Key words:  early detection, molecular diagnosis, prognosis, molecular staging, colorectal cancer

Supported in part by the Early Detection Research Network (EDRN) 1U24 CA86359-01.  Presented at the 2003 Annual Meeting, United States and Canadian Academy of Pathology

ABSTRACT:
The molecular features of colorectal neoplasia can be used in early detection of neoplastic processes, in diagnosis of specific subtypes of CRC, in determining the prognosis of CRC, surrogate endpoint biomarkers in therapeutic and preventive interventions, as well as in multiple other types of biomedical analyses.

Clinically the early detection of colorectal adenocarcinoma (CRC) has relied on the relatively insensitive and non-specific detection of occult blood or blood products in the stool or by the relatively expensive and invasive procedure of endoscopy.  Experimentally, the detection of mutations in p53, K-ras, and/or oncofetal proteins in stool have been used to detect CRC.  These techniques of molecular detection have been improved by the magnetic separation of epithelial cells from the stool and by evaluating multiple molecular markers in the separated cells.  With the development of surface enhanced laser desportion/ionization time of flight mass spectroscopy (SELDI-TOF-MS), early tumors may be detected based on fingerprints of proteins in samples of serum as well as in other non-invasive samples including stool.

The diagnosis of tumors (e.g., separation of mucinous versus non-mucinous CRC) has relied on the histopathologic pattern.  Such diagnoses also may be improved by the use of molecular features of the tumors.

The gold standard for clinical outcome of most cancers has been the clinical and pathologic staging of the tumors after surgery.  For CRC, nodal involvement at the time the primary tumor is resected has been the most reliable indicator of clinical outcome; however, we have reported combinations of molecular markers that are equivalent to pathologic or clinical staging of CRC in predicting clinical outcome.  In addition, molecular markers can be used in conjunction with clinical or pathologic staging to provide a stronger indicator of clinical outcome than staging alone.  We propose that "molecular staging" be added to pathologic staging to aid in predicting clinical outcome and to aid in therapeutic decisions for colorectal cancers, especially Stage II CRCs.

We have reported that the clinical usefulness of most molecular markers vary with the race of patients and the anatomic location of CRCs; this complicates the evaluation of diagnostic or prognostic biomarkers requiring much larger numbers of cases to be evaluated.  Nevertheless, nuclear accumulation of p53 (p53^nac) and phenotypic expression of Bcl-2, MUC-1 and p27^kip-1 may be molecular markers approaching acceptance for use in molecular staging of specific subgroups of colorectal cancers.  Similarly, these same molecular features are useful in separating mucinous from non-mucinous CRCs and may be complemented in multiple uses by protein-fingerprints determined by SELDI-TOF-MS.

INTRODUCTION:
The molecular features of colorectal neoplasia can be used in early detection of neoplastic processes, in diagnosis of specific subtypes of CRC, in determining the prognosis of CRC, as surrogate endpoints in therapeutic or preventive interventions as well as in multiple other types of biomedical analyses^1-5  .

The early detection of colorectal adenocarcinomas (CRCs) has relied clinically on the relatively insensitive and non-specific detection of occult blood or blood products in the stool or on the expensive and invasive method of endoscopy.  The detection in stool of atypical molecules examined in a large proportion of CRCs such as mutated K-ras⁶  or of oncofetal proteins^{e.g., carcinoembryologic antigen (CEA)}  may aid in the early detection of CRCs⁷  (7 reviewed in 3) but these atypical molecules also can be detected in other gastrointestinal malignancies⁸  as well as in very early preinvasive neoplastic lesions such as aberrant crypt foci^9,10  .  However, the sensitivity and specificity of such methods of molecular detection may be improved by concomitantly measuring multiple molecular markers on colorectal epithelial cells that have been separated from feces.

A more recent method which may prove both sensitive and specific for the early detection of colorectal neoplasia relies on the detection of the effects of neoplasia on molecular products in the serum as detected by surface enhanced laser desorption/ionization (SELDI) time of flight mass spectroscopy (TOF-MS).  The use of SELDI-TOF-MS has been reported to be both sensitive and specific for the detection of ovarian cancer¹¹  and of prostate cancer¹²  .

Molecular markers may also be useful in the diagnosis of subsets of CRCs.  For example, mucinous CRCs have been diagnosed rather arbitrarily as having approximately a 50% component of mucin.  Our results based on molecular markers suggest that a 10% component of mucin best correlates with the molecular phenotype of these lesions based on the expressions of MUC-1, MUC-2, p53^nac , Bcl-2, and p27^kip-1 (unpublished data).

Using molecular features to determine recurrence of CRC also strengthens the general approach of molecular diagnosis.  Typically, oncofetal proteins such as CEA have been measured in serum to detect recurrence of CRCs.  We propose that SELDI-TOF-MS may be a more sensitive and specific method of detecting CRC recurrence.

