Received Date: April 18, 2012; Accepted Date: May 28, 2012; Published Date: June 02, 2012
Citation: Calvo Felipe A, Claudio S, Emilio A, Ferrer C, Ochoa E (2012) Prognostic and Predictive Biomolecular Markers in Rectal Cancer. J Nucl Med Radiat Ther S2:006. doi:10.4172/2155-9619.S2-006
Copyright: © 2012 Calvo Felipe A, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Visit for more related articles at Journal of Nuclear Medicine & Radiation Therapy
Prognostic and predictive; Markers; Rectal cancer
In an era in which advances in multimodality treatment strategies have contributed significantly to the improvement of outcome in rectal cancer patients [1-6], treatment decisions are mainly based on tumornode- metastasis (TNM) classification and the circumferential resection margin (CRM) at time of diagnosis (Figure 1). Nonetheless each group requires a different therapeutic approach based on the risk of disease metastasis [2,7]. It is clinically relevant to identify group-specific prognostic markers (markers that provide information on the natural history of the disease and/or have an association with clinical outcome, typically a time-to-event outcome) and predictive markers (markers that are used as indicators of the likely benefit of a specific treatment) besides the TNM staging might simplify the treatment decision process [8,9]. Prior identification of patients who have a higher likelihood of responding to preoperative radio-chemotherapy could help to select those who can benefit from the treatment. Furthermore, the current and future developments of these markers and the issues related to their clinical implementation are discussed. Total mesorectal excision (TME) technique adopted in the mid-90s caused local recurrence rates in rectal cancer patients to drop from 20 to around 10% [4,10,11]. The addition of preoperative radiotherapy (RT) and/or chemotherapy has a beneficial effect on local control and, in some cases, sphincter preservation [5,7,12-15]. A survival benefit of the addition of chemotherapy in the preoperative setting remains unproven; many expert guidelines have implemented preoperative (chemo) radiation as standard of care [1,6,15]. Unfortunately, surgery, preoperative RT, and chemotherapy are associated with significant acute and long-term toxicity and occasionally mortality [16-20]. The peculiar aspects of this approach are related to clinical over staging, which may result in an unnecessary neoadjuvant treatment in almost one-fifth of patients . In the context of a decrease in local recurrences, distant metastases now predominate, with incidences around 30% for locally advanced rectal cancers . Therefore, optimal treatment of rectal cancer is nowadays aimed at prevention of both local and distant tumour recurrence as well as minimizing the risk of side effects due to overtreatment. These strategies will be reviewed along with the predictive and prognostic markers used to determine the optimal treatment strategy.
Organ-saving procedures, intensification of treatment, and the awareness of negative side effects of treatment, call for identification of prognostic and predictive markers. Now a days local recurrence in the quality surgery scenario of practice describes  a local recurrence rate is of approximately 6% to 10% for most rectal cancer patients. The number needed to treat of preoperative RT to prevent one local recurrence (LR) varies between 10 and 18 patients. A subgroup of patients has an LR risk of 4%, and limited treatment in those patients seems justified. The LR risks for more advanced tumours vary between 8% and 20%, depending on T-stage, N-stage, and CRM involvement. Prediction of these tumour characteristics will help to identify specific patient groups and select them for appropriate treatment. The current developments for prognostic and predictive markers in rectal cancer are analyzed.
Prognostic markers are elements that predict the risk of developing a local or distant recurrence. In rectal cancer, prognostic evaluation comprises preoperative markers that predict tumour extent and nodal involvement, and determine treatment of the primary tumour. Prognostic markers in rectal cancer are based on imaging and histopathology and increasingly in molecular markers. Each type of marker will be discussed in this article.
