Survival implications of the age-associated tumor and normal adjacent tissue microbiome among colorectal cancer patients
- Open Access
- 23.12.2025
- Research
Erschienen in:
Abstract
Translational relevance of the work.
Emerging evidence has highlighted the important role of the microbiome in colorectal cancer (CRC). Since the 1990s, there has been an increase in cases of early-onset colorectal cancer. However, there is still a limited understanding of the risk factors contributing to this rise. Investigating the associations between the microbiome of tumors and normal adjacent tissue in relation to aging offers a unique perspective on potential modifiable factors. Notably, our study has shown that age-related changes in the abundance of bacteria originating from the oral cavity, such as Porphyromonas, Peptostreptococcus, and Prevotella, are linked to CRC prognosis. These findings suggest that changes in the tissue microbiome with age may serve as prognostic markers for CRC and could help inform future prevention strategies that consider dietary and oral health interventions.
Anzeige
Introduction
The overall incidence of colorectal cancer (CRC), the second leading cause of cancer death for women and men combined in the United States, has declined over the past three decades thanks to screening and treatment improvements [1, 2]. However, CRC incidence among individuals younger than 50 years of age, herein termed early-onset CRC (EoCRC; based on prior recommendations for CRC screening in adults aged 50 to 75 years) [3], has nearly doubled since the 1990s. EoCRC now represents approximately 10–12% of all newly diagnosed CRC cases [2, 4, 5]. Some evidence supports that EoCRC tumors may have distinct molecular profiles from later-onset CRC tumors (LoCRC) [6, 7]; however, significant gaps remain in our understanding of the composition of potentially modifiable biomarkers, such as the tissue microbiome, across the age spectrum.
The gut microbiome encompasses both fecal and colorectal tissue counterparts, the latter of which may reflect microbes that colonize the mucosa [8]. There is strong biological plausibility for the role of the mucosal microbiome in the progression of colorectal carcinomas, such as through its role in regulating immunity [9]. The microbiome also changes as individuals age and, in addition, may be influenced by a host of factors that differ across birth cohorts, such as obesity, alcohol intake, and antibiotic use [10, 11]. Aging has also been shown to be associated with declines in microbiome diversity, composition, and function, potentially compromising the host-immune response and cellular integrity in tissues [10]. Recently published studies found that certain tissue bacteria like Ruminococcaceae and Fusobacterium were differentially abundant in tumor tissue among EoCRC compared to LoCRC patients [12‐14]. To date, no studies have assessed differences between the two age groups in normal adjacent tissue, which may more so reflect the microbial environment prior to tumor development [15]. Furthermore, the survival implications of such age-related microbiome differences are unclear, especially while considering important confounders and mediators, like stage, tumor anatomic site, and treatment. Studying the tissue microbiome also has unique methodological challenges, given the low amounts of microbial DNA relative to human DNA [16]. Several efforts have been made to disseminate the best practices for microbiome studies [17]. However, decontamination practices are not yet widely implemented, particularly in retrospective studies without quality control samples collected at the time of collection. Contamination is of particular concern as even small amounts of contaminant microbes can appear to be true biological signatures [18]. Though contamination is non-differential in nature, it is highly beneficial to proactively address potential contamination issues to provide a clearer picture of microbiome-disease associations.
Herein, using a careful approach to address contamination, we investigated the composition of the tumor and normal adjacent tissue microbiome across age groups among a cohort of CRC patients. We then investigated clinical implications of these microbiome differences by investigating associations with overall survival among CRC patients.
Materials and methods
Study population
We included individuals who were at least 18 years old and diagnosed with CRC at Moffitt Cancer Center (MCC). We included 46 EoCRC patients (defined as a patient diagnosed with CRC before the age of 50 [19]) and 101 LoCRC patients (diagnosed with CRC at or after the age of 50) who were frequency-matched to EoCRC patients based on sex, tumor site, race (non-Hispanic White vs. other), stage, and year of surgery (+/- 5 years). Patients included in this study comprised newly diagnosed stage I-IV colon or rectal cancer patients who had surgery at Moffitt and with sufficient archival tumor or normal adjacent tissue available for research and follow-up information available in the electronic medical record (EMR). Participants’ surgeries took place between 1995 and 2020, with data available until death, loss to follow-up, or last medical chart review on November 15, 2023. Participants provided written informed consent to the Total Cancer Care™ (TCC) protocol [20]. The Scientific Review Committee and the Institutional Review Board at Moffitt Cancer Center reviewed and approved this retrospective study (Advarra IRB #Pro00056038).
Anzeige
Data collection
Clinical and demographic data included tumor staging, lifestyle factors (i.e., smoking and alcohol), family history of cancer, treatment (including systemic and surgery), and vital status (assessed until November 2023). We use the American Joint Committee on Cancer (AJCC) Cancer Staging Manual (8th version, 2017) for staging [21, 22]. Moffitt ascertains vital status annually via the Florida Cancer Data System, National Death Index, cancer registry, and family reporting of patients’ passing [23].
Sample collection and processing
Tissue biospecimens were collected between 1995 and 2020. A total of 262 tumor and normal adjacent tissues (determined by gross diagnosis and including n = 106 macrodissected and n = 156 unprocessed tissues and n = 21 replicates) across n = 147 patients were collected at resection, snap frozen, and stored long-term in liquid nitrogen. Among these patients, 101 had both normal adjacent and tumor tissue (31 EoCRC, 70 LoCRC), eight patients had tumor tissue only, and 38 patients had normal adjacent tissue only (see Supplementary Fig. S1 for study selection flow chart). The normal adjacent tissue obtained was grossly uninvolved tissue located approximately 3 centimeters from the tumor tissue taken during resection and identified by a licensed pathologist. The selected tissue samples were pulled and aliquoted into ~ 25 mg chunks behind a protection shield using sterile tools and liquid nitrogen to keep the tissue from thawing.
DNA extraction and sequencing
We used 22–25 mg of fresh-frozen tumor and normal adjacent tissue from 46 EoCRC and 101 frequency-matched LoCRC patients for DNA extraction, including a total of 248 tumor and normal adjacent tissue samples. Samples from the same subject/environmental sampling site were included in the same extraction batch and otherwise were randomly distributed across batches. We also included three replicate samples dissected from the same tumor from seven patients across batches to assess intra-subject/tumor variability. The Tissue Service Core at H. Lee Moffitt Cancer Center manually extracted the samples for DNA using a modified protocol (supplementary methods) [26, 27].
After DNA extraction, 248 study samples qualified for further analyses based on sufficient DNA yield for 16 S rRNA gene sequencing. We sent 30uL aliquots of DNA on dry ice to MR DNA (Shallowater, TX, USA). DNA was treated with mammalian blockers (mitochondrial PNA, PNA Bio Inc, Thousand Oaks, CA), according to manufacturer’s instructions, to reduce the background of non-microbial DNA in the samples before undergoing 16 S rRNA gene sequencing. The 16 S rRNA gene V4 hypervariable regions (515f/806r) was sequenced in 30–35 cycles on the Illumina MiSeq 2 × 300, targeting 20,000 reads per sample. We multiplexed the DNA samples using unique dual indices and pooled samples together in equal proportions based on molecular weight and DNA concentration. We purified the samples using calibrated Ampure XP beads and used the pooled purified PCR product to prepare an Illumina DNA library.
Bioinformatics
The bioinformatics pipeline was performed by the Mr. DNA Molecular Research Laboratory. We joined and filtered out sequences < 150 bp or with ambiguous base calls. Sequences were quality-filtered using a maximum expected error threshold of 1.0 and dereplicated. The dereplicated or unique sequences were denoised; unique sequences identified with sequencing and/or PCR point errors and removed, followed by chimera removal, thereby providing a denoised amplicon sequence variant (ASVs). Final ASVs were taxonomically classified using BLASTn against a curated and routinely updated database derived from NCBI (www.ncbi.nlm.nih.gov). We used ASVs to generate relative abundance tables from phylum to genus level and to generate alpha diversity metrics (Shannon Index, observed ASVs, and Faith’s phylogenetic diversity) using the Phyloseq package [28] in R Studio (4.4.2). We also generated Bray-Curtis, weighted UniFrac, and unweighted UniFrac beta diversity distance matrices. The first principal coordinate analysis (PCoA) axes explained 4.7%, 10.3%, and 11.2% of the variation in the beta diversity matrix for Bray Curtis, weighted UniFrac, and unweighted UniFrac, respectively. We empirically determined the rarefaction threshold for alpha and beta diversity estimates based on a balanced assessment of rarefaction curves and retention of study samples with a final rarefaction level of 1000 reads based on the decontaminated data (Supplementary Fig. S2). Rarefaction removed 28 tumor and 24 normal adjacent samples from the analyses (final sample size N = 131 patients with a combined N = 196 tumor and/or normal adjacent tissue samples). Alpha diversity estimates were calculated based on the average of 10 rarefaction samplings and were standardized to a mean of 0 and a standard deviation of 1.
We estimated the relative abundance and presence/absence of multiple microbes. We selected a priori bacteria (e.g., Bacteroides, Fusobacterium, Porphyromonas, Parvimonas, Peptostreptococcus, Gemella, Prevotella, Solobacterium, and Dialister) based on literature on previously identified CRC intratumoral and surrounding tissue microbiomes [27, 29], and conducted exploratory analyses of (1) the relative abundance of bacteria present in ≥ 40% of samples at an average relative abundance of ≥ 0.1%; and (2) the presence/absence of bacteria present in 5% to 95% of the population. We normalized relative abundances using the centered-log ratio and standardized this value to a mean of 0 and a standard deviation of 1.
