benefit for these patients. Using miRNA microarrays, we analysed 40 paired stage II colon cancer tumours and adjacent normal mucosa tissues, and identified 35 miRNAs that were differentially expressed between tumours and normal tissue. Using paraffin-embedded specimens from a further 138 patients with stage II colon cancer, we confirmed differential expression of these miRNAs using qRT-PCR. We then built a six-miRNA-based classifier using the LASSO Cox regression model, based on the association between the expression of every miRNA and the duration of individual patients' disease-free survival. We validated the prognostic and predictive accuracy of this classifier in both the internal testing group of 138 patients, and an external independent group of 460 patients. Using the LASSO model, we built a classifier based on the six miRNAs: miR-21-5p, miR-20a-5p, miR-103a-3p, miR-106b-5p, miR-143-5p, and miR-215. Using this tool, we were able to classify patients between those at high risk of disease progression (high-risk group), and those at low risk of disease progression (low-risk group). Disease-free survival was significantly different between these groups in every set of patients. In the initial training group of patients, 5-year disease-free survival was 89% (95% CI 77·3-94·4) for the low-risk group, and 60% (46·3-71·0) for the high-risk group (hazard ratio [HR] 4·24, 95% CI 2·13-8·47; p<0·0001). In the internal testing set of patients, 5-year disease-free survival was 85% (95% CI 74·3-91·8) for the low-risk group, and 57% (42·8-68·5) for the high-risk group (HR 3·63, 1·86-7·01; p<0·0001), and in the independent validation set of patients, was 85% (79·6-89·0) for the low-risk group and 54% (46·4-61·1) for the high-risk group (HR 3·70, 2·56-5·35; p<0·0001). The six-miRNA-based classifier was an independent prognostic factor for, and had better prognostic value than, clinicopathological risk factors and mismatch repair status. In an ad-hoc analysis, the patients in the high-risk group were found to have a favourable response to adjuvant chemotherapy (HR 1·69, 1·17-2·45; p=0·0054). We developed two nomograms for clinical use that integrated the six-miRNA-based classifier and four clinicopathological risk factors to predict which patients might benefit from adjuvant chemotherapy after surgery for stage II colon cancer. Our six-miRNA-based classifier is a reliable prognostic and predictive tool for disease recurrence in patients with stage II colon cancer, and might be able to predict which patients benefit from adjuvant chemotherapy. It might facilitate patient counselling and individualise management of patients with this disease. Natural Science Foundation of China.

Aberrant DNA methylation characterizes juvenile myelomonocytic leukemia with poor outcome

AbstractAberrant DNA methylation contributes to the malignant phenotype in virtually all types of cancer, including myeloid leukemia. We hypothesized that CpG island hypermethylation also occurs in juvenile myelomonocytic leukemia (JMML) and investigated whether it is associated with clinical, hematologic, or prognostic features. Based on quantitative measurements of DNA methylation in 127 JMML cases using mass spectrometry (MassARRAY), we identified 4 gene CpG islands with frequent hypermethylation: BMP4 (36% of patients), CALCA (54%), CDKN2B (22%), and RARB (13%). Hypermethylation was significantly associated with poor prognosis: when the methylation data were transformed into prognostic scores using a LASSO Cox regression model, the 5-year overall survival was 0.41 for patients in the top tertile of scores versus 0.72 in the lowest score tertile (P = .002). Among patients given allogeneic hematopoietic stem cell transplantation, the 5-year cumulative incidence of relapse was 0.52 in the highest versus 0.10 in the lowest score tertile (P = .007). In multivariate models, DNA methylation retained prognostic value independently of other clinical risk factors. Longitudinal analyses indicated that some cases acquired a more extensively methylated phenotype at relapse. In conclusion, our data suggest that a high-methylation phenotype characterizes an aggressive biologic variant of JMML and is an important molecular predictor of outcome.