Clinical counseling of patients and their familiesÕ as well as therapeutic decisions have been based upon the pathologic and/or clinical stages of colorectal cancers (CRCs).  For example, adjuvant therapy with 5-fluorouracil alone or in several combinations and routes of administration with Leucovorin, Irinotecan or Oxaliplatin is increasingly used for advanced CRCs^13-16  .  In the future, the molecular phenotype of CRCs may be used to determine the "molecular stage" of CRCs and thus to aid in counseling and therapeutic decisions, especially if adjuvant therapy is to be used for aggressive subgroups of Stage I or II lesions.

Several molecular phenotypes have been reported to be associated with aggressive subtypes of CRCs including some molecular markers that are independent of pathologic stage.  These include the nuclear accumulation of p53 as identified by immunohistochemistry (p53^nac) and the phenotypic expression of Bcl-2, MUC-1 (mucin core protein) or p27^kip-1, a cell cycle inhibitor.

The manuscript reviews the status of multiple molecular markers and discusses the potential future uses of these markers in the early detection of CRC, in diagnosis of subtypes of CRC, and in determining the prognosis of patients with CRCs (molecular staging).

A relatively new method for the early detection of neoplastic processes that relies on detecting specific protein patterns or protein fingerprints in serum or other biological fluids/tissues has been used successfully in the early detection of cancers of the ovary and prostate.^{11, 12}    The detection of such "protein fingerprints" using surface enhanced laser desorption/ionization time-of-flight mass spectrometry relies on using a group of patented ProteinChip^ñ arrays from Ciphergen, Inc. Each array consists of modular interlocking, finger-length strips of metal coated at eight sample sites with an active chemical substance with classical properties of chromatography agents.  Proteins adhere based upon, for example, hydrophilic or hydrophobic phases, ion exchange properties or immobilized metal chelation.  In addition, protein chip arrays with pre-activated surfaces permit the covalent immobilization of antibodies, receptors, DNA, glycoproteins, etc., which provide for affinity capture of molecules in samples.  Energy-adsorbing molecules (EAM) are subsequently overlaid at each of the eight sample sites with the bound proteins.  The chip is placed in the SELDI chip reader and is pulsed with a laser causing the UV-absorbent EAM to be ejected, carrying the protein molecules.  The predominately single charged ions of peptides and proteins are analyzed by a time-of-flight mass spectrometer that separates the proteins based on velocity (Figure 1).  The TOF spectrum for different ions is recorded and converted into a mass/charge spectrum.  Lighter ions reach the detector more quickly than heavier ions.  The TOF spectrum for different ions is recorded and converted into a mass spectrum.  There are 100 addressable regions on each sample site and each addressable region can be irradiated multiple times by the laser.  With the aid of SELDI-TOF-MS software, a retentate map is generated depicting the mass/charge, which in most cases corresponds to the molecular weight/charge and a matching amplitude that is correlated with the amount of protein based to the sampled site.  These resulting protein profiles can be evaluated using many different types of classification algorithms.¹² 

In an initial collaborative study between the University of Alabama (UAB) and Eastern Virginia Medical Center (EVMS), we evaluated, using SELDI-TOF-MS, 66 cases of serum from patients who were confirmed pathologically to have either cancer¹⁷  or other gastrointestinal pathological abnormalities¹⁸  .  The other category included diseases such as diverticulitis, inflammatory bowel disease, and other gastrointestinal conditions serious enough to require surgery.  A copper-chelate chip surface was loaded with 20 ul of serum using previously optimized conditions,¹²  and resulting protein profiles were evaluated using a decision-tree classification algorithm. Representative TOF spectra for peptides/proteins in the 1-15 kDa range obtained from the different serum samples are shown in Figure 2.  Of the non-cancer cases, only 2 of 42 were misclassified, and of the cancer cases 5 of 24 were misclassified, resulting in a 95% sensitivity and 79% specificity. These initial results are encouraging enough to pursue SELDI-TOF-MS analysis of a much larger data set in order to allow the development of a more sophisticated discrimination algorithm that a larger series would support.

Typical spectra are shown in Figure 2.  Of the non-cancer cases, only 2 of 42 were misclassified and of the cancer cases 5 of 24 were misclassified (unpublished data).  Thus, the initial study was encouraging with a 95% sensitive and 79% specificity even though it did not permit the development of a sophisticated discrimination algorithm that a larger series would support.

Several subtypes of CRCs have been proposed based on their histopathologic pattern including medullary-type, lymphoepitheliomatous type and mucinous adenocarcinoma.  Of interest is that the lymphoepithelomatous type of CRC has been associated with Epstein-Barr (EB) viral infections while the medullary type has been associated with micro-satellite instability (MSI).  Both of these tumors which may overlap have a better prognosis than CRCs with more typical patterns of histopathology.

The prognostic importance of classifying CRCs as to their mucin content is controversial with some studies indicating that mucinous tumors have a worse prognosis than non-mucinous CRCs^19-21  ; however, while others a better prognosis^{22, 23}  .