Quality control analysis
We compared results from aliquoted tissue samples from the same patient and assessed the environmental controls for potential contaminants [24, 25]. In the blank extraction samples, there was a median of 7,431 reads across 13 blank samples (11 blanks remained after rarefaction). Clustering by batch was shown most strongly via Unweighted Unifrac PCoA distance (though we included samples from the same participant in the same batch, which may have contributed to some of the clustering; Fig. 1A). We defined seven Unweighted UniFrac clusters based on visual inspection of the distances between PCoA axes 1 and 2 and using Ward’s Hierarchical clustering method with Euclidean distances [30]. We estimated which bacteria were differentially abundant between the clusters (e.g., Streptococcus, Ralstonia, and Bradyrhizobium) using Kruskal Wallis’ test. Some of these bacteria were also highly abundant in environmental control samples (e.g., Corynebacterium, Methylobacterium, Staphylococcus, Streptococcus, and Propionibacterium, Actinobaculum, Bradyrhizobium) and extraction blanks (e.g., Ralstonia, Staphylococcus, Paracoccus, Dechloromonas, Bibersteinia, Sphingobium, Rhizobium) (Fig. 1B). Using the Kruskal Wallis’ test, we identified several bacteria (e.g., Gaiella, Candidatus cloacimonas, Phyllobacterium, Methanomethylovorans, Pseudomonas, Rhizobium, Bibersteinia, Declomonas, Sphingobium, Actinobaculum) that were statistically significantly more abundant in environmental controls/extraction blanks than patient tissue samples (Supplementary Table S1). We also conducted a literature review to identify commonly reported contaminants in microbiome tissue studies. Among the most abundant genera in our environmental/blank samples, we specifically examined the relative abundance of Novosphingobium, Enterococcus, Aquabacterium, Sphingomonas, Granulicatella, Stenotrophomonas, Enhydrobacter, Brevundimonas, Tyzzeralla, and Thermatoga in environmental and extraction blanks [18, 24]. Based on these collective findings, we systematically removed potential contaminant genera, both highly abundant in our controls and based on the literature, and used the R package decontam [31] in R software to further identify potential contaminants using ASV and DNA concentration classification methods. We used a dual approach to remove ASVs identified as potential contaminants using the frequency-based score statistic and the prevalence-based score statistic methods (Fig. 1C) [32]. Briefly, the frequency-based score utilizes the sum of squared residuals and the number of observations to identify potential ASV contaminants based on a P statistic with a threshold of 0.1. This method flagged three ASVs as potential contaminants. In contrast, the prevalence-based score compares the prevalence of taxa in negative controls to that in actual samples, using a Chi-squared statistic P-value to identify potential contaminants; this method identified 190 ASVs. Removing contaminating bacterial genera and ASVs reduced clustering between batches in the Unweighted UniFrac distance PCoA plot (Fig. 1D).
Anzeige
Fig. 1
Quality control decontamination process for 109 tumor and 139 normal adjacent tissue microbiome samples from 147 CRC patients and quality controls. Summary of the decontamination process for (A) identifying clusters using Principal Coordinate Analysis based on Unweighted UniFrac distances. Seven clusters by batch were differentiated. Linear regression analyses were used to identify genera associated with sample type (environmental control, extraction blanks, and samples). (B) Bar plots of the relative abundance of the top 10 bacteria in the environmental control and extraction blank samples. (C) Prevalence threshold for the classification of potential contaminant ASVs using the R package “Decontam”. (D) Principal Coordinate analysis based on Unweighted UniFrac after decontamination processes and rarefaction showing improvement of batch effect. Rarefaction after decontamination steps removed 28 tumor and 24 normal adjacent tissue samples (N = 131 patients). QC, Quality controls
The percentage of the variability in beta diversity explained by subject and tissue characteristics before and after the decontamination process and rarefaction are presented in Fig. 2 as assessed using permutational ANOVA (PERMANOVA) tests with the Adonis Vegan function [33]. Variation was primarily explained by subject (e.g., unweighted UniFrac 82%, P = 0.01), while minimally explained by the other technical, clinical and tumor characteristics. Further, the percentage explained by batch was reduced after decontamination, especially for unweighted UniFrac distance.
Fig. 2
Percentage of variability before (N = 147) decontamination and rarefaction based on (A) Bray Curtis, (B) WeightedUniFrac, and (C) Unweighted UniFrac and after (N = 131) decontamination and rarefaction based on (A) Bray Curtis, (B) WeightedUniFrac, and (C) Unweighted UniFrac estimated by permutational multivariate analysis of variance PERMANOVA using the adonis function in the vegan package in R among CRC patients treated at Moffitt Cancer Center for surgical resection from 1995–2020
Statistical analyses
We compared microbiome metrics between tumor and normal adjacent tissue using a linear mixed model with participant ID as a random effect and tissue type as a fixed effect. We compared baseline participant demographics and clinical characteristics between EoCRC and LoCRC patients using Chi-squared tests and analysis of variance (ANOVA) for categorical and continuous variables, respectively, and Kruskal-Wallis tests for non-normally distributed continuous variables.
Anzeige
Given that age 50 is a somewhat arbitrary cutoff based on prior CRC screening recommendations [3], we used multivariable linear regression to estimate associations of age continuously (transformed into 10-year increments) with the microbiome metrics. We also used case-case multivariable logistic regression models to estimate odds ratios (ORs) and 95% confidence intervals (95% CI) for the multivariable associations of the microbiome features with EoCRC (relative to LoCRC) using age cut points at 1) </≥50 years old and 2) ≤ 45 years old and ≥ 60 years old.
To assess the clinical implications of the age-related differences in the tissue microbiome, we estimated associations of the microbiome metrics with overall survival. Follow-up time was calculated as the number of days from CRC surgery to the last date known alive or date of death according to medical records, National Death Index, Cancer Registry, or family member reports. We then estimated hazard ratios (HRs) and 95% confidence intervals (95% CI) using multivariable Cox proportional hazards regression models. We tested for the proportionality assumption using Schoenfeld residuals [34]. In tumor tissue models, standardized beta diversity PC axis 2 of Bray Curtis and 3 of unweighted UniFrac did not meet the proportionality assumption and were time-transformed with an interaction term with time in years [35]. To make the study population more representative of the source population, we calculated the inverse of the sampling fraction within each stratum of sex, race, year of surgery (5 years), tumor stage, and tumor location and used this value as weights in the proportional hazards models [36, 37].
Covariates were chosen based on biological plausibility, causal diagrams, and prior literature. These included age (in the survival models), sex (female or male), tumor stage (I, II, III, IV), tumor site (colon or rectum), antibiotic use within the prior year, NSAID use within the prior year, race/ethnicity (non-Hispanic-White, Hispanic, non-Hispanic Black, missing or refused), family history of cancer, neoadjuvant, adjuvant treatment, sequencing batch, resection tissue type (macrodissected or unprocessed), smoking status (never, former, current), alcohol use (never, former, current), and body mass index (BMI; kg/m2) [38‐42]. We conducted survival models stratified by tumor site (colon vs. rectum) given the different physiologies and trajectories of treatment. For the exploratory analyses, we adjusted p-values using the Bonferroni correction method. All statistical analyses were conducted using R, version 4.4.2.
Results
Population characteristics
Baseline characteristics of EoCRC and LoCRC patients are summarized in Table 1. Briefly, the full population was primarily male (59%) and non-Hispanic White (79%). Approximately 39% were treated with neoadjuvant therapies, comprising mostly (95%) rectal cancers. Three patients (two EoCRC and one LoCRC) presented an inherited syndrome and were excluded from the linear, logistic, and survival analyses. There were no statistically significant differences between EoCRC and LoCRC cases based on demographic, clinical, or tumor characteristics. The overall median survival time from surgery was 7.15 years (interquartile range of 6.46 years). Between the tumor and normal adjacent tissue, we detected statistically significant differences in the relative abundance of Observed ASV, Faith’s PD alpha diversity, Bacteroides, Fusobacteria, and Parvimonas (Supplementary Table S2). The final microbial dataset contained 437 unique genera and 22 phyla.
