Risk In Colorectal Cancer Patients Research Articles

Establishing and clinically validating a machine learning model for predicting unplanned reoperation risk in colorectal cancer.

Colorectal cancer significantly impacts global health, with unplanned reoperations post-surgery being key determinants of patient outcomes. Existing predictive models for these reoperations lack precision in integrating complex clinical data. To develop and validate a machine learning model for predicting unplanned reoperation risk in colorectal cancer patients. Data of patients treated for colorectal cancer (n = 2044) at the First Affiliated Hospital of Wenzhou Medical University and Wenzhou Central Hospital from March 2020 to March 2022 were retrospectively collected. Patients were divided into an experimental group (n = 60) and a control group (n = 1984) according to unplanned reoperation occurrence. Patients were also divided into a training group and a validation group (7:3 ratio). We used three different machine learning methods to screen characteristic variables. A nomogram was created based on multifactor logistic regression, and the model performance was assessed using receiver operating characteristic curve, calibration curve, Hosmer-Lemeshow test, and decision curve analysis. The risk scores of the two groups were calculated and compared to validate the model. More patients in the experimental group were ≥ 60 years old, male, and had a history of hypertension, laparotomy, and hypoproteinemia, compared to the control group. Multiple logistic regression analysis confirmed the following as independent risk factors for unplanned reoperation (P < 0.05): Prognostic Nutritional Index value, history of laparotomy, hypertension, or stroke, hypoproteinemia, age, tumor-node-metastasis staging, surgical time, gender, and American Society of Anesthesiologists classification. Receiver operating characteristic curve analysis showed that the model had good discrimination and clinical utility. This study used a machine learning approach to build a model that accurately predicts the risk of postoperative unplanned reoperation in patients with colorectal cancer, which can improve treatment decisions and prognosis.

Correlation between PD-L1 Expression, Clinicopathological Factors, and Metastasis Risk in Colorectal Cancer Patients

This study investigated the correlation between PD-L1 expression, metastasis, and survival in colorectal cancer (CRC) patients. PD-L1 expression was not significantly associated with overall survival, disease-free survival, or mortality rate. However, a significant difference was observed between PD-L1 positive and negative patients regarding the presence of metastasis, which was higher in the PD-L1 positive group. These findings suggest that PD-L1 expression may impact metastasis in CRC patients but not overall survival.

Rapid detection of liver metastasis risk in colorectal cancer patients through blood test indicators.

Colorectal cancer (CRC) is one of the most common malignancies, with liver metastasis being its most common form of metastasis. The diagnosis of colorectal cancer liver metastasis (CRCLM) mainly relies on imaging techniques and puncture biopsy techniques, but there is no simple and quick early diagnosisof CRCLM. This study aims to develop a method for rapidly detecting the risk of liver metastasis in CRC patients through blood test indicators based on machine learning (ML) techniques, thereby improving treatment outcomes. To achieve this, blood test indicators from 246 CRC patients and 256 CRCLM patients were collected and analyzed, including routine blood tests, liver function tests, electrolyte tests, renal function tests, glucose determination, cardiac enzyme profiles, blood lipids, and tumor markers. Six commonly used ML models were used for CRC and CRCLM classification and optimized by using a feature selection strategy. The results showed that AdaBoost algorithm can achieve the highest accuracy of 89.3% among the six models, which improved to 91.1% after feature selection strategy, resulting with 20 key markers. The results demonstrate that the combination of machine learning techniques with blood markers is feasible and effective for the rapid diagnosis of CRCLM, significantly im-proving diagnostic ac-curacy and patient prognosis.

A Retrospective Clinical Study on Cardiovascular Complications from Colorectal Cancer.

