A Systematic Review on HOX Genes as Potential Biomarkers in Colorectal Cancer: An Emerging Role of HOXB9

Emerging evidence shows that Homeobox (HOX) genes are important in carcinogenesis, and their dysregulation has been linked with metastatic potential and poor prognosis. This review (PROSPERO-CRD42020190953) aims to systematically investigate the role of HOX genes as biomarkers in CRC and the impact of their modulation on tumour growth and progression. The MEDLINE, EMBASE, Web of Science and Cochrane databases were searched for eligible studies exploring two research questions: (a) the clinicopathological and prognostic significance of HOX dysregulation in patients with CRC and (b) the functional role of HOX genes in CRC progression. Twenty-five studies enrolling 3003 CRC patients, showed that aberrant expression of HOX proteins was significantly related to tumour depth, nodal invasion, distant metastases, advanced stage and poor prognosis. A post-hoc meta-analysis on HOXB9 showed that its overexpression was significantly associated with the presence of distant metastases (pooled OR 4.14, 95% CI 1.64–10.43, I2 = 0%, p = 0.003). Twenty-two preclinical studies showed that HOX proteins are crucially related to tumour growth and metastatic potential by affecting cell proliferation and altering the expression of epithelial-mesenchymal transition modulators. In conclusion, HOX proteins may play vital roles in CRC progression and are associated with overall survival. HOXB9 may be a critical transcription factor in CRC.


Introduction
Colorectal Cancer (CRC) is the most common gastrointestinal malignancy and the third leading cause of cancer-related death worldwide [1,2]. Despite significant advances in diagnostic and therapeutic strategies, the prognosis for CRC patients remains poor, indicating that cancerous cells are not entirely eradicated by current therapies, thus leading to metastatic disease which is the primary cause of cancer-related mortality [3]. CRC arises as a result of the accumulation of genetic and epigenetic changes, which transform normal glandular epithelial cells into invasive carcinomas and eventually progress into metastatic disease [4]. Colorectal carcinogenesis is a complex multistep process involving the dysregulation of oncogenes or tumour suppressor genes related to initiation, progression and resistance to therapy [4,5]. Therefore, there is an urgent need to identify novel biomarkers which could be used to predict prognosis and act as therapeutic targets.
In recent decades, a lot of attention has been paid to the role of homeobox (HOX) genes in cancer [6]. HOX genes encode a highly conserved family of homeodomain-containing transcription factors which play essential roles in embryonic development including morphogenesis, organogenesis and differentiation [7]. HOX proteins control various cellular processes by regulating the expression of several downstream target genes; hence they can alter cell behaviour such as proliferation, invasion and migration [7]. The human Int. J. Mol. Sci. 2021, 22, 13429 2 of 18 genome contains 39 HOX genes which are classified into four clusters, HOXA, HOXB, HOXC and HOXD based on their sequence similarity and chromosomal position, as shown in Figure 1 [8]. HOX genes are characterized as master regulators during development and alterations in their expression leads to developmental abnormalities and has been reported to be associated with an increased incidence of malignant tumours in humans [9,10]. Numerous studies have shown that HOX genes can act either as oncogenes or tumour suppressors depending on cancer type. For instance, HOXA9 has an oncogenic role in acute leukaemia whereas it acts as a tumour suppressor in breast cancer by regulating the expression of Breast Cancer gene 1 (BRCA1) [11,12]. Additionally, HOXB9 has been shown to inhibit cancer cell proliferation in gastric carcinoma whilst it demonstrates an oncogenic role in breast cancer [13,14].
