CRC microenvironment (CRCME) is a heterogeneous microenvironment containing a variety of immune cells. Numerous studies have linked high infiltration of CD8⁺ T cells, CD4⁺ type-1 helper T cells (Th1), follicular helper T cells (Tfh), M1 macrophages, natural killer (NK) cells, and dendritic cells (DCs) with good prognosis in CRC. Contrarily, high infiltration of myeloid-derived suppressor cells (MDSCs), B cells, M2 macrophages, and Th17 cells are associated with poor prognosis (20–23). MDSCs induce pre-metastatic niches, increase angiogenesis, migration, and invasion of tumor cells, and make tumor chemoresistant (24–26). MDSCs also suppress the antitumor activity of NK and T cells and recruit other immunosuppressive cells such as regulatory T (Treg) cells into the tumor site (27, 28). The frequency of MDSCs and Treg cells in the blood and tumor of CRC patients is higher than healthy individuals and healthy tissues around the tumor, respectively (21, 22, 29, 30). However, the relationship between the amount of Treg cells and disease prognosis is still controversial, with some studies associating it with a bad prognosis and some with a good prognosis (31–34).

CRCME is immunologically active in which a variety of immune cells are infiltrated, and different tumor-associated antigens (TAA) and tumor-specific antigens (TSA) are expressed. Besides, numerous molecular markers and receptors such as VEGF, EGFR, insulin-like growth factor-1 receptor (IGF-1R), human epidermal growth factor receptor-2 (HER2), integrins, mucin 5AC (MUC5AC), death receptor-5 (DR5), cytotoxic T-lymphocyte-associated protein-4 (CTLA-4), programmed cell death protein-1 (PD1) are overexpressed in the CRCME (35–43). Hence, immunotherapy targeting these molecules could be a promising therapeutic candidate for CRC treatment (18). Various immunotherapy methods, including monoclonal antibodies (mAbs), immune-checkpoint inhibitors (ICIs), adoptive T cell therapy (ACT), and cancer vaccines, are currently used in CRC (44).

Approval of mAbs such as Bevacizumab, Cetuximab, Panitumumab, and Ramucirumab as the first and second line of treatment with chemotherapy in mCRC led to greater interest in CRC immunotherapy (5, 45). In 2004, two mAbs, Bevacizumab and Cetuximab, were approved for the treatment of mCRC (5, 18). Bevacizumab is a humanized antibody against VEGF that inhibits angiogenesis, and combined with chemotherapy, increases overall survival (OS) and progression-free survival (PFS) in mCRC patients (46). Other anti-angiogenic mAbs include Ramucirumab and Tanibirumab, which are anti-VEGF receptors (VEGFR). The former has been approved as the second line of mCRC therapy combined with chemotherapy, and the latter has not been approved thus far but has shown positive results in reducing tumor growth (47, 48). Vanucizumab is a bispecific antibody against VEGF and angiopoietin-2 that suppresses angiogenesis and metastasis (49). Aflibercept is a fusion protein containing the extracellular domains of VEGFR1,2 fused to Fc of human IgG, which inhibits VEGF and, in combination with chemotherapy, improves the OS of mCRC patients (50).

Another group of mAbs approved for mCRC is anti-EGFR mAbs, including Cetuximab, Panitumumab, and Necitumumab (18). The first two mAbs are confirmed as the first and second line of treatment in mCRC patients with wild-type Ras, respectively (51, 52). Necitumumab has a higher affinity for EGFR than the other two mAbs. It has shown positive results in reducing tumor growth in mCRC patients, which should be approved in further trials (53). Various studies have reported that patients with mutations in KRAS and other RAS genes are resistant to anti-EGFR mAbs, suggesting the need to evaluate RAS mutations before selecting appropriate treatment (5, 54–56). Other mAbs used in mCRC, including anti-DR5 and anti-IGF-1R mAbs, showed transient effects as monotherapy or in combination with chemotherapy (57).