To investigate the incidence and risk factors of cardiovascular complications amongst patients with colorectal cancer (CRC). A retrospective cohort study was conducted on 2085 patients diagnosed with CRC in two tertiary hospitals in China between 2015 and 2020. The patients' medical records were reviewed to identify cardiovascular complications, including myocardial infarction, heart failure, stroke, hypertension, coronary heart disease, heart failure, and arrhythmia. The incidence rate of cardiovascular complications was calculated, and Cox proportional hazards regression analysis was used to identify risk factors. Of the 2085 CRC patients, 329 (15.8%) experienced cardiovascular complications during the follow-up period, with an incidence rate of 17.4 cases per 1000 person-years. The risk was significantly higher in patients who were older than 60 years (adjusted hazard ratio [HR] 2.04, 95% confidence interval [CI] 1.22-3.41), had a higher level of low-density lipoprotein cholesterol (LDL-C) (adjusted HR 2.32, 95% CI 1.31-4.10), had higher levels of serum C-reactive protein (CRP) (adjusted HR 1.57, 95% CI 1.21-2.04), or who underwent chemotherapy or radiotherapy. CRC patients with cardiovascular complications had significantly higher levels of oxidative stress markers, including malondialdehyde (MDA) (5.8 ± 1.2 μmol/L vs. 3.4 ± 0.9 μmol/L, p < 0.001), lower levels of superoxide dismutase (SOD) (85.2 ± 15.6 U/mg protein vs. 112.5 ± 21.3 U/mg protein, p < 0.001), and lower levels of glutathione peroxidase (GPx) (15.6 ± 3.2 U/mg protein vs. 20.4 ± 4.1 U/mg protein, p < 0.001) compared to those without complications. A progressive increase was observed in the proportion of CRC patients with cardiovascular complications over time, rising from 10% in the first year to 38% by the tenth year of follow-up. Cardiovascular complications pose a high risk in CRC patients, particularly amongst older patients and those with higher levels of LDL-C or CRP. Regular monitoring of cardiovascular function should be considered in the management of patients with CRC.

Improved semiparametric estimation of the proportional rate model with recurrent event data.

Owing to its robustness properties, marginal interpretations, and ease of implementation, the pseudo-partial likelihood method proposed in the seminal papers of Pepe and Cai and Lin etal. has become the default approach for analyzing recurrent event data with Cox-type proportional rate models. However, the construction of the pseudo-partial score function ignores the dependency among recurrent events and thus can be inefficient. An attempt to investigate the asymptotic efficiency of weighted pseudo-partial likelihood estimation found that the optimal weight function involves the unknown variance-covariance process of the recurrent event process and may not have closed-form expression. Thus, instead of deriving the optimal weights, we propose to combine a system of pre-specified weighted pseudo-partial score equationsvia the generalized method of moments and empirical likelihood estimation. We show that a substantial efficiency gain can be easily achieved without imposing additional model assumptions. More importantly, the proposed estimation procedures can be implemented with existing software. Theoretical and numerical analyses show that the empirical likelihood estimator is more appealing than the generalized method of moments estimator when the sample size is sufficiently large. An analysis of readmission risk in colorectal cancer patients is presented to illustrate the proposedmethodology.

A novel senescence-associated LncRNA signature predicts the prognosis and tumor microenvironment of patients with colorectal cancer: a bioinformatics analysis