Emerging evidence shows that HOX transcription factors have significant contributions in the hallmarks of cancer and especially in the proliferative advantage as well as in the invasion and metastasis element, and therefore may play important roles in tumour progression [15]. Expression of HOX genes is dysregulated and often reported to be associated with aggressive nature of tumour biology and poor survival in various types of cancers [16][17][18][19]. In CRC, several studies have revealed that some HOX genes demonstrate aberrant expression in cancerous tissues suggesting that they be considered as potential biomarkers. Although, there is increasing evidence that HOX genes may be important in CRC, a systematic understanding regarding their role in CRC progression and their clinicopathological significance is still lacking. To better understand the existing evidence with regard to the prognostic and functional role of HOX genes in CRC, the authors performed a systematic review of the current literature. Specifically, this review aims to answer the following research questions: (a) What are the clinicopathological and prognostic significance of HOX genes dysregulation in CRC and (b) What is the functional role of HOX genes in CRC progression? Figure 1. HOX genes in the human genome. Adapted from Durston et al. [8] The colour coding represents the correspondence between the genomic order of each HOX gene in the chromosomal cluster and the segmental identity in a human embryo (Microsoft PowerPoint software was used to create this figure). Emerging evidence shows that HOX transcription factors have significant contributions in the hallmarks of cancer and especially in the proliferative advantage as well as in the invasion and metastasis element, and therefore may play important roles in tumour progression [15]. Expression of HOX genes is dysregulated and often reported to be associated with aggressive nature of tumour biology and poor survival in various types of cancers [16][17][18][19]. In CRC, several studies have revealed that some HOX genes demonstrate aberrant expression in cancerous tissues suggesting that they be considered as potential biomarkers. Although, there is increasing evidence that HOX genes may be important in CRC, a systematic understanding regarding their role in CRC progression and their clinicopathological significance is still lacking. To better understand the existing evidence with regard to the prognostic and functional role of HOX genes in CRC, the authors performed a systematic review of the current literature. Specifically, this review aims to answer the following research questions: (a) What are the clinicopathological and prognostic significance of HOX genes dysregulation in CRC and (b) What is the functional role of HOX genes in CRC progression?

Search Strategy
A literature search was conducted for eligible studies in the Medline (https://pubmed.ncbi. nlm.nih.gov/advanced/, 12 July 2020), EMBASE (https://www.embase.com, 19 July 2020), Web of Science (https://www.webofknowledge.com, 25 July 2020) and Cochrane Database (https://www.cochranelibrary.com/, 2 August 2020) search engines in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [20]. The end date of the retrieval period was the 1 July 2021. A search strategy was developed in Medline (https://pubmed.ncbi.nlm.nih.gov/advanced/, 31 March 2020) including keywords, Medical Subject Headings (MESH) and synonyms related to HOX genes, colorectal and neoplasms. The strategy was adapted for the other databases using separate algorithms for each search engine (supplementary material, ESM_1). This systematic review was registered in the international prospective register of systematic reviews (PROSPERO, https://www.crd.york.ac.uk/prospero/) with the identification number CRD42020190953 (registration date was 29 June 2020).

Eligibility Criteria and Study Selection
Three reviewers (EM, GF and AA) independently selected and identified eligible English language studies based on predefined inclusion criteria according to the research question. Discrepancies between the three reviewers were resolved by discussion or with a 4th author.
For the first research question, studies conducted on individuals over 18 years old with sporadic colorectal adenocarcinoma and reporting on HOX genes were included. Specifically, studies reporting at least one of the following criteria were included: (i) studies on HOX dysregulation between cancer and normal tissue, (ii) studies on the association of HOX genes with clinicopathological characteristics of CRC, and (iii) studies reporting outcome measures such as OS and DFS. Exclusion criteria were studies reporting the dysregulation of HOX genes in cell lines or animal tissues. Comparison groups were selected to be the following: cancer to normal colorectal tissue as well as high versus low HOX gene expression patient group. Outcomes were defined as: HOX gene dysregulation, tumour depth, lymph node status, metastases, stage of colorectal cancer disease, grade of disease, disease free survival (DFS) and overall survival (OS) rates.
For the second research question, laboratory-based and animal research studies were included reporting on the effect of HOX gene expression in CRC cell growth. The intervention was considered gene expression editing to either suppress or overexpress the HOX gene of interest. The intervention group in the included studies consisted of either human colorectal cancer cell lines or animal models which had an altered expression of HOX genes. Therefore, the comparison groups were defined as edited versus non-edited human CRC cell lines or animal models. Outcomes were selected to be cell proliferation, cell migration, cell invasion in vitro as well as tumour growth and metastases in vivo.