The definition of mucinous tumors is arbitrary with most studies using ³ 50% mucin content to classify tumors as mucinous.  We have correlated the extent of mucin in tumors with p53^nac and the phenotypic expression of multiple molecular markers including MUC-1, MUC-2, Bcl-2, and p27 ^kip-1.  These molecular markers demonstrate that phenotypic differences correlate with a mucin content of 10-20% rather than 50% (unpublished data).

Colorectal cancer characteristically kills patients by metastatic destruction of the liver and/or by obstruction of the gastrointestinal tract usually by adhesions in the peritoneal cavity.  The primary causes of death can be viewed as shown in figure 3.

Because the primary pathway causing death secondary to CRC is via lymphatic spread, it is not surprising that in the majority of studies involvement by tumor of lymph nodes has been identified statistically as the gold standard with respect to determining prognosis of patients with CRCs.  Similarly, many molecular markers that predict clinical outcome might not be statistically independent of pathologic stage or the nodal component of stage (pN) because involvement of lymph nodes is in the main pathway by which CRCs cause the death of patients.  Thus, in multivariate analysis to determine whether molecular markers predict prognosis, the predictive value of a molecular marker may be lost as independent variable (i.e., a molecular marker may provide the same prognostic information as the nodal component of stage).  Such an interaction between the lymph node involvement and the expression of Bcl-2 in CRC was demonstrated in our recent study²⁴  .

Even though a molecular marker may provide the same prognostic information as stage or a component of stage such a molecular marker may still be useful clinically; for example, the molecular marker can be evaluated on the initial diagnostic biopsy of the tumor and may indicate the aggressiveness of the tumor well before the tumor is resected or before the pathologic evaluation of the resected specimen or the clinical staging of the patient. Similarly, molecular markers that are correlated strongly with stage and/or progression may be useful as molecular markers whose changes with therapeutic or preventive interventions indicate successful interventions (i.e., surrogate endpoint biomarkers).  As our approaches to prevention and/or therapy of tumors advance, including the increased use of novel therapies, e.g., immunotherapy and genetic therapy, it may be important before definitive surgery to be able to predict the aggressiveness of CRCs.  Such information may be useful in counseling patients/families and in selecting, planning or scheduling novel therapies, especially therapies with severe complications.  In addition, some of these molecular markers may aid in predicting responses to specific therapies as the expression of p185^erbB-2 predicts the response of ductal adenocarcinomas of the breast to adramycin^{25, 26}  and to HERCEPTIN^{25, 27} 

The most valuable prognostic molecular markers are those that provide information as to the clinical aggressiveness of tumors beyond that indicated by the stage or components of stage of tumors. Probably, such molecular markers are related more to the biology of the tumor than just to how advanced the lesion is at surgery and/or to the pathways tumors most commonly follow during widespread dissemination.  An example of a molecular pathway affecting the aggressiveness of tumors is that tumors of the mutator phenotype pathway which exhibit microsatellite instability behave less aggressively clinically than other CRCs even though they tend to be of greater size when surgically removed.

Prognostic molecular markers independent of stage not only can provide important information prior to surgical removal and staging of the tumor, but such markers may add an additional informational component to stage - a molecular component which can be added to pathologic/clinical stage in the calculation of hazard ratios.  In addition, such molecular markers can provide aid in the decision concerning whether or not to use adjuvant and/or novel therapies for Stage II lesions.

To date, several molecular markers including p53, p27^kip-1, Bcl-2 and MUC-1 show promise as being useful prognostically in CRCs, but each of these markers awaits more complete validation.  In this manuscript, we review the recent advances in evaluating the prognostic usefulness of these molecular markers alone and in combination.  It is our view that no one molecular marker will be adequate for evaluating all CRCs.  Rather, groups of molecular markers will be useful for specific subsets of CRCs as subdivided by patient race and/or anatomic location of the primary tumor.

While the molecular phenotypes of early lesions of hereditary colorectal neoplasia have been described as part of either the suppressor gene pathway of familial adenomatosis polyposis coli or mutator phenotype causing microsatellite instability, few advances have been made in the phenotypic description of sporadic colorectal cancers; especially molecular changes involved in the invasive stages of these lesions.  Our goal has been to identify reliable molecular information that can be used in early detection of colorectal neoplasia, in diagnosis of CRC subtypes and in predicting clinical outcome of CRCs; specifically to characterize the components of molecular phenotypes of CRCs that identify aggressive subtypes of CRCs.

One of the approaches that has been used to improve the diagnosis of tumor subtypes or to identify more aggressive subtypes is to use gene arrays to identify forms of mRNA that may identify genes that are differentially expressed among subtypes.  Another approach is to predict based on the literature and the biological pathways of cellular biology which molecular features are likely to be important in, for example, progression.  Such pathways include those involved in apoptosis (e.g., Bax, Bcl-2, p53, Caspase 3, TUNEL), in proliferation (e.g., Ki-67, p27 ^kip-1 , cyclin D, cyclin E), in signal transduction (TGFα, EGFR, p185^erbB-2 ), in cellular adhesion (ECAM, PCAM, MUC-1, MUC-2), in immunoregulation (MUC-1), and in oncofetal transformation (TAG-72, CEA).  Until we began to use SELDI-TOF-MS, our approach has been the latter and our results related to prognosis using this approach are reviewed subsequently.