BackgroundAccumulating evidence suggests that cellular senescence promotes tumor formation and that long non-coding RNAs (lncRNAs) expression predicts tumor prognosis. However, senescence-related variables, particularly lncRNAs, are still largely unknown. Therefore, the present study developed a novel senescence-associated lncRNA signature to predict colorectal cancer (CRC) prognosis.MethodsA co-expression network of senescence-associated mRNAs and lncRNAs was built using RNA-sequence data from The Cancer Genome Atlas (TCGA). By using the prognosis outcomes data of overall survival (OS) and disease-free survival (DFS) from TCGA, we constructed a prognostic senescence-associated lncRNA signature (SenALSig). The OS and DFS were compared between the low- and high- risk groups defined by SenALSig. A single-sample gene set enrichment analysis (ssGSEA) and CIBERSORT algorithm were used to investigate the relationship between the predictive signature and immune status. Finally, the relationship between SenALSig and drug treatment options was investigated. An independent CRC cohort and three CRC cell lines were recruited to perform real-time quantitative reverse transcription polymerase chain reaction (RT-qPCR) analysis to validate the results discovered in silico.ResultsA prognostic risk model consisting of six senescence-associated lncRNAs was constructed, including SNHG16, AL590483.1, ZEB1-AS1, AC107375.1, AC068580.3, and AC147067.1. High-risk scores according to the SenALSig were significantly associated with poor OS (hazard ratio =1.218, 95% confidence interval: 1.140–1.301; P<0.001). The model’s accuracy was further supported by receiver operating characteristic (ROC) curves (the area under the curve is 0.714) and a principal component analysis (PCA). In univariate and multivariate Cox regression analyses, SenALSig was further found to be a prognostic factor independent of other clinical factors. Furthermore, we discovered that immune checkpoint expression and response to chemotherapy and targeted therapy differed significantly between the SenALSig-stratified high- and low-risk groups. Finally, the qPCR results revealed that the expression levels of the six senescence-associated lncRNAs differed significantly between tumor and normal tissues and between the CRC cell lines and a normal human colon mucosal epithelial cell line.ConclusionsSenALSig can better predict survival and risk in CRC patients, as well as help develop new anti-cancer treatment strategies for CRC.

Development of a Prognostic Model Based on an Immunogenomic Landscape Analysis of Colorectal Cancer

Screening and therapeutic programs for colorectal cancer (CRC) are invasive or not effective and unable to meet patient needs. Major advances in immunogenomics may change this status but need more exploration. Differentially expressed genes and immune-related genes (IRGs) were identified by computational methods. A prognostic model was established and validated based on survival-related IRGs via stepwise multivariate Cox regression analysis. Nine IRGs were selected and identified as survival-related genes. A 7-gene prognostic model could offer a preliminary and valid determination of risk in CRC patients. The area under the curve of the receiver operating characteristic was 0.672. The 7-gene prognostic model might be used as a novel prognostic tool in CRC patients.

Development and internal validation of a nine-lncRNA prognostic signature for prediction of overall survival in colorectal cancer patients.

BackgroundColorectal cancer remains a serious public health problem due to the poor prognosis. In the present study, we attempted to develop and validate a prognostic signature to predict the individual mortality risk in colorectal cancer patients.Materials and MethodsThe original study datasets were downloaded from The Cancer Genome Atlas database. The present study finally included 424 colorectal cancer patients with wholly gene expression information and overall survival information.ResultsA nine-lncRNA prognostic signature was built through univariate and multivariate Cox proportional regression model. Time-dependent receiver operating characteristic curves in model cohort demonstrated that the Harrell’s concordance indexes of nine-lncRNA prognostic signature were 0.768 (95% CI [0.717–0.819]), 0.778 (95% CI [0.727–0.829]) and 0.870 (95% CI [0.819–0.921]) for 1-year, 3-year and 5-year overall survival respectively. In validation cohort, the Harrell’s concordance indexes of nine-lncRNA prognostic signature were 0.761 (95% CI [0.710–0.812]), 0.801 (95% CI [0.750–0.852]) and 0.883 (95% CI [0.832–0.934]) for 1-year, 3-year and 5-year overall survival respectively. According to the median of nine-lncRNA prognostic signature score in model cohort, 424 CRC patients could be stratified into high risk group (n = 212) and low risk group (n = 212). Kaplan–Meier survival curves showed that the overall survival rate of high risk group was significantly lower than that of low risk group (P < 0.001).DiscussionThe present study developed and validated a nine-lncRNA prognostic signature for individual mortality risk assessment in colorectal cancer patients. This nine-lncRNA prognostic signature is helpful to evaluate the individual mortality risk and to improve the decision making of individualized treatments in colorectal cancer patients.