Anzeige
Table 1
Participant characteristics among EoCRC (n = 46) and locrc (n = 101) patients treated at Moffitt cancer center for surgical resection from 1995–2020
Patient characteristics | EoCRC, N = 46 | LoCRC, N = 101 | P-value1 | ||
|---|---|---|---|---|---|
Mean (SD) | N (%) | Mean (SD) | N (%) | ||
Sex, female | 15 (32.6) | 46 (45.5) | 0.20 | ||
Age at diagnosis | 40.44 (7.0) | 64.62 (10.4) | < 0.001 | ||
BMI, kg/m2 | 28.46 (6.9) | 28.54 (5.9) | 0.94 | ||
Education | 0.06 | ||||
≤ Highschool | 10 (21.7) | 39 (38.6) | |||
≥ College | 23 (50.0) | 32 (31.7) | |||
Race/Ethnicity | 0.38 | ||||
Non-Hispanic White | 35 (76.1) | 81 (81.0) | |||
Non-Hispanic Black | 6 (13.0) | 5 (5.0) | |||
Hispanic | 3 (6.5) | 9 (9.0) | |||
Other | 2 (4.3) | 5 (5.0) | |||
Follow-up time (years) | 8.12 (6.03) | 16.70 (4.58) | 0.56 | ||
Family history of cancer, yes | 15 (32.6) | 20 (19.8) | 0.14 | ||
Antibiotic use within a year from treatment start, yes | 6 (13.0) | 21 (20.8) | 0.37 | ||
NSAID use within a year of treatment start, yes | 12 (26.1) | 35 (34.7) | 0.40 | ||
Smoking status | 0.08 | ||||
Never | 28 (60.9) | 44 (43.6) | |||
Former | 9 (19.6) | 38 (37.6) | |||
Current | 9 (19.6) | 19 (18.8) | |||
Alcohol use | |||||
Never | 14 (30.4) | 25 (24.8) | 0.76 | ||
Former | 9 (19.6) | 20 (19.8) | |||
Current | 23 (50.0) | 56 (55.4) | |||
Proton Pump Inhibitor use, yes | 11 (23.9) | 22 (21.8) | 0.94 | ||
Tumor stage | 0.98 | ||||
I | 6 (13.0) | 13 (12.9) | |||
II | 9 (19.6) | 22 (21.8) | |||
III | 22 (47.8) | 49 (48.5) | |||
IV | 9 (19.6) | 17 (16.8) | |||
Tumor site | 0.34 | ||||
Colon | 15 (32.6) | 43 (42.6) | |||
Rectum | 31 (67.4) | 58 (57.4) | |||
Distant metastasis at diagnosis, yes | 9 (19.6) | 15 (15.0) | 0.65 | ||
Macrodissected tumor tissue | 29 (90.6) | 63 (81.8) | 0.39 | ||
Macrodissected normal adjacent tissue | 1 (2.2) | 9 (9.6) | 0.22 | ||
Neoadjuvant treatment | 0.45 | ||||
Yes | 18 (54.5) | 39 (44.8) | |||
No | 15 (45.5) | 48 (55.2) | |||
Adjuvant treatment | 1.00 | ||||
Yes | 17 (60.7) | 47 (62.7) | |||
No | 11 (39.3) | 28 (37.3) | |||
Lynch Syndrome/Inherited syndromes, yes | 2 (4.3) | 1 (1.0) | 0.48 | ||
MSI, Instable | 1 (4.0) | 5 (10.0) | 0.65 | ||
KRAS, Positive | 4 (33.3) | 10 (50.0) | 0.58 | ||
Vital status | 0.01 | ||||
Alive | 36 (78.3) | 55 (54.5) | |||
Deceased | 10 (21.7) | 46 (45.5) | |||
Age associations with EoCRC Microbiome
Associations of tumor and normal adjacent tissue microbiome metrics with age are presented in (Supplementary Table S3). No associations were statistically significant in tumor tissue; however, in normal adjacent tissue, for every 10-year increase in age, there was a 1-SD increase in the relative abundance of the a priori oral-origin genera Porphyromonas (Beta = 0.14, P = 0.03), Peptostreptococcus (Beta = 0.14, P = 0.03), and Prevotella (Beta = 0.13, P = 0.04). In case-case multivariable logistic regression models (Fig. 3A-D; Supplementary Table S4), considered per 1-SD increase in the CLR transformed abundance, Fusobacterium in normal adjacent tissue was statistically significantly more abundant among EoCRC relative to LoCRC cases (OR = 1.65; 95% CI = 1.03, 2.78; P = 0.05). In exploratory analyses, Bacillus in normal adjacent tissue (OR = 0.58; 95% CI = 0.35, 0.92; P = 0.03) was less abundant among EoCRC patients, though not statistically significantly when adjusting for multiple testing. There were overall differences in the normal adjacent tissue microbiome composition of EoCRC compared to LoCRC individuals based on the third axis of Bray Curtis. Comparing those ≤ 45 years old to those ≥ 60 years old, alpha diversity in both tumor and normal adjacent tissue was lower among individuals ≤ 45 years old (e.g., for Observed ASV’s in normal adjacent tissue (OR = 0.87; 95% CI = 0.42, 1.70; P = 0.69) though not statistically significantly (Supplementary Table S5). Comparing these age groups, the other associations for a priori- and exploratory-selected relative abundances were not statistically significant after Bonferroni correction.
Fig. 3
Associations of tumor and normal adjacent tissue microbiome with age at onset of colorectal cancer. Volcano plots of beta coefficients were estimated via linear regression for A) tumor and B) normal adjacent tissue and odds ratios were estimated via logistic regression for C) tumor and D) normal adjacent tissue. Models were adjusted for sex (female or male), tumor stage (I, II, III, IV), tumor site (colon or rectum), antibiotic use before surgery (yes, no), and NSAID use before surgery (yes, no), race/ethnicity (NHW, NHB, H, other), smoking, and alcohol use within a year from surgery. Analyses were conducted separately for tumor tissue (n rarefied= 81; n = 109) and normal adjacent tissue (n rarefied= 115, n = 139) across alpha diversity (Observed ASVs, Shannon Index, Faith’s PD), beta diversity (PC axes 1–3 of Bray-Curtis, Weighted/Unweighted UniFrac), exploratory genera (selected if present in ≥ 40% of samples with ≥ 0.01% average relative abundance), and a priori genera (selected from CRC literature: Fusobacterium, Bacteroides, Porphyromonas, Parvimonas, Peptostreptococcus, Gemella, Prevotella, Solobacterium, Dialister). A priori and exploratory bacteria were z-score standardized, with abundance transformed using center-log ratio (logistic regression) or multiplied by 100 (linear regression). Age was modeled in 10-year increments. A priori genera, alpha, and beta diversity metrics were tested at a significance threshold of α = 0.05 (red dotted line), while exploratory bacteria P-values were Bonferroni-adjusted at α = 0.002 (blue dotted line). Sample sizes for alpha/beta diversity analyses were based on rarefied data (n = 128; Tumor = 80, Normal = 113), and relative abundance analyses (n = 144; Tumor = 106, Normal = 137). Analyses included EoCRC (n = 39–44) and LoCRC (n = 90–100) with LoCRC as the reference level. PD, Phylogenetic distance; CI, Confidence intervals; PC, Principal coordinate; BC, Bray-Curtis; NHW, Non-Hispanic White; NHB, Non-Hispanic Black; H, Hispanic
We further compared the prevalence of a priori genera among CRC patients ≤ 45, 46–59, and ≥ 60 years old (Fig. 4A-B) to explore the association in relation to updated CRC screening guidelines, which recommend colonoscopy starting at age 45 for individuals at risk [3]. Despite no statistically significant differences, given the small sample sizes, younger individuals had a higher prevalence of Fusobacterium in tumor and a lower prevalence of Parvimonas, Peptostreptococcus, and Prevotella in tumor and normal adjacent tissue.
Anzeige
Fig. 4
Bar graphs of the association of prevalence percentage of a priori selected bacteria genera among A) tumor tissue and B) normal adjacent tissue samples among individuals ≤ 45, 46–59, and ≥ 60 years old at diagnosis treated at Moffitt Cancer Center for surgical resection from 1995–2020, N = 95. Percent prevalence was estimated as the number of samples with a relative abundance > 0 divided by the total number of samples in each age group. No statistically significant differences were observed between age groups
Tumor and normal adjacent tissue Microbiome associations with mortality
Multivariable associations (adjusted for age and other confounders) of tumor and normal adjacent tissue microbiome metrics with mortality are presented in Table 2. We found that multiple age-associated microbiome features were associated with higher risk for mortality. For every 1-SD increase in the CLR-transformed relative abundance of a priori-selected Prevotella in normal adjacent tissue, there was a 47% higher risk of mortality (95% CI = 1.19, 1.81; P < 0.001). Age-associated Porphyromonas, Peptostreptococcus, and Fusobacterium were not strongly and statistically significantly associated with mortality. In normal adjacent tissue, for every 1-SD increase in relative abundance of Bacillus (an exploratory-selected bacterium whose relative abundance was inversely associated with age), there was a 57% lower risk of mortality (HR = 0.43; 95% CI = 0.28, 0.68; P < 0.001). The tumor tissue-based third axis of the Weighted UniFrac PCoA, which indicated differences in overall microbiome composition between age groups, was also associated with mortality (HR = 5.71; 95% CI = 1.82, 17.91; P = 0.003).
