Hao Cai

Differential expression analysis for individual cancer samples based on robust within-sample relative gene expression orderings across multiple profiling platforms

The highly stable within-sample relative expression orderings (REOs) of gene pairs in a particular type of human normal tissue are widely reversed in the cancer condition. Based on this finding, we have recently proposed an algorithm named RankComp to detect differentially expressed genes (DEGs) for individual disease samples measured by a particular platform. In this paper, with 461 normal lung tissue samples separately measured by four commonly used platforms, we demonstrated that tens of millions of gene pairs with significantly stable REOs in normal lung tissue can be consistently detected in samples measured by different platforms. However, about 20% of stable REOs commonly detected by two different platforms (e.g., Affymetrix and Illumina platforms) showed inconsistent REO patterns due to the differences in probe design principles. Based on the significantly stable REOs (FDR<0.01) for normal lung tissue consistently detected by the four platforms, which tended to have large rank differences, RankComp detected averagely 1184, 1335 and 1116 DEGs per sample with averagely 96.51%, 95.95% and 94.78% precisions in three evaluation datasets with 25, 57 and 58 paired lung cancer and normal samples, respectively. Individualized pathway analysis revealed some common and subtype-specific functional mechanisms of lung cancer. Similar results were observed for colorectal cancer. In conclusion, based on the cross-platform significantly stable REOs for a particular normal tissue, differentially expressed genes and pathways in any disease sample measured by any of the platforms can be readily and accurately detected, which could be further exploited for dissecting the heterogeneity of cancer.

Robust transcriptional tumor signatures applicable to both formalin-fixed paraffin-embedded and fresh-frozen samples

Formalin-fixed paraffin-embedded (FFPE) samples represent a valuable resource for clinical researches. However, FFPE samples are usually considered an unreliable source for gene expression analysis due to the partial RNA degradation. In this study, through comparing gene expression profiles between FFPE samples and paired fresh-frozen (FF) samples for three cancer types, we firstly showed that expression measurements of thousands of genes had at least two-fold change in FFPE samples compared with paired FF samples. Therefore, for a transcriptional signature based on risk scores summarized from the expression levels of the signature genes, the risk score thresholds trained from FFPE (or FF) samples could not be applied to FF (or FFPE) samples. On the other hand, we found that more than 90% of the relative expression orderings (REOs) of gene pairs in the FF samples were maintained in their paired FFPE samples and largely unaffected by the storage time. The result suggested that the REOs of gene pairs were highly robust against partial RNA degradation in FFPE samples. Finally, as a case study, we developed a REOs-based signature to distinguish liver cirrhosis from hepatocellular carcinoma (HCC) using FFPE samples. The signature was validated in four datasets of FFPE samples and eight datasets of FF samples. In conclusion, the valuable FFPE samples can be fully exploited to identify REOs-based diagnostic and prognostic signatures which could be robustly applicable to both FF samples and FFPE samples with degraded RNA.

An individualized prognostic signature and multi‑omics distinction for early stage hepatocellular carcinoma patients with surgical resection

Previously reported prognostic signatures for predicting the prognoses of postsurgical hepatocellular carcinoma (HCC) patients are commonly based on predefined risk scores, which are hardly applicable to samples measured by different laboratories. To solve this problem, using gene expression profiles of 170 stage I/II HCC samples, we identified a prognostic signature consisting of 20 gene pairs whose within-sample relative expression orderings (REOs) could robustly predict the disease-free survival and overall survival of HCC patients. This REOs-based prognostic signature was validated in two independent datasets. Functional enrichment analysis showed that the patients with high-risk of recurrence were characterized by the activations of pathways related to cell proliferation and tumor microenvironment, whereas the low-risk patients were characterized by the activations of various metabolism pathways. We further investigated the distinct epigenomic and genomic characteristics of the two prognostic groups using The Cancer Genome Atlas samples with multi-omics data. Epigenetic analysis showed that the transcriptional differences between the two prognostic groups were significantly concordant with DNA methylation alternations. The signaling network analysis identified several key genes (e.g. TP53, MYC) with epigenomic or genomic alternations driving poor prognoses of HCC patients. These results help us understand the multi-omics mechanisms determining the outcomes of HCC patients.

