Although collectively referred to as CRC, there are specific differences between RCC, LCC, and RC. Firstly, in terms of anatomy, the pattern of blood supply to the colon and rectum is quite different. Venous blood from the lower two-thirds of the rectum flows into the lungs via the internal iliac vein and the inferior vena cava, and venous blood from the upper third of the rectum flows into the liver via the inferior mesenteric vein. The venous outflow from the colon flows into the liver only through the lower and upper mesenteric veins. The arterial blood supply to the descending colon, sigmoid colon, and upper third of the rectum originate from the inferior mesenteric artery, while the rest of the colon is supplied by the superior mesenteric artery. The lower two-thirds of the rectum receives its arterial blood supply through the internal iliac artery [25–29]. In terms of pathology, mucosal lesions of the polypoid, non-depressed type are more common in the colon, while mucosal lesions with villous components, mucosal and submucosal lesions of the depressed type, occur more frequently in the rectum [30]. Gut microbes can be linked to the blood circulation through the gut barrier and are consequently involved in the pathogenesis of many diseases [31–34]. Because colon and rectal cancers are associated with different vascular anatomy and mucosal lesions, different blood circulation patterns and intestinal barriers may lead to differences in the microbiological composition of colon and rectal cancers.
From a molecular biology perspective, the RCC has a different molecular profile from the LCC and RC due to the different germinal origins. Microsatellite instability (MSI) is more common in proximal colon cancers than in distal colon and rectal cancers [35–37]. Chromosomal instability (CIN) [38, 39] and microsatellite stability (MSS) [40–42] are more frequent in distal colon and rectal cancers than in proximal colon cancers. KRAS and PI3K mutations are more frequent in proximal colon cancer than in distal colon and rectal [43, 44], and activation of the Wnt pathway is more frequent in distal colon and rectal cancer [45]. Right-side and left-side CRC also exhibited different consensus molecular subtypes (CMS). CMS1 tumors were mainly right-sided lesions, and CMS2 were mainly left-sided [46]. Microsatellite status and KRAS mutation status are relevant to treating CRC. The MSI tumor is often accompanied by high immunogenicity and high T-cell infiltration microenvironment [47, 48], so immune checkpoint inhibitors (ICIs) are more effective in treating RCC than LCC and RC [49, 50]. KRAS is involved in signaling pathways activated by EGFR [51], and mutations of KRAS can lead to tumor resistance to anti-EGFR antibodies [52]. As a result, proximal colon cancer does not benefit as much as distal colon and rectal cancer when treated with anti-EGFR antibodies [53, 54]. Therefore, whether there is an association between molecular signatures and microbial enrichment in the case of different tumor primary locations is also worth exploring.
In view of the clinical outcome, it is clear that there are differences in recurrence and survival prognosis between LCC and RCC. Research of mortality by LCC versus RCC for stages I to III of those who underwent curative surgery showed that no significant difference in mortality for all stages combined or for stage I cancers, stage II RCC had lower mortality, and stage III RCC had higher mortality [55]. But another research of outcomes of LCC and RCC after curative resection demonstrated that RCC was an independent risk factor for recurrence in stage II colon cancer [56]. A meta-analysis of 1,437,846 patients from 66 studies revealed that LCC was associated with a lower risk of death independently [57]. Therefore, RCC is more probably a predictor of poor recurrence and survival prognosis in more situations. But it is also possible that the prognosis of LCC is worse in stage II CRC.
In this study, we integrated metagenomic data from three studies to explore the relationship between tumor primary locations and microbiome. The species co-occurrence network illustrates that CRCs at various primary locations have some common hub species and unique hub species, suggesting that there are both similar and different patterns of microbial interconnections among CRCs of different tumor primary locations. A random forest model based on microbial features could classify RC accurately while performing poorly in classifying LCC and RCC. It suggests that there may be more significant differences in microbial features between rectal and colon cancer, while the microbial features of LCC and RCC tended to be similar. Therefore, we used MAASLIN2 and LDA to identify the species enriched in rectal and colon cancer. It was followed by further stratification of the CC to identify species enriched in the RCC and LCC. Some of the results we obtained are consistent with previous studies. A previous study has found that Prevotella, Pyramidobacterium, Selenomonas, and Peptostreptococcus had a higher abundance in proximal CRC, while Fusobacterium, Escherichia Shigella in distant. In this study, Peptostreptococcus stomatis was enriched in RCC, Fusobacterium nucleatum, and Escherichia coli (the same genus as Escherichia Shigella) were enriched in RC [44]. Q. Sheng et al. found genera Veillonella to be enriched in proximal colon cancer, and our study also showed that Veillonella sp T11011 6 tended to be more enriched in CC compared to RC [58]. Miyake T et al. used 16s amplicon microbiome analysis to demonstrate that Fusobacterium predominated in LCC, Blautia, Eryspelotrichales, Holdemanella, Faecalibacterium, Subdoligranulum, and Dorea constituted the dominant intestinal flora in right-side CRC. We also found that Fusobacterium nucleatum was enriched in RC more than CC, and Dorea sp CAG 317 was enriched in RCC more than LCC. But we also found Blautia hydrogenotrophica was enriched in LCC, which is inconsistent with Miyake T et al. study.