Studies such as case reports, editorials, opinions, conference abstracts, reviews and other secondary research studies were excluded. Studies not using the universal HOX chromosomal cluster terminology as described in Figure 1 for reporting findings were also excluded for both research questions.

Data Extraction, Synthesis and Quality Assessment
For each study, the following details were extracted on publication year, the surname of the first author, study design, participant characteristics (sample size, gender), study characteristics (intervention and control group, endpoint assays) and outcomes. The quality of each eligible primary study involving human participants was assessed using the National Heart, Lung and Blood Institute (NIH) study quality assessment tools for case studies, available online at https://www.nhlbi.nih.gov/health-topics/study-qualityassessment-tools (last accessed 20 December 2020), (Table S1). Preclinical animal studies were assessed using the Systematic Review Centre for Laboratory animal experimentation risk of bias tool (SYRCLE'S), (Table S2) [21,22]. For cell line studies, no established quality assessment tool is currently available [21]. The results were summarised narratively according to each research question using a qualitative data synthesis approach. We conducted a post-hoc random-effects meta-analysis to summarise effect estimates when at least two peer-reviewed studies were available and were sufficiently homogenous in terms of subjects involved, interventions and outcomes [23]. Results were expressed as odds ratio (OR) with 95% confidence intervals (CIs) using Review Manager Version 5.4.1 (Cochrane collaboration) I 2 statistics and p-value were calculated to assess the heterogeneity between the studies. Forest plots were produced to present the ORs with 95% CIs, the percentage weight and the heterogeneity between studies included in each meta-analysis.

Study Selection
The initial literature search identified 2548 eligible citations. Following duplicate citation removal 1498 studies were screened for eligibility. For research questions 1 and 2, a total of 25 and 22 studies met our inclusion criteria, respectively and were included in the final analysis. The process of literature retrieval, according to PRISMA guidelines, is shown in detail in Figure 2. Study characteristics and findings are presented for each research question separately. assessed using the Systematic Review Centre for Laboratory animal experimentation risk of bias tool (SYRCLE'S), (Table S2,) [21,22]. For cell line studies, no established quality assessment tool is currently available [21]. The results were summarised narratively according to each research question using a qualitative data synthesis approach. We conducted a post-hoc random-effects meta-analysis to summarise effect estimates when at least two peer-reviewed studies were available and were sufficiently homogenous in terms of subjects involved, interventions and outcomes [23]. Results were expressed as odds ratio (OR) with 95% confidence intervals (CIs) using Review Manager Version 5.4.1 (Cochrane collaboration) I 2 statistics and p-value were calculated to assess the heterogeneity between the studies. Forest plots were produced to present the ORs with 95% CIs, the percentage weight and the heterogeneity between studies included in each meta-analysis.

Study Selection
The initial literature search identified 2548 eligible citations. Following duplicate citation removal 1498 studies were screened for eligibility. For research questions 1 and 2, a total of 25 and 22 studies met our inclusion criteria, respectively and were included in the final analysis. The process of literature retrieval, according to PRISMA guidelines, is shown in detail in Figure 2. Study characteristics and findings are presented for each research question separately.

Study Characteristics
A total of 3003 patients with stage I-IV CRC between 20 and 90 years old were included in the studies . All 25 studies were single-centre and were published between 1997 and 2020. Two studies included patients who either received or did not receive
The impact of HOX protein expression on OS and DFS in patients with CRC was investigated by 12 [24][25][26]28,36,[38][39][40]43,[46][47][48] and 4 [28,35,36,42] studies, respectively, as shown in Table 1. OS and DFS rates in patients with high versus low expression of the HOX protein of interest were compared. Most of the studies used a semiquantitative IHC approach to score the expression levels of HOX proteins, whereas two studies used the median mRNA level as a cut-off value [25,28].