In recent decades, ICIs have achieved astonishing success in treating solid tumors that had not previously responded well to treatment (2). ICs physiologically suppress the overreaction of immune cells to prevent autoimmunity. By increasing the expression of ICs, tumor cells misuse their inhibitory signals to escape the antitumor immune responses (58, 59). ICIs block ICs or their ligands on the surface of immune cells or tumor cells to restore the antitumor immune responses (2).

The use of ICIs in unclassified CRC patients did not provide an acceptable overall response. It was observed that some patients responded appropriately to the ICIs while the others were ICI-resistant (60, 61). Further studies revealed that hypermutant tumors express a high level of ICs, leading to appropriate responses to ICIs (62). In fact, in tumors with a high mutational burden, neoantigens are more likely to be produced and presented, and immune cells are more likely to infiltrate hypermutant tumors. That is probably why such tumors have a good prognosis and responses to various immunotherapies (63). CRCME in dMMR/MSI-H tumors is highly infiltrated with CD8⁺ T cells, Th1 cells, macrophages, and enriched with IFN-I (63, 64). Interestingly, the amount of somatic mutations in CRC is associated with treatment response (65). Given the promising results of ICIs in increasing OS and PFS of dMMR/MSI-H patients, the Chinese Society of Clinical Oncology (CSCO) guideline recommends anti-PD1 (Nivolumab and Pembrolizumab) as the first line of treatment for dMMR/MSI-H mCRC patients (66). Besides dMMR/MSI-H patients, POLE-mutated patients also have a high immune cell infiltration and IC expression, resulting in a favorable prognosis and response to ICIs (67).

Unlike hypermutant patients, ICI in pMMR-MSI-L patients did not show promising results, possibly due to the lack of tumor infiltration with immune cells (2, 68). The combination of ICI with radio/chemotherapy and anti-angiogenic agents is currently being investigated in these patients (NCT02563002, NCT03122509). Radio/chemotherapy causes the release of neoantigens through direct damages to the tumor cells (69, 70). The released neoantigens and damage-associated molecular patterns (DAMPs) recruit immune cells into the tumor site, leading to patients’ better response to immunotherapy (69–72). Inhibition of the RAS-AMP activated protein kinase (RAS-AMPK) pathway also increases the tumor infiltration of immune cells and improves the response of pMMR/MSI-L patients to ICI (73–75). Though, it did not show significant changes in the OS of pMMR/MSI-L patients compared to other treatments (73–75).

Bispecific antibodies against TAA and CD3 are another way of increasing the infiltration of immune cells into the CRCME to increase the responses to ICI. Since carcinoembryonic antigen (CEA) is a highly expressed TAA in CRC, CEA-TCB (RG7802) binds to the CEA on the surface of tumor cells and CD3 on the surface of T cells, recruiting more T cells into the tumor site. The combination of CEA-TCB with Atezolizumab (anti-PDL1) was encouraging in pMMR/MSI-L patients, and further investigations are ongoing (NCT02324257, NCT02650713) (76). Targeting other ICs such as T-cell immunoglobulin and mucin domain-containing protein-3 (TIM3), Lymphocyte activation gene-3 protein (LAG3), T cell immunoreceptor with immunoglobulin and ITIM domains (TIGIT), and V-domain immunoglobulin suppressor of T-cell activation (VISTA), as well as the use of agonist for costimulatory molecules such as CD27, OX40, 4-1BB, glucocorticoid-induced tumor necrosis factor receptor (GITR), and CD40 are under investigation in both pMMR and dMMR patients with positive results (2, 64, 77–84).

Besides pMMR/MSI-L patients, the use of ICIs in a large proportion of dMMR/MSI-H patients was not clinically significant (85, 86). For example, tumors with mutations in β2M have defects in the expression of β2-microglobulin with major histocompatibility complex (MHC)-I, leading to antigen presentation deficiency and immune escape of tumors (87). β2M mutation causes resistance to ICI even in dMMR tumors (65, 87). Mutations in genes involved in the antigen processing and presentation pathways as well as in IFN-γ signaling pathways also cause resistance to ICI (64, 88, 89). To sum up, the above cases show that the use of immunotherapy regardless of the tumor and patient conditions leads to treatment inefficiency, which indicates the need to examine each individual’s tumor to select the appropriate treatment option.

