Article Text

Download PDFPDF
Predictive markers of radiotherapy-induced rectal cancer regression
  1. J-S Shin1,2,3,
  2. T-G Tut2,3,
  3. V Ho4,5,
  4. C S Lee1,2,3
  1. 1Department of Anatomical Pathology, Liverpool Hospital, University of Western Sydney, Liverpool BC, New South Wales, Australia
  2. 2Cancer Pathology & Cell Biology Laboratory, Ingham Institute for Applied Medical Research, University of Western Sydney, Liverpool BC, New South Wales, Australia
  3. 3Disciplines of Pathology, University of Western Sydney, Liverpool BC, New South Wales, Australia
  4. 4Department of Gastroenterology, Campbelltown Hospital, University of Western Sydney, Liverpool BC, New South Wales, Australia
  5. 5Department of Medicine, School of Medicine, and Molecular Medicine Research Group, University of Western Sydney, Liverpool BC, New South Wales, Australia
  1. Correspondence to Dr Joo-Shik Shin, Department of Anatomical Pathology, South Western Area Pathology Service, Locked Mail Bag 7090, Liverpool BC, NSW 1871, Australia; j.shin{at}


Patients with locally advanced rectal cancer receive preoperative radiotherapy to reduce the probability of recurrence and to possibly improve overall survival. However, this appears dependent on the extent of histological tumour regression seen in the resected bowel, which can be highly variable between individuals. No predictive marker that can stratify patient management in this regard is currently available. Experimental data implicates a variety of factors that are involved in the DNA damage response following radiation injury, tumour tissue oxygenation, autoimmune antitumour response triggered by radiotherapy and in the pathogenesis of colorectal cancer, as potential indicators of radiation sensitivity. These details are presented in this review, which may serve as targets for clinical validation studies aiming to find predictors of radiotherapy response in rectal cancer.


Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Preoperative radiotherapy is often used in the curative treatment of rectal cancer, but the variable extent of histological tumour regression seen in the resected bowel and its influence on the local disease control and possibly overall survival, demands a judicious use of this otherwise toxic treatment. However, clinically applicable markers that are predictive of rectal cancer regression remain elusive. This review will aim to detail the current understanding of molecular interactions in tumour cell response to radiation injury and in rectal cancer pathogenesis, which may serve as targets for predictive assessment.

Preoperative radiotherapy for rectal cancer

Colorectal cancer (CRC) is typically an adenocarcinoma without differentiating morphological features between the large bowel segments. Clinically, however, an important distinction is made between colonic and rectal cancers. Representing about one-third of CRCs,1 ,2 the latter tumours pose additional challenges in complete surgical clearance, with the lack of an enveloping free serosa distally and with close proximity to the anus and other organs within the confined pelvis. This is reflected by poor improvement in the survival outcomes of rectal cancer patients, in comparison with their colonic counterparts.1 Thus, preoperative downstaging of the primary tumour, reflected by its histological tumour regression in the resected bowel, is often sought through radiotherapy with or without chemotherapy. This additional treatment imposes risks to the patient that range from relatively innocuous nausea and diarrhoea to acute urinary retention, thromboembolic disease, lumbosacral plexopathy,3 bowel incontinence4 and sexual dysfunction5 that can diminish the quality of life.6

While studies have shown a positive impact of preoperative radiotherapy on the local disease control of rectal cancer, it is less clear whether this also improves overall patient survival.711 As preoperative radiotherapy is aimed at the primary tumour, it has no direct tumouricidal effects on the potentially involved regional lymph nodes and other metastatic sites which further govern prognosis.11 Furthermore, tumour regression following radiotherapy is inconsistent between individuals, even with similarly sized tumours and the same treatment schedule.