Distance of the tumour to the mesorectal fascia has been shown to be a more important factor than T-stage, for LR because T-stage does not discriminate tumours with a high chance of CRM involvement. MRI has been the only imaging modality that has been found useful for the prediction of the CRM. A meta-analysis of several single- centre studies demonstrated that the sensitivity varies between 60% and 88%, with specificity varying between 73% and 100% . Endoscopic ultrasound (EUS) is useful in the diagnostic process in early rectal cancer. Its sensitivity for detecting T1 and T2 lesions is reported to be between 69% and 97%, but the limitations lie within distinguishing between T2 and T3 lesions [24-26]. An additional advantage of EUS is the ability of EUS-guided fine-needle aspirations that can be carried out directly on lesions or suspicious-looking lymph nodes. The preoperative availability of tumour material will facilitate the incorporation of biomarkers into the process of prognostic stratification and prediction. For lymph node involvement prediction, no significant difference in performance between the different imaging modalities (EUS, computed tomography [CT] scan, or MRI) has been demonstrated . Sensitivity varied between 55% and 67% with specificities of 74% to 78%. The use of additional morphological MRI criteria, such as extra capsular extension of disease and signal heterogeneity, could improve the detection of lymph node metastases. Sensitivity increases to 85% and specificity to 97% for nodal metastases in nodes ≥ 3 mm . Specific MRI contrast agents also seems promising, but has not found its way to routine clinical practice yet [29,30].
Local tumour extent is traditionally classified through the pathologic T-staging system; Nodal involvement and CRM involvement also can be reliably determined by histopathology. Lymphovascular invasion, extramural venous invasion, serosal involvement, and poor differentiation are other factors related to an increased local recurrence risk . The depth of extramural disease and the presence of extramural venous invasion have strong associations with nodal disease . Histopathological evaluation can also be used to accurately determine the extent of response to CRT. Several studies demonstrated favourable long-term outcome for patients with a pCR. A combined analysis of 14 datasets, including 3105 locally advanced rectal cancer patients treated with preoperative CRT and TME surgery, 484 patients had ypCR (complete pathologic response after CRT). 5-year risk for local recurrence was 2.8% in the pCR group and 9.7% in the no pCR group. The 5-year distant metastasis-free survival was 89% and 75%, respectively [33,34].
Until now, not many specific prognostic molecular markers for rectal cancer patients have been identified. Data on prognostic molecular markers come from studies including both colon and rectal cancer (CRC) patients. Within these CRC series, the predictive and prognostic roles of several molecular markers involved in proliferation, apoptosis, neoangiogenesis, DNA mismatch repair and 5-fluorouracil metabolism have been investigated [35-38]. In addition, several studies describe prognostic genomic markers that represent DNA mutations or other DNA aberrations that comprise a larger genomic area. A recent review of molecular markers in CRC concluded that genomic instability markers, such as loss of heterozygosity at chromosome 18q and microsatellite instability as well as the B-Raf V600E mutation, are valuable prognostic markers for outcome . A pooled analysis of 27 studies, including 2189 CRC patients, demonstrated a significantly worse OS in patients with either allelic imbalance at 18q or loss of DCC expression, irrespective of adjuvant treatment or study method . Twenty-four of the twenty-seven studies, included both colon as well as rectal cancer cases. Rectal cancer cases specific results were not presented separately in any of the included studies, and validation of any of these particular markers in large series of rectal cancer patients only is lacking. A methylation profiling study  in a series of early rectal cancer patients identified 5 new markers (ESR1, CDH13, CHFR, APC, and RARB) that showed significantly higher tumour DNA methylation in T1-T2N0 patients compared with stage IV patients. PIK3CA mutations have been found in 10% to 30% of colorectal cancers [42-47]. Because there is a close interaction between RAS and PI3Ks, [48-50] activation of the PI3K/AKT signalling pathway via PIK3CA mutations, or the