Cancer vaccines are immune system boosters that invigorate patients’ immune responses against cancer by exposing tumor antigens to immune cells (44, 146). The immunotherapies discussed so far is targeting TAAs that are overexpressed in the CRCME. Most TAAs targeted in CRC include CEA, EGFR, VEGFR1/2, survivin, mucin-1 (MUC-1), melanoma antigen gene (MAGE), Wilms tumor antigen-1 (WT1), transmembrane-4 superfamily member-5 protein (TM4SF5), mitotic centromere-associated kinesin (MCAK), ring-finger protein-43 (RNF43), translocase of the outer mitochondrial membrane-34 (TOMM34), squamous cell carcinoma antigen recognized by T cells-3 (SART3), insulin-like growth factor II m-RNA-binding protein-3 (IMP3), kinase of the outer chloroplast membrane-1 (KOC1), 5T4, guanylyl cyclase C (GUCY2C), and human telomerase reverse transcriptase (hTERT) (44, 147–160).

Another type of cancer antigen is TSAs or neoantigens, which are nucleotide or polypeptide sequences that are mutated and might be identified as non-self antigens (44). Neoantigens are expressed only on the surface of tumor cells and significantly reduce the risk of adverse effects of vaccination (15, 161). Hypermutant tumors have a higher chance of producing neoantigens, which is one reason for the increase in immune cell infiltration and good prognosis of these tumors (110, 162). The specificity of neoantigens and their association with the prognosis have made neoantigens ideal targets for cancer immunotherapy, especially for personalized cancer vaccines (64). Neoantigens are generated from frameshift and framework mutations, single nucleotide or structural variations, insertions or deletion, and alternative splicing (146). Most of the neoantigens produced in CRC are derived from TTN gene mutations. Mutations in other genes such as TGFBR2, TAF1B, HT001, MARCKS1/2, CDX2, AIM2, TCF7L2, KRAS, PCNXL2, BAX-1, MUC16, SOX9, RNF43, KMT2D, ARID1A, APC, ZFP5N2 could also produce neoantigens (146). Mutations that occur in tumors are divided into driver and passenger types. The products of driver mutations are critical for the survival of tumor cells, and therefore targeting these mutations reduces the likelihood of immune escape by the tumor (15).

Vaccination with short peptides induces a limited immune response. Therefore, the cocktail of several long peptides is usually used in vaccines to induce both class-I and class-II MHC-mediated responses (150, 163, 164). Frameshift mutations generate long neoantigenic peptides with predictable sequences that can be used in vaccines. Interestingly, the number of vaccine peptides that induce the CD8⁺ T cell responses is directly related to patient survival (151). It has been demonstrated that vaccines with more than two neoantigens were able to control the tumor for a long time in preclinical and clinical models (165). In a trial, vaccination with three frameshift-derived neoantigens was performed on advanced-stage CRC patients, which induced a response in all patients and stopped the disease in one patient (166). Due to the high heterogeneity of CRC patients, some studies are attempting to use cocktail neoantigens or establish a library of DNA, RNA, and peptide vaccines as off-the-shelf vaccines (150, 151, 167–169). The list of clinical trials in neoantigen-based off-the-shelf vaccines and ACT is provided in Table 1 .

Neoantigen vaccines are extensively investigated in melanoma and glioblastoma with promising results in increasing patients’ survival (173–175). Given that glioblastoma has a low mutational and neoantigenic burden (8, 176, 177), these results suggest that neoantigen-based vaccines are effective even in tumors with low mutation rates such as pMMR/MSI-L (64, 175). Neoantigen vaccines may improve the infiltration of immune cells into CRCME and increase the immunoscore of tumors, leading to increased tumor responses to ICIs and other immunotherapies (64). The combination of neoantigen vaccines with ICI is currently being investigated in clinical trials (NCT02600949, NCT03289962).