Histological tumour regression

The foregoing is implicitly acknowledged in the current American Joint Committee on Cancer recommendation to histologically grade the degree of tumour loss in the subsequently resected bowel.12 Their 4-tier ‘tumour regression grade (TRG)’, ranging from ‘0’ for complete response with no residual tumour cells found, to ‘4’ with minimal or no evidence of tumour death, follows in the same vein as several other reported grading systems.1318 These have various names, including the original ‘TRG’, a 5-tier grading system initially described in oesophageal cancers by Mandard et al18 and subsequently adapted for rectal malignancies.19 But the term TRG is also used loosely in reference to a number of other grading criteria in the literature. Potential for confusion for both the pathologists and the clinicians is further enhanced by the fact that some of these criteria disparately assign either progressively worse or better regressed tumours with increasing grade.17 ,18 Although the premise of gauging residual tumour volume relative to areas of tumour-free fibrosis within the mass lesion appears simple, there can be poor reproducibility in this semiquantitative process.20 Histological distinction between fibrosis induced by invasive tumour versus that from radiotherapy induced regression is not always clear. Additionally, there are inconsistencies in the volume of macroscopic tumour sampled for histological assessment among the pathologists and institutions and in the time elapsed between the end of radiotherapy and subsequent surgery. If this period is shortened, either for logistic reasons such as operating theatre/surgeon availability or, perhaps, as part of the treatment protocol, then it may be insufficient for maximum tumour regression to take effect. We have previously shown that, as expected, tumour regression was greater when assessed at 4–6 weeks versus 1 week following chemoradiotherapy for locally advanced rectal cancer.21

These concerns may be alleviated by dichotomising the grades of tumour regression into good and poor responders, as often seen in studies attempting to correlate with clinical outcomes.21 ,22 And in spite of the above reservations, some cohorts have, nonetheless, shown a relationship between histological tumour regression grading and patient survival.2326

Tumour cell response to radiation injury

Hypoxia and HIF-1α related radiation resistance

The molecular basis of tumour regression is cell death from DNA damage, elicited by oxygen-derived free radicals generated from ionising radiation exposure. The process is hence dependent on the adequate oxygenation of target tissue, with hypoxia and the related expression of hypoxia-inducible factor-1α (HIF-1α) serving as negating factors in the effectiveness of radiotherapy.27 HIF-1 is the major transcription factor mediating cell response to hypoxia. In an adequately oxygenated environment, its α subunit is degraded by interaction with the von Hippel Lindau complex. This is compromised in hypoxic conditions, leading to the accumulation and nuclear localisation of HIF-1α, with subsequent activation of genes producing growth and antiapoptotic effects, contributing to tumour cell survival following radiation.28 Overexpression of HIF-1α also appears related to phosphatidylinositol-3-kinase (PI3K), a family of lipid kinases constitutively activated in many tumours, with downstream effectors including Akt and mammalian target of rapamycin (mTOR) complex 1 (mTORC1). The net effect of PI3K pathway activation is to provide a tumourigenic proliferative stimulus, with the transcription of multiple genes including myc and cyclin D1.29 Additionally, repression of the PI3K pathway in human tumour cells have shown to downregulate the expression of HIF-1α,30 ,31 suggesting a role for the PI3K/Akt/mTORC1 and other factors upstream and downstream of this signalling as potential markers of radiation response.

To circumvent the issue of hypoxia-related treatment resistance, radiation is administered in fractionated doses to initially kill the better oxygenated areas of tumour, leaving for previously hypoxic cancer cells to reoxygenate with reduction in the total tumour volume. The intervals between radiation also allows the exposed normal cells to recover from injury, while cancerous cells with assumed defects in DNA damage repair are instead destroyed. The other important factor to consider is the ability of tumour to regenerate after radiotherapy, which is dependent on the resistance and survival of the cancer stem cell subpopulation.27

However, the critical step in the cellular response to radiation damage is the ability of the cancer cells to repair the radiotherapy-inflicted DNA damage, mainly the severance of both strands of the DNA helix (DNA double strand breaks (DSB)), which determines their sensitivity to treatment.