Findings
Regarding HOX gene dysregulation and its clinicopathological significance in CRC, 26 out of 39 HOX genes were identified in this systematic review to be differentially expressed in cancerous versus normal colon tissues with 15 of them being overexpressed and six being downregulated, as shown in Table 1 [28,31,32,[37][38][39][44][45][46][47][48]. Discrepancies were reported between studies for five HOX genes (A4, B8, B9, B13 and D10). Among the dysregulated HOX genes and their protein products, several were found to demonstrate potential clinical significance in CRC. Three studies reported that patients with high HOX (B7, B9, C6) protein expression levels demonstrated significantly advanced T status in comparison with patients with low expression [26,42,43] (Table 1). Post-hoc meta-analysis for HOXB9 protein found no statistical association between high HOXB9 and tumour depth (OR 0.92, 95% CI: 0.21-3.97, p = 0.910), (Figure 3a). Similarly, regarding N status, six studies showed that the percentage of patients with regional lymph node invasion was significantly increased in the high HOX protein expression group for the following: HOXA3, A9, B8, B9, B13, C6, D1 and D9 [24,27,37,40,43,48]. Post-hoc meta-analysis for HOXB9 revealed no statistical association between high HOXB9 and lymph node invasion (OR 1.55, 95% CI: 0.45-5.34, p = 0.490), (Figure 3b). Inverse correlation between HOX expression and nodal invasion was demonstrated only for HOXD10 by Wang et al. [44]. Eight studies investigated the presence of metastatic disease according to HOX expression levels [28,40,42,43,[45][46][47][48]. Two studies by Liao et al. [26] and Huang et al. [48] for HOXB7 and B9, respectively, reported that significantly more patients had metastatic disease in the high HOX expression group in comparison with the low expression one (Table 1). Three studies (Huang et al., Zhan et al. and Carbone et al.) were included in the post-hoc meta-analyses regarding HOXB9 and the presence of metastatic disease. High HOXB9 expression was found to be associated with a significant risk for metastatic disease (OR 4.14, 95% CI: 1.64-10.43, p = 0.003), (Figure 3c). Findings remained significant when sensitivity analyses were performed by excluding one study each time (data not shown herein). Eleven studies have reported findings regarding the stage of the disease and level of HOX expression [24,26,28,30,35,36,38,40,45,47,48]. For HOXA3, A9, B7 and D9, high expression groups correlated with an increased number of patients with advanced disease in comparison with the low expression groups [24,26,36,40]. For HOXB9 studies by Zhan et al. [47] and Huang et al. [48] reported antithetical findings (Table 1). For HOXB9, post-hoc meta-analysis showed no statistical correlation between advanced stage and high HOXB9 expression (OR 1.43, 95% CI: 0.49-4.21, p = 0.520), (Figure 3d). Most studies reported no significant association between tumour differentiation and HOX expression level, except for D9, where Liu et al. reported that high HOXD9 levels were significantly associated with poor differentiation [24]. The studies by Hoshino et al. [46] and Zhan et al. [47] reported contradictory results regarding HOXB9 expression levels and their association with CRC differentiation (Table 1). Given the limited available data, it appears that there is no significant association between age, sex and CEA with HOX expression levels. Table 1. Included studies reporting on clinicopathological and prognostic significance of HOX genes in CRC. *: p < 0.05, **: p < 0.01, ***: p < 0.001, ↑: increased, ↓: decreased. Arrows without a * or NR symbol represent findings that are characterized as borderline significant with a p-value ranging between 0.051 and 0.1.   Odds ratios and 95% confidence intervals (95% CIs), were pooled using random-effect meta-analysis. Blue squares indicate the effect size for each study (ORs between high and the low HOXB9 expression group) and the length of the lines indicate the 95% CIs. The size of the square represents its weight in the analysis. The black diamond on the bottom of the forest plot indicates the overall weighted effect size. I 2 indicates between-study heterogeneity. Detailed characteristics of each study are available in Table 1  The prognostic role of HOX dysregulation (A3, B7, B8, C6, C11, D9, D10) was investigated by 14 studies [24][25][26]28,35,36,[38][39][40]42,43,[46][47][48]. Specifically, patients who were characterised by high HOX expression levels had significantly worse survival rates in comparison with the ones with low expression levels. Only two studies by Ji et al. [43] and Liao et al. [26] conducted multivariable analysis and reported that HOXC6 and HOXB7 are independent prognostic markers in patients with CRC (Table 1). HOXB9 was investigated by four studies that reported contradictory results with regard to its positive or negative impact on survival [38,[46][47][48]. In terms of DFS, studies showed that high expression levels of HOXA3, A10, B8 and B9 were significantly associated with worse DFS rates in patients with CRC [28,35,36,42]. HOXA10 and B9 were additionally reported to serve as independent risk factors for worse DFS by Yuan et al. [35] and Carbone et al. [42], respectively, in a multivariable analysis model (Table 1).