In neoantigen vaccines, prioritizing the selection of neoantigens is a critical step influencing the effectiveness of vaccination. Some studies prioritize selecting those with a high prevalence in CRC patients (146, 167, 168). However, in personalized immunotherapy, the goal is to identify the neoantigens expressed in the patient’s tumor and select the most immunogenic and appropriate ones for vaccination. The application of the NGS technique to perform whole-exome sequencing, whole-genome sequencing, and RNA sequencing, together with the use of omics data, helps us identify mutated regions and neoantigens created in the patient tumor to be used in the personalized vaccine (146).

Due to the presentation of all antigens in the body by MHCs, determining patient MHC profile and selecting appropriate peptides with high binding affinity to the patient MHCs should also be considered to select immunogenic neoantigens (146). Many personalized vaccine studies consider the presence of targeted neoantigens in the patient’s tumor as a criterion for vaccine design. The next step in achieving an effective vaccine is to assess the patient’s immune status in terms of the presence or absence of previous immune response to the selected neoantigen (178, 179). Tumor-derived neoantigens that have not been able to stimulate the immune responses cannot induce strong immune responses in the vaccine formulation (180, 181). Therefore, after selecting the ideal neoantigens and matching them with the patient MHC, it is necessary to evaluate the pre-existing immune responses against those neoantigens by determining the frequency of antigen-specific cytotoxic T cells (CTLs) as well as the antigen-specific IgG titer in the serum (15, 182). Considering the pre-existing immunity in neoantigens selection could enhance the vaccination efficacy and reduce the adverse events (15). Clinical trials using this type of personalized vaccination have shown encouraging results in inducing an antitumor immune response and increasing OS even in chemoresistant CRC patients (178, 179, 182). The process of personalized immunotherapy in CRC is illustrated in Figure 1 .

Neoantigen vaccines can be designed and produced in various forms. Peptide/protein vaccines are safe and have a convenient and cost-effective production process but usually do not have sufficient immunogenicity and require adjuvant (178, 183, 184). In addition, the antigenic escape necessitates the need to use antigen cocktails (151, 178, 184). The AutoSynVax vaccine is a personalized vaccine using neoantigens and QS-21 adjuvant that is undergoing a phase 1 clinical trial in advanced cancers (NCT02992977).

DNA and mRNA vaccines are highly immunogenic due to the presentation of produced antigens by both MHC-I and -II pathways. They also stimulate the innate immune sensors in the cytosol (185–187). The mRNA vaccines are safer than DNA vaccines due to the lack of concerns about unwilling integration to the host genome (44, 188). Also, the convenience, speed, and inexpensiveness of mRNA manipulation have led to the widespread acceptance of mRNA vaccines in personalized immunotherapies, which has yielded promising results in CRC (188–190). DCs are ideal candidates to carry the neoantigens into the tumor because they can present neoantigens via both MHC-I/-II pathways and provide costimulatory molecules required for optimum immune responses. They can be pulsed ex vivo with neoantigens or mRNA, matured with cytokines, and then returned to the patient as autologous DC vaccines (191). These vaccines are being tested in clinical trials ( Table 2 ).

Viral, bacterial, and yeast vectors can also be loaded with genes encoding neoantigens and induce robust immune responses against the tumor (44). These vectors release pathogen-associated molecular patterns (PAMPs) that stimulate the innate and adaptive immune response and increase the infiltration of immune cells into the tumor site (206–210). Adenoviruses, lentiviruses, retroviruses, and Poxviruses are important viral vectors in vaccine design (211). Clinical trials with viral vectors loaded with neoantigens and TAAs have been performed on mCRC patients with promising results in inducing the immune responses and improving OS (212, 213). However, viral vector safety concerns such as the possibility of mutation and becoming a pathogen still remain and need further studies (214, 215). The use of engineered bacteria as vectors of neoantigens in clinical trials has also yielded positive results. ADXS-NEO is a personalized neoantigen vaccine with Listeria monocytogenes vector that increased CD8⁺ T cell infiltration in CRCME in 40% of patients and induced specific responses against 90% of neoantigens (NCT03265080). Yeast vectors are a promising future option for cancer vaccines due to their high safety, ease of large-scale production, no need for adjuvants, and induction of effective responses against neoantigens (207, 216). Several clinical trials are investigating the effectiveness of yeast vectors in mCRC vaccines (44).