DNA damage response following radiation injury

DNA damage response (DDR) broadly refers to the complex molecular pathways that aim to detect and repair DNA damage from a variety of causes, including cell metabolism, replication errors and ionising radiation.32 The involved proteins are subgrouped into sensors, transducers and effectors, reflecting their respective roles in relaying the intracellular damage signal to the key outcomes of cell cycle arrest, DNA repair and, should the injury be irreparable, apoptosis.33 Following radiation-induced DNA DSBs, the sensor complex of meiotic recombination 11 (MRE11), a trimer of MRE11, RAD50 and NBS1 (also known as XRS2) molecules bind to the site of DNA damage and potentiate the signalling for cell cycle arrest by generating single-stranded DNA regions with its nuclease activity.34 This recruits the central transducer PI3K-like (PIKK) protein ataxia telangiectasia mutated (ATM, also known as TEL1),35 ,36 which amplifies the checkpoint signalling and cell cycle arrest34 via controlling the phosphorylation of a number of effector proteins, including checkpoint kinase 1 and 2 (CHK1 and CHK2, respectively), as well as p53 which also influences apoptosis.32

DNA DSBs can then be resolved by either homologous recombination that requires a DNA template from the sister chromatid or another chromosome, or non-homologous end joining (NHEJ). For radiation-induced and other non-cell replication-associated DSBs, the exposed DNA ends are directly ligated through NHEJ.36 In this process, the Ku70/80 heterodimeric proteins recognise and bind to DSBs, attracting and activating the DNA-dependent protein kinase catalytic subunit. Follow-up recruitment of other molecules, including DNA ligase IV that processes and ultimately rejoins the DNA ends, ensues.36 ,37 Should the repair be unsuccessful, the DDR is directed towards p53 mediated apoptosis. Counteracting this signalling in the interim and, thereby, encouraging cell survival is nuclear factor-κB (NF-κB).38 ,39

Described above is an orderly but a much simplified version of the DDR process following radiation injury. There are a variety of further molecules and interactions cross-linking the three effector pathways of cell cycle arrest, apoptosis and DNA repair, as well as between the sensor, transducer and effector roles of these proteins, such as the additional role of MRE11 complex in NHEJ.40 Regardless, the critical molecules that have been highlighted here represent those that may reflect the cancer cells’ ability to overcome radiation-induced DNA damage, hence serving as potential markers of treatment response.36

Predictive markers of sensitivity to radiotherapy

However, proving the clinical value of these proteins as predictors of radiotherapy response remains difficult. For instance, absent or reduced expression of the MRE11 complex, which is the sensor of DNA damage that initiates the downstream signalling in DDR, should in theory reflect heightened radiosensitivity. This has been previously observed in cell studies, where skin fibroblasts with truncating mutation in the MRE11 gene were unusually sensitive to radiation.41 However, more than a decade on from this finding, the clinical applicability of MRE11, especially in rectal cancer that has undergone radiotherapy, remains unproven. The few recent translational studies exploring MRE11 as a predictive marker have, instead, examined this in other tumours, such as muscle invasive bladder and breast cancers, where the unexpected reverse trend relating reduced MRE11 immunohistochemistry (IHC) staining with poorer response to radiotherapy was seen.42 ,43

The central role of MRE11 in DDR is shared by ATM, which triggers various key effector pathways involving critical factors, such as p53, NF-κB and Ku70/80. Found mutated with compromised function in patients with ataxia-telangiectasia, a progressive neurodegenerative condition with predisposition to lymphomas and radiosensitivity,44 ATM loss in tumours should also presumably relate to greater radiotherapy response. However, while cells, including that of a CRC cell line,45 with mutated/lost ATM function are sensitive to radiation in vitro46 and drug induction of radiosensitivity relates to ATM suppression,47 only sporadic translational studies48 have tested its applicability to clinical samples and outcome in CRC, especially in the predictive assessment of radiotherapy response.

