A qualitative transcriptional signature for determining the grade of colorectal adenocarcinoma

T2-weighted imaging-based radiomic-clinical machine learning model for predicting the differentiation of colorectal adenocarcinoma

BACKGROUND

The study on predicting the differentiation grade of colorectal cancer (CRC) based on magnetic resonance imaging (MRI) has not been reported yet. Developing a non-invasive model to predict the differentiation grade of CRC is of great value.

AIM

To develop and validate machine learning-based models for predicting the differentiation grade of CRC based on T2-weighted images (T2WI).

METHODS

We retrospectively collected the preoperative imaging and clinical data of 315 patients with CRC who underwent surgery from March 2018 to July 2023. Patients were randomly assigned to a training cohort (n = 220) or a validation cohort (n = 95) at a 7:3 ratio. Lesions were delineated layer by layer on high-resolution T2WI. Least absolute shrinkage and selection operator regression was applied to screen for radiomic features. Radiomics and clinical models were constructed using the multilayer perceptron (MLP) algorithm. These radiomic features and clinically relevant variables (selected based on a significance level of P < 0.05 in the training set) were used to construct radiomics-clinical models. The performance of the three models (clinical, radiomic, and radiomic-clinical model) were evaluated using the area under the curve (AUC), calibration curve and decision curve analysis (DCA).

RESULTS

After feature selection, eight radiomic features were retained from the initial 1781 features to construct the radiomic model. Eight different classifiers, including logistic regression, support vector machine, k-nearest neighbours, random forest, extreme trees, extreme gradient boosting, light gradient boosting machine, and MLP, were used to construct the model, with MLP demonstrating the best diagnostic performance. The AUC of the radiomic-clinical model was 0.862 (95%CI: 0.796-0.927) in the training cohort and 0.761 (95%CI: 0.635-0.887) in the validation cohort. The AUC for the radiomic model was 0.796 (95%CI: 0.723-0.869) in the training cohort and 0.735 (95%CI: 0.604-0.866) in the validation cohort. The clinical model achieved an AUC of 0.751 (95%CI: 0.661-0.842) in the training cohort and 0.676 (95%CI: 0.525-0.827) in the validation cohort. All three models demonstrated good accuracy. In the training cohort, the AUC of the radiomic-clinical model was significantly greater than that of the clinical model (P = 0.005) and the radiomic model (P = 0.016). DCA confirmed the clinical practicality of incorporating radiomic features into the diagnostic process.

CONCLUSION

In this study, we successfully developed and validated a T2WI-based machine learning model as an auxiliary tool for the preoperative differentiation between well/moderately and poorly differentiated CRC. This novel approach may assist clinicians in personalizing treatment strategies for patients and improving treatment efficacy.

Development of machine learning-based predictors for early diagnosis of hepatocellular carcinoma

Hepatocellular carcinoma (HCC) remains a formidable malignancy that significantly impacts human health, and the early diagnosis of HCC holds paramount importance. Therefore, it is imperative to develop an efficacious signature for the early diagnosis of HCC. In this study, we aimed to develop early HCC predictors (eHCC-pred) using machine learning-based methods and compare their performance with existing methods. The enhancements and advancements of eHCC-pred encompassed the following: (i) utilization of a substantial number of samples, including an increased representation of cirrhosis tissues without HCC (CwoHCC) samples for model training and augmented numbers of HCC and CwoHCC samples for model validation; (ii) incorporation of two feature selection methods, namely minimum redundancy maximum relevance and maximum relevance maximum distance, along with the inclusion of eight machine learning-based methods; (iii) improvement in the accuracy of early HCC identification, elevating it from 78.15 to 97% using identical independent datasets; and (iv) establishment of a user-friendly web server. The eHCC-pred is freely accessible at http://www.dulab.com.cn/eHCC-pred/ . Our approach, eHCC-pred, is anticipated to be robustly employed at the individual level for facilitating early HCC diagnosis in clinical practice, surpassing currently available state-of-the-art techniques.

Identifying individualized prognostic signature and unraveling the molecular mechanism of recurrence in early-onset colorectal cancer

BACKGROUND

The incidence and mortality of early-onset colorectal cancer (EOCRC; < 50 years old) is increasing worldwide, with a high recurrence rate. The inherent heterogeneity of EOCRC makes its treatment challenging. Hence, to further understand the biology and reveal the molecular mechanisms of EOCRC, a recurrence risk signature is needed to guide clinical management.

METHODS

Based on the relative expression orderings (REOs) of genes in each sample, a prognostic signature was developed and validated utilizing multiple independent datasets. The underlying molecular mechanisms between distinct prognostic groups were explored via integrative analysis of multi-omics data.

