A Combination of Faecal and Intratumour Microbial Community Profiling Reveals Novel Diagnostic and Prognostic Biomarkers for Pancreatic Tumours.
Wei Wang,Cheng Qian,Ting Wang,Yuetong Jiang,Yiran Zhou,Kaiyu Liu,Zhiyang Ma,Pengyi Liu,Yichi Wu,Leying Chen,Huaizhi Wang,Tingting Zhou
DOI: https://doi.org/10.1002/ctm2.1726
IF: 8.554
2024-01-01
Clinical and Translational Medicine
Abstract:Our findings provide a novel interpretive framework for diagnostic and prognostic biomarkers for pancreatic tumours. A multi-step combination of faecal and intratumour microbial classifiers has the potential to be clinical biomarkers in the early detection of pancreatic cancer, which has less invasiveness, higher sensitivity and specificity. Pancreatic tumours are extremely lethal diseases characterized by difficult early diagnosis and rapid metastasis.1 Histopathological examination of the biopsy specimens of pancreatic tumours obtained by endoscopy or surgery is considered as a gold diagnosis standard of pancreatic tumors.2 Nonetheless, pathological analysis requires very high accuracy in biopsy localization with multiple sampling, and the accuracy of sampling has a significant impact on misjudgment.3 Given diagnostic delays, cancer metastasis and adverse reactions of chemotherapy or radiotherapy, it is urgent to explore new strategies which are required to develop effective diagnostic and treatment options for improving clinical outcomes of pancreatic tumours. As a novel component in shaping the immune system, the microbiome has been increasingly shown to be important in the maintenance of homeostasis in several physiologic processes.4 Here, we used 193 faecal samples (156 patients with different grade of human pancreatic tumours and 37 healthy volunteers) and 362 fresh pancreatic tissues of malignant (PDAC), premalignant (IPMN1 and IPMN2) and benign (SCN) patients from different hospitals (training and validation cohorts) to identify the link between the abundance and diversity changes of faecal and intra-pancreatic microbes and the prognosis and prognostic of pancreatic tumours. The basic statistical conditions are as follows: the difference between the study group and the healthy control group is tested with a confidence level of 80% and a ratio of .1000 between groups. The study group is assumed to be .05 and .15 under the null hypothesis and alternative hypothesis, respectively, while the healthy control group is assumed to be 0.05. Use the method of bilateral Z-test and combined variance to test the statistics. The statistical difference was set to .05. Based on previous foreign data and research on gut microbiota in adolescents with idiopathic chronic pancreatitis, two lowest and most conservative data, a positive rate of 30% in the study group and 10% in the healthy control group, were used for estimation. The required sample size for the study group (PDAC) was 59 cases. Considering factors such as errors, the final number of cases was determined to be 90. At the phylum level, faecal samples from different grade of pancreatic lesion and healthy control groups are relatively close, indicating that the sample composition is similar (Figure 1A, B). To further investigate these findings, we contrasted the microbial genera between healthy controls and pancreatic tumour patients including SCN, IPMN1, IPMN2 and PDAC. Interestingly, linear discriminant analysis (LDA) revealed that the abundance of specific bacterial genera in stools, including Faecalibacterium, Eubacterium, Bacteroides, Blautia, Streptococcus and Ruminococcus, may be used as non-invasive biomarkers to early diagnose the multistep progression of pancreatic tumours (Figure 1C–E). To determine the differences of the microbiome composition among SCN, IPMN1, IPMN2 and PDAC pancreatic tumours, we interrogated the bacterial profiles of fresh pancreatic tissues samples from PDAC and various non-malignant pancreatic tumour patients. Although groups did not separate on PCA based on their microbiomes, similar to the observations that are derived from alpha-diversity index data, PDAC and the related NAT groups showed the greatest separation from the other groups (Figure S1). Considering the potential connection between pancreatic microbiome diversity and malignant degree of pancreatic tumours, we next need to determine the differences of the pancreas microbial composition between non-malignant pancreatic tumours and PDAC. Notably, the distribution of genus-level revealed marked changes in the bacterial composition of the different pancreatic tumour types (Figure 2A). Bacteria belonging to the Pseudomonas and Ralstonia genera were the most abundant species in PDAC and their NATs. In contrast, Rhodococcus and Brucella were more abundant in the non-malignant pancreas (IPMN1, IPMN2, SCN) and their related NAT groups. We also detected the proportions of significantly increased genera in different pancreatic tumour types (Figure 2B). Pseudomonas was the dominant genus, but no significant differences were observed