Further downstream of the DDR, other molecules that have also shown some but as yet inconclusive evidence as a predictor marker include the DNA DSB repair related Ku 8049 and apoptosis-related bcl-250 and survivin.51

Beyond the confines of DNA DDR, other specific targets for predictive assessment in rectal cancer that has undergone radiotherapy relate to additional/alternative theories of radiation sensitivity. For instance, tissue oxygenation that facilitates the DNA damaging effect of radiotherapy could be reflected by the vascular endothelial growth factor (VEGF) expression in tumour. Its loss correlated with complete radiotherapy response in a cohort of 62 rectal cancer patients,50 and anti-VEGF therapy may normalise the tumour vascularity, providing a period of improved tumour oxygenation that enhances radiation sensitivity.52

Another possibility is based on the notion that tumour death via radiotherapy may also occur through triggering an immune response against cancer cells.52 For example, induction of autoantibodies against tumour-specific antigens were seen in the sera of patients with prostate cancer following radiotherapy.53 In rectal cancer, the extent of tumour-infiltrating lymphocytes may indicate those tumours that are already primed for immunological response by the body, and hence may be a clue to radiation sensitivity.54

So far, none of the above or others are currently applied to clinical practice, with evidence for each potential predictive marker seen only in small volume and contradicting between reports. This is highlighted by Kuremsky et al,55 who identified 36 potential biomarkers of rectal cancer chemoradiotherapy response following a review of 1204 published articles. As many of the markers were reported in small single studies, the authors of the review focused their attention on gene products with more than five studies in the literature. Six biomarkers met their criteria: p53, epidermal growth factor receptor (EGFR), thymidylate synthase, ki-67, p21 and bax/bcl-2. The most frequently studied of these was unsurprisingly p53, but even for this biomarker the results of the included reports were inconsistent. Both bcl-2 and ki-67 were similarly deemed non-useful in the clinical decision process based on their review.

Part of the problem may be the paucity of prospective studies correlating various biomarkers with morphological and clinical outcomes. Investigations that have aimed to characterise the genetic profile of radiosensitive tumours have also shown inconsistent results in rectal cancer,5658 though larger comprehensive ongoing efforts using this approach may provide better answers for the predictive assessment of tumour response to radiotherapy.59

Linking established molecular characteristics of CRC with sensitivity to radiotherapy

Of all the known molecular traits of CRC, only a few sets of markers are currently used in their pathological assessment for prognostication and prediction to adjuvant therapy. Of these, the most established is the set of DNA mismatch repair (MMR) proteins MLH-1, MSH-2, MSH-6 and PMS-2, the lack of which defines the subset of CRCs with microsatellite instability (MSI).

Representing approximately 15% of sporadic cases and associated with better survival,60 MSI tumours are more likely to be proximal, poorly or undifferentiated with immune response seen as either tumour-infiltrating lymphocytes and/or Crohn's disease-like lymphoid aggregates, with greater tendency to be mucinous, signet ring or medullary histological subtypes.61 ,62 They are usually sporadic with promotor methylation of the MLH-1 gene, but can also bear germline mutation of one of the MMR protein genes in the Lynch syndrome,63 with autosomal dominant expression of multiple early onset primary tumours, including colorectal, endometrial, gastric and ovarian types.61 These CRCs have diploid or near-diploid cell chromosomes61 and are thought to evolve through a ‘caretaker pathway’. Instead of the conventional pathogenetic progression from traditional adenomas with mutations or loss of key genes such as adenomatous polyposis coli affecting the nuclear localisation of β-catenin and KRAS in the EGFR pathway, an alternative set of molecular changes involving genes such as WNT in β-catenin signalling and BRAF downstream of KRAS are seen.64

MSI and radiotherapy response

MSI-related tumours also show discordant lengths of microsatellites, loci of tandemly repeated DNA motifs through the genome, in comparison with normal cells.63 This results from the accumulation of nucleotide base-pairing errors during cell divisions, as there is a lack of DNA MMR proteins and hence, the ability to detect and repair single-base mismatches, insertions or deletions is compromised.65 Although this is not the type of DNA damage induced by radiation, there are, nonetheless, molecular findings supporting the view that DNA MMR proteins exert influence at multiple levels of the DDR following radiation injury.

On a broad genomic scale, DNA MMR deficiency presumably leads to greater instability with new mutations which may involve the key factors of DDR. Reports of increased mutation rates of proteins including ATM66 and MRE11,66 ,67 as well as DNA-PKcs68 involved in NHEJ have been documented in tumour cells with MSI.