O3-17-1 - Importance to clarify germline hMLH1 K618A missense mutation for clinical surveillance and genetic counseling

Abstract Colorectal cancer (CRC) is the third common cancer and the fourth leading cause of cancer death worldwide. Subproportion CRC with defect function in one of the mismatch repair gene (MMR) are classed as inherited non-polyposis colorectal cancer (HNPCC) later called Lynch syndrome (LS). Disease causing germline mutations have so far been identified in five MMR genes. Among other MMR genes, germline mutations in hMLH1 account up to 50% LS patients and lead to increased risk of cancer development. hMLH1 K618A variant was first identified in our lab and was considered as a disease causing pathogenic alteration, our further investigation also support our previous findings. Although this variant has been investigated intensively during past decade, however the pathogenicity still remain unclear due to controversial data published up to date. In present study, hMLH1 K618A variant were genotyped in 2,743 individuals (1,087 sporadic CRC patients, 239 CRC patients with at least one affected close relative, 69 CRC cases with familial history and 1,348 controls) by TaqMan genotyping assay. Our results demonstrate that 30 individuals were heterozygous for the K618A variant; thirteen normal control had this variants and measurement of allele frequency was 0.96 % which is remarkably higher than previous published data, while eight were in sporadic CRC cases and allele frequency was 0.73 %, six were in CRC patients with at least one affected relative and allele frequency was 2.51 %, three were in CRC patients with familial history and allele frequency was 4.34 %. We conclude that hMLH1 K618A variant may not have effect in sporadic CRC development, in contrast that K618A may significantly increase CRC risk in CRC patients with strong familial history in Swedish population.

The effects of optical sensor-tissue separation in endocavitary photoplethysmography

Objective: Intestinal anastomotic failure that occurs mainly due to ischaemia is a serious risk in colorectal cancer patients undergoing surgery. Surgeons continue to rely on subjective methods such as visual inspection to assess intestinal viability during surgery and there are no clinical tools to directly monitor viability postoperatively. A dual-wavelength reflectance optical sensor has been developed for continuous and dynamic monitoring of intestinal viability via the intestinal lumen. Maintaining direct contact between the sensor and the inner intestinal wall can be difficult in an intraluminal design, therefore impacting on signal acquisition and quality. This paper investigates the effect of direct contact versus variable distances between the sensor and the tissue surface of the buccal mucosa as a surrogate. Approach: The in vivo study involved 20 healthy volunteers to measure the effect of optical sensor-tissue distances on the ability to acquire photoplethysmography signals and their quality. Signals were acquired from the buccal mucosa at five optical sensor-tissue distances. Main results: Distances between 0 mm (contact) to 5 mm were the most optimal, producing signals of high quality and signal-to-noise ratio, resulting in reliable estimations of the blood oxygen saturation. Distances exceeding 5 mm compromised the acquired signals, and were of poor quality, thereby unreliably estimating the blood oxygen saturation. Significance: The developed optical sensor proved to be reliable for acquiring photoplethysmography signals for cases where distances between the optical sensor-tissue may arise during the assessment of intraluminal intestinal viability.

Risk and predictors of suicide in colorectal cancer patients: a Surveillance, Epidemiology, and End Results analysis.