Another strategy for personalized vaccination is the use of autologous tumor cells or tumor lysates. Vaccination with whole tumor lysate reduces the possibility of immune escape due to the presence of all tumor antigens (44). However, many of these antigens are normal self-antigens that do not stimulate the immune response and reduce the effectiveness of whole lysate vaccines. Strategies such as using oncolytic viruses in vivo and in vitro or tumor cells infected with oncolytic viruses release PAMPs, which increases immune responses (217, 218). The results of phase I and II trials showed that the use of Newcastle virus-infected tumor cells reduced recurrence and increased OS in CRC patients (44, 219). However, the high levels of self-antigens present in tumor lysates cause the lack of specificity of immune responses and increase the possibility of promoting autoimmune responses, limiting the use of this method in susceptible individuals (220, 221).

ACT is a cancer immunotherapy method in which T cells are collected from the tumor, lymph nodes, or peripheral blood of a patient and returned to the patient’s body after proliferation and selection of tumor-specific T cells. ACT can be performed with unmanipulated cells or engineered cells that express chimeric antigen receptors (CAR-T cells). CAR-T cells are independent of MHCs, and due to carrying costimulatory domains, they could induce strong antitumor responses (2). CAR-T cells are mainly against TAAs that overexpress in CRC, including CEA, EGFR, mesothelin, MUC-1, NKG2D ligand, HER2, c-met, CD133, GUCY2C, epithelial cell adhesion molecule (EpCAM), and Tumor-associated glycoprotein (TAG)-72 (18, 157). These CAR-T cells contain immune activating domains of CD28 and CD137. In the context of mCRC, CAR-T cells as monotherapy or in combination with cytokines such as IL-12 had encouraging effects such as tumor reduction and long-term disease stability in some patients (222–225). However, challenges such as on-target/off-tumor toxicity and damage to other organs due to the lack of specificity of target antigens are seen. Identification of TSAs is one of the current challenges in CAR-T cell therapy (18). A second concern is cytokine release syndrome due to the CAR-T cells activation following binding to antigens in both tumor cells and normal cells (226).

The use of tumor-specific unmanipulated cells has also yielded positive results in CRC. In one study, tumor-infiltrating lymphocytes (TILs) were collected from metastatic lesions of a patient carrying KRAS-G12D mutation. The mutation-specific CD8⁺ T cells were selected and returned to the patient, resulting in the elimination of 85% of metastatic lesions (227). In another study, sentinel lymph node T was used instead of TILs, which resulted in complete response and disease stability in some patients and partial response in others (228, 229). The combination of ACT with chemotherapy and bevacizumab caused 80% overall response, 26.7% complete response, and stopped tumor progression in stage IV CRC patients (230). Various trials are investigating the effects of ACT alone or combined with other immunotherapies such as ICI in CRC patients (NCT03935893, NCT02757391, NCT01174121, NCT03904537). In these trials, various omics data are used to identify personalized antigens (146). These primary successes suggest that ACT, along with neoantigen vaccines, could be promising candidates for personalized immunotherapy in CRC.

Human microbiome is defined as all the microbiota in and on the human body plus their genomes, structural elements, and productions (231). The microbiota composition depends on genetic and environmental factors such as diet, exposure to chemicals, drugs, especially antibiotics, and is unique to each individual (232). Gut microbiota plays an essential role in the formation and regulation of gastrointestinal immune responses by producing various metabolites (233). Thus, dysbiosis in gut microbiota composition alters their metabolites such as short-chain fatty acids SCFAs, polyphenols, tryptophan catabolites, and some vitamins, and is associated with the pathogenesis of many diseases, including CRC (234). The researchers found several pathogen-derived proteins in the gastrointestinal tract of CRC patients (234, 235). They also showed that the high levels of antibodies against these proteins might have a diagnostic value for CRC patients (235, 236).