Table 2
Weighted and unweighted associations1 of tumor and normal adjacent alpha diversity, a priori, exploratory CLR transformed relative abundance, and the first three principal coordinate axes of beta diversity, with overall survival in CRC patients, N = 144
Unweighted | Weighted | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
Tumor, n = 106 | Normal, n = 137 | Tumor, n = 106 | Normal, n = 137 | ||||||||
Microbiome features | Tumor Median (Min-Max) | Normal Median (Min-Max) | HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | |
Alpha diversit2,6 | |||||||||||
Observed | 1.62 (0.13, 4.67) | 1.86 (0.05, 4.45) | 1.10 (0.74, 1.63) | 0.64 | 0.79 (0.57, 1.09) | 0.15 | 1.16 (0.80, 1.68) | 0.44 | 0.69 (0.49, 0.98) | 0.04 | |
Shannon | 26.35 (3.60, 169.20) | 36.50 (3.90, 146.70) | 1.00 (0.67, 1.48) | 0.98 | 0.76 (0.53, 1.08) | 0.12 | 1.03 (0.77, 1.39) | 0.82 | 0.66 (0.43, 1.02) | 0.06 | |
Faith’s PD | 127.77 (31.90, 467.50) | 159.04 (32.45, 428.06) | 0.94 (0.61, 1.45) | 0.78 | 0.74 (0.52, 1.06) | 0.1 | 1.00 (0.71, 1.41) | 1.00 | 0.66 (0.45, 0.98) | 0.04 | |
A priori bacteria3 | |||||||||||
Bacteroides | 2.13 (0.01, 99.69) | 2.11 (0.01, 43.11) | 0.70 (0.43, 1.15) | 0.16 | 1.29 (0.97, 1.71) | 0.08 | 0.57 (0.30, 1.10) | 0.09 | 1.22 (0.84, 1.78) | 0.29 | |
Dialister | 0.02 (0.00, 94.90) | 0.08 (0.00, 23.43) | 0.87 (0.62, 1.21) | 0.4 | 0.97 (0.70, 1.34) | 0.87 | 0.80 (0.60, 1.08) | 0.14 | 0.81 (0.56, 1.19) | 0.28 | |
Fusobacterium | 0.76 (0.00, 97.35) | 1.45 (0.00, 62.03) | 1.09 (0.77, 1.52) | 0.63 | 0.88 (0.62, 1.25) | 0.47 | 0.84 (0.56, 1.27) | 0.41 | 0.72 (0.49, 1.08) | 0.11 | |
Gemella | 0.23 (0.00, 93.83) | 0.24 (0.00, 50.60) | 1.07 (0.74, 1.56) | 0.71 | 1.23 (0.88, 1.71) | 0.23 | 1.04 (0.68, 1.60) | 0.86 | 1.01 (0.68, 1.50) | 0.96 | |
Parvimonas | 0.34 (0.00, 97.80) | 0.28 (0.00, 22.28) | 0.90 (0.63, 1.28) | 0.54 | 1.15 (0.86, 1.54) | 0.36 | 0.67 (0.42, 1.07) | 0.09 | 1.19 (0.79, 1.81) | 0.41 | |
Peptostreptococcus | 0.001 (0.00, 5.97) | 0.001 (0.00, 12.38) | 0.81 (0.53, 1.24) | 0.33 | 1.02 (0.74, 1.41) | 0.9 | 0.61 (0.38, 0.99) | 0.05 | 0.92 (0.52, 1.64) | 0.78 | |
Porphyromonas | 0.12 (0.00, 37.55) | 0.12 (0.00, 99.26) | 0.75 (0.48, 1.18) | 0.22 | 1.28 (0.99, 1.65) | 0.06 | 0.62 (0.34, 1.15) | 0.13 | 1.17 (0.91, 1.49) | 0.21 | |
Prevotella | 0.001 (0.00, 31.85) | 0.001 (0.00, 3.72) | 1.22 (0.88, 1.70) | 0.24 | 1.47 (1.11, 1.94) | 0.007 | 1.31 (0.94, 1.83) | 0.11 | 1.47 (1.19, 1.81) | < 0.001 | |
Solobacterium | 0.001 (0.00, 0.06) | 0.001 (0.00, 0.31) | 0.99 (0.69, 1.42) | 0.96 | 1.02 (0.72, 1.45) | 0.91 | 1.00 (0.81, 1.23) | 0.96 | 1.26 (0.65, 2.43) | 0.5 | |
Exploratory bacteria4 | |||||||||||
Acinetobacter | 0.03 (0.00, 98.47) | 0.05 (0.00, 97.72) | 1.37 (0.97, 1.94) | 0.07 | 1.31 (0.97, 1.76) | 0.08 | 1.23 (0.86, 1.75) | 0.26 | 1.44 (1.07, 1.93) | 0.02 | |
Actinomyces | 0.08 (0.00, 12.78) | 0.05 (0.00, 28.10) | 0.68 (0.42, 1.11) | 0.13 | 1.03 (0.76, 1.41) | 0.84 | 1.09 (0.61, 1.94) | 0.77 | 1.01 (0.73, 1.40) | 0.95 | |
Alistipes | 0.01 (0.00, 7.59) | 0.01 (0.00, 28.98) | 0.77 (0.49, 1.21) | 0.26 | 1.09 (0.79, 1.50) | 0.59 | 0.80 (0.50, 1.28) | 0.36 | 1.17 (0.84, 1.64) | 0.34 | |
Bacillus | 0.06 (0.00, 96.58) | 0.02 (0.00, 73.00) | 1.11 (0.79, 1.57) | 0.54 | 0.46 (0.30, 0.70) | < 0.001 | 1.00 (0.69, 1.44) | 1.00 | 0.43 (0.28, 0.68) | < 0.001 | |
Bifidobacterium | 0.06 (0.00, 46.56) | 0.13 (0.00, 53.15) | 1.21 (0.83, 1.77) | 0.31 | 0.77 (0.53, 1.12) | 0.18 | 1.12 (0.76, 1.67) | 0.56 | 0.68 (0.39, 1.20) | 0.18 | |
Blautia | 0.29 (0.00, 37.67) | 0.73 (0.00, 96.90) | 0.60 (0.35, 1.05) | 0.07 | 1.31 (0.99, 1.74) | 0.06 | 0.43 (0.26, 0.73) | 0.002 | 1.10 (0.74, 1.64) | 0.64 | |
Cloacibacterium | 0.00 (0.00, 39.22) | 0.00 (0.00, 26.62) | 0.96 (0.70, 1.33) | 0.83 | 0.97 (0.69, 1.36) | 0.86 | 0.78 (0.54, 1.12) | 0.18 | 0.85 (0.55, 1.32) | 0.47 | |
Clostridium | 0.04 (0.00, 80.31) | 0.06 (0.00, 80.14) | 1.02 (0.65, 1.58) | 0.94 | 0.79 (0.57, 1.10) | 0.17 | 0.97 (0.59, 1.62) | 0.91 | 0.74 (0.53, 1.04) | 0.08 | |
Collinsella | 0.03 (0.00, 3.68) | 0.10 (0.00, 47.70) | 0.96 (0.66, 1.38) | 0.82 | 1.16 (0.90, 1.49) | 0.25 | 1.00 (0.72, 1.38) | 0.98 | 1.11 (0.85, 1.47) | 0.44 | |
Coprococcus | 0.02 (0.00, 17.72) | 0.03 (0.00, 5.83) | 0.98 (0.67, 1.43) | 0.91 | 0.93 (0.69, 1.25) | 0.64 | 1.06 (0.67, 1.68) | 0.81 | 0.93 (0.69, 1.25) | 0.61 | |
Dietzia | 0.03 (0.00, 43.04) | 0.08 (0.00, 96.90) | 0.85 (0.58, 1.26) | 0.43 | 0.87 (0.63, 1.19) | 0.38 | 1.17 (0.83, 1.64) | 0.38 | 1.02 (0.75, 1.39) | 0.91 | |
Dorea | 0.01 (0.00, 31.65) | 0.03 (0.00, 7.38) | 0.98 (0.65, 1.46) | 0.91 | 1.09 (0.84, 1.41) | 0.53 | 1.00 (0.61, 1.64) | 0.99 | 1.06 (0.81, 1.38) | 0.68 | |
Eubacterium | 0.03 (0.00, 19.98) | 0.11 (0.00, 52.46) | 0.67 (0.42, 1.06) | 0.09 | 0.97 (0.71, 1.31) | 0.84 | 0.59 (0.39, 0.91) | 0.02 | 0.91 (0.59, 1.41) | 0.68 | |
Faecalibacterium | 0.02 (0.00, 19.23) | 0.04 (0.00, 18.51) | 0.77 (0.50, 1.20) | 0.25 | 0.92 (0.68, 1.25) | 0.61 | 0.65 (0.34, 1.25) | 0.20 | 0.96 (0.65, 1.41) | 0.82 | |
Fusicatenibacter | 0.00 (0.00, 1.39) | 0.01 (0.00, 43.83) | 0.83 (0.57, 1.22) | 0.35 | 0.94 (0.71, 1.26) | 0.69 | 0.95 (0.64, 1.43) | 0.82 | 0.95 (0.75, 1.21) | 0.69 | |
Klebsiella | 0.02 (0.00, 36.54) | 0.02 (0.00, 10.35) | 1.14 (0.77, 1.67) | 0.52 | 0.87 (0.62, 1.21) | 0.4 | 1.21 (0.87, 1.68) | 0.26 | 0.93 (0.59, 1.46) | 0.74 | |
Lachnoclostridium | 0.07 (0.00, 28.01) | 0.16 (0.00, 83.97) | 0.75 (0.45, 1.25) | 0.27 | 1.22 (0.91, 1.63) | 0.19 | 0.65 (0.43, 0.97) | 0.03 | 1.19 (0.81, 1.73) | 0.38 | |
Lactobacillus | 0.01 (0.00, 93.97) | 0.00 (0.00, 99.02) | 1.05 (0.73, 1.51) | 0.78 | 1.20 (0.80, 1.78) | 0.38 | 1.10 (0.77, 1.58) | 0.60 | 1.06 (0.57, 1.97) | 0.84 | |