The risk of suicide is higher for patients with colorectal cancer (crc) than for the general population. Given known differences in morbidity and sites of recurrence, we sought to compare the predictors of suicide for patients with colon cancer and with rectal cancer. Using the U.S. Surveillance, Epidemiology, and End Results database, adult patients with confirmed adenocarcinoma of the colon or rectum during 1973-2009 were identified. Parametric and nonparametric tests were used to assess selected variables, and Cox proportional hazards regression models were used to determine predictors of suicide. The database identified 187,996 patients with rectal cancer and 443,368 with colon cancer. Compared with the rectal cancer group, the colon cancer group was older (median age: 70 years vs. 67 years; p < 0.001) and included more women (51% vs. 43%, p < 0.001). Suicide rates were similar in the colon and rectal cancer groups [611 (0.14%) vs. 337 (0.18%), p < 0.001]. On univariate analysis, rectal cancer was a predictor of suicide [hazard ratio (hr): 1.26; 95% confidence interval (ci): 1.10 to 1.43]. However, after adjusting for clinical and pathology factors, rectal cancer was not a predictor of suicide (hr: 1.05; 95% ci: 0.83 to 1.33). In the colon cancer cohort, independent predictors of suicide included older age, male sex, white race, and lack of primary resection. The aforementioned predictors, plus metastatic disease, similarly predicted suicide in the rectal cancer cohort. The suicide risk in crc patients is low (<0.2%), and no difference was found based on location of the primary tumour. Sex, age, race, distant spread of disease, and intact primary tumour were the main predictors of suicide among crc patients. Further studies and interventions are needed to target these high-risk groups.

Influence of type 2 diabetes mellitus on Khorana venous thromboembolism risk in colorectal cancer patients.

Many studies have documented the association between venous thromboembolism (VTE) and colorectal cancer (CRC). The Khorana model is a VTE risk assessment model for predicting cancer-associated thrombosis. Type 2 diabetes (T2DM) has also been reported to increase the risk of VTE. The aim of this study was to investigate the influence of T2DM on Khorana VTE risk in CRC patients and to explore the relationship between Khorana VTE category and CRC clinicopathological factors. This analysis included 615 CRC patients (205 with T2DM). Fibrinogen and D-dimer levels were compared within each group. A comparison was made of the proportion of patients in different Khorana VTE risk categories in CRC patients with and without T2DM. The association between Khorana VTE risk category and clinicopathological factors among all the CRC patients was evaluated. Fibrinogen levels of CRC patients with T2DM were significantly higher than those of non-diabetes patients (4.13 ± 1.06 vs 3.94 ± 0.98, p < 0.001). A higher proportion of CRC patients with T2DM were in the Khorana intermediate-to-high risk category (H = 4.749, p = 0.029). Female sex, diabetes, colon location (compared with rectum), larger tumor size, advanced pT stage and pN stage were correlated with the intermediate-to-high Khorana VTE risk category, with odd ratios (95% confidence intervals [CI]) of 1.537 (1.064-2.220), 1.499 (1.027-2.186), 2.313 (1.588-3.370), 2.284 (1.542-3.383), 4.429 (2.088-9.396) and 1.822 (1.230-2.698), respectively. T2DM increases Khorana VTE risk in CRC patients. Female sex, diabetes, colon location, large tumor size and poor stage are associated with the intermediate-to-high Khorana VTE risk category.

P1-11-1 - Importance to Clarify Germline Hmlh1 K618A Missense Mutation for Clinical Surveillance and Genetic Counseling

Colorectal cancer (CRC) is the third common cancer and the fourth leading cause of cancer death worldwide. Subproportion CRC with defect function in one of the mismatch repair gene (MMR) are classed as inherited non-polyposis colorectal cancer (HNPCC) later called Lynch syndrome (LS). Disease causing germline mutations have so far been identified in five MMR genes. Among other MMR genes, germline mutations in hMLH1 account up to 50% LS patients and lead to increased risk of cancer development. hMLH1 K618A variant was first identified in our lab and was considered as a disease causing pathogenic alteration, our further investigation also support our previous findings. Although this variant has been investigated intensively during past decade, however the pathogenicity still remain unclear due to controversial data published up to date.In present study, hMLH1 K618A variant were genotyped in 2,743 individuals (1,087 sporadic CRC patients, 239 CRC patients with at least one affected close relative, 69 CRC cases with familial history and 1,348 controls) by TaqMan genotyping assay. Our results demonstrate that 30 individuals were heterozygous for the K618A variant; thirteen normal control had this variants and measurement of allele frequency was 0.96 % which is remarkably higher than previous published data, while eight were in sporadic CRC cases and allele frequency was 0.73 %, six were in CRC patients with at least one affected relative and allele frequency was 2.51 %, three were in CRC patients with familial history and allele frequency was 4.34 %. We conclude that hMLH1 K618A variant may not have effect in sporadic CRC development, in contrast that K618A may significantly increase CRC risk in CRC patients with strong familial history in Swedish population.