Besides, microbiota composition plays a critical role in patients’ treatment responses (237, 238). Presumably, the microbiota diversity among patients is one of the reasons for interpatient variety in response to cancer treatment (238). Microbiota can affect the response to chemotherapy. It has been observed that some bacteria metabolize and reduce the effects of chemotherapy at the tumor site (239). Besides, microbiota are chief players in regulating inflammatory responses in GI, through which they can influence the response of CRC patients to immunotherapy (240–242). Evidence suggests that exposure to antibiotics during anti-PD1 treatment alters the patients’ responses to anti-PD1 (241). Comparing the microbiota of patients who were well-responsive to anti-PD1 with those of patients who were poor-responsive to anti-PD1 showed that the presence of some bacteria (such as Akkermansia, Faecalibacterium, and Bifidobacterium) was associated with a better response to immunotherapy (238, 241). Contrarily, high levels of bacteria such as Bacteroidale are associated with a low response to anti-PD1 therapy (238, 241). This finding indicates that disruption of the microbiota network and reduction of beneficial bacteria reduce the patients’ responses to immunotherapy.

Interestingly, by manipulating the microbiota composition, responses to immunotherapy can be improved. Fecal microbiota transplant (FMT) is a well-known way to manipulate the network of GI microbiota. In FMT, the stool of patients who have responded well to treatment is transferred to the GI tract of non-responsive patients. Chen 2019 Matson 2021 Zhou 2021. It was observed that FMT of mice from anti-PD1-sensitive patients induced a favorable response to anti-PD1 in transplanted mice (241, 243). On the other hand, mice that received FMT from anti-PD1-resistant patients were still resistant to anti-PD1 after transplantation (241, 243). The use of microbiota in improving treatment response has entered clinical studies, and several clinical trials are investigating the combination of FMT with chemotherapy and immunotherapy (NCT04130763, NCT03782428). These studies, which are still in early phases, are investigating the role of FMT or probiotic supplements along with ICIs in CRC patients who have not previously responded well to immunotherapy.

In a nutshell, the evidence suggests that the microbiome is a determinant variable in immunotherapy and is involved in all aspects of CRC, including the diagnosis, prognosis, monitoring, and response to immunotherapy. Therefore, microbiomics should be used along with other omics data in optimizing personalized immunotherapy (238). In this era, molecular pathological epidemiology (MPE) has been recently developed to investigate the relationship between gut microbiota, host, environmental factors, diet, disease etiology and pathogenesis to investigate the role of microbiota and its factors in disease development and treatment response (244–246).

Although personalized immunotherapy approaches promise a new horizon in the treatment of CRC, longtime challenges of researchers with cancer treatment show that cancer is the most complex disease that is not supposed to cure with monotherapy (247). Regarding the high heterogeneity of CRC, it seems that it could escape even from personalized immunotherapy (44). Therefore, multi-aspect combinations of immunotherapy, chemotherapy, radiotherapy and other targeted therapies are required to fight cancers. It has been revealed that the combination of immunotherapies such as ICIs and mAbs with chemotherapy and radiotherapy could reinvigorate the antitumor responses and increase the patients’ survival (247). Personalized immunotherapy could also be combined with other therapeutic approaches to maximize the antitumor effects. In this regard, combination therapy using cancer vaccines (peptide, DNA, RNA, DC, and viral vector vaccines), adoptive cell therapy, ICIs, cytokines, chemotherapy, and radiotherapy increased antitumor effects leading to improved overall survival (151, 213, 227, 247–255). A phase I clinical trial of a neoantigen-encoding mRNA cancer vaccine combined with PD-1 blocking showed acceptable safety and well induction of neoantigen-specific immunity leading to partial and complete response in MSI-H CRC patients (189).

Interestingly, combination therapy is also better to be personalized to achieve the optimum response. The systems biology and simulation models can help find the best combination therapy in each patient. More recently, a quantitative systems pharmacology model predicted the efficacy of monotherapy with a bispecific T cell engager, a PD-L1 blocker, and combination therapy in CRC patients based on their individual characteristics (256). The development of these models could offer the best combination therapy for each patient to improve clinical efficacy.