Our laboratory has described phenotypic molecular patterns of invasive colorectal neoplasia that are correlated with clinical outcomes.  Unfortunately, the phenotypic patterns are complicated because the usefulness of molecular phenotypes in predicting clinical outcomes varies with race of the patient and anatomic location of the tumor.  Variation of molecular markers with racial groups is not unexpected because diet is likely to be important in the pathogenesis of CRCs and dietary norms vary with race²⁸  .  Anatomic location is also an expected variable because of differing embryologic origins, vascular and lymphatic patterns, mucosal functions, and phenotypic patterns of epithelial biomarkers in normal mucosa (reviewed in Grizzle et al. ²⁸  ).  A practical problem is that the variations of biomarkers with race and anatomic location of the tumor affect the numbers of specimens that must be studied in validation studies of molecular markers.  For example, racial and ethnic differences will require ultimately in the U.S. population studies of Caucasian, African-American, Native American, Asian Ð Indian Subcontinent, Asian Ð Chinese/Japanese/other and Hispanic sub-populations as a minimum.  Similarly, because of differing dietary requirements, religious subgroups may require separate study, as do some groups with relatively uniform genetic patterns (e.g., Ashkenazi Jews in the study of Breast or Ovarian cancers).

In Caucasian patients, the major molecular phenotype of aggressive tumors located in the proximal colon includes p53^nac, Bcl-2, MUC-1 and p27^kip-1.  Molecular phenotypes associated with aggressive subtypes of CRC based on race and anatomic locations are specified in Table 1.

Nuclear accumulation of p53:
Using our methods of immunohistochemistry, colorectal tumors whose cells demonstrate nuclear accumulation of p53 (p53^nac) typically have single point mutations in p53 when the complete p53 gene is sequenced or is analyzed by SSCP (Grizzle, et al ²  ).  In our initial studies, we reported that p53^nac in our overall patient population was associated with a poor prognosis; however, on subsequent more detailed study, we identified that p53^nac was important prognostically primarily in proximal tumors of Caucasian patients²⁹  (Figure 4).  The effect of anatomic location of tumors on the interaction of molecular markers with clinical outcome suggests that proximal tumors are biologically different from distal tumors^{30, 31}  .  This also is supported by the effect of the phenotypic expression of Bcl-2 on prognosis.  The literature is very confusing with respect to the prognostic importance of p53 in CRCs as reviewed by Manne et al.³²  , Grizzle et al.^{4, 28}  , and McLeod and Murray³³  .  This confusion may have resulted in part because of variations of p53^nac with complex populations including differences among populations in race and anatomic location of tumors as well as by the methods used to analyze and evaluate the nuclear accumulation of p53.  The method of analysis of p53 is especially important since different points of cut-off in evaluations may vary extensively depending upon whether or not antigen recovery methods are used prior to immunostaining^{2, 4, 24, 28-33}  . 

Bcl-2: 
When the effect of phenotypic expression of Bcl-2 on prognosis was evaluated, our laboratory initially found in our complete population that the phenotypic expression of Bcl-2 was important prognostically, and that the combination of p53^nac plus Bcl-2 was even more useful prognostically³²  .

In a more detailed study in a larger population, we subsequently found that Bcl-2 was important in CRCs from primarily Caucasian populations.  The strongest correlation of the expression Bcl-2 with clinical outcome was initially reported in distal CRCs;³²  however, analysis in a larger Caucasian population (n = 299), Bcl-2 was found to be important prognostically in both distal and proximal CRCs (Figure 5).

In an initial group of African-Americans, Bcl-2 was found to be useful prognostically only in distal tumors.  This group of patients (n = 103) had a number of patients with a relatively good outcome (median survival = 49 months).  In a subsequent randomly selected patient population (n = 201), the median survival was 35 months.  In this group with a poor clinical outcome we could not find a statistical association of phenotypic expression of Bcl-2 and clinical outcome.  Because of variation with race, the combinations of p53^nac and Bcl-2 were an important prognostic combination only in overall Caucasians²⁴  as well as in proximal and distal subgroups.