Influence of tumor microenvironment on prognosis in colorectal cancer: Tissue architecture-dependent signature of endosialin (TEM-1) and associated proteins.

Tumor survival is influenced by interactions between tumor cells and the stromal microenvironment. One example is Endosialin (Tumor Endothelial Marker-1 (TEM-1) or CD248), which is expressed primarily by cells of mesenchymal origin and some tumor cells. The expression, as a function of architectural masking, of TEM-1 and its pathway-associated proteins was quantified and examined for association with five-year disease-specific survival on a colorectal cancer (CRC) cohort divided into training (n=330) and validation (n=164) sets. Although stromal expression of TEM-1 had prognostic value, a more significant prognostic signature was obtained through linear combination of five compartment-specific expression scores (TEM-1 Stroma, TEM-1 Tumor Vessel, HIF2α Stromal Vessel, Collagen IV Tumor, and Fibronectin Stroma). This resulted in a single continuous risk score (TAPPS: TEM-1 Associated Pathway Prognostic Signature) which was significantly associated with decreased survival on both the training set [HR=1.76 (95%CI: 1.44-2.15); p<0.001] and validation set [HR=1.38 (95%CI: 1.02-1.88); p=0.04]. Importantly, since prognosis is a critical clinical question in Stage II patients, the TAPPS score also significantly predicted survival in the Stage II patient (n=126) cohort [HR=1.75 (95%CI: 1.22-2.52); p=0.002] suggesting the potential of using the TAPPS score to assess overall risk in CRC patients, and specifically in Stage II patients.

PCN28 - Bevacizumab-Based Chemotherapy and Thrombotic Events Risk in Colorectal Cancer Patients: A Meta-Analysis Study Of Randomized Controlled Trials

PCN28 - Bevacizumab-Based Chemotherapy and Thrombotic Events Risk in Colorectal Cancer Patients: A Meta-Analysis Study Of Randomized Controlled Trials

Risk and predictors of suicide in colorectal cancer patients: A SEER analysis.

9596 Background: Colorectal cancer (CRC) patients have a higher risk of suicide as compared with the general population. Due to differences in the sites/morbidity of recurrences as well as ostomy rates, we sought to evaluate the distribution and predictors of suicide among patients with colon and rectal cancer. Methods: A retrospective analysis was undertaken using the Surveillance, Epidemiology, and End Results (SEER) database from 1973-2009. Patients included were &gt;18yrs and had confirmed adenocarcinoma of the colon or rectum. Results: Included in this analysis were 187,996 rectal cancer and 443,368 colon cancer patients. Colon cancer patients were older (median age 71 vs. 67 yrs, p &lt;0.001) and included more females (51 vs. 43%, p &lt;0.001) as compared to rectal cancer patients. Suicide rates were similar (611 [0.14%] vs. 337 [0.18%], p &lt;0.001), as was the median time to suicide for colon vs. rectal cancer patients respectively (37 vs.32 months, p = 0.13). On univariate analysis, having rectal cancer was a predictor of suicide (HR 1.26; 95% CI: 1.10-1.43). However after adjustment for age, sex, race, marital, primary site surgery, stage and one primary, rectal cancer was not a predictor of suicide (HR 1.05; CI: 0.83- 1.33). In the combined CRC cohort, independent predictors of suicide included age &gt;70 (HR 1.55; CI: 1.23-1.94), male gender (HR 7.56; CI: 5.34-10.70), being single (HR 1.56; CI: 1.14- 2.13), distant metastases at diagnosis (HR 1.58; CI: 1.13- 2.21), and white race (HR 3.21; CI: 1.75- 5.88). Also, lack of resection of primary tumor was associated with increased risk of suicide (HR 2.83; CI: 1.97- 4.05). Among colon cancer cohort, older age, male gender, and white race as well as lack of primary resection were independent predictors of suicide. Similarly, the aforementioned predictors as well as metastatic disease on presentation were the independent predictors of suicide in the rectal cohort. Conclusions: The suicide risk in CRC patients is low (&lt; 0.2%) and no difference was found based on location of primary tumor. Gender, age, distant spread of disease, intact primary tumour and race are the main predictors of suicide among colorectal patients. Future studies and interventions are needed to target these high risk groups.