Ten studies investigated the effect of HOX genes differential expression in the metastatic potential in CRC [24,28,34,41,[47][48][49][50]52,55]. The primary outcomes were the percentage of cells that showed invasion and/or migration in the relevant assays between the control versus the intervention group. Secondary outcomes were molecular markers involved in the Epithelial-Mesenchymal Transition (EMT), being reported by six studies [24,34,41,48,52,55]. Five studies provided additional results from in vivo experiments by assessing the number of lung/liver metastases as the primary outcome [24,28,47,48,55], ( Table 2). One study reported markers involved in angiogenesis and vessel formation in vivo [46].

Findings
Regarding HOX dysregulation and tumour growth in CRC, of the eighteen HOX genes and their protein products that have been investigated to date, fifteen were found to have oncogenic properties whereas only three were reported to exert tumour suppressive functions. Specifically, loss of function in vitro studies showed that knockdown of HOXA1, A3, A4, A9, A10, B8, C6, C11, C13 and D3 resulted in a reduced proliferation rate of CRC cells [25,28,35,36,40,43,50,51,53,54]. Additionally, overexpression in vitro experiments showed that increased levels of HOXA6, B2, B7 and D9 resulted in increased proliferation rates indicating the tumour promoting properties of the above gene products [24,26,34,49]. The findings from in vivo studies, which have been conducted for HOXA3, B7, B8, B13, C6 and D9, agreed with the functional role observed in vitro [24,26,28,36,43,56]. Findings for HOXB9 by Hoshino et al. [46] and Zhan et al. [47] report contradictory results, with the former study reporting tumour promoting and the latter showing tumour-suppressive properties in CRC.
Regarding HOX dysregulation and metastatic potential in CRC, nine HOX genes and their proteins have been reported to affect CRC disease progression in vitro [24,28,34,41,[47][48][49][50]52,55]. The knockdown of HOXA1 and HOXB8 resulted in a decreased number of invasive cells [28,50]. Similarly, the overexpression of HOXA6, B2 and D9 led to a significantly increased number of invasive and migratory cells in the intervention group [24,34,49]. On the contrary, overexpression of HOXA5, A10 and D8 resulted in a decreased number of invasive and migratory cells in the intervention group [41,52,55], (Tables 2 and 3). In vivo findings agreed with the in vitro results regarding HOXA10 as mice who were injected with cells overexpressing A10 developed fewer metastases than the control group indicating a protective role of HOXA10 in CRC progression [55,56]. On the other hand, altered expression of HOXB8 and D9 showed the metastasis-promoting effects of these genes in vivo [24,28], ( Table 3). The expression of important EMT markers (E-cadherin and vimentin) known to increase the invasive behaviour of cancer cells facilitating metastasis, was altered between the intervention and control group. Specifically, downregulation of E-cadherin with subsequent upregulation of vimentin was observed in the studies investigating metastasis-enhancer HOX genes (A6, D9, B9) [24,34,48] whereas the opposite pattern was observed for the metastases-suppressors ones (A5, A10, D8) [41,52,55] (Tables 2 and 3). Three studies reported findings regarding the role of HOXB9 in CRC progression; however, the results are contradictory with Zhan et al. showing a metastatic promoting function whereas Huang et al. reported a tumour-suppressive function [46][47][48].