RAS-RAFMAPK (mitogen-activated protein kinase) signalling, is considered to be one of the most common mechanisms involved in colorectal carcinogenesis [43,51]. In accordance with this, mutations in KRAS gene and BRAF gene are commonly observed in colorectal cancers with frequencies of 30% to 40% and 5% to 22%, respectively [52-56]. A large population-based study in colon cancer suggested that activation of the PI3K/AKT or the RAS-RAF-MAPK pathway by mutation of at least 1 of the 3 genes predicted poor clinical outcome, but the effect of mutations in PIK3CA alone was not addressed . In several cohort studies, PIK3CA mutations were associated with poor prognosis among patients with resectable stage I-III colon cancer [54,56]. In a large cohort of rectal cancer patients it has been determined a prognostic effect of PIK3CA, although different from what has been established in CRC, because PIK3CA mutations were predictive for LR, and not DM, in this series. The PIK3CA mutation was present in 7.9% of the patients and indicative of increased local recurrence risk at 5 years-27.8% for PIK3CA -mutated patients versus 9.4% in the PIK3CA wild-type population . The most important reason for the lack of studies of prognostic molecular markers in rectal cancer patients is the limited availability of suitable tumour material. The tumour tissue obtained after preoperative treatment at surgery is often highly fibrotic or contains little viable tumour cells for further processing in either immunohistochemical- or DNA-based studies. Even more problems are encountered when the tissue needs to be processed for RNA extraction. Despite these limitations, several studies have determined the value of gene expression profiles in rectal cancer patients. The majority of these studies were able to demonstrate a variety of gene expression alterations during colorectal carcinogenesis when comparing expression profiles of normal colorectal tissue with adenoma or carcinoma [58-61]. A few small studies were able to identify gene expression profiles that were correlated to lymph node status and could be used as prognostic markers. A study comparing 6 stage II with 6 stage III tumours, reported 77 genes identified through gene expression analysis that were associated with lymph node metastasis of colorectal carcinomas . A second study comparing 10 stage II tumours with 8 stage III tumours (both colon and rectum) was able to identify a profile of 9 genes only that were differentially expressed between these node-positive and node-negative patients . A third study comparing 41 stage I and II and 25 stage III tumours found a gene expression profile for lymphatic metastasis with a specificity of 76% to 83% and a sensitivity of 38% to 48%. All 3 studies show that gene expression profiling may predict nodal involvement in tumour growth. However, the small number of samples and the mixture of rectum and colon tumours indicate that these results should be interpreted with care if extrapolated to rectal cancer patients only. Especially, because a gene expression profiles study comparing cecum cancer with (recto) sigmoid carcinoma has demonstrated a difference in gene expression between left- and right-sided carcinomas as well as between left- and right-sided normal tissue . Furthermore, a recent Japanese study confirmed the differences in gene expression between colon and sigmoid/rectum cancer in 89 tumour samples, of which 30% were stage III and 21% stage IV tumours. In addition, by stratifying for the anatomical location of the tumour, they established a gene expression profile for lymphatic metastases with accuracies of 100% for colon and 95.8% for recto sigmoid carcinomas . Validation experiments of the gene expression signatures predictive of nodal involvement, therefore, need to focus on rectal cancer only before these signatures become of true clinical value. Another set of molecular markers focuses on the underlying biology of the process of tumorigenesis, as described by Hanahan and Weinberg . They hypothesize that 6 biological capabilities are acquired during the development of human tumours. Of these, the ability to evade apoptosis and several immune systems– related parameters have been studied and shown to be of prognostic relevance in rectal cancer [67-69]. Especially, apoptosis, defined by the activity of caspase-3, was highly prognostic for the development of LR. There are several promising molecular markers in rectal cancer, but the lack of validation studies hampers the clinical implementation.