Mycobacterium | 0.01 (0.00, 28.64) | 0.01 (0.00, 94.33) | 1.40 (1.02, 1.91) | 0.04 | 0.82 (0.57, 1.16) | 0.26 | 1.53 (1.07, 2.20) | 0.02 | 0.96 (0.73, 1.28) | 0.8 | |
Oscillibacter | 0.02 (0.00, 91.44) | 0.04 (0.00, 13.87) | 0.89 (0.67, 1.19) | 0.44 | 1.13 (0.84, 1.51) | 0.42 | 0.87 (0.68, 1.10) | 0.24 | 1.00 (0.66, 1.52) | 0.98 | |
Parabacteroides | 0.05 (0.00, 19.66) | 0.19 (0.00, 79.73) | 0.90 (0.60, 1.34) | 0.59 | 1.04 (0.77, 1.39) | 0.81 | 0.68 (0.42, 1.11) | 0.12 | 0.93 (0.64, 1.35) | 0.71 | |
Peptoniphilus | 0.01 (0.00, 13.64) | 0.02 (0.00, 91.82) | 1.60 (1.15, 2.23) | 0.006 | 1.08 (0.75, 1.55) | 0.68 | 1.64 (1.17, 2.31) | 0.004 | 1.27 (0.73, 2.22) | 0.39 | |
Ruminiclostridium | 0.00 (0.00, 33.39) | 0.00 (0.00, 8.22) | 1.00 (0.72, 1.40) | 0.98 | 1.00 (0.72, 1.38) | 1 | 1.12 (0.82, 1.53) | 0.49 | 0.88 (0.54, 1.45) | 0.62 | |
Ruminococcus | 0.00 (0.00, 69.28) | 0.02 (0.00, 94.17) | 0.86 (0.55, 1.36) | 0.53 | 0.82 (0.57, 1.16) | 0.26 | 0.81 (0.50, 1.31) | 0.38 | 0.77 (0.48, 1.24) | 0.28 | |
Schlegelella | 0.00 (0.00, 97.01) | 0.00 (0.00, 90.38) | 0.82 (0.52, 1.32) | 0.42 | 0.69 (0.49, 0.99) | 0.04 | 0.79 (0.33, 1.91) | 0.61 | 0.69 (0.47, 1.01) | 0.06 | |
Beta diversity5,6 | |||||||||||
BC PC1 | 0.06 (−0.55, 0.19) | 0.06 (−0.55, 0.20) | 0.72 (0.38, 1.36) | 0.31 | 1.14 (0.85, 1.53) | 0.39 | 0.57 (0.22, 1.47) | 0.24 | 1.30 (0.96, 1.76) | 0.09 | |
BC PC2 | −0.04 (−0.21, 0.52) | −0.04 (−0.22, 0.53) | 0.0096 (0.0018, 0.05) | < 0.001 | 0.82 (0.59, 1.15) | 0.25 | 0.01 (0.002, 0.11) | < 0.001 | 0.91 (0.60, 1.37) | 0.65 | |
BC PC3 | 0.05 (−0.37, 0.13) | 0.04 (−0.41, 0.13) | 0.96 (0.57, 1.61) | 0.88 | 1.42 (0.95, 2.12) | 0.09 | 1.11 (0.69, 1.79) | 0.68 | 1.24 (0.82, 1.88) | 0.31 | |
Weighted UniFrac PC1 | 0.01 (−0.40, 0.31) | 0.04 (−0.41, 0.30) | 1.02 (0.69, 1.50) | 0.93 | 1.01 (0.70, 1.46) | 0.94 | 0.98 (0.67, 1.42) | 0.91 | 0.88 (0.60, 1.29) | 0.52 | |
Weighted UniFrac PC2 | −0.03 (−0.48, 0.27) | 0.01 (−0.47, 0.27) | 1.16 (0.71, 1.87) | 0.55 | 0.85 (0.63, 1.16) | 0.32 | 1.30 (0.74, 2.28) | 0.36 | 0.77 (0.57, 1.06) | 0.11 | |
Weighted UniFrac PC3 | 0.04 (−0.44, 0.24) | −0.01 (−0.43, 0.22) | 4.81 (1.85, 12.51) | 0.001 | 1.45 (1.02, 2.08) | 0.04 | 5.71 (1.82, 17.91) | 0.003 | 1.65 (1.08, 2.52) | 0.02 | |
Unweighted UniFrac PC1 | 0.08 (−0.33, 0.32) | −0.02 (−0.33, 0.36) | 1.19 (0.72, 1.95) | 0.49 | 1.27 (0.92, 1.77) | 0.15 | 1.12 (0.66, 1.91) | 0.68 | 1.38 (1.01, 1.91) | 0.05 | |
Unweighted UniFrac PC2 | −0.01 (−0.27, 0.28) | −0.01 (−0.27, 0.27) | 0.81 (0.52, 1.26) | 0.36 | 0.80 (0.57, 1.14) | 0.22 | 0.87 (0.52, 1.45) | 0.6 | 0.84 (0.54, 1.30) | 0.43 | |
Unweighted UniFrac PC3 | 0.01 (−0.26, 0.31) | 0.00 (−0.27, 0.29) | 1.22 (0.81, 1.85) | 0.34 | 1.04 (0.73, 1.48) | 0.84 | 1.20 (0.79, 1.84) | 0.39 | 1.00 (0.69, 1.43) | 0.98 | |
In stratified analyses (Table 3), there were no substantial variations in associations for the age-associated microbiome metrics by anatomic site. Differences in the associations by anatomic site may reflect the influence of treatment, as 61% of rectal cases received neoadjuvant treatment, whereas colon cancer cases largely represent treatment-naïve tissue. Notably, the relative abundance of Prevotella in normal adjacent tissue showed similar results across individuals with tumors in the colon and rectum anatomic sites. Beyond those bacteria that were age-associated in the respective tissue site described above, tumor tissue Fusobacterium was more strongly associated with a higher risk for mortality among rectal compared to colon cancer patients (P-int = 0.004), whereas tumor tissue Parvimonas was inversely associated with mortality among those with colon cancer (P-int = 0.01).
Anzeige
Table 3
Weighted associations1 of tumor and normal adjacent alpha diversity, a priori, exploratory CLR transformed relative abundance, and the first three principal coordinate axes of beta diversity, with overall survival stratified tumor site in CRC patients, N = 144
Tumor, n = 106 | Normal Adjacent, n = 137 | ||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Colon | Rectum | Colon | Rectum | ||||||||||||||||||
Microbiome features | N | Median (Min-Max) | HR (95% CI) | P | N | Median (Min-Max) | HR (95% CI) | P | P-inter | N | Median (Min-Max) | HR (95% CI) | P | N | Median (Min-Max) | HR (95% CI) | P | P-inter | |||
Alpha diversity2,6 | |||||||||||||||||||||
Observed | 35 | 1.79 (0.15, 4.67) | 0.86 (0.55, 1.36) | 0.52 | 45 | 1.36 (0.13, 4.35) | 1.19 (0.75, 1.88) | 0.46 | 0.87 | 43 | 1.63 (0.09, 4.45) | 0.55 (0.32, 0.95) | 0.03 | 70 | 2.05 (0.05, 4.42) | 0.84 (0.48, 1.48) | 0.55 | 0.29 | |||
Shannon | 35 | 31.40 (7.50, 157.70) | 1.12 (0.67, 1.87) | 0.66 | 45 | 23.10 (3.60, 169.20) | 0.93 (0.56, 1.53) | 0.76 | 0.17 | 43 | 42.30 (7.40, 139.80) | 0.64 (0.33, 1.27) | 0.20 | 70 | 34.00 (3.90, 146.70) | 0.79 (0.44, 1.43) | 0.44 | 0.62 | |||
Faith’s PD | 35 | 157.34 (53.67, 422.69) | 1.15 (0.67, 1.98) | 0.61 | 45 | 117.98 (31.90, 467.50) | 0.89 (0.50, 1.56) | 0.67 | 0.13 | 43 | 180.98 (48.41, 405.38) | 0.70 (0.38, 1.30) | 0.26 | 70 | 156.58 (32.45, 428.06) | 0.80 (0.46, 1.40) | 0.44 | 0.76 | |||
A priori bacteria3 | |||||||||||||||||||||
Bacteroides | 46 | 1.84 (0.01, 37.02) | 0.39 (0.19, 0.77) | 0.007 | 60 | 2.76 (0.01, 99.69) | 0.95 (0.49, 1.82) | 0.87 | 0.009 | 51 | 1.19 (0.01, 43.11) | 0.95 (0.55, 1.66) | 0.86 | 86 | 2.32 (0.04, 41.76) | 1.54 (1.02, 2.33) | 0.04 | 0.21 | |||
Dialister | 46 | 0.04 (0.00, 15.24) | 1.13 (0.70, 1.84) | 0.62 | 60 | 0.01 (0.00, 94.90) | 0.86 (0.63, 1.17) | 0.34 | 0.92 | 51 | 0.08 (0.00, 13.45) | 0.75 (0.39, 1.43) | 0.38 | 86 | 0.07 (0.00, 23.43) | 0.88 (0.59, 1.31) | 0.53 | 0.78 | |||