Discussion
In recent years with the change in people's lifestyle and dietary factors, the incidence of CRC has been increasing, especially in the younger population, making this disease a public health burden [57]. Since the recurrence and development of distant metastases are the major causes of cancer-related mortality, it is crucial to investigate and discover new molecular markers that contribute to CRC aggressiveness and which may affect survival. We conducted this systematic review to investigate the clinicopathological and prognostic significance of HOX genes in CRC and determine the impact of their altered expression in CRC disease progression.
The present systematic review indicates that HOX genes become dysregulated in CRC in comparison with normal tissue and are a diverse group of genes, as some may favour disease progression, whereas others act as tumour suppressors in CRC. The combination of clinical and preclinical findings of the studies included revealed that HOXA3, A9, B7, B8, C6, C11 and D9 were found to be upregulated in CRC tissues [24][25][26]36,40,43]. Their high expression was correlated with adverse clinicopathological characteristics of CRC and worse survival outcomes suggesting an oncogenic role which was supported by the in vitro and in vivo experimental observations. On the other hand, HOXB13 and HOXD10 were found to be downregulated in CRC, and preclinical studies indicated a protective role towards disease progression [31,44]. Among the dysregulated HOX genes, most of them favour an oncogenic behaviour promoting disease progression, rather than acting as tumour suppressors. Similar findings with our study were reported by a recent systematic review by Jin et al. on HOX genes in gastric cancer (GC) which demonstrated diversity in the dysregulation profile of HOX genes with most of them acting as potential oncogenes and are associated with worse disease characteristics and worse OS [58].
HOXB9 was the most frequently investigated protein and from our post-hoc metaanalysis, we found that HOXB9 high expression was associated with an increased risk for metastatic disease indicating that it may predispose to worse OS. However, studies (n = 5) reported contradictory findings regarding its prognostic role in CRC with Carbone et al. [42], Hoshino et al. [46] and Huang et al. [48] showing a negative association with survival outcomes whereas Song et al. [38] and Zhan et al. [47] reported a positive one. A post-hoc meta-analysis regarding HOXB9 expression and OS could not be conducted due to the insufficient data provided by the studies. Similarly, experimental findings were also opposing between studies regarding its tumour promoting or suppressive role, highlighting that HOX proteins may also have a dual role in CRC progression depending on which mechanism is activated that regulates their function [59,60]. For instance, Wan et al. identified the acetylation of HOXB9 as an important post-translational modification which caused suppression of transcription of the HOXB9 target gene Jumonji domaincontaining protein 6 (JMJD6), leading to the inhibition of tumour growth and the migration of lung adenocarcinoma cells, in vitro and in vivo [61]. HOXB9 acetylation was also shown by Song et al. to potentially be responsible for the HOXB9 potential protective role in CRC progression [38,61]. Another potential explanation for the contradictory findings between studies could be the different methodological approaches and categorization of high and low HOXB9 expression groups. For instance, Zhan et al. and Huang et al. evaluated using IHC the association of HOXB9 expression levels with OS. Both studies used IHC as a methodological approach; however, the comparison groups categorized based on IHC staining intensity was different between studies.
HOX proteins contribute to a plethora of functionalities and can be regulated by transcriptional expression, regulating micro-RNAs and post-translational modifications that add complexity in understanding their role [62]. Therefore, the exact mechanism of how HOX proteins promote CRC growth, invasion and metastasis has not yet been elucidated. HOX proteins, as conserved developmental proteins, have the ability to control various cellular functions responsible for cell survival and in many cancers seem to participate in cell proliferation [6]. In CRC Liao et al., showed that HOXB7 could accelerate the transition from G1 to S phase in the cell cycle through the activation of the PI3K/AKT and MAPK pathways resulting in the upregulation of cyclin D1 [26]. Additionally, Zhang et al. found that HOXA3 can serve as an apoptosis-suppressor for cancer development through regulation of apoptosis-related factors (Bcl-2 andcaspase 3) and the activation of the EGFR/Ras/Raf/MEK/ERK pathway [36]. CRC progression to invasive and metastatic disease is characterised by the EMT process, which involves the transition of the stationary cancerous epithelial cells into motile mesenchymal ones enabling them to detach and metastasise [5]. HOX proteins have been found to play an essential role in the EMT, promoting cell invasion and migration. In CRC, Liu et al. reported that HOXD9 promoted CRC cell invasion and migration through enhancing EMT by upregulating vimentin while downregulating E-cadherin [24]. This study also showed that HOXD9 might promote cell invasion and migration through the transforming growth factor-beta (TGF-β) pathway, which is an important pathway in the EMT process in CRC [5]. Our post-hoc meta-analysis identified that HOXB9 was significantly associated with the presence of metastatic disease. There is emerging evidence supporting the role of HOXB9 as a promoter of tumour invasion and metastasis by activating the EMT process through important pathways such as the TGF-β1/Smad2/Slug signalling pathway [63].