Identification of reproducible drug-resistance-related dysregulated genes in small-scale cancer cell line experiments

Researchers usually measure only a few technical replicates of two types of cell line, resistant or sensitive to a drug, and use a fold-change (FC) cut-off value to detect differentially expressed (DE) genes. However, the FC cut-off lacks statistical control and is biased towards the identification of genes with low expression levels in both cell lines. Here, viewing every pair of resistant-sensitive technical replicates as an experiment, we proposed an algorithm to identify DE genes by evaluating the reproducibility of the expression difference or FC between every two independent experiments without overlapping samples. Using four small datasets of cancer cell line resistant or sensitive to a drug, we demonstrated that this algorithm could efficiently capture reproducible DE genes significantly enriched in biological pathways relevant to the corresponding drugs, whereas many of them could not be found by the FC and other commonly used methods. Therefore, the proposed algorithm is an effective complement to current approaches for analysing small cancer cell line data.

Circumvent the uncertainty in the applications of transcriptional signatures to tumor tissues sampled from different tumor sites

The expression measurements of thousands of genes are correlated with the proportions of tumor epithelial cell (PTEC) in clinical samples. Thus, for a tumor diagnostic or prognostic signature based on a summarization of expression levels of the signature genes, the risk score for a patient may dependent on the tumor tissues sampled from different tumor sites with diverse PTEC for the same patient. Here, we proposed that the within-samples relative expression orderings (REOs) based gene pairs signatures should be insensitive to PTEC variations. Firstly, by analysis of paired tumor epithelial cell and stromal cell microdissected samples from 27 cancer patients, we showed that above 80% of gene pairs had consistent REOs between the two cells, indicating these REOs would be independent of PTEC in cancer tissues. Then, by simulating tumor tissues with different PTEC using each of the 27 paired samples, we showed that about 90% REOs of gene pairs in tumor epithelial cells were maintained in tumor samples even when PTEC decreased to 30%. Especially, the REOs of gene pairs with larger expression differences in tumor epithelial cells tend to be more robust against PTEC variations. Finally, as a case study, we developed a gene pair signature which could robustly distinguish colorectal cancer tissues with various PTEC from normal tissues. We concluded that the REOs-based signatures were robust against PTEC variations.

An individualized prognostic signature for gastric cancer patients treated with 5-Fluorouracil-based chemotherapy and distinct multi-omics characteristics of prognostic groups

5-Fluorouracil (5-FU)-based chemotherapy is currently the first-line treatment for gastric cancer. In this study, using gene expression profiles for a panel of cell lines with drug sensitivity data and two cohorts of patients, we extracted a signature consisting of two gene pairs (KCNE2 and API5, KCNE2 and PRPF3) whose within-sample relative expression orderings (REOs) could robustly predict prognoses of gastric cancer patients treated with 5-FU-based chemotherapy. This REOs-based signature was insensitive to experimental batch effects and could be directly applied to samples measured by different laboratories. Taking this unique advantage of the REOs-based signature, we classified gastric cancer samples of The Cancer Genome Atlas (TCGA) into two prognostic groups with distinct transcriptional characteristics, circumventing the usage of confounded TCGA survival data. We further showed that the two prognostic groups displayed distinct copy number, gene mutation and DNA methylation landscapes using the TCGA multi-omics data. The results provided hints for understanding molecular mechanisms determining prognoses of gastric cancer patients treated with 5-FU-based chemotherapy.

