Colorectal Cancer (CRC) is the third most common cancer globally, and the second leading cause of cancer deaths in the UK 1. The establishment of new surgical procedures, targeted therapy and effective use of chemotherapy over the last decade, has brought about significant improvements in the prognostic outcome in advanced CRC 2-4. The standard treatment procedure includes surgery for resectable tumours and adjuvant chemotherapy for tumour, node and metastasis classification system (TNM) equal or greater than stage II (See Appendix 1 for diagnosis and management of CRC) 2. Although surgery represents the most successful treatment modality for CRC patients, long term outcomes are vastly dependent upon therapeutic efficacy of systemic anti-cancer agents 5,6. The discovery of epidermal growth factor receptor (EGFR) inhibition therapy has expanded the spectrum of therapeutic options available for metastatic CRC (mCRC) 7-10. EGFR is commonly overexpressed in CRC and causes subsequent uninterrupted cellular proliferation 9,10. Cetuximab is an anti-EGFR monoclonal antibody agent that exert its therapeutic effects by inhibiting the extracellular domain of EGFR 9. Several studies have demonstrated the benefit of cetuximab as a monotherapy or in combination with chemotherapy for patients whom tumours show KRAS wild type (wt) status 11-17.
In 2008, the EPOC trial commenced as a rational extension of studies that demonstrate the benefit of cetuximab 18. The trial evaluated the efficacy of cetuximab as an adjunct to chemotherapy and surgery for operable colorectal liver metastases 18. The results of the trial were unexpected, a shorter progression free survival (PFS) were affiliated among patients treated with cetuximab; despite clinical trials demonstrating its use in inoperable disease 18. The patients selected in the EPOC trial were restricted to an all RAS wt subgroup; despite such restriction the negative effects persisted 18. Although cetuximab is commonly used in clinical practice, only 25- 40% of patients with KRAS wt tumours respond to this treatment 19-21. To avoid further repercussions in future trials, these results emphasise the importance of furthering our knowledge in understanding the nature of tumour biology and genomic heterogeneity of CRC 18. Additional biomarkers are needed to help patient selection to avoid prescribing ineffective, expensive, and potentially harmful therapy.
Since its discovery in 1993 22, an abundance of compelling evidence demonstrate that microRNA (miRNA) expression is dysregulated in multiple diseases, including CRC 23-25. MiRNAs are small non-coding RNA molecules that negatively regulate 30-40% of gene expression at the posttranscriptional level of protein synthesis 21,26,27. miRNA-31, a frequent miRNA deregulated in CRC, has both oncogenic and tumour suppressive functional roles ascribed to it in differing contexts 28,29. High expression of miRNA-31 is a prognostic marker that correlates with advanced disease 19,21,30-32. Current papers refer to miRNA-31-3p, mature sequence of miRNA-31, to be strongly associated with PFS in KRAS wt CRC patients treated with anti-EGFR therapy 19,21,33. Furthermore, functional studies illustrate the regulatory role of the other mature sequence of miRNA-31, miRNA-31-5p, in proliferation, survival, apoptosis and migration/invasion 19,34,35. This study sought to explore the role of miR-31-5p in the downstream signalling pathway of EGFR in KRAS wt CRC, with a specific focus on the detriment observed in the new EPOC study.
5.2 Precision oncology- The need for predictive biomarker:
Precision oncology is becoming increasingly favoured in CRC treatment. Targeted therapies mainly include the use of monoclonal antibodies that act on EGFR and vascular endothelial growth factor (VEGF) 36,37. However, the establishment of targeted therapies for personalised medicine has also given rise to large variation in treatment responses. An aspect of medicine that is continually under development in field of cancer is the search for molecular markers that can predict treatment outcome in patient’s receipt of anti-cancer agents. This isn’t, inevitably, an incentive for cost-effective research but a precautionary measure that avoids putting patient’s life in jeopardy; Patients chosen through a broad selection criteria are more susceptible to overtreatment and adverse effects. Thus, predictive markers that are able to differentiate between responders and non-responders are in great demand.