Comparable with prognostic markers, predictive factors can be derived from histopathological studies, imaging modalities, or molecular markers. A review concluded that conventional histopathologically based indices, such as tumour stage and grade, were not sufficiently capable of predicting response to CRT . Treatment response prediction with imaging techniques shows more promising results . However, staging accuracy is reduced after preoperative treatment compared with untreated cases. This might be due to post radiation oedema, inflammation, fibrosis, or necrosis. FDG-PET staging and restaging in rectal cancer patients treated with preoperative CRT, showed that initial SUVmax has a prognostic value related to long-term patient outcome (Figure 2) [72,73]. Post treatment staging of twenty-three T3/4 rectal cancer patients undergoing preoperative CRT demonstrated a sensitivity for identifying response of 100% for fluorodeoxyglucose positron emission tomography (FDG-PET), compared with 71% for MRI and 54% for CT scan, whereas the specificity was 60% as opposed to 80% with CT and 67% with MRI . The MERCURY trial (Magnetic Resonance Imaging in Rectal Cancer European Equivalence Study) evaluated the prognostic relevance of MRI after neoadjuvant treatment. MRI assessed tumour regression grade (mrTRG) after preoperative therapy predicted significantly better OS and DFS in responders compared with non-responders. The results for T staging were consistent with those reported in other series, with an overall accuracy of MRI-determined T-stage of 79% after CRT [75,76]. From the MERCURY data, it appeared that degree of tumour replacement by fibrosis correlated better with survival than T-stage after treatment . Combination of modalities may help to provide valuable information before surgery. A good example is the combination of preoperative FDG-PET CT combined with diffusion-weighted MRI to predict responses after CRT. In twenty-two rectal cancer patients, a specificity of 94% and a sensitivity of 100% were recently reported for the prediction of pCR when the results from the pre-treatment diffusion-weighted MRI were combined with the results of the FDGPET CT before and during CRT . Although this study needs confirmation in a larger patient group in a multicenter setting, the potential of this approach of combining imaging techniques is promising. Molecular biomarkers that can predict treatment response are mainly being studied in the high- and intermediate-risk rectal cancer patients, but they have not been yet tested in prospective studies. In a review of 1204 articles, it has been identified 36 putative biomarkers . Because many markers were only reported in small single studies, they chose to focus on gene products with more than 5 studies in the literature. Six biomarkers met these criteria: p53, EGFR, thymidylate synthase (TYMS), ki-67, p21, and bax/bcl2. The most frequently studied biomarker was p53 (Table 1). Because the results of these studies were contradicting, the authors concluded that p53 is unlikely to serve as a good predictor of response to CRT. Similarly, for both bcl2 and ki-67, the authors concluded that their usefulness in the clinical decision process is unlikely, based on the results in the literature reviewed. The expression of EGFR (Table 2) determined as either present or not present was not found to be of sufficient predictive value in this review. However, the authors acknowledged that this marker might be a promising predictor of response when expression is studied more quantitatively than just as presence or absence of expression. Other study showed that a single nucleotide polymorphism: G to T at position 216 in the promoter region of EGFR resulted in variation in the level of expression . Furthermore, in the GG homozygote patients, only 33% were major responders (TRG 1-2) to CRT, whereas in the GT or TT population, this number was significantly higher at 64%. With respect to TYMS (Table 3), 4 of the 5 studies reporting on TYMS described significant associations of expression with outcome . However, this evidence was not sufficient to support TYMS protein quantification as a predictor of outcome because the results of the studies were contradictory, the study populations were generally small, and differences in treatment schedules might have biased results. A recent study on tissue microarrays of 38 rectal cancer patients treated with CRT found ki-67 to be the only predictive marker for treatment response in multivariate analyses. These conflicting results demonstrate once again that the value of a single marker approach is probably limited in the prediction of treatment response. Gene expression profiling may be more promising, but has only been performed in a few clinical studies in rectal cancer, with the main focus on response to CRT. In a study of 30 biopsies of locally advanced rectal cancers treated with CRT, a gene expression profile was established that correctly predicted response in 25 of the 30 cases when T-down staging was taken as response definition . When an adapted tumour regression grade system was used as response definition, no profile could be found. However, a recent update demonstrated that after a median follow- up of 59 months, 7 of 8 patients classified as non-responders developed a recurrence compared with only 1 of the patients classified as responder . A Japanese group used biopsies from 52 rectal cancer patients treated with preoperative RT, who where either classified as responder or