Several oncogenes and tumor suppressor genes such as Bcl-2 and p53 are involved in regulating programmed cell death and cellular proliferation.  The dysregulation and/or alteration of Bcl-2 and p53 genes have been identified during the development and progression of CRCs^34-39  .  In addition, lack of Bcl-2 expression has been correlated with local invasion by tumors, metastasis and recurrence in CRC^{17, 40, 41}  .  Patients with CRCs who exhibit high levels of Bcl-2 expression have been reported to have a good clinical outcome^{37, 42-46}  ; however, some studies have not observed such an association^{17, 35, 47, 48}  .  Conversely, in a small group of CRC patients (n=48), Bhatavdekar et al.⁴⁹  correlated Bcl-2 expression with poor prognosis.  Although the reasons for this controversy (e.g., ethnic/religious and dietary differences) are not known, we hypothesize that the prognostic importance of Bcl-2 expression in CRCs may be limited to specific subgroups of patients similar to the clinical usefulness of p53^nac in CRC which may vary with racial/ethnic groups.

To further clarify the prognostic usefulness of Bcl-2 in CRCs, we performed a meta-analysis of the literature evaluating the relationship of Bcl-2 expression and overall survival.  Specifically we requested original data on all comparable studies from the corresponding authors that had evaluated previously the prognostic usefulness of Bcl-2 in predicting overall survival.  Several authors for various reasons could not supply us with their data.  Ultimately, 9 studies incorporating a total of 1989 patients provided either the hazard ratio (HR) and 95% confidence interval (95% CI) or information from which the HR and its variance could be estimated.  The meta analysis of these studies indicates that Bcl-2 is a useful prognostic marker for CRCs.  The results of the meta analysis are outlined in Table 2.

MUC-1:
One of the characteristic features of glandular epithelial tissues is synthesis and secretion of mucins, which are large glycoproteins that play important roles in protecting epithelial surfaces. Alterations in mucins with regard to the rate of their production and the extent of their glycosylation have been reported in several human malignancies⁵⁰  , including colorectal neoplasia^{51, 52}  . Among the several mucin antigens, MUC-1 is the best characterized. Several earlier studies in colorectal neoplasia have demonstrated that higher expression of MUC-1 was correlated with increased incidence of regional lymph node metastasis and liver metastasis^53-55  . Although expression of MUC-1 has been associated with the aggressiveness of CRCs, its prognostic usefulness in colorectal neoplasia has not been evaluated adequately. Studies from our laboratory⁵³  as well as others^{54, 55}  have suggested that increased expression of the core peptide of MUC-1 is associated with a poor prognosis in CRC.

We have evaluated the prognostic importance of MUC-1 and MUC-2 in CRCs.  We found that the phenotypic expression of MUC-2 was not useful prognostically in CRCs, but that the expression of both MUC-1 and MUC-2 were useful markers in defining the mucinous subtype of CRCs.  In contrast, the phenotypic expression of MUC-1 was important prognostically in CRCs of Caucasians but not of African-Americans (Figure 6A).  We found no variation of the prognostic importance of MUC-1 based on the anatomic location of the tumor.⁵³ 

Just like Bcl-2, MUC-1 interacts prognostically with p53^nac so that in Caucasians the combination of p53^nac plus MUC-1 is a useful prognostic combination (Figure 7).

Because both MUC-1 and p53 are important prognostically in Caucasian patients with CRCs, we have evaluated the association of the combination of MUC-1, Bcl-2 and p53^nac with prognosis in CRCs. The results are demonstrated in Figure 8.

Proliferation and p27:
p27^kip-1inhibits the activity of other cyclin-dependent kinases ( cdks), and like p21^waf-1, plays a key role in preventing progression into S phase of the cell cycle. Decreased p27^kip-1protein expression has been associated with large size CRCs, with positive lymph nodes, and with poor patient survival^56-58  . Furthermore, a small study (n=41) demonstrated that p27^kip-1 is a predictive indicator for tumor metastasis and patient clinical outcome in right-sided colon tumors¹⁸  . Based on the results of others, we considered that p27^kip-1 would be a likely prognostic biomarker in CRCs.  We performed a preliminary analysis of the importance of the proliferative index biomarkers Ki67/MIB-1 and p27^{kip-1 1} in 48 African-American and in 54 Caucasian patients with CRCs which had been analyzed previously for p53. In this study we observed similar proportions of CRCs, collected from African-Americans and Caucasians, exhibited increased expression of p27^kip-1 (50% & 54%, respectively) and Ki67 (54% & 50%, respectively). Univariate Kaplan-Meier survival analyses demonstrated that African-Americans with CRCs exhibiting higher expression of p27^kip-1 without p53^nac had marginally better overall survival (log rank test, P = 0.055) than any other combination of p27^kip-1 and p53^nac.  In Caucasians, the lower expression of p27^kip-1 with p53^nac demonstrated the lowest probability of overall survival (log rank test, P = 0.011). No prognostic value was found for p27 or Ki-67 alone or the combination of Ki-67 with p27 or p53^nac.  Overall, our studies suggested that higher expression of p27^kip-1, and lack of p53^nac in CRCs is a valuable indicator of good prognosis in both African-Americans and Caucasians (Manne et al. unpublished findings). However, further studies are needed to understand the prognostic importance of p27^kip-1in CRC based on the anatomic location of the tumor and the patient ethnicity.