CRCs in different primary locations exhibit different patterns of biological behavior and microbial features, suggesting the possibility that gut microbes are associated with the biological behavior of CRCs in different locations. Microbes enriched in different locations of CRC are potentially associated with different molecular carcinogenesis. The molecular mechanisms involved in proximal CRC include High MSI, KRAS mutation, and CMS1, which are related to CC or RCC enriched species. KRAS gene mutation was positively correlated with the level of Actinomyces [59]. R. V. Purcell et al. and Thyra Löwenmark et al. found the high frequency of Parvimonas micra and Peptostreptococcus stomatis in CMS1, and Parvimonas micra was associated with activated CD69 T cells, antigen-presenting HLA-DR B cells M1 and M2 macrophage traits promoting immune response in CRC [60, 61]. On the other hand, the molecular mechanisms involved in distal CRC include High CIN, activation of Wnt pathways, and CMS2, which are related to LC or RC enriched species. Escherichia coli could induce DNA damage in vivo and drive CIN in CRC cells [62]. Fusobacterium nucleatum could promote CRC by inducing Wnt/β-catenin modulator Annexin A1 [63].
Concerning the immune microenvironment, RCC had more immune cell infiltration and immune responsiveness than LCC and RC. Interestingly, we found a part of the genera and species enriched in RCC or CC in this study that was reported to have a role in promoting immune response or enhancing the efficacy of ICIs in various cancers by other studies. In studies of ICIs-treated lung cancer and hepatocellular carcinoma, Veillonella was predominant in lung cancer and hepatocellular carcinoma patients with responses [64, 65]. Yu Huang et al. enhanced the tumor suppressive effect of anti-PD-1 by administration of Megasphaera to mice [66]. Both Streptococcus mutans and Streptococcus mitis have been associated with immunotherapy of tumors. Streptococcus mutans was associated with the effective response of nivolumab treatment in hepatocellular carcinoma patients [67]. In oral squamous cell carcinoma, Streptococcus mitis-reactive CD8 + T cells have lower PD-1 expression, and non-recurrence patients had a higher frequency of Streptococcus mitis-reactive CD8 + T cells suggesting Streptococcus-reactive CD8 T cell responses might contribute to anti-tumor immunity [68]. Peptostreptococcus could improve the effect of ICIs in oral squamous cell carcinoma [69]. The presence of Bifidobacterium breve in the gastrointestinal (GI) tracts of cancer patients could create a more responsive microenvironment and boost the efficacy of immunotherapy [70, 71]. In contrast, Clostridium genera are enriched in RC, some of which are involved in immunosuppressive responses contributing to the progression of CRC [72]. Overall, this suggests a new hypothesis: alterations in gut microbes may also be responsible for the different molecular carcinogenesis and immune microenvironment of CRC in different primary locations.
In addition, some of RCC-enriched and LCC-enriched species in this study were reported to be associated with the progression and prognosis of CRC. Parvimonas micra, which was one of the RCC-enriched species in this study, is an independent risk factor of poor survival in CRC patients. Zhao L et al. transplanted Parvimonas micra into mice significantly promoted colonocyte proliferation and up-regulated genes involved in cell proliferation, stemness, angiogenesis and invasiveness/metastasis [73]. LCC-enriched species reported the association with prognosis of CRC included Blautia, Eubacterium and Slackia. Two studies reported different effects of Blautia on CRC. Silva-Reis R et al. found that Blautia was with high abundance when dimethylhydrazine induced colorectal carcinogenesis [57]. But Kumar R et al. considered that the enrichment of Blautia was correlated with anti-CRC efficacy of dietary rice bran [74]. Increasing abundance level of Eubacterium also was associated with anti-CRC efficacy of berberine [75]. Eubacterium and Slackia are associated with worse OS or DFS in CRC [76]. Since the interpretation of the role of these species is not entirely consistent across reports, it remains to be explored whether altered microbial enrichment is a cause of the different clinical outcomes of RCC and LCC.
Surprisingly, we found that many genera enriched in distal CRC could produce butyrate, including Roseburia, Clostridium, Eubacterium, and Blautia [77–79]. As a metabolite of gut microbes fermenting fiber-rich food residue in colon, butyrate can promote the assembly of tight junction of gut epithelial barrier to inhibit the occurrence and development of CRC [80]. Although many anti-tumor mechanisms have been reported for butyrate, the relationship between butyrate and different primary locations of CRC has not been investigated. Moreover, there are various metabolites presented in the GI tract, some of which are mediators for the gut microbes to perform their functions. There have also been few studies focusing on the metabolomic landscape of CRC at different primary locations. So, the relationship between the primary location of CRC and microbial metabolites still needs to be explored further.
The present study has the following limitations. Although this study included more samples than previous studies and used higher resolution metagenomic data for the meta-analysis, there was a heterogeneity among the different studies. Studies from different sources produced biases in the data: for example, there was no RCC case reported in the study by Ankit Gupta et al. [17]; only the study by Jun Yu et al. reported smoking history [19]; different studies were in different countries and used different sequencing platform; and unknown factors such as the treatment and dietary habits, may also affect microbiome or locations of CRC, leading to the bias in this study. We have used the MAASLIN2 trying to attenuate the bias, but this heterogeneity due to differences in studies is difficult to avoid altogether, which may be one of the reasons our results are not entirely similar to previous similar studies. In addition, the present study revealed a relationship between gut microbiome and the primary location of CRC and that this relationship may involve molecular carcinogenesis and tumor microenvironment. However, the mechanisms through which gut microbes interact with the primary location of CRC need to be studied with methods such as microbes implantation and molecular experiments.