Recently, the journal Nature Communications published an article entitled 'The DNA methylome of cervical cells can predict the presence of ovarian cancer'. The authors used 850K microarray data of cervical exfoliated cells from ovarian cancer patients and healthy individuals to find differential methylation. Finally, the 14,000 sites with the most significant methylation differences were included to construct a classification index (WID-OC-index). WID-OC-index can effectively distinguish ovarian cancer samples from normal samples. The WID-OC-index correlated with age and BRAC1 mutations, however, the authors found that the classification performance of the WID-OC-index was independent of the presence or absence of ovarian cancer, i.e., in the absence of ovarian cancer DNA in cervical exfoliated cell samples, The WID-OC-index was still able to distinguish ovarian cancer samples from normal samples, which the authors ultimately attributed to an apparent developmental defect during development.
The research is part of the Forcee programme, which is funded by the European Commission and led by the UCL Women's Cancer Unit. Its vision is to develop a screening test designed to prevent four high incidence gynecological cancers - breast, uterine, ovarian and cervical. A total of 2747 clinical cervical exfoliated cell samples from 5 centers were included in this study (Figure 1). The Illunima 850K chip was used to detect the methylation level of the samples. The data set was divided into a discovery set and an external validation set. In the external validation set, except for ovarian Cancer datasets, which also include endometrial cancer, breast cancer, and datasets carrying BRAC mutations.
figure 1.
In the process of searching for differentially methylated sites, the authors removed the interference of immune cells and only selected CG sites with the most significant differences in epithelial cells (Figure 2b and c) and sorted them, using Lasso respectively in the internal validation set and Ridge regression to evaluate the AUC values of the top CG loci with different numbers. The results show that the classification effect of Ridge regression is better than that of Lasso (Figure 2d). When the number of CG loci is 14,000, the AUC value is the largest. The model is used as the classification indicator WID-OC-index. Further, the AUC of WID-OC-index to distinguish ovarian cancer and normal samples in the internal validation set was 0.78 (Fig. 2e).
figure 2
In the external validation set, the AUC value of WID-OC-index to distinguish ovarian cancer and normal samples was 0.76 (Fig. 3b), the AUC value to distinguish endometrial cancer samples from normal samples was 0.90 (Fig. 3d), and the AUC value to distinguish between breast cancer samples and normal samples was 0.90 (Fig. 3d). The AUC value of normal samples was 0.68 (Fig. 3f), and the AUC value of distinguishing healthy individuals with and without BRAC1 mutation was 0.62 (Fig. 3h).
image 3
The WID-OC-index was positively correlated with age (Fig. 4a) and correlated with tumor stage, and the WID-OC-index of advanced patients was greater than that of early-stage patients (Fig. 4f), and there was no significant correlation with other clinical factors.
Figure 4
The above shows that WID-OC-index has certain diagnostic value for ovarian cancer, endometrial cancer and breast cancer. However, surprisingly, the author used HEpiDISH algorithm to analyze the proportion of ovarian cancer DNA in cervical exfoliated cells and found that Ovarian cancer DNA was absent from cervical exfoliated cells from both ovarian cancer patients and healthy individuals (Figure 5). The authors speculate that the WID-OC-index signal may originate from epigenetic developmental defects.
Figure 5
WID-OC-index may be an imprint of embryonic development. Given its positive correlation with age, it may also be affected by acquired factors. As a quantitative indicator of ovarian cancer risk, its clinical application value still needs further evaluation.
About Emerson Life Technology
Emerson Life Technology focuses on the early detection of high-incidence malignant tumors, providing digestive system tumors (intestinal cancer, gastric cancer, liver cancer, esophageal cancer, pancreatic cancer), gynecological tumors (cervical cancer, endometrial cancer), urinary system tumors (bladder cancer) , Pan-cancer species and other high-incidence malignant tumor genetic testing services.
The company has been based on local development for many years, and its core technology is independent and controllable. After several years of accumulation, a diversified product detection system with methylation navigation technology as the core has been established; after tens of thousands of clinical sample verifications, the screening sensitivity and specificity have reached the industry-leading level! The company has applied for 55 patents and published more than 10 cutting-edge research results. It has become one of the few multi-type cancer screening service providers in the world and an early tumor detection company that licenses external technology license-out.
