The morphology of nerves in the large intestine significantly changes during CRC development. Previous studies have explored nerve changes in CRC using immunohistochemical methods in small samples. However, there has been no systematic analysis of nerve diameters and densities using immunofluorescence staining with large sample data. In our study, more than 20,000 nerves were counted after staining for specific markers of nerve fibers and glial cells in 155 patients, and analyses based on different clinicopathological characteristics were performed to obtain a global view of nerve changes in CRC.
Our study revealed that as the depth of tumor invasion increased, the original neural structure significantly deteriorated. This destruction led to atrophy and the development of disorders in the submucosal plexus, muscular plexus, and extrinsic nerve bundle (Fig. 4B). This is also the reason why the nerve densities at on-tumor sites were decreased compared to those at off-tumor sites. Another interesting finding was that the nerves at the on-tumor sites were thicker than those at the off-tumor sites, even though the nerve diameters in the core region and peritumoral region were comparable. This finding was consistent with the findings of Iwasaki et al. in pancreatic cancer, in which nerve density was decreased but nerves became hypertrophic[19]. As a part of the TME, nerves have complicated interactions with tumors. Tumor cells produce a series of neurotrophins (NGF/BDNF), which can be secreted into the TME in the form of paracrine or exosomal mediators to promote axon generation and the growth of nerves[20]. During the process of tumor invasion, the originally tightly bound nerve bundles gradually become loose. These two effects result in more loose nerve bundles and thicker nerve diameters at on-tumor sites. For the same reasons, when PNI occurs, the nerve bundles in close contact with tumor cells become more enlarged. As the distance decreases, the neurotrophins from the tumor cells are more likely to reach the nerves, resulting in the enlargement of nerves and an increase in the number of nerves. Therefore, when the PNI was positive in this study, the nerves in the core region were thicker and denser. After analyzing the damaged nerves, we hypothesized that PNI may be involved in the process of tumor invasion into the nerves. In the PNI stage, the nerve density is greater than that in subsequent stages. Notably, neurotransmitters and neurotrophins produced by nerves also play important roles in the processes of tumor cell proliferation, invasion, and metastasis (Fig. 4A)[21–26]. Furthermore, the application of neural-specific staining resulted in more PNI being detected in the immunofluorescence analysis than in the HE staining analysis, which suggested that neural-specific antibodies should be employed when identifying PNI in routine pathological tests, regardless of whether immunohistochemistry or immunofluorescence detection is used.
In addition, because of the enormous destructive power of tumor cells, regardless of pathological parameters, the nerve densities in the core region were almost lower than those in the peritumoral region at on-tumor sites. The prognosis of CRC patients with KRAS mutations is poorer than that of patients with wild-type KRAS due to the continuous stimulation of downstream signaling pathways and resistance to anti-EGFR therapy induced by KRAS mutations[27]. The nerve density in the core region in the KRAS mutation group was lower than that in the wild-type group. One possible reason is that KRAS drives an invasive and metastatic phenotype of CRC[28]. KRAS-mutant tumor cells are more likely to erode surrounding nerves, causing damage to neural structures. Another reason is that KRAS can lead to an immunosuppressive microenvironment through various mechanisms, such as the activation of tumor-associated macrophages (TAMs) and hypoxia, which are more likely to lead to the destruction of nerves[29, 30]. According to a recent single-cell analysis, the KRAS-mutated group was more enriched in tumorigenesis and immune evasion pathways than the wild-type group[31].
The nerve density in the core region was greater in the < 2 cm group than in the larger group. We speculated that when the tumor size is still small, it cannot cause significant nerve changes. In addition, in patients with distant metastasis (M0 or AJCC-IV), the nerve density in the core region was greater. The reason underlying this phenomenon may be related to tumor-nerve interactions; nerve cells secrete a variety of neurotrophins, such as NGF, which can promote CRC cell metastasis through the TrkA/MAPK/Erk axis (Fig. 4a) [19]. Similarly, in the early stages of CRC, such as when tumors are smaller or have lower stages (T stage or lower AJCC stage), the tumors have not yet contacted and damaged peripheral nerves, allowing more nerves to be preserved, resulting in a greater density of peritumoral nerves. Moreover, due to the relatively abundant nerves in the rectum, the densities of the peritumoral nerves in the rectum were greater than those in the left colon. This finding is in line with Albo's research[32]. Similarly, in the late stage of local CRC (T2/T3 or AJCC-II/III), due to the involvement of more tumor cells, more hypertrophic nerves were found at on-tumor sites. Furthermore, the histological grade also affected the thickness of the nerves. The peritumoral nerve diameters in the Grade 2 group were greater than those in the Grade 3 group. Another study revealed a similar phenomenon: the relative areas of both the Auerbach and Meissner plexuses decreased with increasing histological grade[33].
Previous studies have also investigated alterations in intestinal nerves induced by CRC. Consistent with our findings, Godlewski et al. conducted an immunohistochemistry study on 15 patients and observed nerve damage in CRC[34]. In addition, MPs located close to the invading cancer were significantly smaller and had a lower number of neurons per plexus than those located farther from the tumor[35]. Similarly, the size of ganglions is proportional to the distance from the invading tumor and inversely proportional to the synaptophysin content, which may be related to degenerative changes and dysfunction of ENS cells[36]. Electron microscopy was used to demonstrate how the ultrastructure of the muscle plexus is affected by CRC cells. Scattered apoptotic neurons and glial cells were observed, accompanied by mast cells and plasma cells[37]. However, immunofluorescence staining of caspase 3 or caspase 8 in the ENS plexuses at on-tumor sites and off-tumor sites suggested that the atrophy of the myenteric plexuses due to CRC invasion was not caused by apoptosis or necrosis[38]. Atrophy of the MP at on-tumor sites was probably the combined result of stroma reactions, tumor mass-related increased local tissue pressure, and hypoxia[39]. Tissue decomposition may be induced by the secretome and/or extracellular vesicles containing molecules that are released into the TME by tumor cells[39]. Given the important role of the nervous system in maintaining intestinal peristalsis and homeostasis, secretion, and infection resistance, neurological damage that occurs during the development of CRC is inevitably accompanied by a series of symptoms. Patients with Hirschsprung disease (HSCR) and chagasic disease often experience megacolon due to loss of the enteric nervous system[40, 41]. Similarly, the symptoms of pain, diarrhea, constipation, and changes in bowel frequency in CRC patients[42] may also be related to damage to the nervous system in the intestine.
In previous studies, in addition to PNI, neurogenesis, which manifests as an increase in nerve density and has been confirmed in many malignant tumors, such as colorectal cancer, pancreatic cancer, head and neck tumors, and prostate cancer, has also been reported to be a marker of malignant tumors[32, 43, 44]. This seems to contradict our research. However, it should be noted that neurogenesis often refers specifically to small newly grown nerve fibers, which are significantly different from the nerve trunks above 10 µm in our study. In addition, other studies in CRC patients did not distinguish between the core and peritumoral regions, as we did in this study.
Our research has several limitations: each patient had only one pathological section, and thus, the samples did not represent the neural distribution of the entire tumor. In the future, a three-dimensional perspective may provide more in-depth data. In addition, our neural-specific markers NF-L and S100β failed to distinguish nerves of different classes, and more specific markers are needed in the future to distinguish between intrinsic or extrinsic nerves (sympathetic, parasympathetic, or sensory). In addition, the nerves in the submucosal and muscular layers are uneven in thickness, but we did not separate the nerves in these two areas for statistical analysis. This is an inadequacy of our article. The subsequent research methods can be improved, for example, by analyzing the enteric nervous system and peripheral nervous system separately.