Molecular Staging Using a Combination of Molecular Markers:
It is unlikely that only one or two molecular markers will be useful in developing an approach to molecular staging of CRC; rather a combination of multiple molecular markers may be necessary.  Ultimately p27^kip-1 will be one of these markers; however, due to our limited data on p27^kip-1, to date, for Caucasians this combination would be limited to p53, MUC-1 and Bcl-1.  Similarly because of the limitations of our data we must lump both proximal and distal tumors together.  The survival analysis for this combination is shown in Figure 8 in which the best phenotype (p53nac negative, MUC-1 negative, Bcl-2 positive) is compared with the worst (p53nac positive, MUC-1 positive, Bcl-1 negative) phenotype from the standpoint of survival.

Demographic and Other Features and Models of Prognosis
In developing computer models that can predict the aggressiveness of CRCs, we identified that age ( £ 65 years) of patients had a very important impact on clinical outcome of patients with CRCs^{59 2001}  .  Thus, not only molecular features but also demographic features may impact the clinical outcome of patients with CRCs.  We look forward to developing models of CRC in which age and other demographic features together with pathologic, molecular and clinical stages can be used to predict accurately the clinical outcome of patients with all cancers as well as to aid in selecting specific therapies for subgroups of cancers.

SUMMARY:
The use of molecular and other markers for the early detection of colorectal cancer or for the identification of recurrence of CRC is advancing rapidly with new methods based on analysis of stool and of serum; these include molecular analysis of the separated cellular components of stool for multiple mutations and for multiple oncofetal molecules.  Another potentially exciting approach is the use of SELDI-TOF-MS to identify protein-fingerprints in serum that correlate with the absence or presence of CRC in patients from whom serum samples were obtained.

Molecular markers also are being identified that aid in a more accurate diagnostic separation of tumor subtypes such as the mucinous colorectal tumor.  Similarly, more aggressive subtypes of CRCs can be identified by using selected molecular markers including p27 ^kip-1 , MUC-1, p53^nac, and Bcl-2.  The molecular features vary in their prognostic usefulness based on race/ethnicity and on anatomic location of the tumor.  In combination these molecular markers are more useful than nodal status in predicting clinical outcome and can be used together with stage to predict clinical outcome.  Similarly, we expect that protein-fingerprints as determined by SELDI-TOF-MS may prove useful in identifying patients with new and/or aggressive subtypes of CRC.