Identifying Reproducible Molecular Biomarkers for Gastric Cancer Metastasis with the Aid of Recurrence Information

To precisely diagnose metastasis state is important for tailoring treatments for gastric cancer patients. However, the routinely employed radiological and pathologic tests for tumour metastasis have considerable high false negative rates, which may retard the identification of reproducible metastasis-related molecular biomarkers for gastric cancer. In this research, using three datasets, we firstly shwed that differentially expressed genes (DEGs) between metastatic tissue samples and non-metastatic tissue samples could hardly be reproducibly detected with a proper statistical control when the metastatic and non-metastatic samples were defined by TNM stage alone. Then, assuming that undetectable micrometastases are the prime cause for recurrence of early stage patients with curative resection, we reclassified all the “non-metastatic” samples as metastatic samples whenever the patients experienced tumour recurrence during follow-up after tumour resection. In this way, we were able to find distinct and reproducible DEGs between the reclassified metastatic and non-metastatic tissue samples and concordantly significant DNA methylation alterations distinguishing metastatic tissues and non-metastatic tissues of gastric cancer. Our analyses suggested that the follow-up recurrence information for patients should be employed in the research of tumour metastasis in order to decrease the confounding effects of false non-metastatic samples with undetected micrometastases.

Identifying clinically relevant drug resistance genes in drug-induced resistant cancer cell lines and post-chemotherapy tissues.

Until recently, few molecular signatures of drug resistance identified in drug-induced resistant cancer cell models can be translated into clinical practice. Here, we defined differentially expressed genes (DEGs) between pre-chemotherapy colorectal cancer (CRC) tissue samples of non-responders and responders for 5-fluorouracil and oxaliplatin-based therapy as clinically relevant drug resistance genes (CRG5-FU/L-OHP). Taking CRG5-FU/L-OHP as reference, we evaluated the clinical relevance of several types of genes derived from HCT116 CRC cells with resistance to 5-fluorouracil and oxaliplatin, respectively. The results revealed that DEGs between parental and resistant cells, when both were treated with the corresponding drug for a certain time, were significantly consistent with the CRG5-FU/L-OHP as well as the DEGs between the post-chemotherapy CRC specimens of responders and non-responders. This study suggests a novel strategy to extract clinically relevant drug resistance genes from both drug-induced resistant cell models and post-chemotherapy cancer tissue specimens.

Tamoxifen therapy benefit predictive signature coupled with prognostic signature of post-operative recurrent risk for early stage ER+ breast cancer

Two types of prognostic signatures for predicting recurrent risk of ER+ breast cancer patients have been developed: one type for patients accepting surgery only and another type for patients receiving post-operative tamoxifen therapy. However, the first type of signature cannot distinguish high-risk patients who cannot benefit from tamoxifen therapy, while the second type of signature cannot identify patients who will be at low risk of recurrence even if they accept surgery only. In this study, we proposed to develop two coupled signatures to solve these problems based on within-sample relative expression orderings (REOs) of gene pairs. Firstly, we identified a prognostic signature of post-operative recurrent risk using 544 samples of ER+ breast cancer patients accepting surgery only. Then, applying this drug-free signature to 840 samples of patients receiving post-operative tamoxifen therapy, we recognized 553 samples of patients who would have been at high risk of recurrence if they had accepted surgery only and used these samples to develop a tamoxifen therapy benefit predictive signature. The two coupled signatures were validated in independent data. The signatures developed in this study are robust against experimental batch effects and applicable at the individual levels, which can facilitate the clinical decision of tamoxifen therapy.

Discriminating cancer-related and cancer-unrelated chemoradiation-response genes for locally advanced rectal cancers

For patients with locally advanced rectal cancer (LARC) treated with preoperation chemoradiation (pCRT), identifying differentially expressed (DE) genes between non-responders and responders is a common approach for investigating mechanisms of chemoradiation resistance. However, some of such DE genes might be irrelevant to cancer itself but simply reflect the pharmacokinetic differences of the normal tissues. In this study, we adopted the RankComp algorithm to identify DE genes for each of LARC sample compared with its own normal state. Then, we identified genes with significantly different deregulation frequencies between the non-responders and responders, defined as cancer-related pCRT-response genes. Pathway enrichment and protein-protein interaction analyses showed that these genes specifically and intensively interacted with currently known effective genes of pCRT, involving in DNA replication, cell cycle and DNA repair. In contrast, after excluding the cancer-related pCRT-response genes, the other DE genes between non-responders and responders were enriched in many pathways of drug and protein metabolisms and transports, and interacted with both the known effective genes and pharmacokinetic genes. Hence, these two types of DE genes should be distinguished for investigating mechanisms of pCRT response in LARCs.