Fusobacterium | 46 | 1.13 (0.00, 91.53) | 0.77 (0.51, 1.16) | 0.21 | 60 | 0.63 (0.00, 97.35) | 1.72 (0.93, 3.19) | 0.09 | 0.004 | 51 | 0.98 (0.00, 62.03) | 0.72 (0.44, 1.19) | 0.20 | 86 | 1.71 (0.00, 42.84) | 0.77 (0.45, 1.33) | 0.35 | 0.70 | |||
Gemella | 46 | 0.36 (0.00, 20.81) | 1.36 (0.77, 2.41) | 0.28 | 60 | 0.17 (0.00, 93.83) | 1.16 (0.61, 2.20) | 0.66 | 0.07 | 51 | 0.17 (0.00, 31.66) | 1.14 (0.64, 2.03) | 0.65 | 86 | 0.30 (0.00, 50.60) | 1.11 (0.74, 1.67) | 0.60 | 0.36 | |||
Parvimonas | 46 | 0.42 (0.00, 97.80) | 0.80 (0.49, 1.29) | 0.36 | 60 | 0.26 (0.00, 92.06) | 1.24 (0.67, 2.30) | 0.50 | 0.01 | 51 | 0.23 (0.00, 22.28) | 0.87 (0.43, 1.78) | 0.71 | 86 | 0.35 (0.00, 9.36) | 1.20 (0.91, 1.57) | 0.19 | 0.5 | |||
Peptostreptococcus | 46 | 0.001 (0.00, 3.65) | 1.02 (0.66, 1.57) | 0.94 | 60 | 0.001 (0.00, 5.97) | 0.73 (0.49, 1.10) | 0.14 | 0.83 | 51 | 0.001 (0.00, 12.38) | 0.72 (0.40, 1.29) | 0.26 | 86 | 0.001 (0.00, 5.44) | 1.15 (0.76, 1.72) | 0.51 | 0.43 | |||
Porphyromonas | 46 | 0.16 (0.00, 33.86) | 0.63 (0.33, 1.23) | 0.17 | 60 | 0.11 (0.00, 37.55) | 0.88 (0.36, 2.15) | 0.77 | 0.83 | 51 | 0.20 (0.00, 51.58) | 0.99 (0.46, 2.12) | 0.98 | 86 | 0.08 (0.00, 99.26) | 1.05 (0.79, 1.38) | 0.75 | 0.75 | |||
Prevotella | 46 | 0.001 (0.00, 31.85) | 1.05 (0.60, 1.84) | 0.87 | 60 | 0.001 (0.00, 1.01) | 1.28 (0.94, 1.76) | 0.12 | 0.82 | 51 | 0.001 (0.00, 1.21) | 1.42 (0.89, 2.25) | 0.14 | 86 | 0.001 (0.00, 3.72) | 1.54 (1.16, 2.04) | 0.003 | 0.95 | |||
Solobacterium | 46 | 0.001 (0.00, 0.06) | 0.80 (0.61, 1.06) | 0.12 | 60 | 0.001 (0.00, 0.03) | 0.79 (0.44, 1.42) | 0.42 | 0.78 | 51 | 0.001 (0.00, 0.08) | 3.35 (1.62, 6.95) | 0.001 | 86 | 0.001 (0.00, 0.31) | 0.93 (0.55, 1.58) | 0.79 | 0.004 | |||
Exploratory bacteria4 | |||||||||||||||||||||
Acinetobacter | 46 | 0.04 (0.00, 17.64) | 1.10 (0.66, 1.82) | 0.72 | 60 | 0.03 (0.00, 98.47) | 1.97 (1.16, 3.35) | 0.01 | 0.14 | 51 | 0.06 (0.00, 97.72) | 1.24 (0.75, 2.03) | 0.40 | 86 | 0.03 (0.00, 24.40) | 1.38 (0.91, 2.11) | 0.13 | 0.63 | |||
Actinomyces | 46 | 0.13 (0.00, 12.78) | 0.75 (0.29, 1.94) | 0.55 | 60 | 0.07 (0.00, 11.90) | 0.70 (0.16, 3.00) | 0.63 | 0.21 | 51 | 0.04 (0.00, 28.10) | 1.51 (0.84, 2.71) | 0.16 | 86 | 0.05 (0.00, 13.51) | 0.87 (0.65, 1.17) | 0.36 | 0.06 | |||
Alistipes | 46 | 0.01 (0.00, 7.59) | 0.33 (0.11, 1.01) | 0.05 | 60 | 0.01 (0.00, 5.38) | 0.57 (0.25, 1.29) | 0.18 | 0.98 | 51 | 0.01 (0.00, 7.06) | 1.10 (0.34, 3.50) | 0.88 | 86 | 0.02 (0.00, 28.98) | 1.14 (0.84, 1.54) | 0.41 | 0.27 | |||
Bacillus | 46 | 0.06 (0.00, 19.88) | 1.13 (0.75, 1.71) | 0.55 | 60 | 0.05 (0.00, 96.58) | 0.95 (0.60, 1.50) | 0.82 | 0.81 | 51 | 0.03 (0.00, 23.08) | 0.27 (0.11, 0.64) | 0.003 | 86 | 0.02 (0.00, 73.00) | 0.49 (0.26, 0.93) | 0.03 | 0.75 | |||
Bifidobacterium | 46 | 0.06 (0.00, 8.19) | 1.14 (0.62, 2.12) | 0.67 | 60 | 0.05 (0.00, 46.56) | 1.09 (0.60, 1.97) | 0.78 | 0.10 | 51 | 0.15 (0.00, 17.33) | 0.97 (0.53, 1.77) | 0.93 | 86 | 0.11 (0.00, 53.15) | 0.46 (0.22, 0.95) | 0.04 | 0.15 | |||
Blautia | 46 | 0.54 (0.00, 37.67) | 0.42 (0.18, 0.96) | 0.04 | 60 | 0.21 (0.00, 16.98) | 0.87 (0.45, 1.66) | 0.67 | 0.08 | 51 | 0.63 (0.00, 71.80) | 0.71 (0.33, 1.57) | 0.40 | 86 | 0.89 (0.00, 96.90) | 1.55 (1.06, 2.28) | 0.03 | 0.07 | |||
Cloacibacterium | 46 | 0.00 (0.00, 39.22) | 0.64 (0.39, 1.04) | 0.07 | 60 | 0.00 (0.00, 11.39) | 1.34 (0.58, 3.10) | 0.49 | 0.08 | 51 | 0.00 (0.00, 25.75) | 1.31 (0.72, 2.37) | 0.37 | 86 | 0.00 (0.00, 26.62) | 0.69 (0.44, 1.08) | 0.10 | 0.06 | |||
Clostridium | 46 | 0.06 (0.00, 80.31) | 1.22 (0.72, 2.07) | 0.45 | 60 | 0.03 (0.00, 10.99) | 0.71 (0.32, 1.57) | 0.40 | 0.75 | 51 | 0.04 (0.00, 80.14) | 0.76 (0.29, 2.00) | 0.58 | 86 | 0.08 (0.00, 35.04) | 0.80 (0.56, 1.16) | 0.25 | 0.48 | |||
Collinsella | 46 | 0.04 (0.00, 3.54) | 0.56 (0.29, 1.07) | 0.08 | 60 | 0.02 (0.00, 3.68) | 1.17 (0.76, 1.78) | 0.48 | 0.10 | 51 | 0.08 (0.00, 47.70) | 0.95 (0.63, 1.43) | 0.82 | 86 | 0.10 (0.00, 12.22) | 1.20 (0.85, 1.72) | 0.3 | 0.61 | |||
Coprococcus | 46 | 0.02 (0.00, 17.72) | 0.89 (0.47, 1.69) | 0.72 | 60 | 0.01 (0.00, 1.60) | 0.41 (0.18, 0.93) | 0.03 | 0.03 | 51 | 0.02 (0.00, 3.25) | 0.51 (0.30, 0.89) | 0.02 | 86 | 0.03 (0.00, 5.83) | 1.27 (1.01, 1.61) | 0.04 | 0.001 | |||
Dietzia | 46 | 0.03 (0.00, 12.59) | 0.82 (0.49, 1.36) | 0.44 | 60 | 0.03 (0.00, 43.04) | 1.06 (0.58, 1.92) | 0.86 | 0.42 | 51 | 0.08 (0.00, 89.50) | 1.36 (0.82, 2.25) | 0.23 | 86 | 0.09 (0.00, 96.90) | 0.88 (0.62, 1.25) | 0.47 | 0.32 | |||
Dorea | 46 | 0.02 (0.00, 3.86) | 0.61 (0.31, 1.20) | 0.16 | 60 | 0.00 (0.00, 31.65) | 1.11 (0.62, 1.99) | 0.71 | 0.9 | 51 | 0.03 (0.00, 3.25) | 0.68 (0.40, 1.14) | 0.14 | 86 | 0.04 (0.00, 7.38) | 1.48 (1.08, 2.04) | 0.02 | 0.04 | |||
Eubacterium | 46 | 0.06 (0.00, 7.05) | 0.46 (0.26, 0.83) | 0.01 | 60 | 0.02 (0.00, 19.98) | 0.91 (0.48, 1.74) | 0.78 | 0.98 | 51 | 0.11 (0.00, 52.46) | 0.61 (0.37, 1.00) | 0.05 | 86 | 0.11 (0.00, 19.69) | 1.59 (0.92, 2.75) | 0.09 | 0.06 | |||
Faecalibacterium | 46 | 0.06 (0.00, 19.23) | 0.85 (0.31, 2.35) | 0.75 | 60 | 0.01 (0.00, 2.05) | 1.10 (0.52, 2.36) | 0.80 | 0.8 | 51 | 0.03 (0.00, 18.51) | 0.27 (0.12, 0.61) | 0.002 | 86 | 0.04 (0.00, 13.84) | 1.35 (1.00, 1.81) | 0.05 | 0.005 | |||
Fusicatenibacter | 46 | 0.00 (0.00, 1.19) | 0.45 (0.22, 0.93) | 0.03 | 60 | 0.00 (0.00, 1.39) | 0.97 (0.56, 1.69) | 0.92 | 0.84 | 51 | 0.00 (0.00, 43.83) | 0.67 (0.28, 1.59) | 0.37 | 86 | 0.01 (0.00, 18.29) | 1.13 (0.90, 1.42) | 0.30 | 0.10 | |||
Klebsiella | 46 | 0.02 (0.00, 36.54) | 0.65 (0.30, 1.39) | 0.26 | 60 | 0.02 (0.00, 19.53) | 1.54 (0.78, 3.03) | 0.21 | 0.01 | 51 | 0.01 (0.00, 7.13) | 1.17 (0.70, 1.97) | 0.54 | 86 | 0.05 (0.00, 10.35) | 0.76 (0.47, 1.24) | 0.27 | 0.66 | |||
