Editors: Donna E. Hansel, MD, PhD, chair of pathology, Oregon Health and Science University, Portland; Richard D. Press, MD, PhD, professor and director of molecular pathology, OHSU; James Solomon, MD, PhD, assistant professor, Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York; Sounak Gupta, MBBS, PhD, senior associate consultant, Mayo Clinic, Rochester, Minn.; Tauangtham Anekpuritanang, MD, molecular pathology fellow, Department of Pathology, OHSU; Hassan Ghani, MD, molecular genetic pathology fellow, Department of Pathology, OHSU; and Fei Yang, MD, assistant professor, Department of Pathology, OHSU.
Tumor microbiome diversity and composition: influence on pancreatic cancer outcomes
November 2019—Pancreatic adenocarcinoma has a dismal prognosis, with a high incidence of relapse and a median overall survival of 24 to 30 months. Only nine percent of patients are alive five years after surgery. Genome-wide mutational landscape studies to decipher the factors that contribute to long-term survival have been futile. Recent studies in patients with melanoma and lung cancer have shown that the gut microbiota can mediate tumor responses to chemotherapy and immunotherapy, influencing overall outcome. Previous data (Balachandran, et al.) have shown that pancreatic adenocarcinoma (PDAC) tumors from long-term survivors harbor a high quantity of neoantigens with a profile similar to that of microbial epitopes. To further explore this concept, Riquelme, et al., performed a comprehensive analysis at the MD Anderson Cancer Center, in which they compared the tumor microbiome of PDAC patients with long-term survival (more than five years; average overall survival, 10.14 years) and short-term survival (less than five years; average overall survival, 1.62 years) using the 16S ribosomal community profiling method. The results revealed that the long-term survivors’ tumors had significantly higher intratumor microbiome diversity compared with the short-term survivors’ tumors. Furthermore, the long-term survivors’ tumors had a distinctive tumor microbiota signature that is enriched for specific bacterial taxa: Saccharopolyspora, Pseudoxanthomonas, and Streptomyces. Adding a fourth bacterial species, Bacillus clausii, which is also enriched in the long-term survivors’ tumors, created a microbiome signature that was strongly associated with long-term survivorship in PDAC patients. The same microbiota signature was also found to be predictive for long-term PDAC survival in an independent validation cohort from a geographically different medical center, Johns Hopkins Hospital. The area under the curve for predicting long-term survival was 97.51 and 99.17 percent in the discovery and validation cohorts, respectively, using the signature of the four bacterial species. To assess whether the gut microbiome could be modulated and in turn influence the intratumor microbiota and affect tumor growth, Riquelme, et al., conducted a series of experiments. First, the authors compared the microbiome of the PDAC tumor and adjacent normal tissue to that in matched stool samples and identified the presence of gut microbiome within the tumor, suggesting there is crosstalk between the human gut and the PDAC intratumor microbiota. Second, fecal microbial transplantation (FMT) studies were performed in mice that were subsequently challenged with PDAC implantation. Interestingly, the mice that received the FMT from the long-term survivor patients demonstrated a 70 percent reduction in tumor growth compared with the mice that received the FMT from the short-term survivor donors. The authors then assessed the role of the immune system in modulating tumor growth. They found that the long-term survivor patients had an immunoactivation profile with higher tissue density of CD8+ T cells and higher levels of serum cytokines than the short-term survivor patients. Similar findings were observed in the mice that received FMT from the long-term survivor patients. Bacterial ablation using short-term antibiotics or CD8+ T-cell depletion with neutralizing antibodies induced larger tumors in these mice. Statistical analysis revealed that CD8+ T-cell tissue density was strongly correlated with tumor microbiome diversity, the unique microbiome signature in long-term survivor patients, and the overall survival of PDAC patients. These results suggest that tumor microbiome diversity may contribute to anti-tumor immune response by favoring recruitment of CD8+ T cells to the tumor milieu and their activation. In summary, Riquelme, et al., identified a tumor microbial signature unique to pancreatic cancer long-term survival that could potentially be used as a predictive biomarker. Moreover, due to the gut-to-tumor microbial crosstalk, fecal microbial transplant can be considered a potential therapeutic option to create a favorable tumor microenvironment in pancreatic cancer patients.