among different cohorts. However, Rhodococcus was abruptly enriched in the nonmalignant pancreas cohorts, whereas Brucella was more abundant in SCN and their NAT. Although Acinetobacter was present in low abundance in the pancreas, it increased obvious abundance in the IPMN1 cohorts (Figure 2C). We used the patient's risk score as a variable to observe whether the abundance of OUTs at the genus level can segregate the intra-tumour microbiome (Figure 3A). Next, we attempted to assess the mean relative abundance differences of species from both benign and premalignant to malignant pancreatic tumours which might increase potential value of diagnosing the grades of malignancy. We visualized a significant differential segregation between the communities of PDAC and non-malignant pancreatic tumours according to the risk scores, and found PDAC was associated with a higher-risk value than the control groups (Figure 3B and Figure S2). The 12-bacteria signature (Rhodococcus, Brucella, Pseudomonas, Kiebsiella, Lactococcus, Ruminococcus, Brevundimonas, Paracoccus, Amaricoccus, Ralstonia, Acinetobacter and Sphingomonas) had a more favourable performance in the trained cohorts, which then validated in the tested cohorts. The results were further confirmed by the external validation study including another centre to determine diagnostic biomarkers (Figure 3C). We next tested pancreatic tissues to confirm our findings obtained in intra-tumour microbiom by histologic analysis. These results all indicated that these twelve genera with different abundances had an obvious effect in distinguishing PDAC and malignant pancreatic tissues and could be used as diagnostic biomarkers with a certain degree of accuracy. To explore the relationship between the human intra-pancreatic microbiome composition and the survival rate of PDAC patients, we carried out Kaplan–Meier survival analysis and machine learning analysis. The result suggested that the survivorship of PDAC patients could be influenced and predicted by the abundance of 11 taxonomic communities, including Blautia, Faecalibacterium, Comamonas, Coprococcus, Erysipelatoclostridium, Sulfuritalea, Thermomonas, Phenylobacterium, Raoultibacter, Erysipelothrix, Thiothrix (Figure S3). There is evidence to suggest that changes in the composition of the gut microbiome are a characteristic of metabolic disorders, meanwhile the metabolic syndrome has a significant impact on our health and could increase the probability of developing cancer.5-7 To assess if the faecal and pancreatic microbiota have relationship with the metabolic pathways, reference genome databases were used to predict metagenomes through PICRUSt analysis (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States). Ten core functional modules were identified in faecal samples, meanwhile 30 core functional modules were identified in tissue samples of SCN, IPMN1, IPMN2 and PDAC cohorts including related NAT cohorts (Figure 4A,B). Twenty distinct intrapancreatic phyla were detected in PDAC Grade2 and Grade3, and the potential interactions were predicted between them, suggesting that the distinct phyla may support PDAC progression through various metabolic mechanisms(Figure 4C–F). In conclusion, composite microbial communities may serve as diagnostic and prognostic biomarkers for pancreatic tumours in the early stage, and this strategy is a promising approach with less invasiveness, higher sensitivity and specificity. Wei Wang, Cheng Qian, Ting Wang, Yuetong Jiang, Yiran Zhou, Kaiyu Liu, Zhiyang Ma, Pengyi Liu, Yichi Wu, Leying Chen, Huaizhi Wang and Tingting Zhou conducted the experiment and analysed the data. Wei Wang carried out experiments. Cheng Qian wrote the initial manuscript. Ting Wang analysed the data. Yuetong Jiang helped revised the manuscript and designed the graphical abstract. Yiran Zhou, Kaiyu Liu, and Zhiyang Ma helped analyse 16S rRNA sequencing data. Pengyi Liu, Yichi Wu, Leying Chen helped to provide clinical sample and performed data analyses. Tingting Zhou conceived and initiated the project. Wei Wang, Tingting Zhou and Huaizhi Wang revised the manuscript. Tingting Zhou and Cheng Qian directed the project. Wei Wang was responsible for the drafts of the manuscript, as well as all requested revisions. This work was financially supported by the National KEY R&D Program of China (2017YFC1308600), the National Natural Science Foundation of China (82072723), the National Natural Science Foundation of China (81170332) and Guangci Distinguished Young Scholars Training Program (GCQN-2019-B16). The authors declared no competing interests. This study was approved by the Institutional Review Board (IRB) of Shanghai Jiao Tong University, and the protocols were approved by the Committee of Human Subjects Protection of Ruijin Hospital. Informed consent was obtained from the parents of all recruited children. The clinical trial registry number is NCT03809247 (https://register.clinicaltrials.gov/). Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.