References

1. Grizzle, W.E., R.B. Myers, and U. Manne, The use of biomarker expression to characterize neoplastic processes. Biotech Histochem, 1997. 72(2): p. 96-104.
2. Grizzle, W.E., et al., Immunohistochemical evaluation of biomarkers in prostatic and colorectal neoplasia., in John Walker's Methods in Molecular Medicine-Tumor Marker Protocols., M. Hanausek and Z. Walaszek, Editors. 1998, Humana Press: Totowa, NJ. p. 143-160.
3. Grizzle, W.E., D. Shibata, and U. Manne, Molecular and histopathologic changes in the development of colorectal neoplasia., in Molecular pathology of early cancer, S. S, H. D, and A.F. Gazdar, Editors. 1999, IOS Press: Amsterdam. p. 197-220.
4. Grizzle, W.E., et al., The molecular characterization of colorectal neoplasia in translational research. Archives of Pathology and Laboratory Medicine, 2001. 125(1): p. 91-98.
5. Grizzle, W.E., et al., Molecular staging of colorectal cancer in African-American and Caucasian patients using phenotypic expression of p53, BCL-2, MUC-1 and p27^Kip-1. Int J Cancer, 2002. 97(3).
6. Forrester, K., et al., Detection of high incidence of K-ras oncogenes during human colon tumorigenesis. Nature, 1987. 327(6120): p. 298-303.
7. Sidransky, D., et al., Identification of ras oncogene mutations in the stool of patients with curable colorectal tumors. Science, 1992. 256(5053): p. 102-5.
8. Caldas, C., et al., Detection of K-ras mutations in the stool of patients with pancreatic adenocarcinoma and pancreatic ductal hyperplasia. Cancer Res, 1994. 54(13): p. 3568-73.
9. Smith, A.J., et al., Somatic APC and K-ras codon 12 mutations in aberrant crypt foci from human colons. Cancer Res, 1994. 54(21): p. 5527-30.
10. Pretlow, T.P., et al., K-ras mutations in putative preneoplastic lesions in human colon {see comments}. Journal of the National Cancer Institute, 1993. 85(24): p. 2004-7.
11. Petricoin, E.F., et al., Use of proteomic patterns in serum to identify ovarian cancer. Lancet, 2002. 359(9306): p. 572-7.
12. Adam, B.L., et al., Serum protein fingerprinting coupled with a pattern-matching algorithm distinguishes prostate cancer from benign prostate hyperplasia and healthy men. Cancer Res, 2002. 62(13): p. 3609-14.
13. de Gramont, A., et al., Leucovorin and fluorouracil with or without oxaliplatin as first-line treatment in advanced colorectal cancer. Journal of Clinical Oncology, 2000. 18(16): p. 2938-47.
14. Bogliolo, G., et al., Advanced colorectal cancer: quality of life and toxicity in patients after weekly 24-hour continuous infusions of biomodulated 5-fluorouracil. Anticancer Research, 2000. 20(1B): p. 501-4.
15. Giacchetti, S., et al., Phase III multicenter randomized trial of oxaliplatin added to chronomodulated fluorouracil-leucovorin as first-line treatment of metastatic colorectal cancer. Journal of Clinical Oncology, 2000. 18(1): p. 136-47.
16. Saltz, L.B., et al., Irinotecan plus fluorouracil and leucovorin for metastatic colorectal cancer. Irinotecan Study Group {see comments}. New England Journal of Medicine, 2000. 343(13): p. 905-14.
17. Giatromanolaki, A., et al., Combined role of tumor angiogenesis, bcl-2, and p53 expression in the prognosis of patients with colorectal carcinoma. Cancer, 1999. 86(8): p. 1421-30.
18. Liu, D.F., et al., p27 cell-cycle inhibitor is inversely correlated with lymph node metastases in right-sided colon cancer. Journal of Clinical Laboratory Analysis, 1999. 13(6): p. 291-5.
19. Pihl, E., et al., Mucinous colorectal carcinoma: immunopathology and prognosis. Pathology, 1980. 12(3): p. 439-47.
20. Umpleby, H.C., D.L. Ranson, and R.C. Williamson, Peculiarities of mucinous colorectal carcinoma. Br J Surg, 1985. 72(9): p. 715-8.
21. Nozoe, T., et al., Clinicopathological characteristics of mucinous carcinoma of the colon and rectum. J Surg Oncol, 2000. 75(2): p. 103-7.
22. Halvorsen, T.B. and E. Seim, Degree of differentiation in colorectal adenocarcinomas: a multivariate analysis of the influence on survival. J Clin Pathol, 1988. 41(5): p. 532-7.
23. Hermanek, P., I. Guggenmoos-Holzmann, and F.P. Gall, Prognostic factors in rectal carcinoma. A contribution to the further development of tumor classification. Diseases of the Colon & Rectum, 1989. 32(7): p. 593-9.
24. Manne, U., H. Weiss, and W.E. Grizzle, Bcl-2 expression is associated with improved prognosis in patients with distal colorectal adenocarcinomas. International Journal of Cancer, 2000. 89: p. 423-30.
25. Ross, J.S. and J.A. Fletcher, The HER-2/neu oncogene in breast cancer: prognostic factor, predictive factor, and target for therapy. Stem Cells, 1998. 16(6): p. 413-28.
26. Paik, S., et al., erbB-2 and response to doxorubicin in patients with axillary lymph node-positive, hormone receptor-negative breast cancer {see comments}. Journal of the National Cancer Institute, 1998. 90(18): p. 1361-70.
27. Kumamoto, H., et al., Chromogenic in situ hybridization analysis of HER-2/neu status in breast carcinoma: application in screening of patients for trastuzumab (Herceptin) therapy. Pathol Int, 2001. 51(8): p. 579-84.
28. Grizzle, W.E., et al., Molecular and histopathologic changes in the development of colorectal neoplasia., in Molecular Pathology of Early Cancer, S. Srivastava, D. Henson, and A. Gazdar, Editors. 1999, IOS Press: Amsterdam, Netherlands. p. 197-220.