Lachnoclostridium | 46 | 0.19 (0.00, 28.01) | 0.53 (0.28, 1.02) | 0.06 | 60 | 0.04 (0.00, 5.45) | 0.55 (0.17, 1.74) | 0.31 | 0.93 | 51 | 0.23 (0.00, 16.81) | 0.98 (0.51, 1.86) | 0.95 | 86 | 0.16 (0.00, 83.97) | 1.48 (1.12, 1.94) | 0.006 | 0.27 | |||
Lactobacillus | 46 | 0.05 (0.00, 42.64) | 1.49 (0.73, 3.08) | 0.28 | 60 | 0.00 (0.00, 93.97) | 1.32 (0.86, 2.02) | 0.20 | 1.00 | 51 | 0.01 (0.00, 99.02) | 1.55 (0.43, 5.59) | 0.50 | 86 | 0.00 (0.00, 13.39) | 1.01 (0.63, 1.62) | 0.97 | 0.50 | |||
Mycobacterium | 46 | 0.01 (0.00, 4.02) | 1.81 (1.07, 3.06) | 0.03 | 60 | 0.01 (0.00, 28.64) | 1.15 (0.80, 1.65) | 0.45 | 0.46 | 51 | 0.01 (0.00, 2.51) | 3.45 (1.34, 8.86) | 0.01 | 86 | 0.02 (0.00, 94.33) | 0.67 (0.46, 0.96) | 0.03 | 0.02 | |||
Oscillibacter | 46 | 0.04 (0.00, 91.44) | 0.97 (0.73, 1.28) | 0.83 | 60 | 0.01 (0.00, 6.45) | 0.69 (0.33, 1.42) | 0.31 | 0.13 | 51 | 0.05 (0.00, 9.47) | 0.71 (0.44, 1.14) | 0.16 | 86 | 0.03 (0.00, 13.87) | 1.17 (0.79, 1.75) | 0.43 | 0.29 | |||
Parabacteroides | 46 | 0.12 (0.00, 19.66) | 1.02 (0.62, 1.68) | 0.93 | 60 | 0.04 (0.00, 4.22) | 0.37 (0.14, 0.99) | 0.05 | 0.66 | 51 | 0.12 (0.00, 39.67) | 0.96 (0.52, 1.78) | 0.90 | 86 | 0.45 (0.00, 79.73) | 1.02 (0.74, 1.40) | 0.90 | 0.95 | |||
Peptoniphilus | 46 | 0.00 (0.00, 1.97) | 2.06 (1.06, 4.02) | 0.03 | 60 | 0.01 (0.00, 13.64) | 1.22 (0.75, 1.96) | 0.42 | 0.15 | 51 | 0.02 (0.00, 91.82) | 1.74 (1.14, 2.66) | 0.01 | 86 | 0.03 (0.00, 17.66) | 0.99 (0.52, 1.88) | 0.98 | 0.37 | |||
Ruminiclostridium | 46 | 0.00 (0.00, 33.39) | 0.94 (0.49, 1.79) | 0.85 | 60 | 0.00 (0.00, 1.93) | 1.17 (0.53, 2.60) | 0.70 | 0.54 | 51 | 0.00 (0.00, 8.22) | 0.70 (0.36, 1.37) | 0.29 | 86 | 0.01 (0.00, 1.81) | 0.99 (0.48, 2.04) | 0.97 | 0.66 | |||
Ruminococcus | 46 | 0.03 (0.00, 9.76) | 0.75 (0.30, 1.84) | 0.53 | 60 | 0.00 (0.00, 69.28) | 1.09 (0.52, 2.26) | 0.82 | 0.21 | 51 | 0.02 (0.00, 94.17) | 0.57 (0.32, 1.02) | 0.06 | 86 | 0.01 (0.00, 9.31) | 1.29 (0.78, 2.13) | 0.33 | 0.04 | |||
Schlegelella | 46 | 0.00 (0.00, 97.01) | 1.08 (0.79, 1.47) | 0.63 | 60 | 0.01 (0.00, 79.43) | 0.64 (0.15, 2.83) | 0.56 | 0.29 | 51 | 0.01 (0.00, 90.38) | 0.53 (0.26, 1.10) | 0.09 | 86 | 0.00 (0.00, 1.77) | 0.85 (0.55, 1.34) | 0.49 | 0.44 | |||
Beta diversity5,6 | |||||||||||||||||||||
Bray Curtis, PC1 | 35 | 0.05 (−0.55, 0.15) | 0.44 (0.21, 0.94) | 0.03 | 45 | 0.07 (−0.50, 0.19) | 1.86 (0.48, 7.16) | 0.37 | 0.07 | 43 | 0.06 (−0.52, 0.18) | 1.72 (1.03, 2.85) | 0.04 | 70 | 0.05 (−0.55, 0.20) | 1.09 (0.67, 1.77) | 0.74 | 0.27 | |||
Bray Curtis, PC2 | 35 | −0.04 (−0.20, 0.39) | 0.85 (0.26, 2.83) | 0.8 | 45 | −0.04 (−0.21, 0.52) | 0.83 (0.43, 1.58) | 0.57 | 0.62 | 43 | −0.04 (−0.22, 0.53) | 0.68 (0.36, 1.30) | 0.24 | 70 | −0.04 (−0.22, 0.53) | 0.96 (0.63, 1.48) | 0.86 | 0.35 | |||
Bray Curtis, PC3 | 35 | 0.06 (−0.31, 0.12) | 6.87 (1.59, 29.68) | 0.01 | 45 | 0.04 (−0.37, 0.13) | 0.93 (0.41, 2.14) | 0.87 | 0.04 | 43 | 0.03 (−0.41, 0.13) | 0.74 (0.41, 1.31) | 0.30 | 70 | 0.04 (−0.37, 0.13) | 3.17 (1.05, 9.56) | 0.04 | 0.05 | |||
Weighted UniFrac PC1 | 35 | 0.00 (−0.35, 0.31) | 0.46 (0.22, 0.95) | 0.04 | 45 | 0.01 (−0.40, 0.30) | 1.07 (0.71, 1.60) | 0.76 | 0.04 | 43 | 0.03 (−0.38, 0.30) | 0.35 (0.16, 0.74) | 0.006 | 70 | 0.04 (−0.41, 0.30) | 1.26 (0.78, 2.04) | 0.34 | 0.04 | |||
Weighted UniFrac PC2 | 35 | 0.02 (−0.48, 0.27) | 0.90 (0.51, 1.59) | 0.71 | 45 | −0.04 (−0.37, 0.23) | 1.45 (0.80, 2.63) | 0.23 | 0.35 | 43 | −0.01 (−0.47, 0.26) | 0.84 (0.49, 1.44) | 0.52 | 70 | 0.04 (−0.40, 0.27) | 0.84 (0.50, 1.42) | 0.52 | 0.85 | |||
Weighted UniFrac PC3 | 35 | 0.02 (−0.43, 0.20) | 1.31 (0.68, 2.54) | 0.42 | 45 | 0.06 (−0.44, 0.24) | 1.22 (0.82, 1.83) | 0.33 | 0.48 | 43 | 0.01 (−0.43, 0.20) | 2.25 (1.00, 5.08) | 0.05 | 70 | −0.02 (−0.36, 0.22) | 1.25 (0.74, 2.11) | 0.40 | 0.20 | |||
Unweighted UniFrac PC1 | 35 | 0.02 (−0.30, 0.29) | 0.83 (0.41, 1.67) | 0.60 | 45 | 0.10 (−0.33, 0.32) | 1.30 (0.71, 2.38) | 0.39 | 0.11 | 43 | −0.03 (−0.32, 0.27) | 1.24 (0.75, 2.05) | 0.40 | 70 | −0.00 (−0.33, 0.36) | 1.18 (0.72, 1.96) | 0.51 | 0.89 | |||
Unweighted UniFrac PC2 | 35 | −0.01 (−0.25, 0.27) | 0.39 (0.13, 1.17) | 0.09 | 45 | −0.01 (−0.27, 0.28) | 1.32 (0.65, 2.69) | 0.44 | 0.43 | 43 | 0.00 (−0.26, 0.26) | 0.63 (0.39, 1.04) | 0.07 | 70 | −0.02 (−0.27, 0.27) | 1.01 (0.62, 1.63) | 0.97 | 0.09 | |||
Unweighted UniFrac PC3 | 35 | 0.01 (−0.24, 0.31) | 1.56 (0.71, 3.42) | 0.27 | 45 | 0.02 (−0.26, 0.28) | 0.90 (0.49, 1.64) | 0.73 | 0.32 | 43 | 0.01 (−0.27, 0.29) | 1.22 (0.70, 2.11) | 0.48 | 70 | 0.00 (−0.27, 0.27) | 0.94 (0.58, 1.52) | 0.81 | 0.74 | |||
Discussion
We characterized the bacterial composition of colorectal tumor and normal adjacent tissue among a cohort of CRC patients, among which a third were considered EoCRC patients. We investigated the associations of age with these bacteria and then assessed the mortality implications for these age-associated bacteria. We found several bacteria were associated with age diagnosis, considered both as a continuous variable and in comparisons of EoCRC and LoCRC. Notably, bacteria of oral-origin previously implicated in CRC development and progression, such as Porphyromonas, Peptostreptococcus, and Prevotella [43, 44], were among those identified to be associated with older ages in normal adjacent tissue. In turn, normal adjacent tissue abundance of Prevotella, and moderately Porphyromonas, were associated with a higher risk of mortality in the overall study population. In contrast, normal adjacent tissue relative abundance of Fusobacterium and Bacillus was inversely associated with age and, in turn, the latter was inversely associated with mortality. Taken together, our study suggests that age-associated microorganisms around CRC tumors should be further interrogated as they may relate to CRC prognosis.