Angiogenesis plays a vital role in the progression of cancer, and various HOX proteins have been shown to function in promoting the formation of new vessels in solid tumours by upregulating angiogenic genes [6,63]. In CRC, Hoshino et al. showed that overexpression of HOXB9 resulted in upregulation of angiogenic factors such as interleukin 8 (IL8) and vascular endothelial growth factor (VEGF) both in vitro and in vivo [46]. HOXB9 was also found to be of important clinical significance, as patients with high expression levels appeared to respond better to anti-angiogenic therapy with bevacizumab demonstrating longer OS and DFS in comparison with those who had low HOXB9 levels [46]. Interestingly, Carbone et al. reported the same effect of HOXB9 in the expression of angiogenic factors as Hoshino et al.; however, in vivo models showed that HOXB9 positive nude mice showed resistance to treatment with bevacizumab [42]. Either way, both studies demonstrate that HOXB9 could serve as a potential marker for selecting treatment with anti-angiogenic chemotherapeutic drugs. The possible synergistic role of HOX genes modulation with chemotherapy treatment was shown by Yuan et al. for HOXA10, which was found to promote tumour progression in vitro and knockdown resulted in increased sensitivity to 5-Fluorouracil therapy in vitro and in vivo [35].
This systematic review is the first to provide cumulative current evidence regarding the role of HOX genes and their protein products in CRC progression, their clinicopathological and prognostic significance. This study outlines the heterogeneity among studies, as many have only investigated a specific HOX gene out of the 39 in the human genome. Since the use of HOX genes as future biomarkers in CRC has recently started to attract research interest, further studies are warranted on the subject to fully explore the function of each HOX gene. This systematic review showed that few studies had been conducted to date that combine clinical and preclinical data (in vitro and in vivo) to thoroughly investigate the clinicopathological and functional role of a HOX gene in CRC progression. Moreover, studies demonstrated diversity in the study population characteristics included as well as variability in methodological approaches used. For instance, population characteristics varied between studies in terms of neoadjuvant chemotherapy administration. Furthermore, there was no established 15 standardized cut-off point between a high and low expression level and inconsistent criteria were used between studies to investigate the clinicopathological and prognostic role of the HOX gene of interest. It is worth highlighting the lack of power sample size reporting in both clinical and preclinical studies, as well as sample size information in some studies. For reasons which include the diversity between studies and the lack of detailed and robust data a meta-analysis could be conducted only for HOXB9 on specific outcomes.

Conclusions
In conclusion, this review provides for the first time systematic evidence that HOX genes are dysregulated in patients with CRC and their aberrant expression is related to clinicopathological characteristics and survival. Moreover, this systematic review shows that altered expression of HOX genes affects CRC progression in vitro and in vivo. These findings suggest that HOX genes may serve as potential biomarkers in CRC and their differential expression may be a candidate hallmark for survival outcomes [64]. The potential clinical application of the findings of this review is that HOX genes may be considered as future targets for the development of anticancer therapeutic agents. HOXB9 protein overexpression was identified to be associated with the presence of metastatic disease indicating that it may be a critical transcription factor in CRC. Nevertheless, due to the complexity and heterogeneity of the HOX gene family, further well-conducted and even larger-scale or multicentre clinical and preclinical studies with robust methodology are needed to elucidate the role of each gene and especially HOXB9 in CRC thus determining the validity of their role as potential biomarkers or therapeutic targets in CRC.