RESULTS

The prognostic signature consisting of 6 gene pairs (6-GPS) could predict the recurrence risk for EOCRC at the individual level. High-risk EOCRC classified by 6-GPS showed a poor prognosis but a good response to adjuvant chemotherapy. Moreover, high-risk EOCRC was characterized by epithelial-mesenchymal transition (EMT) and enriched angiogenesis, and had higher mutation burden, immune cell infiltration, and PD-1/PD-L1 expression. Furthermore, we identified four genes associated with relapse-free survival in EOCRC, including SERPINE1, PECAM1, CDH1, and ANXA1. They were consistently differentially expressed at the transcriptome and proteome levels between high-risk and low-risk EOCRCs. They were also involved in regulating cancer progression and immune microenvironment in EOCRC. Notably, the expression of SERPINE1 and ANXA1 positively correlated with M2-like macrophage infiltration.

CONCLUSION

Our results indicate that 6-GPS can robustly predict the recurrence risk of EOCRC, and that SERPINE1, PECAM1, CDH1, and ANXA1 may serve as potential therapeutic targets. This study provides valuable information for the precision treatment of EOCRC.

StemSC: a cross-dataset human stemness index for single-cell samples

Background

Stemness is defined as the potential of cells for self-renewal and differentiation. Many transcriptome-based methods for stemness evaluation have been proposed. However, all these methods showed low negative correlations with differentiation time and can’t leverage the existing experimentally validated stem cells to recognize the stem-like cells.

Methods

Here, we constructed a stemness index for single-cell samples (StemSC) based on relative expression orderings (REO) of gene pairs. Firstly, we identified the stemness-related genes by selecting the genes significantly related to differentiation time. Then, we used 13 RNA-seq datasets from both the bulk and single-cell embryonic stem cell (ESC) samples to construct the reference REOs. Finally, the StemSC value of a given sample was calculated as the percentage of gene pairs with the same REOs as the ESC samples.

Results

We validated the StemSC by its higher negative correlations with differentiation time in eight normal datasets and its higher positive correlations with tumor dedifferentiation in three colorectal cancer datasets and four glioma datasets. Besides, the robust of StemSC to batch effect enabled us to leverage the existing experimentally validated cancer stem cells to recognize the stem-like cells in other independent tumor datasets. And the recognized stem-like tumor cells had fewer interactions with anti-tumor immune cells. Further survival analysis showed the immunotherapy-treated patients with high stemness had worse survival than those with low stemness.

Conclusions

StemSC is a better stemness index to calculate the stemness across datasets, which can help researchers explore the effect of stemness on other biological processes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13287-022-02803-5.

A novel qualitative signature based on lncRNA pairs for prognosis prediction in hepatocellular carcinoma

Background

Prognostic assessment is imperative for clinical management of patients with hepatocellular carcinoma (HCC). Most reported prognostic signatures are based on risk scores summarized from quantitative expression level of candidate genes, which are vulnerable against experimental batch effects and impractical for clinical application. We aimed to develop a robust qualitative signature to assess individual survival risk for HCC patients.

Methods

Long non-coding RNA (lncRNA) pairs correlated with overall survival (OS) were identified and an optimal combination of lncRNA pairs based on the majority voting rule was selected as a classification signature to predict the overall survival risk in the cancer genome atlas (TCGA). Then, the signature was further validated in two external datasets. Besides, biomolecular characteristics, immune infiltration status, and chemotherapeutics efficacy of different risk groups were further compared. Finally, we performed key lncRNA screening and validated it in vitro.

Results

A signature consisting of 50 lncRNA pairs (50-LPS) was identified in TCGA and successfully validated in external datasets. Patients in the high-risk group, when at least 25 of the 50-LPS voted for high risk, had significantly worse OS than the low-risk group. Multivariate Cox, receiver operating characteristic (ROC) curve and decision curve analyses (DCA) demonstrated that the 50-LPS was an independent prognostic factor and more powerful than other available clinical factors in OS prediction. Comparison analyses indicated that different risk groups had distinct biomolecular characteristics, immune infiltration status, and chemotherapeutics efficacy. TDRKH-AS1 was confirmed as a key lncRNA and associated with cell growth of HCC.

Conclusions

The 50-LPS could not only predict the prognosis of HCC patients robustly and individually, but also provide theoretical basis for therapy. Besides, TDRKH-AS1 was identified as a key lncRNA in the proliferation of HCC. The 50-LPS might guide personalized therapy for HCC patients in clinical practice.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12935-022-02507-z.