Furthermore, there may be direct effects of MMR proteins at multiple stages in the DDR postradiation that lead to cell cycle arrest, subsequent DSB repair and/or apoptosis.69 MLH-1 and MSH-2 may contribute to the initial recognition of radiation-induced DNA DSBs,70 ,71 and the inability to mobilise MRE11 and increased death have been observed in MSH-2 negative irradiated cells.72 Subsequent efficient G2/M transition arrest also appears to require MLH-1 and MSH-2.70 ,72 Observations in human fibroblasts suggest the interaction of MLH-1 and PMS-2 with p53, which may guide the direction of p53-mediated signalling between cell cycle arrest and apoptosis.71 MLH-1 loss also correlates with the inactivation of NF-κB,38 an antiapoptotic effector in the DDR cascade after radiation injury.39 MMR proteins may also ensure the accuracy of final DNA repair by checking for correct base pairing of the amended DSBs during the error prone NHEJ.70

Unfortunately, the scant number of clinical studies that have attempted to replicate this theory have failed to show a relationship between the MSI status and radiotherapy response in rectal cancers, including a recent study of a larger cohort of 316 locally advanced rectal cancers, 25 (7.9%) of which were MSI-related tumours.73 ,74 Despite the inability of the aforementioned studies to confirm the relationship between the MSI status and radiotherapy response in rectal cancers, IHC evaluation of the DNA MMR proteins is increasingly used for routine prognostication of CRCs75 and is comparable in reliability for MSI detection to the assessment of microsatellite lengths via PCR amplification.76 In suspected cases of Lynch syndrome, it is also able to direct the subsequent mutation analysis to the specific protein.77


Rectal cancer treatment frequently uses preoperative radiotherapy for tumour regression/downstaging, with variable success between individuals. Predictive assessment for patient stratification is currently lacking. This is despite what has been detailed of the molecular pathways involved in the tumour cells’ ability to overcome radiation injury and the potential markers that have been highlighted in the process, as presented in this review and summarised in figure 1. While cell-based studies have shown correlation between a number of such factors and radiosensitivity, these findings are yet to be convincingly replicated in clinical cohorts of rectal cancers. Although their MSI status is promising in this regard, and also tempting in that it already has a proven value in prognostication and can be routinely assessed, this again has failed to show a strong relationship with morphological tumour response in treated rectal cancers. However, efforts are currently underway to amass large tissue banks and data banks to characterise radiotherapy response in both normal and tumorous tissues,59 not to mention many researchers continuing to focus on the predictive assessment of rectal and other cancers. The latter should emphasise on translating the current body of experimental data to clinical studies, where their applicability to patient management can be proven.

Figure 1

Potential markers of radiotherapy response in rectal cancer. Within the depicted framework of main processes involved in cellular response to ionising radiation, examples of factors with possible relationship to treatment sensitivity/resistance are highlighted in red. DSB, double strand break; MSI, microsatellite instability; TILs, tumour infiltrating lymphocytes; MMR, mismatch repair; VEGF, vascular endothelial growth factor; HIF, hypoxia-inducible factor; PI3K, phosphatidylinositol-3-kinase; mTOR, mammalian target of rapamycin; ATM, ataxia telangiectasia mutated; MRE, meiotic recombination.

Take home messages

  • Preoperative radiotherapy is used for local disease control in rectal cancer management.

  • Radiotherapy response in rectal cancer can be histologically assessed by tumour regression grading, which can be highly variable between individuals following the same treatment.

  • Predictive markers of better radiotherapy response is needed for patient stratification.

  • A number of factors involved in tumour oxygenation, DNA damage response, colorectal cancer pathogenesis and immunological anti-tumour response may serve as potential predictive biomarkers of radiotherapy response in rectal cancer.



  • Contributors First author J-SS is responsible for the overall content of the manuscript as a guarantor. J-SS and CSL conceptualised the topic and overall direction of the manuscript. T-GT and VH contributed to the review of previously published markers of radiotherapy response in rectal cancer. J-SS wrote the initial draft, which was edited by the remaining coauthors.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.