non-responder . They divided their patient population into a training set, containing 35 samples, based on which they built their signature and a testing set, containing 17 samples, which they used for validation. The predictive signature that resulted from the statistical analysis in the training set contained 33 genes of which the expression differed significantly between responders and nonresponders. The signature was eventually able to correctly classify 14 of the 17 validation cases as either responders or non-responders. A German group based their signature of 43 genes on 42 patients in whom pre-treatment biopsies were available. Their signature could effectively identify a responder to treatment in 71% of the cases and a non-responder in 86% . Although these results seem promising, all 3 signatures were determined on small sample sizes, did not show any overlap in genes involved, and neither 1 of them was prospectively validated in an independent cohort. A new development in the field of predictive markers in rectal cancer has been the use of stem cell markers to predict response to CRT, in particular CD133. In colon cancer, CD133-positive cells have been shown to be a prognostic marker of poor survival and over expression on the tumour membrane was related to chemotherapy resistance in recurrent and stage IV tumours [86-87]. Recently, several studies investigated the predictive potential of the marker in rectal cancer patients treated with CRT. The frequency of CD133 staining in CRT specimens was significantly higher than that of non-CRT specimens . Two Japanese groups [89,90] were able to show that the presence of CD133-positive cells in the resection specimen of rectal cancer patients after CRT treatment was associated with distant recurrence and poor DFS. In conclusion, stem cell markers, such as CD133, have not been investigated widely in rectal cancer but could harbour potential as predictive biomarkers in rectal cancer. The addition of biologically active targeted therapies to CRT has made the development of predictive markers inevitable. The high level of patient variation in response to these therapies and the potentially toxic side effects speak strongly to the need for better patient selection. Fortunately, some studies have been already been performed to identify these much needed markers. For example, when evaluating the effect of anti-EGFR therapies, these studies focused on downstream markers in the EGFR pathway or changes in the DNA of the EGFR gene, in particular, to identify good responding patients. Two studies showed more tumour regression in wild-type K-Ras patients than in mutant patients when EGFR inhibition was added to conventional CRT [91,92]. In addition, EGFR copy number seems to be a significant predictor of increased tumour regression . By the other hand molecular assessment of radiation-induced effects in rectal cancer identifies a heterogeneous pattern of response both in bio imaging and IHC identification [93,94] (Figure 3). The fact that none of these biomarkers is currently used in the clinical setting is probably not only because of a lack of relevance or reproducibility. Several other factors hamper the implementation of these markers. Ideally, the optimal moment to determine the status of a predictive marker would be on pre-treatment biopsies. Unfortunately, for research purposes, pretreatment biopsies are often not available or are very small. In posttreatment samples, (nearly) complete tumour regression or extensive damage to the tissue in response to therapy limits the availability of sufficient tumour material. Besides, almost all assays currently studied are expensive and time-consuming procedures. Because of the difficult technical aspects of the assays, the majority would have to be executed at a dedicated laboratory. These costs- and logistics related issues put a major constraint on the clinical applicability of these markers.
|Chang||130||IHC||50.4 Gy 5-
|Sturm||66||IHC||45 Gy Heat
shock hypertermia 5-FU+LV
|TNM downstaging||No correlation|
|Lin||70||IHC||45 Gy +/- 5-FU||TNM downstaging||Lack of p53 expression associated with poor response|
|Bertolini||91||IHC||50 Gy 5-FU||TRG; TNM
|Kudrimoti||17||IHC||50.4-59.4 Gy 5-
|pCR vs. PR||No correlation|
|Jakob||22||IHC||50.4 5-FU||TRG||No correlation|
|Terzi||37||IHC||45 Gy 5-FU||TRG; TNM
|Negri||57||IHC||40-45 Gy +/- 5-
|Moral||39||IHC||42 Gy 5-FU +
|TNM downstaging||No correlation|
|Tumor size||No correlation|
Table 1: p53 expression and prediction of response to preoperative radio chemotherapy in rectal cancer patients.
|pCR;DFS/OS;Metastasis-free survival||EGFR expression associated with decreased pCR rate|
|Kim||183||IHC||50 Gy 5-
|TRG; TNM downstaging||Low EGFR expression associated with TNM downstaging|
|TRG||EGFR Sp1-216 associated with tumor response|
|TRG||Combination of TS 2R/2R and EGFR 61A/G or EGFR Sp1-216T associated with tumor regresión|
|Bertolini||91||IHC||50 Gy 5-FU||TRG; TNM downstaginf;DFS/OS||No correlation|
|Toiyama||40||PCR/RNA||20 Gy 5-
|TNM downstaging||Low EGFR expression associated with high response rate|
|50.4 Gy 5-
|TRG||High EGFR and wild-type KRAS associated with response to treatment|
|Debucquoy||41||IHC||50.4 Gy 5-
|TRG; TNM downstaging||No correlation|
|Bengala||146||IHC;FISH;PCR/DNA||50 Gy 5-
|TRG;DFS/OS||No association of EGFR, GCN and KRAS with TRG and OS|
Table 2: Epidermal growth factor receptor (EGFR) expression and prediction of response to preoperative radio-chemotherapy in rectal cancer patients.