5.3 EGFR signalling pathway
The EGFR is a single chain transmembrane glycoprotein receptor that is a member of the ErbB family, and subfamily member of other receptor tyrosine kinases: ErbB-1, ErbB-2 (HER2), ErbB-3(HER3) and ErbB-4(HER4) 38. The EGFR (ErbB-1 / HER1) comprise a large extracellular ligand-binding region, a spanning transmembrane domain and an intracellular tyrosine-kinase region 9,38. The epidermal growth factor (EGF) was discovered by Stanley C Et al. Overtime, studies would demonstrate how EGF would bind to the cell surface of EGFR to induce its biological effect 39. In addition to EGF, current papers have established the interaction of other binding ligands, seven precisely, that are complimentarily to EGFR 38. Ligand binding to the extracellular domain of EGFR induces dimerization of the receptor to form heterodimers and homodimers. This activates downstream tyrosine kinases, followed by autophosphorylation of specific tyrosine kinase residues that activate multiple signal transduction pathways. RAS-RAF-MEK-ERK MAPK and AKT-PI3K-mTOR are among the main intracellular pathways activated. Activation of these pathways yields various transcription factors that regulate important cellular processes like that of proliferation, survival, and differentiation. Aberrant activation of EGFR (overexpression) and mutation in the RAS-RAF-MEK-ERK MAPK pathway have been associated with poor tumour progression and poor survival in various cancers 9.
5.4 EGFR Overexpression:
The dynamics of this pathway has been investigated in great depths, both experimentally and computationally. Through the use of immunohistochemical staining, overexpression of EGFR has been reported in 25-82% of CRC cases 40,41. In the last decade, following controversial results of trials, the prognostic value of EGFR expression in predicting cancer progression was placed under scrutiny; Numerous studies generated conflicting results showing that overexpression of EGFR is not associated with poorer prognosis. The study conducted by Mckay JA et al found no association between EGFR overexpression and poor tumour differentiation in 134 tumours 42, while the study by Resnick MB et al found 249 EGFR overexpressed tissue samples to be poorly differentiated (P=0.014) 43. In addition, whilst other studies found an association between EGFR overexpression and survival 44, others have not 41. However, most recent papers have settled this controversy with current literature following the notion that there is a link between EGFR positivity and poor tumour differentiation and prognostic outcome 45,46.
5.5 Anti-EGFR Therapy
Cetuximab and Panitumumab are two anti-EGFR monoclonal antibodies used to specifically treat CRC 9. Cetuximab is a chimeric igG1 class antibody that binds to the extracellular domain of EGFR- inhibiting intracellular signal transduction downstream to the receptor 9. Cetuximab was first approved in 2004 to be used in CRC patient’s refractory to irinotecan 9. Panitumumab is a human monoclonal antibody (igG2) that exerts its therapeutic effects similar to that of Cetuximab 9. Panitumumab was first approved in 2006, and was the first antibody to demonstrate the use of KRAS profiling as a predictive marker for anti-EGFR inhibiton.47.
5.6 KRAS status and anti-EGFR therapy response:
Current functional studies have unravelled numerous mechanisms that cause anti-EGFR therapy resistance. Amongst those are mutations in KRAS and BRAF gene in the downstream signalling pathway of EGFR. The first study to provide promising data showing a link between KRAS status and response to anti-EGFR monoclonal antibody (panitumumab) was by Amado RG et al 47. In addition, the CRYSTAL study published by Eric VC et al demonstrated similar results with Cetuximab 48. Since then, numerous studies have validated the significance of the relationship between these two variables. Thus, KRAS mutation testing is recommended in patient selection and stratification for anti-EGFR therapy.
However, studies that demonstrate the benefit EGFR inhibition in KRAS wt CRC have yet to provide a viable explanation for non-responders with the same genetic status 19-21. This paradoxical phenomenon underpins the gap of knowledge yet to be fulfilled in the field of molecular biology. CRC, much like any other cancer, differs in clinical presentation, morphological characteristic, and prognosis conveyed to patients. In addition, the cancer acquired, either hereditary, sporadically or a combination of both is unique to the patient. There has been no historical record of two tumours being alike. This accounts for the vast treatment responses observed among patients in clinical trials.
Over the last decade, there have been a wealth of new discoveries that have linked molecular features of CRC cells to the efficacy of anti-EGFR therapy. This has been made possible through integrating genetic, epigenetics, micro-environmental factors and molecular subtyping among the comprehensive classification of CRC heterogeneity 49. This includes the discovery of miRNAs, and its expression levels have been implicated in various cancers. Can miRNA expression level serve as a potential biomarker candidate for precision oncology, and overcome the shortcomings of just assessing KRAS mutational status for anti-EGFR therapy?