Colorectal cancer and risk of atrial fibrillation and flutter: a population-based case–control study

Colorectal cancer has recently been associated with an increased atrial fibrillation risk, but evidence is very sparse. So, we conducted a population-based case-control study in northern Denmark (population 1.7 million) during 1998-2006 to estimate the atrial fibrillation/flutter risk in colorectal cancer patients. We identified 28,333 atrial fibrillation/flutter cases and 283,260 sex-, age-, and county-matched population controls. We searched the databases for a prior colorectal cancer diagnosis, a prior cancer diagnosis other than colorectal cancer, and performance of surgery within 30 days prior to atrial fibrillation/flutter. We used conditional logistic regression to estimate the OR of atrial fibrillation/flutter in patients with colorectal cancer, cancers other than colorectal and in patient with surgery. Among cases, 0.59% (n = 168) had a colorectal cancer diagnosis within 90 days before their atrial fibrillation/flutter diagnosis, compared with 0.05% (n = 155) of controls (adjusted OR = 11.8; 95% CI 9.3-14.9). Beyond the first 90 days after a colorectal cancer diagnosis, atrial fibrillation/flutter risk was no longer increased. There was likewise an increased atrial fibrillation/flutter risk in patients diagnosed with another cancer form in the prior 90 days (OR = 7.0, 95% CI 6.3-7.8). Furthermore, the atrial fibrillation/flutter risk was elevated fivefold in patients who had undergone surgery, whether or not cancer-related. We therefore conclude that colorectal cancer patients are at increased atrial fibrillation/flutter risk exclusively in the first 90 days after cancer diagnosis, but to no greater an extent than are patients with other cancers. The performance of surgery probably plays an important role in this association.

Semi-supervised learning improves gene expression-based prediction of cancer recurrence

Gene expression profiling has shown great potential in outcome prediction for different types of cancers. Nevertheless, small sample size remains a bottleneck in obtaining robust and accurate classifiers. Traditional supervised learning techniques can only work with labeled data. Consequently, a large number of microarray data that do not have sufficient follow-up information are disregarded. To fully leverage all of the precious data in public databases, we turned to a semi-supervised learning technique, low density separation (LDS). Using a clinically important question of predicting recurrence risk in colorectal cancer patients, we demonstrated that (i) semi-supervised classification improved prediction accuracy as compared with the state of the art supervised method SVM, (ii) performance gain increased with the number of unlabeled samples, (iii) unlabeled data from different institutes could be employed after appropriate processing and (iv) the LDS method is robust with regard to the number of input features. To test the general applicability of this semi-supervised method, we further applied LDS on human breast cancer datasets and also observed superior performance. Our results demonstrated great potential of semi-supervised learning in gene expression-based outcome prediction for cancer patients. bing.zhang@vanderbilt.edu. Supplementary data are available at Bioinformatics online.

How accurate is the Cleveland Clinic Foundation model in predicting operative risk in colorectal cancer patients?

How accurate is the Cleveland Clinic Foundation model in predicting operative risk in colorectal cancer patients?