Balachandran VP, Luksza M, Zhao JN, et al. Identification of unique neoantigen qualities in long-term pancreatic cancer survivors. Nature. 2017;551(7681):512–516.
Correspondence: Dr. Vinod P. Balachandran at balachav@mskcc.org
Riquelme E, Zhang Y, Zhang L, et al. Tumor microbiome diversity and composition influence pancreatic cancer outcomes. Cell. 2019;178(4):795–806.
Correspondence: Dr. Florencia McAllister at fmcallister@md anderson.org
Dynamic risk profiling using serial tumor biomarkers for personalized outcome prediction
Prognostication of outcomes for cancer patients using traditional clinical, pathological, and imaging techniques has been greatly improved with the advent of personalized oncology, in which genomic data further defines patient subgroups for targeted therapies or systemic risk-stratified therapies based on prognosis. Despite such advances, a substantial minority of cancer patients do not respond optimally to their therapeutic regimen due, in part, to significant heterogeneity in most cancer subtypes. The traditional cancer risk-stratification tools have primarily focused on pretreatment biomarkers. Other innovative tools, such as liquid biopsies, multiparameter flow cytometry, and PET/CT imaging, allow on-treatment serial monitoring of tumor responses to specific therapies. However, these tools provide risk stratification based on an assessment at a fixed time point, acting as independent predictors. The authors of this study proposed a new prognostic framework—the Continuous Individualized Risk Index (CIRI)—that integrates pretreatment and on-treatment biomarkers and estimates outcomes at any given time for a patient. The CIRI model provides a personalized prediction of outcomes over time that uses an initial naive Bayes method to predict outcomes at a fixed endpoint, plus a second method of calculation based on Bayesian analysis and proportional hazard assumptions. In each approach, the model is continuously updated as information is gathered over a disease course. In a proof-of-concept analysis, the authors applied the CIRI model to three different malignancies that have well-established prognostic systems: diffuse large B-cell lymphoma (DLBCL), breast adenocarcinoma, and chronic lymphocytic leukemia (CLL). Using the current prediction indices as the gold standard, the composite CIRI model demonstrated superior performance in outcome prediction for each disease. Based on the validation data in these three cancer models, the authors confirmed that several features of the CIRI model are more advanced than those of the conventional risk models. First, the CIRI model can leverage a limited number of predictors or prior knowledge of the biomarkers without the need for a large number of training cases. This feature is especially useful for utilizing emerging biomarkers that are not long-established in the literature. Second, the performance of the CIRI model remained robust even with missing data, which is commonplace in the clinic. Third, the predictive biomarkers, such as cfDNA and multiparameter flow cytometry for minimal residual disease monitoring, often provide prognostic value but are unable to inform the follow-up therapeutic regimen. Conversely, the CIRI model could aid in therapy selection by providing quantitative estimates of likely outcomes, as illustrated in CLL. With the latter disease, the CIRI-CLL model calculated the probability of therapeutic benefit based on interim data on minimal residual disease and identified a subset of patients who would preferentially benefit from an FCR (fludarabine, cyclophosphamide, and rituximab) treatment regimen compared to alternative immunochemotherapies. By retrospectively applying a novel dynamic risk profiling model to patient cohorts for three types of cancer, the authors demonstrated the advantages of continuous or longitudinal integration of a diverse source of risk factors when making predictions and selecting therapy. Prospective clinical trials would be the next step to confirm the clinical utility of the CIRI model in personalized outcome prediction.
Kurtz DM, Esfahani MS, Scherer F, et al. Dynamic risk profiling using serial tumor biomarkers for personalized outcome prediction. Cell. 2019;178(3):699–713.
Correspondence: Dr. Maximilian Diehn at diehn@stanford.edu or Dr. Ash A. Alizadeh at arasha@stanford.edu