We found that age was associated with multiple microbiome metrics among CRC patients. This is consistent with the literature in healthy populations demonstrating that the tissue microbiome differs across age groups, though this has been mostly studied via fecal samples [45]. Specifically, in our study, age was positively associated with normal adjacent tissue relative abundance of the common oral pathogens Porphyromonas, Peptostreptococcus, and Prevotella. As age increases, the gut barrier may become progressively disrupted allowing the formation, attachment, and invasion of bacterial biofilms to the colonic epithelium [27, 46]. It is currently not well known whether or how these bacteria travel through blood and enteral routes [47], but prior examples support that oral microbes, like Porphyromonas gingivalis, may enter vasculature, causing disruptions to endothelial integrity, promoting inflammation, and increasing susceptibility to other viruses and bacteria [48]. In recent studies, other oral bacteria (e.g., Fusobacterium spp.) survived acidic conditions and adhered to epithelial cells in the gut [49, 50]. Other potential biological explanations could involve the increasing prevalence of periodontal disease with age, which may render the oral cavity susceptible to these oral pathogens. Future studies may need to account for dental history to address this hypothesis [51, 52]. Taken together, there is strong biological plausibility to support differences in the tissue microbiome across age groups.
Comparing EoCRC to LoCRC patients, we observed multiple bacterial differences in the normal adjacent tissue microbiome, but less so in the tumor tissue microbiome. Specifically, Fusobacterium was more abundant, and Bacillus and Brevundimonas less abundant, in normal adjacent tissue of EoCRC patients. Only a few studies to date have investigated tumor tissue microbiome differences between EoCRC and LoCRC cases, but none to our knowledge have investigated normal adjacent microbiome differences. In a study of 15 EoCRC and 70 LoCRC cases, using whole-genome sequencing of colonic tumor tissue, there were differences in several taxa such as Enterobacteriaceae and Ruminococcaceae, but no genera that overlapped with our study [12]. Another study employing 16 S rRNA gene sequencing of colonic tumor tissue identified differences in genera Bacteroides and Akkermansia and reported higher alpha diversity among EoCRC cases, the latter of which was similar to our findings, though ours were not statistically significant [13]. Finally, another study investigated the tumor tissue microbiome among 20 EoCRC and 44 LoCRC. Using 16 S rRNA gene sequencing and partial least squares discriminant analysis (PLS-DA), they identified bacteria that successfully discriminated between EoCRC and LoCRC. Specifically, Fusobacterium was statistically significantly more abundant among EoCRC cases [14], which was similar to our findings. Differences between our study and these prior studies of CRC tumor tissue may be attributed to many factors, most notably tissue type under investigation, differences in extraction, sequencing, contamination control, and the patient population.
Similar to many prior studies [53‐55], we noted the potentially harmful role of oral-originating bacteria as it relates to prognosis, though we are among the first to associate aging with these bacteria using both tumor and normal adjacent tissue. Specifically, we found that the relative abundance of Porphyromonas and Prevotella in normal adjacent tissue were strongly associated with a higher mortality risk. In vitro studies have elucidated how Porphyromonas species, such as Porphyromonas gingivalis, adhere to the colon’s mucosal lining, invade host cells, and promote cancer cell proliferation by deregulation of protein synthesis through the MAPK/ERK signaling pathway [43]. A study using 16 S rRNA gene sequencing, co-abundance analysis, quantitative PCR, and cell and tissue cultures from tumor and adenoma tissues among 116 CRC patients, found that the co-existence of Prevotella intermedia and Fusobacterium nucleatum exerted an additive effect on the migration and invasion of CRC cells and together they were more strongly associated with CRC progression [56]. Several strains within the genus Bacillus have been shown to promote anti-inflammatory and anticancer responses by producing antimicrobial peptides [57, 58]. A recent study isolated four strains belonging to the genus Bacillus from fecal samples of 72 CRC patients. In CRC cell lines, these Bacillus strains were shown to significantly inhibit cancer cell proliferation, suggesting a potential protective effect against progression [59].
This study has several strengths. Unlike many microbiome studies, we had access to detailed information on clinical covariates to adjust for potential confounders. Further, we were able to provide a more comprehensive overview of the tissue microbiome by leveraging paired normal adjacent and tumor tissue samples. It was intriguing that associations with survival differed in some cases by specimen type. This could reflect specimen processing procedures (e.g., macrodissection for tumors), or dysregulations in tumor tissue compared to normal adjacent tissue. Of note, while our findings in normal adjacent tissue may somewhat reflect the tissue prior to the development of a carcinoma, it could also somewhat be altered by the tumor microenvironment.
Our study also had limitations. Rectal cancer patients were mostly treated with neoadjuvant therapy, so the intratumor microbiome among these patients was not ‘pre-treatment’. Tissue microbiome samples tend to have lower microbial yield and a higher risk of contamination, which requires additional precautions as microbial contamination can influence true biological signals [60]. Long-term storage of the samples and changing clinical protocols could also possibly influence microbiome composition. However, we used a rigorous approach to account for potential environmental and processing contaminants that could have influenced our results by leveraging prospectively collected quality control samples. There may be spatial heterogeneity within the tumor that could have led to inadequate characterization of the full intratumor microbiome [61]. We were able to include samples from multiple tumor sections in a subset of the seven patients, and still found that intersubject variability explained a high percentage of variability [60]. Our rarefaction depth of 1,000 reads, while chosen to retain samples and reduce bias from uneven sequencing in our diversity analyses, may overestimate the Shannon index and underestimate richness, potentially limiting the utility of these diversity measures. Of note, while our findings in normal adjacent tissue may somewhat reflect the tissue prior to the development of a carcinoma, it could also somewhat be altered by the tumor microenvironment. Finally, our sample size was limited within strata of patient and tumor characteristics, and our population was primarily non-Hispanic White. This necessitates replication in more diverse cohorts with sample sizes of at least 250 individuals.
In conclusion, higher age at diagnosis was associated with higher abundances of several bacterial genera in normal adjacent tissue that were implicated previously in CRC. In turn, some of these age-associated bacteria were associated with worse survival. Longitudinal studies with larger sample sizes among diverse populations of CRC patients with carefully collected tumor and paired normal adjacent tissue, in addition to oral samples given the oral bacteria associations, are needed to confirm and extend these findings. Overall, our findings highlight a potential microbiome-mediated mechanism whereby aging may influence survival outcomes among CRC patients.
Acknowledgements
This work has been supported in part by the Collaborative Data Services Core, the Quantitative Imaging Core, and the Tissue Core at the H. Lee Moffitt Cancer Center & Research Institute, a comprehensive cancer center designated by the National Cancer Institute and funded in part by Moffitt’s Cancer Center Support Grant (P30-CA076292). This work was also supported by the protocol: Gastrointestinal Oncology Transformative Initiatives for Translation (GOTIT): A Cloud-based Datamart for Rapid and Efficient Acquisition of Data to Enhance Clinical Care and Inform Research Advances (MCC 21064) at Moffitt Cancer Center. DNA extraction methods were modified based on input from Dr. Cindy Sears’ lab at Johns Hopkins.
Declarations
Ethics approval and consent to participate
This research was conducted in accordance with the Declaration of Helsinki and was reviewed and approved by Moffitt Cancer Center’s Scientific Review Committee and the Institutional Review Board (Advarra IRB #Pro00056038).
Consent for publication
Participants provided written informed consent to the Total Cancer Care™ (TCC) protocol at Moffitt Cancer Center, which included consent for the use and publication.
Competing interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
