Presented by: Taylor Lander
View Abstract
Background: Several studies have reported the importance of the human microbiome in the overall health of its host. While recent studies have explored the microbiome’s role in various types of cancer compared to healthy patients, this study narrows the focus to pancreatic cancer. This study aims to characterize the skin microbiomes on the forehead and cheek of individuals from three groups: 1) patients with pancreatic cancer, 2) patients with other forms of cancer, and 3) patients without any form of cancer. The goal is to determine if the results from this trial could provide insight on associations of microbial flora with the state or severity of cancer, status of host immune system, or progress of an ongoing therapy, which could have therapeutic applications.
Methods: A total of 58 participants were enrolled in the study. Participants were given a questionnaire that prompted them to provide information including age, gender, ethnicity, race, weight, height, and status of skin health. An additional 60 control samples were drawn from an existing broader database of healthy skin samples at ProdermIQ to supplement the analysis. The participants were enrolled from three groups: cancer patients with pancreatic cancer, cancer patients with other types of cancer, and individuals without cancer. Skin microbiome samples from the forehead and cheek collection sites were processed and then analyzed by incorporating both statistical methods and machine learning techniques.
Results: A total of 150 samples were analyzed, including 79 samples from subjects with cancer and 71 samples from control subjects. The mean age of the control group was 60 years, and the mean age of the cancer group was 63 years. Characterization of the two analysis groups was further refined using observed features and alpha diversity metrics. The cancer group displayed a significantly higher mean alpha diversity compared to the control group. Our analysis showed that the following organisms were the most abundant across all samples: Cutibacterium acnes PMH5, Streptococcus sanguinis SK353, Staphylococcus aureus subsp. aureus NN50, Streptococcus mitis SK642, Snograssella alvi wkB12, Staphylococcus epidermidis NW32, Streptococcus anginosus ChDC B695, Streptococcus gordonii Challis CH1, Kingella oralis UB-38, Streptococcus porci DSM 23759, Cutibacterium acnes HL411PA1, Corynebacterium kroppenstedtii DSM 44385, Corynebacterium diphtheriae sv. mitis B-D-16-78, Gardnerella vaginalis 315-A, and Cutibacterium acnes HL053PA1. Organisms such as Streptococcus
mitis SK642, Snograssella alvi wkB12, and Streptococcus gordonii Challis CH1 were seen in abundance within the pancreatic and other cancer groups but not within the no cancer group. Streptococcus porci DSM 23759 and Kingella oralis UB-38 were seen significantly within the pancreatic and no cancer group but not within the other cancer group. Additionally, a machine learning classification model built on the microbiome data demonstrated a median F1 Score of 0.761 for accurately classifying the cancer (all types) versus control samples. Given that F1 scores above 0.70 are generally regarded as satisfactory, this result indicates the skin microbiome can be predictive of cancer status.
Conclusion: This analysis showed that there were significant differences in the skin microbiome of cancer patients versus patients without cancer. The cancer groups showed an increase in alpha diversity versus the no cancer group, and the machine learning model achieved a satisfactory F1 Score for differentiating the control and cancer samples . This could indicate the presence of dysbiosis in cancer subjects’ skin microbiomes due to their clear differentiation from the healthy skin microbiomes. Additional research could provide potential opportunities to develop biomarkers that can identify pancreatic and other types of cancer.
If you have any questions regarding the poster, feel free to reach out here.