NcRNAs regulate gene expression at transcriptional and post-transcriptional level 50. Two groups of ncRNA, categorised by their nucleotide length as either short or long ncRNA, appear to be involved with epigenetic processes 50. As our knowledge in molecular science continuous to thrive, short and long ncRNA have taken the spotlight for therapeutic exploitation in various cancers. MiRNAs are an abundant class of short ncRNA molecules that are 18-25 nucleotides in length, and play a crucial role in the post-transcriptional phase of protein synthesis 27.
Our understanding of the regulatory and functional role this non-coding RNA molecule possess have been made possible through high-throughput sequencing technologies, and use of computational and bioinformatics predictive methods 51. Most recent miRNA quantification and functional analysis report have revealed how just a handful are involved in cellular proliferation, apoptosis, fat mobilisation, neuronal mapping, governing immune cells and epithelial-mesenchymal transition 24,52. To date, more than 2500 miRNAs have been identified 53. More than 20 studies have confirmed the aberrant miRNA expression in CRC 53. Following further study reports, a total of 160 miRNAs have been found to be dysregulated in CRC 53. The study by Michael et al, was among the first study to provide conclusive data demonstrating how miR-143 and miR-145 is implicated in CRC pathogenesis 54; miR-143 and miR-145 is downregulated in CRC, which frequently precedes APC (Adenomatous polyposis coli) gene mutation, contributing to the development of CRC 54. Furthermore, it’s believed that miR-31 regulates the EGFR signalling pathway and aberrant expression of this miRNA is the equivalence of a mutation at any point of the EGFR RAS-RAF-MEK-ERK MAPK downstream signalling pathway 21. Elevated levels of miR-31 could be indicative of anti-EGFR therapy resistance, and studies have yet to refute this 55.
5.8 MiRNA biogenesis and function:
The functional unit of miRNA, mature miRNA, is generated through a two-step cleavage of primary miRNA 27. The mature miRNA binds to mRNA to form the effector complex: RNA-induced silencing complex (RISC) 27. The distinct gene silencing mechanism employed out of the two mechanisms is determined by the mRNA target site and biological interaction between miRNAs 27. This process of base-pairing induces negative gene regulation by cleavage of target mRNA with subsequent degradation or translation inhibition 27 (See figure 3).
5.9 MiRNA profiling:
There is a general understanding of the regulatory and functional roles of miRNAs in the development of cancer. However, there is limited information on the biological function of individual miRNAs, including miRNA- 31. Expression profiling of miR-31 in CRC has become more expansive. miRNA profiling gives researchers an insight into the expression level and kinetic interactions of miRNAs. This of which allows reflection on their biological output in normal cellular function and disease state 56. MiRNA profiling is performed by three main methods: quantitative reverse transcription PCR, high-throughput RNA sequencing and hybridisation based methods (DNA microarrays). Furthermore, functional studies have manipulated the expression level of miRNAs in CRC cells in vitro, to assess its biological impact on cell growth, apoptosis, metastases and differentiation. Although the mechanism that is involved in the regulation of miRNA expression is unknown, aberrant expression is proving to be clinically relevant in the diagnosis, progression and predictive outcome of treatment in various cancers.
5.10 MicroRNA- 31 and Anti-EGFR therapy response.
The study published by Manceau et al, had looked into miRNAs that could predict response to anti-EGFR therapy in KRAS wt CRC refractory to chemotherapy 21. The study screened the expression level of 1,145 miRNAs in fresh frozen (FF) and formalin-fixed paraffin-embedded (FFPE) samples from training set 1 and 2 consisting of 87 patients in total, and a validation cohort consisting of a total 45 FF and FFPE samples collected prospectively. The primary outcome variable was PFS. The study found only one miRNA, miR-31-3p, to have a significant association with PFS in both the training set FF samples with HRs being 1.8; 95% confidence interval (CI), 1.1–2.9; P = 0.01 and 2.5; 95%, (CI), 1.3–4.5; P = 0.002, respectively. In addition, the study performed a Super-PC analysis combining cox proportional hazard model to find a cut off value score of those at high or low risk of progression. The median PFS in the training set was = high. risk vs low risk: 9 vs 35.3 weeks with HR, 4.10: 95% CL, 1.3-13.2; P = 0.018. The validation cohort showed similar results, with median PFS High.risk vs. low risk: 23 vs 48.8 weeks = HR, 4.1:95% CI 1.1-15.3; P = 0.04. Ultimately, the study found that those with a high