29. Manne, U., et al., Nuclear accumulation of p53 in colorectal adenocarcinoma: prognostic importance differs with race and location of the tumor. Cancer, 1998. 83: p. 2456-67.
30. Sinicrope, F.A., et al., Apoptotic and mitotic indices predict survival rates in lymph node-negative colon carcinomas. Clinical Cancer Research, 1999. 5(7): p. 1793-804.
31. Kern, S.E., et al., Clinical and pathological associations with allelic loss in colorectal carcinoma {published erratum appears in JAMA 1989 Oct 13;262(14):1952}. Journal of the American Medical Association, 1989. 261(21): p. 3099-103.
32. Manne, U., et al., Prognostic significance of Bcl-2 expression and p53 nuclear accumulation in colorectal adenocarcinoma. International Journal of Cancer, 1997. 74(3): p. 346-58.
33. McLeod, H.L. and G.I. Murray, Tumour markers of prognosis in colorectal cancer. British Journal of Cancer, 1999. 79(2): p. 191-203.
34. Hague, A., et al., BCL-2 expression in human colorectal adenomas and carcinomas. Oncogene, 1994. 9(11): p. 3367-70.
35. Bosari, S., et al., bcl-2 oncoprotein in colorectal hyperplastic polyps, adenomas, and adenocarcinomas. Human Pathology, 1995. 26(5): p. 534-40.
36. Sinicrope, F.A., et al., bcl-2 and p53 oncoprotein expression during colorectal tumorigenesis. Cancer Research, 1995. 55(2): p. 237-41.
37. Baretton, G.B., et al., Apoptosis and immunohistochemical bcl-2 expression in colorectal adenomas and carcinomas. Aspects of carcinogenesis and prognostic significance. Cancer, 1996. 77(2): p. 255-64.
38. Hao, X.P., M. Ilyas, and I.C. Talbot, Expression of Bcl-2 and p53 in the colorectal adenoma-carcinoma sequence. Pathobiology, 1997. 65(3): p. 140-5.
39. Yang, H.B., et al., The role of bcl-2 in the progression of the colorectal adenoma-carcinoma sequence. Anticancer Research, 1999. 19(1B): p. 727-30.
40. Ilyas, M., et al., Loss of Bcl-2 expression correlates with tumour recurrence in colorectal cancer. Gut, 1998. 43(3): p. 383-7.
41. Ishijima, N., et al., The immunohistochemical expression of BCL-2 oncoprotein in colorectal adenocarcinoma. Surgery Today, 1999. 29(7): p. 682-4.
42. Sinicrope, F.A., et al., Prognostic value of bcl-2 oncoprotein expression in stage II colon carcinoma. Clinical Cancer Research, 1995. 1(10): p. 1103-10.
43. Ofner, D., et al., Immunohistochemically detectable bcl-2 expression in colorectal carcinoma: correlation with tumour stage and patient survival. British Journal of Cancer, 1995. 72(4): p. 981-5.
44. MacLean, G.D., M.A. Reddish, and B.M. Longenecker, Prognostic significance of preimmunotherapy serum CA27.29 (MUC-1) mucin level after active specific immunotherapy of metastatic adenocarcinoma patients. Journal of Immunotherapy, 1997. 20(1): p. 70-8.
45. Kaklamanis, L., et al., Bcl-2 protein expression: association with p53 and prognosis in colorectal cancer. British Journal of Cancer, 1998. 77(11): p. 1864-9.
46. Buglioni, S., et al., Evaluation of multiple bio-pathological factors in colorectal adenocarcinomas: independent prognostic role of p53 and bcl-2. International Journal of Cancer, 1999. 84(6): p. 545-52.
47. Schneider, H.J., et al., Bcl-2 expression and response to chemotherapy in colorectal adenocarcinomas. British Journal of Cancer, 1997. 75(3): p. 427-31.
48. Tollenaar, R.A., et al., Immunohistochemical detection of p53 and Bcl-2 in colorectal carcinoma: no evidence for prognostic significance. British Journal of Cancer, 1998. 77(11): p. 1842-7.
49. Bhatavdekar, J.M., et al., Coexpression of Bcl-2, c-Myc, and p53 oncoproteins as prognostic discriminants in patients with colorectal carcinoma. Diseases of the Colon & Rectum, 1997. 40(7): p. 785-90.
50. Jass, J.R. and A.M. Roberton, Colorectal mucin histochemistry in health and disease: a critical review. Pathology International, 1994. 44(7): p. 487-504.
51. Ho, S.B., et al., Heterogeneity of mucin gene expression in normal and neoplastic tissues. Cancer Research, 1993. 53(3): p. 641-51.
52. Hanski, C., et al., Altered glycosylation of the MUC-1 protein core contributes to the colon carcinoma-associated increase of mucin-bound sialyl-Lewis(x) expression. Cancer Research, 1993. 53(17): p. 4082-8.
53. Manne, U., H. Weiss, and W.E. Grizzle, Racial differences in the prognostic usefulness of MUC1 and MUC2 in colorectal adenocarcinomas. Clinical Cancer Research, 2000. 6: p. 4017 - 4025.
54. Aoki, R., et al., MUC-1 expression as a predictor of the curative endoscopic treatment of submucosally invasive colorectal carcinoma. Diseases of the Colon & Rectum, 1998. 41(10): p. 1262-72.
55. Hiraga, Y., et al., Immunoreactive MUC1 expression at the deepest invasive portion correlates with prognosis of colorectal cancer. Oncology, 1998. 55(4): p. 307-19.
56. Loda, M., et al., Increased proteasome-dependent degradation of the cyclin-dependent kinase inhibitor p27 in aggressive colorectal carcinomas {see comments}. Nature Medicine, 1997. 3(2): p. 231-4.
57. Jessup, J.M., et al. Multivariate analysis of tissue-based prognostic markers in stage II-III colorectal carcinoma. in Proc Annu Meet Am Soc Clin Oncol. 1997.
58. Tenjo, T., et al., Prognostic significance of p27^Kip1 protein expression and spontaneous apoptosis in patients with colorectal adenocarcinomas. Oncology, 2000. 58(1): p. 45-51.
59. Jhala, N.C., et al. Nuclear accumulation of p53 in colorectal adenocarcinoma: prognostic importance differs with race, age and location of the tumor. in Proc 92nd Annu Meet AACR. March 24 - 28, 2001. New Orleans, LA.