|Terrazino||125||PCR/DNA||45-50.4 Gy 5-FU or 5-FU+LV or 5-FU+Oxa or 5-FU+ carboplatin||TRG||No correlation|
|Bertolini||91||IHC||50 Gy 5-FU||TRG; TNM downstaging; OS/DFS||No correlation|
|Spindler||60||PCR/DNA||65 Gy UFT+LV||TRG||TS 2R/2R associated with tumor regression|
|Jakob||22||PCR/RNA||50.4 Gy 5-FU||TRG||Low TS expression associated with tumor regression|
|Stoehlmacher||40||PCR/DNA; PCR/RNA||50.4 Gy 5-FU||TRG||TS 3-UTR 6 bp deletion slightly associated with tumor response|
|Negri||57||IHC||40-45 Gy +/- 5-FU + oxa||pCR||High TS expression associated with higher rate of response|
|Kikuchi||60||IHC||45 Gy Irinotecan||TRG||Higher TS expression associated with better response|
|Carlomagno||46||IHC||45 Gy Capecitabine + Oxa||TRG||Low TS expression assocoated with low response|
|Hur||44||IHC; PCR/DNA||45 Gy 5-FU||TRG; TNM downstaging||Low-expression genotypes associated with TNM downstaging; no correlation with TRG|
|Paez||51||PCR/DNA||45 Gy 5-FU||TRG; TNM downstaging||Genotipe 3R/3R associated with tumor response|
Table 3: TS expression and prediction of response to preoperative radio-chemotherapy in rectal cancer patients.
Future initiatives should be aimed at the incorporation of prognostic and predictive markers derived from preoperative imaging, histopathological studies as well as molecular studies into one “tool.” To be able to base treatment decisions on this tool, it should be effective in determining disease stage and outcome, as well as easily applicable in current clinical practice. At this moment, the best example of such a comprehensive application has been developed by Valentini et al. . A model to predict LR, DR, and survival was established, based on individual patient data accrued in 5 European randomized trials. The developed nomogram is available online and can be used as a decision support tool to select patients for specific treatment and follow-up strategies. Differential expression in inflammatory-related genes after preoperative CRT in normal rectal tissue and rectal tumour may play a role in the tumour response and side effects to RT are warranted . Unfortunately, all models include pathologic staging and are, therefore, not suitable as pre-treatment decision support. These examples pointed out the new scenario of taking decision based on individual patient features exploiting the numbers of nomograms prediction models. Predictive nomograms such as these will complement existing Consensus or Guidelines reports. The latter identify the possible components of a multidisciplinary approach, the nomograms allow more patient specific selection from the options menu and possibility to share decision with patients based on an objective evaluation of risks. These prediction models are structured to allow the integration of new data from molecular imaging and biology, which will help to improve the reliability of their prediction. They also provide an objective subgroup identification to address clinical trial for more homogenous groups of patients. Clinicians must be aware of the limits of these prediction models related to the need of internal validation techniques and on the quality of the collected data.
The development of additional and alternative treatment strategies to standard TME surgery has increased the need for better patient selection. Nowadays, patient selection for treatment is a balanced process of weighing the pros and cons of each modality. This patienttailored approach demands the availability of reliable information about CRM and lymph node involvement, biological tumour behaviour, and the expected response to treatment. The future of the implementation of new treatment strategies lies, therefore, within the development of new prognostic and predictive markers and validation of known prognostic and predictive markers that can provide us with this information. For each risk category, different motives determine the consideration of treatment options and, therefore, every subgroup is in need of its own predictive and prognostic markers. In this review, we have provided an overview of the different treatment options and the role that prognostic and predictive markers (may) play. From the update of their current status, it is clear that prediction of rectal cancer is still in its infancy and requires more research. However, recent advances in molecular biology–based studies, imaging modalities and initiatives, such as the development of nomograms, are important steps in the right direction of optimal patient-tailored treatment.