expression of miR-31-3p were more susceptible to have a worse PFS when treated with anti-EGFR therapy in contrast to low expressers. Furthermore, the study had transfected 3 CRC cell lines with miR-31-3p mimic and scramble. Expression profile and bioinformatics analysis of the transfected CRC cell lines identified 47 genes downregulated (fold change < 0.7; P < 0.05) and 27 genes significantly upregulated (fold change < 1.3; P < 0.05). The study theorised that the high miR-31-3p expression causing anti-EGFR resistance is due the absence of EGFR when subjected to hypoxia. Furthermore, among the genes downregulated included AMFR and EPB41L4B. The latter is associated with PFS, as EPB41L4B promotes cellular adhesion, migration and induces cellular differentiation 57. However, the study had failed to demonstrate the predictive role of miR-31-3p as there was no control arm consisting of patients without anti-EGFR therapy. In addition, the study had only screened a fraction of miRNAs encoded in the human genome and may have ruled out other miRNAs of interest. Furthermore, the study published by Mlcochova and colleagues proposed the same aim as that of Manceau et al 19. The study found that 9 miRNAs were differentially expressed in responders and non-responders to cetuximab therapy (P ? 0.01) 19. Following further evaluation, miR-31-3p (P<0.001) and miR-31-5p (P<0.001) expression were classed as being strongly associated with time to progression (TTP) in KRAS wt CRC treated with cetuximab, but not panitumumab. In addition, evaluation of the complete cohort of cetuximab patients (N = 69), miR-31-3p (HR, 5.10; 95% CI, 2.52–10.32; P < 0.001) and miR-31-5p (HR, 4.80; 95% CI, 2.50–9.24; P < 0.001) correlated with TTP with high level of significance 19. The median TTP in patients with high and low expression of miR-31-3p and miR-31-5p were 14 vs 44 weeks (P < 0.0001, HR 5.099, 95%CI 2.520 to 10.317) and 16 vs 49 weeks (P < 0.001, adjusted HR 35.051, 95% CI 6.887 to 178.412), respectively. However, the study hypothesised that miR-31-5p is the leading strand, and expression of miR-31-3p is not representative of significant functional consequences. The study had also transfected 3 CRC cell lines with miR-31-5p mimic and scramble, and found 148 genes significantly downregulated (P < 0.01) and 84 genes upregulated (P < 0.01). The genes that were putatively targeted by miR-31-5p were involved in immune system processes, cytokine receptor interaction and chemokine signalling pathway. However, the insufficient sample size (N=93) was not enough to obtain a significant statistical power, hence a larger independent population is needed for further validation. Moreover, the study by Mosakhani et al had also found miR-31 to be differentially expressed in KRAS wt CRC 33. The study performed qRT-PCR to analyse genes implicated by high expression of miR-31, of which revealed lower mRNA levels of ATN1 and SLC26A3 33. Low expression of ATN1 has been reported in chemoradiation-resistant CRC cell lines, but its role in anti-EGFR therapy resistance remains unknown 58. SLC26A3 encodes a chloride ion found in the intestine 59, and could act as a tumour suppressor gene. A post-hoc analysis of the EPOC trial was performed by Pugh et al 55. Patients that were recruited in the EPOC trial were classed into different subgroups depending on their miR-31-3p expression levels. High expressers of miR-31-3p (N=50) treated with cetuximab had a worse PFS compared to the chemotherapy alone arm (median PFS 12.4 versus 26.7 months respectively HR=2.73, 1.17; 6.40, p=0.02). No significant difference was noted among low expressers (N=50) in both treatment arms (median PFS 20.3 versus 18.9 months respectively HR=1.06 0.43; 2.61 p=0.9). Patients with an intermediate expression (N=49) were also analysed and the results closely resembled that of high expressers, hence both subgroups were combined yielding a shorter PFS in these set of patients treated with cetuximab (HR= 2.28 1.27; 4.09, p=0.006). Although the study demonstrates a shorter PFS in those with high miR-31-3p expression treated with cetuximab, it was unable to classify a set of patients that would benefit from cetuximab depending on their miR-31-3p expression. In addition, as there was no notable difference in PFS in low expressers of miR-31-3p treated with cetuximab, further validation is needed to determine the predictive value of this miRNA. However, the biological interactions of miR-31-5p in the EGFR signalling pathway remains unknown. The miRNA is thought to be associated with 25 other genes. The role of these genes is poorly understood, but recently there has been a link between specific miRNA maturation and EGFR kinase activity. Once the biological mechanism is understood from a molecular level, the expression of such microRNAs could be used as a potential biomarker in clinical practice.